NOUVELLES APPLICATIONS DE LA SPECTROSCOPIE EN PROCHE INFRAROUGE POUR
L’ÉVALUATION DE L’ÉCOLOGIE NUTRITIONNELLE DU CERF DE VIRGINIE
par
Pierre-Olivier Jean
thèse présentée au Département de biologie en vue
de l’obtention du grade de docteur ès sciences (Ph.D.)
FACULTÉ DES SCIENCES
UNIVERSITÉ DE SHERBROOKE
Sherbrooke, Québec, Canada
Septembre 2015
Le 15 septembre 2015
Le jury a accepté le mémoire de Monsieur Pierre-Olivier Jean
dans sa version finale
Membres du jury
Professeur Robert L. Bradley
Directeur de recherche
Département de Biologie, Université de Sherbrooke
Professeur Jean-Pierre Tremblay
Co-directeur de recherche
Département de Biologie, Université Laval
Professeur William J. Foley
Évaluateur externe
Australian National University, Canberra, Australie
Robert Berthiaume
Évaluateur interne
Valacta/Agriculture Canada
Professeur Mark Vellend
Président-rapporteur
Département de Biologie, Université de Sherbrooke
SOMMAIRE
L’écologie nutritionnelle des populations de cervidés revêt une grande importance en contexte
de population surabondante.
L’intense pression d’herbivorie de certaines populations de
cervidés nuît à la régénération de la flore endémique et entraîne une baisse de la biodiversité.
On explique encore mal la sélection alimentaire chez ces herbivores, ce qui nuit d’une part à
notre capacité de prédire d’éventuels changements dans les trajectoires floristiques, et d’autre
part à notre capacité d’anticiper la réponse démographique de ces herbivores. Or, notre capacité
à acquérir des connaissances sur l’écologie nutritionnelle des herbivores peut être limitée par
les contraintes logistiques et méthodologiques nécessaires à l’évaluation de la qualité
nutritionnelle dans un contexte d’étude écologique, de même que par un manque d’objectivité
dans la définition de devises alimentaires appropriées pour l’animal d’étude. Les principaux
objectifs de mon projet de doctorat sont donc de développer de nouvelles méthodologies dans
l’étude des interactions entre un grand cervidé, le cerf de Virginie (Odocoileus virginianus) et
son habitat, et d’implémenter ces nouveux outils dans le cadre d’études écologiques.
J’ai effectué mes recherches à l’île d’Anticosti, qui est un laboratoire naturel bien adapté à
l’étude des populations abondantes de cervidés. Le cerf de Virginie y a été introduit en l’absence
de prédateurs naturels il y a plus de 100 ans, et cette population de cervidés est l’une des les plus
denses au monde. La forte pression d’herbivorie a profondément changé la composition
floristique sur l’île. Puisque l’industrie forestière et le tourisme issu de la chasse au cerf y sont
les deux activités économique les plus importantes, l’étude des interactions cerf-forêt y est
cruciale, tant sur le plan scientifique que pour le développement régional.
Cette thèse se décline en 4 chapîtres centraux (i.e. chapitres 1,2, 3 et 4) écrits sous formes
d’articles. Les deux premiers articles sont consacrés à l’élaboration de nouvelles méthodes
i
facilitant le monitoring de la nutrition du cerf de Virginie à l’île d’Anticosti. Dans un premier
temps (chapitre 1), j’utilise des modèles prédictifs utilisant la spectroscopie dans le proche
infrarouge ainsi que certaines propriétés chimiques des fèces de cerf afin de prédire son régime
alimentaire. Je compare ensuite (chapitre 2) différentes méthodes afin de prédire des variables
alimentaires obtenues par des digestions In vitro de différents fourrages qui utilisent la liqueur
ruminale de cerf. Les deux chapitres suivants sont intégrateurs et propose des implémentations
de nouvelles méthodes dans le cadre d’études écologiques à l’échelle du paysage. Je propose
dans le chapitre 3 un nouvel indice de qualité alimentaire qui utilise la spectroscopie dans le
proche infra-rouge, les géostatistiques ainsi que des données d’inventaires floristiques afin
d’estimer les patrons de qualité alimentaire à l’échelle du paysage. Finalement, dans le chapitre
4, je cherche à expliquer les patrons de régénération en sapin beaumier dans certains secteurs
de l’île d’Anticosti en dépit des densités élevées de cerf de Virginie. Pour ce faire, je teste une
hypothèse qui stipule que les conditions édaphiques des parcelles en régénération affectent
négativement la qualité alimentaire des tissus foliaires.
Mots-clés : nutrition des herbivores, gestion des populations animales, monitoring, suivi de
populations, qualité des fourrages, interactions plante-herbivore, spectroscopie dans le proche
infrarouge
ii
REMERCIEMENTS
Cette thèse fut possible grâce au soutien de très nombreuses personnes. Tout d’abord, merci
aux camarades des labos Bradley, Shipley et Vellend, qui ont été mes comparses des cinq
dernières années : Mathieu Dufresne, Cédric Frenette-Dussault, Mathieu Gauthier, Mélanie
Drouin, Sylvain Pelletier-Bergeron, Sébastien Auger, Pierre-Philippe Lachapelle, Mark Jewel,
Antoine Tardif, Maurice Aulen, Martine Fugère, Charline Binguier, Pierre-Luc Chagnon,
Michaël Belluau, Antoine Becker-Scarpitta. Les opinions constructives, les blagues salées et
les cafés partagés m’ont vraiment aidé à progresser à travers ce projet. Merci aux profs et au
personnel de soutien que j’ai côtoyé, et qui m’ont souvent aidé à y voir plus clair : Bill Parsons,
Bill Shipley, Marc Bélisle, Benoît Lapointe, Richard Fournier, Fanie Pelletier. Merci à la gang
de la Chaire, à mes assistants de terrain et aux gens d’Anticosti, avec qui j’ai eu de précieux
échanges, tant sur le plan scientifique qu’humain: Sonia de Bellefeuille, Yolande, Émilie
Champagne, Michaël Bonin, Édouard Bélanger, Cécile Meyer, Joëlle Castonguay, Marianne
Bachand, Bert Hidding, Gaëlle Darmon, Jean-François Adam, Gaétan Laprise et Danielle
Morin. Merci aux gens du RECSUS, et au Vert et Or Natation, qui avez mis de l’équilibre dans
ma vie d’étudiant gradué.
Merci à mes conseillers et précieux collaborateurs, qui ont beaucoup contribué à faire avancer
mes travaux : Steeve Côté, Mark Vellend, Robert Berthiaume, Sophie Calmé et Marie-Andrée
Giroux. Merci à mes directeurs de recherches, qui m’ont guidé tout le long de la gestation de
ce long projet. À Robert, qui m’a presque appris à écrire. À J-P, pour sa passion contagieuse
pour la recherche.
Merci à ma famille, qui a su me supporter à travers ce projet, mais aussi à travers les épreuves
que je traversais en parallèle. Merci à Kate, pour sa compréhension et sa patience.
Merci à mon Papa qui sait tout. Merci à Flavie, qui apprend si vite …
iii
TABLE DES MATIÈRES
SOMMAIRE ................................................................................................................................ i
REMERCIEMENTS ................................................................................................................. iii
LISTE DES ABRÉVIATIONS ............................................................................................... viii
LISTE DES TABLEAUX .......................................................................................................... x
LISTE DES FIGURES .............................................................................................................. xi
INTRODUCTION ...................................................................................................................... 1
Objectifs .............................................................................................................................. 4
Contexte de l’étude ............................................................................................................. 4
Utilisation de la spectroscopie dans le proche infrarouge comme outil prédictif de
variables écologiques .......................................................................................................... 5
Fonctionnement de la SPIR ......................................................................................................... 5
Applications de la SPIR ...................................................................................................... 7
CHAPITRE 1 - UTILISATION DE LA SPECTROSCOPIE DANS LE PROCHE
INFRAROUGE ET DE LA CHIMIE FÉCALE COMME PRÉDICTEURS DE LA
COMPOSITION DE LA DIÈTE DU CERF DE VIRGINIE ..................................................... 9
Avant-propos ..................................................................................................................... 10
Résumé .............................................................................................................................. 12
Abstract ............................................................................................................................. 13
Introduction ....................................................................................................................... 14
Study area .................................................................................................................................. 17
Fecal samples ............................................................................................................................ 18
iv
Fecal chemical properties .......................................................................................................... 19
Near infrared spectroscopy ........................................................................................................ 19
Fecal chemistry models ............................................................................................................. 20
Results ............................................................................................................................... 20
Discussion ......................................................................................................................... 24
Implications ....................................................................................................................... 26
Acknowledgements ........................................................................................................... 26
References ......................................................................................................................... 27
CHAPITRE 2 - MÉTHODES DE SUBSTITUTION AUX PROTOLES DE DIGESTION IN
VITRO POUR L’ÉTUDE DE L’ÉCOLOGIE NUTRITIONNELLE DES RUMINANTS
SAUVAGES ............................................................................................................................. 32
Avant-propos ..................................................................................................................... 33
Résumé .............................................................................................................................. 35
Abstract ............................................................................................................................. 36
Introduction ....................................................................................................................... 37
Material and Methods ....................................................................................................... 39
Study area .................................................................................................................................. 39
Sampling.................................................................................................................................... 40
In vitro digestion assays using white-tailed deer inoculum ....................................................... 40
Proxies to estimate white-tailed deer in vitro digestion ............................................................ 41
Statistical analysis ..................................................................................................................... 42
Results ............................................................................................................................... 43
Proxies for true digestibility, total gas and VFA production using deer inoculum ................... 43
Discussion ......................................................................................................................... 49
Proxies for IVTD, total gas and VFA production using deer inoculum .................................... 49
Acknowledgements ........................................................................................................... 54
References ......................................................................................................................... 55
v
CHAPITRE 3 - UTILISATION DE LA SPECTROSCOPIE DANS LE PROCHE
INFRAROUGE ET DES GÉOSTATISTIQUES POUR GÉNÉRER UNE CARTE DE LA
QUALITÉ ALIMENTAIRE À L’ÉCHELLE DU PAYSAGE ................................................ 60
Avant-propos ..................................................................................................................... 61
Résumé .............................................................................................................................. 63
Abstract ............................................................................................................................. 64
Introduction ....................................................................................................................... 65
Materials and methods ...................................................................................................... 67
Geography and history of the study area ................................................................................... 67
Collection and spectral analysis of fecal samples ..................................................................... 69
Sorting fecal samples by discriminant analysis ......................................................................... 71
Mapping habitat nutritional quality ........................................................................................... 72
Fecal chemical properties .......................................................................................................... 73
Inferring habitat attributes responsible for nutritional quality................................................... 73
Results ............................................................................................................................... 75
Discussion ......................................................................................................................... 78
Acknowledgements ........................................................................................................... 81
References ......................................................................................................................... 81
Supporting information ..................................................................................................... 88
CHAPITRE 4 - TEST DE L’HYPOTHÈSE DU CONTRÔLE DE L’HERBIVORIE PAR LES
RESSOURCES DANS UN ÉCOSYSTÈME À FORTE DENSITÉ D’HERBIVORES ......... 89
Avant-propos ..................................................................................................................... 90
Résumé .............................................................................................................................. 92
Abstract ............................................................................................................................. 93
Introduction ....................................................................................................................... 94
Material and methods ........................................................................................................ 96
Study area .................................................................................................................................. 96
Sampling.................................................................................................................................... 97
vi
Forage and soil analyses ............................................................................................................ 98
Statistical analyses..................................................................................................................... 99
Results ............................................................................................................................. 100
Discussion ....................................................................................................................... 103
Acknowledgments ........................................................................................................... 104
References ....................................................................................................................... 105
Supporting information ................................................................................................... 108
CONCLUSION ...................................................................................................................... 113
Valeur scientifique .......................................................................................................... 113
Futures recherches .......................................................................................................... 114
1-Implémentation de la SPIR dans les programmes de monitoring écologique ...................... 115
2- Les sous-produits de digestion ruminale pour étudier l’écophysiologie des ruminants
sauvages .................................................................................................................................. 115
3- Étude des interactions entre le microbiome ruminal et l’environnement ............................ 116
BIBLIOGRAPHIE ................................................................................................................. 118
vii
LISTE DES ABRÉVIATIONS
ADF
Acid detergent fibers
ADL
Acid detergent lignin
AICc
Akaike Information criterion for small sample size
DISTEX
Mahalanobis distance of a spectrum to the centroid of exclosure
spectra
Gmax
Maximal gas production
IVTD
In vitro true digestibility
KM
Time to reach ½ of the maximal gas production
MLR
Mixed-linear regression
N
Nitrogen
NDF
Neutral detergent fibers
NIR
Near infrared
NIRS
Near infrared spectroscopy
P
Phosphorus
viii
PC
Principal component
PCA
Principal component analysis
PIR
Proche infra-rouge
PLSR
Partial-least squares regression
RMSEC
Root mean squared error of calibration
RMSEP
Root mean squared error of prediction
RPD
Ratio of prediction to deviation
SPIR
Spectroscopie dans le proche infra-rouge
VFA
Volatile fatty acids
ix
LISTE DES TABLEAUX
Chapitre 1
1. Best-fitting mixed-linear regression models predicting the fecal content of various
forages based on fecal chemical properties (N = 89); ADL = Acid detergent lignin.
23
Chapitre 2
1. Results of Student’s two sided t-tests comparing the effect of deer vs. cow inocula
on Km and Gmax values.
49
Chapitre 4
AP. B. Summary of mineral soil properties for 21 sampling plots located on three
geological deposits on Anticosti Island, QC, Canada.
108
AP. C. Results of multivariate analyses testing the relationship between plant nutritional
quality of 4 forage species and soil properties.
109
AP. D. Results of univariate tests testing the effect of the geological deposit on the
nutritional quality of 4 forage species. The analyses are achieved by using the
Tukey's method in conjunction with one-way analysis of variance (ANOVA).
110
x
LISTE DES FIGURES
Chapitre 1
1. Results of PC-based discriminant analysis showing the Mahalanobis distances
between the NIR spectra of fecal samples collected from captive deer that had
been fed 2 different diets (Control diet = 70% balsam fir + 20% white spruce +
10% lichens; Poor diet = 50% balsam fir + 40% white spruce + 10% lichens). A
total of 102 samples were used for model calibration and 30 samples for
validation.
2.
21
Correlation between coniferous content of fecal samples as determined by
microhistology and that predicted by a partial least-squares regression model of
fecal NIR spectra. A total of 75 samples were used for model calibration and 15
samples for validation. RMSEC = root mean squared error of calibration;
RMSEP = root mean squared error of prediction.
22
Chapitre 2
1. Relationship between in vitro true digestibility (IVTD) of 4 forage species
determined using a deer inoculum and (a) IVTD values using a cow inoculum, (b)
forage acid detergent fibers (ADF), (c) 48 h CO2 production from soil-forage
mixtures, and (d) prediction equations based on near infrared spectra (NIRS) of
forage. Cross bars in frame (c) represent standard errors on replicated measures.
xi
NIRS predictive model was calibrated using 56 samples and validated by
calculating root-mean square error (RMSEVAL) of remaining samples.
45
2. Relationship between in vitro gas production of 4 forage species determined using
a deer inoculum and (a) in vitro gas production using a cow inoculum, (b) forage
acid detergent fibers (ADF), (c) 48 h CO2 production from soil-forage mixtures,
and (d) prediction equations based on near infrared spectra (NIRS) of forage.
Cross bars represent standard errors on replicated measures. NIRS predictive
model was validated using a leave-one-out cross-validation procedure in order to
calculate the root-mean square error value (RMSEVAL) of the 12 composite forage
samples.
3.
46
Relationship between in vitro volatile fatty acids (VFA) production of 4 forage
species determined using a deer inoculum and (a) in vitro VFA production using
a cow inoculum, (b) forage acid detergent fibers (ADF), (c) 48 h CO2 production
from soil-forage mixtures, and (d) prediction equations based on near infrared
spectra (NIRS) of forage. Cross bars represent standard errors on replicated
measures. NIRS predictive model was validated using a leave-one-out crossvalidation procedure in order to calculate the root-mean square error value
(RMSEVAL) of the 12 composite forage samples.
47
4. Cumulative gas production profiles from (a) deer and (b) cow inocula during a 48
h in vitro digestion assay.
5.
48
Results of Student’s two sided t-tests comparing the effect of deer vs. cow inocula
on the relative production of acetate, propionate and butyrate (first 3 bar clusters
from the left), as well as on the total amount of volatile fatty acids (right bar
cluster) produced during a 48 h in vitro digestion assay of 4 forage species.
50
xii
Chapitre 3
1. (a) Map of the study area showing the various land cover classes obtained from
the mapping geodatabase of the Quebec Ministry of Natural Resources and
Wildlife (MRNFQ 2010) as well as each sampling location. (b) Map of our study
area showing the occurrence of cold-spots (in blue) and hot-spots (in red) of forage
nutritional quality, as determined by Gi* analysis.
67
2. Photo showing differences in vegetation within (right) and outside (left) an
exclosure. The photo was taken 10 years after erecting the 12 foot fence. Photo
credit: S. de Bellefeuille.
69
3. Results of a regression tree analysis using the Gi* statistic as the response variable,
and the percentage of cover of different habitat types in a 300 ha surrounding each
collected fecal sample as the explanatory variables.
74
4. Boxplots showing the distribution of Gi* (Fig. 4a, b) or fecal acid detergent lignin
(ADL) (Fig.4c, d) when Canada bunchberry or tamarack were available within a
300 ha neighborhood surrounding each fecal sample. Gi* values deviating from 0
by  1.65 respectively correspond to cold-spots and hot-spots.
75
5. Relationship between fecal acid detergent lignin (ADL) and the mean DISTEX
values of 300 ha buffers surrounding each collected fecal sample.
76
AP. A. Map of the study area showing cold-spots (in blue) and hot-spots (in red) as
determined by a Gi* analysis, using DISTEX values computed solely from (a)
the first or (b) the second of the 2 exclosures.
87
xiii
Chapitre 4
1. Ordination biplots generated from principal component analyses discriminating
sampling plots according to (a) soil chemical properties, and (b) white spruce
foliar chemistry. Polygons circumscribe plots found within each of the 3
geological deposits.
98
2. Results of one-way analyses and Tukey’s tests describing the effects of geological
deposit on (a) in vitro true digestibility of white spruce foliage, (b) neutral
detergent fiber concentration of white spruce foliage, and (c) in vitro true
digestibility of Canada bunchberry foliage. Statistically significant differences in
forage quality are denoted by different lower-case letters.
101
AP. A. Location of 21 sampling plots distributed over 3 geological deposits on Anticosti
Island, QC, Canada. The island covers a total area of 7,9463 km2.
107
xiv
INTRODUCTION
La qualité et l’abondance des ressources alimentaires ont été identifiées comme des variables
écologiques clés pour la qualité des habitats des herbivores. Mysterud (1998) a estimé qu’un
ruminant tempéré passait en moyenne plus de 50 % de son budget total d’activité à se nourrir,
ce qui illustre à quel point son alimentation affecte toutes les facettes de ses traits d’histoire de
vie, incluant sa croissance (White, 1983; Crête et Huot, 1993), sa reproduction (Cook et al.,
2004) et sa survie (Bender et al., 2008; Cook et al., 2004). Dans un premier temps, la qualité
alimentaire des fourrages ingérés par un ruminant détermine à quel degré ces derniers seront
digérés et intégrés aux différentes voies métaboliques nécessaires au maintien de l’homéostasie
de l’animal (Chilliard et al., 2008). Dans un deuxième temps, la disponibilité de ces mêmes
ressources dans leur domaine vital détermine la quantité d’énergie qu’il doit consacrer à sa quête
alimentaire. C’est d’ailleurs la balance entre la qualité et la disponibilité relative des ressources
alimentaires qui constitue le pilier de la théorie de l’approvisionnement optimal décrite par
MacArthur et Pianka (1966), qui stipule qu’un animal cherche à maximiser ses intrants en
nutriments en trouvant le meilleur compromis entre la qualité des ressources et le coût nécessaire
pour y avoir accès.
S’il est possible de calculer objectivement l’abondance des ressources alimentaires disponibles
au moyen de différentes techniques d’inventaires dans le cadre d’études écologiques (e.g.
Catchpole et Wheeler, 1992), il est plus complexe de statuer objectivement sur la « qualité » de
ces mêmes ressources. Les besoins en nutriments d’un herbivore sont complexes et varient
énormément au fil des saisons (e.g. Berteaux et al., 1998), d’une espèce à l’autre (e.g. Codron
et al., 2007), entre des individus d’âge (Provenza et al., 1992) et de sexes différents (Mysterud,
2000). La plupart des études en écologie nutritionnelle cherchant à généraliser les patrons de
sélection de diète des herbivores sauvages assument généralement que ces derniers cherchent à
1
optimiser l’acquisition de devises limitantes telles que des substrats énergétiques (e.g. Courant
et Fortin, 2012) ou des nutriments tels que l’azote (e.g. Miyashita et al., 2007). Bien que
considérer la nutrition d’une perspective unidimensionnelle soit souhaitable afin de simplifier
la modélisation des patrons de sélection des diètes, cette approche ne considère pas le caractère
multidimensionnel inhérent à la chimie des plantes et aux besoins nutritionnels complexes d’un
herbivore. Les métabolites secondaires contenus en différentes concentrations dans les plantes
peuvent notamment influencer le choix alimentaire des herbivores, et ce indépendamment de la
teneur en énergie et ou en azote de celles-ci. À titre d’exemples, les tanins sont reconnus pour
diminuer la disponibilité de l’azote ingéré (i.e. Robbins et al., 1987a) et la digestibilité du bolus
alimentaire (i.e. Robbins et al., 1987b) chez les ruminants, alors que la consommation de
certains alcaloïdes est reconnue pour exiger un coût métabolique de détoxification (McLean et
Duncan, 2006).
Malgré d’importantes avancées dans l’identification de composés
antinutritionnels et des processus évolutif ayant contribué à leur apparition (e.g. Pichersky et
Gang 2000), nous avons encore une compréhension incomplète des processus qui relient la
composition chimique des plantes ingérées à la performance des herbivores (De Gabriel et al.,
2014).
Cette complexité inhérente à l’analyse des besoins nutritionnelle des herbivores sauvages a
orienté la recherche vers de nouvelles approches théoriques. Raubenheimer et Simpson (1993)
ont notamment introduit le Cadre Conceptuel Géométrique pour l’étude de la nutrition des
herbivores, qui intègre de multiples devises nutritionnelles susceptibles de limiter leur
performance. Cette approche consiste à observer le choix de différents individus confrontés à
des options nutritionnelles différant de par leurs contenus en éléments nutritifs. Cette approche
permet de définir une cible nutritionnelle pour l’herbivore dans un espace multivarié contraint
par la physiologie propre à chaque espèce d’herbivore. Bien que cette approche permette de
prédire avec précision le patron de sélection de diète de différents herbivores en captivité (e.g.
Kohler et al., 2012), son application nécessite de pouvoir échantillonner les plantes disponibles
et observer de près les comportements d’alimentation (e.g. Raubenheimer et al., 2014), ce qui
pose d’évidentes contraintes logistiques.
2
Dans un contexte d’étude écologique, la recherche est évidemment limitée par les aléas des
contraintes du travail terrain. Dans un premier temps, le comportement cryptique de certaines
espèces rend difficile la quantification du régime alimentaire par données observationnelles (e.g.
Ruckstuhl et al., 2003). On doit alors estimer la diète plus ou moins précisément, soit en se
basant sur la disponibilité des ressources alimentaires de l’habitat (e.g. Hanley, 2012) ou en
analysant les contenus ruminaux (McInnis et al., 1983) ou fécaux (Dearden et al., 1975). De
plus, l’accès même au territoire d’étude des populations naturelles est parfois problématique, ce
qui limite grandement les tailles d’échantillonnage ainsi que le type d’analyse qu’il est possible
d’effectuer. À titre d’exemple, il est bien documenté que plusieurs composés volatils tels que
les terpénoïdes (Vourc’h et al., 2002) et les flavonoïdes (Levin, 1976) peuvent moduler
l’appétence des fourrages offerts au bétail. Or, dans le cadre d’études écologiques, il peut être
difficile de doser ces composés volatils puisqu’ils sont instables et s’évaporent lors d’un
entreposage prolongé.
3
Objectifs
Puisque d’une part, le concept même de qualité nutritionnelle est difficile à définir pour un
herbivore sauvage, et que d’autre part, les méthodes existantes pour évaluer la nutrition des
herbivores dans le cadre d’études écologiques comportent de nombreuses contraintes
logistiques, cette thèse de doctorat comporte deux principaux objectifs :
1. Développer des méthodes afin d’évaluer le contenu de la diète d’un herbivore et sa
qualité.
2. Implémenter la méthodologie développée dans le cadre d’études nutritionnelles à
l’échelle du paysage.
Contexte de l’étude
Mon aire d’étude, l’île d’Anticosti, est un territoire propice pour l’étude des interactions planteherbivore, mais son éloignement complexifie la mise en place d’études écologiques. Le cerf de
Virginie y a été introduit à la fin du 19ième siècle en l’absence de prédateurs naturels, et la
population s’y maintient à de très hautes densités depuis près de 80 ans (Marie-Victorin et
Rolland-Germain, 1969). La pression d’herbivorie constante a amené l’extirpation de
nombreuses espèces végétales (e.g. Acer spicatum, Corylus cornuta, Taxus canadensis) (Potvin
et al., 2003). La préférence du cerf pour les espèces décidues et le sapin beaumier par rapport
aux épinettes blanche et noire amène un changement des communautés végétales à l'échelle du
paysage. Les sapinières à bouleau blanc endémiques sont graduellement remplacées par des
4
pessières blanches et noires. Cette situation amène des nouvelles dynamiques de succession dont
les effets sont complexes et imprévisibles. Vu l’importance des ressources tant forestières que
fauniques à l’île, un suivi de la qualité de l’habitat du cerf est fort souhaitable afin d’élaborer
des mesures de gestion adaptatives efficaces.
Utilisation de la spectroscopie dans le proche infrarouge comme outil
prédictif de variables écologiques
Durant mes études, j’ai attribué une grande place à l’utilisation de la spectroscopie dans le
proche infrarouge (SPIR) afin de réaliser mes objectifs. Afin de faciliter la compréhension des
articles qui suivent cette introduction, je présente ici une brève explication de cette technique
Fonctionnement de la SPIR
Dans tout matériel, les liens chimiques entre les atomes vibrent à la façon d’un mouvement
harmonique simple et la fréquence des oscillations dépendent de la masse des atomes liés ainsi
que de la force des liens qui les unit (Osborne, 1986; Bertrand, 2002). Un spectroscope dans
l’infrarouge proche balaie un échantillon avec différentes longueurs d’ondes comprises entre
780 et 2500 nm. Lorsque la fréquence de la vibration émise par l’appareil correspond à celle de
la liaison chimique ou à un multiple de celle-ci, les molécules liées passent à un niveau excité.
Ce transfert d’énergie peut-être mesuré et analysé sous la forme d’un spectre de réflectance. La
spectroscopie en proche infrarouge est particulièrement sensible aux vibrations des liens C-H,
N-H et O-H. La forme du spectre PIR provenant d’un échantillon constitué de molécules
complexes est donc fonction des différents groupements fonctionnels chimiques (e.g. –OH, 5
NH2) qui le composent. Il est toutefois difficile de relier un pic de réflectance d'un spectre à un
groupement fonctionnel précis puisque les valeurs mesurées correspondent aux différents
niveaux d'énergie de plusieurs combinaisons de molécules excitées. Un spectre PIR correspond
ainsi à une empreinte chimique complexe donnée par l’oscillation des différents liens chimiques
soumis au balayage de différentes longueurs d’ondes.
À l’instar de la spectrophotométrie classique, la SPIR permet d’associer l’information spectrale
à une variable réponse. On obtient une calibration en utilisant une méthode de référence (e.g.
dosage de l’azote par la méthode de Kjeldahl). On bâtit un modèle prédictif en associant la
variable réponse d’intérêt (e.g. l’azote total) à l’information spectrale de l’échantillon dont la
composition est connue. Pour développer une méthode prédictive, on utilise généralement un
certain nombre d’échantillons pour bâtir le modèle de calibration et un plus petit nombre
d’échantillons pour valider le modèle (Mayer et Butler, 1993). On utilise finalement le jeu de
validation externe au modèle afin d’en évaluer le pouvoir prédictif.
La prédiction de la variable d’intérêt se fait par modèle de régression multivarié sur les valeurs
des spectres PIR. Dans le traitement statistique d’un spectre PIR, chaque valeur de réflectance
à une longueur d’onde précise correspond à une variable explicative. Qui plus est, ces variables
sont fortement corrélées entre elles (i.e. réflectance à 1300 nm ≈ réflectance à 1301 nm). Les
méthodes de régression en composantes principales (Massy, 1965) et par moindres carrés
partiels (Wold, 1966) permettent de contourner les problèmes de saturation (i.e. « over-fitting »)
du modèle qui sont engendrés par la colinéarité des nombreuses variables. Les zones du spectre
expliquant le plus de variabilité dans la variable réponse sont condensées en variables latentes
afin de réduire la dimensionnalité du jeu de données.
6
Applications de la SPIR
Utiliser la SPIR pour bâtir des modèles prédictifs présente plusieurs avantages. L’analyse est
rapide, non-destructive, ne nécessite que très peu de matériel et ne produit aucun déchet toxique.
D’abord utilisée dans l’industrie agroalimentaire en contrôle qualité, cette efficacité analytique
finit par intéresser les écologistes, pour qui le traitement d'un grand nombre d'échantillon peut
être long et coûteux. Parmi les premiers écologistes à utiliser la SPIR dans le cadre de leurs
recherches, Brooks et Anderson (1984) ont mesuré les différentes fractions de fibres de certaines
plantes sauvages consommées par le wapiti (Cervus elaphus) alors que Joffre et al. (1992) font
de même avec de la litière en décomposition.
L’utilité de la SPIR pour des analyses chimiques de routine n’est plus à prouver, mais les
écologistes s’intéressent souvent davantage aux attributs d’un échantillon qu’à sa chimie. Ce
besoin a contribué à orienter la SPIR vers une nouvelle approche dite holistique. Cette approche
vise à prédire une propriété d’un échantillon (e.g. la résistance à l’herbivorie d’une plante)
plutôt que la proportion d’un de ses constituants (e.g. phénols totaux). Un lien direct entre un
métabolite de la plante et la résistance à l’herbivorie est rarement direct puisqu’un grand nombre
de composés chimiques sont impliqués (Foley et al., 1998). L’avantage de la SPIR est qu’elle
considère l'ensemble de l'empreinte chimique de l'échantillon, ce qui est impossible de faire
dans le cadre d’analyses chimiques traditionnelles.
L’approche holistique est prometteuse pour l’écologie, parce que n’importe quel attribut
fonctionnel d’un échantillon relié à sa composition chimique est susceptible d’être modélisé par
la SPIR (Foley et al., 1998). Rutherford et al. (1996) furent parmi les premiers à utiliser cette
approche en expliquant 54 % de la variation de la résistance à un ravageur par les spectres PIR
de la cire des tiges de canne à sucre. McIlwee et al. (2001), quant à eux, ont correctement prédit
7
le taux d’ingestion de différentes espèces d’eucalyptus par des marsupiaux en utilisant des
spectres NIR des feuilles. On note un intérêt croissant à démontrer de nouvelles applications de
la SPIR dans plusieurs domaines de l’écologie, notamment en écologie des sols (Cécillon et al.,
2009), en écologie végétale (Stolter et al., 2006) ou animale (Tolleson et al., 2005).
8
CHAPITRE 1
UTILISATION DE LA SPECTROSCOPIE
DANS LE PROCHE INFRAROUGE ET
DE LA CHIMIE FÉCALE COMME
PRÉDICTEURS DE LA COMPOSITION
DE LA DIÈTE DU CERF DE VIRGINIE
9
Avant-propos
L’objectif de ce premier chapitre était de développer un outil afin d’effectuer le monitoring de
la diète hivernale du cerf de Virginie. La méthode de prédilection généralement utilisée pour
quantitifer le régime alimentaire de cet herbivore est la micro-histologie fécale (e.g. Lefort et
al. 2007), une technique qui demande de nombreuses heures en laboratoire et une excellente
connaissance de la micromorphologie des plantes consommées. La méthode proposée dans ce
chapitre permet de réduire considérablement le temps requis pour avoir des estimés de contenu
de la diète. On y parvient en calibrant des modèles prédictifs qui utilisent propriétés chimique
fécales comme variable explicative et des données micro-histologiques acquises de façon
usuelle comme variable réponse.
Pour développer mon approche, j’ai utilisé des jeux de données et des échantillons provenant de
deux études réalisées antérieusement dans le cadre de travaux de recherches menés à Anticosti
: celles de Joëlle Taillon (Taillon et Côté 2006) et de Marie-Andrée Giroux (Giroux et al. 2012).
La rédaction de ce chapitre sous forme d’article s’est fait en collaboration avec Robert Bradley
(U. Sherbrooke), Marie-Andrée Giroux, Jean-Pierre Tremblay et Steeve D. Côté (U. Laval), et
a été publié dans le numéro de mars 2014 du journal Rangeland Ecology and Management :
Jean, P.-O., Bradley, R.L., Giroux, M.-A., Tremblay, J.-P. et Côté, S.D. (2014). Near Infrared
Spectroscopy and Fecal Chemistry as Predictors of the Diet Composition of White-Tailed
Deer. Rangeland Ecology & Management 67: 154–159.
10
Near Infrared Spectroscopy and Fecal Chemistry as Predictors of
the Diet Composition of White-Tailed Deer
Pierre-Olivier JEAN*1, Robert L. BRADLEY1, Marie-Andrée GIROUX2, Jean-Pierre
TREMBLAY2 & Steeve D. CÔTÉ2
1
Département de biologie, Université de Sherbrooke, 2500 boulevard de l'Université,
Sherbrooke, Québec, Canada J1K 2R1.
2
Département de biologie, Université Laval, Québec, QC, G1V0A6, Canada.
Keywords: fecal analysis; diet quality; over-browsing; white-tailed deer; fecal
microhistology; near infrared spectroscopy
Abbreviations: NIR, near infrared; PLSR, partial-least squares regression; MLR, mixed-linear
regression; ADL, acid detergent lignin, NDF, neutral detergent fibers; ADF, acid detergent
fibers; N, nitrogen; RPD, ratio of prediction to deviation; AICc, Akaike Information criterion
for small sample size, PC, principal component, RMSEC, root mean squared error of calibration;
RMSEP, root mean squared error of prediction.
Reference: Jean, P.-O., Bradley, R.L., Giroux, M.-A., Tremblay, J.-P. and Côté, S.D. (2014).
Near Infrared Spectroscopy and Fecal Chemistry as Predictors of the Diet Composition of
White-Tailed Deer. Rangeland Ecology & Management 67 : 154–159.
11
Résumé
La pression élevée d’herbivorie par le cerf de Virginie (Odocoileus virginianus Zimmermann)
sur l’île d’Anticosti (Canada) nous amène à développer des méthodes efficaces afin d’estimer
leur patrons de sélection alimentaire. Nous avons testé le potentiel de la spectroscopie dans le
proche infra-rouge (NIR) et de propriétés chimiques de fèces pour prédire la composition de la
diète de différents individus. Nous avons d’abord utilisé une analyse discriminante en
composantes principales pour séparer les spectres NIR de fèces (n=102) obtenus de deux
groupes d’individus nourris de deux diètes différentes ne différant que de par leur abondance
relative en sapin beaumier (Abies balsamea (L.) P.Mill.) et en épinette blanche (Picea glauca
(Moench) Voss.). Le modèle calibré a permis de classer 28 échantillons de validation sur 30
(93,3 %) dans la bonne catégorie. Dans une deuxième étude, nous avons tenté de prédire les
proportions de conifères, d’essences décidues, d’herbacées et de lichens dans la diète du cerf
telles que déterminées par micro-histologie fécale.
Pour ce faire, nous avons comparé
l’utilisation des spectres PIR et différentes propriétés chimiques de fèces comme variables
explicatives dans des modèles prédictifs de la composition de la diète. Les spectres NIR ont été
analysés en utilisant une méthode de régression en moindres carrés partiels (PLSR), alors que
les propriésés chimiques de fèces ont été ont étés analysés en utilisant des modèles linéaires
mixte (MLR). Les modèles PLSR ont été robustes (R2=0,89, Ratio des erreurs de prédiction sur
la variation = 3,2) pour prédire la proportion des fragments de conifères, mais pas pour prédire
les proportions de sapin beaumier, d’épinette blanche, des essences décidues et du lichen dans
les fèces. Les modèles MLR ont révélés une relation positive (47 % de variance expliquée)
entre la concentration en lignine (ADL) fécale et la proportion de fragments de conifères dans
les fèces. La concentration en ADL et en cellulose fécales ont expliqué 24 % de la variation
dans la proportion de fragments d’essences décidues dans les fèces, alors que la concentration
en ADL a permis d’expliquer 22 % de la variance dans le contenu en fragments de sapin
beaumier dans les fèces. Ces résultats suggèrent l’utilité de ces deux approches dans le contexte
de l’île d’Anticosti, notamment afin de mesurer le degré de variation dans les diètes à l’intérieur
12
d’un même domaine vital, de même que pour mesurer l’importance des conifères dans la diète
hivernale du cerf de Virginie.
Abstract
Over-browsing by white-tailed deer (Odocoileus virginianus Zimmermann) on Anticosti Island
(Canada) created a need to develop efficient methods for estimating their foraging patterns. We
tested the ability of near infrared (NIR) spectra of feces and of fecal chemical properties to
predict diet composition of different individuals. We first used a principal component-based
discriminant analysis to sort the NIR spectra of fecal samples (n=102) obtained from 2 groups
of captive deer that had been fed 2 different diets. The diets differed only in their relative
abundance of balsam fir (Abies balsamea (L.) P.Mill.) and white spruce (Picea glauca (Moench)
Voss.) foliage. The calibrated model allowed us to assign 28 of 30 validation fecal samples (93.3
%) to the correct diet. In a second study, we attempted to estimate the proportion of coniferous,
deciduous, herbaceous and lichenous forages in diets of free-ranging white-tailed deer, as
determined by fecal microhistology. Both NIR spectra and chemical properties of feces were
used as predictors of diet composition. NIR spectra were analyzed using partial least-squares
regression (PLSR), whereas fecal chemical properties were analyzed using mixed-linear
regressions (MLR). The PLSR models were robust (R2 = 0.89, Ratio of prediction to deviation
= 3.2) for predicting the amount of coniferous fragments, but not for predicting the relative
amounts of balsam fir, white spruce, deciduous and lichenous fragments within feces. MLR
models revealed a positive relationship (47% variance explained) between acid detergent lignin
(ADL) and coniferous fragments within feces. ADL and cellulose explained 24% of variance in
deciduous fecal fragments, whereas ADL alone explained 22% of variance in balsam fir fecal
fragments. These results suggest that NIRS and fecal chemical properties have several
applications on Anticosti Island, such as measuring the degree of variation in diets within a
given home range or determining dietary conifer intake during winter.
13
Introduction
Studying the foraging behaviour of large wild herbivores in heterogeneous landscapes informs
us on habitat selection, seasonal movements, or the impacts of herbivory on plant communities.
To this end, a variety of methods have been developed over past decades to describe the diet of
wild herbivores. These include fecal analyses (Stewart 1967), the probing of stomach contents
(McInnis et al. 1983), browse surveys (Shipley et al. 1998), behavioral observations in the field
(Parker et al. 1999), as well as isotope fractionation of animal tissues (Dalerum and Angerbjörn
2005). Among these techniques, fecal analyses are an interesting option because fecal samples
are easy to obtain in high density populations and do not require the handling or killing of
animals. The main disadvantage of this approach is that highly digestible plant species could be
under-represented in the estimated diets, whereas less digestible plants could be overrepresented. Nevertheless, Hanley and McKendrick (1985) showed 65% to 77% similarity
between the estimated plant species composition of rumen contents and fecal samples of Sitka
black-tailed deer (Odocoileus hemionus sitkensis Rafinesque).
Estimating the diets of large herbivores through fecal analyses usually entails the
microhistological analysis of plant fragments (Dearden et al. 1975; Lefort 2002). This method
is, however, expensive and time consuming and it requires a good knowledge of the
micromorphology of different plant tissues. More rapid proxy methods have been developed,
such as determining fecal n-alkanes (Dove and Mayes 2005; Bugalho et al. 2004) or DNA
barcoding of plant fragments within feces (Valentini et al. 2009a), but these remain relatively
expensive and their ability to accurately quantify the proportion of different plant species in
complex diets is uncertain (Bugalho 2002; Valentini 2009b). There is, therefore, a need to
develop rapid and cost-efficient ways for estimating the diets of large herbivores through fecal
analyses.
14
Recent research has shown that near infrared spectroscopy (NIRS) of fecal samples could be
used to assess the dietary composition of free-ranging domestic livestock (Dixon and Coates
2009; Walker et al. 2010). NIRS is a non-destructive procedure requiring minimal sample
preparation, and it produces a spectrum expressing the entire chemical makeup of the sample.
NIRS may thus be used to predict both the chemical (Malley et al. 2002; Stuth et al. 2003) and
functional (McIlwee et al. 2001; Stolter et al. 2006) properties of materials. There is an
opportunity, therefore, to explore the potential and limitations of NIRS to discriminate different
diets of wild herbivores through the spectral analysis of their feces.
A second potentially rapid and cost-effective approach to assess the diets of wild herbivores
would be to use fecal chemical properties such as fecal nitrogen (N) or proximate carbon (C)
fractions. These have traditionally been used to assess the nutritional and energetic quality of
different diets or habitats, but not to estimate the actual botanical composition of the diet. For
example, fecal N may be a good index of seasonal variations in diet quality (Verheyden et al.
2011), provided that the amount of tannins in the diet is low (Leslie et al. 2008). Likewise,
Hodgman and Davitt (1996) proposed that neutral detergent fiber (NDF) content of feces be
used as a predictor for digestible energy of the diet. Given that different forage types vary in
their nutritional and energetic qualities, there is an opportunity to explore the ability of fecal
chemical properties to estimate the actual botanical composition of the diets of wild herbivores.
For example, fecal N might be used to detect the intake of nitrophilous plant species, whereas
NDF might be used to identify the most digestible plant species. We posit that this potential to
detect diet composition via fecal chemical properties increases as the number of available forage
types decreases.
Using NIRS or fecal chemical properties to estimate the diets of wild herbivores first requires a
calibration dataset of fecal samples from animals with known diets. Knowledge of an animal’s
recent food intake may be obtained directly through feeding trials on captive individuals, or by
reliable microhistological data. Feeding trials may yield more accurate data than microhistogical
15
analyses, but are relatively expensive and logistically complicated to perform on wild animals
in remote regions. Furthermore, feeding trials usually comprise a limited number of distinct
diets, which under-represents the wide range of diets that are possible in the wild. On the other
hand, microhistological data have been criticized as being imprecise (e.g. Wam et al. 2010),
leading Walker et al. (2010) to speculate that these cannot be used as calibration data sets for
NIRS. This remains, however, a contentious matter given that Taillon et al. (2006) found an
almost perfect concordance between the amounts of fir and spruce needles fed to captive whitetailed deer (Odocoileus virginianus Zimmermann) fawns and the percentages of these 2 forage
types within feces. There is, therefore, a need to test the resolution at which NIRS or fecal
chemical properties can predict simple diets of wild herbivore based on both types of calibration
data sets (i.e., feeding trials and microhistological data).
We report on a study where we evaluated both the potential and limitations of NIRS and fecal
chemical properties to predict the proportion of different plant groups composing the winter
diets of white-tailed deer on Anticosti Island, Canada. The island boasts a high population of
this herbivore, which makes it a prized area for game hunting. There are concerns, however,
that over-browsing of certain winter forage types could result in an important crash of the
population before 2100 (Potvin et al. 2004). For this reason, it is important to propose rapid and
cost-effective tools that could help rangeland managers on Anticosti Island to monitor the
depletion of resources during winter. Fecal samples were obtained, therefore, from 2 prior
studies; one in which distinct diets had been administered to captive deer, the other in which
diets of free-ranging deer had been estimated from microhistological analyses. Materials and
methods
16
Study area
Anticosti Island (49°28'N 63°00'W) is located in the Gulf of St-Lawrence, Canada. The entire
island (7,943 km2) is located in the Eastern balsam fir-White birch bioclimatic zone (Grondin
et al. 1996). The sub-boreal maritime climate provides cool summers (mean of 16° in July) and
cold winters (mean of -11° in January), with an average annual precipitation of 630 mm of rain
and 406 cm of snow (Environment Canada 2006).
About 200 white-tailed deer were introduced onto the island in 1896 and the population
increased rapidly in the absence of natural predators. Accordingly, the average white-tailed deer
density over different sectors of the island has been 20 to 60 individuals km-2 for the past 80
years (Potvin et al. 2004). Heavy browsing has resulted in a loss of palatable species in the shrub
layer such as Mountain maple (Acer spicatum Lam.), Beaked hazel (Corylus cornuta Marsh.)
and Canadian yew (Taxus canadensis Marsh) (Potvin et al. 2003), and in the spread of browse
tolerant graminoids such as Canada bluejoint (Calamagrostis canadensis (Michx.); Dufresne et
al. 2009). Balsam fir (Abies balsamea (L.) P.Mill.) is the preferred winter forage (Lefort et al.
2007), such that the recruitment of balsam fir seedlings has substantially diminished in the
understory giving way to a higher recruitment of less palatable white spruce seedlings (Picea
glauca (Moench) Voss.; Hidding et al. 2013). In accordance with this shift in the understory
vegetation, it is expected that white-tailed deer will increase their winter consumption of white
spruce seedlings in coming years (Sauvé 2005; Taillon et al. 2006).
17
Fecal samples
In study 1, Taillon et al. (2006) tested the effects of two winter diets on the behaviour and body
condition of captive white-tailed deer fawns. In 2004, 7 fawns were fed a control winter diet,
approximating the proportional intake of balsam fir (70%), white spruce (20%) and arboreal
lichens (10%) for white-tailed deer on Anticosti Island (Lefort et al. 2007). Six other fawns were
fed a “poor quality” diet consisting of 50% balsam fir, 40% white spruce and 10% arboreal
lichens. Feces were collected on a daily basis (control diet: n = 69, poor diet: n = 63) and kept
frozen at -20 C until further analyses.
In study 2, performed in 2009-2010, Giroux et al. (2012) conducted a field survey on Anticosti
Island to determine the influence of winter diet on body condition of free-ranging white-tailed
deer. Using a combination of telemetry and the bio-marking of feces with food dyes, they
followed 27 individuals and recovered fresh feces (N = 90). Within 6 h after collection, fecal
samples were dried at 20°C and subsequently stored at -4°C. A subsample of each fecal sample
was kept for NIRS scanning and chemical analyses (explained below), whereas the remaining
fecal material was analyzed for its plant content by microhistological analysis (Lefort et al.
2007). Coniferous, deciduous, as well as herbaceous and lichenous fragments were identified
by comparing with reference samples under a microscope. Among the coniferous fragments,
balsam fir and white spruce needles were distinguished based on their micromorphology. The
proportion of lichens was estimated at 16x magnification, whereas the proportions of the other
forage types were estimated at 100x magnification, according to the protocol developed by
Lefort (2002).
18
Fecal chemical properties
The stored fecal subsamples from study 2 ( Giroux et al. 2012) were thawed, dried again in an
air-draft oven (50 °C), and ground in a mortar to pass a 2 mm sieve. Fecal N was measured by
high temperature combustion followed by gas analysis, using a Vario Macro C&N Analyzer
(Elementar Analysensysteme Corp., Hanau, Germany). Neutral detergent fibers, acid detergent
fibers (ADF) and acid detergent lignin (ADL) were measured following the protocol of Goering
and Van Soest (1970), using an Ankom Fiber Analyzer 200 (Ankom Technology, Macedon,
USA). Cellulose was estimated as the difference between ADL and ADF (Van Soest 1994).
Near infrared spectroscopy
Air-dried and sieved fecal subsamples of both study 1 and study 2 were scanned for the
reflectance of wavelengths ranging from 1,000 to 2,500 nm, using a Nicolet Antaris FT-NIR
spectroscope supported with Omnic software (Thermofisher Scientific, Waltham, USA). In
study 1, a principal component (PC) based discriminant analysis using Mahalanobis distances
was used to discriminate the NIR spectra of feces from white-tailed deer fed with control versus
poor quality winter diets. This predictive model was calibrated using 102 fecal subsamples and
the remaining 30 subsamples were used for its validation. The non-randomness in the
classification of fecal samples among the 2 diets was evaluated using a Chi-square test. For
study 2, we used partial least-squares regressions (PLSR; Wold 1966) to build predictive models
relating microhistological data to the NIR spectra of 75 fecal samples. First derivatives and
mathematical smoothing features from the TQ Analyst software (Thermofisher Scientific,
Waltham, USA) were used to improve the predictive power of the calibrated PLSR models
(Savitzky and Golay 1964; Williams and Norris 1987). The remaining spectra from 15 randomly
selected fecal samples were used to validate each PLSR model and evaluate its robustness based
on the ratio of prediction to deviation (RPD = Ratio of the standard deviation of predictions to
19
the observed standard deviation). According to Williams (2001), model calibrations are deemed
reliable only if R2 > 0.49 and RPD > 2.3.
Fecal chemistry models
Linear mixed-effects models (nlme R package; Pinheiro and Bates 2000) were used to relate
fecal chemistry to microhistological data stemming from study 2 (Giroux et al., 2012). We
considered the identity of the deer as a random effect variable, which required us to drop one
fecal sample the identity of which was unknown (N=89). We limited the number of explanatory
variables to ADL, cellulose and fecal N in order to avoid multi-collinearity among fecal
chemical properties and to maintain variance inflation factors below 2.5 (Graham 2003). ADL
and cellulose were expressed on a NDF basis in order to reduce the bias induced by variations
in labile substrates in the feces (e.g., microbial biomass, endodermal losses, etc.; Van Soest
1994). Microhistological data were either arcsine or arcsine square root-transformed, in order
to meet the assumption of homoscedasticity (Gilbert 1989). We used the Akaike information
criterion (Akaike 1974; Hurvich and Tsai 1989) corrected for small sample size (AICc) to select
the best fitting model for each forage type.
Results
PC-based discriminant analysis is a modeling method that classifies each fecal sample into 1 of
2 predefined classes (i.e. diets), based on a set of principal components that were derived from
NIR absorbance values. The classification of a given fecal sample is based on the shortest
Mahalanobis distance (i.e. H value) to the mean of each class. Thus, PC-based discriminant
analysis of NIR spectra of feces from 2 different diets used in study 1 correctly classified 28 of
20
30 validation samples (Fig. 1). Altogether, 115 of 132 fecal samples were correctly assigned to
the correct diet (2 = 72.8, P < 0.001).
Figure 1. Results of PC-based discriminant analysis showing the Mahalanobis distances
between the NIR spectra of fecal samples collected from captive deer that had
been fed 2 different diets (Control diet = 70% balsam fir + 20% white spruce +
10% lichens; Poor diet = 50% balsam fir + 40% white spruce + 10% lichens). A
total of 102 samples were used for model calibration and 30 samples for
validation.
Microhistological analyses from study 2 reported fecal samples with 47–100% conifer content
(𝑋̅= 81%), 0–44% deciduous content (𝑋̅= 7%), and 0–17% lichen content (𝑋̅= 5%). Among the
conifer fragments, balsam fir and white spruce respectively represented 0–93% (𝑋̅= 40%) and
0–90% (𝑋̅ = 41%) of the total forage fragments in the feces. PLSR regression models using NIR
spectral information successfully predicted (R2 = 0.89, RPD = 3.2) the proportion of conifer
21
fragments as determined by microhistological analyses (Fig. 2). However, PLSR models were
not robust for predicting the proportions of deciduous (R2 = 0.64, RPD = 1.1), balsam fir (R2 =
0.68, RPD = 1.4), white spruce (R2 = 0.55, RPD = 1.1) or lichen (R2 = 0.36, RPD = 1.0) content
of fecal samples.
Figure 2. Correlation between coniferous content of fecal samples as determined by
microhistology and that predicted by a partial least-squares regression model of
fecal NIR spectra. A total of 75 samples were used for model calibration and 15
samples for validation. RMSEC = root mean squared error of calibration;
RMSEP = root mean squared error of prediction.
Linear mixed-models applied to fecal chemistry data from study 2 best predicted the conifer
content of fecal samples, with 47% of the variance explained when using ADL as the only
explanatory variable (Table 1). For predicting the deciduous content of feces, the best fitting
22
model explained 24% of the variance, and included ADL and cellulose as explanatory variables.
Fecal ADL explained 22% of the variance in balsam fir content, whereas ADL and cellulose
could only explain 8% of the variance for white spruce. No model could explain the variance in
lichen content by fecal chemistry. Fecal N was not a predictor of microhistological data in any
of the models.
Table 1. Best-fitting mixed-linear regression models predicting the fecal content of various
forages based on fecal chemical properties (N = 89)
Pvalue
Explained
variance
(%)
Response
variable
Explanatory
variable
Coefficient
Standard
error
Conifers
ADL
Intercept
5.7
-1.0
0.9
0.3
<0.001
Deciduous
ADL
Cellulose
Intercept
-4.2
3.2
0.6
1.3
1.6
0.8
0.001
0.049
0.468
24
Balsam fir
ADL
Intercept
7.3
-1.4
2.1
0.7
<0.001
<0.001
22
White spruce
ADL
Cellulose
Intercept
-2.6
-4.9
4.0
1.6
2.2
1.0
0.111
0.028
<0.001
8
Lichens
Intercept
(NULL model)
0.38
0.02
<0.001
0.003
47
ADL = Acid detergent lignin.
23
Discussion
Our study showed that NIR spectra of feces from study 1 could discriminate between animals
consuming 2 diets differing only in the relative proportion of balsam fir and white spruce. In the
context of Anticosti Island, this may have practical implications, given that the relative
abundance of white spruce to balsam fir should continue to increase over coming years. NIRS
could, therefore, be used to track a possible concomitant shift in white-tailed deer winter diets.
For example, it would be relatively inexpensive and easy to collect a large number of winter
fecal samples over consecutive years, and to scan these by NIRS. Using our PC-based
discriminant model, the average statistical distance of fecal samples to each calibration diet type
could then be computed in order to detect an increasing use of white spruce in the winter diets
of white-tailed deer.
The potential for NIRS to discriminate between two related simple diets could also be applied
to answer more fundamental questions on the ecology of white-tailed deer on Anticosti Island.
For example, one of the outcomes of the “Optimal Foraging Theory” (MacArthur and Pianka
1966) is that individuals sharing a common home range should converge towards the same diet.
Empirical evidence has shown, however, that non-random variations may occur in the feeding
behaviour of neighbouring individuals of the same age-group (Searle et al. 2010). Hanley (1997)
discussed the possible role of learned behavior and other social factors that may create this
variation in diet choices among individuals. Thus, analyzing the multivariate dispersion among
fecal NIR spectra (e.g., Anderson 2004) could be a novel and cost-efficient way of measuring
the degree of variation in diets within a given home range.
Although PC-based discriminant analysis successfully sorted fecal spectral data according to 2
different diets used in study 1, a regression approach to analyzing both NIR spectra and fecal
24
chemical properties in study 2 had limited success in estimating the relative abundance of
different forage types. This is consistent with past studies that had limited success in developing
robust predictive models for estimating the specific forage composition of diets based on NIR
spectra (e.g. Walker et al. 1998; Landau et al. 2004). It may be argued that our limited success
was due to the unreliability of microhistological data as a calibration tool for NIRS, as suggested
by Walker et al. (2010). But, as previously mentioned, Taillon et al. (2006) reported a high
correlation between the spruce and fir contents of fecal samples as determined by
microhistology, and the actual proportion of these 2 forage types in the diets of captive whitetailed deer on Anticosti Island. These forage types are especially relevant to rangeland
management on Anticosti Island, as they are among the main winter staples for white-tailed deer
and because they differ in nutritional quality (Sauvé and Côté 2007). It is, therefore, unfortunate
that NIRS and fecal chemical properties could not accurately predict the proportion of these 2
forage fragments in fecal samples from study 2.
The salient finding of our study is that both NIRS and fecal chemical properties were relatively
good at predicting the coniferous content of feces from free ranging white-tailed deer (i.e. study
2). We speculate that coniferous forage is a lot less digestible than the other forage types found
on Anticosti Island, which facilitated its detection by NIRS. This is a relatively new application
for fecal NIR spectra, as previous research on wild ruminants has so far only succeeded in using
NIRS to estimate diet nutritional quality rather than the botanical composition of different diets
(Kamler et al. 2004; Showers et al. 2006). The fact that we succeeded is probably due to the
fact that ecosystems on Anticosti Island are relatively simple, with only a few winter forage
types available to browsers. Although we were hoping to predict specific forage types, being
able to predict the percent conifer content is nevertheless useful, as it is a good indicator of
quality food shortages during the winter (Häsler and Senn 2012).
The fact that fecal ADL was positively correlated to conifer fragments (i.e. needles and twigs),
and negatively correlated to deciduous fragments (i.e., twigs and young bark), suggests that the
25
former may have been richer in lignin than the latter. We have difficulty explaining, however,
why fecal ADL correlated to balsam fir fragments but not to white spruce fragments, as both
forage types contain approximately the same amounts of lignin (Sauvé and Côté 2007). Dumont
et al. (2005) demonstrated a positive relationship between different assumed bite sizes of balsam
fir forage by white-tailed deer and the ADL concentration of each morsel. Thus, the relationship
between fecal ADL and balsam fir fragments may reflect larger bites by white-tailed deer when
browsing balsam fir, their preferred winter forage type.
Implications
Our study has shown the potential and limitations of fecal NIR spectra, and of fecal chemical
properties, as rapid and cost-efficient indices of the winter diet composition of white-tailed deer
on Anticosti Island. Both of these tools could be used, therefore, to launch a low cost monitoring
program of winter resources for the white-tailed deer in Anticosti. Only a few fecal samples per
relevant home range would be required each year to detect changes in winter feeding behaviours.
The low costs associated to both analytical methods allow rangeland managers to substantially
increase their sampling efforts.
Acknowledgements
We thank J. Taillon, W. F. J. Parsons, S. Rioux-Paquette, S. Rivard, B. Baillargeon, S. De
Bellefeuille, C. Hins, D. Duteau, J.-F. Adam, B. Chénard, J.-P. Harnois, A. Hemery, J. HénaultRichard, K. Ladhiri, G. Laprise, L. L’Italien, J. Malouin, D. Morin, G. Reuchet, L. Richard, M.
O’Connor, M. Sicilia, R. Théoret-Gosselin and A. Veillette for technical and logistical support.
26
References
Akaike, H. 1974. A new look at the statistical model identification. IEEE Transactions on
automatic control 19:716–723.
Anderson, M.J. 2004. PERMDISP: a FORTRAN computer program for permutational analysis
of multivariate dispersions (for any two-factor ANOVA design) using permutation tests.
Department of Statistics, University of Auckland, New-Zealand.
Bugalho, M.N., H. Dove, and W. Kelman. 2004. Plant wax alkanes and alcohols as herbivore
diet composition markers. Rangeland Ecology & Management 57:259–268.
Bugalho, M.N., R.W. Mayes, and J.A. Milne. 2002. The effects of feeding selectivity on the
estimation of diet composition using the n-alkane technique. Grass and Forage Science 57:224–
231.
Dalerum, F. and A. Angerbjörn. 2005. Resolving temporal variation in vertebrate diets using
naturally occurring stable isotopes. Oecologia 144:647–58.
Dearden, B.L., R.E. Pegau, and R.M. Hansen. 1975. Precision of microhistological estimates of
ruminant food habits. The Journal of Wildlife Management 39:402–407.
Dixon, R., and D. Coates. 2009. Review: Near infrared spectroscopy of faeces to evaluate the
nutrition and physiology of herbivores. Journal of Near Infrared Spectroscopy 17:1-31.
Dove, H., and R.W. Mayes. 2005. Using n-alkanes and other plant wax components to estimate
intake, digestibility and diet composition of grazing/browsing sheep and goats. Small Ruminant
Research 59:123–139.
Dufresne, M., R.L. Bradley, J.-P. Tremblay, M. Poulin, and S. Pellerin. 2009. Clearcutting and
deer browsing intensity interact in controlling nitrification rates in forest floor. Ecoscience
16:361–368.
Dumont, A., J.-P. Ouellet, M. Crête, and J. Huot. 2005. Winter foraging strategy of white-tailed
deer at the northern limit of its range. Ecoscience 12:476–484.
Environment Canada. 2006. Climate normals and averages daily data reports of Port-Menier’s
station from 1995 to 2005. Available at http://www.climate.weatheroffice.ec.gc.ca, Accessed
June 8, 2006.
Gilbert, N. 1989. Biometrical interpretation. Making sense of statistics in biology. New York,
NY, USA: Oxford University Press. 146 p.
27
Giroux, M.-A., C. Dussault, N. Lecomte, J.-P. Tremblay, and S.D. Côté, 2012. A new way of
assessing foraging behaviour at the individual level using faeces marking and satellite telemetry.
PloS one 7:e49719. doi:10.1371/journal.pone.0049719
Goering, H.K., and P.J. Van Soest. 1970. Forage fiber analysis apparatus (reagents, procedures,
and some applications). Agriculture handbook No.379. US Department of Agriculture.
Agricultural Research Service, Washington, DC, USA. 20 p.
Graham, M.H., 2003. Confronting multicollinearity in ecological multiple regression. Ecology
84:2809–2815.
Grondin, P., C. Ansseau, L. Bélanger, J.-F. Bergeron, Y. Bergeron, A. Bouchard, J. Brisson, L.
De Grandpré, G. Gagnon, C. Lavoie, G. Lessard, S. Payette, P. J. H. Richard, J. P. Saucier, L.
Sirois, and L. Vasseur. 1996. Cadre bioclimatique de référence des régions écologiques du
Québec. In: Manuel de Foresterie. Les Presses de l’Université Laval. Québec. Canada. p.134279. (in French)
Hanley, T.A., 1997. A Nutritional View of Understanding and Complexity in the Problem of
Diet Selection by Deer (Cervidae), Oikos 79: 209-218.
Hanley, T.A., and J.D. McKendrick. 1985. Potential nutritional limitations for black-tailed deer
in a spruce-hemlock forest, southeastern Alaska. The Journal of Wildlife Management 49:103–
114.
Häsler, H., and J. Senn. 2012. Ungulate browsing on European silver fir Abies alba : the role of
occasions, food shortage and diet preferences. Wildlife Biology 18:67–74.
Hidding, B., J.-P. Tremblay, and S.D. Côté. 2013. A large herbivore triggers alternative
successional trajectories in the boreal forest. Ecology (in press).
Hodgman, T.P., and B.B. Davitt. 1996. Monitoring mule deer diet quality and intake with fecal
indices. Journal of Range Management 49:215–222.
Hurvich, C.M., and C.L. Tsai. 1989. Regression and time series model selection in small
samples. Biometrika 76:297–307.
Kamler, J., M. Homolka, and D. Cizmár. 2004. Suitability of NIRS analysis for estimating diet
quality of free-living red deer Cervus elaphus and roe deer Capreolus capreolus. Wildlife
Biology, 3:235-240.
Landau, S., T. Glasser, L. Dvash, and A. Perevolotsky. 2004. Fecal NIRS to monitor the diet of
Mediterranean goats. South African Journal of Animal Science 34:76–80.
28
Landau, S., T. Glasser, and L. Dvash. 2006. Monitoring nutrition in small ruminants with the
aid of near infrared reflectance spectroscopy (NIRS) technology: A review. Small Ruminant
Research, 61:1-11.
Lefort, S. 2002. Habitat hivernal du cerf de Virginie (Odocoileus virginianus) à l’île d'Anticosti
[thesis]. Québec, QC, Canada. Université Laval. 87 p. (in French)
Lefort, S.,J.-P. Tremblay, J.-P. and F. Fournier. 2007. Importance of balsam fir as winter forage
for white-tailed deer at the northeastern limit of their distribution range. Ecoscience 14 :109–
116.
Leslie, D.M., Jr., R.T. Bowyer, and J.A. Jenks. 2008. Facts from feces: Nitrogen still measures
up as a nutritional index for mammalian herbivores. Journal of Wildlife Management 72 :1420–
1433.
MacArthur, R.H., and E.R. Pianka. 1966. On optimal use of a patchy environment. American
Naturalist 100:603–609.
Malley, D.F., L. Yesmin, and R.G. Eilers. 2002. Rapid Analysis of Hog Manure and Manureamended Soils Using Near-infrared Spectroscopy. Soil Science Society of America Journal
66:1677.
McIlwee, A.M., I.R. Lawler, S.J. Cork, and W.J. Foley. 2001. Coping with chemical complexity
in mammal-plant interactions: near-infrared spectroscopy as a predictor of Eucalyptus foliar
nutrients and of the feeding rates of folivorous marsupials. Oecologia 128:539–548.
McInnis, M.L., M. Vavra, and W.C. Krueger. 1983. A comparison of four methods used to
determine the diets of large herbivores. Journal of Range Management 36:302–306.
Parker, K.L., M.P. Gillingham, T.A. Hanley, and C.T. Robbins. 1999. Energy and protein
balance of free-ranging black-tailed deer in a natural forest environment. Wildlife Monographs
143:3-48.
Pinheiro, J. C., and D. M. Bates. 2000. Mixed-Effects Models in S and S-PLUS. New-York:
Springer. 528 p.
Potvin, F., P. Beaupré, and G. Laprise. 2003. The eradication of balsam fir stands by whitetailed deer on Anticosti Island , Québec : A 150-year process. Ecoscience 10 :487–495.
Potvin, F., P. Beaupré, and C. Dussault. 2004. La population de cerfs de l'île d'Anticosti:
élaboration de modèles densité-habitat et prévisions pour l'horizon 2000-2100, Report;
November 2004; Québec, QC, Canada. NSERC-Produits forestiers Anticosti Industrial
Research Chair. 60 p. (in French)
29
Sauvé, D., 2005. Les effets de la détérioration de la qualité de l’alimentation d'hiver sur la
condition corporelle du cerf de Virginie de l'Île d'Anticosti [thesis]. Québec, QC, Canada,
Université Laval. 116 p. (in French)
Sauvé, D.G., and S.D. Coté. 2007. Winter forage selection in white-tailed deer at high density:
Balsam fir is the best of a bad choice. Journal of Wildlife Management 71: 911–914.
Savitzky, A. and M.J.E. Golay.1964. Smoothing and differentiation of data by simplified least
squares procedures. Analytical Chemistry 36:1627–1639.
Searle, K.R., L.P. Hunt and I.J. Gordon. 2010. Individualistic herds: Individual variation in
herbivore foraging behavior and application to rangeland management. Applied Animal
Behaviour Science 122:1–12.
Shipley, L.A., S. Blomquist, and K. Danell. 1998. Diet choices made by free-ranging moose in
northern Sweden in relation to plant distribution, chemistry, and morphology. Canadian Journal
of Zoology 76:1722–1733.
Stewart, D.R.M. 1967. Analysis of plant epidermis in faeces: a technique for studying the food
preferences of grazing herbivores. Journal of Applied Ecology 4:83–111.
Stolter, C., R. Julkunen-Tiitto, and J.U. Ganzhorn. 2006. Application of near infrared reflectance
spectroscopy (NIRS) to assess some properties of a sub-arctic ecosystem. Basic and Applied
Ecology 7:167–187.
Stuth, J.W., A. Jama, and D.R. Tolleson. 2003. Direct and indirect means of predicting forage
quality through near infrared reflectance spectroscopy. Field Crops Research 84:45–56.
Taillon, J., D.G. Sauvé, and S.D. Coté. 2006. The effects of decreasing winter diet quality on
foraging behavior and life-history traits of white-tailed deer fawns. Journal of Wildlife
Management 70:1445–1454.
Valentini, A., C. Miquel, M.A. Nawaz, E. Bellemain, E. Coissac, F. Pompanon, L. Gielly, C.
Cruaud, G. Nascetti, P. Wincker, J.E. Swenson, and P. Taberlet. 2009a. New perspectives in
diet analysis based on DNA barcoding and parallel pyrosequencing: the trnL approach.
Molecular Ecology Resources 9:51–60.
Valentini, A., F. Pompanon, and P. Taberlet. 2009b. DNA barcoding for ecologists. Trends in
Ecology and Evolution 24:110–117.
Van Soest, P.J. 1994. Nutritional ecology of the ruminant. Ithaca, NY, USA.: Cornell University
Press.476 p.
30
Verheyden, H., L. Aubry, J. Merlet, P. Petibon, B. Chauveau-Duriot, N. Guillon, and P. Duncan.
2011. Faecal nitrogen, an index of diet quality in roe deer Capreolus capreolus? Wildlife Biology
17:166–175.
Walker, J. W., D. H. Clark, and S. D. McCoy. 1998. Fecal NIRS for predicting percent leafy
spurge in diets. Journal of Range Management 51:450–455.
Walker, J.W., E.J. Campbell, R.W. Kott, S. Landau, C.J. Lupton, C.B. Scott, L. Surber, C.A.
Taylor Jr, and W.R. Whitworth. 2010. Chapter 5. Fecal NIRS for Predicting Botanical
Composition of Herbivore Diets. In J. W. Walker, D. R. Tolleson, and S. Byrns [eds.]. Shining
light on manure improves livestock and land management. Jefferson, MO, USA: Texas AgriLife
Research and Society for Range Management, p. 53–65.
Wam, H.K., and O. Hjeljord. 2010. Moose Summer Diet From Feces and Field Surveys: A
Comparative Study. Rangeland Ecology & Management 63:387–395.
Williams, P. and K.H. Norris. 1987. Near-infrared technology in the agricultural and food
industries. St-Paul, MN, USA: American Association of Cereal Chemists. 330 p.
Williams, P.C. 2001. Implementation of near-infrared technology. In Williams P.C and K.H.
Norris [eds.]. Near-infrared Technology in the Agricultural and Food Industries. St-Paul,
MN,USA: American Association of Cereal Chemists, p. 145–170.
Wold, H. 1966. Estimation of principal components and related models by iterative least
squares. In P. R. Krishnaiah [ed.] Multivariate analysis. New-York, NY, USA: Academic press,
p. 391–420.
31
CHAPITRE 2
MÉTHODES DE SUBSTITUTION AUX
PROTOLES DE DIGESTION IN VITRO
POUR L’ÉTUDE DE L’ÉCOLOGIE
NUTRITIONNELLE DES RUMINANTS
SAUVAGES
32
Avant-propos
Différents protocoles ont été développés afin de simuler la digestion ruminale en utilisant un
incubateur et un inoculum provenant du rumen d’un animal donneur. Grâce à ces méthodes, on
peut mesurer la digestibilité ainsi que les différents produits de fermentation générés lors de la
digestion des fourrages, ce qui permet du même coup de statuer sur la qualité nutritionnelle de
ces derniers. Lors de mes lectures, je me suis rapidement aperçu que les différentes techniques
disponibles étaient bien adaptées à la nutrition du bétail, mais que leur utilisation dans le cadre
d’études écologiques était susceptible d’entraîner des biais importants.
Puisqu’il est
problématique d’effectuer ce genre d’étude avec une liqueur ruminale provenant d’un animal
sauvage, je me suis intéressé dans ce chapitre au développement de méthodes de substitution en
mettant l’emphase sur l’amélioration de notre compréhension de l’écologie nutritionnelle des
ruminants sauvages.
Les données obtenues pour la rédaction de cet article résultent d’un travail de longue haleine
réalisé lors des automnes 2010, 2011 et 2012. La rédaction a été effectuée en collaboration avec
mes directeurs de recherche, Robert L. Bradley et Jean-Pierre Tremblay, de même qu’avec
Robert Berthiaume, expert en nutrition des ruminants. Cet article a été soumis à la revue Wildlife
Society Bulletin le 7 avril 2015, puis resoumis le 7 août 2015 en version révisée.
33
Evaluating old and novel proxies for in vitro digestion assays in wild
ruminant ecology
Pierre-Olivier JEAN*1, Robert L. BRADLEY1, Robert Berthiaume2 & Jean-Pierre
TREMBLAY3
1
Département de biologie, Université de Sherbrooke, 2500 boulevard de l'Université,
Sherbrooke, Québec, Canada J1K 2R1.
2
Dairy and Swine Research and Development Centre, Agriculture and Agri-Food Canada,
Lennoxville, QC, J1M 1Z3, Canada; 3Valacta Dairy Production, Centre of Expertise for Quebec
and the Atlantic, Sainte-Anne-de-Bellevue, QC, H9X 3R4, Canada (present affiliation)
3
Département de biologie, Université Laval, Québec, QC, G1V0A6, Canada.
Keywords: dairy cow, forage quality, in vitro digestibility, monitoring tool, near infrared
spectroscopy, rumen gas production, ruminant nutrition, volatile fatty acids, white-tailed deer.
Abbreviations: IVTD, In vitro true digestibility; VFA, volatile fatty acids; ADF, acid detergent
fibers; NIRS, near infrared spectroscopy; RMSEVAL, root-mean square error of validation; Gmax,
maximal gas production; KM , time corresponding to ½ Gmax .
34
Résumé
Différents protocoles de digestion In vitro fournissent des données utiles sur l’écologie
nutritionnelle des ruminants sauvages. Ces études restent cependant difficiles à conduire en
régions éloignées avec une liqueur ruminale provenant de ruminants sauvages. Dans cette étude,
nous évaluons des méthodes de substitution aux digestion In vitro pour mesurer la digestibilité
vraie des fourrages (IVTD), le dégagement de gaz et la production d’acides gras volatils (VFA)
lorsque déterminés avec une liqueur ruminale de cerf de Virginie. Ces recherches ont été menées
à l’île d’Anticosti, Canada, où le cerf de Virginie est l’herbivore principal. Des tissus foliaires
de 2 plantes consommées en été (Cornouiller quatre-temps; Maianthème du Canada) et 2 plantes
consommées principalement en hiver (sapin beaumier, épinette blanche) ont été amassées dans
une aire d’étude de 2500 km2. Des liqueurs ruminales de cerf de Virginie fraîchement abattus
à la chasse ont été utilisées afin de déterminer l’IVTD ainsi que les valeurs totales de
dégagement de gaz et de production de VFAs lors de la fermentation de ces 4 fourrages
consommés par le cerf. Puis, nous avons évalués 4 méthodes de substitution permettant de
prédire ces variables : (1) L’utilisation de liqueur ruminal provenant de vache laitière, (2) la
mesure de la concentration en ADF (i.e. acid-detergent fibers), (3) la production de CO2 d’un
mélange plante-sol et (4) l’utilisation modèles prédictifs utilisant la spectroscopie dans le proche
infrarouge (NIRS). Les modèles NIRS furent les plus performants pour prédire les valeurs
d’IVTD, de gas et de VFA, tout en étant peu biaisés dans leurs prédictions pour différentes
espèces de plantes. Les concentrations en ADF ainsi que le dégagement de CO2 de mélanges
plante-sol étant relativement bien corrélés avec le montant de gas et de VFA générés.
L’utilisation d’un inoculum de vache laitière a relativement bien performé pour prédire les
valeurs d’IVTD obtenus avec un inoculum de cerf (R2=0,94), mais pas pour prédire le montant
de gas et de VFA générés (R2<0,63). Des modèles NIRS calibrés ont le potentiel d’améliorer la
prédiction de la digestibilité des fourrages ainsi que les produits de digestion tels que le gas et
les VFAs. Si toutefois il est impossible d’accéder à cette technique, des mesures d’ADF ou de
dégagement de CO2 d’un mélange plante-sol sont des alternatives intéressantes.
La
35
comparaison de la cinétique de la digestion ainsi que les mesures de production de VFA lors de
digestion In vitro d’herbivores sauvages et domestiques est une technique prometteuse afin de
mieux comprendre l’écophysiologie de la digestion par différents herbivores.
Abstract
In vitro digestion assays provide useful data for wild ruminant ecologists. It remains highly
constraining, however, to conduct these assays in the field. We evaluated various proxies for in
vitro digestion assays that use ruminal liquor from wild ruminants. The ruminal liquor of freshly
killed deer was used to measure in vitro true digestibility (IVTD), total gas and volatile fatty
acids (VFA) production. We then evaluated the ability of 4 proxies to predict these digestibility
variables: (1) IVTD, total gas and VFA produced with cow ruminal liquor; (2) acid detergent
fiber (ADF) concentration; (3) CO2 production from soil-forage mixtures; (4) predictive models
based on near infrared spectra (NIRS) of forages. NIRS predictions were best correlated with
all deer digestibility variables (each R2 > 0.93) with limited bias across forage species. Forage
ADF and CO2 production from soil-forage mixtures correlated relatively well with all
digestibility variables (R2 = 0.62 to 0.95). Digestion assays using cow ruminal liquor was
relatively good for predicting IVTD (R2 = 0.94) from deer ruminal liquor, but the weakest for
predicting total gas (R2 = 0.63) and VFA production (R2 = 0.44). Calibrated NIRS models could
substantially improve the prediction of in vitro digestibility variables. If access to NIRS
technology or the ability to collect reference data with deer inocula for NIRS calibrations is
impractical, forage ADF or CO2 production from soil-forage mixtures remain interesting
alternatives. Comparing the digestion kinetics and VFA production of wild and domestic
ruminants is an innovative avenue for better understanding the adaptive ecophysiology of
digestion by different herbivores.
36
Introduction
In vitro digestibility assays using authentic ruminal liquor have extensively been used by both
livestock scientists and ecologists to study the nutritional quality of different forages for
ruminants. These methods have been useful to optimize feed energy conversion (Reynold et al.
2011) as well as to minimize greenhouse gas emissions by domestic ruminants (Bodas et al.
2008). For ecologists working with wild herbivores, in vitro digestibility assays have been used
to describe the nutritional quality of forages in order to explain diet selection (Crawford 1982,
Côté 1998) and to characterize the suitability of habitats (Hanley et al. 2012). In short, these
assays consist of measuring forage mass loss during an incubation with ruminal liquor, followed
by a pepsin bath or neutral detergent treatment to simulate post-ruminal digestion (Tilley and
Terry 1963, Goering and Van Soest 1970).Complementary assays, such as measuring
fermentation gases and volatile fatty acids (VFA) using ruminal liquor, can also yield
information on the adaptive ecophysiology of wild ruminants (Hofmann 1989) as well as on the
metabolic pathways that are favoured during digestion (Smith and Crouse 1984, Getachew et
al. 2004). For example, Russell and Strobel (1989) showed that fermentation rates and high
production of propionate were correlated with energy use efficiency during digestion. It remains
highly constraining, however, to conduct in vitro digestibility assays in remote areas, where
fresh ruminal liquor from free-ranging wild ruminants is difficult to obtain, and where the
absence of infrastructure requires the transport of a makeshift laboratory. To cope with these
logistical constraints, ecologists have looked for reliable proxies that could reflect differences
in digestibility and energy yielding values of various forage types.
A first option for predicting in vitro digestibility of forages by wild ruminants is to use a more
accessible alternative, such as ruminal liquor from domestic animals (e.g. Harvey and Fortin
2013; Tollefson et al. 2011). In doing so, the researcher assumes that the digestive performance
of the ruminal microflora is the same for both the wild and domestic subjects. This assumption
may be tenuous, however, as studies have shown strong differences in the catabolic profile of
37
ruminal liquor in different species, as well as in different individuals of the same species that
were fed different diets (Campa et al. 1984, Clary et al. 1988). A second option for predicting
in vitro digestibility of forages by wild ruminants has been to quantify the carbon fractions of
forages usually suspected of limiting digestibility. For example, acid detergent fibres (ADF)
represent structural carbohydrates that limit digestibility (Van Soest 1978, Lippke 1980), and
this variable has been used to estimate forage quality for free-ranging ruminants (e.g. Marell et
al. 2006, McCuistion et al. 2014). The main advantage of using forage ADF is that the method
is inexpensive and rapid. A third option for predicting in vitro digestibility of forages by wild
ruminants would be to measure the aerobic decomposition rate of forages. For example,
Cornelissen et al. (2004) successfully correlated mass loss of different forages in a soil litter bed
to in vitro digestion rates using ruminal liquor. Although soil and rumen microbial communities
are starkly different in most aspects, the chemical attributes of forages that regulate the speed at
which they are utilized by both microbial guilds should roughly be the same (Chesson 1997).
By far, this is the most simple and least expensive proxy for in vitro digestibility of forages.
Finally, a fourth option for predicting in vitro digestibility of forages by wild ruminants is to
develop predictive models based on near infrared spectroscopy (NIRS). This technique consists
of regressing in vitro digestibility values on the near infrared absorbance spectra of forages
(Decruyenaere et al. 2009, Andueza et al. 2011). The advantage, relative to forage ADF, is that
NIRS considers the entire chemical composition of forages that may affect digestibility (Foley
et al. 1998), compounds such as tannins for example (Robbins et al. 1987b). However, unlike
the three previous methods, this approach unconditionally requires a reference dataset of in vitro
digestibility values using wild ruminant liquor.
Motivated by a need to gather reliable data on forage quality for a remote population of whitetailed deer (Odocoileus virginianus Zimmermann), we performed a study to evaluate which of
the aforementioned proxies (i.e. digestibility using cow inoculum, forage ADF, aerobic
decomposition rates and NIRS) could best predict in vitro digestibility and digestion byproducts (i.e. gas and VFA production) using white-tailed deer ruminal liquor. To do so, we
used 4 forage plant species representing important summer and winter staples for white-tailed
38
deer in our study area. Our second objective was to compare the kinetics and the types of VFAs
produced by deer and cow inocula, with an emphasis on the ecological constraints that drive
wild ruminant nutrition.
Material and Methods
Study area
Anticosti Island (49°28'N 63°00'W) is located in the Gulf of St-Lawrence, Québec, Canada. The
entire island (7,943 km2) lies within the Eastern Balsam Fir-White Birch bioclimatic zone
(Grondin et al. 1996). The sub-boreal maritime climate results in cool summers (mean July
temperature of 16 °C) and cold winters (mean January temperature of -11 °C), while the average
annual precipitation is 917 mm (Environment Canada 2006) of which about one-third falls as
snow.
About 200 white-tailed deer were introduced on the island in 1896, and the population increased
rapidly in the absence of natural predators. Accordingly, white-tailed deer has been drastically
transforming plant communities on Anticosti island over the past 80 years (Marie-Victorin and
Rolland-German 1969), leading to a decrease in forage quality (Lefort et al. 2007). Heavy
browsing has resulted in a loss of palatable species in the shrub layer such as Mountain maple
(Acer spicatum Lam.), Beaked hazel (Corylus cornuta Marsh.) and Canadian yew (Taxus
canadensis Marsh) (Potvin et al. 2003), and in the persistence of browse tolerant herbaceous
species such as Canada bunchberry (Cornus canadensis L.), Canada mayflower (Maianthemum
canadense Desf.) and graminoids such as Canada bluejoint (Calamagrostis canadensis
(Michx.)) (Tremblay et al. 2006). Balsam fir (Abies balsamea (L.) P.Mill.) is the preferred
winter forage (Lefort et al. 2007) but its low recruitment has resulted in an increased
39
consumption of less palatable white spruce foliage (Picea glauca (Moench) Voss.) (Hidding et
al. 2013).
Sampling
On each of 21 randomly selected sampling sites, within a 2,500 km2 area in the western sector
of Anticosti island, we collected foliar samples of 2 summer forage species (Canada bunchberry
and Canadian mayflower) in mid-July, and 2 winter forage species (balsam fir and white spruce)
in mid-September. The 84 forage samples were oven-dried at 50 °C for 1 week and ground using
a centrifugal mill (Retsch, Haan, Germany) equipped with a 2 mm mesh sieve.
In vitro digestion assays using white-tailed deer inoculum
In mid-September, we accompanied hunters on Anticosti Island and collected the rumen of
freshly killed deer (i.e. 7 rumens over 2 weeks). Each rumen was ligatured to maintain
anaerobiosis and conserved in a 40 °C water-filled insulated chamber for transport to a
makeshift laboratory. Time of transport never exceeded 2 h. The rumen content was promptly
filtered through 4 layers of cheese cloth and in vitro true digestibility (IVTD) of the 84 foliar
Incubator (Ankom Technology, Fairport NY) according to the protocol of Goering and Van
Soest (1970). These steps were repeated for each of the 7 rumens from which we calculated the
average IVTD for each foliar sample.
In vitro gas and VFA production were measured following the protocol described by Theodorou
et al. (1994). Because these analyses are time consuming and require a large amount of plant
40
material, we pooled the remaining foliar material into 3 composite samples per forage type
(N=12). Ruminal liquor from 3 additional deer was obtained as previously described, and each
was repeated for each of the 3 deer ruminal liquors. Fermentation gas production was measured
11 times during each incubation using a syringe and a manometer fitted to a luer-lock valve.
Following each incubation, the content of each vial was kept frozen until the liquid portion was
analyzed for acetate, propionate, butyrate and total VFA concentrations using a Varian 3700 gas
chromatograph (Varian Specialities Ltd., Brockville, Canada). Each of these VFAs was then
reported as a proportion of total VFA concentration.
Proxies to estimate white-tailed deer in vitro digestion
IVTD, in vitro gas and VFA production using cow inoculum were measured using the same
protocols as for the deer inoculum. Cow inoculum was obtained from a single fistulated Holstein
cow at the Agriculture and Agri-Food Canada experimental farm in Lennoxville, Canada. IVTD
measurements using cow inoculum were repeated 3 times whereas in vitro gas and VFA
production were repeated twice.
Forage ADF concentration of the 84 forage samples was determined in duplicate assays
(Goering and Van Soest 1970) using an Ankom Fiber Analyzer (Ankom Technology, Macedon,
NY).
For the 12 composite forage samples, CO2 production of soil-forage mixtures were evaluated in
duplicate assays. We mixed 1 g of each composite forage sample to 50 g of fresh old-field fallow
soil (sandy loam texture; pH in H2O = 5.7; organic matter = 13.7%; extractable P = 0.41 ppm;
41
total N = 0.6 % w/w) and transferred these mixtures into 250 ml jars with air-tight lids equipped
with rubber septa. CO2 concentration in the headspace was measured after 48 h incubation at
room temperature using a Varian 431-GC gas chromatograph (Agilent, Santa Clara, California).
In order to generate NIR spectra, the 84 individual forage samples as well as the 12 composite
forage samples (i.e. used to measure gas and VFA production) were scanned for their light
absorbance at wavelengths ranging from 1,000 to 2,500 nm, using a Nicolet Antaris FT-NIR
spectroscope supported with Omnic software (Thermofisher Scientific, Waltham, USA).
In order to generate NIR spectra, the 84 individual forage samples as well as the 12 composite
forage samples (i.e. used to measure gas and VFA production) were scanned for their light
absorbance at wavelengths ranging from 1,000 to 2,500 nm, using a Nicolet Antaris FT-NIR
spectroscope supported with Omnic software (Thermofisher Scientific, Waltham, USA).
Statistical analysis
Regression type II models (Legendre 2013, R package lmodel2) were used to test the suitability
of cow inoculum, forage ADF and CO2 production from soil-forage mixtures as proxies for
IVTD and digestion by-products (i.e. total gas and total VFAs) using deer inoculum. The slope
of the relationship was estimated using the ordinary least squares method. The regression
coefficients for these models were determined by permutation, because the distributional
assumptions of parametric testing were not satisfied (Edgington and Onghena 2007). The
potential for NIRS to predict the same 3 variables was investigated using partial least square
regressions, as implemented in TQ Analyst software (Thermofisher Scientific, Waltham,USA).
Regions of the spectra that correlated best with the response variable were automatically
42
selected by the software through an iterative process. NIRS predictive models for IVTD were
evaluated using 2/3 of the dataset for model calibration and 1/3 for validation (N=84), while
making sure each forage species was equally represented in both validation and calibration sets.
Because of our small sample size (N=12) for the estimation of total gas and VFA production,
we validated the models using a leave-one-out cross-validation technique (Lachenbruch and
Mickey 1968). The final predictive models were selected for their lowest root-mean square error
of validation (RMSEVAL).
To compare the gas production kinetics of deer and cow inocula, we first fitted a classic
Michaelis-Menten equation to the gas production profiles of each of the 12 composite forage
samples, using the drc R package (Ritz and Strebig 2014). We then estimated the maximal gas
production (Gmax) as well as the time (KM) corresponding to ½ Gmax for each forage species.
The mean values of each of these 2 parameters obtained with deer or cow inocula were compared
using two-sided t-tests. Similarly, we compared the effects of each inocula on the relative
amount of each VFA (i.e. acetate, propionate and butyrate) and total VFA production using twosided t-tests.
Results
Proxies for true digestibility, total gas and VFA production using deer inoculum
All 4 proxies correlated significantly with IVTD from deer inoculum (Fig. 1), with NIRS being
the best predictor (R2=0.99) and soil CO2 production being the weakest predictor (R2=0.65).
Both IVTDcow and soil CO2 production overestimated IVTDdeer values for Canada mayflower
43
and underestimated those for Canada bunchberry. Moreover, soil CO2 production
underestimated IVTDdeer values for balsam fir.
All 4 proxies correlated significantly with total gas and VFA production using deer inoculum
(Fig. 2 and 3). For these 2 variables, forage ADF and NIRS were the 2 best proxies (R2= 0.93
and 0.95) whereas gas and VFA production using cow inoculum were the weakest predictors
(R2= 0.63 and 0.44). Once again, cow inoculum overestimated values for Canada mayflower
and underestimated values for Canada bunchberry. Soil CO2 production was a better predictor
of total gas (R2=0.83) and VFA (R2=0.73) production than it had been for IVTD.
44
Figure 1. Relationship between in vitro true digestibility (IVTD) of 4 forage species
determined using a deer inoculum and (a) IVTD values using a cow inoculum,
(b) forage acid detergent fibers (ADF), (c) 48 h CO2 production from soil-forage
mixtures, and (d) prediction equations based on near infrared spectra (NIRS)
of forage. Cross bars in frame (c) represent standard errors on replicated
measures. NIRS predictive model was calibrated using 56 samples and
validated by calculating root-mean square error (RMSEVAL) of remaining 28
samples.
45
Figure 2. Relationship between in vitro gas production of 4 forage species determined using
a deer inoculum and (a) in vitro gas production using a cow inoculum, (b) forage
acid detergent fibers (ADF), (c) 48 h CO2 production from soil-forage mixtures,
and (d) prediction equations based on near infrared spectra (NIRS) of forage.
Cross bars represent standard errors on replicated measures. NIRS predictive
model was validated using a leave-one-out cross-validation procedure in order
to calculate the root-mean square error value (RMSEVAL) of the 12 composite
forage samples.
46
Figure 3. Relationship between in vitro volatile fatty acids (VFA) production of 4 forage
species determined using a deer inoculum and (a) in vitro VFA production using
a cow inoculum, (b) forage acid detergent fibers (ADF), (c) 48 h CO2 production
from soil-forage mixtures, and (d) prediction equations based on near infrared
spectra (NIRS) of forage. Cross bars represent standard errors on replicated
measures. NIRS predictive model was validated using a leave-one-out crossvalidation procedure in order to calculate the root-mean square error value
(RMSEVAL) of the 12 composite forage samples.
47
Digestion kinetics and VFA production
KM values for balsam fir, Canada mayflower and white spruce were about twice as low for deer
inoculum compared to cow inoculum (Fig. 4, Table 1). On the other hand, Gmax values using
deer inoculum were lower for Canada mayflower and white spruce, but higher for Canada
bunchberry (Table 1). With the exception of Canada bunchberry, deer inoculum produced a
lower relative amount of acetate, and a higher relative amount of propionate and butyrate, than
the cow inoculum (Fig. 5). For balsam fir and Canada bunchberry, total VFA production was
higher for deer than for cow inoculum (Fig. 5).
Figure 4. Cumulative gas production profiles from (a) deer and (b) cow inocula during a
48 h in vitro digestion assay.
48
Table 1: Results of Student’s two sided t-tests comparing the effect of deer vs. cow
inocula on Km and Gmax values.
Deer
Cow
Forage
Balsam fir
KM (h)
6.0
KM (h)
12.0
df
4
T
-9.44
p-value
>0.001
White spruce
4.8
9.5
4
-18.38
>0.001
Canada bunchberry
19.9
19.3
4
0.15
0.886
Canada mayflower
5.7
10.7
4
-7.52
0.002
Balsam fir
Gmax(ml)
139
Gmax(ml)
146
4
-0.79
0.473
White spruce
101
128
4
-4.69
0.009
Canada bunchberry
245
169
4
17.90
>0.001
Canada mayflower
241
323
4
-22.93
>0.001
Where Gmax = asymptotic gas production and KM = time required to reach ½ Gmax
Discussion
Proxies for IVTD, total gas and VFA production using deer inoculum
Our results show that in vitro digestibility assays using cow inoculum are not an ideal proxy for
IVTD, total gas and VFA production using deer inoculum. The fact that values for Canada
bunchberry were underestimated whereas those for Canadian mayflower were overestimated
suggests that assays that use cow inoculum run the risk of improperly ranking the nutritional
quality of some forages. The fact that in vitro digestibility assays using cow inoculum was the
poorest predictor of total gas and VFA production with deer inoculum reflects strong differences
49
in the metabolic potential of ruminal microflora from each ruminant species. This was pointed
out
by Hofmann
(1989)
when
he
discussed
the
adaptive
differences
in
the
amylolytic:cellulolytic enzyme ratio of ruminal microflora between wild and domestic
ruminants.
Figure 5. Results of Student’s two sided t-tests comparing the effect of deer vs. cow
inocula on the relative production of acetate, propionate and butyrate (first 3
bar clusters from the left), as well as on the total amount of volatile fatty acids
(right bar cluster) produced during a 48 h in vitro digestion assay of 4 forage
species.
50
Forage ADF was a good predictor of IVTD, and a better predictor of total gas and VFA
production than in vitro digestibility values using cow inoculum. In fact, the correlations
between forage ADF and both total gas and VFA production were much higher for deer than
those reported in the literature for domestic ruminants (Getachew et al. 2004, Kamalak et al.
2005, Cerrillo and Juaréz 2004). These higher correlations suggest that deer ruminal digestion
rates on Anticosti Island might be limited by structural carbohydrate concentrations (i.e. ADF)
rather than by other forage quality attributes such as tannins (see discussion below). Whether
forage ADF is a good proxy of digestion for other wild ruminants, or by white-tailed deer found
in other ecosystems, remains to be seen. Nevertheless, the low cost of forage ADF analyses, and
the fact that forage ADF data can be gleaned from the literature, makes it an interesting proxy
to consider when undertaking ecological studies on the nutritional quality of forages.
The relatively weak correlation between CO2 production from soil-forage mixtures and IVTD
might be due to large differences in the process rates of each assay. For example, in vitro
digestion of balsam fir needles in our study resulted in 66 % mass loss after 48 hours; by
contrast, Strukelj et al. (2012) found that balsam fir needles decomposing in the forest floor lost
42 % of its original mass after 1 year. Thus, it is likely that CO2 production from a soil-forage
mixture over a short 48 hour period mainly reflects microbial degradation of labile moieties of
the forage, rather than the degradation of ADF and other recalcitrant compounds. As this assay
is perhaps the easiest among the 4 proposed proxies, it might be worth testing the strength of its
correlation with IVTD when the assay is performed over longer periods. On the other hand, the
higher correlations between CO2 production from soil-forage mixtures and total gas as well as
VFA production, suggests that useful information can be collected from this simple approach
even with a short assay period.
Taken collectively, our data revealed that partial least square models based on NIRS is the most
successful approach for predicting IVTD, total gas and VFA from a wild deer inoculum. The
successful prediction of IVTD with NIRS is not surprising, as NIRS has already been shown to
51
accurately predict in vitro digestibility for domestic ruminants (Decruyenaere et al. 2009,
Andueza et al. 2011). As for total gas and VFA production, we are the first to demonstrate a
potential for NIRS to predict these 2 variables for both domestic and wild ruminants. As these
2 variables were measured for only 12 forage samples, it could be argued that the high
correlation coefficients that were found arose from over-fitted partial least-squares models.
However, the low root mean square error terms associated to leave-one-out cross-validations
(RMSEVAL) provide strong evidence that our models are indeed robust. But before NIRS can be
chosen over other methods, it must be possible for ecologists to collect reliable reference data
sets based on in vitro digestibility measurements using wild ruminant inoculum. This may pose
a logistical challenge, but the returns on that investment are great considering that calibrated
NIRS predictive models can be used repeatedly and at minimal user costs.
Digestion kinetics and VFA production
Comparing digestion kinetics and VFA production of cow and white-tailed deer ruminal liquors
gives us insights as to the differences in nutritional ecophysiology of these 2 ruminants. Deer
inoculum displayed higher early fermentation rates (i.e. smaller KM values) for 3 of the 4 forage
species, in line with the comparative physiology of ruminants of different sizes (Hofmann 1989).
Small browsers such as white-tailed deer have higher food passage rates than large grazers such
as cows, due to differences in rumen size (Van Soest 1994, Shipley 1999). White-tailed deer
would benefit, therefore, from higher fermentation rates to optimize the conversion of labile
forage fractions into energy, despite the relatively high ADF content of their diets (Hofmann
1989, Shipley 1999). The cow inoculum, on the other hand, may be more adapted to maximizing
energy conversion from forages that are rich in digestible fibers (e.g. graminoids), which require
a longer fermentation period (Shipley 1999).
52
Different VFAs are precursors of different energetic pathways in ruminants (Van Soest
1994). Monitoring the relative amounts of different VFAs produced during digestion can inform
us, therefore, on the digestive capacities of the rumen flora that will drive ruminant energy
acquisition and use. For example, acetate is the main precursor of lipogenesis, hence ruminal
liquors that produce a high proportion of acetate may lead to an increase in the amount of milk
fat (e.g. van Knegsel et al. 2007). Accordingly, the higher amount of acetate produced by the
cow inoculum is consistent with the anabolic conversion of forage into body and milk fat. On
the other hand, propionate is a main precursor for glucose synthesis, hence the higher amount
of propionate produced by deer inoculum reflects a more efficient conversion of forage to
readily metabolizable energy (e.g. Russel 1998, Sutton et al. 2003). As for butyrate, this VFA
is a precursor to the synthesis of ketone bodies during fasting. Thus, the higher amount of
butyrate produced by deer inoculum suggests a rumen flora that is more adapted to energydeficient diets.
The largest difference in total VFA production between inocula occurred with Canada
bunchberry, with deer inoculum producing about 40 % more VFAs than cow inoculum. Among
forb species, Canada bunchberry leaves are reported to contain relatively large amounts of
secondary metabolites such as tannins (Van Horne 1988), terpenes (Stermitz and Krull 1998).
Moreover, Hassan (2013) reported total phenolics content for Canada bunchberry that were
about 16 times higher than what was reported for both balsam fir white spruce (36.0 mg/g vs ~
2.2 mg/g; Sauvé and Côté 2007). These compounds are reputed to reduce the activity of ruminal
microbial communities (i.e. Robbins et al 1987b) and protein digestion (i.e. Robbins et al.
1987a), and may thus reduce total VFA production. This effect could, however, be less severe
for deer than for cows, as studies have shown that deer inocula are more efficient to digest
phenolic-rich forages, either through tannin-binding proteins in the saliva (Mole et al. 1990) or
through ruminal microfloral resistance to phenolic toxicity (McSweeney et al. 2001).
53
In retrospect, our study demonstrated the usefulness of NIRS as a proxy for determining
IVTD, gas and total VFA production using a wild ruminant’s inoculum. The ease and minimal
cost with which forage samples can be analyzed by calibrated NIRS models would increase the
sample size that ecologists can strive for, and would thus dramatically improve the accuracy of
their predictions. NIRS predictions of forage quality could then be implemented in low-cost
monitoring programs of remote wild ruminant populations. On the other hand, NIRS technology
may not be accessible to some, just as collecting reliable reference data sets based on in vitro
assays using wild ruminant liquor may be logistically difficult. If these constraints preclude the
use of NIRS, then the use of forage ADF or CO2 production from soil-forage mixtures are both
interesting alternatives, especially as proxies of gas and total VFA production. Finally, our data
warn against the use of cow inoculum as a proxy for studying wild ruminant nutritional ecology.
If wild ruminant inoculum is available, we emphasize the interest of comparing digestion
kinetics and VFA production between different inoculum sources as an innovative and
promising avenue to better understand the adaptive ecophysiology of digestion by different
ruminant herbivores.
Acknowledgements
We thank S.D. Côté, C. Meyer, J. Castonguay, M. Bonin, J. Faure-Lacroix, S. De La Croix de
La Valette, S. Provencher, W.F.J. Parsons, G. Laprise, D. Morin, C.Frenette-Dussault and S. De
Bellefeuille for technical and logistical support. The project was funded by an NSERC
Discovery Grant awarded to R. L. Bradley and an NSERC Collaborative Research and
Development Grant (S. Côté et al.).
54
References
Andueza, D., F. Picard, M. Jestin, J. Andrieu, and R. Baumont. 2011. NIRS prediction of the
feed value of temperate forages: efficacy of four calibration strategies. Animal 5:1002–13.
Bodas, R., S. López, M. Fernández, R. García-González, A. B. Rodríguez, R. J. Wallace, and J.
S. González. 2008. In vitro screening of the potential of numerous plant species as
antimethanogenic feed additives for ruminants. Animal Feed Science and Technology
145:245–258.
Campa III, H., D. Woodyard, and J. Haufler. 1984. Reliability of captive deer and cow in vitro
digestion values in predicting wild deer digestion levels. Journal of Range Management
37:468–470.
Cerrillo, M., and R. Juarez. 2004. In vitro gas production parameters in cacti and tree species
commonly consumed by grazing goats in a semiarid region of North Mexico. Livestock
Research for Rural Development 16:21.
Chesson, A. 1997. Plant degradation by ruminants: parallels with litter decomposition in soils.
Pages 47-66 in G. Cadisch, and K.E. Giller, editors. Driven by Nature: Plant Litter Quality
and Decomposition. CAB International,Wallingford, UK.
Clary, W. P., B. L. Welch, and G. D. Booth. 1988. In vitro digestion experiments: importance
of variation between inocula donors. The Journal of Wildlife Management 52:358–361.
Cornelissen, J. H. C., H. M. Quested, D. Gwynn-Jones, R. S. P. Van Logtestijn, M. A. H. De
Beus, A. Kondratchuk, T. V Callaghan, and R. Aerts. 2004. Leaf digestibility and litter
decomposability are related in a wide range of subarctic plant species and types. Functional
Ecology 18:779–786.
Coté, S. D. 1998. In Vitro Digestibilities of Summer Forages Utilized by the Rivière George
Caribou Herd. Arctic 51:48–54.
Crawford, H. 1982. Seasonal food selection and digestibility by tame white-tailed deer in central
Maine. The Journal of Wildlife Management 46:974–982.
Decruyenaere, V., P. Lecomte, C. Demarquilly, J. Aufrere, P. Dardenne, D. Stilmant, and A.
Buldgen. 2009. Evaluation of green forage intake and digestibility in ruminants using near
infrared reflectance spectroscopy NIRS: Developing a global calibration. Animal Feed
Science and Technology 148:138-156.
Edgington, E., and P. Onghena. 2007. Randomization tests, 4th edn. Chapman and Hall/CRC,
Boca Raton, FL.
55
Environment Canada. 2006. Climate normals and averages, daily data reports of Port-Menier’s
station from 1995 to 2005. Environment Canada, Québec, Canada.
Foley, W. J., A. McIlwee, I. Lawler, L. Aragones, A. P. Woolnough, and N. Berding. 1998.
Ecological applications of near infrared reflectance spectroscopy – a tool for rapid, costeffective prediction of the composition of plant and animal tissues and aspects of animal
performance. Oecologia 116:293–305.
Getachew, G., P. H. Robinson, E. J. DePeters, and S. J. Taylor. 2004. Relationships between
chemical composition, dry matter degradation and in vitro gas production of several ruminant
feeds. Animal Feed Science and Technology 111:57–71.
Goering, H., and P. J. Van Soest. 1970. Forage fiber analysis, USDA Research Service,
Handbook number 379. U.S. Department of Agriculture, Washington, DC.
Grondin, P., C. Ansseau, L. Bélanger, J.-F. Bergeron, Y. Bergeron, A. Bouchard, J. Brisson, L.
De Grandpré, G. Gagnon, C. Lavoie, G. Lessard, S. Payette, P.J.H. Richard, J. P. Saucier, L.
Sirois, and L. Vasseur. 1996. Cadre bioclimatique de référence des régions écologiques du
Québec. Pages 134-279 in Manuel de Foresterie. Les Presses de l’Université Laval. Québec.
Canada. [in French.]
Hanley, T. A., D. E. Spalinger, K. J. Mock, O. L. Weaver, and G. M. Harris. 2012. Forage
resource evaluation system for habitat—deer: an interactive deer habitat model, Technical
Report PNW-GTR-85. Forest Service, U.S. Department of Agriculture, Washington, D.C.
Harvey, L., and D. Fortin. 2013. Spatial heterogeneity in the strength of plant-herbivore
interactions under predation risk: the tale of bison foraging in wolf country. PloS one
8:e73324.
Hassan, M. 2013. In vitro antioxidant and antibacterial activity of twenty-one Northern Ontario
medicinal plants. M. Sc. Thesis, Lakehead University, Ontario, Canada.
Hidding, B., J.-P. Tremblay, and S. D. Côté. 2013. A large herbivore triggers alternative
successional trajectories in the boreal forest. Ecology 94:2852-2860.
Hofmann, R. R. 1989. Evolutionary steps of ecophysiological adaptation and diversification of
ruminants: a comparative view of their digestive system. Oecologia 78:443–457.
Kamalak, A., O. Canbolat, Y. Gurbuz, O. Ozay, and E. Ozkose. 2005. Chemical composition
and its relationship to in vitro gas production of several tannin containing trees and shrub
leaves. Asian-Australasian Journal of Animal Sciences 18:203–208.
56
Lachenbruch, P., and M. Mickey. 1968. Estimation of error rates in discriminant analysis.
Technometrics 10:1–11.
Lefort, S., J.-P. Tremblay, F. Fournier, F. Potvin, and J. Huot. 2007. Importance of balsam fir
as winter forage for white-tailed deer at the northeastern limit of their distribution range.
Ecoscience 14:109–116.
Legendre, P. 2013. lmodel2 : model II regression. Version 1.7-2.
http://cran.r-project.org/web/packages/lmodel2/index.html
Lippke, H. 1980. Forage characteristics related to intake, digestibility and gain by ruminants.
Journal of Animal Science 50:952–961.
Mårell, A., A. Hofgaard, and K. Danell. 2006. Nutrient dynamics of reindeer forage species
along snowmelt gradients at different ecological scales. Basic and Applied Ecology 7:13–30.
Marie-Victorin, F., and F. Rolland-Germain. 1969. Flore de l’Anticosti-Minganie. Presses de
l’Université de Montréal, Montréal, Québec, Canada. [in French.]
McCuistion, K., M. Grigar, D. B. Wester, R. Rhoades, C. Mathis, and L. Tedeshi. 2014. Can
We Predict Forage Nutritive Value With Weather Parameters? Rangelands 36:2-9.
McSweeney, C. S., B. Palmer, D. M. McNeill, and D. O. Krause. 2001. Microbial interactions
with tannins: nutritional consequences for ruminants. Animal Feed Science and Technology
91:83-93.
Mole, S., L. G. Butler, and G. Iason. 1990. Defense against dietary tannin in herbivores: A
survey for proline rich salivary proteins in mammals. Biochemical Systematics and Ecology
18:287–293.
Potvin, F., P. Beaupré, and G. Laprise. 2003. The eradication of balsam fir stands by whitetailed deer on Anticosti Island , Québec : A 150-year process. Ecoscience 10: 487–495.
Reynolds, C., L. Crompton, and J. Mills. 2011. Improving the efficiency of energy utilisation in
cattle. Animal Production Science 51:6–12.
Ritz, C., and J. Strebig. 2014. drc: Analysis of dose response curve data. Version 2.3-96.
http://cran.r-project.org/web/packages/drc/index.html.
Robbins, C., T. Hanley, A. Hagerman, O. Hjelford, D. L. Baker, C. C. Schwartz, and W. W.
Mautz. 1987a. Role of tannins in defending plants against ruminants: reduction in protein
availability. Ecology 68:98–107.
57
Robbins, C. T., S. Mole, A. E. Hagerman, and T. A. Hanley. 1987b. Role of Tannins in
Defending Plants Against Ruminants: Reduction in Dry Matter Digestion? Ecology 68:1606–
1615.
Russell, J. B. 1998. The importance of pH in the regulation of ruminal acetate to propionate
ratio and methane production in vitro. Journal of Dairy Science 81:3222–30.
Russell, J. B., and H. J. Strobel. 1989 Effect of ionophores on ruminal fermentation. Applied
and Environmental Microbiology 55:1-5.
Sauvé, D. G., and S. D. Côté. 2007. Winter forage selection in white-tailed deer at high density:
Balsam fir is the best of a bad choice. Journal of Wildlife Management 71:911-914.
Shipley, L. 1999. Grazers and browsers: how digestive morphology affects diet selection. Pages
20-27 in K.L. Launchbaugh, K.D. Sanders, and L.C. Mosley, editors. Grazing behavior of
livestock and wildlife. University of Idaho, Moscow, ID.
Smith, S., and J. Crouse. 1984. Relative contributions of acetate, lactate and glucose to
lipogenesis in bovine intramuscular and subcutaneous adipose tissue. The Journal of
Nutrition 114:792–800.
Stermitz, F. R., and R. E. Krull. 1998. Iridoid glycosides of Cornus canadensis: a comparison
with some other Cornus species. Biochemical Systematics and Ecology 28:845-849.
Strukelj, M., S. Brais, S. A. Quideau, and S.-W. Oh. 2012. Chemical transformations of
deadwood and foliar litter of mixed boreal species during decomposition. Canadian Journal
of Forest Research 42:772–788.
Sutton, J. D., M. S. Dhanoa, S. V. Morant, J. France, D. J. Napper, and E. Schuller. 2003. Rates
of production of acetate, propionate, and butyrate in the rumen of lactating dairy cows given
normal and low-roughage diets. Journal of Dairy Science 86:3620–33.
Theodorou, M. K., B. A. Williams, M. S. Dhanoa, A. B McAllan, and J. France. 1994. A simple
gas production method using a pressure transducer to determine the fermentation kinetics of
ruminant feeds. Animal Feed Science and Technology 48:185–197.
Tilley, J., and R. Terry. 1963. A two-stage technique for the in vitro digestion of forage crops.
Grass and Forage Science 18:104–111.
Tollefson, T. N., L. A. Shipley, W. L. Myers, and N. Dasgupta. 2011. Forage Quality's Influence
on Mule Deer Fawns. Journal of Wildlife Management 75: 919-928.
Tremblay, J.-P., J. Huot, and F. Potvin. 2006. Divergent nonlinear responses of the boreal forest
field layer along an experimental gradient of deer densities. Oecologia 150:78–88.
58
Van Horne, B., T. Hanley, R. G. Cates, J. D. McKendrick, and J. D. Horner. 1988. Influence of
seral stage and season on leaf chemistry of southeastern Alaska deer forage. Canadian Journal
of Forest Research 18:90–99.
van Knegsel, A. T. M., H. van den Brand, J. Dijkstra, W. M van Straalen, M. J. W. Heetkamp,
S. Tamminga, and B. Kemp. 2007. Dietary Energy Source in Dairy Cows in Early Lactation:
Energy Partitioning and Milk Composition. Journal of Dairy Science 90:1467-1476.
Van Soest, P. J. 1978. Dietary fibers: their definition and nutritional properties. The American
Journal of Clinical Nutrition 31:12–20.
Van Soest, P.J. 1994. Nutritional Ecology of the Ruminant. 2nd edn. Cornell University Press,
Ithaca, NY.
59
CHAPITRE 3
UTILISATION DE LA SPECTROSCOPIE
DANS LE PROCHE INFRAROUGE ET
DES GÉOSTATISTIQUES POUR
GÉNÉRER UNE CARTE DE LA
QUALITÉ ALIMENTAIRE À
L’ÉCHELLE DU PAYSAGE
60
Avant-propos
Les 2 chapitres précédents visaient à améliorer le monitoring de la composition et de la qualité
de la diète du cerf. Dans ce présent chapitre, je cherche à considérer la composante spatiale de
la qualité de la diète, d’une part, en intégrant en un nouvel indice l’aspect multidimentionnel de
la nutrition du cerf. D’autre part, je cherche à faire le lien entre la qualité nutritionnelle des
plantes ingérées par le cerf et le type d’habitat qu’il fréquente en identifiant les patrons spatiaux
de qualité de la diète à l’échelle du paysage. Identifier des zones de haute et de basse qualité
alimentaire permettra de mieux cibler des zones à préserver et de mieux prédire les changements
dans la qualité de la diète pour le cerf selon les trajectoires floristiques anticipées.
Ce chapitre sous forme d’article a été rédigé en collaboration avec mes directeurs Robert L.
Bradley et Jean-Pierre Tremblay, de même qu’avec Steeve D. Côté (U. Laval), et il est publié
dans la revue Ecological Applications (septembre 2015):
Jean, P., Bradley, R.L., Tremblay, J.P. et Côté, S.D. (2015). Combining near infrared spectra of
feces and geostatistics to generate forage nutritional quality maps across landscapes. Ecological
Applications 25:1630-1639.
61
Combining near infrared spectra of feces and geostatistics to
generate forage nutritional quality maps across landscapes
Pierre-Olivier JEAN*1, Robert L. BRADLEY1, Jean-Pierre TREMBLAY2& Steeve D.
CÔTÉ2
1
Département de biologie, Université de Sherbrooke, 2500 boulevard de l'Université,
Sherbrooke, Québec, Canada J1K 2R1.
2
Département de biologie, Université Laval, Québec, QC, G1V0A6, Canada.
Keywords: Forage nutritional quality; Free-ranging ungulates; Landscape ecology; Mapping
tool; Near infrared spectroscopy; Wildlife management.
Abbreviations: DISTEX, mahalanobis distance of a fecal spectrum to the centroid of exclosure
fecal spectra; NDF, neutral detergent fibers; ADF, neutral detergent fibers; ADL, acid detergent
lignin, N, nitrogen; NIR, near infrared; RMSEP, root mean square error of prediction;
Reference: Jean, P., Bradley, R.L., Tremblay, J.P. and Côté, S.D. (2015). Combining near
infrared spectra of feces and geostatistics to generate forage nutritional quality maps across
landscapes. Ecological Applications. In press.
62
Résumé
Une condition essentielle à une gestion éclairée des populations d’ongulés sauvages est de
connaître la distribution spatiale des ressources alimentaires disponibles. On doit également
connaître quelles plantes sont consommées et évaluer leur qualités nutritionnelles, ce qui est
souvent problématique dans le cadre d’études écologiques. Nous proposons une méthode rapide
et économique qui combine la spectroscopie dans le proche infrarouge, les géostatistiques et des
données de base sur l’abondance des plantes disponibles afin de générer une carte de la qualité
nutritionelle de la diète à l’échelle du paysage. Notre approche repose sur la prémisse que la
chimie de fèces échantillonnées à l’intérieur d’habitats à faible densité d’herbivores (i.e. exclos)
reflète la qualité optimale de diète dans notre aire d’étude. On définit donc la qualité alimentaire
en se basant sur la distance de Mahalanobis des spectres dans le proche infrarouge de fèces à
travers le paysage au centroïde des spectres amassés dans les exclos (DISTEX). En utilisant la
statistique spatiale Gi*, il est alors possible de détecter des patrons de points chauds et de points
froids de qualité alimentaire à l’échelle du paysage. Ous avons testé cette approche dans un
paysage hétérogène de l’Île d’Anticosti (Québec, Canada), où le cerf de Virginie (Odocoileus
virginianus) s’est maintenu a des très hautes densités de population pour les 80 dernières années,
entrainant du même coup des changements profonds dans la composition des forêts. Nos
résultats suggèrent que des points chauds de qualité alimentaire surviennent lorsque plus de 39,8
% de sapinière matures une zone avoisinante de 300 ha, et que des points froids de qualité
alimentaire surviennent principalement les zones de transition entre la tourbière et la forêt. La
présence de Cornouiller quatre-temps (Cornus canadensis) était fortement corrélée avec la
présence des points chauds, alors que la présence de mélèze laricin (Larix laricina) était
fortement corrélée avec la présence des points froids. Les valeurs de DISTEX étaient en
moyenne corrélées positivement et significativement avec la concentration en NDF (i.e. neutraldetergent fibers) et ADL (i.e. acid-detergent lignin) des fèces. Bien que notre approche nécessite
encore des études indépendantes avant qu’elle soit totalement validée, sa facilité d’utilisation et
63
son faible coût la rendent fort utile pour contribuer à notre compréhension de l’écologie des
grands herbivore
Abstract
An important asset for the management of wild ungulates is recognizing the spatial distribution
of forage quality across heterogeneous landscapes. To do so typically requires knowledge of
which plant species are eaten, in what abundance they are eaten, and what their nutritional
quality might be. Acquiring such data, however, may be difficult and time consuming. Here, we
are proposing a rapid and cost-effective forage quality monitoring tool that combines near
infrared (NIR) spectra of fecal samples and easily obtained data on plant community
composition. Our approach rests on the premise that NIR spectra of fecal samples collected
within low population density exclosures reflect the optimal forage quality of a given landscape.
Forage quality can thus be based on the Mahalanobis distance of fecal spectral scans across the
landscape relative to fecal spectral scans inside exclosures (referred to as DISTEX). The Gi*
spatial autocorrelation statistic can then be applied among neighboring DISTEX values to detect
and map “hot-spots” and “cold-spots” of nutritional quality over the landscape. We tested our
approach in a heterogeneous boreal landscape on Anticosti Island (Québec, Canada), where
white-tailed deer (Odocoileus virginianus) populations over the landscape have ranged from 20
to 50 individuals km-2 for at least 80 years, resulting in a loss of most palatable and nutritious
plant species. Our results suggest that hot-spots of forage quality occur when 300 ha
neighborhoods comprise > 39.8 % old-growth balsam fir stands, whereas cold-spots occur in
laggs (i.e. transition zones from forest to peatland). In terms of ground level indicator plant
species, the presence of Canada bunchberry (Cornus canadensis) was highly correlated with
hot-spots, whereas tamarack (Larix laricina) was highly correlated with cold-spots. Mean
DISTEX values were positively and significantly correlated with the neutral detergent fiber and
acid detergent lignin contents of feces. While our approach would need more independent field
64
trials before it is fully validated, its low cost and ease of execution should make it a valuable
tool for advancing both the basic and applied ecology of large herbivores.
Introduction
Successful wildlife management is contingent upon the ability to predict spatial patterns of
habitat quality based on biotic and abiotic landscape features. To this end, habitat suitability
indices (Brooks 1997; Roloff and Kernohan 1999) and resource selection functions (Manly et
al. 2002; Hebblewhite and Merrill 2008) have been used to identify areas of conservation
interest where large herbivore populations feed, aggregate or reproduce. These tools use relevant
habitat data to explain changes in population densities (e.g. McLoughlin et al. 2010) or the
fitness of individuals across the landscape (e.g. Fortin et al. 2008). Some of these studies
suggested that the abundance of quality forage is an important factor guiding habitat selection
(Godvik et al. 2009; Samelius et al. 2013; Maklakov et al. 2008). Forage quality may vary over
time, however, due to browsing, disturbance or natural succession (Coley 1987; Hawkes and
Sullivan 2001; Mysterud et al. 2001; Persson et al. 2007). There is, therefore, a need to develop
rapid and cost-effective monitoring tools to describe changing patterns in forage nutritional
quality across landscapes (Searle et al. 2007; DeGabriel et al. 2014).
The nutritional quality of forage is intrinsically linked to its chemical attributes in relation to
nutritional needs of herbivores and the physiology of their digestive systems. For example, past
studies on captive animals suggested a negative effect of plant secondary metabolites such as
tannins and other phenolic compounds, and a positive effect of essential macronutrients such as
nitrogen (N) and digestible energy, on diet selection, digestibility and reproduction (e.g. Dearing
et al. 2005; McArt et al. 2009; Estell 2010). In the case of wild herbivores that feed across
heterogenous landscapes, the types and concentrations of nutrients and secondary metabolites
65
may vary within and between multiple plant species (Moore et al. 2010; DeGabriel et al. 2014).
Hence, spatial patterns of forage quality for wild herbivores may be hard to deduce from a few
chemical traits measured on a few conspicuous forage types.
One approach, which has long been used to study forage quality and diet choice by wild
herbivores, has been to collect and analyze fecal samples (e.g. Johnson 1994; Wam and Wjeljord
2010). Collecting fecal samples does not bias herbivore behavior as field observations of feeding
animals may do. Furthermore, fecal abundances can inform us on population densities, while
fecal chemical properties may reveal information on diet selection and diet quality. Codron et
al. (2007) have shown a significant negative relationship between fecal lignin and the amount
of highly digestible grasses in diets of many species of wild ungulates. Likewise, Jean et al.
(2014) showed a positive correlation between fecal lignin and the amount of lignaceous forage
in the diet of white-tailed deer. Similarly, several studies have shown total fecal fiber content to
be negatively related to forage energy yields for wild herbivores (Hodgman et al. 1996; Brown
et al. 1995). On the other hand, some fecal chemical properties are more difficult to interpret
because of the confounding factors driving forage quality and digestive metabolism. For
example, fecal N may be a good indicator of dietary N (Leslie and Starkey 1985), but not when
a diet is rich in tannins that precipitate protein in the gut (i.e. leading to an overestimation of
dietary N; Hobbs 1987). An alternative option to fecal chemical analyses has been to identify
consumed plant species using microhistological analyses (Dearden et al.1975), plant wax alkane
markers (Bugalho et al. 2004) or plant DNA bar-coding (Valentini et al. 2009a). These methods,
however, are imprecise at estimating the proportion of each plant species in the original diet
(Bugalho et al. 2002; Valentini et al. 2009b) and the information they provide is difficult to
translate into indices of nutritional quality (DeGabriel et al. 2014).
In summary, patterns in forage nutritional quality across landscapes are difficult to describe
based on current methodological approaches. Even remote sensing methods developed to
estimate spatial patterns of foliar protein and polyphenols (e.g. Skidmore et al. 2010) are
66
relatively impractical in landscapes where forage is mainly found underneath tree canopies. For
this reason, we are proposing a new approach that combines near infrared (NIR) spectra of fecal
samples, geostatistical tools and easily obtained data on plant community composition to
generate forage nutritional quality maps across heterogeneous landscapes. Ours is a holistic
approach that explores the entire chemical signature of fecal samples using near infrared
spectroscopy (NIRS), rather than focusing on specific compounds or plant fragments.
Materials and methods
Geography and history of the study area
Anticosti Island (7,943 km2) is located in the Gulf of St-Lawrence – Canada (49°28'N,
63°00'W). It is part of the eastern balsam fir–white birch bioclimatic zone (Grondin et al. 2007).
The sub-boreal maritime climate results in cool summers (mean July temperature of 16 °C) and
cold winters (mean January temperature of -11 °C), while the average annual precipitation is
917 mm (Environment Canada 2006) of which about one-third falls as snow.
About 200 white-tailed deer were introduced on the island in 1896 and the population increased
rapidly in the absence of natural predators. Accordingly, the average white-tailed deer density
over different sectors of the island has ranged from 20 to >50 individuals km-2 for the past 80
years (Gingras et al. 1993; Potvin 2000). Such elevated densities have stimulated economic
activity on the island via tourism and game hunting (Potvin 2000). However, elevated whitetailed deer densities have also resulted in the over-browsing and loss of palatable summer plant
species such as mountain maple (Acer spicatum Lam.), beaked hazel (Corylus cornuta Marsh.)
and Canadian yew (Taxus canadensis Marsh.) (Potvin et al. 2003), and in the spread of browse67
tolerant grasses such as Canada bluejoint (Calamagrostis canadensis (Michx.)) (Dufresne et al.
2009). Balsam fir (Abies balsamea (L.) Mill.) currently is the preferred winter forage (Lefort et
al. 2007), but the recruitment of balsam fir seedlings in the understory is substantially
diminishing due to overbrowsing. On the other hand, the recruitment of less palatable and less
nutritious white spruce (Picea glauca (Moench) Voss) seedlings has been increasing (Hidding
et al. 2013). Given these ongoing changes in habitat quality, the sustainability of elevated whitetailed deer populations on Anticosti Island may be compromised in coming years (Potvin et al.
2004).
Figure 1. (a) Map of the study area showing the various land cover classes obtained from
the mapping geodatabase of the Quebec Ministry of Natural Resources and
Wildlife (MRNFQ 2010) as well as each sampling location. (b) Map of our study
area showing the occurrence of cold-spots (in blue) and hot-spots (in red) of
forage nutritional quality, as determined by Gi* analysis.
68
Collection and spectral analysis of fecal samples
We established an 84 km2 sampling grid in an area of Anticosti Island with a diverse landscape
dominated by old-growth balsam fir stands in the south, peatlands in the center, and white and
black spruce (Picea mariana (Mill.) BSP) stands in the north (Fig. 1a). This sector was also in
proximity of 2 deer exclosures (680 and 560 ha) that had been erected in 2001 and 2003. In each
exclosure, 70 % of the area had been clearcut and 30 % kept as old-growth forest the year prior
to erecting a 12-foot-high fence and culling the deer population inside the exclosures. At the
time of sampling, white-tailed deer populations within the exclosures were estimated at 2-8
individuals km-2 (Laprise 2014). The herbaceous vegetation and deciduous shrubs and saplings
were strikingly more abundant within than outside the exclosures (Bachand et al. 2014; Fig. 2)
and comprised an abundance of plant species with high nutritional quality such as fireweed
(Chamerion angustifolium (L.) Holub.) (Dostaler et al. 2011). Consequently, the deer within
these exclosures were in better physical condition than those living outside the exclosures
(Giroux et al. 2014). Sampling locations (N = 77) within the sampling grid were spaced 800 m
apart, which approximately corresponds to the average diameter of the summer home range of
females (~ 42 ha) on Anticosti Island (Massé and Côté 2009). Since females generally have
smaller home ranges than males (Nelson and Mech 1984), we considered an 800 m spacing as
the largest allowable sampling step to minimize our sampling effort without compromising the
reliability of the data (Fortin and Dale 2005).
69
Figure 2. Photo showing differences in vegetation within (right) and outside (left) an
exclosure. The photo was taken 10 years after erecting the 12 foot fence.
Photo credit: S. de Bellefeuille.
In the first 2 weeks of August 2011, we collected 2–4 fresh fecal samples within a 100 m
diameter on each of the sampling locations (N = 236 outside exclosures; N = 49 within
exclosures). We then identified the dominant ground layer vegetation within a 5 m radius around
each fecal sample. In all, 10 ground layer plant species or genera were identified outside the
exclosures. These included thistle (Cirsium spp.), fireweed, sedges (Carex spp.), tufted vetch
(Vicia cracca L.), Canada bunchberry (Cornus canadensis L.), clovers (Trifolium spp.), blue
flag (Iris versicolor L.), graminoids, ferns and mosses. We also identified seedlings of tamarack
(Larix laricina (Du Roi) K. Koch), black spruce, white spruce and ericaceous shrubs. Fecal
samples were immediately frozen at -20 °C, transported to the Soil Ecology Laboratory
(Université de Sherbrooke) where they were freeze-dried, ground with a mortar and pestle, and
passed through a 2 mm mesh sieve. Each fecal sample was scanned with an Antaris II Fourier
Transform-NIR spectroscope (Thermo-Fisher Scientific, Waltham, MA), which measures the
absorbance of 1,557 wavelengths at 0.4 to 2.5 nm intervals in the near infrared spectral region
(1,000–2,500 nm). The spectroscope was supported by Omnic acquisition and processing
70
software (Thermo-Fisher Scientific, Waltham, MA). Each sample was scanned 32 times from
different angles, using the Antaris II Sample Cup Spinner, and the 32 spectra were subsequently
averaged to cope with potential sample heterogeneity.
Sorting fecal samples by discriminant analysis
We first performed a principal component-based discriminant analysis, using the TQ analyst
software (Thermo-Fisher Scientific, Waltham, MA), to discriminate between the NIR spectra of
feces from white-tailed deer foraging inside vs. outside the exclosures. The raw spectral data
were transformed into first order derivative functions to remove baseline and spectral noise
(Rinnan et al. 2009). The software algorithm subsequently converged on 2 regions of the spectra
(1,453–1,726 nm and 1,913–2,288 nm) that were used in our discriminant analysis. The first 15
principal components, which explained 98 % of the total spectral variation, were then used to
calculate the Mahalanobis distance of each fecal sample to the centroid of fecal samples
collected within vs. outside the exclosures. Mahalanobis distances controlled for multiple
correlations within the spectral datasets and reduced dispersion-related classification errors that
are frequently encountered with multivariate classification techniques (De Maesschalk et al.
2000). Our discriminant model thus classified each fecal sample according to its shortest
Mahalanobis distance to one or the other centroid. To verify the robustness of our discriminant
model, we used 143 samples as a calibration data set and the other 142 samples as a validation
data set. All validation samples were successfully classified into the correct class (i.e., inside vs.
outside exclosures).
71
Mapping habitat nutritional quality
Our approach to describe the nutritional quality of the landscape rests on the premise that NIR
spectra of fecal samples collected within exclosures reflect the optimal forage quality on
Anticosti Island. Thus, nutritional quality of fecal samples collected outside the exclosures was
based on the Mahalanobis distance of their NIR spectral scans to the centroid of spectral scans
of samples collected within exclosures (i.e., computed using the output of the discriminant
model previously described). For the sake of brevity, this distance is hereafter referred to as
DISTEX.
We tested for local spatial autocorrelation among neighboring DISTEX values using the Gi*
statistic (Getis and Ord 1992) as implemented in ArcGIS software version 10.1 (ESRI 2013).
The Gi* statistic, expressed as a Z-score, can detect “hot-spots” (i.e.,high nutritional quality)
and “cold spots” (i.e., low nutritional quality) over the landscape by measuring the degree of
clustering of neighboring samples based on their similarity to the mean of the global sample set
(Nelson and Boots 2008). Thus, Gi* values were calculated using a ~ 300 ha “buffer” (i.e.,
neighborhood with a circular radius of 977 m) around each sampling point. This radius was
selected because we considered it unlikely that a deer would process a food passage cycle in an
area larger than 300 ha, based on reported home range sizes (Nelson and Mech 1984; Massé and
Côté 2009) and gut passage rates (Mautz and Petrides 1971) for white-tailed deer. Gi* values
deviating from 0 by + 1.65 and + 1.96 respectively, corresponded to 90 and 95 % confidence
levels. To visualize the data, the calculated Gi* values were mapped using a kriging procedure
implemented in ArcGIS software version 10.1 (ESRI 2013).
72
Fecal chemical properties
We randomly selected 92 of the 285 fecal samples to characterize chemical properties related
to diet quality. We estimated the neutral detergent fiber (NDF) and acid detergent lignin (ADL)
concentrations by sequential extractions (Goering and Van Soest 1970) using an Ankom
FiberAnalyzer 200 (Ankom Technology, Macedon, NY). Total-N was measured by high
temperature combustion followed by thermo-conductometric detection, using a Vario-Macro
CN Analyzer (Elementar Analysensysteme, Hanau, Germany). From these data, we used the
TQAnalyst software to perform partial least-squares regressions for estimating NDF, ADL and
total-N from the NIR spectra of these 92 fecal samples. The robustness of these models was
evaluated by using 68 samples for calibration and 24 samples for validation. Calibrations were
robust for NDF (R2 = 0.94, root mean square error of prediction (RMSEP) = 3.2), for ADL (R2
= 0.91, RMSEP = 1.7) and for total-N (R2 = 0.91, RMSEP = 0.18). According to Williams
(2001), this goodness-of-fit was suitable for predicting chemical properties of the remaining 193
fecal samples.
Inferring habitat attributes responsible for nutritional quality
To infer habitat attributes responsible for nutritional quality, we correlated Gi* to the percent
cover of clearcuts, peatlands, balsam fir stands, white spruce stands and black spruce stands
within the 300 ha buffer around each sampling location. These land cover classes were obtained
using the mapping geodatabase of the Quebec Ministry of Natural Resources and Wildlife
(MRNFQ 2010). We used univariate regression tree analysis (R party package – ctree function;
Hothorn et al. 2006) to develop a conditional inference tree model with land cover classes as
explanatory variables and Gi* as the response variable. The most parsimonious conditional
73
inference tree was set to identify habitat attributes responsible for hot-spots (Gi* < -1.65) and
cold-spots (Gi* > 1.65).
As an extension of the mapping exercise, we attempted to identify indicator plant species and
indicator fecal chemical properties of hot- and cold-spots. To do this, we used two-sided Student
t-tests to compare the distribution of the Gi* statistic and the distribution of each fecal chemical
property in the presence or absence of individual plant species within the 300 ha buffers. We
also applied type II regression models to test the relationships between DISTEX and fecal
chemical properties, as the latter were random explanatory variables (Legendre and Legendre
2012). The slope of the relationship was estimated using the ranged major axis (RMA) method.
The regression coefficients for these models were determined by permutation.
74
Results
Figure 3. Results of a regression tree analysis using the Gi* statistic as the response
variable, and the percentage of cover of different habitat types in a 300 ha
surrounding each collected fecal sample as the explanatory variables.
Significant hot-spots of nutritional quality were found in the southeast sector of the study area,
whereas cold-spots were found in the northern sector (Fig. 1b). Conditional inference tree
analysis showed that hot-spots occurred where old-growth balsam fir stands comprised more
75
than 39.8 % of the land cover (Fig. 3). On the other hand, cold-spots occurred in areas with more
than 5.4 % black spruce stands and less than 11.8 % old-growth balsam fir stands. It should be
noted, however, that black spruce cover was no more than 11 % in any of the 300 ha
neighborhoods (see Discussion below). Other types of land cover classes did not significantly
explain the occurrence of hot- or cold-spots.
Figure 4. Boxplots showing the distribution of Gi* (Fig. 4a,b) or fecal acid detergent lignin
(ADL) (Fig.4c,d) when Canada bunchberry or tamarack were available within
a 300 ha neighborhood surrounding each fecal sample. Gi* values deviating
from 0 by  1.65 respectively correspond to cold-spots and hot-spots.
76
Forage quality was significantly higher (i.e., significantly lower Gi* values: t = -12.5, df = 234,
p < 0.01) in the presence of Canada bunchberry (Fig. 4a). In contrast, forage quality was
significantly lower (i.e., significantly higher Gi* values: t = 8.8, df = 234, p < 0.01) in the
presence of tamarack (Fig. 4b). Fecal ADL was significantly lower (t = -12.5, df = 234, p <
0.01) in the presence of Canada bunchberry (Fig. 4c), and significantly higher (t = 8.8, df = 234,
p < 0.01) in the presence of tamarack (Fig. 4d).
The mean DISTEX value of all 300 ha neighborhoods correlated significantly and positively
with fecal ADL (R2 = 0.25, y = 0.028x+0.94, p = 0.01) (Fig. 5) and with fecal NDF (R2 = 0.15,
y=0.017x+0.69, P = 0.01) (not shown), but not with fecal N (R2 = 0.01, p > 0.05).
Figure 5. Relationship between fecal acid detergent lignin (ADL) and the mean DISTEX
values of 300 ha buffers surrounding each collected fecal sample.
77
Discussion
In a recent review of the methods and concepts required for studying the nutritional ecology of
wild herbivores, DeGabriel et al. (2014) pointed out the importance of knowing which plant
species are eaten, in what abundance they are eaten, and what their nutritional quality might be.
The same authors go on to explain why these objectives are difficult to achieve using
conventional methods. The approach that we are proposing departs from convention in that NIR
spectra of fecal samples allow us to circumvent the collection of data on feeding habits and plant
chemistry to describe the nutritional quality of landscapes. The robustness of NIR spectral
analysis lies in its ability to detect similar multidimensional signals in fecal samples derived
from different forages. In other words, small DISTEX values (i.e., statistically significant hotspots) are not necessarily indicative of diets comprising the same plant species as those
consumed by white-tailed deer living within exclosures. Rather, the passage of food through the
gut acts as a filter defining the complex chemical attributes of non-assimilated material,
endogenous losses and microbial tissues found in feces, from which NIR spectral analysis is
able to detect levels of similarity between samples. One may argue that our designated coldspots simply reflect large differences in forage chemistry from the exclosures, regardless of
nutritional quality. There is, therefore, a need to validate our underlying assumption that
DISTEX is indeed correlated with forage quality. To do so would require that we relate DISTEX
to the physical condition of each individual that produced each fecal sample, which is clearly
not realistic. However, the fact that DISTEX was related positively to NDF, and even more so
to ADL, is strong evidence of its ability to locate hot- and cold-spots of nutritional quality across
the sampled landscape.
As revealed by the conditional inference tree, statistically significant hot-spots of nutritional
quality over the sampled landscape were associated with balsam fir stands located in the
southern sector of the study area. One may argue that the vegetation found in these hot-spots
was not spatially independent from the vegetation found in the nearby exclosure. To
78
demonstrate that this was not the case, we performed a similar analysis using feces from only
one exclosure at a time as the reference sample set from which to compute DISTEX (see
Appendix A). In both cases, we found a similar spatial pattern of hot- and cold-spots, in spite of
the fact that the second exclosure lies 10 km west of the study area and differs in terms of its
plant community composition compared to the first exclosure (MRNFQ 2010). Furthermore,
the vegetation within any exclosure on Anticosti Island is unequivocally different from the
vegetation outside exclosures, even at close proximity (Fig. 2). Thus, our results call for an
ecological explanation as to why areas with a preponderance of mature balsam fir stands
represent a higher nutritional value over the sampled landscape. For one, these stands are at least
90 years of age (MRNFQ 2010), corresponding to an old-growth senescent stage in balsam fir
stand development (Burns et al. 1990). Balsam fir stands of this age-class are particularly prone
to windthrow (Ruel 2000), which creates patches of ground level vegetation not seen in the
younger ( 40 years) stands of our study area. The particularly strong correlation of hotspots
with Canada bunchberry, one of the few herbaceous plants still available to deer on Anticosti
Island, corroborates this point. This perennial forb can tolerate a wide range of light regimes,
and its rhizomatous propagation is particularly responsive to canopy openings (Hall and Sibley
1976). Previous studies have also shown that the nutritional quality of Canada bunchberry
increases with the age of the forest. For example, Van Horne et al. (1988) assessed leaf
chemistry of this species as a function of the seral stage of hemlock–spruce forests in Alaska.
They found significantly higher concentrations of N and K, significantly lower concentrations
of polyphenolics, and significantly lower astringency of Canada bunchberry leaves collected in
80–450 year old stands than in regenerating stands of 5–11 years of age. They attributed this
observation to lower light availability, which could decrease plant investments in herbivore
defense (Bryant et al. 1983).
Our conditional inference tree also revealed statistically significant cold-spots of nutritional
quality in areas associated with black spruce stands located in the northern sector of the study
area. The designation of “black spruce stand” is in fact an aberration arising from the legend
79
presented in the MRNFQ’s forest cover map, as these areas had less than 11 % black spruce
cover. These areas would better be classified as laggs, that is transition zones from forest to
peatland. Laggs on Anticosti Island consist of open woodlands dominated by graminoids,
sedges, ericaceous shrubs and stunted coniferous trees such as black spruce and tamarack. The
significant correlation between cold-spots and tamarack corroborates this point, as this species
is almost exclusively found in laggs on Anticosti Island (Grondin et al. 2007). Plant species
occurring in these laggs are richer in lignin and fibers compared to forbs such as Canada
bunchberry (Oldemeyer et al. 1977), or richer in tannins and other phenolic compounds
(Joanisse et al. 2009). Massé and Côté (2009) reported that laggs on Anticosti Island provide
more forage biomass and are more frequently browsed by white-tailed deer during the snowfree season compared to other habitats. Our results thus suggest that high population densities
drive wild ungulates to feed where forage is abundant (Coulombe et al. 2011) rather than where
forage is of the highest quality.
The approach that we are proposing for mapping the nutritional quality of landscapes may
significantly improve landscape and wildlife management, if certain challenges can be met. The
first challenge, and the most important one, is to provide further validation of hotspots and
coldspots as comprising areas of contrasting forage quality. The best direct evidence would be
body size and fitness of wild ungulates feeding exclusively in different habitat types. Obviously,
this is difficult to achieve with free-ranging, furtive animals that are in constant movement over
the landscape. On the other hand, our approach is original and alleviates the need to perform
extensive vegetation surveys and chemical analyses. It is a rapid and cost-effective approach
which is relevant and useful to wildlife management where the landscape is rapidly changing
and needs to be continuously monitored. Coupled with other data, our approach could provide
valuable information on the ecology of large herbivores. For example, combining nutritional
quality maps with spatial patterns of fecal abundance would allow us to determine the
relationship between habitat nutritional quality and feeding patterns. Alternatively, fecal
samples could be analyzed for steroids to determine gender (Barja et al. 2008), which would
then allow us to explore how males and females react to changes in available forage. Correlating
80
a wider hormonal profile of feces with DISTEX values would also allow us to explore
relationships between forage nutritional quality and physiological stress. In fact, NIRS
predictive models could themselves be calibrated to assess gender (Tolleson et al. 2005) or
hormonal content (Santos et al. 2014) without the need for lengthy laboratory analyses. In
summary, the approach that we are proposing is innovative and relevant to both the basic and
applied ecology of large herbivores.
Acknowledgements
We thank M. Gauthier, W.F.J. Parsons, G. Laprise, D. Morin, R. Fournier, P. Legendre, R. Van
Hulst, C.Frenette-Dussault and S. De Bellefeuille for technical and logistical support. The
project was funded by an NSERC Discovery Grant awarded to R. L. Bradley and an NSERC
Collaborative Research and Development Grant (S. Côté et al.).
References
Bachand, M., S. Pellerin, S. D. Côté, M. Moretti, M. De Cáceres, P.-M. Brousseau, C. Cloutier,
C. Hébert, É. Cardinal, J.-L. Martin, and M. Poulin. 2014. Species indicators of ecosystem
recovery after reducing large herbivore density: Comparing taxa and testing species
combinations. Ecological Indicators 38:12–19.
Barja, I., G. Silvan, S. Rosellini, A. Pineiro, M. J. Illera, and J.C. Illera. 2008. Quantification of
sexual steroid hormones in faeces of Iberian wolf (Canis lupus signatus): A non-invasive sex
typing method. Reproduction in Domestic Animals 43:701-707.
Brooks, R. P. 1997. Improving habitat suitability index models. Wildlife Society Bulletin
25:163-167.
Brown, R. D., E. C. Hellgren, M. Abbott, D. C. Ruthven III, and R. L. Bingham. 1995. Effects
of dietary energy and protein restriction on nutritional indices of female white-tailed deer. The
Journal of Wildlife Management 59:595–609.
81
Bryant, J. P., F. S. Chapin III, and D. R. Klein. 1983. Carbon/nutrient balance of boreal plants
in relation to vertebrate herbivory. Oikos 40:357–368.
Bugalho M. N., R.W. Mayes, and J. A. Milne. 2002. The effects of feeding selectivity on the
estimation of diet composition using the n-alkane technique. Grass and Forage Science 57:224231.
Bugalho, M. N., H. Dove, W. Kelman, J. T. Wood, and R. W. Mayes. 2004. Plant wax alkanes
and alcohols as herbivore diet composition markers. Rangeland Ecology and Management
57:259-268.
Burns, R. M., and B. H. Honkala, editors. 1990. Silvics of North America: 1. Conifers; 2.
Hardwoods. Agriculture Handbook 654. U.S. Department of Agriculture, Forest Service,
Washington, DC, USA.
Codron, D., J. A. Lee-Thorp, M. Sponheimer, J. Codron, D. De Ruiter, and J. S. Brink. 2007.
Significance of diet type and diet quality for ecological diversity of African ungulates. The
Journal of Animal Ecology 76:526–37.
Coley, P. D. 1987. Interspecific variation in plant anti-herbivore properties: the role of habitat
quality and rate of disturbance. New Phytologist 106:251–263.
Coulombe, M.-L., J. Huot, A. Massé, and S. D. Côté. 2011. Influence of forage biomass and
cover on deer space use at a fine scale: A controlled-density experiment. Écoscience 18:262272.
Dearden, B. L., R. E. Pegau, and R. M. Hansen. 1975. Precision of microhistological estimates
of ruminant food habits. The Journal of Wildlife Management 39:402–407.
Dearing, M. D., W. J. Foley, and S. McLean. 2005. The influence of plant secondary metabolites
on the nutritional ecology of herbivorous terrestrial vertebrates. Annual Review of Ecology,
Evolution, and Systematics 36:169–189.
DeGabriel, J. L., B. D. Moore, A. M. Felton, J. U. Ganzhorn, C. Stolter, I. R.Wallis, C. N.
Johnson, and W. J. Foley. 2014. Translating nutritional ecology from the laboratory to the field:
milestones in linking plant chemistry to population regulation in mammalian browsers. Oikos
123:298-308.
De Maesschalck, R., D. Jouan-Rimbaud, and D. L. Massart. 2000. The Mahalanobis distance.
Chemometrics and Intelligent Laboratory Systems 50:1–18.
82
Dostaler, S., J.-P. Ouellet, J.-F. Therrien, and S. D. Côté. 2011. Are feeding preferences of
white-tailed deer related to plant constituents? The Journal of Wildlife Management 75:913918.
Dufresne, M., R. L. Bradley, J.-P. Tremblay, M. Poulin, and S. Pellerin. 2009. Clearcutting and
deer browsing intensity interact in controlling nitrification rates in forest floor. Ecoscience
16:361–368.
ESRI 2013. ArcGIS Desktop: Release 10.1. Environmental Systems Research Institute.
Redlands, CA, USA.
Estell, R. E. 2010. Coping with shrub secondary metabolites by ruminants. Small Ruminant
Research 94:1–9.
Environment Canada. 2006. Climate normals and averages, daily data reports of Port-Menier’s
station from 1995 to 2005. Environment Canada, Québec, Canada.
Fortin, D., D. Morris, and P. McLoughlin. 2008. Habitat selection and the evolution of
specialists in heterogeneous environments. Israel Journal of Ecology and Evolution 54:311–
328.
Fortin, M. J., and M. R. T. Dale. 2005. Spatial analysis: a guide for ecologists. Cambridge
University Press. Cambridge, UK.
Getis, A. and J. K. Ord. 1992. The analysis of spatial association by use of distance statistics.
Geographical Analysis 24:189-206.
Gingras, A., F. Potvin, and B. Rochette. 1993. Inventaire aérien du cerf dans trois zones
forestières de l'île d'Anticosti, en relation avec une épidémie appréhendée d'arpenteuse de la
pruche. Ministère du Loisir, de la Chasse et de la Pêche, Gouvernement du Québec, Report
2132. 31 p. (in French)
Giroux, M.-A., J.-P. Tremblay, M.-A. Simard, N. G. Yoccoz, and S. D. Côté. 2014. Foragemediated density and climate effects on body mass in a temperate herbivore: a mechanistic
approach. Ecology 95:1332–1340.
Godvik, I. M. R., L. E. Loe, J. O. Vik, V. Veiberg, R. Langvatn, and A. Mysterud. 2009.
Temporal scales, trade-offs, and functional responses in red deer habitat selection. Ecology
90:699–710.
Goering, H. K. and P. J. Van Soest. 1970. Forage fiber analysis apparatus (reagents, procedures,
and some applications). Agriculture handbook No.379. US Department of Agriculture.
Agricultural Research Service, Washington, DC, USA.
83
Grondin, P., J.-P. Berger, Y. Landry, and P. Leboeuf. 2007. Guide de reconnaissance des types
écologiques des régions écologiques 5j – Île d’Anticosti et îles de Mingan de même que 5k–
Îles de la Madeleine, Québec, ministère des Ressources naturelles et de la Faune, Forêt Québec,
Direction des inventaires forestiers, Division de l’analyse et de la diffusion des informations
forestières et écologiques. (in French)
Hall, I. V. and J. D. Sibley. 1976. The biology of Canadian weeds. 20. Cornus canadensis L.
Canadian Journal of Plant Science 56: 885-892.
Hodgman, T. P., and B. B. Davitt. 1996. Monitoring mule deer diet quality and intake with fecal
indices. Journal of Range Management 49:215–222.
Hothorn T, K. Hornik, and A. Zeileis. 2006. Unbiased recursive partitioning: a conditional
inference framework. Journal of Computational and Graphical Statistics 15:651-674.
Hawkes, C. and J. Sullivan. 2001. The impact of herbivory on plants in different resource
conditions: a meta-analysis. Ecology 82:2045-2058.
Hebblewhite, M. and E. Merrill. 2008. Modelling wildlife-human relationships for social
species with mixed-effects resource selection models. Journal of Applied Ecology 45:834–844.
Hidding, A., J.-P. Tremblay, and S. D.Côté. 2013. A large herbivore triggers alternative
successional trajectories in the boreal forest. Ecology 94:2852–2860.
Hobbs, N. T. 1987. Fecal indices to dietary quality: a critique. The Journal of Wildlife
Management 51:317–320.
Jean, P.-O., R. L. Bradley, M.-A. Giroux, J.-P. Tremblay, and S. D. Côté. 2014. Near infrared
spectroscopy and fecal chemistry as predictors of the diet composition of white-tailed deer.
Rangeland Ecology and Management 67:154–159.
Joanisse, G. D., R. L. Bradley, C. M. Preston, and G. D. Bending. 2009. Sequestration of soil
nitrogen as tannin–protein complexes may improve the competitive ability of sheep laurel
(Kalmia angustifolia) relative to black spruce (Picea mariana). The New Phytologist 181:187198.
Johnson, C. N. 1994. Nutritional ecology of a mycophagous marsupial in relation to production
of hypogeous fungi. Ecology 75:2015-2021.
Laprise, G. 2014. Base de données du succès de chasse au cerf de Virginie dans les exclos du
Lac Simonne et du Lac Claude. Ministère de la Forêt, de la Faune et des Parcs. Gouvernement
du Québec. (In French)
84
Leslie Jr., D. M. and E. E. Starkey. 1985. Fecal indices to dietary quality of cervids in oldgrowth forests. The Journal of Wildlife Management 49:142-146.
Lefort, S., J.-P. Tremblay, F. Fournier, F. Potvin, and J. Huot. 2007. Importance of balsam fir
as winter forage for white-tailed deer at the northeastern limit of their distribution range.
Ecoscience 14:109–116.
Legendre P. and L. Legendre. 2012. Numerical ecology. 3rd edition. Elsevier, Amsterdam, The
Netherlands. 1006 p.
Maklakov, A. A., S. J. Simpson, F. Zajitschek, M. D. Hall, J. Dessmann, F. Clissold, D.
Raubenheimer, R. Bonduriansky, and R. C. Brooks. 2008. Sex-specific fitness effects of nutrient
intake on reproduction and lifespan. Current Biology 18:1062-1066.
Manly, B. F. J., L. L. McDonald, D. L. Thomas, T. L. McDonald, and W. P. Erickson. 2002.
Resource selection by animals: statistical analysis and design for field studies. Second Edition.
Kluwer, Boston, Massachusetts, USA.
Massé, A. and S. D. Coté. 2009. Habitat selection of a large herbivore at high density and
without predation: Trade-off between forage and cover? Journal of Mammalogy 90:961–970.
Mautz, W. W. and G. A. Petrides. 1971. Food passage rate in the white-tailed deer. The Journal
of Wildlife Management 35:723–731.
McArt, S., D. Spalinger, W. B. Collins, E. R. Shoen, T. Stevenson, and M. Bucho. 2009.
Summer dietary nitrogen availability as a potential bottom-up constraint on moose in southcentral Alaska. Ecology 90:1400–1411.
McLoughlin, P. D., D. W. Morris, D. Fortin, E. Vander Wal, and A. L. Contasti. 2010.
Considering ecological dynamics in resource selection functions. The Journal of Animal
Ecology 79:4–12.
Moore, B. D., I. R. Lawler, I. R. Wallis, C. M. Beale, and W. J. Foley. 2010. Palatability
mapping: a koala’s eye view of spatial variation in habitat quality. Ecology 91:3165–3176.
MRNFQ. 2010. Base de données du troisième inventaire écoforestier. Ministère des Ressources
naturelles et de la Faune. Direction des inventaires forestiers. Gouvernement du Québec. (in
French)
Mysterud, A., R. Langvatn, N. Yoccoz, and N. C. Stenseth. 2001. Plant phenology, migration
and geographical variation in body weight of a large herbivore: the effect of a variable
topography. Journal of Animal Ecology 70:915–923.
85
Nelson, M. E. and L. D. Mech. 1984. Home-range formation and dispersal of deer in
northeastern Minnesota. Journal of Mammalogy 65:567–575.
Nelson, T. A. and B. Boots. 2008. Detecting spatial hot spots in landscape ecology. Ecography
31:556–566.
Oldemeyer, J., A. Franzmann, A. L. Brundage, P. D. Arneson, and A. Flynn. 1977. Browse
quality and the Kenai moose population. The Journal of Wildlife Management 41:533-542.
Persson, I.-L., R. Bergström, and K. Danell. 2007. Browse biomass production and regrowth
capacity after biomass loss in deciduous and coniferous trees: responses to moose browsing
along a productivity gradient. Oikos 116:1639–1650.
Potvin, F. 2000. Le cerf et les sapinières de l’île d'Anticosti, Direction de la faune et des habitats,
Société de la faune et des parcs du Québec, Québec, Canada. (in French)
Potvin, F., P. Beaupré, and C. Dussault. 2004. La population de cerfs de l'île d'Anticosti :
élaboration de modèles densité-habitat et prévisions pour l'horizon 2000- 2100. Rapport de
recherche, Chaire de recherche industrielle CRSNG-Produits forestiers Anticosti, Université
Laval, Québec, Canada. (in French)
Potvin, F., P. Beaupré, and G. Laprise. 2003. The eradication of balsam fir stands by whitetailed deer on Anticosti Island, Québec: A 150-year process. Écoscience 10:487–495.
Rinnan, Å., F. Van Den Berg, and S. B. Engelsen. 2009. Review of the most common preprocessing techniques for near-infrared spectra. Trends in Analytical Chemistry 28:1201–1222.
Roloff, G. J. and B. J. Kernohan. 1999. Evaluating reliability of habitat suitability index models.
Wildlife Sostciety Bulletin 27:973-985.
Ruel, J.-C. 2000. Factors influencing windthrow in balsam fir forests: from landscape studies to
individual tree studies. Forest Ecology and Management 135:169-178.
Samelius, G., H. Andrén, P. Kjellander, and O. Liberg. 2013. Habitat selection and risk of
predation: re-colonization by lynx had limited impact on habitat selection by roe deer. PLoS
ONE 8:e75469.
Santos, J. P.V., J. Vicente, M. Villamuelas, E. Albanell, E. Serrano, J. Carvalho, C. Fonseca, C.
Gortázar, and J. R. López-Olvera. 2014. Near infrared reflectance spectroscopy (NIRS) for
predicting glucocorticoid metabolites in lyophilised and oven-dried faeces of red deer.
Ecological Indicators 45:522–528.
86
Searle, K., N. Hobbs, and I. Gordon. 2007. It’s the“ foodscape”, not the landscape: using
foraging behavior to make functional assessments of landscape condition. Israel Journal of
Ecology and Evolution 53:297–316.
Skidmore, A. K., J. G Ferwerda, O. Mutanga, S. E. Van Wieren, M. Peel, R. C. Grant, H. H. T.
Prins, F. B. Balcik, and V. Venus. 2010. Forage quality of savannas — Simultaneously mapping
foliar protein and polyphenols for trees and grass using hyperspectral imagery. Remote Sensing
of Environment 114:64–72.
Tolleson, D., R. D. Randel, J. W. Stuth, and D. A. Neuendorff. 2005. Determination of sex and
species in red and fallow deer by near infrared reflectance spectroscopy of the faeces. Small
Ruminant Research 57:141–150.
Valentini, A., C. Miquel, M. A. Nawaz, E. Bellemain, E. Coissac, F. Pompanon, L. Gielly, C.
Cruaud, G. Nascetti, P. Wincker, J. E. Swenson, and P. Taberlet. 2009a. New perspectives in
diet analysis based on DNA barcoding and parallel pyrosequencing: the trnL approach.
Molecular Ecology Resources 9:51–60.
Valentini, A., F. Pompanon, and P. Taberlet. 2009b. DNA barcoding for ecologists. Trends in
Ecology and Evolution 24:110–117.
Van Horne, B., T. Hanley, R. G. Cates, J. D. McKendrick and J. D. Horner. 1988. Influence of
seral stage and season on leaf chemistry of southeastern Alaska deer forage. Canadian Journal
of Forest Research 18:90–99.
Wam, H. K. and O. Hjeljord. 2010. Moose summer diet from feces and field surveys: a
comparative study. Rangeland Ecology and Management 63:387–395.
Williams, P. 2001. Implementation of near-infrared technology. Pages 145-170 In Williams P.
and K. Norris, editors. Near-infrared Technology in the Agricultural and Food Industries.
American Association of Cereal Chemist, St-Paul, MN.
87
Supporting information
Appendix A. Map of the study area showing cold-spots (in blue) and hot-spots (in red) as
determined by a Gi* analysis, using DISTEX values computed solely from
(a) the first or (b) the second of the 2 exclosures.
88
CHAPITRE 4
TEST DE L’HYPOTHÈSE DU
CONTRÔLE DE L’HERBIVORIE PAR
LES RESSOURCES DANS UN
ÉCOSYSTÈME À FORTE DENSITÉ
D’HERBIVORES
89
Avant-propos
La forte pression de broutement entraîne une disparition graduelle de la sapinière endémique à
l’île d’Anticosti. En général, les peuplements de sapinière naturels y sont âgés de plus de 80
ans, et leur recrutement est faible ou inexistant. On observe cependant des parcelles en
régénération, et ce presqu’exclusivement sur une région géologique appelée Chicotte. Dufresne
et al. (2011) avaient émis l’hypothèse selon laquelle les conditions édaphiques des parcelles en
régénération conféraient aux plantes qui s’y trouvent une meilleure défense face aux herbivores.
Le chapitre qui suit vise à tester cette hypothèse en utilisant plusieurs des techniques abordées
et perfectionnées dans le cadre des chapitres précédents (i.e. digestibilité In vitro, teneur en
fibres des fourrages).
Rédigé en collaboration avec Robert L. Bradley et Jean-Pierre Tremblay, l’article découlant de
ce chapitre a été soumis le 6 avril 2015 et est présentement sous presse dans la revue Wildlife
Biology.
90
Testing for bottom-up effects in an overbrowsed boreal landscape
Pierre-Olivier JEAN*1, Robert L. BRADLEY1 & Jean-Pierre TREMBLAY2
1
Département de biologie, Université de Sherbrooke, 2500 boulevard de l'Université,
Sherbrooke, Québec, Canada J1K 2R1.
2
Département de biologie, Université Laval, Québec, QC, G1V0A6, Canada.
Keywords: Bottom-up effects; top-down effects; plant-herbivore interactions, boreal forest
regeneration; overbrowsing
Abbreviations: NDF, neutral detergent fibers; ADF, acid detergent fibers; ADL, acid detergent
lignin; IVTD, in vitro true digestibility; N, nitrogen; P, phosphorus; PCA, principal component
analysis.
91
Résumé
Sur l’île d’Anticosti, la pression de broutement exercée par le cerf de Virginie (Odocoileus
virginianus) compromet la régénération naturelle des peuplements endémiques de sapin
beaumier (Abies balsamea (L.) Mill.). Des événements localisés de régénération surviennent
cependant sur une petite portion du territoire, sur une région géologique appelée Chicotte.
Certains ont émis l’hypothèse selon laquelle les conditions édaphique trouvées sur les parcelles
en régénération avaient un impact négatif sur la qualité alimentaire sur les plantes qui s’y
trouvent, un phénomène qui, à son tour, peut amener une diminution de la pression de
broutement et augmenter le recrutement du sapin. Afin de tester cette hypothèse, nous avons
mis en relation les propriétés chimiques et alimentaires de 4 espèces de fourrages consommés
par le cerf de Virginie (sapin beaumier, Abies balsamea (L.) Mill.; épinette blanche, Picea
glauca (Moench) Voss; maianthème du Canada, Maianthemum canadense Desf.; cornouiller
quatre-temps, Cornus canadensis L.) avec des propriétés édaphiques du sol en utilisant des tests
de Mantel et de Procrustes. Nous avons également testé pour des différences significatives dans
les qualités chimiques et alimentaires des fourrages en fonction de leurs différentes formations
géologiques en utilisant une analyse de la variance (i.e. « one-way ANOVA) et de tests de
Tukey. Nous avons identifié des corrélations significatives (p < 0,05) entre les propriétés du sol
et les propriétés chimiques et alimentaires pour l’épinette blanche avec les tests de Mantel et de
Procrustes. L’épinette blanche montrait une plus haute teneur en fibre et des valeurs plus faibles
de digestibilité In vitro sur les sites dont la disponibilité en nutriments était à son plus faible.
Nos résultats, cependant, ne permettent pas d’expliquer la régénération du sapin beaumier sur
la région géologique de Chicotte, mais suggèrent un rôle potentiel des conditions édaphiques
sur la qualité nutritionnelle de certaines plantes.
92
Abstract
On Anticosti Island (Quebec, Canada), overbrowsing by white-tailed deer (Odocoileus
virginianus) has substantially modified plant communities and reduced the recruitment of
balsam fir (Abies balsamea) seedlings over most of the territory. An exception to this
phenomenon has been observed in localised patches occurring on a single geological deposit
named Chicotte, where the natural recruitment of balsam fir is occurring even in the presence
of a large white-tailed deer population. We hypothesized that edaphic properties within the
Chicotte deposit could result in lower forage quality, which in turn could reduce browsing
pressure and allow fir regeneration to occur (i.e. bottom-up effects). To test this hypothesis, we
measured soil properties and foliage chemistry of 4 forage species (balsam fir, white spruce
(Picea glauca), Canada mayflower (Maianthemum canadense) and Canada bunchberry (Cornus
canadensis)) collected on each of 3 geological deposits on Anticosti Island: Chicotte, Becscie
and Jupiter (the latter two considered as controls). Contrary to expectation, results from principal
component analysis suggested that Chicotte was the most fertile, whereas Becsie was the least
fertile, of the 3 deposits. Furthermore, balsam fir foliage chemistry did not respond to geological
deposit. Conversely, Mantel et Procrustes tests revealed a significant correlation between soil
properties and forage quality for white spruce, consistent with the Carbon-Nutrient Balance
Hypothesis. Univariate tests confirmed that neutral detergent fiber concentrations in white
spruce were higher on the Becscie than on the Chicotte deposit. Likewise, in vitro true
digestibility of both white spruce and Canada bunchberry foliage were lower on the Becscie
than on the Chicotte deposit. Although we failed to demonstrate why balsam fir recruitment
occurs on the Chicotte deposit, our data demonstrate that edaphic properties may affect the
quality of some forage types, which potentially affect foraging patterns in overbrowsed boreal
landscapes.
93
Introduction
Forest diversity and structure may be influenced by both top-down and bottom-up trophic
interactions (Hunter and Price 1992; Vucetich and Peterson 2004). For instance, high population
densities of herbivores may exert a top-down pressure on plant communities by limiting the
recruitment, growth and survival of forage plants while encouraging the expansion of less
palatable species (Kielland and Bryant 1998). The resulting plant community structure may, in
turn, affect soil properties by controlling rhizosphere interactions, nutrient uptake and plant litter
quality (Wardle et al. 2004; Dufresne et al. 2009). Conversely, bottom up interactions may result
from low soil fertility promoting plant chemical defenses against herbivores (Bryant et al. 1983;
Coley et al. 1985; Dufresne et al. 2011), which in turn can limit the amount of energy transferred
to higher trophic levels. While both top-down and bottom-up processes may occur
concomitantly, most studies in forest landscapes characterized by overbrowsing have focused
on top-down control (e.g. Chollet et al. 2013).
On Anticosti Island, heavy browsing from a large population of introduced white-tailed deer
prevents the natural recruitment of native balsam fir (Abies balsamea (L.) Mill.) (Potvin et al.,
2003), the preferred winter forage species. This, in turn, has favoured the recruitment of white
spruce (Picea glauca (Moench) Voss.) seedlings (Hidding et al. 2013; Barrette et al. 2014), a
less palatable winter forage species for white-tailed deer. Although these top-down effects seem
to control forest succession on most of Anticosti Island, natural regeneration of balsam fir stands
can commonly be observed within localized patches over one specific geological deposit named
Chicotte (Chouinard and Fillion 2005), which covers approximately 700 km2 (Appendix A).
These patches of balsam fir regeneration correspond to an area that had previously been
dominated by mature balsam fir stands, but was subsequently ravaged in 1971-72 by an eastern
hemlock looper (Lambdina fiscellaria fiscellaria Guenée) epidemic. The epidemic covered a
total of 1,165 km2 (15% of the island) and extended far beyond the Chicotte formation, into
94
areas where balsam fir regeneration is uncommon. We thus hypothesized that balsam fir
regeneration on the Chicotte formation is driven by low soil fertility, compared to other
geological deposits with infrequent occurrence of balsam fir regeneration. Low soil fertility is
expected to stimulate carbon-based plant defenses and reduce their digestibility and palatability
to herbivores (Bryant et al. 1983).
Overbrowsing on Anticosti Island by white-tailed deer has also reduced the diversity and
abundance of preferred summer forage species. For example, many deciduous trees such as
paper birch (Beluta papyrifera Marsh.), as well as a large number of palatable understory plants,
have been eradicated on most of the island. Two notable exceptions are Canada mayflower
(Maianthemum canadense Desf.) and Canada bunchberry (Cornus canadensis L.), two
herbaceous forage species preferred by deer (e.g. Crawford 1982; Rooney 1997) that remain
abundant on Anticosti Island. To date, there is no assessment of how edaphic properties may
affect the nutritional quality of these two summer forage species.
We report on a study performed on Anticosti Island, where we tested for bottom-up controls on
forage quality of 4 forage species. We related the chemical quality and in vitro digestibility of
forages to soil properties measured within the Chicotte geological region, as well as within two
other geological deposits (i.e. controls) named Becscie and Jupiter. Our data provide
presumptive evidence that bottom-up effects may influence foraging patterns in overbrowsed
boreal landscapes.
95
Material and methods
Study area
Anticosti Island (7,943 km2) is located in the Gulf of St-Lawrence, Canada (49°28'N, 63°00'W).
It is part of the eastern balsam fir–white birch bioclimatic zone (Grondin et al. 1996). The subboreal maritime climate brings cool summers (mean July temperature of 16 °C) and cold winters
(mean January temperature of -11 °C). The average annual precipitation is 917 mm
(Environment Canada 2006), of which approximately one-third falls as snow. About 200 whitetailed deer were introduced on the island in 1896, and the population increased rapidly in the
absence of natural predators. Accordingly, white-tailed deer herbivory pressure has been
drastically transforming plant communities since the late 1920s (Marie-Victorin and RollandGerman 1969). Heavy browsing has resulted in a loss of the most palatable plant species at the
ground level, across the entire island.
Our study was conducted on 3 distinct geological deposits. The Chicotte deposit originates from
the upper Llandovery (Telychian) and is characterized paleontologically as a speciose and
crinoid-rich sand-shoal complex (Desrochers 2006).
The older Jupiter deposit dates back to the late Llandoverian and is composed primarily of
highly fossiliferous limestone (Duffield 1985). The oldest Becscie deposit is predominantly
limestone with calcareous shale and siliciclastics, and is characterized by a distinct earliest
Silurian (early Rhuddanian) brachiopod fauna (Sami and Desrochers 1992). A map of these 3
geological deposits is provided in Appendix A.
96
Sampling
On each geological deposit, we established 7 sampling plots (700 m2) that each contained the 4
forage species of interest (balsam fir, white spruce, Canada mayflower and Canada bunchberry).
The minimal distance between plots within each deposit was 1 km, which far exceeds the spatial
dependence of forest soil chemical properties (Qian and Klinka 1995). Foliar samples of the 2
summer forage species (Canada mayflower and Canada bunchberry) were collected between
July 10-14, whereas winter forage species (balsam fir and white spruce) were sampled between
September 1-7. For the winter forage species, only mature trees were sampled given the low
recruitment of this species on the Jupiter and Becscie deposits. The 84 plant samples (i.e. 7 plots
x 3 geological deposits x 4 species) were dried in an air-draft oven at 45 °C and ground in a
Centrifugal mill to pass a 2 mm mesh (Retsch GmbH, Haan, Germany). For the Becscie deposit,
we removed one sample of Canada bunchberry and Canada mayflower from the subsequent
analyses due to insufficient plant material.
In the centre of each plot, we established a 9-point soil sampling grid (10 m x 10 m) from which
we noted the average thickness of the organic forest floor. We then collected 1 mineral soil core
(0–10 cm depth) at each sampling point, and bulked these 9 cores into one composite sample
per plot.
97
Forage and soil analyses
To evaluate forage quality, we measured both chemical properties and in vitro true digestibility
(IVTD) of each forage type. Proximate C fractions (neutral detergent fibers (NDF), acid
detergent fibers (ADF) and acid detergent lignin (ADL)) were determined based on the protocol
of Goering and Van Soest (1970), using an Ankom Fiber Analyzer 200 (Ankom Technology,
Macedon, NY). Hemicellulose concentration in forage was estimated as the difference between
NDF and ADF concentrations, while cellulose concentration was estimated as the difference
between ADF and ADL concentrations. Foliar N was measured by high temperature combustion
followed by gas analysis, using a Vario Macro C&N Analyzer (Elementar Analysensysteme,
Hanau, Germany). Total P was determined using a magnesium acetate dry ashing method to
avoid volatilisation of organic P (Schulte et al. 1987); ashes were then dissolved into nitric acid,
evaporated to dryness, redissolved into a Mehlich III matrix and measured by an ascorbic acidmolybdate colorimetric method (Murphy and Riley 1962).
We determined IVTD of each forage sample based on the protocol described by Goering and
Van Soest (1970), using F57 filter bags and a Daisy Incubator (Ankom Technology, Macedon,
NY, USA). To do so, we accompanied hunters on Anticosti Island and collected the rumen of 7
freshly killed deer. Each rumen was ligatured to maintain anaerobiosis and conserved in a 40
°C water-filled insulated chamber for transport, which never exceeded 2 h. The rumen content
was filtered through cheese cloths and IVTD (i.e. weight loss) was measured following a 48 h
incubation period at 39.5 °C
Fresh soil subsamples were extracted in aqueous 1.0 N KCl and analysed colorimetrically for
NH4+ (salicylate–nitroprusside-hypochlorite) and NO3− (cadmium reduction) concentrations,
using a Technicon Autoanalyser (Pulse Instrumentation, Saskatoon, Canada). The pH of air98
dried and sieved (2 mm mesh) mineral soil was measured in 1:2 (w:w) mixtures with deionized
water. Air-dried samples were extracted in Mehlich III solution to which 2% lanthanum oxide
had been added (Tran and Simard, 1993), and the extracts were analyzed for base cations (Ca2+,
Mg2+, K+) using an AAnalyst 100 atomic absorption spectrometer (Perkin Elmer Corp.,
Norwalk, CT). Total C and N were analyzed by high temperature combustion followed by gas
analysis, using a Vario Macro C and N Analyzer (Elementar Analysensysteme, Hanau,
Germany). Mineral soil texture was determined by the hydrometer method (Bouyoucos 1936)
after removing the organic matter by combustion in a muffle furnace (400 °C for 24 h).
Statistical analyses
We first performed principal component analysis (PCA) in order to discriminate sampling plots
based either on soil properties or on forage quality. For each PCA, we retained the 7 variables
that explained the maximum amount of variance in the first 2 principal components, in order to
keep a 3:1 ratio between sites and variables (Grossman et al. 1991; Williams and Titus 1988).
From each PCA, we extracted a distance matrix based on the scores along the first 2 principal
components. We then tested for the correlation between soil properties and forage quality among
sites by correlating the 2 score sets using both a Mantel test (Mantel 1967) and a Procrustean
approach (Jackson 1995). PCAs, Mantel tests and Procrustes tests were performed using the R
vegan package (Oksanen et al. 2014). Finally, for each plant species, we tested for statistically
significant differences in forage quality between geological deposits, using one-way analyses of
variance and Tukey’s tests (R Core Development Team 2014).
99
Results
Figure 1. Ordination biplots generated from principal component analyses discriminating
sampling plots according to (a) soil chemical properties, and (b) white spruce
foliar chemistry. Polygons circumscribe plots found within each of the 3
geological deposits.
Results of PCA based on soil properties suggest higher soil nutrient concentrations, higher soil
pH, and thinner forest floors on the Chicotte than on the Becscie deposit, while soil properties
on the Jupiter deposit are intermediate between Chicotte and Becscie (Fig.1a; Appendix B).
Both Mantel (r = 0.21, p = 0.04) and Procrustes tests (r = 0.52, p = 0.002) revealed significant
100
correlations between PCA scores of soil properties and those of forage quality variables for
white spruce. Based on these results, we present PCA biplots for white spruce (Fig.1b), which
show positive correlations between fiber concentrations (NDF, ADF, ADL, hemicellulose) and
the Becscie deposit, and a positive correlation between IVTD values and the Chicotte deposit.
The results of both Procrustes and Mantel tests are summarized in Appendix C. One-way
analyses of variance showed significant correlations between geological deposits and
digestibility for white spruce (F2,18= 5.03; p=0.02) and Canada bunchberry(F2,17 = 3.89; p=0.04),
and between geological deposits and NDF concentration for white spruce (F2,18= 7.3; p=0.005).
Subsequent Tukey’s tests confirmed that white spruce foliage collected on the Becscie deposit
had significantly higher concentrations of NDF than foliage collected on both Jupiter (p=0.04)
and Chicotte (p = 0.004) deposits (Fig.2a), as well as significantly lower IVTD values (p = 0.02)
than foliage collected on the Jupiter deposit (Fig.2b). Canada bunchberry also showed
significantly higher IVTD values (p = 0.02) on the Chicotte than on the Becscie deposit (Fig.2c).
The detailed results of univariate tests are provided in Appendix D.
101
Figure 2. Results of one-way analyses and Tukey’s tests describing the effects of geological
deposit on (a) in vitro true digestibility of white spruce foliage, (b) neutral
detergent fiber concentration of white spruce foliage, and (c) in vitro true
digestibility of Canada bunchberry foliage. Statistically significant differences in
forage quality are denoted by different lower-case letters.
102
Discussion
Our results do not provide evidence that recruitment of balsam fir on the Chicotte deposit
resulted from bottom-up controls limiting deer herbivory, as previously hypothesized by
Dufresne et al. (2011). On the contrary, balsam fir forage quality did not respond to geological
deposits. Furthermore, our data suggest higher soil fertility, and a potential for higher forage
quality and higher IVTD on the Chicotte than on the Becscie deposit. Thus, balsam fir
regeneration on the Chicotte deposit may have resulted from historical factors. For example, the
current balsam fir stands on Chicotte originated from a 1971-72 eastern hemlock looper
(Lambdina fiscellaria fiscellaria Guenée) epidemic, which produced large quantities of coarse
woody debris (Harmon et al. 1986). Casabon and Pothier (2007) found that coarse woody debris
on Anticosti Island could provide provisional safe sites against deer herbivory and increase the
recruitment of seedlings by 650 %. A higher number of safe sites, combined with higher vertical
growth rates resulting from higher soil fertility, may have allowed regenerating balsam fir on
the Chicotte formation to escape deer browsing (Coley 1988; Skarpe and Hester 2006). A study
from McLaren (1996) supports this hypothesis, as it showed that balsam fir was more resilient
to moose herbivory in forest gaps where light availability and growth rates were higher.
In spite of our unexpected results regarding soil fertility on Chicotte relative to the other 2
deposits, our data did provide presumptive evidence that bottom-up pressures could control
forage quality on Anticosti Island, consistent with the Carbon-Nutrient Balance Hypothesis
(Bryant et al. 1983). This hypothesis stipulates that plants growing under low nutrient
availability are more prone to exhibit carbon-based defenses that can reduce their digestibility
and palatability to herbivores. The Chicotte and Becscie deposits represent opposite ends in a
soil fertility gradient, with Chicotte plots being positively correlated with soil nutrient
concentrations and Becscie plots being positively correlated with forest floor depth (i.e. amount
of sequestered nutrients). Results from multivariate Mantel and Procrustes tests suggest that
103
these differences in soil properties affected forage quality of white spruce. Univariate Tukey’s
tests confirmed significantly higher NDF content in white spruce foliage that was collected
within the Becscie deposit. NDF are structural carbon-based compounds and their concentration
in forages is generally negatively correlated with digestibility and food intake (Van Soest 1994).
Furthermore, univariate tests confirmed a lower IVTD for white spruce and Canada bunchberry
foliage from the Becscie than from the Chicotte deposit. The question remains as to why such
an effect was not observed for the other 2 forage species in our study (i.e. Canada mayflower
and balsam fir). Notwithstanding this conundrum, our data illustrate that soil properties may
affect forage quality and digestibility, which in turn could affect foraging patterns in
overbrowsed boreal landscapes.
Further research is needed to explain how localized events of balsam fir regeneration on
Anticosti Island are occurring. For example, it would be useful to correlate geological deposits
with plant secondary metabolites that affect forage quality for herbivores, such as tannins
(Robbins et al. 1987) and other phenolics (e.g. Stolter et al. 2010). Such knowledge could help
forest managers limit the loss of native balsam fir forests, as well as help ecologists better
understand plant-herbivore interactions.
Acknowledgments
The project was funded by a NSERC Collaborative Research and Development Grant and by a
NSERC Discovery Grant. We are thankful to Bill Parsons, Cédric Frenette-Dussault, Cécile
Meyer, Sonia Van Wijk, Valérie Massé, Sonia DeBellefeuille and Steeve Côté for technical and
logistical support.
104
References
Barrette, M. et al. 2014. Cumulative effects of chronic deer browsing and clear-cutting on
regeneration processes in second-growth white spruce stands. - For. Ecol. Manage. 329: 69–78.
Bouyoucos, G. 1936. Directions for making mechanical analyses of soils by the hydrometer
method. - Soil Sci. 42:225-230.
Bryant, J. P. et al. 1983. Carbon/nutrient balance of boreal plants in relation to vertebrate
herbivory. - Oikos 40: 357–368.
Casabon, C. and Pothier, D. 2007. Browsing of tree regeneration by white-tailed deer in large
clearcuts on Anticosti Island, Quebec. - For. Ecol. Manage. 1: 112–119.
Chollet, S. et al. 2013. Importance for forest plant communities of refuges protecting from deer
browsing. - For. Ecol. Manage. 289: 470–477.
Chouinard, A., and Filion, D. 2005. Impact of introduced white-tailed deer and native insect
defoliators on the density and growth of conifer saplings on Anticosti Island, Québec. Ecoscience 12:506–518.
Coley, P. 1988. Effects of plant growth rate and leaf lifetime on the amount and type of antiherbivore defense. - Oecologia 74: 531–536.
Coley, P. D. et al. 1985. Resource availability and plant antiherbivore defense. – Science. 230:
895–899.
Crawford, H. 1982. Seasonal food selection and digestibility by tame white-tailed deer in central
Maine. - J. Wildlife Manage. 46:974–982.
Desrochers, A. 2006. Rocky shoreline deposits in the Lower Silurian (upper Llandovery,
Telychian) Chicotte Formation, Anticosti Island, Quebec. - Can. J. Earth Sci. 43: 1205–1214.
Duffield, S. 1985. Land-derived microfossils from the Jupiter Formation (upper Llandoverian),
Anticosti Island, Quebec. - J. Paleontol. 59: 1005–1010.
Dufresne, M. et al. 2009. Clearcutting and deer browsing intensity interact in controlling
nitrification rates in forest floor. - Ecoscience 16: 361–368.
Dufresne, M. et al. 2011. Evidence that Soil Depth and Clay Content Control the PostDisturbance Regeneration of Balsam Fir and Paper Birch Under Heavy Browsing from Deer. Ecoscience 18: 363–368.
105
Environnement Canada. 2006. Climate normals and averages daily data reports of Port-Menier’s
station from 1995 to 2005. Available at: http://www.climate.weatheroffice.ec.gc.ca. Accessed 8
June 2006.
Goering, H. K. and Van Soest, P. J. 1970. Forage fiber analyses (apparatus, reagents, procedures
and some applications). — Agric. Res. Serv., US Dept Agric., Washington, DC, USA.
Grossman, G. et al. 1991. Principal component analyses of assemblage structure data: utility of
tests based on eigenvalues. - Ecology 72: 341–347.
Harmon, M. E. et al. 1986. Ecology of coarse woody debris in temperate ecosystems. - Adv.
Ecol. Res. 15:133–302.
Hidding, B. et al. 2013. A large herbivore triggers alternative successional trajectories in the
boreal forest. - Ecology 94: 2852–2860.
Hunter, M. and Price, P. 1992. Playing chutes and ladders: heterogeneity and the relative roles
of bottom-up and top-down forces in natural communities. - Ecology 73: 724–732.
Jackson, D. 1995. PROTEST: a PROcrustean randomization TEST of community environment
concordance. - Ecoscience 2: 297–303.
Kielland, K. and Bryant, J. 1998. Moose herbivory in taiga: effects on biogeochemistry and
vegetation dynamics in primary succession. - Oikos 82: 377–383.
Mantel, N. 1967. The detection of disease clustering and a generalized regression approach. Cancer Res. 27: 209–220.
Marie-Victorin, F. and Rolland-Germain, F. 1969. Flore de l’Anticosti-Minganie. - Presses de
l’Université de Montréal. (In French).
McLaren, B.E. 1996. Plant-Specific Response to Herbivory: Simulated Browsing of Suppressed
Balsam Fir on Isle Royale. - Ecology 77:228-235.
Murphy, J. and Riley, J. 1962. A modified single solution method for the determination of
phosphate in natural waters. - Anal. Chim. Acta 27: 31–36.
Oksanen, J. et al. 2014. vegan: community ecology package
http://CRAN.R-project.org/package=vegan
Potvin, F. et al. 2003. The eradication of balsam fir stands by white-tailed deer on Anticosti
Island , Québec : A 150-year process. - Ecoscience 10: 487–495.
106
Qian, H. and Klinka, K. 1995. Spatial variability of humus forms in some coastal forest
ecosystems of British Columbia. - Ann. Sci. For. 52: 653–666.
R Core Team 2014. R: A Language and Environment for Statistical Computing. Vienna,
Austria.
Rooney, T. 1997. Escaping herbivory: refuge effects on the morphology and shoot demography
of the clonal forest herb Maianthemum canadense. - J. Torrey Bot. Soc. 124:280–285.
Sami, T. and Desrochers, A. 1992. Episodic sedimentation on an early Silurian, storm‐
dominated carbonate ramp, Becscie and Merrimack formations, Anticosti Island, Canada. Sedimentology 39: 355–381.
Schulte, E. E. et al. 1987. Wisconsin procedures for soil testing, plant analysis, and feed and
forage analysis. Department of Soil Science, University of Wisconsin—Madison, WI.
Skarpe, C. and Hester, A. 2008. Plant Traits , Browsing and Grazing Herbivores , and Vegetation
Dynamics. - In: Iain J. Gordon, I. J. and H. H. T. Prins (eds.). The Ecology of Browsing and
Grazing. Springer, pp. 217–261.
Stolter, C. et al. 2010. Herbivores and variation in the composition of specific phenolics of
boreal coniferous trees: a search for patterns. – Chemoecology 20:229-242.
Tran, T. S. and Simard, R. R. 1993. Mehlich III-extractable elements. - In: Carter, M. R. (ed.),
Soil Sampling and Methods of Analysis. Canadian Society of Soil Science, Lewis Publishers,
CRC Press, pp. 43–49.
Van Soest, P. J. 1994. Intake. - In: Nutritional ecology of the ruminant. 2th ed. Cornell
University Press, pp. 337–353.
Vucetich, J. and Peterson, R. 2004. The influence of top–down, bottom–up and abiotic factors
on the moose (Alces alces) population of Isle Royale. - Proc. R. Soc. London Ser. B-Biological
Sci. 271: 183–189.
Wardle, D. et al. 2004. Ecological linkages between aboveground and belowground biota. –
Science 304: 1629–1633.
Williams, B. and Titus, K. 1988. Assessment of sampling stability in ecological applications of
discriminant analysis. - Ecology 69: 1275–1285.
Grondin, P., C. et al. 1996. Cadre bioclimatique de référence des régions écologiques du
Québec. - In: Manuel de Foresterie. Les Presses de l’Université Laval. Québec. Canada. pp.
134-279. (in French)
107
Supporting information
Appendix A. Location of 21 sampling plots distributed over 3 geological deposits on
Anticosti Island, QC, Canada. The island covers a total area of 7,9463 km2.
108
Appendix B. Summary of mineral soil properties for 21 sampling plots located on three geological deposits on Anticosti
Island, QC, Canada
Becscie 1
Becscie 2
Becscie 3
Becscie 4
Becscie 5
Becscie 6
Becscie 7
Mean Becscie
Chicotte 1
Chicotte 2
Chicotte 3
Chicotte 4
Chicotte 5
Chicotte 6
Chicotte 7
pH
4.59
6.28
5.71
4.39
6.00
3.43
6.31
5.24
6.95
5.75
6.68
6.74
6.92
7.04
6.94
N
(mg/g)
0.267
0.190
0.132
0.183
0.140
0.190
0.415
0.217
0.315
0.482
0.241
0.285
0.327
0.178
0.304
C
(mg/g)
3.04
2.53
1.23
3.16
1.70
3.79
8.10
3.37
6.02
8.93
6.56
10.13
4.32
10.13
9.57
Silt
(%)
38
37
24
31
28
41
56
36
13
39
31
39
36
20
19
Sand
(%)
57
58
74
64
67
54
42
59
83
54
64
54
58
78
75
Clay
(%)
5
5
3
5
5
5
3
4
5
7
5
7
6
3
6
Mean Chicotte
Jupiter 1
Jupiter 2
Jupiter 3
Jupiter 4
Jupiter 5
Jupiter 6
Jupiter 7
6.72
6.75
6.69
6.92
6.31
4.90
4.98
6.87
0.305
0.229
0.221
0.176
0.190
0.268
0.312
0.208
7.95
3.44
5.52
2.56
1.80
4.22
5.51
3.81
28
35
29
38
46
33
19
23
66
62
70
57
46
63
78
75
Mean Jupiter
6.20
0.229
3.84
32
64
Sampling plot
Na
Ca
Mg
K
(mg/kg) (mg/kg) (mg/kg) (mg/kg)
0.0596
0.224
1.95
0.114
0.0374
0.218
3.12
0.121
0.0698
0.219
2.68
0.073
0.0721
0.230
2.60
0.093
0.0384
0.213
2.92
0.056
0.0621
0.228
0.94
0.088
0.0553
0.224
2.63
0.117
0.0564
0.222
2.40
0.094
0.0940
0.265
22.34
0.204
0.0830
0.260
8.13
0.157
0.0679
0.211
18.48
0.130
0.0587
0.235
19.42
0.185
0.0673
0.222
13.05
0.199
0.0275
0.191
19.95
0.104
0.0590
0.224
20.84
0.139
NH4
(mg/g)
0.016
0.007
0.004
0.004
0.003
0.005
0.004
0.006
0.044
0.031
0.044
0.032
0.003
0.020
0.026
NO3
(mg/g)
0.0018
0.0168
0.0036
0.0211
0.0127
0.0000
0.0667
0.0175
0.0047
0.0078
0.0080
0.0058
0.0329
0.0158
0.0219
Forest floor
depth (cm)
6
23
24
15
16
15
7
15
8
11
10
9
17
9
12
6
4
2
6
8
5
4
3
0.0653
0.1180
0.0841
0.0986
0.0894
0.0933
0.1590
0.1095
0.230
0.203
0.214
0.201
0.226
0.202
0.290
0.214
17.46
10.42
16.13
12.46
3.18
3.26
5.50
11.84
0.160
0.098
0.140
0.088
0.043
0.098
0.260
0.147
0.029
0.002
0.048
0.008
0.005
0.008
0.017
0.031
0.0139
0.0343
0.0028
0.0042
0.0036
0.0118
0.0107
0.0030
11
12
11
53
34
13
9
10
4
0.1074
0.221
8.97
0.125
0.017
0.0101
20
*Methods used for measuring the variables are described in the Material and Methods section above.
109
Appendix C. Results of multivariate analyses testing the relationship between plant
nutritional quality of 4 forage species and soil properties.
Type of test
Procrustes
Mantel
Species
Balsam fir
White spruce
Canada bunchberry
Canada mayflower
Balsam fir
White spruce
Canada bunchberry
Canada mayflower
r
0.22
0.52
0.29
0.42
0.09
0.21
0.04
0.14
p-value
0.7
0.004
0.4
0.07
0.8
0.02
0.3
0.08
*The analyses are achieved using the first two principal components of PCAs that
describe the variance of either the plant nutritional quality or soil properties.
110
Appendix D: Results of univariate tests testing the effect of the geological deposit on the nutritional quality of 4 forage
species. The analyses are achieved by using the Tukey's method in conjunction with one-way analysis of
variance (ANOVA).
Plant
species
Balsam fir
White
spruce
Forage quality
variable (%)
nitrogen
carbon
phosphorus
IVTD
NDF
ADF
ADL
hemicellulose
cellulose
nitrogen
carbon
phosphorus
IVTD
NDF
ADF
ADL
hemicellulose
cellulose
Variance explained by
deposit
(i.e. in ANOVA)
R2
0.03
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.11
0.03
0.29
0.38
0.45
0.13
0.10
0.20
p-value
(i.e. in
ANOVA)
0.3
0.5
0.6
0.7
0.4
0.4
0.4
0.6
0.8
0.5
0.1
0.3
0.02
0.005
0.001
0.1
0.1
0.06
Mean values
Becscie
Chicotte
Jupiter
1.18
51.66
0.110
66.3
39.7
37.4
23.2
2.30
14.2
1.15
50.48
0.117
52.7a
55.6a
45.6a
27.2
9.9
18.4
1.08
51.81
0.097
66.3
40.9
39.6
25.9
1.27
13.7
1.10
50.67
0.106
54.4ab
51.7b
42.5b
25.5
9.2
17.0
1.19
52.01
0.108
65.6
39.5
38.5
24.9
0.96
13.7
1.11
51.02
0.101
56.0b
52.9b
44.4a
27.1
8.4
17.3
111
Canada
bunchberry
Canada
mayflower
nitrogen
carbon
phosphorus
IVTD
NDF
ADF
ADL
hemicellulose
cellulose
nitrogen
carbon
phosphorus
IVTD
NDF
ADF
ADL
hemicellulose
cellulose
0.00
0.00
0.00
0.23
0.00
0.00
0.00
0.00
0.04
0.09
0.00
0.04
0.12
0.09
0.00
0.00
0.07
0.12
0.5
0.6
0.9
0.04
0.5
0.6
0.7
0.4
0.3
0.2
0.4
0.3
0.1
0.2
0.5
0.7
0.21
0.1
1.68
43.87
0.14
87.4a
41.1
31.9
11.1
9.3
20.8
1.99
44.82
0.17
87.2
38.6
28.2
9.9
10.4
18.3
1.79
43.55
0.15
88.8b
39.9
30.7
10.3
9.2
20.4
1.82
44.49
0.15
87.6
44.6
29.2
9.7
15.4
19.6
1.79
43.63
0.15
88.4ab
43.9
31.0
11.4
12.9
19.6
1.96
43.95
0.17
86.2
39.3
28.3
10.3
11.0
18.0
*Bold characters show significant differences in quality between forage collected within different geological deposits.
Statistically significant differences in forage quality as shown by Tukey's tests (p<0.05) are denoted by different lowercase letters.
112
CONCLUSION
Valeur scientifique
Le travail réalisé dans le cadre de ces travaux de doctorat m’ont permis d’atteindre les objectifs
généraux de ma thèse, soit de (1) développer des méthodes afin d’évaluer le contenu de la diète
d’un herbivore et sa qualité et de (2) implémenter la méthodologie développée dans le cadre
d’études écologiques à l’échelle du paysage.
J’ai en premier lieu développé une méthode qui utilise la spectroscopie dans le proche infrarouge
(SPIR) afin de prédire des composantes de la diète du cerf à partir de fèces. S’il est implémenté
dans un programme de suivi de la qualité alimentaire de la diète du cerf à Anticosti, cet outil
permettra d’améliorer l’effort d’échantillonnage tout en réduisant les coûts d’analyse. Puisque
la SPIR a été plus performante pour prédire la proportion des composantes les moins digestibles
de la diète, elle a un fort potentiel d’être applicable dans des écosystèmes soumis à des fortes
densités et où les herbivores consomment une diète comparable (e.g. écosystèmes nordiques à
fortes densité d’orignaux).
J’ai par la suite développé et testé de nouvelles méthodes afin d’évaluer la qualité nutritionnelle
de fourrages pour le cerf de Virginie. J’ai pu d’une part émettre des recommandations destinées
aux gestionnaires et aux écologiques désireux d’évaluer la qualité nutritionnelle de la faune
ruminante et d’en faire un monitoring sur une base régulière. D’autre part, ce volet de mon
projet m’a permis de mettre en lumière le potentiel de mesures les produits de digestions in vitro
113
(i.e. gas et acides gras volatils) dans le cadre d’études écologiques sur les ruminants afin d’en
apprendre davantage sur leur écophysiologie.
J’ai finalement transposé mes recherches à l’échelle du paysage en appliquant, d’une part, une
toute nouvelle approche qui combine la SPIR et les géostatistiques afin de cartographier la
qualité nutritionnelle de la diète consommée par le cerf à l’échelle du paysage. Dans une récente
revue de littérature, De Gabriel et al. (2014) ont pu identifier différents défis de taille dans
l’étude de l’écologie nutritionnelle des herbivores. Les auteurs ont notamment identifié que la
complexité inhérente à la chimie des plantes rend difficile l’évaluation de celles-ci pour leur
qualité nutritionnelle sur la base d’un seul critère ou devise (e.g. azote, énergie). De plus, il est
difficile de transposer l’évaluation de la qualité alimentaire de fourrages du laboratoire à
l’écosystème, puisque la distribution spatiale de la disponibilité et la qualité des ressources
disponibles est souvent mal décrite. Ce volet de mon doctorat s’avère une contribution originale
puisqu’elle s’attaque de front à ces deux défis majeurs. Puis, finalement, j’ai testé une hypothèse
selon laquelle les conditions édaphiques des fourrages disponibles pour le cerf affectaient la
qualité nutritionnelle de celles-ci, ce qui a contribué à améliorer nos connaissances sur les
interactions multi-trophiques entre le sol, les plantes et les herbivores à l’île d’Anticosti.
Futures recherches
Si mon projet a permis de développer de nouvelles méthodologies, l’étape suivante sera
certainement de les transposer dans le cadre d’études écologiques et de programmes de suivi
des populations d’herbivores. Je vois trois champs de recherche découlant de mon projet de
doctorat qui constituent, à mon avis, des avenues forts intéressantes pour contribuer à la
recherche sur l’écologie nutritionnelle des herbivores :
114
1-Implémentation de la SPIR dans les programmes de monitoring écologique
Depuis près de vingt ans, des écologistes démontrent avec succès les avantages de la technologie
SPIR dans la prédiction de variables écologiques associées à des échantillons de plantes (e.g.
Rutherford et al., 1996; Foley et al., 1998), de sols (e.g. Cécillon et al., 2009) ou de matière
fécale (Dixon et Coates 2009). Bien que d’une part, il soit essentiel de continuer à développer
de nouvelles applications de la SPIR afin de démontrer le potentiel de cette technique, je constate
un manque à gagner important dans l’implémentation de ces nouvelles méthodes dans des
programmes de monitoring écologique. Un des rares projets de longue haleine en ce sens a été
entrepris par l’équipe de Doug Tolleson de l’Université de l’Arizona, qui utilise des appareils
SPIR portables pour effectuer le suivi de la qualité des pâturages pour le bétail (e.g. Tolleson
2010). Ces nouveaux appareils SPIR, plus pratiques et moins onéreux, sont disponibles depuis
quelques années, et devraient contribuer à populariser la SPIR, notamment auprès des
gestionnaires du milieu naturel, des écologistes et des forestiers.
2- Les sous-produits de digestion ruminale pour étudier l’écophysiologie des ruminants
sauvages
Plusieurs chercheurs en sciences animales s’intéressent à mesurer les sous-produits de
fermentation puisqu’ils en tirent une foule d’informations sur la cinétique et le métabolisme de
la digestion des ruminants. Ces connaissances permettent à leur tour d’augmenter la productivité
des animaux de bétail (Getachew et al., 2004). Tel que mentionné dans le chapitre 2 de cette
thèse, il serait hautement souhaitable d’utiliser la théorie développée en sciences animales afin
de transposer ces techniques dans une perspective écologique. Par exemple, on pourrait émettre
l’hypothèse qu’une femelle lactante sélectionne préférentiellement des fourrages lipogéniques
115
(i.e. pour la synthèse du gras du lait), qui génèrent une grande proportion d’acétate dans le rumen
(e.g. van Knegsel et al., 2007). On pourrait également émettre l’hypothèse que le cerf de
Virginie de l’île d’Anticosti possède un inoculum ruminal qui permet une digestion plus rapide
que son homologue du continent, puisqu’il doit ingérer davantage de fourrage de faible qualité
pour subvenir à ses besoins énergétiques. À condition de pouvoir utiliser des liqueurs ruminales
adaptées à l’animal d’étude, cette approche permettrait de voir sous l’angle physiologique les
interactions plantes-herbivores.
3- Étude des interactions entre le microbiome ruminal et l’environnement
On sait que les ruminants forment une symbiose obligatoire avec leur flore ruminale, qui leur
permet de digérer des plantes très riches en fibres par fermentation. On sait aussi que les
bactéries présentes à la surface des plantes sont souvent les mêmes qui composeront la flore
bactérienne à l’intérieur du rumen. Si certaines études ont démontré une coévolution potentielle
entre les plantes consommées par les ruminants et les bactéries présentes sur les feuilles (e.g.
Hess et al., 2011; Brulc et al., 2009), on connaît cependant encore peu les impacts potentiels
des relations entre la flore ruminale et les processus écosystémiques (e.g. sur les processus du
sol, sur l’assemblage des communautés végétales etc.). En raison de l’importance de cette flore
pour la survie des populations de ruminants (et par extension, des autres herbivores
fermenteurs), et du rôle prépondérant de ces mêmes herbivores dans bon nombre d’écosystèmes,
je crois que l’étude de ces processus permettrait d’intéressantes percées dans l’étude des
interactions multi-trophiques.
Hackmann et Spain (2010) soulignaient qu’une collaboration étroite entre chercheurs en
production animale et écologistes pouvaient être grandement utile au développement de
nouvelles stratégies de production animale. De la même manière je crois qu’une meilleure
116
collaboration entre écologistes et chercheurs en sciences animales est cruciale pour mieux
comprendre l’écologie nutritionnelle des ruminants sauvages.
Ce projet de recherche
multidisciplinaire m’a poussé à explorer une vaste littérature dans le domaine des sciences
animales, de l’agriculture et de la chimie analytique. Comme mot de la fin, j’encourage donc le
lecteur de cette thèse à approfondir ses lectures et à s’intéresser à des sujets qui sortent
complètement de son domaine d’expertise. On en sort parfois avec de drôles d’idées, mais ce
sont ces mêmes idées qui permettent de percevoir une problématique sous un angle différent.
117
BIBLIOGRAPHIE
Bender, L. C., J. G. Cook, R. C. Cook, et P. B. Hall. (2008). Relations between nutritional
condition and survival of North American elk Cervus elaphus. Wildlife Biology 14:70–80.
Berteaux, D., M. Crête, J. Huot, J. Maltais, et J. P. J.-P. Ouellet. (1998). Food choice by whitetailed deer in relation to protein and energy content of the diet: A field experiment. Oecologia
115:84–92.
Bertrand, D. (2002). La spectroscopie proche infrarouge et ses applications dans les industries
de l’alimentation animale. Productions Animales 15:209–219.
Brooks III, J., et M. J. Anderson. (1984). Infrared reflectance analysis of forage quality for elk.
The Journal of Wildlife Management 48:254–258.
Brulc, J. M., D. A. Antonopoulos, M. E. B. Miller, M. K. Wilson, A. C. Yannarell, E. A.
Dinsdale, R. E. Edwards, E. D. Frank, J. B. Emerson, P. Wacklin, P. M. Coutinho, B. Henrissat,
K. E. Nelson, et B. A. White. (2009). Gene-centric metagenomics of the fiber-adherent bovine
rumen microbiome reveals forage specific glycoside hydrolases. Proceedings of the National
Academy of Sciences USA 106:1948–1953.
Catchpole, W. R., et C. J. Wheelert. (1992). Review Estimating plant biomass : A review of
techniques. Australian Journal of Ecology 17:121–131.
Cécillon, L., B. G. G. Barthès, C. Gomez, D. Ertlen, V. Genot, M. Hedde, a. Stevens, et J. J. J.
Brun. (2009). Assessment and monitoring of soil quality using near-infrared reflectance
spectroscopy (NIRS). European Journal of Soil Science 60:770–784.
Chilliard, Y., A. Ferlay, Y. Faulconnier, M. Bonnet, J. Rouel, et F. Bocquier. (2008). Adipose
tissue metabolism and its role in adaptations to undernutrition in ruminants. Proceedings of the
Nutrition Society 59:127–134.
Codron, D., J. A. Lee-Thorp, M. Sponheimer, J. Codron, D. D. E. Ruiter, et J. S. Brink. (2007).
Significance of diet type and diet quality for ecological diversity of African ungulates. The
Journal of Animal Ecology 76:526–37.
Cook, J. G., B. Johnson, et R. C. Cook. (2004). Effects of summer‐autumn nutrition and
parturition date on reproduction and survival of elk. Wildlife Monographs 155:1–61.
Courant, S., et D. Fortin. (2012). Time allocation of bison in meadow patches driven by potential
energy gains and group size dynamics. Oikos 121:1163–1173.
118
Crête, M., et J. Huot. (1993). Regulation of a large herd of migratory caribou: summer nutrition
affects calf growth and body reserves of dams. Canadian Journal of Zoology 71:2291–2296.
Dearden, B. L., R. E. Pegau, et R. M. Hansen. (1975). Precision of microhistological estimates
of ruminant food habits. The Journal of Wildlife Management 39:402–407.
DeGabriel, J., B. Moore, et A. Felton. (2014). Translating nutritional ecology from the
laboratory to the field: milestones in linking plant chemistry to population regulation in
mammalian browsers. Oikos 3:298–308.
Dixon, R., et D. Coates. (2009). Review: Near infrared spectroscopy of faeces to evaluate the
nutrition and physiology of herbivores. Journal of Near Infrared Spectroscopy 17:1-31.
Foley, W. J., A. McIlwee, I. Lawler, L. Aragones, A. P. Woolnough, et N. Berding. (1998).
Ecological applications of near infrared reflectance spectroscopy - A tool for rapid, costeffective prediction of the composition of plant and animal tissues and aspects of animal
performance. Oecologia 116:293–305.
Getachew, G., P. H. Robinson, E. J. DePeters, et S. J. Taylor. (2004). Relationships between
chemical composition, dry matter degradation and in vitro gas production of several ruminant
feeds. Animal Feed Science and Technology 111:57–71.
Giroux, M. A., C. Dussault, N. Lecomte, J. P. Tremblay, et S. D. Côté. (2012). A new way of
assessing foraging behaviour at the individual level using faeces marking and satellite telemetry.
PloS one 7:e49719.
Hackmann, T. J., et J. N. Spain. (2010). Invited review: ruminant ecology and evolution:
perspectives useful to ruminant livestock research and production. Journal of Dairy Science
93:1320–34.
Hanley, T. A., D. E. Spalinger, K. J. Mock, O. L. Weaver, et G. M. Harris. (2012). Forage
resource evaluation system for habitat—deer: an interactive deer habitat model (Washington:
U.S. Department of Agriculture).
Hess, M., A. Sczyrba, R. Egan, T.-W. Kim, H. Chokhawala, G. Schroth, S. Luo, D. S. Clark, F.
Chen, T. Zhang, R. I. Mackie, L. A. Pennacchio, S. G. Tringe, A. Visel, T. Woyke, Z. Wang, et
E. M. Rubin. (2011). Metagenomic discovery of biomass-degrading genes and genomes from
cow rumen. Science 331:463–467.
Joffre, R., D. Gillon, P. Dardenne, R. Agneessens, et R. Biston. (1992). The use of near-infrared
reflectance spectroscopy in litter decomposition studies. Annales des Sciences Forestieres
49:481–488.
119
Köhler, A., D. Raubenheimer, et S. W. Nicolson. (2012). Regulation of nutrient intake in nectarfeeding birds: insights from the geometric framework. Journal of Comparative Physiology. B,
Biochemical, Systemic, and Environmental Physiology 182:603–11.
Lefort, S., J.-P. Tremblay, F. Fournier, F. Potvin, et J. Huot. (2007). Importance of balsam fir
as winter forage for white-tailed deer at the northeastern limit of their distribution range.
Ecoscience 14:109–116.
Levin, D. A. (1976). The Chemical Defenses of Plants to Pathogens and Herbivores. Annual
Review of Ecology and Systematics 7:121–159.
MacArthur, R. H., et E. R. Pianka. (1966). On optimal use of a patchy environment. American
Naturalist 100:603–609.
Marie-Victorin, F., et F. Rolland-Germain. (1969). Flore de l’Anticosti-Minganie (Montréal;
Presses de l’Université de Montréal).
Massy, W. F. (1965). Principal components regression in exploratory statistical research.
Journal of the American Statistical Association 60:234–256.
Mayer, D. G., et D. G. Butler. (1993). Statistical validation. Ecological modelling 68:21–32.
McIlwee, A. M., I. R. Lawler, S. J. Cork, et W. J. Foley. (2001). Coping with chemical
complexity in mammal-plant interactions: near-infrared spectroscopy as a predictor of
Eucalyptus foliar nutrients and of the feeding rates of folivorous marsupials. Oecologia
128:539–548.
McInnis, M. L., M. Vavra, et W. C. Krueger. (1983). A comparison of four methods used to
determine the diets of large herbivores. Journal of Range Management 36:302–306.
McLean, S., et A. J. Duncan. (2006). Pharmacological perspectives on the detoxification of plant
secondary metabolites: implications for ingestive behavior of herbivores. Journal of Chemical
Ecology 32:1213–28.
Miyashita, T., M. Suzuki, M. Takada, G. Fujita, K. Ochiai, et M. Asada. (2007). Landscape
structure affects food quality of sika deer (Cervus nippon) evidenced by fecal nitrogen levels.
Population Ecology 49:185–190.
Mysterud, A. (2000). The relationship between ecological segregation and sexual body size
dimorphism in large herbivores. Oecologia 124:40–54.
Mysterud, A. (1998). The relative roles of body size and feeding type on activity time of
temperate ruminants. Oecologia 113:442–446.
120
Osborne, B. G. (1986). Near-infrared spectroscopy in food analysis. Encyclopedia of Analytical
Chemistry 1:1-14.
Pichersky, E., et D. R. Gang. (2000). Genetics and biochemistry of secondary metabolites in
plants: An evolutionary perspective. Trends in Plant Science 5:439–445.
Potvin, F., P. Beaupré, et G. Laprise. (2003). The eradication of balsam fir stands by whitetailed deer on Anticosti Island, Québec : A 150-year process. Ecoscience 10:487–495.
Provenza, F. D., J. A. Pfister, et C. D. Cheney. (1992). Mechanisms of learning in diet selection
with reference to phytotoxicosis in herbivores. Journal of Range Management 45:36–45.
Raubenheimer, D., G. E. Machovsky-Capuska, C. a Chapman, et J. M. Rothman. (2014).
Geometry of nutrition in field studies: an illustration using wild primates. Oecologia 177:223–
234.
Raubenheimer, D., et S. J. Simpson. 1993. The geometry of compensatory feeding in the locust.
Animal Behaviour 45:953–964.
Robbins, C., T. Hanley, A. Hagerman, O. Hjelford, D. L. Baker, C. C. Schwartz, et W. W.
Mautz. (1987a). Role of tannins in defending plants against ruminants: reduction in protein
availability. Ecology 68:98–107.
Robbins, C. T., S. Mole, A. E. Hagerman, et T. A. Hanley. (1987b). Role of Tannins in
Defending Plants Against Ruminants: Reduction in Dry Matter Digestion? Ecology 68:1606–
1615.
Ruckstuhl, K. E., M. Festa-Bianchet, et J. T. Jorgenson. (2003). Bite rates in Rocky Mountain
bighorn sheep (Ovis canadensis): effects of season, age, sex and reproductive status. Behavioral
Ecology and Sociobiology 54:167–173.
Rutherford, R. S. S., J. Van Staden, et J. Staden. (1996). Towards a rapid near-infrared technique
for prediction of resistance to sugarcane borer Eldana saccharina walker (Lepidoptera:
Pyralidae) using stalk surface wax. Journal of Chemical Ecology 22:681–694.
Stolter, C., R. Julkunen-Tiitto, et J. U. Ganzhorn. (2006). Application of near infrared
reflectance spectroscopy (NIRS) to assess some properties of a sub-arctic ecosystem. Basic and
Applied Ecology 7:167–187.
Taillon, J., D. G. Sauvé, et S. D. Coté. (2006). The effects of decreasing winter diet quality on
foraging behavior and life-history traits of white-tailed deer fawns. Journal of Wildlife
Management 70:1445–1454.
121
Tolleson, D. R., R. D. Randel, J. W. Stuth, et D. A. Neuendorff. (2005). Determination of sex
and species in red and fallow deer by near infrared reflectance spectroscopy of the faeces. Small
Ruminant Research 57:141–150.
Van Knegsel, a T. M., H. van den Brand, J. Dijkstra, W. M. van Straalen, R. Jorritsma, S.
Tamminga, et B. Kemp. (2007). Effect of glucogenic vs. lipogenic diets on energy balance,
blood metabolites, and reproduction in primiparous and multiparous dairy cows in early
lactation. Journal of Dairy Science 90:3397–409.
Vourc’h, G., M. De Garine-Wichatitsky, A. Labbé, D. Rosolowski, J. L. Martin, et H. Fritz.
(2002). Monoterpene effect on feeding choice by deer. Journal of Chemical Ecology 28:2411–
2427.
Walker, J. W., et D. R. Tolleson. (2010). Shining Light on Manure Improves Livestock and
Land Management (College Station: Texas AgriLife Research and Society for Range
Management).
White, R. G. (1983). Foraging Patterns and Their Multiplier Effects on Productivity of Northern
Ungulates. Oikos 40:377–384.
Wold, H. (1966). Estimation of principal components and related models by iterative least
squares. Dans Multivariate Analysis, P. R. Krishnaiaah, éd. (New-York: Academic Press).
122