Ecology, 95(2), 2014, pp. 495–504
Ó 2014 by the Ecological Society of America
Environmental context influences both the intensity of seed
predation and plant demographic sensitivity to attack
TOVE
2
VON
EULER,1 JON ÅGREN,2
AND
JOHAN EHRLÉN1,3
1
Department of Ecology, Environment and Plant Sciences, Stockholm University, SE-106 91 Stockholm, Sweden
Plant Ecology and Evolution, Department of Ecology and Genetics, Evolutionary Biology Centre, Uppsala University,
Norbyvägen 18 D, SE-752 36 Uppsala, Sweden
Abstract. Variation in mutualistic and antagonistic interactions are important sources of
variation in population dynamics and natural selection. Environmental heterogeneity can
influence the outcome of interactions by affecting the intensity of interactions, but also by
affecting the demography of the populations involved. However, little is known about the
relative importance of environmental effects on interaction intensities and demographic
sensitivity for variation in population growth rates. We investigated how soil depth, soil
moisture, soil nutrient composition, and vegetation height influenced the intensity of seed
predation as well as host plant demography and sensitivity to seed predation in the perennial
herb Primula farinosa. Intensity of seed predation ranged from 0% to 80% of seeds damaged
among the 24 study populations and was related to soil moisture in two of four years. The
effect of seed predation on plant population growth rate (k) ranged from negligible to a
reduction in k by 0.70. Sensitivity of population growth rate to predation explained as much of
the variation in the reductions in population growth rate due to seed predation as did
predation intensity. Plant population growth rate in the absence of seed predation and
sensitivity to predation were negatively related to soil depth and soil moisture. Both intensity
of predation and sensitivity to predation were positively correlated with potential population
growth rate and, as a result, there was no significant relationship between predation intensity
and realized population growth rate. We conclude that in our study system environmental
context influences the effects of seed predation on plant fitness and population dynamics in
two important ways: through variation in interaction intensity and through sensitivity to the
effects of this interaction. Moreover, our results show that a given abiotic factor can influence
population growth rate in different directions through effects on potential growth rate,
intensity of biotic interactions, and the sensitivity of population growth rate to interactions.
Key words: biotic interactions; demography; environmental context; integral projection models; plant–
herbivore interactions; population dynamics; Primula farinosa; sensitivity.
INTRODUCTION
Variation in the intensity of mutualistic and antagonistic interactions constitutes an important source of
variation in population dynamics, natural selection, and
coevolutionary dynamics (e.g., Thompson 2005, Maron
and Crone 2006, Ågren et al. 2008, Gómez et al. 2009,
Armbruster et al. 2010, Vanhoenacker et al. 2013).
However, we have yet to understand fully why effects
vary over time and space, and to determine when and
where effects on fitness and population dynamics are
weak or strong. One important component of such an
understanding is the realization that the effects of
interactions on fitness and population growth rates
depend not only on the intensity of interactions, but also
on the demography of the interacting populations. The
demography determines how sensitive total fitness of an
organism is to interaction-driven changes in components
Manuscript received 19 March 2013; revised 5 July 2013;
accepted 11 July 2013. Corresponding Editor: J. Weiner.
3 Corresponding author. E-mail: [email protected]
in fitness (Caswell 2001, Kolb et al. 2007b, Horvitz et al.
2010, Maron et al. 2010). Variation in sensitivities
should weaken the relationship between interaction
intensity and effects on total fitness and population
growth. These relationships can be further weakened if
interaction intensities or sensitivities are positively
correlated with population growth rate in the absence
of the focal interaction. To understand how environmental context influences the outcome of interactions
and to understand the causes of variation in population
dynamics and selection caused by biotic interactions, it
is thus necessary to simultaneously explore links
between environmental factors and interaction intensities and links between environmental factors and
demography.
In plants, antagonistic interactions such as herbivory
and seed predation may have important effects on
population growth rate in some systems (e.g., Crawley
1989, Louda and Potvin 1995, Rose et al. 2005, Kolb et
al. 2007b), but weak or no effects in other systems
(Ehrlén 1996, Fröborg and Eriksson 2003, Weppler and
495
496
TOVE VON EULER ET AL.
Stoecklin 2006). Also, variation in intensity of interactions between the same pair of species is common, and
has been linked to variation in environmental variables
such as altitude (Louda 1982, Rodriguez et al. 1994) and
canopy cover (Louda and Rodman 1996, Leimu et al.
2002, Arvanitis et al. 2007), as well as to global
environmental change (Tylianakis et al. 2008). Variation
in interaction intensities linked to environmental variation has in some cases also been shown to translate into
variation in the effects on population growth rate (Kolb
et al. 2007b, Vandegehuchte et al. 2010). At the same
time, many studies have documented how plant
demography is influenced directly by environmental
factors such as nutrient availability (Gotelli and Ellison
2002, Brys et al. 2005, Dahlgren and Ehrlén 2009),
vegetation height (Lindborg and Ehrlén 2002, QuintanaAscencio et al. 2009, Sletvold et al. 2010), and water
availability (Casper 1996, Eckstein 2005, Schleuning et
al. 2008, Toräng et al. 2010). Such environmental effects
on demography have sometimes been shown to influence
the sensitivity of population growth rate to changes in
the vital rates imposed by biotic interactions (Horvitz
and Schemske 1995, Horvitz et al. 2005, Kolb et al.
2007b, Åberg et al. 2009). For instance, in populations
where seedling recruitment is rare, a reduction in seed
output due to predation might have smaller effects on
population growth rates than in populations where
recruitment is frequent (Turnbull et al. 2000, Maron and
Crone 2006). The sensitivity of population growth rate
to effects of antagonistic interactions on seed production
may often be positively correlated with population
growth rate (Silvertown et al. 1993), influencing the
net relationship between the intensity of the interaction
and host plant population growth rates (Münzbergová
2006, Kolb et al. 2007b). Although environmental effects
on intensity of interactions and dynamics of populations
have been widely studied, we still know little about the
effects of demographic sensitivity on spatial variation in
the consequences of interactions for total plant fitness
and population growth rates.
In this study, we monitored 24 populations of Primula
farinosa during four years to investigate how environmental context influences the impact of pre-dispersal
seed predation on population growth rate (k). The seed
predator, a small tortricid moth, Falseuncaria ruficiliana,
feeds on the developing seeds, and sometimes causes
considerable reductions in seed production (Ehrlén et al.
2002, Vanhoenacker et al. 2009). The study sites varied
in soil depth, soil moisture, soil nutrient availability, and
vegetation height. We used integral projection models
(Easterling et al. 2000, Ellner and Rees 2006) combined
with lasso regressions (Tibshirani 1996) to study the
relationship between environmental factors, intensity of
seed predation, and demographic parameters such as
potential (in the absence of seed predation) and realized
(in the presence of seed predation) population growth
rates, and the sensitivity of population growth rate to
reductions in seed production. Specifically, we asked (1)
Ecology Vol. 95, No. 2
How much does predation intensity vary among
populations? (2) How does pre-dispersal seed predation
affect population growth rates? (3) What is the relative
importance of variation in predation intensity vs.
sensitivity to predation for effects of predation on
population growth rate? (4) How do environmental
factors influence intensity of seed predation, potential
population growth rate and sensitivity of population
growth rate to seed predation? and (5) Are predation
intensity, sensitivity to predation and population growth
rate in the absence of predation positively correlated,
thus weakening the relationship between predation and
population growth rate?
MATERIALS
AND
METHODS
Study system
Primula farinosa (see Plate 1) is a small, rosetteforming, outcrossing, perennial herb, native to northern
Europe (Tutin 1972). It is favored by grazing and occurs
in moist calcareous grasslands (Hambler and Dixon
2003). The flowers are arranged in an umbel-like
inflorescence and each individual usually produces 3–
20 flowers. Primula farinosa is pollinated by butterflies
and solitary bees. Fruits and seeds are sometimes
attacked by the moth Falseuncaria ruficiliana, which
oviposits in the developing fruit, usually resulting in all
seeds being consumed by the larvae (Ehrlén et al. 2002,
Vanhoenacker et al. 2009). Seeds may germinate directly
or remain in a seed bank for one or more years (Toräng
et al. 2010). In the study area, P. farinosa is dimorphic
for scape length and occurs in a long-scaped (3–20 cm)
and a short-scaped (0–3 cm) morph. The two scape
morphs differ in reproductive traits, pollen limitation,
risk of seed predation, and grazing damage (Ehrlén et al.
2002, Ågren et al. 2006, 2013, Vanhoenacker et al. 2009),
but as far as known, not in any other demographic
parameters. The study populations were situated in the
northern part of the great Alvar on the island Öland off
the southeast cost of Sweden. The study area is
characterized by shallow, calcareous soils, and water
levels are known to fluctuate between seasons and years.
Large parts of the area are grazed by cattle, sheep, and
horses (Ekstam 2002). The study sites vary in soil depth,
soil nutrient availability, and grazing intensity, and
therefore in water availability and vegetation height
(Appendix). Where soil moisture is low, vegetation is
usually short, whereas, at sites with high soil moisture,
vegetation may be short or tall depending on the
intensity of grazing.
Data collection
In 2007, 2–17 (depending on plant density) permanent 1 3 1 m plots were established in each of 24 P.
farinosa populations. At the start of the demographic
monitoring, a minimum of 200 individuals were
mapped in each population. Each year, from 2007 to
2010, we recorded survival, vegetative size (rosette
diameter), number of flowers, and number of intact
February 2014
ENVIRONMENT, DEMOGRAPHY, AND PREDATION
fruits, as well as the occurrence of grazing damage and
the number of fruits consumed by seed predators for all
mapped individuals. We used number of intact fruits 3
mean number of seeds per intact fruit as an estimate of
total seed production. We calculated the mean number
of seeds per intact fruit in 80 flowering individuals in
each of 16 of the 24 study populations in 2000 and
2001. The reason that we were not able to obtain
estimates of the number of seeds per fruit from the
remaining eight populations was that too few individuals with intact fruits were available outside the
permanent plots, and that we did not want to interfere
with the local population dynamics by collecting seeds
within or in the near vicinity of permanent plots. The
population mean values of the number of seeds per
fruit varied relatively little among the 16 populations
(mean ¼ 49.9, SD ¼ 6.6). We therefore used the overall
mean as an estimate of seed number per fruit in all
populations. Only rarely do a few seeds survive fruit
predation, and here we assumed that fruits with moth
larvae produced no viable seeds. Any new individuals
appearing within the plots during the study were
mapped and their status recorded in the following
years. Soil depth and soil moisture were measured
during two consecutive days in July 2010. Soil depth
was measured by pushing a steel rod into the soil until
it hit bedrock. Soil moisture was measured using a
Theta Probe sensor connected to a HH2 moisture meter
(Delta-T Devices, Cambridge, UK). Sixteen measurements of soil depth and soil moisture were taken at
each of the 24 sites where the study populations
occurred, distributed over the area where permanent
plots were located. Mean values of soil depth and soil
moisture from each site were used in the statistical
analyses. Although soil moisture levels differ among
and within years due to differences in temperature and
precipitation, we assumed the rank order among
populations to be fairly similar over time, since they
are related to more permanent, site-specific factors such
as soil depth and topography. Still, it is true that
differences in soil moisture are likely to be largest at
intermediate levels of drought and smaller during very
wet or very dry conditions. Our measurements of soil
moisture were therefore carried out during such
intermediate conditions, three weeks after the most
recent rainfall. For soil nutrient analyses, nine soil
samples were collected at each site in May 2008. The
samples were pooled by site and analyzed for phosphorous, potassium, calcium, and magnesium using the
Al extraction method, and for total amounts of carbon
and nitrogen using elemental analysis. Analyses were
carried out at the Division of Soil Fertility and Plant
Nutrition at the Swedish University of Agricultural
Sciences (SLU), Uppsala, Sweden. Vegetation height
was measured each year in July (2007–2010). At each
site, three measurements were taken in the vicinity of
each of 30 flowering individuals.
497
Integral projection models
We used integral projection models to estimate
population growth rates given attack by seed-feeding
insects (realized population growth rates) and to
estimate what the growth rate of those populations
would be in the absence of predation (potential
population growth rates). Since seed predators destroyed virtually all seeds in attacked fruits, we assumed
the per fruit seed loss to equal the average number of
seeds in un-attacked fruits. This will underestimate
losses caused by seed predation if seed predators
preferentially oviposit in large flowers with many ovules,
and thus result in a conservative estimate of the effect of
seed predation on population growth rate. Integral
projection models are based on fitted functions of vital
rates, which, in turn, depend on state variables, such as
size. This modeling approach is well suited for plant
populations without clearly defined discrete life cycle
stages and allow for statistical approaches to parameterize models (Ellner and Rees 2006). The models were
constructed using data on survival, rosette diameter,
probability of flowering, seed production, and germination rates collected each year in the permanent plots.
Separate models were created for each population and
yearly transition (2007–2008, 2008–2009, and 2009–
2010). Models were based on the following equations:
Z U
SB ðt þ 1Þ ¼ SB ðtÞpssb þ
pf ðxÞfsn ptosb nðx; tÞdx ð1Þ
L
Ss ðt þ 1Þ ¼ SB ðtÞpgsb þ
Z
U
pf ðxÞfsn pgd nðx; tÞdx
ð2Þ
L
nðy; t þ 1Þ ¼ SS ðtÞps pd ðyÞ þ
Z
U
sðxÞgðy; xÞnðx; tÞdx ð3Þ
L
where L and U represent the lower and upper limit of
possible sizes (rosette diameter), x represents size in year
t, and y represents size in year t þ 1. In Eq.1, SB
represents seeds present in the seed bank and is given by
the number of seeds in the seed bank, the probability of
a seed surviving in the seed bank, pssb, and the
probability of flowering in year t, pf (x), the number of
seeds per flowering individual in year t, fsn, and the
probability of a seed entering the seed bank, ptosb. In Eq.
2, SS represents the number of seedlings and is
composed of the number of seeds in the seed bank
multiplied by the probability of germinating from the
seed bank, pgsb, the probability of flowering in year t,
pf (x), the number of seeds per flowering individual in
year t, fsn and the probability of a seed germinating
directly, pgd. In Eq. 3, n(y, t þ 1) is the size distribution
and is given by SS multiplied by the probability of
seedling survival, ps, and seedling size distribution, given
by a normal distribution, pd, and the size distribution of
plants in the previous year (n(x, t)), the probability of
survival of established plants from year t to year t þ 1 as
a function of rosette diameter in year t, and a growth
498
TOVE VON EULER ET AL.
Ecology Vol. 95, No. 2
PLATE 1. Dense population of Primula farinosa. Photo credit: J. Ågren.
function, g(y, x), giving the probability of a size x
individual attaining size y, based on a regression of
rosette diameter at time t þ 1 on rosette diameter at time
t in established plants. Because no specific information
about vital rates of seeds in the soil was available from
our recordings, we used estimated seed bank survival
from an earlier demographic study on this system
(Toräng et al. 2010). We assumed that the probability
of germinating from the seed bank was similar to the
probability of newly produced seeds germinating directly (changing this assumption had negligible effects on
the results). R code used for combining Eqs. 1–3 into a
matrix describing the transition kernel was based on
code provided in Ellner and Rees (2006). (For a
description of the code used in the models, see the
Supplement). The asymptotic growth rate (k) was
calculated as the dominant right eigenvalue of the
matrix.
Intensity of predation was quantified as the mean
proportion of fruits consumed per individual within
each population. This estimate was strongly correlated
with the proportion of the total seed production that
was consumed within a population (r ¼ 0.94–0.97 during
the four study years). The effect of predation on
population growth rate was quantified as the difference
between potential and realized asymptotic population
growth rate, Dk. We defined sensitivity of population
growth rate to predation as Dk divided by the mean
proportion of fruits consumed. This was because we
were interested in the sensitivity integrated over the
entire interval from observed levels of seed predation to
no predation rather than point estimates of sensitivity at
a specific level of predation.
Statistical analyses
To examine relationships between environmental
variables and predation intensity, potential k and
sensitivity to predation during the three transition
intervals, we used lasso regressions (R package lars).
The lasso regression (least absolute shrinkage and
selection operator) is a shrinkage method, where a cap
(maximum value) is put on the sum of the absolute
values of all regression coefficients and variable selection
is achieved by shrinking coefficients for the variables
that contribute least to the model (Tibshirani 1996).
Non-zero coefficients are retained in the model, whereas
coefficients shrunk to zero are left out. This method
largely avoids over-fitting, which is often a problem
when dealing with small data sets (Dahlgren 2010). The
environmental variables used in the lasso regressions
were soil depth, soil moisture, and soil nutrient
composition and vegetation height. Because vegetation
height was highly correlated between years, means of
yearly population averages were used for analyses. To
identify main axes of variation in soil concentrations of
nutrients, we performed principle component analysis
February 2014
ENVIRONMENT, DEMOGRAPHY, AND PREDATION
(PCA), based on the correlation matrix, and used the
two main axes for further analyses. The first axis (PC1)
explained 53% of the variation and mainly represented
soil nitrogen and potassium concentrations. The second
axis (PC2) explained 24% of the variation and mainly
represented calcium and magnesium concentrations.
Previous studies with this system have shown that the
incidence and intensity of predation may be related to
population size and the relative frequency of the longscaped morph (Vanhoenacker et al. 2009). In the
populations included in this study, however, predation
intensity was not related to plant population size or
scape morph frequencies, and those variables were thus
not included as predictors in the models.
To explore relationships among demographic parameters, intensity of predation, and effects of predation on
population growth rate, we plotted the pairwise
relationships between potential population growth rate
(potential k), predation intensity, realized population
growth rate (realized k), Dk, and the sensitivity of
population growth rate to predation in each of the three
transition intervals (2007–2008, 2008–2009, 2009–2010).
To summarize our results and investigate the different
ways in which environmental context influenced antagonistic interactions and their effects on plant population
growth rates, we used a path diagram. For this, the
coefficients of the lasso regressions of potential k,
predation intensity and sensitivity to predation on the
environmental variables were combined with linear
regression coefficients between predation intensity and
Dk, as well as between sensitivity to predation and Dk.
In the path diagram, the relationship between potential
and realized k was set to 1 and the relationship between
Dk and realized k to 1 (cf. Conner 1996). Lastly, to
examine the net relationship between seed predation and
population growth rate we calculated the correlation
between these two variables. All statistical analyses were
performed in R 2.11.1 (R Development Core Team
2011).
RESULTS
Seed predation intensity varied considerably among
populations and years (two-way ANOVA, population,
F23,68 ¼ 4.9, P , 0.001; year, F3,68 ¼ 3.5, P ¼ 0.019), and
so did potential and realized population growth rate.
The proportion of fruits damaged by seed predators
ranged from 0 to 0.84 (0.31 6 0.22 [mean 6 SD]) and
predation was absent in only 7.3% of the population by
year combinations (N ¼ 72 combinations). Potential
population growth rate, in the absence of seed predation, ranged from 0.60 to 2.36 (1.15 6 0.25), whereas
realized population growth rate ranged from 0.60 to 1.66
(1.07 6 0.16; Fig. 1). Reductions in population growth
rate due to seed predation (Dk) ranged from 0 to 0.70
(0.08 6 0.12), and increased with increasing intensity of
seed predation. This relationship appeared to be
nonlinear with very small reductions at low predation
499
FIG. 1. Variation in potential population growth rates (k)
(in the absence of seed predation) and realized population
growth rates (k) (in the presence of seed predation) of 24
populations of Primula farinosa during three years of demographic monitoring.
intensities and very large reductions at high intensities
(Fig. 2a).
Variation in the reduction in population growth rate
caused by predation was not only the result of
differences in the intensity of seed predation, but also
of differences in the sensitivity of population growth
rates to seed predation. The sensitivity of population
growth rates to seed predation ranged from 0 to 0.98
(0.20 6 0.17). Sensitivity to predation explained as much
of the variation in Dk (56–97, N ¼ 3 transitions) as did
predation intensity (52–85%). Both predation intensity
and sensitivity to seed predation were also positively
correlated with potential k (Fig. 2e and f ).
Lasso regression and path analyses indicated that
environmental variables affect realized k in several ways:
by influencing potential k, predation intensity, and
sensitivity to predation. Effects of soil depth and soil
moisture were recorded in multiple years, whereas
effects of principal components reflecting soil nutrient
composition were only observed in single years, and in
500
TOVE VON EULER ET AL.
Ecology Vol. 95, No. 2
FIG. 2. Pairwise correlations between potential population growth rate (k), predation intensity, difference between potential
and realized population growth rate (Dk), and sensitivity of k to predation: (a) Dk vs. predation intensity (2007–2008, r ¼ 0.92, P ,
0.001; 2008–2009, r ¼ 0.72, P , 0.001; 2009–2010, r ¼ 0.86, P , 0.001), (b) Dk vs. sensitivity (2007–2008, r ¼ 0.75, P , 0.001; 2008–
2009, r ¼ 0.98, P , 0.001; 2009–2010, r ¼ 0.93, P , 0.001), (c) predation intensity vs. sensitivity (2007–2008, r ¼ 0.59, P ¼ 0.005;
2008–2009, r ¼ 0.64, P ¼ 0.002; 2009–2010, r ¼ 0.69, P ¼ 0.001), (d) Dk vs. potential k (2007–2008, r ¼ 0.67, P , 0.001; 2008–2009, r
¼ 0.94, P , 0.001; 2009–2010, r ¼ 0.83, P , 0.001), (e) predation intensity vs. potential k (2007–2008, r ¼ 0.47, P ¼ 0.02; 2008–2009,
r ¼ 0.55, P ¼ 0.006; 2009–2010, r ¼ 0.62, P ¼ 0.001), and (f ) sensitivity vs. potential k (2007–2008, r ¼ 0.87, P , 0.001; 2008–2009, r ¼
0.97, P , 0.001; 2009–2010, r ¼ 0.90, P , 0.001).
no year was a significant effect of vegetation height
recorded (Table 1, Fig. 3). Potential population growth
rate was negatively related to soil depth in 2007–2008,
and to soil moisture in 2008–2009 and 2009–2010.
Predation intensity was negatively related to soil depth
in 2009 and to soil moisture in 2009 and 2010 (Table 1).
Finally, sensitivity to seed predation was negatively
related to soil depth in 2007–2008 and 2009–2010, and
to soil moisture in 2008–2009 and 2009–2010. Thus,
both increased soil depth and increased soil moisture
affected k negatively via a decreased potential k, but
positively via a decreased predation intensity and
decreased sensitivity to predation.
In all three transition intervals, both predation
intensity and sensitivity to predation were positively
correlated with potential population growth rate (Fig. 2e
and f ), and as a result there was no significant
relationship between predation intensity and realized k
(Fig. 4).
DISCUSSION
The results of this study provide a clear example of
how environmental context influences the effects of
antagonistic interactions on host plant population
growth rate in multiple ways. Intensity of pre-dispersal
seed predation, potential population growth rate, and
sensitivity of population growth rate to seed predation
varied among Primula farinosa populations and were
significantly related to environmental factors, such as
soil moisture and soil depth. The reduction in popula-
tion growth rate caused by predation ranged from
negligible to very large, and was strongly correlated with
both sensitivity to predation and predation intensity.
Because both high predation intensity and high sensiTABLE 1. Lasso regressions of the effects of soil depth, soil
moisture, soil nutrient availability (PC1, nitrogen and
potassium, and PC2, calcium and magnesium) and vegetation height on pre-dispersal seed predation, potential
population growth rate, and sensitivity of population growth
rate to pre-dispersal seed predation in Primula farinosa (N ¼
24 populations).
Soil
Soil
Time period depth moisture PC1 PC2 R 2 df Rss
Predation
2007
2008
2009
2010
Potential k
2007–2008
2008–2009
2009–2010
Sensitivity
2007–2008
2008–2009
2009–2010
0.10
0.12
0.34
0.28
0.86 2.08
1.27 1.65
0.75 3.36
1.00 1.37
þ
0.32 3 0.43 1.65
0.34 2 1.38 1.25
0.59 2 0.65 0.92
þ
0.34 3 0.12 1.27
0.50 2 0.46 0.58
0.52 4 0.32 3.35
1
1
3
3
Cp
Notes: Minus signs () denote negative effects, and plus signs
(þ) denote positive effects of the respective predictor variable in
the respective year. Effects of vegetation height were never
included in final models. Rss stands for residual sum of squares.
Cp is Mallow’s Cp, a criterion to assess fits when models with
different numbers of parameters are being compared.
February 2014
ENVIRONMENT, DEMOGRAPHY, AND PREDATION
501
FIG. 3. Path diagram showing the effects of environmental factors on potential population growth rate (potential k, without
the effects of seed predation), predation intensity, and the sensitivity of population growth rate to seed predation, and the effects of
predation intensity and sensitivity to predation on Dk (the difference between potential and realized population growth rate), and
the effects of potential k and Dk on realized population growth rate (k, after the effects of seed predation) in 24 populations of
Primula farinosa during three transition intervals. Effects of environmental factors were estimated by coefficients from lasso
regressions. Effects of predation intensity and sensitivity to predation on Dk were estimated by linear regression coefficients. The
relationship between potential and realized k was set to 1, and the relationship between Dk and realized k to 1.
tivity to predation were associated with high potential
population growth rate, reductions in lambda were
highest at high population growth rates, and variation in
realized population growth rate was markedly lower
than variation in potential growth rate. As a consequence, realized population growth rate was not
significantly related to predation intensity despite the
fact that predation sometimes caused large reductions in
population growth rate.
The proportion of fruits destroyed by seed predation
varied greatly among populations, from 0% to 84%.
Seed predation intensity increased with decreasing soil
502
TOVE VON EULER ET AL.
FIG. 4. Correlations between predation intensity and
realized population growth rate in 24 populations of Primula
farinosa during three years of demographic recordings (2007–
2008, r ¼ 0.24, P ¼ 0.27; 2008–2009, r ¼ 0.32, P ¼ 0.13; 2009–
2010, r ¼ 0.26, P ¼ 0.22).
moisture in two years. This may either be because soil
moisture or unmeasured environmental variables associated with soil moisture, such as temperature, affect the
abundance of the seed predator directly or because they
affect flowering phenology or other characteristics of the
host plant population influencing attractiveness to the
seed predator. Many previous studies have documented
large variation in the intensity of seed predation and
herbivory, and linked this variation to variation in
environmental factors (Louda 1982, Rodriguez et al.
1994, Louda and Rodman 1996, Leimu et al. 2002,
Arvanitis et al. 2007). Environmental context thus seems
to have important effects on the intensity of antagonistic
interactions in many natural systems.
The impact of predation on population growth rate,
i.e., the difference between potential and realized
population growth rate ranged from 0 to 0.70 in our
study (mean 0.08; Fig. 2). Relatively few studies have
previously assessed the effects of pre-dispersal seed
predation on the population growth rates of perennial
plants and these studies have only very rarely found
reductions in growth rates of 0.1 or more (Ehrlén 1995,
Kolb et al. 2007a, Maron et al. 2010). The effects of seed
predation on population growth rates of P. farinosa
were thus in some cases very large compared to those
observed in previously studied systems.
The effect of predation on population growth rate of
P. farinosa was positively related to both predation
intensity and sensitivity to predation. Reductions in
population growth rate due to seed predation increased
with increasing intensity of seed predation. The apparent nonlinearity of this relationship can at least partly be
attributed to the positive correlation between predation
intensity and sensitivity of population growth to
predation. Overall, sensitivity explained as much of the
variation in Dk as did predation intensity. This suggests
that, in this system, the outcome of the host-plant–seed
predator interaction is as much affected by the
Ecology Vol. 95, No. 2
demographic properties of the host plant population as
by variation in the amount of damage caused by the seed
predator. Variation in the sensitivity to predation has
been found to have important effects on host plant
population growth rates also in other systems (Ehrlén
1996, Münzbergová 2006, Kolb et al. 2007b). In Primula
veris, predation intensity increased with canopy cover,
while the sensitivity of population growth rate to
predation decreased, causing a nonlinear relationship
between seed predation and population growth rate
(Kolb et al. 2007b). Taken together, the results of this
and previous studies suggest that not only the intensity
of biotic interactions but also variation in demography
and sensitivity to interaction-governed variation in vital
rates are important for the outcome of interactions. This
implies that differences in interaction-mediated selection
can be the result both of differences in interaction
intensity and of differences in demography, and that
variation in the life histories of interacting organisms by
themselves can result in spatial variation in coevolutionary processes (cf. Thompson 2005).
In the present study, both estimated population
growth rate in the absence of seed predation and
sensitivity of population growth rate to predation varied
among populations and this variation was partly
explained by differences in soil depth and soil moisture.
Moreover, variation in the sensitivity of population
growth rate to predation was positively correlated with
potential population growth rates, suggesting that
populations are more sensitive to seed loss under more
favorable conditions. High sensitivity to changes in
fecundity at high population growth rates has been
found in several previous studies (Silvertown et al. 1993,
Horvitz and Schemske 1995, Menges and Dolan 1998,
Valverde and Silvertown 1998, Ramula et al. 2008). This
relationship is expected because high population growth
rates in non-clonal species necessarily involve a high rate
of recruitment. In the present study, also the intensity of
seed predation was positively correlated with potential
growth rate. Because both the intensity of predation and
the sensitivity of population growth rate to predation
were positively correlated with potential population
growth rates, realized population growth rate varied less
than did potential growth rate, and was not correlated
with predation intensity. This demonstrates that strong
effects of biotic interactions on population growth rates
cannot always be detected by examining only the
correlation between interaction intensities and realized
population growth rates. This is because the effect of
variation in seed predation on population growth rate is
contingent on the sensitivity to seed predation and
variation in these two variables may partly cancel.
Taken together, the results of this study show that the
extent to which the population growth rate is reduced by
antagonists may strongly depend on the sensitivity of
population growth rate to losses in seed production.
Moreover, environmental factors may not only influence
the intensity of antagonistic interactions, but also how
February 2014
ENVIRONMENT, DEMOGRAPHY, AND PREDATION
sensitive populations are to interactions with antagonists. A general conclusion is thus that environmental
context can influence the effects of biotic interactions on
plant fitness in multiple ways and that effects through
several pathways may be important for variation in
population dynamics.
ACKNOWLEDGMENTS
We thank Johanne Eriksson, Susanne Govella, Veronika
Johansson, Anna Knöppel, Mikiko Skoglund, Lina Sundkvist,
and Cecilia Åldemo for help with data collection, and Johan
Dahlgren for comments on an earlier version of the manuscript.
The research was supported by grants from FORMAS and the
Swedish Research Council to J. Ågren and J. Ehrlén.
LITERATURE CITED
Åberg, P., C. J. Svensson, H. Caswell, and H. Pavia. 2009.
Environment-specific elasticity and sensitivity analysis of the
stochastic growth rate. Ecological Modelling 220:605–610.
Ågren, J., J. Ehrlén, and C. Solbreck. 2008. Spatio-temporal
variation in fruit production and seed predation in a
perennial herb influenced by habitat quality and population
size. Journal of Ecology 96:334–345.
Ågren, J., C. Fortunel, and J. Ehrlén. 2006. Selection on floral
display in insect-pollinated Primula farinosa: effects of
vegetation height and litter accumulation. Oecologia 150:
225–232.
Ågren, J., F. Hellström, P. Toräng, and J. Ehrlén. 2013.
Mutualists and antagonists drive among-population variation in selection and evolution of floral display in a perennial
herb. Proceedings of the National Academy of Sciences USA
110:18202–18207.
Armbruster, W. S., J. Lee, and B. G. Baldwin. 2010.
Macroevolutionary patterns of defense and pollination in
Dalechampia vines: adaptation, exaptation, and evolutionary
novelty. Proceedings of the National Academy of Sciences
USA 106:18085–18090.
Arvanitis, L., C. Wiklund, and J. Ehrlén. 2007. Butterfly seed
predation: effects of landscape characteristics, plant ploidy
level and population structure. Oecologia 152:275–285.
Brys, R., H. Jacquemyn, P. Endels, G. De Blust, and M.
Hermy. 2005. Effect of habitat deterioration on population
dynamics and extinction risks in a previously common
perennial. Conservation Biology 19:1633–1643.
Casper, B. B. 1996. Demographic consequences of drought in
the herbaceous perennial Cryptantha flava: effects of density,
associations with shrubs, and plant size. Oecologia 106:144–
152.
Caswell, H. 2001. Matrix population models: construction,
analysis, and interpretation. Second edition. Sinauer Associates, Sunderland, Massachusetts, USA.
Conner, J. K. 1996. Understanding natural selection: an
approach integrating selection gradients, multiplicative fitness components, and path analysis. Ethology, Ecology and
Evolution 8:387–397.
Crawley, M. J. 1989. Insect herbivores and plant-population
dynamics. Annual Review of Entomology 34:531–564.
Dahlgren, J. P. 2010. Alternative regression methods are not
considered in Murtaugh (2009) or by ecologists in general.
Ecology Letters 13:E7–E9.
Dahlgren, J. P., and J. Ehrlén. 2009. Linking environmental
variation to population dynamics of a forest herb. Journal of
Ecology 97:666–674.
Easterling, M., S. Ellner, and P. Dixon. 2000. Size-specific
sensitivity: applying a new structured population model.
Ecology 81:694–708.
Eckstein, R. L. 2005. Differential effects of interspecific
interactions and water availability on survival, growth and
503
fecundity of three congeneric grassland herbs. New Phytologist 166:525–536.
Ehrlén, J. 1995. Demography of the perennial herb Lathyrus
vernus. 2. Herbivory and population-dynamics. Journal of
Ecology 83:297–308.
Ehrlén, J. 1996. Spatiotemporal variation in predispersal seed
predation intensity. Oecologia 108:708–713.
Ehrlén, J., S. Käck, and J. Ågren. 2002. Pollen limitation, seed
predation and scape length in Primula farinosa. Oikos 97:45–
51.
Ekstam, U. 2002. Svenska Alvarmarker: historia och ekologi.
Naturvårdsverket förl, Stockholm, Sweden.
Ellner, S. P., and M. Rees. 2006. Integral projection models for
species with complex demography. American Naturalist 167:
410–428.
Fröborg, H., and O. Eriksson. 2003. Predispersal seed
predation and population dynamics in the perennial understorey herb Actaea spicata. Canadian Journal of Botany 81:
1058–1069.
Gómez, J., F. Perfectti, J. Bosch, and J. P. M. Camacho. 2009.
A geographic selection mosaic in a generalized plantpollinator–herbivore system. Ecological Monographs 79:
245–263.
Gotelli, N. J., and A. M. Ellison. 2002. Nitrogen deposition and
extinction risk in the northern pitcher plant, Sarracenia
purpurea. Ecology 83:2758–2765.
Hambler, D. J., and J. M. Dixon. 2003. Primula farinosa L.
Journal of Ecology 91:694–705.
Horvitz, C. C., J. Ehrlén, and D. Matlaga. 2010. Contextdependent pollinator limitation in stochastic environments:
Can increased seed set overpower the cost of reproduction in
an understorey herb? Journal of Ecology 98:268–278.
Horvitz, C. C., and D. W. Schemske. 1995. Spatiotemporal
variation in demographic transitions of a tropical understory
herb: projection matrix analysis. Ecological Monographs 65:
155–192.
Horvitz, C. C., S. Tuljapurkar, and J. B. Pascarella. 2005.
Plant–animal interactions in random environments: habitatstage elasticity, seed predators, and hurricanes. Ecology 86:
3312–3322.
Kolb, A., J. Ehrlén, and O. Eriksson. 2007a. Ecological and
evolutionary consequences of spatial and temporal variation
in pre-dispersal seed predation. Perspectives in Plant
Ecology, Evolution and Systematics 9:79–100.
Kolb, A., R. Leimu, and J. Ehrlén. 2007b. Environmental
context influences the outcome of a plant–seed predator
interaction. Oikos 116:864–872.
Leimu, R., K. Syrjänen, J. Ehrlén, and K. Lehtilä. 2002. Predispersal seed predation in Primula veris: among-population
variation in damage intensity and selection on flower
number. Oecologia 133:510–516.
Lindborg, R., and J. Ehrlén. 2002. Evaluating the extinction
risk of a perennial herb: demographic data versus historical
records. Conservation Biology 16:683–690.
Louda, S. M. 1982. Distribution ecology: variation in plant
recruitment over a gradient in relation to insect seed
predation. Ecological Monographs 52:25–41.
Louda, S. M., and M. A. Potvin. 1995. Effect of inflorescencefeeding insects on the demography and lifetime fitness of a
native plant. Ecology 76:229–245.
Louda, S. M., and J. E. Rodman. 1996. Insect herbivory as a
major factor in the shade distribution of a native crucifer
(Cardamine cordifolia A. Gray, bittercress). Journal of
Ecology 84:229–237.
Maron, J. L., and E. Crone. 2006. Herbivory: effects on plant
abundance, distribution and population growth. Proceedings
of the Royal Society B 273:2575–2584.
Maron, J. L., C. C. Horvitz, and J. L. Williams. 2010. Using
experiments, demography and population models to estimate
504
TOVE VON EULER ET AL.
interaction strength based on transient and asymptotic
dynamics. Journal of Ecology 98:290–301.
Menges, E. S., and R. W. Dolan. 1998. Demographic viability
of populations of Silene regia in Midwestern prairies:
relationships with fire management, genetic variation, geographic location, population size and isolation. Journal of
Ecology 86:63–78.
Münzbergová, Z. 2006. Ploidy level interacts with population
size and habitat conditions to determine the degree of
herbivory damage in plant populations. Oikos 115:443–452.
Quintana-Ascencio, P. F., I. Caballero, J. M. Olano, A.
Escudero, and M. J. Albert. 2009. Does habitat structure
matter? Spatially explicit population modelling of an Iberian
gypsum endemic. Population Ecology 51:317–328.
R Development Core Team. 2011. R: a language and
environment for statistical computing. R Project for Statistical Computing, Vienna, Austria. http://www.R-project.org
Ramula, S., T. M. Knight, J. H. Burns, and Y. M. Buckley.
2008. General guidelines for invasive plant management
based on comparative demography of invasive and native
plant populations. Journal of Applied Ecology 45:1124–
1133.
Rodriguez, J., D. Jordano, and J. F. Haeger. 1994. Spatial
heterogeneity in a butterfly–host plant interaction. Journal of
Animal Ecology 63:31–38.
Rose, K. E., S. M. Louda, and M. Rees. 2005. Demographic
and evolutionary impacts of native and invasive insect
herbivores on Cirsium canescens. Ecology 86:453–465.
Schleuning, M., V. Huaman, and D. Matthies. 2008. Flooding
and canopy dynamics shape the demography of a clonal
Amazon understorey herb. Journal of Ecology 96:1045–1055.
Silvertown, J., M. Franco, I. Pisanty, and A. Mendoza. 1993.
Comparative plant demography—relative importance of lifecycle components to the finite rate of increase in woody and
herbaceous perennials. Journal of Ecology 81:465–476.
Sletvold, N., D.-I. Øien, and A. Moen. 2010. Long-term
influence of mowing on population dynamics in the rare
Ecology Vol. 95, No. 2
orchid Dactylorhiza lapponica: the importance of recruitment
and seed production. Biological Conservation 143:747–755.
Thompson, J. N. 2005. The geographic mosaic of coevolution.
University of Chicago Press, Chicago, Illinois, USA.
Tibshirani, R. 1996. Regression shrinkage and selection via the
lasso. Journal of the Royal Statistical Society B 58:267–288.
Toräng, P., J. Ehrlén, and J. Ågren. 2010. Linking environmental and demographic data to predict future population
viability of a perennial herb. Oecologia 163:99–109.
Turnbull, L. A., M. J. Crawley, and M. Rees. 2000. Are plant
populations seed-limited? A review of seed sowing experiments. Oikos 88:225–238.
Tutin, T. G. 1972. Flora Europaea. Volume 3, Diapensiaceae to
Myoporaceae. Cambridge University Press, Cambridge, UK.
Tylianakis, J. M., R. K. Didham, J. Bascompte, and D. A.
Wardle. 2008. Global change and species interactions in
terrestrial ecosystems. Ecology Letters 11:1351–1363.
Valverde, T., and J. Silvertown. 1998. Variation in the
demography of a woodland understorey herb (Primula
vulgaris) along the forest regeneration cycle: projection
matrix analysis. Journal of Ecology 86:545–562.
Vandegehuchte, M. L., E. de la Pena, and D. Bonte. 2010.
Relative importance of biotic and abiotic soil components to
plant growth and insect herbivore population dynamics.
PLoS ONE 5:e12937.
Vanhoenacker, D., J. Ågren, and J. Ehrlén. 2009. Spatial
variability in seed predation in Primula farinosa: local
population legacy versus patch selection. Oecologia 160:77–
86.
Vanhoenacker, D., J. Ågren, and J. Ehrlén. 2013. Non-linear
relationship between intensity of plant–animal interactions
and selection strength. Ecology Letters 16:198–205.
Weppler, T., and J. Stoecklin. 2006. Does pre-dispersal seed
predation limit reproduction and population growth in the
alpine clonal plant Geum reptans? Plant Ecology 187:277–
287.
SUPPLEMENTAL MATERIAL
Appendix
Population mean soil depth (cm), soil moisture (%), PC1 (N, P, K), PC2 (Ca, Mg) and vegetation height (cm) in 24 populations
of Primula farinosa, measured in 2008–2010 (Ecological Archives E095-044-A1).
Supplement
R code used to estimate potential and realized population growth rates of 24 populations of Primula farinosa (Ecological
Archives E095-044-S1).