RAPID COMMUNICATIONS IN MASS SPECTROMETRY
Rapid Commun. Mass Spectrom. 2008; 22: 1673–1680
Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/rcm.3497
Application of nitrogen stable isotope analysis in
size-based marine food web and macroecological
researchy
Simon Jennings1*, Carolyn Barnes1, Christopher J. Sweeting2 and
Nicholas V. C. Polunin3
1
Centre for Environment, Fisheries and Aquaculture Science, Pakefield Road, Lowestoft NR33 0HT, UK
CNR-IAMC, Laboratorio di Ecologia Marina, via G. Da Verrazzano 17, 91014 Castellammare del Golfo (TP), Italy
3
Department of Marine Science and Technology, University of Newcastle upon Tyne NE1 7RU, UK
2
Received 8 February 2008; Accepted 11 February 2008
Interacting human and environmental pressures influence the structure and dynamics of marine food
webs. To describe and predict the effects of these pressures, theoretical advances need to be
supported by a capacity to validate the underlying models and assumptions. Here, we review recent
applications of nitrogen stable isotope analysis in marine food web and macroecological research,
with a focus on work that has paralleled a resurgence of interest in the development and application
of size-based models. Nitrogen stable isotope data have been used to estimate intra- and inter-specific
variation in trophic level, predator-prey size ratios, transfer efficiency, food chain length, relationships between predator and prey species diversity and the dynamics of energy use. Many of these
estimates have contributed to the development, testing and parameterisation of food web and
ecosystem models, some of which have been used to establish baselines for assessing the scale
of human impacts. The interpretation of results depends on assumed fractionation but, when
supported by sensitivity analyses and experimental validation, nitrogen stable isotope data provide
valuable insights into the structuring of marine communities and ecosystems. Copyright # 2008 John
Wiley & Sons, Ltd.
Nitrogen stable isotope analysis provides a powerful tool for
ecologists, giving new insight into the origins and transformations of organic matter and the migrations of animals in
terrestrial and aquatic environments.1–3 This tool has been
used to identify the sources of energy that support
consumers,4 to describe interactions among small groups
of species4 and to investigate migration.5,6 The development
of automated and subsequently more cost-effective methods
for measuring nitrogen stable isotope natural abundance7
has facilitated many of the recent ecological applications of
stable isotope analysis. This is because inherent variability in
life history and ecology among individuals, populations,
communities and ecosystems means that high levels of
replication in space and time are needed to gain sufficient
statistical power to test competing hypotheses.
It is rarely feasible to study the ecology of all species that
comprise food webs in such detail that the results can be used
*Correspondence to: S. Jennings, Centre for Environment, Fisheries
and Aquaculture Science, Pakefield Road, Lowestoft NR33 0HT,
UK.
E-mail: [email protected]
y
Presented at the annual meeting of the Stable Isotopes Mass
Spectrometry Users’ Group (SIMSUG), 20–22 June, 2007, hosted
by the Institute for Research on the Environment and Sustainability (IRES) and the School of Civil Engineering and Geosciences, Newcastle University, UK.
Contract/grant sponsor: Defra; contract/grant number: MF1001.
Contract/grant sponsor: NERC; contract/grant number: NER/
S/A/2003/11886.
to describe community and ecosystem processes. Indeed,
while species are often a focus of food web research at high
trophic levels, low trophic levels are invariably dealt with at
higher levels of aggregation (e.g. phytoplankton or plants).
Given these constraints, an informative process is to assess
the properties of communities and ecosystems in their own
right, irrespective of the identities of component species. In
marine ecology, interest in this approach was stimulated by
the invention of the Coulter counter, which allowed the size
distributions of living particles in the ocean to be described
for the first time.8 The regularity of these distributions
prompted research into the processes responsible,9 and
provided a good example of what was later dubbed the
macroecological approach.10
Here, we focus on recent applications of nitrogen stable
isotope analysis in marine food web and macroecological
research; work that has paralleled the resurgence of interest
in size-based modelling of marine food webs. This is a
necessarily short review to meet the requirements of this
conference volume, and does not cover significant advances
in using stable isotope analysis to study source materials,
pathways and conventional species-based food web structures; as recently reviewed by Fry.2
Size-based analyses of food webs treat body size rather
than species identity as the principal descriptor of an
individual’s role in the food web. In such analyses, a small
individual of a large species is treated as functionally
Copyright # 2008 John Wiley & Sons, Ltd.
1674 S. Jennings et al.
Size-based analysis
A principal assumption of size-based analysis is that body
size accounts for a large proportion of the variance in trophic
level compared with species identity. If this assumption were
not met, then categorisation by species would be necessary,
greatly complicating the analysis of food web structures in
ecosystems that vary widely in species composition. Many
marine food chains are, however, strongly size-based. For
example, analysis of the relative contributions of intra- and
inter-specific differences in trophic level (as indicated by
d15N) to trophic structure14,15 in a North Sea food web
showed that the inter-specific relationships between maximum body mass and d15N at a fixed proportion of that body
mass were weak or non-significant for invertebrates and
fishes. This result was confirmed with a comparative analysis
that accounted for phylogenetic relationships among
species.16 Conversely, when all animals in the communities
were divided into body mass classes without accounting for
species identity, trophic level rose near continuously with
body mass (e.g. Fig. 1).
Additional analyses of intra- and inter-specific relationships between body mass and d15N in fishes and invertebrates have shown that increases in trophic level across the
size spectrum were predominantly a consequence of
intra-specific increases in trophic level with body mass
rather than larger species (species with greater maximum
body mass) feeding at higher trophic levels.15 Thus the
increase in trophic level of individual species with body
size makes an important contribution to the increase in the
trophic level of the community with size (Fig. 2). Analyses of
this type have yet to be done for a complete size spectrum
but, within the size range considered, most of the animals
contributing to the spectrum were included.
Copyright # 2008 John Wiley & Sons, Ltd.
18
(a)
14
15
N (0/00)
16
12
10
4
6
8
10
12
14
16
log2 maximum body mass
18
(b)
N (0/00)
16
14
15
equivalent to a large individual of a small species in the same
body mass class. A strength of the focus on body size is that
body size dictates much of the variation in the biological
properties of individuals11 and underpins predator-prey
interactions.12 Size-based approaches are especially relevant
in aquatic ecosystems where many species grow through five
or more orders of magnitude in body mass and two or more
trophic levels during their life cycle and where most primary
production is attributable to pelagic phytoplankton.
Increases in mean trophic level with size were implied by
theoretical analyses9 and first examined empirically using
nitrogen stable isotope analysis in size-fractionated zooplankton samples.13 If representative sampling of organisms
in different size classes is feasible, then knowledge of
relationships between trophic level, abundance, production,
diversity and size can be used to describe structures and
processes in food webs. Here, we discuss how size-based
nitrogen stable isotope analysis has been used to quantify
intra- and inter-specific variation in trophic levels, predatorprey size ratios, transfer efficiency, food chain length,
relationships between predator and prey species diversity
and the dynamics of energy use by populations and
communities and to establish baselines for assessing the
scale of human impacts on ecosystems. In addition, we
consider the effects of assumptions about fractionation on
the outcomes of such analysis.
12
10
0
5
10
15
log2 body mass class
Figure 1. Relationships between the mean d15N (95% CL)
and maximum body mass of North Sea fish species (a) and
between the mean d15N (95% CL) of fishes of all species
combined into log2 body mass classes (b).14
The relationship between body size and trophic structure
is expected to be stronger in marine food webs that are based
on phytoplankton than in marine food webs based on benthic
algae, macroalgae and detritus or food webs in freshwater
and terrestrial environments.17,18 However, phytoplankton
do account for around 90% of global marine primary
production and around 35% of global primary production.19
Figure 2. Relationships between the d15N of invertebrates/
fishes and log2 body mass class in a size-fractionated community (broken line) and between the d15N of individual
species of invertebrates or fishes and log2 body mass class
(continuous lines).15 The individual species shown are those
that accounted for the greatest proportion of the biomass of
the whole community.
Rapid Commun. Mass Spectrom. 2008; 22: 1673–1680
DOI: 10.1002/rcm
Stable isotope analysis in macroecological research
Predator-prey mass ratios and transfer
efficiency
If the fractionation of d15N with trophic level is known then
relationships between d15N and body mass class can be used
to predict the predator-prey mass ratio (PPMR) in the
community. The PPMR is the ratio of the mean body mass of
predators in a food web to the mean body mass of their prey.
PPMR is an important attribute of a food web because it can
be used to predict the strength of predatory interactions, the
length of food chains and the pathways of energy transfer.
Estimates of the PPMR are necessary inputs to models of the
structure and function of marine food webs.9 Mean PPMR
(m) is a function of the slope (b) of the relationship between
d15N and logn body mass class, and is calculated as m ¼ n(D/b),
where n is the base of the logarithm of the body mass classes
and D the fractionation of d15N.20
The use of the relationships between d15N and body mass
class to calculate the PPMR requires representative sampling
of the animals in each body size class in the food web. This is
challenging but possible in most marine environments.
Representative sampling will usually require different
sampling gears to target different groups of animals. The
mean d15N weighted by biomass for all groups can be
calculated from the abundance and d15N data for each group,
provided that the sampling gears allow abundance to be
determined in a standard way for all groups (e.g. mean
biomass m2).20
Body mass is closely related to other attributes of animals,
such as metabolism, respiration, production, mortality and
intrinsic rates of population increase.11 For this reason, the
distributions of abundance by body mass class can be used to
approximate the distributions of other attributes, provided
that relationships between body mass and these attributes
are known. The relationship between body mass (M) and
individual biomass production (P/M) is approximated by
P/M ¼ aM0.25, where a is a constant. The existence of this
relationship underpins an approach where size-based
nitrogen stable isotope analysis can be used to predict
transfer efficiency (e); where e is defined as the proportion of
prey production that is converted into predator production
(e ¼ Plþ1/Pl, where Plþ1 is predator production and Pl
is prey production).20 Production in each body mass class
is estimated from biomass (B) in the body mass class
[P ¼ B (P/M)] and trophic level in each class from
d15N. The transfer efficiency (TE) can be calculated from
the slope of the relationship between lognP (y) and d15N (x)
where e ¼ nDb.
Total species richness often decreases systematically with
increasing body mass and this allows size-based nitrogen
stable isotope analysis to be used to estimate relationships
between numbers of predator and prey species.20 For a
relationship between logn normalised species richness by
body mass class (y) and d15N by body mass class (x), the ratio
between the number of predator and prey species will be nDb.
The approaches for estimating PPMR, TE and the ratios
between numbers of predator and prey species all rely on
representative sampling of biomass and species richness in
each size class and knowledge of the fractionation of
d15N (Dd15N). The logistical challenge of sampling can be
considerable, but it has been possible to conduct such
Copyright # 2008 John Wiley & Sons, Ltd.
1675
sampling for defined size windows in the overall size
spectrum and estimates of these parameters have been
published for some marine food webs.21,22 These size-based
methods of food web analysis make a number of gross
simplifications about the size-based nature of predator-prey
relationships, the significance of omnivory in marine
ecosystems and the estimation of trophic level from nitrogen
stable isotope analysis.15,20 However, they do provide a basis
for assessing the structure and function of ecosystems at
large spatial scales and the results have been broadly
consistent with those obtained from costly and labourintensive diet and ecosystem modelling studies.23
Food chain length
Food chain length affects rates of nutrient recycling, levels of
contaminants in top predators and pathways of energy flow
in food webs. Nitrogen stable isotope data have been used to
estimate food chain length,24 with much of the existing
emphasis on work in freshwater and terrestrial environments.25–28 In size-based food webs, PPMR and food chain
length are closely linked since, in a community of given size
composition, a smaller PPMR will result in longer food
chains. The drivers that affect food chain length could act on
the PPMR or the length of food chains directly, by removing
top predators. An investigation of the links between PPMR
and food chain length at 74 sites in the North Sea21 showed
that the PPMR was smaller in longer food chains and in less
variable environments. Also notable was the observation
that the heaviest predator at each site rarely fed at the highest
trophic level (Fig. 3) and the longest food chains supported
predators with intermediate body size. This suggested that it
would not be sufficient to sample only the largest predators
when estimating food chain length. On average, trophic level
rose consistently and significantly with body mass (Fig. 4),
and the mean predator-prey mass ratio among sites was
424:1 if Dd15N ¼ 3.4% was assumed.21
Figure 3. The relationship between the maximum trophic
level of any individual and the maximum trophic level of the
largest individual in North Sea food webs.21 Trophic level was
estimated from d15N assuming a fractionation of 3.4% with
each trophic level and that ‘base’ d15N varied according to a
model linking base d15N and environmental variables.29
Rapid Commun. Mass Spectrom. 2008; 22: 1673–1680
DOI: 10.1002/rcm
1676 S. Jennings et al.
biomass scaling of M0.2 in a marine food web was not
significantly different from the scaling of M0.24 predicted
from energy availability at size.22 Applications of the method
to subsets of food webs have provided further evidence of
the central role of energy availability in determining
abundance in body size classes. In some benthic food webs,
for example, the d15N value suggested that trophic level fell
with body size because the smaller size classes in the food
web were dominated by predatory worms and the larger size
classes by deposit and suspension feeders such as bivalves.31
When the positive scaling of E and M was used to predict the
relationship between log B and log M, the expectation that
the slopes would be more positive than M0.25 was
confirmed.31
Figure 4. Relationship between body mass and mean
trophic level at 74 North Sea sites ( SD).21 Trophic level
was estimated from d15N assuming a fractionation of 3.4%1
and that base d15N varied according to a model linking base
d15N and environmental variables.29
Size spectra
Abundance-body size relationships capture the distribution
of body sizes in food webs. These are expressed in two ways.
First, as cross-species relationships between log mean
abundance and log mean M, often for taxonomically or
functionally defined groups of species, such as plants, birds
or mammals. Second, for all individuals in a food web or
specified size range, from relationships between log
abundance in a body mass class and log M at the lower
bound or midpoint of the class.9 The former approach has
principally been explored for terrestrial communities and the
latter for aquatic communities, where abundance-body size
relationships are usually referred to as size spectra.8
In circumstances where there is a linear reduction in
energy availability with size, as in strongly size-based
aquatic food webs, it is straightforward to estimate energy
availability in different size classes of animals from knowledge of trophic level, TE and PPMR.9 The energy available
ultimately constrains abundance and therefore determines
the slope of the size spectrum.10 However, calculations of
slope also depend on knowledge of how abundance changes
with body size when energy is shared. This is predicted by
the energetic equivalence hypothesis, which suggests that
numerical abundance N scales as M0.75 in plants or animals
sharing a resource (and therefore biomass abundance B
as M0.25). Since the metabolic rate is known to scale as
approximately M0.75, energy use is independent of M and
body size does not provide any population with an energetic
advantage.30 When the availability of energy (E) to animals
changes with body size, as is expected when trophic level
changes, then the scaling of N or B with M can be predicted
from the scaling of E and M which is given as E ¼ Mloge/logm.
The slope of the size spectrum (relationship between log B
and log M) can then be calculated as Mloge/logm M0.25.
Based on the preceding theory, nitrogen stable isotope
analysis of animals fractionated into body mass classes has
been used to predict slopes of size spectra in complete food
webs and subsets of those webs. For example, the observed
Copyright # 2008 John Wiley & Sons, Ltd.
Human impacts
Human activities can modify size spectra. Fishing, for
example, increases the slopes of size spectra.32 Comparisons
between slopes that describe food web structures in the
absence of human impacts and slopes that describe impacted
systems provide a measure of the magnitude of human
impacts.
The unimpacted slope of a size spectrum can be predicted
using the methods described in the preceding section
because the PPMR and TE are largely unaffected by fishing.21
For the North Sea, a comparison of the slopes of
contemporary (2001) size spectra and slopes predicted from
measurements of PPMR and TE33 revealed reductions in the
biomass of large fishes, suggesting that the biomass of fishes
weighing 4–16 kg and 16–66 kg, respectively, was 97.4% and
99.2% lower than in the absence of fisheries exploitation
(Fig. 5). The method is advantageous because it provides a
common baseline for assessing fishing impacts in different
regions and because the structure of unexploited communities cannot always be predicted from historical data. This is
because fisheries exploitation usually precedes scientific
investigation and non-fisheries impacts, such as climate
change, modify ecosystems over time.
Figure 5. Predicted slope of an unexploited size spectrum
(dashed line) and the size spectrum for the exploited North
Sea in 2001 (continuous line and data).33 Trophic level was
calculated from body mass and d15N assuming a fractionation
of 3.4% and that ‘base’ d15N varied according to a model
linking base d15N and environmental variables.29 The height
(intercept) of the size spectrum was predicted from an estimate of primary production.33
Rapid Commun. Mass Spectrom. 2008; 22: 1673–1680
DOI: 10.1002/rcm
Stable isotope analysis in macroecological research
1677
Stability of trophic structure
Time-series d15N data are starting to provide insight into the
relative stability of trophic levels of different species and
body size classes in food webs, and of the food web as a
whole.34 In addition, valuable insights into the stability of
trophic structure and feeding strategies are starting to
emerge from comparisons among sites subject to different
levels of human impact. Thus fragmentation of coastal
marine habitats appears to be associated with reductions in
the niche width of predators using those habitats.35
Regular annual sampling of species that dominate the total
biomass of the North Sea fish community shows the range of
trophic strategies employed by these species and their
variation through time (Fig. 6). Indeed, for some species such
as plaice Pleuronectes platessa, trophic level actually decreases
with body size reflecting a shift in diet from small predatory
worms to larger filter-feeding bivalve molluscs.34 Such
strategies are only adopted by relatively few species,
however, and the main pathways of energy flow in the
community are strongly size-based. Further, the overall
relationship between trophic level and body mass, and hence
PPMR, is remarkably consistent despite variations in the
abundance of component species.34
Fractionation
All the preceding work is predicated on the assumption that
d15N provides a good index of trophic level and that Dd15N is
known. For this reason, the effects of assumed Dd15N need to
be understood and Dd15N should ideally be validated in
comparable food webs. A simple approach to understanding
the effects of assumed Dd15N on the results of analyses is to
conduct sensitivity analysis. These show, for example, how
estimated PPMR and TE might vary with Dd15N in a marine
food web (Fig. 7). The effects of the propagation of errors
associated with TE or PPMR can then be assessed by looking,
for example, at their effect on the calculated slope of the size
spectrum (Fig. 8).
Absolute values of trophic level as required in the analyses
of energy flow are also dependent on assumptions about
d15N at the base of food chains (d15Nbase). In large areas of the
marine environment that have been studied d15Nbase is
positively correlated with salinity, probably reflecting the
contribution of sewage-derived nitrogen inputs to coastal
waters. In a study of spatial variation in the d15N of a
bivalve mollusc that filters sea water and is a good integrator
of the very variable short-term d15N signals that characterise
detritus and phytoplankton,36 a model of the relationship
between d15N and environmental variables was used to show
that 51% and 77% of spatial variance in the d15N of two fish
species, and hence apparent trophic level, could be attributed
to differences in d15N at the base of the food chain.29 Since
temperature and salinity are correlated with base d15N, and
since gradients in these physical variables are particularly
pronounced in coastal areas, spatial comparisons of trophic
level in these areas are easily biased if fine-scale information
on base d15N is not available. Conversely, in offshore regions,
where temperature and salinity show little variation over
large areas, variations in base d15N and the associated bias are
expected to be less (Fig. 9).
Copyright # 2008 John Wiley & Sons, Ltd.
Figure 6. Relationships between d15N or trophic level and
body mass for North Sea fishes in the years 2002–2004. Fine
continuous lines indicate the fitted relationships between
d15N or trophic level and body mass for species sampled in
each year, where the start and end points of the line indicate
the minimum and maximum recorded body mass. Closed
circles on each line show the mean trophic level and mean
body mass for each species. Open circles indicate biomass
weighted mean trophic level when individuals were pooled by
body mass class independent of species identity. The broken
line is the fitted cross-species relationship between mean
trophic level and mean body mass while the continuous grey
line is the fitted relationship between biomass weighted mean
trophic level and body mass.34 Trophic level was estimated
from d15N assuming a fractionation of 3.4% and that ‘base’
d15N varied according to a model linking base d15N and
environmental variables.27 Relationships between d15N and
trophic level on the y axes differ slightly among years because
trophic level was re-based year on year.
Understanding the effects of uncertainty in Dd15N on the
outputs of analyses is only one part of a process that should
also include validation; to help determine what the most
likely values of Dd15N would be. In studies that aggregate
across individuals under ‘natural’ conditions it is likely that
some of the extreme variability in Dd15N that is recorded for
individuals in experimental studies will not be relevant37 and
the observation that PPMR, as assessed with d15N and diet
data, seems to be relatively constant across at least ten orders
of magnitude in body mass suggests that strong size-related
effects on Dd15N are unlikely. At the level of individuals and
populations, variation in Dd15N is likely to be greater and will
affect the relationships among calculated trophic levels for
Rapid Commun. Mass Spectrom. 2008; 22: 1673–1680
DOI: 10.1002/rcm
1678 S. Jennings et al.
Figure 7. The effects of assumptions about the degree of
fractionation of d15N (%) with trophic level on transfer efficiency (broken line) and predator-prey body mass ratios
(continuous line).20
different species. Dd15N has often been assumed to be
3.4 þ 1.1% (1 standard deviation (SD)) and although this
estimate was based on a very small number of individual
estimates,38 it proved to be a surprisingly reliable mean value
when further results were compiled.37
In the context of the food web work previously described,
we have placed recent emphasis on experimentally validating Dd15N for fishes under experimental regimes that
approximate natural conditions of temperature, light, food
type and food availability (temperatures and/or temperature variations encountered in the wild, natural photoperiods, low stocking densities, food types eaten in the wild
and feeding rates representing the expected minimum and
maximum for wild fish). For example, when European sea
bass (Dicentrarchus labrax) were reared on constant diets of
dab (Limanda limanda) muscle or whole sandeel (Ammodytes
marinus) for 2 years under natural light and temperature
regimes, mean values of Dd15N were 3.83% and 3.98% for
muscle tissue.39 Fractionation appeared to be independent of
body mass, but there was a very small time effect on muscle
Dd15N, described by a sinusoidal function with a period of
1 year and amplitude 0.3%. Although the Dd15N for bass
muscle on both diets approached 4%, the data from this
study were combined with literature data for other
Figure 8. Sensitivity of the slope of the size spectrum to
changes in PPMR and TE.33
Copyright # 2008 John Wiley & Sons, Ltd.
Figure 9. Predicted spatial variation in queen scallop (Aequipecten opercularis) white muscle d15N in the North Sea, Irish
Sea and Channel as determined from a model linking
d15N and environmental parameters.29 The d15N of queen
scallop white muscle was treated as an indicator of d15Nbase
for trophic level 2.5. This figure is available in colour online at
www.interscience.wiley.com/journal/rcm.
species and these suggested that, when species-specific data
were not available, a mean Dd15N for fish muscle of 3.2%
should be applied (Fig. 10), close to the previously reported
mean values for a wider range of animals.37,38
To further investigate effects of temperature and feeding
rate on Dd15N for bass muscle, bass were also reared on
identical sandeel diets at 11 and 168C and at three ration
levels for 600 days.32 The Dd15N was affected by a temperature ration interaction (Fig. 11), with Dd15N falling from
4.41% at 118C to 3.78% at 168C for all rations combined. The
effects of ration were relatively small. Thus at all rations
Dd15N was higher in the cooler temperature but the
difference in Dd15N between cool and warm varied depending on the ration, with the greatest difference (0.2% per
18C) at medium ration and the least difference at high ration.
Figure 10. Frequency histogram of the magnitude of nitrogen trophic fractionation for fish white muscle based on
literature data.39
Rapid Commun. Mass Spectrom. 2008; 22: 1673–1680
DOI: 10.1002/rcm
Stable isotope analysis in macroecological research
1679
the costs of low replication. If methods for compoundspecific analyses can be more effectively automated and if
amino acids that provide reliable information on trophic
fractionation are identified, then compound-specific analyses
will help to deal with many of the factors that probably
influence bulk signals, such as differences in the amino acid
supply from diet and protein breakdown.45
Acknowledgements
15
Figure 11. Effect of temperature and ration on Dd N of white
muscle of European sea bass fed a constant diet of sandeel.
Data are from samples taken after 200 days equilibration up to
day 600.40 ‘High’, ‘medium’ and ‘low’ refer to the ration levels.
The experimental studies largely confirm the expectation
that d15N is a better indicator of relative rather than absolute
trophic level and that differences in Dd15N among species
should be expected (Fig. 10). Values of Dd15N around 3.4%
still represent the mean expectation and may be realised
when animals of many species are combined in size classes.
In studies of single species the use of a mean value may be
more misleading. These concerns surrounding the value of
Dd15N should be weighed against the recognition that there is
currently no better method for looking at many aspects of
food web structure and function, and that approaches based
on nitrogen stable isotope analysis often require lower levels
of replication than diet studies to provide realistic time- and
space-integrated assessments of food web structure and
function.
CONCLUSIONS
The use of nitrogen stable isotope analysis in size-based
marine food web and macroecological research provides one
example of the many potential applications of this methodology when assessing food web structure and function.2 One
valuable feature of the size-based work, however, is that it
has paralleled a resurgence of interest in the development
and application of size-based models of marine food
webs.11,41,42 and empirical analyses have been guided by
theory, and vice versa. Thus, nitrogen stable isotope data have
been used to quantify intra- and inter-specific variation in
trophic level, predator-prey size ratios, transfer efficiency
and food chain length, and many of these estimates have
been to support the development, testing and parameterisation of models.
The precision, and thus the power to resolve competing
hypotheses, has been much improved by the automation and
falling relative cost of nitrogen stable isotope analysis.
However, the methods of marine macroecological and food
web analysis described would have wider applicability if
there were more confidence in the accuracy of the outputs,
especially when attempting to assess energy fluxes. An
increase in accuracy is most likely to be achieved with
compound-specific rather than bulk analysis,43–44 but at
present there is an obvious requirement to balance the
benefits of more accurate predictions of trophic level against
Copyright # 2008 John Wiley & Sons, Ltd.
Many thanks to the organisers of SIMSUG 2007 for the
invitation to present this review, to the UK Department of
Environment, Food and Rural Affairs and the Natural
Environment Research Council for funding (Grant Nos.
MF1001 and NER/S/A/2003/11886, respectively), to three
referees for helpful suggestions and to the staff of Isoanalytical Ltd., the Newcastle University Mass Spectrometry
Unit and the Scottish Crop Research Institute for their contributions to the studies described.
REFERENCES
1. Owens NJP. Adv. Mar. Biol. 1987; 24: 389.
2. Fry B. Stable Isotope Ecology. Springer: New York, 2006;
308.
3. Peterson BJ, Fry B. Annu. Rev. Ecol. Syst. 1987; 18: 293.
4. Thomas CJ, Cahoon LB. Mar. Ecol. Prog. Ser. 1993; 95: 19.
5. Phillips DL. Oecologia 2001; 127: 166.
6. Hobson KA. Oecologia 1999; 120: 314.
7. Preston T, Owens NJP. Analyst 1983; 108: 971.
8. Sheldon RW, Prakash A, Sutcliffe WH. Limnol. Ocean. 1972;
17: 327.
9. Dickie LM, Kerr SR, Boudreau PR. Ecol. Monographs 1987; 57:
233.
10. Brown JH, Gillooly JF. Proc. Nat. Acad. Sci. USA 2003; 100:
1467.
11. Brown JH, Gillooly JF, Allen AP, Savage VM, West GB.
Ecology 2004; 85: 12.
12. Brose U, et al. Ecology 2006; 87: 2411.
13. Fry B, Quinones RB. Mar. Ecol. Prog. Ser. 1994; 112: 201.
14. Jennings S, Pinnegar JK, Polunin NVC, Boon TJ. J. Anim. Ecol.
2001; 70: 934.
15. Jennings S, Pinnegar JK, Polunin NVC, Warr KJ. Mar. Ecol.
Prog. Ser. 2002; 226: 77.
16. Harvey PH, Pagel MD. The Comparative Method in Evolutionary Biology. Oxford University Press: Oxford, 1991; 239.
17. Layman CA, Winemiller KO, Arrington DA, Jepsen DB.
Ecology 2005; 86: 2530.
18. Al-Habsi SH, Sweeting CJ, Polunin NVC, Graham NAJ. Mar.
Ecol. Prog. Ser. 2008; 353: 55.
19. Duarte CM, Cebrián J. Limn. Ocean. 1996; 41: 1758.
20. Jennings S, Warr KJ, Mackinson S. Mar. Ecol. Prog. Ser. 2002;
240: 11.
21. Jennings S, Warr KJ. Proc. Roy. Soc. B 2003; 270: 1413.
22. Jennings S, Mackinson S. Ecol. Lett. 2003; 6: 971.
23. Polunin NVC, Pinnegar JK. In Handbook of Fish Biology
and Fisheries, Hart PJB, Reynolds JD. Blackwell: Oxford,
2002; 30.
24. Vander Zanden MJ, Fetzer WW. Oecologia 2007; 116: 1378.
25. Kling GW, Fry B, O’Brien JW. Ecology 1992; 73: 651.
26. Cabana G, Rasmussen JB. Proc. Nat. Acad. Sci. USA 1996; 93:
10844.
27. Vander Zanden MJ, Casselman JM, Rasmussen JB. Nature
1999; 401: 464.
28. Post DM, Pace ML, Hairston NG. Nature 2000; 405: 1047.
29. Jennings S, Warr KJ. Mar. Biol. 2003; 142: 1131.
30. Damuth J. Nature 1981; 290: 699.
31. Maxwell TAD, Jennings S. Oecologia 2006; 150: 282.
32. Rice JC, Gislason H. ICES. J. Mar. Sci. 1996; 53: 1214.
33. Jennings S, Blanchard JL. J. Anim. Ecol. 2004; 73: 632.
34. Jennings S, d’Oliveira JAA, Warr KJ. J. Anim. Ecol. 2007; 76:
72.
Rapid Commun. Mass Spectrom. 2008; 22: 1673–1680
DOI: 10.1002/rcm
1680 S. Jennings et al.
35. Layman CA, Quattrochi JP, Peyer CM, Allgeier JE. Ecol. Lett.
2007; 10: 937.
36. Lorrain A, Paulet YM, Chauvaud L, Savoye N, Donval A,
Saout C. J. Exp. Mar. Biol. Ecol. 2002; 275: 47.
37. Post DM. Ecology 2002 83: 703.
38. Minagawa M, Wada E. Geochem. Cosmochim. Acta 1984; 48:
1135.
39. Sweeting CJ, Barry J, Barnes C, Polunin NVC, Jennings S.
J. Exp. Mar. Biol. Ecol. 2007; 340: 1.
Copyright # 2008 John Wiley & Sons, Ltd.
40. Barnes C, Sweeting CJ, Jennings S, Barry JT, Polunin NVC.
Funct. Ecol. 2007; 21: 356.
41. Anderson KH, Beyer JE. Am. Nat. 2006; 168: 54.
42. Maury O, Faugeras B, Shin YJ, Poggaile JC, Ben Ari T,
Marsac F. Prog. Ocean. 2007; 74: 479.
43. Macko SA, Estep MLF, Engel MH, Hare PE. Geochem.
Cosmochim. Acta 1986; 50: 2143.
44. McClelland JW, Montoya JP. Ecology 2002; 83: 2173.
45. Vanderklift MA, Ponsard S. Oecologia 2003; 136: 169.
Rapid Commun. Mass Spectrom. 2008; 22: 1673–1680
DOI: 10.1002/rcm