ArticlePDF Available

The roles of productivity and ecosystem size in determining food chain length in tropical terrestrial ecosystems


Abstract and Figures

Many different drivers, including productivity, ecosystem size, and disturbance, have been considered to explain natural variation in the length of food chains. Much remains unknown about the role of these various drivers in determining food chain length, and particularly about the mechanisms by which they may operate in terrestrial ecosystems, which have quite different ecological constraints than aquatic environments, where most food chain length studies have been thus far conducted. In this study, we tested the relative importance of ecosystem size and productivity in influencing food chain length in a terrestrial setting. We determined that (1) there is no effect of ecosystem size or productive space on food chain length; (2) rather, food chain length increases strongly and linearly with productivity; and (3) the observed changes in food chain length are likely achieved through a combination of changes in predator size, predator behavior, and consumer diversity along gradients in productivity. These results lend new insight into the mechanisms by which productivity can drive changes in food chain length, point to potential for systematic differences in the drivers of food web structure between terrestrial and aquatic systems, and challenge us to consider how ecological context may control the drivers that shape food chain length.
Content may be subject to copyright.
Ecology, 94(3), 2013, pp. 692–701
Ó2013 by the Ecological Society of America
The roles of productivity and ecosystem size in determining food
chain length in tropical terrestrial ecosystems
Biology Department, Stanford University, 371 Serra Mall, Stanford, California 94305 USA
Division of Vertebrate Zoology, Smithsonian Institution, Washington, D.C. 20013 USA
Center for the Environment, Harvard University, 24 Oxford Street, Cambridge, Massachusetts 02138 USA
Hopkins Marine Station, Stanford University, 120 Oceanview Boulevard, Pacific Grove, California 93950 USA
Department of Environmental Science, Policy, and Management, University of California, 130 Mulford Hall,
Berkeley, California 93950 USA
Department of Environmental Earth System Science, Stanford University, Stanford, California 94305 USA
Abstract. Many different drivers, including productivity, ecosystem size, and disturbance,
have been considered to explain natural variation in the length of food chains. Much remains
unknown about the role of these various drivers in determining food chain length, and
particularly about the mechanisms by which they may operate in terrestrial ecosystems, which
have quite different ecological constraints than aquatic environments, where most food chain
length studies have been thus far conducted. In this study, we tested the relative importance of
ecosystem size and productivity in influencing food chain length in a terrestrial setting. We
determined that (1) there is no effect of ecosystem size or productive space on food chain
length; (2) rather, food chain length increases strongly and linearly with productivity; and (3)
the observed changes in food chain length are likely achieved through a combination of
changes in predator size, predator behavior, and consumer diversity along gradients in
productivity. These results lend new insight into the mechanisms by which productivity can
drive changes in food chain length, point to potential for systematic differences in the drivers
of food web structure between terrestrial and aquatic systems, and challenge us to consider
how ecological context may control the drivers that shape food chain length.
Key words: ecosystem size; food chain length; food web structure; islands; Palmyra Atoll; productivity.
Understanding the forces that determine food chain
length (FCL; the number of trophic exchanges between
the top and bottom of a food web) and explain natural
variation in FCL across ecosystems was a prominent
aim of early ecologists (Elton 1927, Lindeman 1942,
Hutchinson 1959) and is an issue that continues to be
debated today (Pimm 1982, Post 2002, Calcagano et al.
2011). Food chain length is a fundamental architectural
property of ecosystems that carries strong implications
for a wide range of ecosystem functions and properties
(DeAngelis et al. 1989, Cabana and Rasmussen 1994,
McIntyre et al. 2007). While multiple potential drivers of
variation in FCL have been proposed (Schoener 1989,
Post 2002, McHugh et al. 2010, Calcagano et al. 2011),
three of them (ecosystem size, productivity, and
environmental stochasticity [disturbance]) have received
the majority of research attention and support.
Productivity was an early and intuitive explanation
for observed variation in food chain length (Hutchinson
1959, Pimm 1982). Based largely on theoretical grounds,
the suggestion of this energy limitation hypothesis was
that more productive ecosystems could support a greater
biomass of both consumers and their predators,
ultimately allowing higher trophic levels to persist.
Results from microcosm experiments and some early,
detailed food web analyses corroborated these theoret-
ical predictions, illustrating that productive ecosystems
could support longer food chains (Jenkins et al. 1992,
Kaunzinger and Morin 1998, Townsend et al. 1998).
Despite the theoretical and small-scale experimental
support for the connections between productivity and
FCL, multiple studies, including several larger scale field
comparisons, have failed to identify productivity alone
as a primary driver of increased FCL (Briand and
Cohen 1987, Vander Zanden et al. 1999, Post et al. 2000,
Sabo et al. 2010). Similarly mixed results have emerged
from the studies of the role of disturbance on FCL, with
studies finding negative (Townsend et al. 1998, McHugh
et al. 2010, Sabo et al. 2010), positive (Parker and Huryn
2006, Marty et al. 2009), and no effects (Takimoto et al.
2008) of disturbance on FCL. In contrast to the highly
variable responses of FCL to productivity alone or to
Manuscript received 4 May 2012; revised 8 October 2012;
accepted 22 October 2012. Corresponding Editor: N. J.
Present address: Center for the Environment, Harvard
University, 24 Oxford Street, Cambridge, Massachusetts
02138 USA. E-mail:
environmental stochasticity, ecosystem size, or size
mediated productivity (called productive space), has
been a relatively reliable and frequently observed driver
of FCL (Schoener 1989, Spencer and Warren 1996, Post
et al. 2000, Post et al. 2007, Vander Zanden and Fetzer
2007, Takimoto et al. 2008, McHugh et al. 2010, Sabo et
al. 2010). These studies do not suggest that ecosystem
size or size-mediated productivity is likely to be the only
explanatory driver of FCL, and indeed, there is growing
agreement that there are generally likely to be multiple
interacting drivers of FCL. However, cumulatively,
these findings have led to the suggestion that ecosystem
size is typically the strongest and most consistent driver
of FCL. Yet, the variable results from large-scale studies
on the drivers of FCL emphasizes how little we still
know about the processes and mechanisms underlying
these relationships and how crucial it is to explore
multiple contexts to understand where and when various
controls on FCL are most important (McHugh et al.
Even in systems where primary drivers of FCL have
been identified, detailed mechanisms underlying these
patterns are generally poorly understood (Post et al.
2000, Takimoto et al. 2008). Three non-mutually
exclusive possible explanations underlying changes in
FCL were proposed by Post and Takimoto (2007): (1)
new higher level predator species are added (additive
mechanism); (2) the diversity or relative abundance of
intermediate consumers increases (insertion mecha-
nism); (3) the consumers diet changes either by growing
larger and feeding on larger and potentially higher
trophic position prey, or by purely behavioral shifts
toward specialization allowing for more trophic ex-
changes within the food web (omnivory mechanism).
However, the relative roles of these mechanisms in
driving FCL, particularly the latter two, have received
little attention. Understanding the mechanisms by which
FCL changes is critically important to any effort to
understand the variability in drivers across ecosystems,
and how ecological context might determine observed
variation in drivers of FCL.
Nearly all large-scale, ecosystem-level FCL studies
exploring the drivers of FCL have come from freshwa-
ter, and mostly lacustrine, ecosystems (Vander Zanden
et al. 1999, Post et al. 2000, Thompson and Townsend
2005, Doi et al. 2009, McHugh et al. 2010, Sabo et al.
2010). Aquatic food webs likely differ fundamentally
from terrestrial ones in many ways that could affect the
functioning of the various mechanisms driving variation
in FCL. For example, aquatic food webs may diverge
regarding the relative importance of top-down ecolog-
ical controls (Shurin et al. 2002, Borer et al. 2005), body
size ratios and the degree of size structuring across food
webs and trophic levels (Brose et al. 2006, Shurin et al.
2006), the relationship between productivity and diver-
sity (Mittelbach et al. 2001, Partel et al. 2007), and in the
relative abundance of generalist consumers (Hairston
and Hairston 1993, Thompson et al. 2007). Moreover,
many terrestrial ecosystems, as well as many marine
ecosystems, are much larger than most lake and riparian
systems, and often lack the relatively clearly defined
boundaries of many freshwater environments, perhaps
reducing the likelihood that ecosystem size might limit
FCL (Post et al. 2007). Given these multiple, systematic
differences between the aquatic systems from which so
many of our conclusions about FCL have been derived,
and other ecological contexts, there is a strong need for
additional examination of the drivers of FCL, and the
mechanisms by which these drivers operate, outside the
freshwater environment.
To take up this task, we first measured average FCL
across a series of topographically and geologically
similar islets of a single tropical atoll. The islets are
positioned across large and independent gradients of
productivity and size, but yet span only a very small
geographic area under the same climatic conditions. We
then compared a series of models including productivity,
ecosystem size, and productive space for their ability to
explain observed variation in total FCL and trophic
position of each consumer species. We found strong
evidence that productivity alone drives FCL in these
systems. We also explored the mechanisms underlying
changes in FCL through detailed surveys of consumer
abundance, diversity, body size, and diet across the
productivity gradient. We found no support for changes
in predator identity in explaining observed gradients in
FCL. However, we did see support for the role of
increased species richness in explaining these changes, as
well as for the role of behavioral and morphological
changes in consumers. Cumulatively these small shifts
may be driving the large overall shift in FCL across this
tropical terrestrial productivity gradient.
The fieldwork was conducted across a series of islets
(n¼23) at Palmyra Atoll (58530N, 1628050W), a remote,
wet tropical atoll in the central Pacific Ocean (see Plate
1). All islets are low lying (,2 m above sea level), consist
of coral reef-derived materials overlying limestone
basement, and are located in close proximity to one
another (all within a 20-km
area). Islet area (used as a
metric of ecosystem size) varied nearly four orders of
magnitude, from 5.29 310
to 2.60 310
. Measured
productivity varied 14-fold, from 2.20 310
to 3.20 3
g foliar Nm
, while available soil nutrients
(‘‘potential productivity,’’ as used in other studies)
varied ;100-fold from 8 to 786 lg plant available
, and PO
per gram of dry mass soil.
Variation in productivity across islets was derived from
variation in abundance of nesting and roosting seabirds
on the atoll (Young et al. 2010a), and there was no
significant relationship between ecosystem (islet) size
and productivity (R
¼0.05, P¼0.76). Due to their very
close geographic proximity, and the open ocean
surroundings, the islets likely experience little natural,
systematic variation in disturbance over long time
Characterization of productivity and ecosystem size
Many different approaches have been used to estimate
productivity or energy availability of systems in FCL
studies, including direct measurements of productivity
(e.g., mg Cm
[Thompson and Townsend 2005,
Sabo et al. 2010]), nutrient limitation (e.g., lg P/L [Post
et al. 2000], precipitation index [Arim et al. 2007]), or,
most recently, realized resource availability (e.g., lg
edible C/L [Doi et al. 2009]). We elected to estimate
productivity as g foliar Nm
produced in con-
trolled growth experiments on site. In Appendix A we
provide details on methods of this growth experiment,
confirm that this metric is strongly correlated to
alternative metrics of productivity, and demonstrate
that the use of alternative metrics, including limiting
nutrients, do not change our conclusions. Ecosystem size
for each islet was directly measured via ArcGIS (ESRI
2008). We used the product of productivity and
ecosystem size as a metric of productive space (Post
Collection and measurement of isotopic samples
The length of the food chain on each islet was
approximated using the trophic position of the highest
level predator in that system (generally geckos or
spiders). The trophic position of consumers was
calculated as the mean increase in d
N between that
consumer and the base of the food web, corrected for
changes in consumer dietary sources using a mixing
model, and calibrated by average fractionation of
consumers (Appendix B). The isotopic baselines of each
food web were defined using an integrated sample of
three common plant species on each islet. The marine
baseline of each food web was characterized as an
integrated sample of marine wrack collected across the
atoll system (Appendix B contains details on isotopic
baseline characterizations). On each islet we attempted
to sample nine different species of the most common
consumers including four predators (Heteropoda venato-
ria,Neoscona theisi,Lepidodactylus lugubris,Lepidodac-
tylus sp. nov), two omnivores (Rattus rattus,Ornebius
sp.), and three herbivores (Stoeberhinus testaceus,
Dysmicoccus sp., and Agonexana argaula larvae).
Between 3 and 10 individuals per species were sampled
at each islet when present, although not all species were
found at all islets. R. rattus was sampled only at a subset
of 10 islets. For isotopic analysis of consumers we used
whole animals for insects and arachnids, except for the
abdominal cavity (to avoid sampling gut contents); for
geckos we used tail tips; and for rats we used whole
muscle tissue. R. rattus and Lepidodactylus spp. tissue
samples were stored frozen, freeze-dried, ground, lipid
extracted (using chloroform–methanol extraction), and
oven dried prior to analysis (Catenazzi and Donnelly
2007). Plant tissue, insects, and marine wrack were dried
at 558C for 48 h, and ground to a powder. All
individuals within a given taxonomic group collected
on a given islet were pooled prior to isotopic analysis,
except for R. rattus and Lepidodactylus spp., where all
individual animal samples were analyzed separately, as
part of another study, and data were then pooled
subsequent to analysis (see Appendix C: Table C1 for
sample sizes). Stable-isotopic ratios of d
C and d
were analyzed using a Carlo Erba CN analyzer coupled
to a ConFlo open split interface (Carlo Erba, Milan,
Italy) feeding either a Thermo Finnigan Delta-Plus
IRMS or a Thermo Delta V Advantage isotope-ratio
mass spectrometer (IRMS; Thermo Fisher Scientific,
Bremen, Germany). Analytical error was ,0.2%for
both d
C and d
N. All samples for these analyses were
collected over a four-month period in 2010.
Consumer abundance, diversity, and diet
To explore mechanisms potentially driving patterns of
increased FCL, we surveyed arthropod (primarily
insect) abundance and diversity across islets using three
different methods: (1) blacklight traps, (2) standardized
vegetation surveys, and (3) targeted surveys of abun-
dance of Dysmicoccus sp. and Phisis holdhausi, two
particularly abundant and easily surveyed insects in the
system (sampling details for all methods in Appendix
C). We focused on arthropods as they make up the vast
majority of free-living terrestrial diversity in this system
and include consumers at all trophic levels. While
certainly there are many taxa not captured via these
three methodologies, it was not feasible to comprehen-
sively inventory biodiversity at all islets. We thus assume
that changes consistently observed across the diverse
group of species that we captured using these surveys are
likely to be representative of changes in entire islet food
webs. We also surveyed body mass (60.01 g) of the
larger predators occurring on the atoll (L. lugubris,
Lepidodactylus sp. nov., H. venatoria, and diurnal
spiders) from hand-collected animals (the only terrestrial
vertebrate that occurs in this system that was not
included in our sampling was Hemidactlyus frenatus,an
invasive gecko that occurs only on a few islets). Finally,
we analyzed stomach contents from a subset of surveyed
individuals of Lepidodactylus spp. (both species pooled,
n¼115), and quantified the frequency with which other
predators were found in the gut contents (details in
Appendix C). In order to limit the amount of lethal
sampling of Lepidodactylus spp., these diet analyses were
conducted only on a subset of extremely high- and low-
productivity islets (defined using top and bottom
quartile of productivity).
Calculating food chain length
We calculated food chain length per islet as the
trophic position of the apex predator. Trophic position
of the organisms examined in this study was calculated
to be kþ(d
)/D, with Dequaling
fractionation and krepresenting the trophic position of
HILLARY S. YOUNG ET AL.694 Ecology, Vol. 94, No. 3
baseline (taken to be 1; e.g., Post et al. 2000, Takimoto
et al. 2008). The base of the food web (d
) used in
determining trophic position for all higher omnivores
and predators (H. venatoria,N. theisi,L. lugubris,
Lepidodactylus sp. nov, and R. rattus, and Ornebius
sp.) was calculated using a two-end member mixing
model (Fry and Sher 1984, Post 2002), with marine
wrack and plant material as two potential sources
according to the following formula:
d15Nbase ¼ðd15 Nplant material 3aþd15Nmarine wrack ð1aÞÞ=D
a¼ðd13Cconsumer d13 Cmarine wrackÞ
ðd13Cplant material d13Cmarine wrack Þ
(Appendix B: Fig. B1 shows isotopic values of end
members and of seabird guano as a reference point.) This
accounts both for changes in plant baseline d
across islets due to varying plant uptake of high d
guano on some islets, and for the potential of diet shifting
by consumers to marine food sources on some islets
(Catenazzi and Donnelly 2007). This model assumes no
trophic fractionation of d
C (as in other studies of FCL,
e.g., Post 2002); previous work has demonstrated that
estimates of trophic position using this methodology are
relatively insensitive to estimates of the trophic fraction-
ation of d
C (Post 2002). For obligate herbivores (S.
testaceus,A. argaula,andDysmicoccus sp.), we used
terrestrial plants as the sole baseline for calculating
trophic position (as they are not known to consume
marine wrack). For all species we estimated fractionation
as 3.4%for d
N (Takimoto et al. 2008). In Appendix C
we demonstrate that using other standard values of
trophic fractionation of d
N or species-specific fraction-
ation values (from literature and directly measured in the
laboratory) do not alter our general conclusions. As C
plants are relatively uncommon on Palmyra (Young et al.
2010b), we assumed no variation in carbon from
variability in d
C in terrestrial plants.
Statistical analyses
To analyze the relative role of each potential driver in
explaining FCL and trophic position of each consumer,
we constructed a series of multiple linear regressions that
included productivity, ecosystem size, and the product
of these terms as predictor variables. We selected the
best model using the stepwise method based on Akaike’s
information criterion (AIC
) and tested if the results fit
models using an Ftest. These models were repeated for
total FCL measurements and for trophic position of
each consumer. We reran analyses with varying assump-
tions of fractionation, foliar nutrient baselines, and
metrics used to estimate productivity, and confirmed our
conclusions (Appendix B: Table B1). We used multino-
mial logistic regressions to test if there was a significant
relationship between either productivity or ecosystem
size and the identity of the predator at the top trophic
position on each islet. For consumer abundance,
diversity, and body size, we used multiple regressions;
abundance data for each consumer type was log-
transformed prior to analysis (except for Dysmicoccus
sp. surveys, as this was not a raw abundance metric, but
rather a metric of the percentage of rosettes occupied).
The R
values presented in Table 1 are for each term,
and the productive space term (product of productivity
and area) does not control for main effects. All
statistical analyses were performed in R version
2.13.12 (R Development Core Team 2012).
We found that FCL varied by approximately three
trophic levels between islets. Investigation of the relative
importance of productivity, ecosystem size, and produc-
tive space on FCL demonstrated that productivity alone
TABLE 1. Effects of productivity, island size (measured in m
), and productive space (the product of these terms) on insect
abundance, insect diversity, and predator body size.
Response metric
No. islets
(g Nm
) log
(island area) Productive space
PCoefficient R
PCoefficient R
Total insect abundance
Blacklight surveys (log biomass) 14 0.31 0.04 244.8 6106.6 0.18 0.13 0.3 60.2 0.41 0.01 0.5 60.2
Vegetation surveys (log biomass) 18 0.29 0.02 268.0 6104.6 0.03 0.51 0.1 60.2 0.15 0.11 0.2 60.1
Scale surveys (%rosettes occupied) 18 0.22 0.05 312.2 6148.3 0.01 0.68 0.2 60.2 0.05 0.36 0.1 60.2
Targeted Phisis holdhausi surveys
(log count)
18 0.65 ,0.001 269.9 650.0 0.01 0.75 0.0 60.1 0.09 0.11 0.1 60.1
Blacklight traps (species richness) 14 0.38 0.02 5.6 62.1 0.01 0.74 1.0 62.9 0.40 0.02 6.0 62.1
Body size
Lepidodactylus lugubris 16 0.26 0.04 123.1 655.9 0.02 0.62 0.1 60.1 0.01 0.77 0.0 60.1
L. sp. nov. 8 0.00 0.87 34.9 6201.8 0.01 0.84 0.0 60.2 0.00 0.97 0.1 60.3
Heteropoda venatoria 16 0.28 0.04 290.1 6125.5 0.01 0.74 0.0 60.2 0.01 0.81 0.0 60.2
Web-building spiders 16 0.05 0.41 100.4 6119.3 0.09 0.23 0.1 60.1 0.06 0.36 0.2 60.1
Notes: Significant effects are shown in boldface. Regression coefficients are shown 6SE.
was the best predictor of variation of FCL (R
¼0.64, P
,0.0001; Fig. 1). Ftest results indicated that the
addition of ecosystem size (F
¼1.50, P¼0.23) or of
productive space (F¼2.39, P¼0.14) to the linear model
did not significantly improve upon the model based on
productivity alone. Use of alternative values for
fractionation, or alternative metrics of productivity, do
not substantively change the conclusions about drivers
of FCL, but do alter the magnitude of total FCL change
observed (Appendix B)
Multinomial logistic regressions showed no relation-
ship between productivity and the identity of the apex
predators (R
¼0.09, v
¼5.07, P¼0.17). Likewise,
adding the identity of the apex predator to the best
models of FCL did not improve overall model fit (F
0.39, P¼0.81). Measurements of the trophic position of
four top predators, two omnivores, and three herbivores
revealed strong increases in trophic position with
increasing productivity for all predators (Fig. 2A) and
a significant increase for one of the two omnivores with
the second omnivore, R. rattus, only marginally
nonsignificant and showing similar patterns (P¼0.06;
Fig. 2B), but no increase in trophic position for any of
the herbivores, as would be predicted given that the
trophic position of obligate herbivores is fixed (Fig. 2C).
There was no effect of productive space or ecosystem
size on trophic position of any consumer (Appendix B:
Table B1).
To understand how changes in productivity could
engender these changes in FCL we examined changes in
consumer abundance, body size, and diet. With regard to
consumer abundance, we found increased abundance of
insects on islets with higher productivity in all three of the
abundance measurements we employed (Table 1). In
contrast, there was no significant effect of either island
area or productive space on most metrics of abundance,
although productive space was positively correlated to
abundance of insects in blacklight surveys (Table 1). With
regard to species richness of insects in blacklight traps,
this was also significantly higher on more productive
islets; species richness was also positively correlated with
productive space, but not with ecosystem size (Table 1).
Species richness was analyzed only for the one trapping
method (blacklight trapping) where we had sufficiently
high taxonomic resolution to analyze diversity.
With regards to morphological or behavioral shifts in
consumers along the productivity gradient, we observed
significant increases in average body size with increasing
islet productivity levels for two of the four top predators
(or predator groups) we examined: Both L. lugubris and
the H. venatoria showed significant increases in body size
with productivity (Table 1). Smaller, web-building
spiders showed no significant increase in mean body
size with productivity (Table 1). Body size of L. sp. nov.
was not significantly correlated with any of the metrics
examined; however, this species is not as widespread at
Palmyra and thus islet sample size was low (n¼8). None
of the predator species showed any relationship of body
size to either productive space or island area.
Finally, we also saw direct evidence of increased intra-
guild predation in the geckos, Lepidodactylus spp., the
only taxa for which diet contents were readily examin-
able. These predatory Lepidodactylus spp. showed small
but significant increases in the frequency of occurrence
of other predators in their diet in high-productivity islets
as compared to low-productivity islets (13%vs. 0%of
individuals contained intra-guild predators in their
stomach contents on high- vs. low-productivity islets;
Fisher’s exact test, P¼0.025).
We found that productivity alone explained .60%of
the variation in FCL observed across our study system.
Neither ecosystem size nor productive space was
FIG. 1. Maximum trophic position by ecosystem properties. Maximum trophic position (highest mean trophic position of any
consumer species) was used to approximate food chain length in our study islets. (A) It had a strong positive relationship with
ecosystem productivity as estimated by g Nm
but showed no significant relationship with (B) ecosystem size (measured in
) or (C) productive space (the product of productivity and ecosystem size). Along the productivity gradient, omnivores (open
triangles) and herbivores (open diamonds) were found occupying the maximum trophic position only at low-productivity sites, but
there were no other significant overall relationships between the maximum trophic position of the islet and the identity of the
predator (Neoscona theisi, light-gray circles; Lepidodactylus spp., medium-gray circles; and Heteropoda venatoria, dark-gray circles).
HILLARY S. YOUNG ET AL.696 Ecology, Vol. 94, No. 3
effective in explaining FCL variability. This result
strongly contrasts with results from many other studies
that found ecosystem size to be a dominant factor in
explaining FCL (reviewed in Takimoto and Post 2012).
While the strong observed response of FCL to
productivity is highly consistent with the early energy
limitation hypotheses, the mechanisms suggested by our
results differ somewhat from those predicted by the
proponents of this theory (e.g., Hutchinson 1959), which
suggest that limitations to FCL by productivity are due
to the fact that low-productivity systems simply cannot
support top predator species because of the energy losses
associated with multiple trophic transfers. In contrast to
the expectations of the energy limitation hypothesis, we
found no difference in identity of top predators across
the productivity gradient and no evidence that the
addition of new top predators played any role in the
observed lengthening of food chains (as observed in Post
et al. [2000], and McHugh et al. [2010]).
If new predators were not added at more productive
sites, the question remains: How were the pronounced
increases in FCL that we observed achieved? The
increases in trophic position within the same species of
predators and omnivores (but not of herbivores) across
the productivity gradient strongly suggest that changes
in FCL took place due to a series of changes within the
existing food web network rather than by new additions
to the top of the food web (consistent with Post and
Takimoto [2007], Post et al. [2007], Takimoto et al.
[2008], Sabo et al. [2010]). Likewise, the strong increases
in total insect abundance we observed in high-produc-
tivity systems using multiple survey techniques suggests
that the increased productivity is directly stimulating
consumer abundance, providing opportunity for food
chain lengthening. This is consistent with early concep-
tual understandings of energetic constraints to FCL
(e.g., Schoener 1989), with more energetically rich islets
supporting more animals and ultimately longer interac-
tion chains. To understand how such changes in
productivity and consumer abundance might translate
to increased FCL, we examined two avenues of change
in food web structure that might change total FCL
without changes in apex predator identity (the additive
mechanism): (1) the insertion mechanism, in which
increases in diversity of consumers in high-productivity
sites supports longer food webs; and (2) ominivory
mechanisms, in which consumers shift in their amount
of omnivory, achieved either by changes in morphology
of consumers facilitating consumption of larger, poten-
tially higher trophic-level prey items, or by behavioral
shifts alone, with consumers targeting higher trophic-
level prey (Appendix C: Fig. C2).
We found some support for each of the two
mechanisms considered. With regards to the ominvory
mechanism, we found increases in predator body size
with increasing productivity, although not for all
FIG. 2. Average trophic position by consumer taxa and ecosystem properties. Average trophic position of a given consumer on
each islet is strongly correlated to productivity (as estimated by g Nm
) for (A) all four predators and for (B) one of the two
omnivores (results are only marginally nonsignificant for R. rattus), but (C) not for any of the measured herbivores. Significant
relationships are marked with an asterisk.
predators. Two of the consumers for which body size
was examined (L. lugubris and H. venatoria)had
significantly larger sizes in more productive islands.
Results were not significant for Lepidodactylus sp. nov.,
but this may be due to the very small sample size for this
consumer; they were also not significant for web-
building spiders, which are less likely to have increased
hunting efficacy with increased body size. An increased
body size of predators could allow a given species of
predator to realize a higher trophic position if it allows
them to increase the size and type of prey they target: By
increasing intra-guild predation and cannibalism, larger
animals can operate functionally like the addition of a
new species (Woodward et al. 2005). Increases in trophic
position with increasing body size have been shown for
many predators, particularly smaller bodied ones
(Cohen et al. 1993, France et al. 1998, Woodward and
Hildrew 2002). We also found significant, albeit small,
changes in the diet of predators for the one group where
it was examined. Lepidodactylus spp. had higher
proportions of intra-guild predators in their stomach
contents when found on high-productivity islets as
compared to low-productivity islets. Sample size was
limited in this analysis, and more extensive diet analysis
in this group of consumers and others would help
confirm this conclusion. However, these results are
consistent with those observed in other studies. Preda-
tors and omnivores have previously been shown to
exhibit greater diet specialization, selectively feeding on
higher trophic-level prey, in more productive systems
(Arim et al. 2007, Stenroth et al. 2008). Increases in body
size may drive this change, as larger body size facilitates
cannibalism and intra-guild predation (Cohen et al.
1993, Woodward and Hildrew 2002).
With regard to the insertion mechanism, in our
examination of changes in consumer diversity with
productivity, we found support for the hypothesis that
diversity may also be elevated in more productive islets
in this system, potentially providing internal diversifica-
tion of food webs. While a wide variety of productivity–
diversity relationships have been observed and the
universality of these relationships is much contested
(Adler et al. 2011), there is some suggestion that this
relationship may be generally positive in terrestrial
tropical systems (Partel et al. 2007). While increases in
diversity need not be correlated with increased FCL,
increases in functional diversity, particularly within
intermediate consumers, is a major proposed mechanism
for increases in FCL (Post and Takimoto 2007). Such
increased FCL in other systems (e.g., Vander Zanden
et al. 1999, Parker and Huryn 2006, McHugh et al.
2010). When such a positive relationship occurs, it will
likely lead to stronger FCL–productivity relationships.
While both insertion and omnivory mechanisms are
likely to result in small individual shifts in trophic
position, when taken cumulatively across multiple steps
in the food web, the additive effect could cause large
effects on FCL without requiring changes in the identity
of the apex predators. However, much more detailed
studies of diet and behavior of consumers in this system
and others would be needed to document the relative
importance of each of these pathways, and to ascertain if
these two mechanisms are sufficiently powerful to
explain observed changes on the productivity gradient.
Equally intriguing as the strong positive response of
FCL to productivity is the lack of any effect of any
metric of ecosystem size (either directly or integrated
into productive space) on FCL. While this lack of
response to ecosystem size was observed in a global
review of FCL and ecosystem size (Vander Zanden and
Fetzer 2007), it is at odds with a robust literature
documenting strong effects of ecosystem size or produc-
tive space on FCL in other, predominantly freshwater,
systems (Takimoto and Post 2012). Indeed, not only is
there no suggestion of any trend toward increasing FCL
with ecosystem size at Palmyra, but some of the largest
islets in the atoll complex actually have some of the
lowest measured FCLs. Why did an explanation that
plays such an obvious and important role in controlling
the food chain length of lakes fail to be important in this
terrestrial context? We argue that the explanation likely
derives from several fundamental differences between
the terrestrial context in which we conducted this study
and other aquatic systems in which the majority of other
work has been conducted.
There are many systematic differences documented, or
hypothesized, in the functioning of terrestrial and
aquatic ecosystems that could account for the variation
of our results from those derived primarily from
freshwater systems (Chase 2000, Shurin et al. 2006).
Here, we focus upon three such potential pathways.
First, differences in feeding constraints in consumers in
aquatic vs. terrestrial systems may make terrestrial
systems less likely to be constrained by size limitations
of predators. In aquatic systems, and particularly for
fish, feeding (particularly maximum prey size) is often
constrained by gape size ratios leading to strongly size-
structured food webs (Jennings et al. 2001). In compar-
ison, terrestrial systems, where grasping and behavioral
innovations (e.g., spider webs) may allow greater
predation within a size class and within a feeding guild
(Hairston and Hairston 1993, Brose et al. 2006), large
size may not be needed to obtain a high trophic position.
Consistent with this mechanism, freshwater systems
show higher predator : prey body size ratios than
terrestrial systems (marine systems are similar to
terrestrial systems in this regard, but this may be
confounded by a preponderance of benthic samples in
these compendiums of marine data which often are not
as heavily fish dominated; Brose et al. 2006). Thus, in
terrestrial systems, relatively small animals, with smaller
energetic needs and home ranges, could occupy higher
trophic positions, potentially dampening the influence
that increases in ecosystem size have on FCL. However,
this difference is not absolute; gape-limited predators in
HILLARY S. YOUNG ET AL.698 Ecology, Vol. 94, No. 3
terrestrial systems should still gain in trophic position
with increased body size. There is some evidence for this
in our results from Palmyra. Lepidodactylus (geckos,
which are likely gape limited) and H. venatoria (a spider
species that does not build webs) both showed body size
increases on more productive islets. By contrast, as
might be expected, primarily web-building spiders did
not show size increases, but still showed similar changes
in trophic position. In summary, it seems that, while
body size may matter less, on average, for terrestrial
system FCL than in aquatic systems, it may still play an
important role in increasing trophic position of individ-
ual consumers and thus increasing overall FCL.
A second possible systemic difference between aquatic
and terrestrial systems of importance to FCL may be in
relationships between productivity and diversity. If
productivity–diversity relationships are generally posi-
tive in terrestrial tropical systems (Partel et al. 2007), as
appears may be the case in this study system too, but
unimodal or nonexistent in many aquatic systems
(Mittelbach et al. 2001), and if changes in species
diversity increases FCL, such variation in productivity–
diversity relationships could explain variation in impor-
tance of productivity as a driver in FCL.
Another final possible important difference between
freshwater and terrestrial systems in driving FCL is in
the nature of ecosystem boundaries and the relative size
of systems, potentially explaining the lack of observed
response of FCL to productive space. Both productivity
and productive space as drivers of FCL rest on similar
theoretical underpinnings: principally that increasing the
amount of energy or limiting resources in a system
should increase the ability of the system to support top
predators. Since productive space encapsulates whole-
system energy availability, it is intuitively more con-
nected to the system-wide metric of FCL than is the per
unit area measurement of productivity. However, at
some spatial scale, which should greatly exceed top
predator home range size, predators should not be
affected by further increases in productive space, as no
additional energy would be available to them with
increasing ecosystem size. In such cases only the per unit
area productivity should matter. We suggest that this
may explain the lack of response of predators to
productivity in this system: Because Palmyra’s top
predators (geckos, spiders), as well as many of the other
consumers in this system, have very small home range
sizes, increases in islet size beyond a very small level are
unlikely to increase the energy resources available to top
predators. Thus, even though Palmyra’s islets are very
small, because the atoll, for biogeographic reasons, has
few far-ranging animals, the relationship between
predator home range and ecosystem size may be more
similar to that found in many natural terrestrial systems
(e.g., Arim et al. 2007), and may deviate from that of
many lacustrine systems of similar area. Also of
importance in terrestrial systems is that many predators
(e.g., migratory passerines and birds of prey, notably
PLATE 1. The study system of Palmyra Atoll. (A) Palmyra consists of 23 islets ranging from (B) 0.05 ha to over 250 ha. The
islets have relatively simple food chains with various species of (C) geckos and spiders as top predators. (D) Islets span a tenfold
gradient in productivity that is driven by variation in seabird density and associated nutrient inputs. A color version of the plate is
available in Appendix D. Photo credits: (A) K. Pollock, (B) G. Carol, (C) S. Hathaway, (D) H. Young.
lacking from Palmyra) will regularly use multiple
habitats, easily crossing spatial and energetic boundaries
(McCauley et al. 2012), thus making ecosystem size
more difficult to delimit. Freshwater systems are, of
course, also intimately energetically linked to surround-
ing terrestrial habitats by predators (e.g., raccoons,
ospreys, bears) and a great deal of nutrient movement
(e.g., litterfall, leaching), making measurements of
ecosystem size difficult even in these most cleanly
demarcated systems. However, the boundaries are
generally much clearer in most freshwater systems than
in many terrestrial and marine habitats, where it is often
challenging to conceptualize productive space (or
ecosystem size; Post et al. 2007). The lack of response
of FCL to ecosystem size or productive space in this
terrestrial system and in a global review (Vander Zanden
and Fetzer 2007) may thus reflect underlying differences
between terrestrial and freshwater systems in the
importance of size constraints on food web structure
and FCL. However, these ideas will require further
substantiation in other terrestrial settings, particularly in
continental environments or in other contexts in which
the scale of ecosystem size exceeds that explored in this
The observations we report from this study of
terrestrial ecosystems contribute to our growing aware-
ness that there is no universal driver of FCL and suggest
that considerations of ecological context must be taken
into account when identifying dominant drivers. Our
results indicate that productivity may play an underap-
preciated role in structuring FCL in certain ecosystems,
particularly in terrestrial settings, consistent with early
energetic-limitation hypotheses. By providing some
preliminary mechanistic support for how such changes
in productivity can effect changes in FCL, these results
offer insight into how different drivers may achieve
primacy in various systems. While it seems likely that
drivers of FCL vary both within and among systems, the
responses observed here seem inconsistent with the
spectrum of responses observed in most aquatic systems
(e.g., Takimoto and Post 2012). However, there may be
significant overlap in drivers of FCL across systems.
Extending these studies to other terrestrial systems, as
well as to other larger systems not constrained by size
and dispersal limitations, will be important to under-
standing in what contexts productivity is likely to have
its strongest impact on trophic architecture. Yet,
cumulatively, these observations suggest that some of
the drivers of FCL that have been derived from aquatic
systems, and held to be largely universal, may not have
the same explanatory power in terrestrial settings.
We thank the National Science Foundation (#0639185), the
Woods Institute for the Environment, the National Geographic
Society, the Bishop Museum, and the U.S. Fish and Wildlife
Service for support on this project. We are grateful to P.
Vitousek, N. Cormier, D. Croll, R. Fisher, S. Hathaway, C.
France, C. Chu, and A. A. Briggs for data, advice, and field
assistance. We also thank D. Post, and three anonymous
reviewers for their careful reviews of earlier versions of the
manuscript, which greatly improved this paper. Vector images
are courtesy of Integration and Application Network, Univer-
sity of Maryland Center for Environmental Science (ian.umces.
edu/imagelibrary). This is publication number PARC-0092 of
the Palmyra Atoll Research Consortium.
Adler, P. B., E. W. Seabloom, E. T. Borer, H. Hillebrand, Y.
Hautier, A. Hector, W. S. Harpole, L. R. O’Halloran, J. B.
Grace, and T. M. Anderson. 2011. Productivity is a poor
predictor of plant species richness. Science 333:1750–1753.
Arim, M., P. A. Marquet, and F. M. Jaksic. 2007. On the
relationship between productivity and food chain length at
different ecological levels. American Naturalist 169:62–72.
Borer, E. T., E. W. Seabloom, J. B. Shurin, K. E. Anderson,
C. A. Blanchette, B. Broitman, S. D. Cooper, and B. S.
Halpern. 2005. What determines the strength of a trophic
cascade? Ecology 68:528–537.
Briand, F., and J. E. Cohen. 1987. Regulation of lake primary
productivity by food web structure. Ecology 68:1863–1867.
Brose, U., et al. 2006. Consumer-resource body-size relation-
ships in natural food webs. Ecology 87:2411–2417.
Cabana, G., and J. B. Rasmussen. 1994. Modeling food chain
structure and contaminant bioaccumulation using stable
nitrogen isotopes. Nature 372:255–257.
Calcagano, V., F. Massol, N. Mouquet, P. Jarne, and P. David.
2011. Constraints on food chain length arising from regional
metacommunity dynamics. Proceedings of the Royal Society
B 1721:3042–3049.
Catenazzi, A., and M. A. Donnelly. 2007. The Ulva connection:
marine algae subsidize terrestrial predators in coastal Peru.
Oikos 116:75–86.
Chase, J. M. 2000. Are there real differences among aquatic and
terrestrial food webs? Trends in Ecology and Evolution
Cohen, J. E., S. L. Pimm, P. Yodzis, and J. Salda ˜
na. 1993. Body
sizes of animal predators and animal prey in food webs.
Journal of Animal Ecology 62:67–78.
DeAngelis, D. L., S. M. Bartell, and A. L. Brenkert. 1989.
Effects of nutrient recycling and food chain length on
resilience. American Naturalist 134:778–805.
Doi, H., K. K. Chang, T. Ando, I. Ninomiya, H. Imai, and S.
Nakano. 2009. Resource availability and ecosystem size
predict food chain length in pond ecosystems. Oikos
Elton, C. 1927. Animal ecology. Sidgwick and Jackson,
London, UK.
ESRI. 2008. ArcGIS. Environmental Systems Research Insti-
tute, Redlands, California, USA.
France, R., M. Chandler, and R. Peters. 1998. Mapping trophic
continua of benthic food webs: body size–d
N relationships.
Marine Ecology Progress Series 174:301–306.
Fry, B., and E. B. Sher. 1984. d
C measurements as indicators
of carbon flow in marine and freshwater ecosystems.
Contributions in Marine Science 27:13–47.
Hairston, N. G., and N. G. Hairston. 1993. Cause-effect
relationships in energy-flow, trophic structure, and interspe-
cific interactions. American Naturalist 142:379–411.
Hutchinson, G. E. 1959. Homage to Santa Rosalia; or why are
there so many kinds of animals? American Naturalist 93:145–
Jenkins, B., R. L. Kitching, and S. L. Pimm. 1992. Productivity
and food web structure at a local spatial scale in experimental
container habitats. Oikos 65:249–255.
Jennings, S., J. K. Pinnegar, N. V. C. Polunin, and T. W.
Boonin. 2001. Weak cross-species relationships between body
size and trophic level belie powerful size-based trophic
HILLARY S. YOUNG ET AL.700 Ecology, Vol. 94, No. 3
structuring in fish communities. Journal of Animal Ecology
70:934– 944.
Kaunzinger, C. M. K., and P. J. Morin. 1998. Productivity
controls food chain properties in microbial communities.
Nature 395:495–497.
Lindeman, R. L. 1942. The trophic-dynamic aspect of ecology.
Ecology 23:399–418.
Marty, J., K. Smokorowski, and M. Power. 2009. The influence
of fluctuating ramping rates on the food web of boreal rivers.
River Research and Applications 25:962–974.
McCauley, D. J., H. S. Young, R. B. Dunbar, J. A. Estes, B. X.
Semmens, and F. Micheli. 2012. Assessing the effects of large
mobile predators on ecosystem connectivity. Ecological
Applications 22:1711–1717.
McHugh, P. A., A. R. McIntosh, and P. G. Jellyman. 2010.
Dual influences of ecosystem size and disturbance on food
chain length in streams. Ecology Letters 13:881–890.
McIntyre, P. B., L. E. Jones, A. S. Flecker, and M. J. Vanni.
2007. Fish extinctions alter nutrient recycling in tropical
waters. Proceedings of the National Academy of Sciences
USA 104:4461–4466.
Mittelbach, G. G., C. F. Steiner, S. M. Scheiner, K. L. Gross,
H. L. Reynolds, R. B. Waide, M. R. Willig, S. I. Dodson, and
L. Gough. 2001. What is the observed relationship between
species richness and productivity? Ecology 82:2381–2396.
Parker, S. M., and A. D. Huryn. 2006. Food web structure and
function in two arctic streams with contrasting disturbance
regimes. Freshwater Biology 51:1249–1263.
Partel, M., L. Laanisto, and M. Zobel. 2007. Contrasting plant
productivity-diversity relationships across latitude: the role
of evolutionary history. Ecology 88:1091–1097.
Pimm, S. L. 1982. Food webs: population and community
biology series. Chapman and Hall, London, UK.
Post, D. M. 2002. The long and short of food chain length.
Trends in Ecology and Evolution 17:269–277.
Post, D. M. 2007. Testing the productive-space hypothesis:
rational and power. Oecologia 153:973–984.
Post, D. M., M. W. Doyle, J. L. Sabo, and J. C. Finlay. 2007.
The problem of boundaries in defining ecosystems: A
potential landmine for uniting geomorphology and ecology.
Geomorphology 89:111–126.
Post, D. M., M. L. Pace, and N. G. Hairston. 2000. Ecosystem
size determines food chain length in lakes. Nature 405:1047–
Post, D. M., and G. Takimoto. 2007. Proximate structural
mechanisms for variation in food chain length. Oikos
R Development Core Team. 2012. R: a language and
environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria.
Sabo, J. L., J. C. Finlay, T. Kennedy, and D. M. Post. 2010.
The role of discharge variation in scaling of drainage area
and food chain length in rivers. Science 12:965–967.
Schoener, T. W. 1989. Food webs from the small to the large.
Ecology 70:1559–1589.
Shurin, J. B., E. S. Borer, E. W. Seabloom, K. Anderson, C. A.
Blanchette, B. Broitman, S. D. Cooper, and B. S. Halpern.
2002. A cross-ecosystem comparison of the strength of
trophic cascades. Ecology Letters 5:785–791.
Shurin, J. B., D. S. Gruner, and H. Hillebrand. 2006. All wet or
dried up? Real differences between aquatic and terrestrial
food webs. Proceedings of the Royal Society B 273:1–9.
Spencer, M., and P. H. Warren. 1996. The effects of habitat size
and productivity on food web structure in small aquatic
microcosms. Oikos 75:419–430.
Stenroth, P., N. Holmqvist, P. Nystro
¨m, O. Berglund, P.
Larsson, and W. Grande
´li. 2008. The influence of produc-
tivity and width of littoral zone on the trophic position of a
large-bodied omnivore. Oecologia 156:681–690.
Takimoto, G., and D. M. Post. 2012. Environmental determi-
nants of food-chain length: a meta-analysis. Ecological
Takimoto, G., D. A. Spiller, and D. M. Post. 2008. Ecosystem
size, but not disturbance, determines food chain length on
islands of the Bahamas. Ecology 89:3001–3007.
Thompson, R. M., M. Hemberg, B. M. Starzomski, and J. B.
Shurin. 2007. Trophic levels and trophic tangles: the
prevalence of omnivory in real food webs. Ecology 88:612–
Thompson, R. M., and C. R. Townsend. 2005. Energy
availability, spatial heterogeneity and ecosystem size predict
food-web structure in streams. Oikos 108:137–148.
Townsend, C. R., R. M. Thompson, A. R. McIntosh, C.
Kilroy, E. Edwards, and M. R. Scarsbrook. 1998. Distur-
bance, resource supply, and food-web architecture in
streams. Ecology Letters 1:200–209.
Vander Zanden, M. J., and W. W. Fetzer. 2007. Global patterns
of aquatic food chain length. Oikos 116:1378–1388.
Vander Zanden, M. J., B. J. Shuter, N. Lester, and J. B.
Rasmussen. 1999. Patterns of food chain length in lakes: a
stable isotope study. American Naturalist 154:406– 416.
Woodward, G., B. Ebenman, M. Emmerson, J. M. Montoya,
J. M. Olesen, A. Valido, and P. H. Warren. 2005. Body size in
ecological networks. Trends in Ecology and Evolution
Woodward, G., and A. G. Hildrew. 2002. Body size determi-
nants of niche overlap and intraguild predation within a
complex food web. Journal of Animal Ecology 71:1063–1074.
Young, H. S., D. J. McCauley, R. D. Dunbar, and R. Dirzo.
2010a. Plants cause ecosystem nutrient depletion via inter-
ruption of bird-derived spatial subsidies. Proceedings of the
National Academy of Sciences USA 107:2072–2077.
Young, H. S., T. K. Raab, D. J. McCauley, A. A. Briggs, and
R. Dirzo. 2010b. The coconut palm, Cocos nucifera, impacts
forest composition and soil characteristics at Palmyra Atoll,
Central Pacific. Journal of Vegetation Science 21:1058–1068.
Appendix A
Evaluating islet productivity (Ecological Archives E094-060-A1).
Appendix B
Calculating food chain length (Ecological Archives E094-060-A2).
Appendix C
Surveying consumer abundance, diversity, and body size (Ecological Archives E094-060-A3).
Appendix D
A color version of Plate 1, showing images from Palmyra atoll study system (Ecological Archives E094-060-A4).
... However, most evidence for FCL determinants is from freshwater ecosystems (e.g. Vander Zanden and Fetzer, 2007;Thompson and Townsend, 2005;Maceda-Veiga et al., 2018), which have more clearly defined boundaries than terrestrial or marine ecosystems; it is unclear how much ecosystem size affects FCL in terrestrial ecosystems (Young et al., 2013;Ward and McCann 2017). Another factor that might significantly affect FCL is the mobility of predators (Post et al., 2007;Takimoto et al., 2012). ...
... Last, the relatively poor support for the disturbance hypothesis (reviewed by Takimoto and Post, 2013) might be due to aquatic ecosystems being able to attenuate the impacts of disturbances on FCL because aquatic species often have wider trophic spectrum than terrestrial taxa (Hairston and Hairston, 1993;Thompson et al., 2007). Therefore, owing to these and other systematic differences between terrestrial and aquatic ecosystems (Shurin et al., 2002;Brose et al., 2006;Young et al., 2013), understanding the generalities of factors affecting FCL requires more study in terrestrial ecosystems. ...
... However, the ecological role of wild boar in the absence of traditional forest management needs further investigation to determine the degree to, and the mechanisms by which, wild boars and humans might be affecting forestarthropod food webs. Our findings are consistent with terrestrial ecosystems being less spatially restricted habitats than freshwater ecosystems, and so, with food webs and FCLs being less influenced by ecosystem size per se (Young et al., 2013). However, it is difficult to identify 'ecological boundaries' in forests because exchanges of energy and biota mean that apparent -to humans -physical boundaries may not correspond well with realized ecological boundaries (see also Post et al., 2007). ...
Food-chain length, or the trophic level of an apex predator, is among the most important properties of food-webs with implications for community structure, ecosystem processes and pollutant accumulation in forests. Three main hypotheses (ecosystem-size, productivity, and disturbance) have been erected to explain variation in food-chain length in freshwater ecosystems, yet the support for these hypotheses in less spatially restricted terrestrial ecosystems has not been extensively studied. Here, we used nitrogen (δ¹⁵N) and carbon (δ¹³C) stable isotopes to explore variation in the realized trophic positions (δ¹⁵N) of a beetle Carabus lineatus lateralis and a wasp Vespula vulgaris in 32 chestnut woodland patches in northwestern Spain, while accounting for the insects’ relative mobility by using inferences based on δ¹³C. We used five potential predictors of ecosystem-size productivity, seven of disturbance and six covariates in anticipation that biological assemblages in woodlands might be influenced by ecosystem-size productivity, and from a change from the human management to a much-increased activity of ecosystem engineers, especially wild boar. Our results provided support for the disturbance hypothesis and suggested that the beetle FCL seems to be more affected by wild boar disturbance than by human-forest management, possibly due to increased forest cover and to rural abandonment in recent decades. Moreover, we found a negative association between the wasp FCL and the ecosystem-size productivity hypothesis, as indicated by the plant Ellengberg’s indicator value for nitrogen, which contrasts to the hypothesis that ecosystem-size productivity should increase FCL. Our findings are discussed in relation to: (1) differences in intensity and frequency between human- and wild boar-induced disturbances; (2) the diets and mobilities of the two predators; and (3) the near lack of hard ecological boundaries in terrestrial ecosystems, such as chestnut woodlands.
... Nonetheless, because our 16 islands differed in size and shoreline:area ratio, we sought to correct for potential marine-derived subsidies. We assumed that our 16 study islands shared a common marine isotopic baseline 69 , which we estimated using average values from 11 macroalgae samples collected from locations across our 13-km string of study islands (Fig. 1d). A previous Bahamian study found that conclusions about trophic position and food-chain length were not sensitive to whether the marine isotopic baseline was estimated using macroalgae averaged across all islands (as we did) or particulate organic matter filtered from seawater at each island 30 , which provides further justification for our approach. ...
... In calculating island-wide food-chain length, we followed previous studies 25,28,30,69 in using the mean trophic-position value of the apical consumer population on each island (see 'Curly-tailed lizards as top predators'). Food-chain length was analysed using a generalized least-squares linear model with main effects of island area and the factorial green-anole and curly-tailed-lizard addition treatments, along with all first-order interactions between these variables (Extended Data Table 2). ...
Full-text available
Biological invasions are both a pressing environmental challenge and an opportunity to investigate fundamental ecological processes, such as the role of top predators in regulating biodiversity and food-web structure. In whole-ecosystem manipulations of small Caribbean islands on which brown anole lizards (Anolis sagrei) were the native top predator, we experimentally staged invasions by competitors (green anoles, Anolis smaragdinus) and/or new top predators (curly-tailed lizards, Leiocephalus carinatus). We show that curly-tailed lizards destabilized the coexistence of competing prey species, contrary to the classic idea of keystone predation. Fear-driven avoidance of predators collapsed the spatial and dietary niche structure that otherwise stabilized coexistence, which intensified interspecific competition within predator-free refuges and contributed to the extinction of green-anole populations on two islands. Moreover, whereas adding either green anoles or curly-tailed lizards lengthened food chains on the islands, adding both species reversed this effect—in part because the apex predators were trophic omnivores. Our results underscore the importance of top-down control in ecological communities, but show that its outcomes depend on prey behaviour, spatial structure, and omnivory. Diversity-enhancing effects of top predators cannot be assumed, and non-consumptive effects of predation risk may be a widespread constraint on species coexistence.
... The disturbance hypothesis, also termed as the dynamic constraints hypothesis, predicts that more frequent or more intense disturbance in ecosystems would shorten FCL, because longer chains are less resilient and thus unlikely to persist in disturbed habitats 13 . Still, among the common FCL hypotheses, the productivity hypothesis has been tested most frequently but with incongruent results from field and laboratory studies 13,14,19,20 . ...
... Recently, nitrogen stable isotope measurements have become the technique most often used for FCL determination, next to gut content analyses [12][13][14][15]20,21 . Nitrogen stable isotope composition reflects the trophic position of consumers 22 . ...
Full-text available
Food-chain length (FCL) is a fundamental ecosystem attribute, integrating information on both food web composition and ecosystem processes. It remains untested whether FCL also reflects the history of community assembly known to affect community composition and ecosystem functioning. Here, we performed microcosm experiments with a copepod (top predator), two ciliate species (intermediate consumers), and bacteria (producers), and modified the sequence of species introduction into the microcosm at four productivity levels to jointly test the effects of historical contingency and productivity on FCL. FCL increased when the top predator was introduced last; thus, the trophic position of the copepod reflected assembly history. A shorter FCL occurred at the highest productivity level, probably because the predator switched to feeding at the lower trophic levels because of the abundant basal resource. Thus, we present empirical evidence that FCL was determined by historical contingency, likely caused by priority effects, and by productivity.
... We expect that the particular environmental and biogeographic conditions of each ecosystem will translate into different network architecture (Song & Saavedra, 2020). We hypothesize that the warmer, more productive, with a more heterogeneous habitat Beagle Channel (Amin et al., 2011;Schloss et al., 2012) will present a more complex food web (Kortsch et al., 2018), measured as species richness (Duffy et al., 2017) and number of trophic interactions, hence, connectance, with more trophic levels (Young et al., 2013), higher omnivory (Thompson et al., 2007) and stronger modularity (Welti & Joern, 2015). It is expected for a more complex food web to present a lower stability against perturbations (May, 1973). ...
Full-text available
Food web structure plays an important role in determining ecosystem stability against perturbations. High-latitude marine ecosystems are being affected by environmental stressors and biological invasions. In the West Antarctic Peninsula these transformations are mainly driven by climate change, while in the sub-Antarctic region by anthropogenic activities. Understanding the differences between these areas is necessary to monitor the changes that are expected to occur in the upcoming decades. Here, we compared the structure and stability of Antarctic (Potter Cove) and sub-Antarctic (Beagle Channel) marine food webs. We compiled species trophic interactions (predator-prey) and calculated complexity, structure and stability metrics. Even if both food webs presented the same connectance, we found important differences between them. The Beagle Channel food web is more complex, but less stable and sensitive to the loss of its most connected species, while the Potter Cove food web presented lower complexity and greater stability against perturbations.
... In longer food chains, dragonfly larvae are intermediate predators often competing for resources with top predators, but in smaller ecosystems with shorter food-chain lengths they tend to be top predators with less competition for resources. Evidence that ecosystem size is a determinant of food chain length has not only been shown in lakes (Post et al. 2000), but in a variety of aquatic ecosystems (McHugh et al. 2010, Takimoto et al. 2008, Young et al. 2013. There is evidence from other freshwater ecosystems that trophic position of the same predatory species does change with increases in ecosystem size (McHugh et al. 2010). ...
Full-text available
Mercury (Hg) is a toxic pollutant, widespread in northeastern US ecosystems. Resource managers’ efforts to develop fish consumption advisories for humans and to focus conservation efforts for fish-eating wildlife are hampered by spatial variability. Dragonfly larvae can serve as biosentinels for Hg given that they are widespread in freshwaters, long-lived, exhibit site fidelity, and bioaccumulate relatively high mercury concentrations, mostly as methylmercury (88% ± 11% MeHg in this study). We sampled lake water and dragonfly larvae in 74 northeastern US lakes that are part of the US EPA Long-Term Monitoring Network, including 45 lakes in New York, 43 of which are in the Adirondacks. Aqueous dissolved organic carbon (DOC) and total Hg (THg) were strongly related to MeHg in lake water. Dragonfly larvae total mercury ranged from 0.016–0.918 μg/g, dw across the study area; Adirondack lakes had the minimum and maximum concentrations. Aqueous MeHg and dragonfly THg were similar between the Adirondack and Northeast regions, but a majority of lakes within the highest quartile of dragonfly THg were in the Adirondacks. Using landscape, lake chemistry, and lake morphometry data, we evaluated relationships with MeHg in lake water and THg in dragonfly larvae. Lakewater DOC and lake volume were strong predictors for MeHg in water. Dragonfly THg Bioaccumulation Factors (BAFs, calculated as [dragonfly THg]:[aqueous MeHg]) increased as lake volume increased, suggesting that lake size influences Hg bioaccumulation or biomagnification. BAFs declined with increasing DOC, supporting a potential limiting effect for MeHg bioavailability with higher DOC.
... Given that trophic limitation affects community dynamics, understanding how island ecosystems are regulated is therefore critical (Terborgh 2010). Island ecosystems appear to be limited differently from continental ecosystems (e.g., Polis & Hurd 1995, Gruner 2004, Young et al. 2013, and this is likely an outcome of their smaller areas and associated reduced productivity, but island studies on trophic dynamics and productivity that would enable us to fully understand the differences are scarce (Yoshida 2008, Terborgh 2010. ...
The rate of non-native species introductions continues to increase, with directionality from continents to islands. It is no longer single species but entire networks of coevolved and newly interacting continental species that are establishing on islands. The consequences of multispecies introductions on the population dynamics and interactions of native and introduced species will depend on the form of trophic limitation on island ecosystems. Freed from biotic constraints in their native range, species introduced to islands no longer experience top-down limitation, instead becoming limited by and disrupting bottom-up processes that dominate on resource-limited islands. This framing of the ecological and evolutionary relationships among introduced species with one another and their ecosystem has important consequences for conservation. Whereas on continents the focus of conservation is on restoring native apex species and top-down limitation, on islands the focus must instead be on removing introduced animal and plant species to restore bottom-up limitation. Expected final online publication date for the Annual Review of Ecology, Evolution, and Systematics Volume 50 is November 4, 2019. Please see for revised estimates.
... Primary production is a fundamental driver of ecosystem complexity and function, with higher productivity linked to higher species diversity [1][2][3], secondary production [4,5], longer food chains [6,7] and more complex food webs [8]. Most primary production, however, is not consumed by herbivores but becomes detritus that may vary in its fate, residence time and lability [9,10]. ...
Full-text available
Detritus can fundamentally shape and sustain food webs, and shredders can facilitate its availability. Most of the biomass of the highly productive giant kelp, Macrocystis pyrifera, becomes detritus that is exported or falls to the seafloor as litter. We hypothesized that sea urchins process kelp litter through shredding, sloppy feeding and egestion, making kelp litter more available to benthic consumers. To test this, we conducted a mesocosm experiment in which an array of kelp forest benthic consumers were exposed to 13C- and 15N-labelled Macrocystis with or without the presence of sea urchins, Strongylocentrotus purpuratus. Our results showed that several detritivore species consumed significant amounts of kelp, but only when urchins were present. Although they are typically portrayed as antagonistic grazers in kelp forests, sea urchins can have a positive trophic role, capturing kelp litter before it is exported and making it available to a suite of benthic detritivores.
... Populations on higher trophic levels would thus be less prone to stochastic extinctions and longer food-chains would be feasible (Yodzis 1984). Despite this simple logic behind the hypothesis, initial studies to clarify the question found that food-chain length was independent of productivity (Briand & Cohen 1987) and the results of later studies varied, with some authors confirming the relationship of food-chain length and productivity (Carpenter et al. 1987;Persson et al. 1992;Young et al. 2013) and others that found no such relationship (Spencer & Warren 1996;Wootton et al. 1996). Plant defensive mechanisms and inedible plant parts like structural carbon components were considered to hamper a consumption of the total primary productivity by some authors and studies. ...
Full-text available
Food webs are the dynamic structure of natural ecosystems. Island biogeographical and metacommunity perspectives have aided food web research in spatially delimiting and opening the the studied object at the same time. While islands and habitat fragments have become empirical representations of spatially delimited focal habitat patches and their dynamics, considering the dispersal of organisms between them has opened new perspectives on community assembly and the maintenance of biodiversity. Furthermore, the spillover of organisms between adjacent habitats has been integrated into studying the dynamics of local ecosystem processes. These ‘spatial subsidies’ can generate strong interdependencies between the food webs of adjoined habitats, as different as terrestrial and aquatic ones. When studying the effects of ecosystem size and spatial subsidies on community and food web properties, islands have a prominent role in testing appropriate ecological hypotheses. In this study, the effect of ecosystem size in determining the food chain length of arthropod communities was investigated on young man-made lake islands of restored lakes in the Lower Rhine area. Moreover, it was analyzed how the importance of spatial subsidies for local food webs varied with island area and which role dispersal and adaptations to local conditions have in determining the community composition of large and small islands. Food chain length and spatial subsidies were tracked with a stable isotope approach. A strong relationship between food chain length and island area could be shown, while the importance of spatial subsidies, which was restricted to a few species, decreased with island size. It could be shown, that the species richness of spiders and ground beetles tracked an increasing spatial heterogeneity with increasing island size and arthropods were significantly assorted to spatially varying environmental conditions.
Full-text available
Invasive rodent eradications are frequently undertaken to curb island biodiversity loss. However, the breadth of rodents' ecological impact, even after eradication, is not always fully recognized. For example, the most widespread invasive rodent, the black rat (Rattus rattus), while omnivorous, eats predominantly seeds and fruit. Yet, the effects of seed predation release after eradication on plant communities and ecological functions are not well understood, posing a gap for island restoration. We examined the role of seed predation release following black rat eradication in changes to tree composition and aboveground biomass across an islet network (Palmyra Atoll) in the Central Pacific. We conducted repeated surveys of seed, juvenile, and adult tree biomass and survival in permanent vegetation plots before and after the eradica-tion of rats. We observed a 95% reduction in seed predation for an introduced, previously cultivated tree population (Cocos nucifera). Juvenile tree biomass of all species increased 14-fold, with C. nucifera increasing the most, suggesting that eradication increased this tree's competitive advantage. Indeed, based on stage-structured demographic models, rat eradication led to a 10% increase in C. nucifera population growth rate. The effect of invasive rodent seed predation varies considerably among the plant species in a community and can shift competitive dynamics, sometimes in favor of invasive plants. These bottom-up effects should be considered in evaluating the costs and benefits of eradication. Documenting the variation in invasive rodent diet items, along with long-term surveys, can help prioritize island eradications where restoration is most likely to be successful.
The origin of variation in food chain length (FCL) has long been a central topic of ecology due to its influences on trophic control, energy flow and bioaccumulation of environmental contaminants. The positive association of FCL and ecosystem size has been consistently supported in natural systems. Recent studies, however, have begun to report no effects of ecosystem size on FCL. The emerging inconsistency among previous studies implies the existence of overlooked factors that can modulate the relationship between FCL and ecosystem size. Here, we developed a mathematical model describing metacommunity dynamics to show that large ecosystems do not necessarily support long food chains in the presence of spatially correlated disturbance. Our simple model predicted that the effect of ecosystem size was strongly dependent on the spatial extent of disturbance synchrony. When the spatial correlation of disturbance was weak, the positive association between FCL and ecosystem size appeared. However, the relationship of FCL and ecosystem size became more complex, including hump‐shaped and multimodal forms, as the spatial extent of disturbance synchrony increases. This pattern emerged because larger ecosystems have wider aerial coverage, so more chances exist that an episodic disturbance hits some of the habitat patches and spreads across the landscape. Our finding highlights the important role of spatial disturbance synchrony in driving FCL, providing insights into why ecosystem size effects are variable across systems. We developed a mathematical model of meta community dynamics that predict food chain length in the presence of spatial disturbance synchrony
Full-text available
We performed whole-lake manipulations of fish populations to test the hypothesis that higher trophic levels regulate zooplankton and phytoplankton community structure, biomass, and primary productivity. The study involved three lakes and spanned 2 yr. Results demonstrated hierarchical control of primary production by abiotic factors and a trophic cascade involving fish predation. In Paul Lake, the reference lake, productivity varied from year to year, illustrating the effects of climatic factors and the natural dynamics of unmanipulated food web interactions. In Tuesday Lake, piscivore addition and planktivore reduction caused an increase in zooplankton biomass, a compositional shift from a copepod/rotifer assemblage to a cladoceran assemblage, a reduction in algal biomass, and a continuous reduction in primary productivity. In Peter Lake, piscivore reduction and planktivore addition decreased zooplanktivory, because potential planktivores remained in littoral refugia to escape from remaining piscivores. Both zooplankton biomass and the dominance of large cladocerans increased. Algal biomass and primary production increased because of increased concentrations of gelatinous colonial green algae. Food web effects and abiotic factors were equally potent regulators of primary production in these experiments. Some of the unexplained variance in primary productivity of the world's lakes may be attributed to variability in fish populations and its effects on lower trophic levels.
Full-text available
The attempt to explain the observed structure of ecological food webs has been one of the recent key issues of theoretical ecology. Unquestionably, many factors are involved in determining food-web structure. The dissipation of available energy from one trophic level to the next has been emphasized by Yodzis as the major factor limiting the length of food chains. However, Pimm and Lawton and Pimm have argued that a decrease in relative stability with increasing food-chain length may also be a factor. By relative stability (more commonly, resilience), we mean the rate at which a stable ecological system returns to a steady state following a perturbation. Resilience can be defined more precisely as the inverse of the return time T{sub R}, the time it takes a systems to return a specified fraction of the way toward a steady state following a perturbation. Besides its possible significance to food-web structure, ecosystem resilience is a factor of practical importance, since it is a measure of the rate at which the ecosystem can recover from disturbances. Our purpose is to investigate resilience in food-chain and food-web models as nutrient input and the trophic structure are varied and to offer explanations of the observed model behaviors. In this paper we present the basic results by first using a simple abstract food-chain model at steady state and then showing that these results hold for a more complex food-web simulation model without a constant steady state solution.
Full-text available
We studied food webs comprising fish, macroinvertebrates, and algae (identified to species or morphospecies) in small streams using a consistent methodology at the same spatial and temporal scales. Our aim was to test a priori hypotheses derived from dynamic-demographic and energetics models concerning the effects of disturbance and resource availability on food-web attributes. The regime of bed disturbance affecting the organisms in the webs was measured in 10 streams. We also derived measures of the supply of resources for animals in the webs in terms of algal primary productivity and detritus standing crop. Both web size and number of links per species were significantly negatively related to mean intensity of bed disturbance. Mean chain length had a significant positive relationship with algal primary productivity but not disturbance. No food-web attribute was related to detritus standing crop.
Full-text available
THE nitrogen pools of animals are enriched in 15N relative to their food1, with the top predators having the highest concentrations of this stable isotope2. The use of δ15N to indicate trophic position depends on the degree to which it reflects variation in the underly-ing food-web structure, rather than variable fractionation along the food chain. Here we compare adult lake trout, a top pelagic predator, from a series of lakes, and find that δ15N values vary from 7.5 to 17.5%o, a surprisingly wide range for one species. The length of the food chain can explain this variation, supporting the idea that δ15N is a food-web descriptor. Food-chain length was measured by the presence or absence of two intermediate trophic levels, pelagic forage fish and the macrozooplankter, Mysis relicta, each of which when present contributes about three δ15N units to the trout signature. We find that δ15N can be used as a continuous, integrative measure of trophic position, which is supported by its correlation to mercury levels in lake trout.
We discuss two mechanisms by which habitat size could affect food web structure: productive space and spatial effects on the persistence of unstable interactions. We assembled communities of bacteria, protozoa, rotifers, microcrustacea and Hydra in laboratory microcosms, and independently manipulated habitat size and total energy input, in order to distinguish between these mechanisms. After 163 d, larger habitats supported food webs with more species, more links per species and longer food chains, even in the absence of differences in total energy input. There were no significant differences in food web structure between energy treatments. Some species' relative abundances were affected by habitat size or energy, but there was no consistent overall pattern of responses. These results do not support the productive space hypothesis, but suggest that spatial effects on the persistence of unstable food webs may be important, although differences in disturbance between different sizes of microcosm may also have been a factor.
Ecologists have long struggled to explain variation in food-chain length among natural ecosystems. Food-chain length is predicted to be shorter in ecosystems subjected to greater disturbance because longer chains are theoretically less resilient to perturbation. Moreover, food-chain length is expected to be longer in larger ecosystems because increasing ecosystem size increases species richness and stabilizes predator-prey interactions, or increases total resource availability. Here we test the roles of disturbance and ecosystem size in determining the food-chain length of terrestrial food webs on Bahamian islands. We found that disturbance affected the identity of top predators, but did not change food-chain length because alternative top predators occupied similar trophic positions. On the other hand, a 106- fold increase in ecosystem size elevated food-chain length by one trophic level. We suggest that the effect of disturbance on food-chain length is weak when alternate top predators are trophic omnivores and have similar trophic positions. This and previous work in lakes suggest that ecosystem size may be a strong determinant of food-chain length in both aquatic and terrestrial ecosystems.
The stable nitrogen isotopic composition (δ15N) of benthic foodwebs was analyzed to determine trophic relationships among 23 taxa composited from 4 proximal oligotrophic lakes and 35 taxa composited from 2 proximal seagrass meadows. Because omnivory was prevalent, animals in both foodwebs existed along continua of trophic positions rather than in discrete trophic levels. These trophic continua were in turn related to organism body size in support of Elton's early suppositions about foodweb structuring. These foodweb delineations developed due to feeding hierarchies being determined by size-related predation, resulting in 'upper triangular' foodweb matrices as posited by Cohen's cascade model. Expressing aquatic foodweb structure in relation to the ataxonomic continuous variables of organismal trophic position (δ15N) and body size has the advantage of providing a convenient common currency for comparative purposes that dispenses with the artificiality and impossibility of constraining omnivorous animals into rigidly defined trophic levels.
Stable isotope ratios provide clues about the origins and transformations of organic matter. A few key reactions control the isotopic composition of most organic matter. Isotopic variations introduced by these reactions are often passed on with little change so that isotopic measurements can indicate natural pathways and flows “downstream” from these key reactions. When chemical and metabolic processes scramble the information content of molecules, isotopic compositions are often preserved. This realization has prompted increasing use of stable isotope analyses as a tool for understanding complex ecological processes.