Printer Friendly

Revisiting the energetic efficiency hypothesis: body mass, metabolism, and food chain length.


For more than a century, ecologists have sought to understand the forces determining food chain lengths in natural communities (Post and Takimoto, 2007) and to explain why food chain lengths differ between aquatic and terrestrial systems (Chase, 2000; Shurin et al., 2006). During that time multiple explanations have been proposed. These include species richness (alone or in combination with ecosystem size; Cohen and Newman, 1991; Post et al., 2000; Takimoto et al, 2008), productivity and nutrient availability (Oksanen et al., 1981; DeAngelis et al., 1989; Kaunzinger and Morin, 1998), disturbance regimes (Pimm and Lawton, 1978; Power et al, 1996; along with productivity in Townsend et al, 1998), ecological stoichiometry (Diehl and Feissel, 2000), size-based patterns in predator-prey interactions (Cohen et al, 1993; Jennings and Warr, 2003; Williams and Martinez, 2004), food-web structure (Long et al, 2011), and metacommunity dynamics (Calcagno et al, 2011). The references offered here are but a small sampling of the attempts to understand the mechanisms affecting food chains. Among the oldest explanations is the idea animal metabolism might determine food chain length. Elton (1927) argued that the general constraint on food chain length may be largely related to the loss of energy in successive trophic levels due to low conversion efficiency in consumer species. This argument extends from the basic thermodynamic principle that energy dissipates from a system as a consequence of inefficient transfer or conversion and has been expanded and refined by Lindeman (1942) and Yodzis (1984), among others.

In ecological systems this is best expressed by Lindeman's concept of ecological efficiency, which is represented as the "ratio of the energy flux into [a] trophic level to the energy flux into the level below" (Colinvaux and Barnett, 1979). This is itself directly related to the conversion efficiency of consumers (Slobodkin, 1962) and leads to the observation that organisms are able to convert only a fraction of consumed resources into their own biomass. Now termed the "trophic efficiency hypothesis" (e.g., Jenkins et al., 1992, or previously the "energy flow hypothesis," Hutchinson, 1959), the basic idea is that some fraction (often very small) of the energy (usually expressed as biomass) entering any trophic level will be fixed as biomass in the subsequent trophic level. This is thought to be the mechanism responsible for the almost-universally observed biomass and abundance pyramids in ecological communities (e.g., Trites, 2003). Essentially, the metabolic explanation for constraints on food chain lengths contends higher trophic levels are simply energy starved, and at some point, cannot support an additional level of consumers (predators). It is important to note such mechanisms might apply both to individual food chains and to the trophic structure of entire communities.

Two predictions extend from the trophic efficiency hypothesis. First, while trophic structure may be organized with respect to gape limitation and body size distributions among consumer and resource species (Roberts, 2003), body size should also play a role in determining food chain lengths. For animals metabolic rate scales with body size as a 3/4 power law (e.g., Brown et al, 2004), and therefore, larger consumer species have substantially larger energetic requirements (the scaling of metabolic rate to body size is different for autotrophs, as well as microbes and protists; see Pretzsch and Dieler 2012 and DeLong et al. 2010 respectively). This, combined with the fact that population density decreases with increasing body size on a similar logarithmic scale, could act to bottleneck energy passing through food chains. For example large herbivorous species might limit energy flow up trophic levels relative to smaller taxa that utilize the same resources, and thereby shorten food chains. Second, metabolic costs associated with endothermy should also influence food chain lengths. Energetic demands are roughly an order of magnitude greater in endotherms relative to ectotherms of the same size (e.g., Humphreys, 1979; Bennet and Ruben, 1979; Hayes, 2010; recently reviewed by Glazier, 2014). Therefore, systems containing more endotherms should display shorter food chain lengths. Additionally, the trophic position of endotherms should impact food chain lengths; endothermic herbivores should result in shorter chain lengths.

In the past three decades, many high-resolution consumer-resource networks (as well as whole food webs) have been produced (e.g., Hall and Raffaelli, 1991, Williams and Martinez, 2000; Brose et al., 2005; Woodward et al., 2008), and have been used to examine, among other things, the patterns and relationships between consumer-resource body sizes and metabolic modes (e.g, Brose et al., 2006; Schmitz and Price, 2011). Further, patterns of variation in metabolic efficiencies between modes (endothermy and ectothermy) have been considered in the context of the trophic structure of ecological networks (e.g, Riede et al., 2011; Moore and de Ruiter, 2012). The possibility of increased endotherm-based constraints on terrestrial food chains was considered by Ryszkowski and French (1982), and Yodzis (1984) attempted to directly explore predictions extending from this hypothesis, using a collection of published food webs available at that time (N = 34, all of which were quite small and poorly resolved by today's standards). Because attributes like body size, metabolic mode and trophic position vary among taxa in different systems, consideration of the food webs available today should offer a more thorough assessment of any relationships between these variables and food chain length. Using the recently completed GlobalWeb database (, augmented with several other large published food webs, we assess many of the predictions that extend from the trophic efficiency hypothesis. Specifically, we test the predictions that: (1) increasing the fraction of endothermic taxa in a system will lead to shorter food chains. (2) Increasing the fraction of endothermic taxa residing at lower trophic levels (e.g., herbivores) will produce shorter average food chains. (3) Increasing the fraction of large consumers at lower trophic levels will shorten food chains. We predict that these effects would be expressed in the mean food chain length (MFCL) of entire systems and therefore could be detected by comparing MFCL across systems. Finally, an alternative prediction consistent with the energy efficiency hypothesis is that systems conforming to one or more of the above scenarios would also contain fewer total food chains than expected given food-web size. To address each prediction individually, we attempt to separate variables related to body size, metabolic mode and trophic position, and determine the strengths of these variables as predictors of food chain length.



Our data were largely drawn from the recently constructed GlobalWeb database (Thompson et al., 2012), which is now the largest single collection of published food webs in the world. Thompson and colleagues discuss in some detail the relationship between food web size (which is tightly correlated with species richness) and resolution. In general smaller food webs produce fewer food chains and shorter MFCLs that larger ones. However, our purpose was not to compare large and small food webs but to compare food webs of equal sizes across ecosystem types. That is, we predicted that differences in MFCL for systems of equal size would correlate with differences in the body size distributions, metabolic modes and the trophic positions of taxa in each system. For our purposes it was necessary that each food web be complete (i.e., not confined to a guild or interaction module) and only 69 webs in the database met this criterion. In addition to these webs, we added several large, recently published food webs, for a total of 77 food webs in the analysis (23 freshwater, 34 marine, and 20 terrestrial systems), ranging in web size from 8-224 taxa. The entire list and relevant descriptors of each web have been made available on-line ( Research/). One web, Jennings Environmental Center, was produced by us for this paper and has also been made available on-line ( with the searchable web matrix available through GlobalWeb ( Because humans and parasites were consumer types that were not represented in all webs (and would significantly bias our metrics for food webs for which they're reported), they were removed entirely from the study. All food webs either already existed as, or were converted to, binary consumer-based matrices (see Dunne, 2006, for a formal description of this type of matrix).

We predicted the relative proportion of endothermic taxa in a food web would be negatively correlated with mean food chain length (MFCL). However, the relative species richness of endotherms is only a coarse metric in this regard. We felt that it was also necessary to consider the relative number of endothermic and ectothermic trophic interactions in each web (i.e., not only to determine how many endothermic taxa are present, but assess their relative contribution to each food-web network). To examine the relationship between the trophic positions of endothermic taxa and the resulting food chain lengths, food webs were deconstructed into collections of unique food chain paths (FCPs). That is, we evaluated all "realized" paths from resources to terminal predators in each food web (here we mean only FCPs derived from each food-web interaction matrix, and not simply all possible FCPs). Trophic position (or height) is typically defined as the one plus the average trophic position of the species' prey or resources. Any heterotrophic species (taxon) could potentially appear in multiple food chains and at different trophic levels. However, none of our analyses sought to assess directly the average trophic position of individual species (taxa) but rather the relative proportion of endotherms at each trophic level across all FCPs in a system. We excluded circularities ("loops") and cannibalism from these calculations. This permitted us to assess the relative contributions of endotherms and ectotherms to the total number of trophic interactions (both within trophic levels and globally), as well as whether or not paths containing more endotherms, or endotherms at particular trophic positions, produced shorter food chain lengths in each web.

We also sought to consider the role that body size plays in determining food chain lengths and perhaps even food-web structure. Therefore, we also estimated the average mass of each species in each web, where such distinctions were possible (some low-resolution groupings did not allow for the approximation of masses). For some webs the precise masses of species have been previously published (e.g., Tuesday Lake, Carpinteria marsh, Raritan river, Muskingum brook, etc.), and they were included here. Where study-specific values were not available, values from existing literature were used. For example was utilized for estimating the average masses of most fishes. Because most food webs either do not separate juveniles and adults (i.e., they are represented as one node or unit), or they do not consider the diets of juveniles at all, we used average masses for adults. Where juveniles were treated separately, we also did so. The average masses of many other vertebrate species were also easily retrievable from the literature. This enabled us to closely approximate the masses of other similar taxa where no records were available. As in Yodzis (1984), our goal was to compare masses between and across taxa at the level of orders of magnitude (i.e., at the log-scale), making any small errors in our estimates of little consequence. We were able to approximate masses for taxa in 72 of the food webs in our dataset. This information was used to supplement each food-web matrix, allowing us to evaluate the effect of body size along with metabolic mode, trophic position, and the relative frequencies of trophic interactions.


For each food web, we examined numerous variables related to food web network structure, metabolic mode, and body size. The following variables were considered: system type (freshwater, marine, or terrestrial), species richness, the number of food chain paths (FCPs), mean food chain length (MFCL), the relative proportion of endotherms at the herbivore level, maximum food chain length (FCL), connectance, linkage density, average diet breadth, average prey vulnerability, average body size at the autotroph, herbivore and apex predator positions, and the difference between predator-prey body size at the herbivore and apex predator positions. Because both MFCL and maximum FCL varied across food webs, it was not possible to compare the proportion of endotherms or the average body sizes of omnivorous and mid-level predators. For example the apex predators in one food web might occupy the fourth trophic level, while those of another food web might occupy the sixth. However, it was possible to look at these attributes in the "apex" predator levels of our webs (regardless of respective trophic level).

As a first pass through the data, we assessed pairwise correlations between all variables using a restricted maximum likelihood (REML) multivariate method offered in JMP 11.0. This method produces unbiased estimates of variance and covariance. We were specifically interested in determining which variables were highly correlated with MFCL, and the degree of covariation among those variables. We also utilized multiple linear regression (MLR) in establishing which variables might serve as good predictor variables for MFCL and which could be excluded. Variables related to system size were most strongly correlated with MFCL. However, several variables necessarily covaried (for example species richness in combination with linkage density directly determines the number of FCPs in any web). Additionally, species richness (as well as the number of FCPs) is known to be heavily biased by taxonomic resolution, which is not a biologically real property of any system. To control for this effect, we used best-fit models for each system type to transform MFCL values so that they were flat with respect to (i.e. not related to) species richness. We then used pairwise Kolmogrov-Smirnov tests to evaluate the role of system type and the number of FCPs in predicting MFCL.

Because of the linked covariation among species richness and the number of FCPs, and the potential sampling bias observed in species richness, we removed theses variables from further analyses. We also excluded variables related to the body size and metabolic mode of apex predators, because they were not significantly correlated with MFCL. Using MFCL as the response variable, we then constructed generalized linear models (GLMs) for both the entire data set (all systems combined) and individual system types (marine, freshwater and terrestrial). In general variable selection (from the remaining variables) for our models was determined in a "backwards" fashion (James et al. 2013), where variables producing the largest P-value were excluded in a sequential way. Based on the optimal variable set (as established by corrected AIC scores as described in Akaike, 1974) for the entire data set, we applied the same variable set to the GLM for each system type. Prior analysis indicated that the distributions of our variables were within the bounds in which the assumptions of a Normal distribution model using the Identity link function were appropriate. Goodness-of-fit was defined as [[summation].sub.i] [w.sub.i][([y.sub.i] - [[mu].sub.i]).sup.2] and statistical significance was evaluated by Pearson's chi-squared statistic.

Finally, we assessed the relative strength and importance of predictor variables using Classification and Regression Tree (CART) models (again excluding species richness and the number of FCPs) to establish a hierarchical series of optimal "splits" in the data based on optimized cut-points for predictor variables. Here, the model seeks to find optimal cut-points for partitioning data based on a response variable (in our case, MFCL). The predictor variable that best allows the partitioning of the data based on MFCL would represent the first cut-point. One of the strengths of this type of analysis is that it allows for the discovery of cut-points along the distribution of one predictor variable. For example the resulting cut-point will not simply be some variable (say, the body size of herbivores) but will be represented as the point along that distribution allowing for the best partitioning of the data (a partition above and below a specific body size). That is, CART analysis allows us not only to determine which variables are the best predictors of MFCL (which our other analyses already do) but also a more refined determination of the range (or thresholds) for which those variables are most important. However, it is important to mention that CART models are known to over-fit data (Gray and Fan, 2008), particularly when multiple hierarchical splits are requested (i.e., the model is capable of continuing to subdivide a single variable repeatedly). We were only interested in the first and second optimal splits, after which predictive power of subsequent variables typically declines precipitously (based on LogWorth values), and therefore overfitting by repeated partitioning of a single variable was not a concern. Because we anticipated marked differences among system types, the same procedures were applied to system-specific partitions of the data (marine, freshwater, terrestrial), in order to examine differences in the predictive power of variables in each system type. These models were also constructed using the statistical software package JMP 11.0.


Overall, mean food chain length (MFCL) ranged from 2.88 to 6.87 across food webs. The largest MFCL for any terrestrial system was 5.40, while nearly a quarter (24.6%) of the MFCLs in aquatic food webs (freshwater and marine) exceeded this value. MFCL increased linearly with the logarithm of species richness but on different slopes for aquatic and terrestrial systems (t-value 6.38, P < 0.0001; Fig 1A). After controlling for species richness, MFCL did not vary significantly between marine and freshwater systems (Kolmogrov-Smirnov test, D = 0.15, P = 0.93) but did when comparing aquatic and terrestrial systems (D = 0.45 and 0.40, P = 0.01 and 0.06 for marine-terrestrial and freshwater-terrestrial pairwise comparisons respectively). That is, system type (aquatic vs. terrestrial) strongly influenced MFCL.

In our assessment of correlations using REML, the number of food chain paths (FCPs) was strongly correlated with MFCL in all system types (statistically significant at P < 0.0001 in all cases). Notably, terrestrial systems added fewer FCPs in relation to species richness (Fig. 1B) than did aquatic systems, connectance values for terrestrial systems were significantly lower than those of aquatic systems (D = 0.49, P = 0.001), and connectance was negatively correlated with species richness (Correlation = -0.41, P = 0.0005). However, there was no statistical difference between the total number of links (trophic interactions) among aquatic and terrestrial systems with respect to species richness (data not shown).

In considering variables related to the energy efficiency hypothesis, terrestrial food webs contained many more endothermic taxa than their aquatic counterparts. On average, 35.8% of the species making up terrestrial webs were endotherms (SE = 4.57), while endotherms comprised just 13.84% (SE = 2.81) of marine webs and only 3.85% (SE = 1.43) of freshwater webs (but see discussion). In terrestrial systems endothermic taxa also occurred more frequently at lower trophic levels. Eighty-five percent of terrestrial webs contained endotherms at the herbivore level (compared to 12.1% and 4.3% in marine and freshwater food webs respectively). In contrast endotherms were scarce in the lower trophic levels of marine and freshwater food webs, but were more common atop food chains (particularly food chain paths that were above the MFCL in each system; Fig. 2). Within terrestrial systems endotherms displayed larger trophic breadths and smaller vulnerability values than ectotherms on the same trophic level (Fig. 3). Body mass distributions varied dramatically within and between trophic levels and across system types. Specifically, the average masses of autotrophic and herbivorous taxa in marine and freshwater systems were significantly smaller than those of terrestrial systems (Fig. 4A). This also resulted in significant differences in the average masses of trophically linked herbivore and autotroph taxa for respective systems (Fig. 4B).

After removing species richness and FCPs, our multivariate analyses indicate the average (log) body mass of herbivorous taxa and, in aquatic systems, the body mass of their predators, best predicted MFCL. System-specific GLMs indicated that the average difference in body masses between linked species in trophic levels two and three was the best predictor of MCFL in aquatic systems ([X.sup.2] = 10.78, P = 0.001 and [X.sup.2] = 14.18, P = 0.0002 for freshwater and marine systems respectively). However, in our GLM for terrestrial systems, the average mass of herbivorous taxa (trophic level two) was decidedly the best predictor of MFCL ([X.sup.2] = 20.60, P < 0.0001). This finding was largely corroborated by our CART analysis, which determined the average mass of herbivorous taxa to be the strongest predictor variable of MFCL in aquatic systems (SS = 11.47, LogWorth = 3.85, cut point at 1.92 and SS = 21.08, LogWorth = 5.49, cut point at 0.55 for freshwater and marine systems respectively). In contrast CART analysis for terrestrial systems found that the relative proportion of endothermic taxa at the herbivore level was the most powerful predictor of MFCL (SS = 3.07, LogWorth = 2.02, cut point at 0.15).


There remains great interest in the factors that determine food chain lengths in natural communities, and a single comprehensive theory remains elusive (Post, 2002; Shurin et al., 2006). In the present study, we used a large dataset of published food webs to examine the relationships between food chain length and variables related to body size and metabolic mode--in the context of food-web structure. Like previous studies (e.g., Yodzis, 1984; Hairston and Hairston, 1993; Shurin et al., 2006), we observed marked differences in the mean food chain lengths (MFCLs) of aquatic and terrestrial food webs. These differences were separable from species richness but were highly correlated with the total number of food chain paths (FCPs). Aquatic food webs displayed more FCPs than terrestrial food webs of equal size (both in terms of species richness and in the total number of trophic links), suggesting a difference in the trophic structuring (connectance values were also lower in terrestrial systems). This might also explain why maximum food chain lengths varied with system type but not with species richness. In aquatic (namely marine) systems, while rare, particularly long food chain paths were possible (as long as 12 links), while the longest terrestrial food chains path was seven links. (It must be noted that these maximum lengths are much larger than the MFCL values for each system, and are rare trophic paths).

One common explanation for the observed differences in MFCL between aquatic and terrestrial systems is the severe constraint on the range of body sizes displayed in terrestrial systems (Cohen et al., 1993; Jennings and Warr, 2003). In aquatic systems food chains are often filled with small prey items (pico-, nano-, microplankton, meiophauna, etc.), and many top predators are much larger than those of terrestrial systems. According to classic cascade models of food webs (i.e., Roberts, 2003), body size and gape limitation organize trophic interactions. For example Brose et al. (2006) assessed 16,863 consumer-resource links spanning ten high-resolution interaction networks (most of which were trophic guilds, as opposed to complete food webs) and found that the differences between the masses of linked consumer and resource taxa in aquatic systems are significantly larger than those of terrestrial systems. Presumably, a wider the range of body sizes moving from autotrophs to top predators in a system permits longer food chains. Our findings support the hypothesis that body size distribution constrains food chain lengths. Aquatic webs contained, on average, significantly smaller autotroph and herbivore taxa and (in the case of marine systems) larger top predators. The importance of this pattern was detected by our multivariate analyses, which singled out the average (log) body mass of herbivores (and in the case of aquatic systems, their predators) as strong predictors of MFCL. That is, the presence of large herbivorous species was strongly associated with shortened MFCLs in systems (regardless of ecosystem type).

Most consumer-resource interactions among small organisms are complete consumption events (as opposed to grazing), and it may be that biomass flux to higher trophic levels is much greater in systems with smaller organisms at their bases. Body size is also strongly (negatively) correlated with generation time and abundance (e.g., Cyr et al., 1997), meaning total biomass production (and flux) for small organisms may be large relative to bigger taxa (Benke and Huryn, 2010), and the microbial loops of aquatic systems are known to be larger than those of terrestrial systems. Therefore, we cannot discount the possibility that body size might still fall within the explanatory umbrella of the nutrient limitation hypothesis (Arim et al., 2007). Additionally, it must be noted that, while bacteria (and other components of the microbial community) are present in terrestrial systems, they are poorly resolved in (and often completely absent from) published terrestrial food webs.

While body size was detected as a major driver in MFCL for all systems, our analyses also indicated metabolic mode in herbivores was important in determining the MFCLs of terrestrial systems. Broadly, endothermy was much more common in terrestrial systems, and endotherms assumed lower trophic positions (namely, herbivory), both of which were negatively correlated with MFCL. More specifically, the presence of endothermic taxa at the herbivore level was determined as the strongest predictor variable of MFLC in terrestrial systems, as opposed to body size in aquatic systems. While the reporting of larger endotherms in terrestrial food chains may be more resolved than for their smaller ectothermic counterparts, it seems evident that there are more endothermic herbivores in terrestrial systems than aquatic ones. In terrestrial systems macroflora are also usually well described and may be over-represented in our food webs. However, any bias likely has little relation to our overall findings, as the average size differences between aquatic and terrestrial autotrophs was not a significant predictor of MFCL.

Terrestrial systems act as a test for the role of endothermy in limiting food chain lengths, as endotherm and ectotherm consumers are often more similar in size in those systems (partially controlling for body size as a variable). Our CART analysis suggested that endothermy (and not larger body size) was the best predictor of MFCL in terrestrial systems. These findings are consistent with Ryszkowski and French (1982), who directly implicated the high energetic cost of endothermy and the reciprocal low conversion efficiency with respect to biomass production as important drivers in the biomass patterns and community structuring of terrestrial ecosystems (though they did not test the hypothesis). This is also consistent with the more recent work of Shurin and Seabloom (2005), in which ectothermic species were found to better propagate trophic cascades than endotherms, as a direct consequence of their increased metabolic efficiency. However, those authors also implicated size differences between herbivores and autotrophs as important predictors of the strength of trophic cascades (see also DeLong et al. 2015).

Given that body size was implicated as the most important driver in aquatic systems (which have larger body size spectra), but not in terrestrial systems, we believe that endothermy may be hierarchically nested within the more overarching role of body size in determining food chain length. It may be that, within the narrow range of body sizes seen in terrestrial systems, endothermy becomes a much more relevant factor in constraining trophic interactions (and energy flux). It is interesting to note that two of the largest MFCL values for terrestrial systems came from food webs that reported only ectotherms (WEB151 and WEB311, additional information on these WEBs are available on-line at, even though WEB151 contained just 23 taxa. Here, the size distribution was considerably constrained. This pattern was not seen in aquatic systems, and supports the idea that endothermy plays a more pivotal role in determining food chain lengths within a narrowed range of consumer-resource body sizes.

Taken together, the body size, relative fraction and trophic position of endotherms in aquatic and terrestrial systems may play into the previously discussed differences in FCPs. Because metabolic requirements are roughly an order of magnitude greater for endotherms than ectotherms of the same size, it may be that endothermic taxa bottleneck energy, limiting the total number of trophic interactions (and chains) for entire food webs. This view is supported by the observation that endotherms have fewer predators (limiting the number of FCPs up food chains) and depend upon more prey items (which might be linked to increased energetic requirements). Our findings are consistent with Yodzis (1984), who demonstrated that ectotherms are more likely to support consumers than endotherms, as well as the data used in Cohen et al. (1993), which contained far more ectothermic trophic interactions than endothermic ones, even when considering only trophic links to vertebrate predators. Therefore, we suggest endothermy may limit food chain lengths by altering food-web structure and complexity. This however does not necessarily explain why many marine food webs with long MFCLs contain endotherms as top predators, and it is worth noting Vander Zanden and Fetzer (2007) found marine food chains containing mammals were longer than those without them. However, we point to the facts these mammals are typically top predators, rather than consumers at the herbivore and omnivore levels (as is seen in terrestrial systems), and that they are likely fewer in number and smaller in population biomass. Still, these findings need further reconciliation going forward.

Acknowledgments.--We thank Jennifer Dunne and Ross Thompson for providing access to the GlobalWebs database. We also thank Kenneth Elgersma and Givonni Strona for advice and discussion regarding analysis. Finally, we thank Brandi L. Miller-Parrish and the PADEP for providing us with access to the survey studies and data for Jennings Environmental Education Center (Slippery Rock, PA, U.S.A.).


Akaike, H. 1974. A new look at the statistical model identification. IEEE Trans. Auto. Contr., 19(6):716-723.

Arim, M., F. Bozinovic, and P. A. Marquet. On the relationship between trophic position, body mass and temperature: reformulating the energy limitation hypothesis Oikos, 116:1524-1530.

Benke, A. C. and A. D. Huryn. 2010. Benthic invertebrate production: facilitating answers to ecological riddles in freshwater ecosystems. J. N. Am. Benthol. Soc., 29:264-285.

Bennet, A. F. and J. A. Ruben. 1979. Endothermy and activity in vertebrates. Science, 9(206):649-654.

Brose, U. el al. 2005. Body sizes of consumers and their resources. Ecology, 86:2545, Ecological Archives E086-135.

--et al. 2006. Consumer-resource body size relationships in natural food webs. Ecology, 87:2411-2417.

Brown, J. H. et al 2004. Toward a metabolic theory of ecology. Ecology, 85:1771-1789.

Calcagno, V. 2011. Constraints on food-chain length arising from regional metacommunity dynamics. Proc. R. Soc. B., 278(1721):3042-3049.

Chase, J. M. 2000. Are there real differences among aquatic and terrestrial food webs? Trends Ecol. Evol, 15:408-412.

Cohen, J. E. and C. M. Newman. 1991. Community area and food chains length: theoretical predictions. Am. Nat., 138(6):1542-1554.

--et al 1993. Body sizes of animal predators and animal prey in food webs. J. Anim. Ecol., 62:67-78.

Colinvaux, P. A. and B. D. Barnett. 1978. Lindeman and the ecological efficiency of wolves. Am. Nat., 114(5):707-718.

Cyr, H., J. A. Downing, and R. H. Peters. 1997. Density-body size relationships in local aquatic communities. Oikos, 79:333-346.

DeAngelis, D. L., S. M. Bartell, and A. L. Brenkert. 1989. Effects of nutrient recycling and food-chain length on resilience. Am. Nat., 134(5):778-805.

DeLong, J. P., J. G. Okie, M. E. Moses, R. M. Sibly, and J. H. Brown. 2010. Shifts in metabolic scaling, production, and efficiency across major evolutionary transitions of life. Proc. Natl. Acad. Sci. USA, 107:12941-12945

Diehl, S. and M. Feissel. 2000. Effects of enrichment on 3-level food chains with omnivory. Am. Nat., 155:200-218.

Dunne, J. A. 2006. The network structure of food webs, p 27-86. In: Pascual, M. and J. A. Dunne (eds.). Ecological networks: linking structure to dynamics in food webs. Oxford University Press.

Elton, C. 1927. Animal ecology. Sidgwick and Jackson, London.

Glazier, D. S. 2014. Metabolic scaling in complex living systems. Systems, 2:451-540.

Gray, J. B. and F. Guangzhe. 2008. Classification tree analysis using TARGET. Comput. Stat. Data An., 52:1362-1372.

Hairston, N. G. J. and N. G. S. Hairston. 1993. Cause-effect relationships in energy-flow, trophic structure, and interspecific interactions. Am. Nat., 142:379-411.

Hall, S. J. and D. Raffaelli. 1991. Food-web patterns--lessons from a species-rich web. J. Anim. Ecol, 60:823-842.

Hayes, J. P. 2010. Metabolic rates, genetic constraints, and the evolution of endothermy. J. Evol. Biol., 23(9):1868-1877.

Humphreys, W. F. 1979. Production and respiration in animal populations. J. Anim. Ecol, 48:427-453.

Hutchinson, G. E. 1959. Homage to Santa Rosalia, or why are there so many kinds of animals? Am. Nat., 93:145-159.

James, G., D. Witten, T. Hastie, and R. Tibshirani. 2013. An introduction to statistical learning: with applications in R. Springer, New York, NY.

Jenkins, B., R. L. Kitching, and S. L. Pimm. 1992. Productivity, disturbance and food web structure at a local spatial scale in experimental container habitats. Oikos, 65:249-255.

Iennings, S. and K. T. Warr. 2003. Smaller predator-prey body size ratios in longer food chains. Proc. R. Soc. B., 270:1413-1417.

Kaunzinger, C. M. K. and P. J. Morin. 1998. Productivity controls food-chain properties in microbial communities. Nature, 395:495-497.

Kuris, A. M., R. F. Hechinger, J. C. Shaw, K. Whitney, L. Aguirre-Macedo, C. Boch, A. P. Dobson, E. J. Dunham, B. L. Fredensborg, T. C. Huspeni, J. Lorda, L. Mababa, F. Mancini, A. Mora, M. Pickering, N. Talhouk, M. E. Torchin, and K. D. Lafferty. 2008. Ecosystem energetic implications of parasite and free-living biomass in three estuaries. Nature, 454:515-518.

Lindeman, R. L. 1942. The trophic-dynamic aspect of ecology. Ecology, 23(4):399-417.

Long, Z. T., J. F. Bruno, and J. E. Duffy. 2011. Food chain length and omnivory determine the stability of a marine subtidal food web. J. Anim. Ecol, 80(3):586-594.

Moore, J. C. and P. C. de Ruiter. 2012. Energetic food webs: an analysis of real and model ecosystems. Oxford University Press, Oxford, UK.

Oksanen, L., S. D. Fretwell, J. Arruda, and P. Niemela. 1981. Exploitation ecosystems in gradients of primary productivity. Amer. Nat., 118:240-261.

Pimm, S. L. and J. H. Lawton. 1978. On feeding on more than one trophic level. Nature, 275:542-544.

Post, D. M. 2002. The long and short of food-chain length. Trend Ecol. Evol., 17:269-277.

--and G. Takimoto. 2007. Proximate structural mechanisms for variation in food-chain length. Oikos, 1176:775-782.

--, M. L. Pace, and N. G. Hairston Jr. 2000. Ecosystem size determines food-chain length in lakes. Nature, 405:1047-1049.

Power, M. E., M. S. Parker, and J. T. Wootton. 1996. Disturbance and food chain length in rivers, p 286-297. In: Polis, G. A. and K, O. Winemiller (eds.). Food webs: integration of patterns and dynamics. Chapman and Hall.

Pretzsch, H. and J. Dieler. 2012. Evidence of variant intra- and interspecific scaling of tree crown structure and relevance for allometric theory. Oecologia 169:637-649.

Riede, J. O., U. Brose, B. Ebenman, U. Jacob, R. Thompson, C. R. Townsend, and T. Jonsson. 2011. Stepping in Elton's footprints: a general scaling model for body masses and trophic levels across ecosystems. Ecol. Lett., 14(2): 169-178.

Roberts, J. M. 2003. Testing the cascade model for food webs. J. Agri. Biol. Environ. Stat., 8(2):196-204.

Rossiter, W. and M. V. K. Sukhdeo. 2014. Habitat-based constraints on food web structure and parasite life cycles. Oecologia, 174(4): 1311-1321.

Riszkowski, L. and N. R. French. 1982. Tophic impact of mammals in terrestrial ecosystems. Acta Theriol, 27(1):3-24.

Schmitz, O. J. and J. R. Price. 2011. Convergence of trophic interaction strengths in grassland food webs through metabolic scaling of herbivore biomass. J. Anim. Ecol, 80(6):1330-1336.

Shurin, J. B. and E. W. Seabloom. 2005. The strength of trophic cascades across ecosystems: Predictions from allometry and energetics. J. Anim. Ecol., 74:1029-1038.

--et al. 2006. All wet or dried up? Real differences between aquatic and terrestrial food webs. Proc. R. Soc. Lon. B., 273:1-9.

Slobodkin, L. 1962. Energy in animal ecology. Adv. Ecol., 1:69-101.

Takimoto, G., D. A. Spiller, and D. M. Post. 2008. Ecosystem size, but not disturbance, determines food-chain length on islands in the Bahamas. Ecology, 89:3001-3007.

Thompson, R. M. et al. 2012. Food webs: reconciling the structure and function of biodiversity. Trend Ecol. Evol., 27(12):689-697.

Townsend, C. R., R. M. Thompson, A. R. McIntosh, C. Kilroy, E. Edwards, and M. R. Scarsbrook. 1998. Disturbance, resource supply, and food-web architecture in streams. Ecol. Lett., 1:200-209.

Trites, A. W. 2003. Food webs in the ocean: who eats whom and how much? p. 125-141. In: M. Sinclair and G. Valdimarsson (eds.). Responsible fisheries in the marine ecosystem. CABI Publishing.

Vander Zanden, M. J. and W. W. Fetzer. 2007. Global patterns of aquatic food chain length. Oikos, 116:1378-1388.

Williams, R. J. and N. D. Martinez. 2000. Simple rules yield complex food webs. Nature, 404:180-183. --and--. 2004. Limits to trophic levels and omnivory in complex food webs: theory and data. Am. Nat., 163(3):458-468.

Woodward G. 2008. Trophic trickles and cascades in a complex food web: impacts of a keystone predator on stream community structure and ecosystem processes. Oikos, 117:683-692.

Yodzis, P. 1984. Energy flow and the vertical structure of real ecosystems. Oecologia, 65:86-88.




Department of Biology, Waynesburg University, 51 W. College St., Waynesburg, Pennsylvania 15320


Department of Biological Sciences, Box 870344 University of Alabama, Tuscaloosa, 35487



Department of Ecology, Evolution and Natural Resources, Rutgers University, 14 College Farm Road, New Brunswick, New Jersey 08901

(1) Corresponding author: Telephone: (732) 947-1574; e-mail:

Caption: Fig. 1.--(A) Distribution of MFCLs against species richness (log-transformed) for aquatic and terrestrial ecosystems. Aquatic systems demonstrate significantly longer MFCLs as species richness increases (see results for statistical tests). (B) Relationships between MFCL and the number of FCPs (log-transformed) in each system type. There was no statistical difference between the slopes for freshwater and marine systems, but terrestrial systems produce significantly smaller MFCLs in relation to increasing FCPs

Caption: Fig. 2.--Average relative proportions of taxa that are endothermic at each trophic position in food chains for respective systems. Endothermy is significantly more prevalent in the lower trophic levels of terrestrial systems, and declined precipitously beyond trophic level three. Aquatic systems contained very few trophic interactions involving endothermic herbivores, but the proportion of endothermic predators increased with trophic level. In all systems, food chains containing more than seven links were rare, but consistently contained endotherms as apex predators in aquatic systems

Caption: Fig. 3.--Mean differences in vulnerability (the number of predators feeding on a prey species) and trophic breadth (the number of prey species a predator feeds on) for ectothermic and endothermic taxa in terrestrial systems. Positive values for vulnerability indicate that, on average, ectothermic species had more predators than did endotherms in the same trophic level. Conversely, negative trophic breadth values indicate that endothermic predators fed on more prey species than did ectotherms in the same trophic level. Differences were calculated for each food web and were averaged across all terrestrial systems. Error bars indicate SE

Caption: Fig. 4.--(A) Average body size (mass) of autotrophs and herbivores (trophic levels one and two), and top predators (T2 and T1) in each system. (B) Average difference in the body masses of resource and consumer species at the bottom two and top two trophic levels for each system type. Because most terrestrial food webs contain five or less trophic levels, those top predators (trophic level five) are compared to their aquatic counterparts, which may sit at much higher trophic levels, but still represent top predators
COPYRIGHT 2017 University of Notre Dame, Department of Biological Sciences
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2017 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Rossiter, Wayne; King, Gabrielle; Johnson, Brian
Publication:The American Midland Naturalist
Article Type:Report
Geographic Code:1USA
Date:Jan 1, 2017
Previous Article:Individual identification of raccoons (Procyon lotor) using track plate foot printing.
Next Article:Temporal activity levels of mammals in patches of early successional and mature forest habitat in Eastern Connecticut.

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |