Metabolic theory and taxonomic identity predict nutrient recycling in a diverse food web

Abstract

Reconciling the degree to which ecological processes are generalizable among taxa and ecosystems, or contingent on the identity of interacting species, remains a critical challenge in ecology. Ecological stoichiometry (EST) and metabolic theory of ecology (MTE) are theoretical approaches used to evaluate how consumers mediate nutrient dynamics and energy flow through ecosystems. Recent theoretical work has explored the utility of these theories, but empirical tests in species-rich ecological communities remain scarce. Here we use an unprecedented dataset collected from fishes and dominant invertebrates (n = 900) in a diverse subtropical coastal marine community (50 families, 72 genera, 102 species; body mass range: 0.04-2,597 g) to test the utility of EST and MTE in predicting excretion rates of nitrogen (E(N)), phosphorus (E(P)), and their ratio (E(NP)). Body mass explained a large amount of the variation in EN and EP but not E(NP). Strong evidence in support of the MTE 3/4 allometric scaling coefficient was found for E(P), and for E(N) only after accounting for variation in excretion rates among taxa. In all cases, including taxonomy in models substantially improved model performance, highlighting the importance of species identity for this ecosystem function. Body nutrient content and trophic position explained little of the variation in E(N), E(P), or E(NP), indicating limited applicability of basic predictors of EST. These results highlight the overriding importance of MTE for predicting nutrient flow through organisms, but emphasize that these relationships still fall short of explaining the unique effects certain species can have on ecological processes.

Keywords: coastal ecosystems; consumer-mediated nutrient cycling; ecological stoichiometry; nitrogen; phosphorus.

Fig. 1.

Linear regression models for excretion rates of N (E_N), P (E_P) (µg/h), and N:P (E_NP) and wet body mass of individual organisms. Families are indicated by a unique combination of color and symbol. All values are log₁₀ transformed. *Slope significantly differs from zero (α < 0.05).

Fig. 2.

Linear regression models for excretion rates of N (E_N), P (E_P) (µg/h), and N:P (E_NP) and wet body mass of individual organisms (n = 675, 372, and 367 for N, P, and N:P, respectively). Subsequent regressions are between excretion-mass model residuals and either δ¹⁵N or body nutrient content (for N, P, or N:P). *Slope significantly differs from zero (α < 0.05). Families are indicated by a unique combination of color and symbol.

Fig. 3.

(A) Random slope and intercept estimates for each family (circles) in the N excretion (E_N) vs. body mass (wet mass) mixed effects model and (B) random intercepts for each family in the P excretion (E_P) vs. body mass mixed effects model. Bars indicate 95% prediction intervals for both the intercept and slope. The vertical black lines indicate the mean intercept and slope. Vertical dashed gray line indicates 0.75, the 3/4 scaling coefficient predicted by MTE. Filled and empty circles indicate vertebrate and invertebrate families, respectively. See Table S2 for family names associated with listed numbers.

Fig. 4.

Random slope and intercept estimates for each species (circles) in the N:P excretion (E_NP) vs. body mass (wet mass) mixed effects model. Bars indicate 95% prediction intervals for both the intercept and slope. The vertical black lines indicate the mean intercept and slope. Vertical dashed gray line indicates a slope of 0. Filled and empty circles indicate vertebrate and invertebrate families, respectively. See Table S2 for species names associated with listed numbers.