Modeling an interaction web with stage-structured variables: Do they improve fit?

Background/Question/Methods

Stage-structured models are a mainstay in population biology, but have only recently been adopted in the study of consumer-resource interactions.  We evaluated the benefit of including stage-structured interactions into models describing variation of interaction webs along resource and disturbance gradients in the Everglades, Florida, USA.   Mesocosm experiments documented intraguild predation and cannibalism as potentially important interactions of common species.  Do these experimental results scale up to field conditions sampled at a landscape scale?  We used structural equation modeling (SEM), to evaluate the relative contribution of hydrology, nutrients, and biotic interactions on density and biomass of small fish and invertebrates.  Nutrient and hydrology data, as well as periphyton biomass and composition, primary consumer (juvenile fish and invertebrate), and secondary consumer (adult fish and invertebrate) data were collected annually for six years at approximately 130 sites.  In addition to comparing the effects of hydrology and nutrients, our SEMs allow examination of the magnitude of key direct and indirect effects over time.   Models were fit separately to density and biomass of each consumer category; results were largely consistent for both data types so only biomass will be discussed.

Results/Conclusions

We found that including stage-structured interactions improved the fit of models to data from all six years examined.  Adding reciprocal effects of consumers on their resources and omnivory also improved model fit.  Top-down effects of carnivore biomass on biomass of juvenile fishes were negative and effects of juvenile fish on adult carnivores were positive.  Adult omnivores had no consistent effect on juvenile fish biomass.  Surprisingly, direct hydrological effects on adult carnivore and herbivore biomass, and on juvenile biomass, were small; direct effects of algal characteristics on herbivore and juvenile biomass were large and positive in four of the six years.  When present, most hydrological effects on consumers were indirect through their effects on algal characteristics.  Indirect nutrient effects on consumers were limited to the first two years of the study and were exclusively through their impact on the relative abundance of edible taxa in periphyton mats; periphyton biomass had little indirect effect on any consumer.   Experimental research in mesocosms and field cages indicated that top-down effects, trophic cascades, nutrient regeneration, and stage-structured interactions could be important at local scales.   Our results support the hypothesis that such local interactions scale up and should be considered in models guiding ecological management and restoration activities, as least in the Everglades.