Main content area

How detectable is predation in stage‐structured populations? Insights from a simulation‐testing analysis

Oken, Kiva L., Essington, Timothy E., Gurney, Bill
The journal of animal ecology 2015 v.84 no.1 pp. 60-70
age structure, biomass, fish, food webs, life history, models, mortality, ontogeny, population dynamics, predation, predators, time series analysis
The potential of predation to structure marine food webs is widely acknowledged. However, available tools to detect the regulation of prey population dynamics by predation are limited, partly because available population data often aggregate a population's age structure into a single biomass or abundance metric. Additionally, many food webs are relatively complex, with prey species subject to different assemblages of predators throughout their ontogeny. The goal of this study was to evaluate the extent to which stage‐structured predation could be reliably detected from time series of total biomass of predators and prey. We simulated age‐structured populations of four mid‐trophic‐level fish species with distinct life‐history traits, exposed them to variable predation at different life stages and fit production models to resulting population biomass to determine how reliably the effects of predators could be detected. Predation targeting early life history and juvenile life stages generally led to larger fluctuations in annual production and was therefore more detectable. However, ecologically realistic levels of observation error and environmental stochasticity masked most predator signals. The addition of predation at a second life stage sharply decreased the ability to detect the effect of each predator. We conclude that the absence of detectable species interactions from biomass time series may be partly due to the interactive effects of environmental variability and complex food web linkages and life histories. We also note that predation signals are most robust for predator–prey systems where predators primarily act on mortality of submature life‐history stages. Simulation testing can be applied widely to evaluate the statistical power of analyses to detect predation effects.