Trophic structure and generalist predator niches in a habitat fragmentation experiment
Habitat loss and fragmentation are major drivers of biodiversity loss. However, a notable gap in our understanding of fragmentation is how it affects trophic structure. Theory often applied to fragmentation predicts that with decreasing habitat amount and in converted matrix habitat, species of higher trophic rank should comparatively suffer greater declines than species of lower trophic rank. Trophic positions and niches of generalist predators that persist in fragments can be used as indicators of changes in food webs caused by fragmentation. In a large-scale experiment, the Wog Wog Habitat Fragmentation Experiment, we compared the trophic position and niche approximated from analysis of both gut contents as well as carbon (δ13C) and nitrogen (δ15N) stable isotopes values of a generalist predator, the pale-flecked garden sunskink (Lampropholis guichenoti). Because L. guichenoti consume a wide variety of prey items, their trophic position and niche are likely good indicators of fragmentation effects on trophic structure of a diverse arthropod community. The experiment consists of replicated eucalypt forest fragments (0.25ha, 0.875ha, and 3.062ha) surrounded by a pine plantation matrix as well as a continuous eucalypt forest reference stand. We predicted that fragmentation would lower L. guichenoti δ15N values and isotopic niche areas.
Contrary to our predictions, we found that L. guichenoti mean δ15N values were 0.9‰ and 1.6‰ higher in fragments and matrix, respectively, than in the continuous eucalypt forest. This pattern suggests that L. guichenoti in the fragments and matrix had higher trophic positions on average than their counterparts in the eucalypt forest. Also contrary to our predictions, isotopic niches that were quantified with Bayesian spatial metrics robust to sample size differences were ~100% and ~170% larger in fragments and matrix, respectively, than in the continuous eucalypt forest. Variance in δ13C values was attributed to a year-treatment interaction where δ13C declined over time from evapotranspiration effects on primary producers as shade increased as the pine matrix matured. Edge effects, season, topography, and body length were poor predictors of δ13C and δ15N values. Gut content analysis revealed that L. guichenoti were largely feeding on ants and spiders across treatments. Order richness of prey items among treatments did not differ when rarefied to equivalent sample sizes. Our findings did not match predictions based on models where fragments are surrounded by an inhospitable matrix, much like islands surrounded by ocean. This underscores the growing realization that this simplification is not adequate for forest fragmentation.