Stable isotope-based community metrics as a tool to identify patterns in food web structure in east African estuaries

1.Quantitative tools to describe biological communities are important for conservation and ecological management. The analysis of trophic structure can be used to quantitatively describe communities. Stable isotope analysis is useful to describe trophic organization, but statistical models that allow the identification of general patterns and comparisons between systems/sampling periods have only recently been developed. 2.Here, stable isotope-based Bayesian community-wide metrics are used to investigate patterns in trophic structure in five estuaries that differ in size, sediment yield and catchment vegetation cover (C3/C4): the Zambezi in Mozambique, the Tana in Kenya and the Rianila, the Betsiboka and Pangalanes Canal (sampled at Ambila) in Madagascar. 3.Primary producers, invertebrates and fish of different trophic ecologies were sampled at each estuary before and after the 2010–2011 wet season. Trophic length, estimated based on δ15N, varied between 3·6 (Ambila) and 4·7 levels (Zambezi) and did not vary seasonally for any estuary. Trophic structure differed the most at Ambila, where trophic diversity and trophic redundancy were lower than at the other estuaries. Among the four open estuaries, the Betsiboka and Tana (C4-dominated) had lower trophic diversity than the Zambezi and Rianila (C3-dominated), probably due to the high loads of suspended sediment, which limited the availability of aquatic sources. 4.There was seasonality in trophic structure at Ambila and Betsiboka, as trophic diversity increased and trophic redundancy decreased from the prewet to the postwet season. For Ambila, this probably resulted from the higher variability and availability of sources after the wet season, which allowed diets to diversify. For the Betsiboka, where aquatic productivity is low, this was likely due to a greater input of terrestrial material during the wet season. 5.The comparative analysis of community-wide metrics was useful to detect patterns in trophic structure and identify differences/similarities in trophic organization related to environmental conditions. However, more widespread application of these approaches across different faunal communities in contrasting ecosystems is required to allow identification of robust large-scale patterns in trophic structure. The approach used here may also find application in comparing food web organization before and after impacts or monitoring ecological recovery after rehabilitation.