Ecological influences on larval chironomid communities in shallow lakes: implications for palaeolimnological interpretations

Summary 1. To correctly interpret chironomid faunas for palaeoenvironmental reconstruction, it is essential that we improve our understanding of the relative influence of ecosystem variables, biotic as well as physicochemical, on chironomid larvae. To address this, we analysed the surface sediments from 39 shallow lakes (29 Norfolk, U.K., 10 Denmark) for chironomid head capsules, and 70 chironomid taxa (including Chaoborus ) were identified. 2. The shallow lakes were selected over large environmental gradients of aquatic macrophytes, total phosphorus (TP) and fish communities. Redundancy analysis (RDA) identified two significant variables that explained chironomid distribution: macrophyte species richness ( P 10 all had relatively diverse (Hill’s N2) chironomid assemblages, and lakes with Hill’s N2 >10 all had TP <250 μg L −1 and total fish densities <2 fish per m 2 . 3. Plant density (PVI), and perhaps more importantly species richness, were primary controls on the distribution of chironomid communities within these lakes. This clearly has implications for palaeoenvironmental reconstructions using zoobenthos remains (i.e. chironomids) and suggests that they could be used to track changes in benthic/pelagic production and could be used as indicators of changing macrophyte habitat. 4. Measuring key biological gradients, in addition to physicochemical gradients, allowed the major controls on chironomid distribution to be assessed more directly, in terms of plant substrate, food availability, competition and predation pressure, rather than implying indirect mechanisms through relationships with nutrients. Many of these variables, notably macrophyte abundance and species richness, are not routinely measured in such studies, despite their importance in determining zoobenthos in temperate shallow lakes. 5. When physical, chemical and ecological gradients are considered, as is often the case with palaeo‐reconstructions rather than training sets chosen to maximise one gradient, complex relationships exist, and attempting to reconstruct a single trophic variable quantitatively may not be appropriate or reliable.