Pattering the variability of chironomid-based metrics: the potential in the bioassessment of non-wadeable rivers

Abstract

Multimetric indices based on the aquatic macroinvertebrates are used in many bioassessment programs of freshwater ecosystems worldwide, simplifying complex biological data, but keeping a sufficient amount of information regarding the ecosystem health. Majority of them do not rely on data of the chironomid community due to the complicated identification and great natural variability which they show along the longitudinal gradient. In addition, all these routine monitoring programs are defined for smaller streams but not for large, non-wadeable rivers. Thus, we conducted the survey on two large rivers, the Danube and Sava River, forming a network of 96 sampling sites. The main goal of the study was to model the variability of chironomid-based metrics on the spatial gradients, and the environmental gradients which indicate different types of pollution. To realize this, macroinvertebrates, including Chironomidae larvae, were sampled applying multihabitat approach and simultaneously 16 environmental parameters were measured. Chironomid community structure was patterned using the Self-organizing map (SOM). This multivariate visualization technique constructed two dimensional neural networks where all sampling sites were ordinated and clustered into the three groups of neurons. Passively, not influencing previous ordination, we introduced 15 biotic metrics based on proportion of sensitive and tolerant chironomid taxa, and 17 environmental parameters, into the SOM. Component planes as an output of SOM analysis present a variability pattern for each passive parameter distributed on the SOM network, and a clearly formed gradient indicates a high influence of the parameter on the community pattern. In this study, natural variability (distance to source), conductivity, water elements from natural (nitrates, dissolved oxygen) and anthropogenic sources (orthophosphate and ammonium) and heavy metal pollution (Zn, Cu, As and Cd), formed clear gradients, indicating their important contribution to the chironomid community pattern. This was confirmed by the Kruskal–Wallis test, since the values of visualized parameters significantly differed (P<0.01) between groups of neurons obtained by SOM. The same analysis revealed 6 biotic metrics which concordantly changed together with significant environmental parameters on the SOM network. The variability of chironomid community was driven by longitudinal gradient and multiple stressors. The proportion of sensitive taxa regularly alternated along longitudinal gradient and became dominant in the community, indicating particular complex of stressors. Chironomid-based metrics showed promising variability pattern since they predictably changed along the both natural and stressors gradients. After the scoring system is established, this type of metrics could be included in the routine monitoring programs and provide useful information regarding the ecosystem health.