Potential pitfalls of aggregating aquatic invertebrate data from multiple agency sources: Implications for detecting aquatic assemblage change across alteration gradients

Abstract

Abstract Data from nine Delaware River basin (DRB) agencies were used to develop procedural steps that harmonise differences in subsample size (organism fixed counts, OFC), taxa identified and lowest taxonomic levels (LTL) used for identifications. These procedural steps are systematically presented to help practitioners avoid some of the pitfalls associated with combining data from multiple agency sources. Harmonising taxa levels and subsample sizes in the DRB data substantially reduced richness metrics with the majority of the reduction associated with harmonising LTLs rather than OFCs. The opposite was true for abundance metrics with LTLs having little effect and OFCs having a large effect. Harmonisation of LTLs and OFCs increased the average similarity among samples threefold suggesting that failing to harmonise greatly overestimated differences in assemblages among sites. A hypothetical assemblage in which the abundance of each taxon decreased across a disturbance gradient as a function of published tolerance values was developed from the DRB data. This hypothetical assemblage was used to evaluate how LTLs and OFCs affected responses to disturbance, as indicated by assemblage metrics and ordination analysis, and to evaluate the efficacy of this approach to data harmonisation. Reducing LTLs (e.g. species to genus) and(or) OFCs (e.g. 300–100) in the hypothetical assemblage reduced measures of taxa richness and made it more difficult to detect responses across the disturbance gradient. Responses of abundance metrics were not affected by LTLs, but were affected by OFCs. More importantly, mixing data with different LTLs and(or) OFCs introduced a second methodological gradient that could be erroneously interpreted as an environmental gradient. The procedures proposed for harmonising LTLs and OFCs were effective at reducing differences and helped to uncover the underlying response of the taxa and assemblage. Failure to harmonise LTLs and OFCs across mixed data sources can overstate differences among assemblages, reduce the ability to detect underlying environmental gradients and/or introduce methodological gradients that can lead to erroneous interpretations of invertebrate responses to hydrologic alteration or urbanisation. The misinterpretation of underlying environmental gradients can strongly affect the applicability of the findings to support criteria development for aquatic life uses or management actions that strive to balance water availability for human and ecological needs. Consequently, great care must be exercised in harmonising taxa and subsample sizes prior to aggregating data across multiple sources.