You are here:
Post-processing of aquatic biodiversity data collected at multiple levels of resolution: Implications for estimates of taxa richness, abundance, and rarefaction curves
Citation:
Meredith, C., A. Trebitz, AND J. Hoffman. Post-processing of aquatic biodiversity data collected at multiple levels of resolution: Implications for estimates of taxa richness, abundance, and rarefaction curves. Freshwater Science. The Society for Freshwater Science, Springfield, IL, 98:137-148, (2019). https://doi.org/10.1016/j.ecolind.2018.10.047
Impact/Purpose:
Data sets of benthic invertebrate composition often include taxa identified at multiple, potentially redundant levels of resolution. These redundancies arise because of variations in specimen physical condition and life-stage that affect the degree they can be morphologically identified, and because for organisms that require time-intensive slide mounting to fully identify, it is common practice to do so for only a subset of the sample. Such redundancies can inflate true taxonomic richness and distort occurrence and relative abundance patterns, so is desirable to remove them prior to such analyses. This manuscript presents a formal assessment of the pros and cons of various methods for resolving taxonomic redundancies, as well as methodological details for implementing the methods. The data sets use to evaluate the methods stem from sampling campaigns conducted at two disparate Great Lakes locations as part of EPA-ORD’s research into early detection monitoring strategies for aquatic invasive species. This manuscript lays the ground work for further analyses of these data sets, and more broadly, informs the data processing choices that typically accompany any benthic invertebrate-based bioassessment effort.
Description:
Biodiversity information is an important basis for ecological research and biological assessment, and can be impacted by choices made in the manipulation and processing of taxonomic composition data. In aquatic studies, macro-invertebrate taxa are often identified at multiple levels of resolution due to the presence of damaged or otherwise unidentifiable specimens (e.g., the parent taxon Hexagenia and its child Hexagenia limbata might both occur in a dataset). The ramifications of choice of method to resolve these inconsistencies on taxa richness and rarity are often not fully considered. We determined how multiple methods of resolving ambiguous taxa, including those implemented at the site and dataset level, influenced macro-invertebrate richness and abundance and overall taxonomic composition for two nearshore areas of Lake Superior. We also determined the influence on projected-taxa-richness (detected + undetected), as obtained from species accumulation theory estimators that hinge on the presence of rare taxa (e.g., taxa that are present only once or twice in the dataset). Variants of assigning parents to the most abundant child (APTC) were best at retaining abundance and richness information compared to methods which removed all ambiguous parents and kept children (RPKC) or removed parents or merged children with parents depending on their ratio (RPMC). APTC methods had little impact on the richness and abundance of taxa used in calculating broad IBI metrics. Conversely, the RPKC methods could affect abundance metrics while the RPMC methods could substantially affect both abundance and richness metrics. The dataset-scale APTC and RPKC methods had similar levels of projected-taxa-richness while the RPMC method resulted in much lower and potentially unrealistic estimates due to the removal of rare taxa when taxa were merged with parents. The number of samples needed to reach projected-taxa-richness showed high sensitivity to number of rare taxa. We suggest the APTC method as appropriate for most applications involving mixed-resolution macroinvertebrate datasets in which both the preservation of both richness and abundance information is required, and we make available R-code for automating this method.
URLs/Downloads:
DOI: Post-processing of aquatic biodiversity data collected at multiple levels of resolution: Implications for estimates of taxa richness, abundance, and rarefaction curves