Length-mass equations for freshwater unionid mussel assemblages: Implications for estimating ecosystem function
Abstract
Biomass is often used to scale the contributions of individuals and their functional traits to a community or ecosystem. However, accurate biomass measurements can require destructive sampling, which is detrimental to longlived organisms such as unionid mussels.We amassed a database of 6684 measurements of length and soft tissue dry mass (STDM) or shell dry mass (SHDM) from 43 species of unionid mussels to reduce the need for destructive sampling. We used these data to produce regression equations that relate maximum shell length to mass (either STDM or SHDM) at 3 taxonomic levels: family, phylogenetic tribe, and species. For 2 widely-distributed unionid species, Amblema plicata and Elliptio complanata, we present length–STDM regression equations from 6 waterbodies and basins. We also used bootstrapping resampling at the family level to develop a universal regression equation for unionid mussels. We compared models within all 3 taxonomic levels to determine if multiplicative (log–log transformation) or additive (non-linear parameter estimation) error structures provided better fits for biomass prediction.Models with multiplicative errors best fit length–STDM data at the family level (STDM 5 6.63 1026 Lmax 2.89 1.13; r2 5 0.94), for 83% of tribes (n55, average r250.95 +- 0.03), and for 90% of species (n533, average r250.93 +- 0.07). For length– SHDM, models with multiplicative errors best fit the family level (SHDM 5 2.98 1024 Lmax 2.98 1.32; r2 5 0.86), all tribes (n 5 5, average r2 5 0.88 +- 0.13), and all species (n 5 27, average r2 5 0.94 +- 0.09). Models with multiplicative errors also provided the best fit (r2 > 0.76) for our 2 wide-ranging species, A. plicata and E. complanata. Finally, we present a case study based on data collected from 19 river sites in Alabama and Oklahoma, USA, to determine the performance of our power relationships (bootstrap resampling versus length–STDM regressions). In both river systems, tribe- and species-specific equations improved the prediction of unionid STDM over the familylevel regression by 2–20%. Finer taxonomic resolution equations produce more accurate mass predictions, but where accurate taxonomic identifications are lacking, our family-level regression for STDM will produce acceptable estimates, which are key parameters when estimating mussel contributions to ecosystem services. Our study provides a toolkit that will allow scientists and managers to non-destructively quantify biomass (with uncertainty) of freshwater unionid mussels for secondary production, ecosystem function, and services estimates.