Skip to main content
Dryad

A unifying framework for analyzing temporal changes in functional and taxonomic diversity along disturbance gradients

Cite this dataset

Larson, Erin et al. (2021). A unifying framework for analyzing temporal changes in functional and taxonomic diversity along disturbance gradients [Dataset]. Dryad. https://doi.org/10.5061/dryad.v6wwpzgw8

Abstract

Frameworks exclusively considering functional diversity are gaining popularity, as they complement and extend the information provided by taxonomic diversity metrics, particularly in response to disturbance. Taxonomic diversity should be included in functional diversity frameworks to uncover the functional mechanisms causing species loss following disturbance events. We present and test a predictive framework that considers temporal functional and taxonomic diversity responses along disturbance gradients. Our proposed framework allows us to test different multidimensional metrics of taxonomic diversity that can be directly compared to calculated multidimensional functional diversity metrics. It builds on existing functional diversity-disturbance frameworks both by using a gradient approach and by jointly considering taxonomic and functional diversity. We used previously unpublished stream insect community data collected prior to, and for the two years following, an extreme flood event that occurred in 2013. Using 14 northern Colorado mountain streams, we tested our framework and determined that taxonomic diversity metrics calculated using multidimensional methods resulted in concordance between taxonomic and functional diversity responses. By considering functional and taxonomic diversity together and using a gradient approach, we were able to identify some of the mechanisms driving species losses following this extreme disturbance event.

Methods

We sampled 14 streams across three drainages in summer 2011 prior to the flood for another study (Harrington et al. 2016) and resampled them again post-flood in the summers of 2014 and 2015 to assess species persistence to this extreme event (Poff et al. 2018). Each site received a normalized channel disturbance score, hereafter referred to as channel disturbance. Values across the 14 sites ranged from 0.2 to 1.0 (See Poff et al. 2018 for more complete methods).

In the current work, we use previously unreported community data to test our predictions. The stream insect community was sampled using a 0.01 m2 Hess sampler, a standard method for quantitatively sampling stream insect density (Hauer and Lamberti 2006). In each year, we collected five substrate samples in run/riffle habitat by agitating the substrate to 10 cm depth. Invertebrate samples were stored in 100% ethanol prior to being cleaned of debris to allow invertebrate identification and enumeration in the lab using a dissecting microscope to the lowest practical taxonomic level. All taxa were identified to genus except for the early instars of chloroperlid stoneflies, simuliid blackflies, elmid beetles, and dytiscid beetles, which were all identified to family and assigned traits at the family-level.

Usage notes

We calculated taxonomic richness, Pielou’s evenness, taxonomic dispersion, distance between non-metric multidimensional scaling points (NMDS distance), and taxonomic distance for each site before and after the disturbance using the ‘vegan’ and ‘FD’ packages in R (Laliberté and Legendre 2010, Laliberté et al. 2014, Oksanen et al. 2017). We selected Pielou’s evenness to compare to functional dispersion because both are sensitive to changes in richness (Mason et al. 2012) and contrasted it with taxonomic dispersion to determine which taxonomic dispersion metric best correlated with functional dispersion. Taxonomic dispersion was calculated by using the ‘FD’ package to calculate dispersion based on a taxonomic dissimilarity matrix that was created using a Linnean classification (Clarke and Warwick 1998, 2001). Taxonomic distance was calculated on the same taxonomic dissimilarity matrix based on the methods from Boersma et al. (2016). Finally, non-metric dimensional scaling (NMDS) distance was calculated based on the Euclidean distance between points for each site at each sampling time point in NMDS space, constructed using Bray-Curtis dissimilarities (Oksanen et al. 2017). Abundance changes were calculated as the change in overall density of individuals at each site between sampling years.

We calculated functional distance between communities from different years at the same site using presence/absence data (Boersma et al. 2016). Using the FD package (Laliberté and Legendre 2010, Laliberté et al. 2014) in R v. 3.4.4 (R Core Team 2018), we calculated abundance-weighted functional richness and functional dispersion for community data in different years at each site using a Gower distance matrix (Gower 1971). Functional richness was the number of unique functional groups, because all of our traits were categorical. We used the same functional space to calculate functional diversity metrics across sites because, given the spatial extent of our study area, we assume all of our sites have a similar regional species pool (Harrington et al. 2016).

Funding

National Science Foundation