Abstract
Univariate and multivariate methods are commonly used to explore the spatial and temporal dynamics of ecological communities, but each has limitations, including oversimplification or abstraction of communities. Rank abundance curves (RACs) potentially integrate these existing methodologies by detailing species-level community changes. Here, we had three goals: first, to simplify analysis of community dynamics by developing a coordinated set of R functions, and second, to demystify the relationships among univariate, multivariate, and RACs measures, and examine how each is influenced by the community parameters as well as data collection methods. We developed new functions for studying temporal changes and spatial differences in RACs in an update to the R package library(“codyn”), alongside other new functions to calculate univariate and multivariate measures of community dynamics. We also developed a new approach to studying changes in the shape of RAC curves. The R package update presented here increases the accessibility of univariate and multivariate measures of community change over time and difference over space. Next, we use simulated and real data to assess the RAC and multivariate measures that are output from our new functions, studying (1) if they are influenced by species richness and evenness, temporal turnover, and spatial variability and (2) how the measures are related to each other. Lastly, we explore the use of the measures with an example from a long-term nutrient addition experiment. We find that the RAC and multivariate measures are not sensitive to species richness and evenness and that all the measures detail unique aspects of temporal change or spatial differences. We also find that species reordering is the strongest correlate of a multivariate measure of compositional change and explains most community change observed in long-term nutrient addition experiment. Overall, we show that species reordering is potentially an understudied determinant of community changes over time or differences between treatments. The functions developed here should enhance the use of RACs to further explore the dynamics of ecological communities.
Original language | English (US) |
---|---|
Article number | e02881 |
Journal | Ecosphere |
Volume | 10 |
Issue number | 10 |
DOIs | |
State | Published - Oct 1 2019 |
Bibliographical note
Funding Information:This work was funded primarily by an LTER Synthesis Postdoctoral Fellowship at the National Socio-Environmental Synthesis center (SESYNC) to MLA. These ideas were originally developed in an LTER synthesis working group led by MLA and KK and further refined in a second funded LTER synthesis working group led by KK, MLA, and KRW from the LTER Network Communication Office LNCO at NCEAS NSF DEB 1545288. We thank members of those working groups: Emily Grman, David S. Johnson, Adam Langley, Bill Bowman, Alan Knapp, and Nathan Lemoine. We also thank Cynthia Chang for help developing the measures and testing them on the Mount St. Helens dataset. SESYNC is funded from the National Science Foundation DBI-1052875. In addition, partial support for this work was provided by NSF DBI 1262458, 1262377, and 1262463.
Publisher Copyright:
© 2019 The Authors.
Keywords
- R package
- codyn
- community composition
- long-term data
- multivariate analysis
- richness
- spatial variability
- temporal variability