Table 2. Descriptions of the outputs obtained from analysis methods for temporal change in community composition with examples from the literature.
| Method (text section, R code) | Description of outputs | Key article(s) using each method |
|---|---|---|
| Raw dissimilarity values (Method 1.1; Code S3) | Various, depending on the method. For example, the dissimilarity-overlap method results in a graph of pairwise dissimilarity values against the proportion of individuals that taxonomically overlap; overlaid on this is a non-parametric, non-linear regression line, the slope of which indicates whether or not samples are changing in similar ways or not; this is compared to a null hypothesis of zero slope. | Bashan et al. (2016) |
| Mean dissimilarity (Method 1.1; Code S3) | Mean dissimilarity values for data subsets. Values can be used in subsequent graphs, analyses or maps. | Collins (1992) |
| Mean dissimilarity as response in regression (Method 1.1; Code S2) | Regression statistics indicating the significance of the relationship between mean dissimilarity and explanatory variables, which includes time if dissimilarity is calculated for different spatial samples, or will only include change in non-time explanatory variables, if dissimilarity is calculated across time. | Collins (1992); Barros et al. (2014) |
| Beta or zeta diversity time-lag graph (Methods 1.2 & 4; Code S4) | Graph showing the relationship between temporal diversity and time lag (distance in time). In the case of zeta and multi-site dissimilarity, time-lag graphs are specific to each number of samples in the subset (e.g. zeta order). See full description of zeta diversity below. | Hui & McGeoch (2014), Mcgeoch et al. (2017) |
| Beta or zeta time lag regression (Methods 1.2 & 4; Code S4) | Regression statistics indicating the significance of the relationship between temporal diversity and time lag (distance in time). In the case of zeta and multi-site dissimilarity, time-lag analyses are specific to each number of samples in the subset (e.g. zeta order). | Mcgeoch et al. (2017) |
| Beta diversity decomposition (nestedness, turnover) (Method 1.3, Code S2) | Beta diversity values for nestedness and turnover that can be used in subsequent graphs (e.g. time lag) or analyses of explanatory variables. | Baselga & Orme (2012) |
| Beta diversity decomposition (SCBD: species contributions to beta diversity; LCBD: local contributions to beta diversity) (Method 1.3; Code S2) | List of relative contributions of taxa to the overall compositional variation. Matrix of LCBD values showing the relative contribution of each sample-time combination to the overall compositional variation. Values can be used in subsequent graphs or analyses of explanatory variables. | Legendre & De Cáceres (2013), Legendre & Gauthier (2014), Lamy et al. (2015) |
| Ordination (constrained) (Method 1.4) | Ordination diagram and table of statistics (see the text section ‘Ordination’ for more details and other possible outputs). | Anderson & Willis (2003), http://ecology.msu.montana.edu/labdsv/R/ |
| Ordination site scores in general linear model or non-linear regression (Method 1.4; Code S2) | Regression statistics indicating the significance of the relationship between species composition and explanatory variables. | Day & Buckley (2013) |
| Temporal coherence (Method 1.4; Code S2) | A single value representing the degree of consistency in temporal change among samples for the dataset or subset considered (see the text section ‘Ordination’ for more details). | Rusak et al. (1999), Angeler & Johnson (2012) |
| Principal response curves (Method 1.4; Code S2) | Principal response curve diagram comparing trajectories in compositional space of different sites or samples over time relative to the baseline or control. | Auber et al. (2017) |
| Pivot days (Method 1.6; Code S2) | A list of time points is identified at which significant shifts in composition have occurred. | Lellouch et al. (2014) |
| Community trajectory analysis (Method 1.7) | Principal coordinates analysis diagram with trajectories showing change in species composition of samples through time. This is paired with diagrams showing the overall dissimilarity in complete trajectories between samples. Trajectory statistics, including shape, size and direction are also obtained. | De Cáceres et al. (2019) |
| Rank abundance distributions (Method 2; Code S4) | Species ranked from highest to lowest relative abundance on graph. Can be combined with dissimilarity methods and rank clocks. | Avolio et al. (2015), Hallett et al. (2016) |
| Venn diagrams, overlap and time cores* (Method 3; Code S2) | Venn diagram showing the number of taxa shared and unshared between two or more subsets of samples. Lists of shared and unshared taxa. Time cores results generate lists of taxa that occur consistently through time for a set of samples. Specifically useful for detecting taxon replacement over time | Vallès et al. (2014), O’Sullivan et al. (2015) |
| Zeta diversity (Method 4; Codes S3 & S4) | Zeta diversity values for each zeta order (number of samples in subsets) that can be used in subsequent graphs (e.g. zeta diversity vs. order) or analyses of explanatory variables | Hui & McGeoch (2014), McGeoch et al. (2019) |
| Synchrony (Method 5; Code S4) | A single value representing the degree to which taxa are changing in a similar way over time, calculated for a given sample by species matrix across a set of times. Can be compared for different time windows or different subsets of samples (e.g. experimental treatments) | Loreau & De Mazancourt (2008), Gouhier & Guichard (2014) |
| Turnover rates (Method 6; Code S3) | Percent change in species composition between time points | Diamond (1969) |
| Joint species distribution models (Method 7) | Observed vs. predicted patterns in taxon associations. Can test spatially and temporally-specific scenarios or hypotheses by modelling taxon associations in relation to environmental variables. Traits and phylogenetic information can be included. Presentation of results depends on the hypothesis of interest, for example predicted and observed changes in taxon communities can be mapped along environmental gradients | Ovaskainen et al. (2017a, 2017b) |
Note:
We provide a reference to the relevant text section within Step 2 and the location of example R code in supplementary material (Codes S1–S4). Where R code is not listed as available in supplementary material, it is available in the article cited for the method. LCBD refers to ‘Local Contributions to Beta Diversity’ and SCBD refers to ‘Species Contributions to Beta Diversity’.