Skip to contents
library(tidydiversity)
library(dplyr)
library(ggplot2)
ants <- filter(png_ants, season == "wet")

The concept

Whittaker (1960) described diversity at three scales:

Scale Meaning Example
Alpha (α\alpha) Diversity within a site Ant species at one elevation
Gamma (γ\gamma) Diversity of all sites pooled Ant species on the whole mountain
Beta (β\beta) How much the sites differ How much the elevations replace each other’s species

With Hill numbers the three are linked by multiplicative partitioning (Jost 2007; Chao et al. 2012):

γ=α×β⇒β=γ/α \gamma = \alpha \times \beta \qquad\Longrightarrow\qquad \beta = \gamma / \alpha

β\beta is then the effective number of completely distinct communities: it ranges from 1 (all sites identical) to NN, the number of sites (no species shared). Because β\beta depends on NN, it is converted to dissimilarities between 0 and 1 that can be compared across studies:

Dissimilarity Weighs Generalises
Sørensen-type (1−CqN1 - C_{qN}) Overlap from the point of view of an average site Sørensen index at q = 0
Jaccard-type (1−UqN1 - U_{qN}) Overlap from the point of view of the pooled community Jaccard index at q = 0

As with alpha diversity (Chapter 3), beta diversity depends on sampling effort, so tidydiversity computes it at an equal sample coverage (Chao et al. 2023).

1. Gamma, alpha and beta along the mountain

bh <- beta_hill(ants, q = 0:2)
knitr::kable(select(bh, q, coverage, gamma, alpha, beta, dis_sorensen, dis_jaccard),
  digits = 3,
  caption = "Multiplicative partitioning at equal coverage across the five elevations."
)
Multiplicative partitioning at equal coverage across the five elevations.
q coverage gamma alpha beta dis_sorensen dis_jaccard
0 0.957 78.929 27.909 2.828 0.457 0.808
1 0.957 39.359 13.628 2.888 0.659 0.659
2 0.957 22.178 7.292 3.041 0.839 0.510

Interpretation. For species richness (q=0q = 0), the mountain holds 78.9 effective species against 27.9 per site, so β\beta = 2.83: the five elevations behave like almost three completely distinct communities. The Sørensen-type dissimilarity of 0.46 means that, on average, a site shares only about 54 % of its species with the other elevations. Dissimilarity is higher at q=1q = 1 and q=2q = 2: the dominant species differ between elevations even more than the species lists do.

2. Turnover or nestedness?

The concept

Two sites can differ in two ways (Baselga 2010):

Component What happens Ecological meaning
Turnover (replacement) Species are replaced by different species Environmental filtering, dispersal limits, history
Nestedness The poorer site holds a subset of the richer site’s species Species loss, e.g. along a stress gradient or after disturbance

The Sørensen dissimilarity splits exactly into these two parts: βSOR=βSIM+βNES\beta_{SOR} = \beta_{SIM} + \beta_{NES}.

pairs <- beta_pair(ants)
knitr::kable(pairs, digits = 3,
  caption = "Pairwise Sørensen dissimilarity (total) split into turnover and nestedness."
)
Pairwise Sørensen dissimilarity (total) split into turnover and nestedness.
site_1 site_2 turnover nestedness total
MW0200 MW0700 0.619 0.054 0.673
MW0200 MW1200 0.800 0.005 0.805
MW0200 MW1700 0.900 0.035 0.935
MW0200 MW2200 1.000 0.000 1.000
MW0700 MW1200 0.550 0.075 0.625
MW0700 MW1700 0.900 0.047 0.947
MW0700 MW2200 1.000 0.000 1.000
MW1200 MW1700 0.600 0.133 0.733
MW1200 MW2200 1.000 0.000 1.000
MW1700 MW2200 0.800 0.067 0.867
multi <- beta_multi(ants)
knitr::kable(multi, digits = 3,
  caption = "The same partition for all five sites together (Baselga 2010)."
)
The same partition for all five sites together (Baselga 2010).
n_sites turnover nestedness total
5 0.79 0.074 0.864

Interpretation. Across all sites, turnover (0.79) makes up 91 % of the total dissimilarity (0.86); nestedness is small. So higher elevations do not simply hold a subset of the lowland fauna: different species replace each other up the mountain. This points to elevation-specific assemblages (filtered by temperature, for example), not to the loss of species from one regional pool.

3. Distance decay

The concept

Communities become less similar as the distance between them grows (Nekola and White 1999), in space or along an environmental gradient. The rate of decay summarises how fast species composition changes. Here the “distance” is the difference in elevation.

decay <- beta_decay(ants, gradient = elevation)
decay
#> # A tibble: 10 × 8
#>    site_1 site_2 turnover nestedness total gradient_1 gradient_2 distance
#>    <chr>  <chr>     <dbl>      <dbl> <dbl>      <int>      <int>    <dbl>
#>  1 MW0200 MW0700    0.619    0.0544  0.673        200        700      500
#>  2 MW0200 MW1200    0.8      0.00488 0.805        200       1200     1000
#>  3 MW0200 MW1700    0.9      0.0355  0.935        200       1700     1500
#>  4 MW0200 MW2200    1        0       1            200       2200     2000
#>  5 MW0700 MW1200    0.55     0.075   0.625        700       1200      500
#>  6 MW0700 MW1700    0.9      0.0474  0.947        700       1700     1000
#>  7 MW0700 MW2200    1        0       1            700       2200     1500
#>  8 MW1200 MW1700    0.6      0.133   0.733       1200       1700      500
#>  9 MW1200 MW2200    1        0       1           1200       2200     1000
#> 10 MW1700 MW2200    0.8      0.0667  0.867       1700       2200      500
ggplot(decay, aes(distance, 1 - total)) +
  geom_point(size = 2.5, colour = "#0072B2") +
  geom_smooth(method = "lm", formula = y ~ x, colour = "black", se = FALSE) +
  labs(x = "Difference in elevation (m)", y = "Similarity (1 - Sørensen)") +
  theme_tidydiversity()

Distance decay of ant community similarity with elevation difference

Interpretation. Similarity falls steadily with elevation difference: sites 500 m apart share far more species than sites 2,000 m apart, which share almost none. For a formal test of the slope (pairs are not independent, so ordinary regression p-values are wrong), use beta_decay_model(), which tests by permutation.

How to report it

Beta diversity was quantified by multiplicative partitioning of Hill numbers at equal sample coverage (Chao et al. 2023) in tidydiversity. Across the five elevations, gamma richness was 78.9 and mean alpha richness 27.9 (beta = 2.83; Sørensen-type dissimilarity 0.46). Partitioning Sørensen dissimilarity (Baselga 2010) showed that turnover accounted for 91 % of compositional differences.

References

Baselga, A. (2010). Partitioning the turnover and nestedness components of beta diversity. Global Ecology and Biogeography, 19, 134-143.

Chao, A., Chiu, C.-H. and Hsieh, T. C. (2012). Proposing a resolution to debates on diversity partitioning. Ecology, 93, 2037-2051.

Chao, A. et al. (2023). Rarefaction and extrapolation with beta diversity under a framework of Hill numbers: the iNEXT.beta3D standardization. Ecological Monographs, 93, e1588.

Jost, L. (2007). Partitioning diversity into independent alpha and beta components. Ecology, 88, 2427-2439.

Nekola, J. C. and White, P. S. (1999). The distance decay of similarity in biogeography and ecology. Journal of Biogeography, 26, 867-878.

Whittaker, R. H. (1960). Vegetation of the Siskiyou Mountains, Oregon and California. Ecological Monographs, 30, 279-338.