Species richness and identity both determine the biomass of global reef fish communities

Abstract

Changing biodiversity alters ecosystem functioning in nature, but the degree to which this relationship depends on the taxonomic identities rather than the number of species remains untested at broad scales. Here, we partition the effects of declining species richness and changing community composition on fish community biomass across >3000 coral and rocky reef sites globally. We find that high biodiversity is 5.7x more important in maximizing biomass than the remaining influence of other ecological and environmental factors. Differences in fish community biomass across space are equally driven by both reductions in the total number of species and the disproportionate loss of larger-than-average species, which is exacerbated at sites impacted by humans. Our results confirm that sustaining biomass and associated ecosystem functions requires protecting diversity, most importantly of multiple large-bodied species in areas subject to strong human influences.

Fig. 1. A conceptual representation of our decomposition quantifying the difference in biomass between two sites: a reference or reference site (B) and a comparison site (F).

Notation: S_c, S_uB, and S_uF refers to the number of species: in common, unique to B and unique to F, respectively; and Z_uB, Z_cB, Z_cF, and Z_uF to average ecological function (in this example, biomass) of species: unique to B, common to both sites when present at B, common to both sites when present at F, and unique to F, respectively. The first term, RICH-L, reflects loss of species from the reference site that are most like the average of the species that are retained. In Contrast I, the comparison site lacks some species present at the reference site (left column), so the species richness is lower at the comparison site (second column). In this case, the species in common between the two sites have the same average contribution to biomass (per species) as species unique to the reference (i.e., z_cB = z_uB, third column). Thus, the difference in biomass between the two sites in Contrast I is entirely captured by the RICH-L effect (last column). The second term, COMP-L, reflects the loss of biomass beyond what is expected given the number of species lost and average per species contribution of the reference species, i.e., the shared species. Unlike in Contrast I, in the case shown for Contrast II the species unique to the reference site are larger and therefore greater contributors (per species) to total biomass than the species that are shared between sites (${\overline{\mathit{z}}}_{\mathit{uB}}\mathit{-}{\overline{\mathit{z}}}_{\mathit{cB}}$ = δ_B > 0). The difference in total biomass between communities in Contrast II would be captured by the COMP-L term (last column) in addition to the RICH-L term (as shown for Contrast I). The RICH-G and COMP-G terms are analogous to RICH-L and COMP-L but arise from species that are present in the comparison site and absent from the reference (as illustrated in Contrasts III and IV). The final term, the “context-dependent effect” or CDE, considers only the species shared among both the reference and comparison sites. Differences in the contributions of shared species among the two sites reflect processes other than changes in richness and composition, such as changes in per species biomass, community size structure, resources, or the abiotic environment (e.g., temperature). In Contrasts I-IV, the shared species did not differ in their average per species contribution between the two sites (i.e., z_uF = z_cB) so there was no CDE. In Contrast V, the shared species differ in their average per species contribution to biomass, resulting in a nonzero CDE. In a real comparison, all five components can occur simultaneously. The five components sum to the observed difference in biomass between the reference and comparison communities.

Fig. 2

A map of study sites included in the analysis.

Fig. 3. Declines in fish biomass between sites are driven primarily by loss of species (RICH-L) and compositional losses (COMP-L).

Panels show the frequency distributions of each component based on differences between sites with the highest biomass and other nearby sites. Values have been standardized to the interval (−1, 1). Black points represent the mean of all observations ±95% confidence intervals (which are too small to be observed). RICH richness, COMP composition, L loss, G gain, CDE context-dependent effect, DIV total diversity effect (= RICH-L + COMP-L + RICH-G + COMP-G).

Fig. 4. Human population size predicts aspects of fish community structure.

Proximity to high densities of humans is associated with: A reduced total fish community biomass; B smaller observed median size classes; and C fewer species. Lines are predicted trends from generalized linear models.

Fig. 5. Contributions of fish size to total fish biomass at comparison and reference sites.

A A much larger proportion of the biomass of species unique to reference sites was attributable to large fishes (>200 cm), as compared to species that were only found at comparison sites. B Among those species that were shared between both the reference and comparison sites, small fish (<50 cm) made up a much larger proportion of the total biomass, especially at the comparison site. Points are medians ±95% quantiles, and the gray shaded area shows the underlying distribution of raw data. The total number of individuals represented in the plot is n = 41,267.