Foundational biodiversity effects propagate through coastal food webs via multiple pathways

Abstract

Relatively few studies have attempted to resolve the pathways through which the effects of biodiversity on ecosystem functioning cascade from one trophic level to another. Here, we manipulated the richness of habitat-forming seaweeds in a western Atlantic estuary to explore how changes in foundation species diversity affect the structure and functioning of the benthic consumer communities that they support. Structural equation modeling revealed that macroalgal richness enhanced invertebrate abundance, biomass, and diversity, both directly by changing the quality and palatability of the foundational substrate and indirectly by increasing the total biomass of available habitat. Consumer responses were largely driven by a single foundational seaweed, although stronger complementarity among macroalgae was observed for invertebrate richness. These findings with diverse foundational phyla extend earlier inferences from terrestrial grasslands by showing that biodiversity effects can simultaneously propagate through multiple independent pathways to maintain animal foodwebs. Our work also highlights the potential ramifications of human-induced changes in marine ecosystems.

Keywords: biodiversity-ecosystem functioning; coastal marine ecosystems; epifaunal invertebrates; foundation species; niche complementarity; nonnative species; plant diversity effects; structural equation modeling.

FIGURE 1

Final observed structural equation model relating experimentally manipulated macroalgal richness to properties of both macroalgae and associated invertebrate consumer communities (see Appendix S1: Figure S2 for the null‐hypothesized model). Arrows represent directed effects (i.e., flow of causality from one variable to another). Standardized regression coefficients are shown next to the arrows in units of SD of mean, such that they can be compared fairly across response variables of differing units. Arrow widths are scaled by the standardized coefficients (Appendix S1: Table S1). Marginal (fixed effects only) and conditional (fixed + random effects) R ² values are also reported for each response variable.

FIGURE 2

Effects of macroalgal species identity and richness on metrics of primary and secondary production. Invertebrate (a) abundance, (b) dry biomass, and (c) taxonomic richness, and (d) macroalgal wet mass. Points are the time‐averaged response of each replicate over the final 4 weeks (n = 5 for all treatments). Colors correspond to the seven experimental macroalgal treatments (see Tukey's honestly significant difference test for the results of post hoc analysis). Gray underlying boxplots represent pooled response of all monocultures and polycultures, respectively (see t‐tests for comparisons of means). The results of one‐way analyses of variance (ANOVAs) are shown near the margin of each panel.

FIGURE 3

Nonmetric multidimensional scaling (NMDS) plot showing composition of invertebrate consumer community that colonized each experimental macroalgal treatment. Colors correspond to the seven experimental macroalgal treatments. Points represent the n‐dimensional response of each replicate and were calculated from the mean consumer abundance after time averaging over the final 4 weeks. Shaded hulls indicate the multidimensional space occupied by the invertebrate community in each treatment (see Appendix S1: Table S2 for results of pairwise planned contrasts). The results of permutational multivariate analysis of variance (PERMANOVA) based on the unweighted species abundances in each treatment are shown near the lower left margin (n = 5 for all treatments).

FIGURE 4

Net biodiversity effect partitioned into its component complementarity and selection effects for each response variable (rows) in multispecies treatments. (a) Invertebrate abundance, (b) biomass, (c) taxonomic richness and (d) macroalgal wet mass. Colors correspond to the three biodiversity effect components as indicated in the legend. Points were calculated from time‐averaged responses of each replicate over the final 4 weeks (n = 5 for all treatments), and all values were square‐root‐transformed with the original sign preserved (Loreau & Hector, 2001). Axis scales for biodiversity effect components are equivalent for each response variable (within rows), but not across response variables (among rows). Responses above the black line at 0 are positive, whereas those below are negative (see Appendix S1: Table S3 for results of t‐tests).