Published April 2, 2024 | Version v1
Dataset Open

Extending trait dispersion across trophic levels: predator assemblages act as top-down filters on prey communities

  • 1. Stanford University


Studies of community assembly typically focus on the effects of abiotic environmental filters and stabilizing competition on functional trait dispersion within single trophic levels. Predation is a well-known driver of community diversity and composition, yet the role of functionally diverse predator communities in filtering prey community traits has received less attention. We examined functionally diverse communities of predators (fishes) and prey (epifaunal crustaceans) in eelgrass (Zostera marina) beds in two Northern California estuaries to evaluate the filtering effects of predator traits on community assembly, and how filters acting on predators influence their ability to mediate prey community assembly. Fish traits related to bottom orientation selected for more clustered epifauna communities, and epifauna were generally overdispersed while fishes were clustered, suggesting that prey may be pushed to disparate areas of trait space to avoid capture by benthic sit-and-wait predators. We also found correlations between the trait dispersions of predator and prey communities that strengthened after accounting for the effects of habitat filters on predator dispersion, suggesting that habitat filtering effects on predator species pools may hinder their ability to affect prey community assembly. Our results present compelling observational evidence that specific predator traits have measurable impacts on the community assembly of prey, inviting experimental tests of predator trait means on community assembly, and explicit comparisons of how the relative effects of habitat filters and intraguild competition on predators impact their ability to affect prey community assembly. Integrating our understanding of traits at multiple trophic levels can help us better predict the impacts of community composition on food web dynamics as regional species pools shift with climate change and anthropogenic introductions.


Code file is written in R; data files are .csv files that can be opened and viewed with Microsoft Excel; phylogeny is a .txt that can be opened in a text editor and manipulated and analyzed in R. Table S1-S3 are .xlsx files viewable in Microsoft Excel. 

Funding provided by: University of California Natural Reserve System
Crossref Funder Registry ID:
Award Number:

Funding provided by: American Philosophical Society
Crossref Funder Registry ID:
Award Number:

Funding provided by: University of California, Davis
Crossref Funder Registry ID:
Award Number:

Funding provided by: National Science Foundation
Crossref Funder Registry ID:
Award Number: OCE-1829992

Funding provided by: National Science Foundation
Crossref Funder Registry ID:
Award Number: OCE-1829976


We sampled fish and epifaunal communities in 6 eelgrass beds in Bodega Harbor and Tomales Bay in the summers of 2019 and 2021. We sampled fishes in 6 sets of a custom beach seine net when the water level was at or below 1 m above the seafloor. The seine sampled a circular area of 11 m2. We counted, identified to the lowest possible taxonomic level (typically species), and released animals retained in the seine.

We sampled epifauna at 12 locations within each site separated by at least 10m and spanning a depth gradient of intertidal to shallow subtidal. We collected each sample by everting an open-mouth drawstring mesh bag (500 µm mesh size) over a clump of shoots in the eelgrass bed so that the mouth of the bag was flush with the sediment surface, cutting the shoots, and closing the drawstring to capture shoots, macroalgae, and associated animals. We transferred the shoots to the laboratory on ice, rinsed, and hand-inspected them to dislodge the epifauna, which we then passed through a 500 µm sieve and ultimately transferred into 70% ethanol. We then identified epifauna to the lowest possible taxonomic level (typically species). We also quantified the biomass of macroalgae in each epifaunal sample. We also measured water temperature, total eelgrass shoot density m-2, flowering shoot density m-2, percent cover, canopy height, and epiphyte dry weight mm-2 eelgrass as described by Aoki et al. (2022).

For the 23 most abundant species of peracarids in our surveys, we assigned values for 11 traits putatively related to predator avoidance and microhabitat niche. We collected three of these traits (maximum body size, shape, and living habit) from the literature. We determined the tube fidelity for each species according to observations of living and preserved specimens along a four-point ordered scale as follows: none (species lacks silk glands to build tubes), low (species has silk glands but was never observed in a tube alive or preserved in ethanol), medium (species has silk glands and was observed in tubes when alive but readily flees tube when exposed to ethanol), and high (species has silk glands, is tubicolous when alive, and is regularly found inside tubes after preservation in ethanol). We measured mean body size (length from rostrum to telson), relative eye diameter, and relative antenna lengths from 10-20 preserved individuals collected across sites and years. We measured activity levels as fractions of time spent swimming, walking, and still (unmoving) from one-minute video recordings of 10-20 live individuals per species across sites and years. We log-transformed peracarid traits where appropriate to conform to a normal distribution.

We assigned two categorical (vertical position and foraging mode) and one continuous trait (trophic level) to the 16 most abundant fishes based on the literature. We fuzzy-coded vertical position and foraging mode among 5 and 3 levels, respectively, to accommodate species that could be classified among multiple levels. We collected linear morphometric measurements of fishes (body and head dimensions, fin lengths, eye size and position, and mouth height and protrusion) from 2-26 specimens per species and size class collected from seines, and standardized them for ease of comparison across species.

This dataset includes average trait data for both peracarids and fishes.

To address the potential effects of evolutionary history on peracarid community responses to predators, we built a phylogeny of our species by subsetting from the ultrametric peracarid supertree published by Ashford et al. (2018). For species in our dataset that were not included in the supertree we substituted congeners or confamilials as needed.

Tables S1-S3 (on Zenodo) show modeled responses of residual peracarid community trait and phylogenetic dispersion to fish community-weighted mean trait values (Table S1), total fish community trait dispersion (Table S2) and residual fish community trait dispersion after accounting for habitat filters (Table S3).


Files (49.3 kB)

Name Size Download all
36.4 kB Preview Download
13.0 kB Preview Download