Habitat fragmentation alters post-dispersal more than pre-dispersal seed predation through edge effects

Smith, Thomas 1 ; Brudvig, Lars2; Burt, Melissa3; Damschen, Ellen1; Gross, Kevin4; Levey, Douglas5; Orrock, John1; Resasco, Julian6; Haddad, Nick2

Published Mar 19, 2026 on Dryad. https://doi.org/10.5061/dryad.0p2ngf2g6

Data files

Mar 19, 2026 version files 653.80 KB

Abstract

In addition to causing loss of biodiversity, habitat fragmentation can alter species interactions. Relative to well-studied effects of fragmentation on mutualistic interactions like pollination, effects on antagonistic interactions remain poorly understood. One such interaction, seed predation, may be particularly important because it can alter plant demographics and trigger cascading changes to both plant and seed predator communities. Previous studies have failed to find consistent effects of fragmentation on seed predation, possibly because seed predation occurs at two distinct stages relative to seed dispersal (pre- and post-dispersal). Working within a large-scale fragmentation experiment, we tested effects of three consequences of fragmentation – edge-to-area ratio, distance to edge, and connectivity (via habitat corridors) - for pre-dispersal seed predation and post-dispersal seed removal of eight plant species of conservation interest. We found that fragmentation effects, especially distance to edge, were more likely significant for post-dispersal seed removal (assumed to positively correlate with seed predation) than for pre-dispersal seed predation, although the significant patterns differed across species. Post-dispersal seed predation declined sharply with seed mass and this decline was greater at the patch edge. This led to larger seeded species having less predation nearest to the edge compared to patch interiors. This experiment highlights how fragmentation can alter seed predation by generating edge effects and how species traits can predict consequences across a community.

Dataset DOI: 10.5061/dryad.0p2ngf2g6

Description of the data and file structure

File List

File 1: predispersal_seedpred.csv

Description: Data on the predispersal seed predation for founder plants in the Corridor Project. Contains information on rates of seed production and predation at the plant level (each row) before dispersal across experimental sites.

File 2: postdispersal_seedpred.csv

Description: Data on the postdispersal seed predation for founder plants in the Corridor Project. Presents the results from postdispersal seed predation trials using seed depots placed at four plot locations with increasing distance from the habitat edge.

Variable List

File 1: predispersal_seedpred.csv

species: These are the 6 letter species codes used to identify plant species. There are five species in this dataset: ARISTR = Aristida beyrichiana, ANTVIL = Anthenantia villosa, CARBEL = Carphephorus bellidifolius, LIASTI = Liatris squarrulosa, SORSEC = Sorghastrum secundum

block: Factor variable for the eight experimental landscape blocks: "EU8", "EU10", "EU52", "EU53N", "EU53S", "EU54N", "EU54S", or "EU57".

patch: Letter associated with the five replicate patches in each landscapes: A-E

patchtype: Factor variable indicating the four patch types: Center, Connected, Winged, and Rectangle.

patchtype2conn: Factor variable for the patch types but with Center patches treated as Connected patches.

corner: Roman numeral for each of the 4 transects at the corners in each patch: I-IV.

plot: Letter for each sampling plot along a transect: a-d.

edgedist: The distance in meters (m) that each plot is from the corner edge: a = 1.5 , b = 14, c = 27, d = 51.

seedsize: Seed size measurement in grams (g) for each species calculated from a random sample of 50 seeds from the study area.

numstalks: The number of reproductive structures (infructescences or culms) collected for an individual plant

develseed: The average number of seeds per reproductive structure that showed evidence of developing through the presence of intact endosperm and embryo

nondevelseed: The average number of seeds potential seeds (achenes/caryopses) that failed to show the presence of an intact endosperm or embryo

estdevelseed: The estimated average number of seeds per reproductive structure that showed evidence of developing but adjusted by the number missing seeds due to dispersal or other non-predator related means.

estnondevelseed: The estimated average number of potential seeds (achenes/caryopses) that failed to show the presence of an intact endosperm or embryo adjusted by the number missing due to dispersal or other non-predator related means.

predispersalpred: The proportion of the nondeveloped potential seeds (achenes/caryopses) that showed signs of predation. This was calculated from up to ten randomly selected nondeveloped achenes/caryopses from a single reproductive structure for the plant.

prednondev: The calculation of the number of nondeveloped potential seeds (achenes/caryopses) that showed predispersal seed predation. This is the variable "predispersalpred" multiplied by the "estnondevelseed".

totdev: The calculation of the total number of seeds that would develop including those that had predispersal seed predation. The variable "estdevelseeds" added to "prednondev".

predrate: Predispersal seed predation rate calculated from number potential seeds showing predispersal seed predation ("prednondev") divided by the total potential number of seeds ("totdev")

plantpredocc: Binary variable of whether a plant showed any predispersal seed predation (1) or none (0).

File 2: postdispersal_seedpred.csv

species: These are the 6 letter species codes used to identify plant species. There are eight species in this dataset: ARISTR = Aristida beyrichiana, ANTVIL = Anthenantia villosa, CARBEL = Carphephorus bellidifolius, GAYDUM = Gaylussacia dumosa, LIASTI = Liatris squarrulosa, NOLGEO = Nolina georgiana, PHYAME = Phytolacca americana, SORSEC = Sorghastrum secundum

round: Sampling round of the two post-dispersal seed predation trials conducted for each species: Either roman numeral "I" or "II"