Untangling the web: How spider traits link with management practices in agroecosystems

This dataset accompanies the manuscript "Untangling the Web: How Spider Traits Link with Management Practices in Agroecosystems". It includes the raw input matrices used to explore the functional composition of spider communities across agroecological and conventional cereal fields in the Swiss lowlands.

The dataset consists of four tables:
(1) L_spiders: Relative abundances of 84 spider species across 22 field sites, derived from standardized multi-method sampling (pitfall traps, sweep netting, vacuum suctioning).
(2) Q_traits: Functional traits of the observed spider species, including hunting guild, dispersal ability, vegetation stratum, preferred habitat, and body size.
(3) R_filters: Site-level environmental and management variables such as pesticide input (TFI), nitrogen application, and landscape metrics.
(4) R_veg: Local vegetation characteristics based on percentage cover of plant species per site. From this, community-weighted means (CWMs) for plant height, specific leaf area, and leaf nitrogen content, as well as Rao’s functional diversity (RaoQ), were computed.

These data were prepared for RLQ and fourth-corner analyses to assess how spider traits respond to environmental filtering through direct and vegetation-mediated pathways.

Methods
Spiders were sampled in 22 cereal fields (wheat, barley, oilseed rape) in the Swiss lowlands as part of the PestiRed project. Each field pair consisted of one conventionally and one agroecologically managed site, matched by crop type. Sampling took place during the 2023 growing season using three complementary methods—pitfall trapping, sweep netting, and vacuum suctioning—to cover different vegetation strata. Only adult spiders were retained and identified to species level. A total of 55 species occurring in at least two sites were included in the analysis.

For the analysis, the spider species abundances across sites have been Hellinger-transformed (L matrix). The Q matrix includes five functional traits: hunting guild, dispersal ability, vegetation stratum, habitat type, and body size. Trait data were compiled from expert sources and online databases, then clustered to reduce dimensionality.

The Renv matrix consists of seven z-transformed environmental variables reflecting field-level management (e.g., pesticide use, mechanical interventions), nitrogen input, vegetation diversity, and landscape structure (e.g., diversity and patchiness of land use within 500m). The Rveg matrix represents local vegetation characteristics, including community-weighted means of specific leaf area, plant height, and leaf nitrogen content, dominant plant family per site, and functional vegetation diversity (RaoQ), all z-transformed.

These matrices were used in RLQ and fourth-corner analyses to explore trait–environment relationships.