Fig 1.
Summary food web (years, seasons combined).
Middle Panel: CONSUMERS (spider families). Solid bars are cursorial spiders, bars with horizontal lines are web spinners. Relative widths of triangles connecting a CONSUMER to a PREY category reflects percentages of prey in the CONSUMER’s diet. Width of a CONSUMER bar indicates number of prey items detected for that family (black scaling bar = 20; this scale also applies to top and bottom bars). Number of prey items detected per spider family ranged from 110 to 182. Total number detected = 1488. Top Panel: Spiders that were PREY, total number = 698. Bottom Panel: Non-spider PREY, total number = 790.
Fig 2.
Food webs for years 2009 through 2012 (seasons combined).
Key as in Fig 1, except black scaling bar represents 10 individuals. The overall length of the middle panel does not vary between years; however, the width of the black scaling bar changes each year, sometimes substantially, because the total number of prey detected each year (N) varied. (a) 2009, N = 223, (b) 2010, N = 266, (c) 2011, N = 663 and (d) 2012, N = 336.
Fig 3.
Food webs for spring, summer and fall (years combined).
Key as in Fig 2. The width of the black scaling bar changes because the total number of prey detected each season (N) varied. (a) Spring, N = 421, (b) Summer, N = 575 and (c) Fall, N = 492.
Fig 4.
Median restricted connectance, with 95% BCI.
(a) Year (seasons pooled) and (b) Season (years pooled). Median values are about ~25% of the maximum possible.
Fig 5.
Median interaction evenness, with 95% BCI.
(a) Year (seasons pooled) and (b) Season (years pooled). Median values are about ~90% of the maximum possible.
Fig 6.
(a) Year (seasons pooled) and (b) Season (years pooled). % IGPrey is the percentage of total prey that is other spiders that are part of an IGP module.
Fig 7.
Pattern of seasonal change in % IGPrey, with 95% CI, across four years.
(a) 2009, (b) 2010, (c) 2011 and (d) 2012. The seasonal shift in % IGPrey was not paralleled by any seasonal change in the proportion of interaction pathways that were IGP modules (S3 Table). The best model for the 4 x 3 x 2 contingency table of changes in % IGPrey did not include an interaction between season and year; thus, there is no evidence suggesting that the pattern of seasonal change in % IGPrey differed across years. The best model did include an effect of season, but no effect of year, on % IGPrey–a result that is consistent with the pattern here and in Fig 6. Details of the modeling results can be accessed on Dryad: https://doi.org/10.5061/dryad.dz08kps43.