Fig 1.
A) Map of California (USA) with the San Francisco Bay-Delta watershed. The inset is a finer-scale map of the Delta, with the focal region in the Cache Slough Complex outlined with a dashed box. B) The Cache Slough complex, with sampling areas circled. Each area contained four sampling stations, each with a different habitat type: SAV, FAV, EAV, or Channel. Maps are the product of the author, and use data from the National Watershed Boundary Dataset [8] and CDFW’s Vegetation Classification and Mapping Program [9].
Table 1.
Sampling stations and habitat types, with average environmental parameters when samples were collected.
DO = dissolved oxygen in milligrams per liter, Temp = surface temperature in degrees Celsius, SC = specific conductance in micro-Siemens per centimeter, Turb. = Turbidity in nephelometric turbidity units, and Depth of water in meters. Latitude and Longitude are in WGS 1984.
Table 2.
Predictor variables for explaining observed differences in catch and taxa richness.
Fig 2.
Distribution of total catch in each area and habitat type.
Sample size in parentheses below boxes. Four outliers in FAV and SAV samples with catch > 1000 not shown. Models support significantly higher catch in SAV than in other habitats and significantly higher catch with the sweep net in FAV (see Table 3).
Table 3.
Coefficients for the negative binomial mixed model with a log link predicting total invertebrate catch of sweep nets and leaf packs Model: Catch ~ Habitat*Sampler + Month + Error(Area/Habitat).
Fig 3.
Distribution of taxa richness for sweep nets and leaf packs in various habitat types.
Sample size is in parentheses along the x-axis. Models support significantly higher richness for samples collected with sweep nets, and significantly higher richness for FAV and SAV samples than EAV or channel samples.
Table 4.
Coefficients for the linear mixed model predicting taxa richness of sweep nets and leaf packs Model: Catch ~ Habitat + Sampletype + Month + Error(Area/Habitat).
The interaction term was not significant and so was dropped from the final model.
Fig 4.
Relative abundance of major taxa in samples collected with leaf packs and sweep nets in various habitats in the three different areas (Liberty, Lindsey, and Miner).
Taxa that made up less than 0.5% of the total catch were combined into the “other” category to simplify the graph. PERMANOVA showed significant differences between habitat types, between areas, and between sample types (Table 5). The “fish” category included juvenile sunfish (family Centrarchidae), Mississippi silversides (Menidia beryllina), and gobies (genus Tridentiger), all non-native species.
Fig 5.
A) Non-metric multidimentional scaling plots (NMDS) of propotional taxonomic abundance with colored elipses showing the standard deviation of group centroids by habitat type and black outlined elipses showing grouping by sample type. Points represent samples, text represents species. Stress = 0.18. B) Plot of the same NMDS as 5A, but with colored elipses showing the standard deviation of group centroids by Area. Black outlined elipses show grouping by sample type.
Table 5.
Results of PERMANOVA performed on the entire data set and on subsets of the dataset using sweep nets only or leaf packs only.
Table 6.
Taxa which were significantly associated with a particular sample type, area, or habitat, mean abundance by group (abund.), proportion of samples where taxa was present (pres.), and statistical significance of the association.
Table 7.
Taxa which were significantly associated with a particular area mean abundance by group (abund.), proportion of samples where taxa was present (pres.), and statistical significance of the association.
Table 8.
Taxa which were significantly associated with a particular habitat, mean abundance by group (abund.), proportion of samples where taxa was present (pres.), and statistical significance of the association.