Statistical analysis

All statistical analyses were performed using R version 4.0.5 with the packages: vegan v.2.5-7 (Oksanen et al., 2020), ade4 v. 1.7-18 (Dray et al., 2007), lme4 v.1.1-27.1 (Bates et al., 2015), glmmADMB v.0.8.3.3. (Bolker et al., 2012), MuMIn v.1.43.17 (Barton, 2016) statMatch 1.4.0 (D’Orazio) and ape v.5.5 (Paradis et al., 2021). Kettle holes were classified into two hydroperiod-types based on the timespan of water containment, i.e., permanent (kettle holes which contained water during more than half of the sampling campaigns) and ephemeral (kettle holes which contained water during half or less of the sampling campaigns). Temporal stability of the species composition of a kettle hole was assessed over two subsequent years (2019 and 2020), by comparing data from water samples collected in the same month across both years (n=60 from 15 kettle holes) using diversity indices (Species richness S, Shannon Index H’, and Simpson Index D1). A non-metric multidimensional scaling (NMDS) approach based on Bray-Curtis dissimilarities in combination with a Permanova was used to assess the difference in community species compositions within kettle holes across years. As these analyses did not reveal significant differences among 2019 and 2020, further analyses focused on the more densely sampled year 2019.
Seasonality in species numbers and composition of the open water zooplankton communities within a year was assessed across the eight sampling campaigns in 2019 (n= 91 metabarcoded samples from 24 kettle holes). Species richness (S) and Shannon Index (H’) were related to environmental parameters (pH, water temperature, surrounding field crops, kettle hole size, kettle hole location, numbers of neighbouring kettle holes in a 500 m radius, average wind direction/wind speed) with a generalized linear mixed effects model (Gelman & Hill, 2006; Zuur et al., 2009), using the lme4 package for Species richness (S) and glmmADMB package for Shannon Index (H’). Kettle hole ID, referring to repeated observations from the same kettle hole, was used as a random effect. The diversity as response variable (H’) and predictor variables (pH, water temperature, kettle hole size etc.) were Tukey-transformed to obtain a distribution approximate to normal. To identify the model that explains most of the variance, model selection was conducted based on the Akaike information criterion (AIC), using the dedgre function in the R package MuMIn (Barton, 2016). A NMDS approach based on Bray-Curtis dissimilarities in combination with Permanovas and environmental parameter correlation fitting (package vegan, function: envfit) was used to assess the difference in species composition and putative drivers of community assembly. To evaluate the influence of the individual kettle hole within a season on community composition, a Permanova based on the Bray-Curtis dissimilarities was conducted. To assess the impact of geographic/environmental proximity/distance based on km and Gower’s distance (Gower, 1971), a Mantel test was performed to correlate community dissimilarities with geographic/environmental distances among all pairs of kettle holes. To assess potential for dispersal from the resting stages located in the sediment (“vertical dispersal”), we compared soil and water samples from the same year (2019), using the same diversity indices and the NMDS approach described above.