Desiccation risk favours prevalence and diversity of tardigrade communities and influences their trophic structure in alpine ephemeral rock pools

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY 4.0 https://creativecommons.org/licenses/by/4.0/ Desiccation risk favours prevalence and diversity of tardigrade communities and influences their trophic structure in alpine ephemeral rock pools © 2022 the Authors Published version Vecchi, Matteo; Ferrari, Claudio; Stec, Daniel; Calhim, Sara Vecchi, M., Ferrari, C., Stec, D., & Calhim, S. (2022). Desiccation risk favours prevalence and diversity of tardigrade communities and influences their trophic structure in alpine ephemeral rock pools. Hydrobiologia, 849(9), 1995-2007. https://doi.org/10.1007/s10750-022-04820-0 2022 Vol.: (0123456789) 1 3 Hydrobiologia https://doi.org/10.1007/s10750-022-04820-0 PRIMARY RESEARCH PAPER Desiccation risk favours prevalence anddiversity oftardigrade communities andinfluences their trophic structure inalpine ephemeral rock pools MatteoVecchi · ClaudioFerrari · DanielStec · SaraCalhim Received: 3 June 2021 / Revised: 13 January 2022 / Accepted: 10 February 2022 © The Author(s) 2022 extracting and classifying them from rock pools in the Italian Apennines. Rock pools were divided into three types, based on maximum duration of their inundation period. Following the patterns usually observed with rock pool invertebrates, we tested the hypothesis that desiccation has a negative effect on prevalence, abundance, and diversity of tardigrades, and thus could alter the trophic structure of their communities. In contrast to what is commonly found for other animal groups in the same type of habitat, we found that tardigrades were more diverse and prevalent in shallower, more desiccation prone, rock pools. Moreover, the trophic structure of tardigrade communities was different among the different pool types. Lastly, we also provide DNA barcodes of the most commonly found taxa. Ultimately, our study demonstrates that tardigrade communities in rock pools provide a valuable model system for the study of abiotic factors influencing meiofauna communities. Keywords Tardigrada· Rock pools· Desiccation· DNA barcoding· Community ecology· Meiofauna Introduction Rock pools are eroded depressions that occur in a matrix of bedrock that accumulate a layer of unconsolidated bottom sediment by trapping material transported by runoff and wind (Brendonck etal., 2016). These habitats occur all over the globe in all major Abstract Rock pools are ephemeral freshwater habitats characterized by their small size, well-defined boundaries, and periodic desiccation, making them ideal model systems to answer numerous ecological questions. Although there are numerous studies on rock pool fauna around the world, tardigrades have only rarely been recorded. We conducted the first tardigrade-focused study on rock pools by quantitatively Handling editor: Dani Boix Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1007/ s10750- 022- 04820-0. M.Vecchi(*)· S.Calhim Department ofBiological andEnvironmental Science, University ofJyvaskyla, PO Box35, 40014Jyvaskyla, Finland e-mail: matteo.m.v[email protected]; matteo.v[email protected] C.Ferrari Department ofChemistry, Life Sciences andEnvironmental Sustainability, University ofParma, Parco Area delle Scienze 33/A, 43124Parma, Italy D.Stec Institute ofSystematics andEvolution ofAnimals, Polish Academy ofSciences, Sławkowska 17, 31-016Kraków, Poland D.Stec Institute ofZoology andBiomedical Research, Jagiellonian University, Gronostajowa 9, 30-387Kraków, Poland Hydrobiologia 1 3 Vol:. (1234567890) biomes and depend mainly on precipitation for filling (Jocqué etal., 2010). Therefore, they are ephemeral, generally oligotrophic (Jocqué etal., 2010) and experience extreme fluctuations in their abiotic parameters (especially pH, temperature, and dissolved oxygen; De Vries, 1996; Brendonck etal., 2000). Rock pools have been proposed as a model system in ecology and evolutionary studies (Srivastava et al., 2004; Brendonck etal., 2010) as they are small, pristine environments with clear boundaries; thus, they can be subjected to experimental manipulations and generally host a simple community. About 460 aquatic animal species have been recorded from freshwater rock pools around the world (Jocqué etal., 2010). These animals can be classified as passive or active dispersers. Passive dispersers (for example tardigrades; Ptatscheck et al., 2018; Fontaneto, 2019) disperse mainly in a resting stage via wind and overflow of water between pools, and thus can be rock pool specialists. Active dispersers are those that can actively migrate (usually only during the adult stage) between different pools, have broad distributions and occur in a wide range of temporary habitats in addition to rock pools. The main factors driving animal community composition in rock pools are the size and hydroregime of the rock pool (Jocqué etal., 2007; Vanschoenwinkel etal., 2009), with passive disperser being more influenced by those environmental parameters than active dispersers (Vanschoenwinkel etal., 2009). Different responses of animal groups have been found to respond differently to the hydroregime. For example, diversity and abundance of insects and crustaceans decreased in rock pools with shorter hydroperiods in a Jamaican coastal system (Therriault & Kolasa, 2001). The opposite pattern was reported for rotifers and tardigrades by Jocqué et al. (2007), who found higher densities in pools with shorter hydroperiods. Developmental constraints, such as development time, can prevent larger organisms and predators from occurring in pools with shorter hydroperiods, which, in turn, allows the proliferation of smaller and less competitive organisms like tardigrades adapted to ephemeral habitats (Tevis, 1966; Jeffries, 1994; Jocqué etal., 2007). Tardigrada (also known as water bears or moss piglets) is a phylum of micrometazoans of about 50–1200µm in size (Møbjerg etal., 2018), with more than 1300 species described worldwide (Guidetti & Bertolani, 2005; Degma & Guidetti, 2007; Degma et al., 2021). When colonizing terrestrial environments, tardigrades require at least a film of water surrounding their bodies to perform activities necessary for life. However, because of their ability to enter cryptobiosis (a temporary and reversible suspension of metabolism) in all the phases of their life cycle, some species of limnoterrestrial tardigrades can withstand desiccation (anhydrobiosis; Wright, 2001) and freezing (cryobiosis; Guidetti etal., 2011). As a result of their small size and metabolic adaptations, tardigrades can colonize a variety of environments, ranging from limnoterrestrial (leaf litter, soil, mosses, lichens) to aquatic habitats (periphyton, sediment) (Nelson etal., 2018) and with extreme climatic conditions (for example glaciers (Zawierucha etal., 2016) and deserts (Neher etal., 2009)). Tardigrade records in rock pools are extremely limited, having been recorded only seven times in the scientific literature. Interestingly, although only few studies have recorded their existence, their occurrence has been cited in rock pools throughout the world (Namibia, Koste (1996); Ivory Coast, Snoeks etal. (2021); South Africa, De Vries (1996); Israel, Spencer et al. (1999); Utah, USA, Jocqué et al. (2007); Spain, Velasco-González et al. (2020); Australia, Boix etal. (2016)), emphasizing a potential association between tardigrades and rock pools. The main aim of this paper is to test if tardigrades in rock pools represent a good model system for the study of the effects of extreme environmental stresses (desiccation) on community composition. Specifically, following the observations of Jocqué etal. (2007) that tardigrades were more abundant in rock pools with shorter hydroperiods, we tested the hypothesis that tardigrade prevalence, abundance and diversity are higher in rock pools that desiccate faster and more often. Materials andmethods Study area The studied rock pools are located inside the “Parco Nazionale dell’Appennino Tosco-Emiliano” (Tuscan-Emilian Apennine National Park), Italy. The area is characterized by a complex geological structure (Mariani etal., 2018). All the main peaks and most of ridge’s summits of the Tuscan-Emilian Hydrobiologia 1 3 Vol.: (0123456789) district are made of sandstones belonging to the Macigno formation. During the Last Glacial Period (115,000–11,700years ago), glaciers covered almost all the peaks of the northern Apennines and the moving ice left different geomorphological formations such as sheepback rocks. All the rock pools studied are made of sandstone in sheepback rocks. In the study area, the vegetation of montane belt is characterized by Fagus sylvatica L., 1753 forests (Foggi & Rossi, 1996) and Vaccinium-heaths for the subalpine belt (Pignatti, 1994; Ferrari & Piccoli, 1997; Tomaselli etal., 2019). The yearly average temperature recorded from the closest climatic station (Lake Paduli, Massa-Carrara; Lat: 44.35 Lon: 10.14) is 8.2°C (min −16.0, max 31.3) with 75days of frost and 2268mm of precipitations (SM.01). Rock pools sampling and characterization Fifty-six rock pools from 11 clusters were selected for this study. Clusters (Fig.1a–b, SM.01) are isolated rocky outcrops where pools (from 1 to 19 per cluster) occurred. Cluster locations are shown in Fig.2. Due to the irregular shape and absence of water during the sampling, the maximum water level (depth) was impossible to measure accurately but was estimated. Pools were divided into three categories based on the approximated maximum depth (Fig. 1c–d, SM.02): “Pans” (< 2 cm), “Intermediates” (> 2 cm, < 5 cm) and “Holes” (> 5cm). We used maximum depth to divide the pools in three classes because this gives a reasonable indication of the maximum duration of the inundation period (Jocqué et al., 2006; Vanschoenwinkel etal., 2009; Anusa etal., 2012). Thus, the three categories also represent three different classes of inundation period maximum duration (short for Pans, intermediate for Intermediates and long for Holes). Rainfall and temperature also influence rock pools inundation period (Marcus & Weeks, 1997), but those environmental factors can be considered almost identical between the analysed pools due to their geographic proximity and similar elevation. Sampling was conducted during April–May 2019 and June 2020 on three occasions, however most of the samples were collected in the last sampling trip (SM.01). Each pool was sampled only once; however, during the multiple sampling visits all the analysed rock pools were completely dry at least once (i.e., none of them hold water permanently). The sediment of each pool was manually homogenized (to avoid Fig. 1 Photographs of rock pools taken during their dry phase. a Location of pool cluster C; b Location of pool cluster E; c Pool C7 from cluster C; d Pool E11 from cluster E. Photos courtesy of Tommaso Sandri Hydrobiologia 1 3 Vol:. (1234567890) sampling error due to low-scale variability in meiofauna distribution; Gansfort etal., 2018) and scraped with a clean metal spoon. The sediments were collected in clean plastic bags (when wet) or paper bags (when dry). Wet samples were left to dry protected from light at room temperature (20 ºC ± 1), then kept frozen at −20°C until use. One aliquot of rock pool E19, which contained water when sampled, was kept frozen instead of being dried, and was later examined for potential predators of tardigrades. Tardigrade extraction Dry samples were sieved through a 1mm screen to remove stones and large organic fragments that could bias the weighing. Aliquots of the sieved dry samples (from 0.05 to 5.39g, according to the amount of available sample; SM.01) were placed in 50mL plastic tubes and left to rehydrate overnight with 10mL of distilled water. Then 30mL of boiling distilled water was added to each tube to kill and relax all the fauna and avoid live tardigrades griping onto sediment fragments and avoiding extraction. Fauna extraction was performed by centrifugation with Ludox according to Bartels & Nelson (2006). From selected samples, an additional portion (not counted for the calculation of tardigrade densities) was processed to extract more specimens for DNA sequencing. Samples for DNA sequencing were chosen to maximise the diversity of sequenced taxa by selecting the ones with abundant populations and target groups that are taxonomically challenging and likely comprise cryptic/pseudocryptic species. Other fauna extraction Rotifers, mites and nematodes were consistently extracted as collaterals of tardigrade extraction. These were only semi-quantitatively analysed and classified as Absent, Rare, Common or Abundant (see definitions in SM.01). When occasionally chironomid carcasses were retrieved, they were recorded (SM.01). The frozen aliquot of pool E19 was thawed and qualitatively examined for the presence of potential predators (SM.01). Microscopy and tardigrade identification Specimens for light microscopy were mounted on microscope slides in a small drop of Hoyer’s medium and secured with a cover slip, following the protocol by Morek etal. (2016a, b). When more than 80 Fig. 2 Locations of the sampled rock pool clusters Hydrobiologia 1 3 Vol.: (0123456789) tardigrades were extracted from a sample (only in pools C12 and E1), only a subsample of about 80 individuals were mounted on slides (the number was determined based on the maximum number of animals per sample that could be analysed with the available time and resources; a similar approach was used in Bartels & Nelson, 2006). Slides were examined under a Zeiss Axio Scope A1 light microscope with phase and Nomarski differential interference contrast (PCM and NCM, respectively; named collectively as light contrast microscopy, LCM). Tardigrades were identified to genus level as it was not possible to identify every individual to the level of nominal species. Genera identification was performed following Ramazzotti & Maucci (1983) and descriptions/redescriptions or recent reviews of the identified genera: Claxton (1998), Pilato (1992), Biserov (1997–98), Bertolani etal. (2014), Morek etal. (2016a, b), Vecchi etal. (2016), Hansen etal. (2017), Stec etal. (2018), Gąsiorek etal. (2019a, b), Gąsiorek & Michalczyk (2020), Tumanov 2020 and Stec etal. (2021). A quantitative faunistic table of the genera found in each sample is provided in SM.01. Tardigrade genera were divided in two main functional feeding groups according to their ability to feed on other animals and thus the ability to exploit them as resources: Carnivorous (Carnivorous + Omnivorous) and Non-Carnivorous (Herbivorous + Microbivorous). Division into feeding groups was done according to Zawierucha etal. (2019). Data analysis Bayesian Generalized Linear Mixed Models (BGLMM) were used to estimate the unbiased prevalence (defined as the probability of finding at least one tardigrade by analysing one gram of dry substrate, the number of individuals per gram of dry substrate (abundance), the number of taxa (diversity), and the proportion of non-carnivorous individuals. The posterior distributions of the unbiased estimates were then used to test the differences between the three groups using the indices of effect existence (Bayesian P value) and significance—Std. Effect size (e.g.: Cohen’s h and Cohen’s d) according to Makowski etal. (2019a) and Cohen (1988). Following Makowski etal. (2019a), in this paper we define the significance of an effect as a combination of P value (existence) and effect size; therefore, we jointly evaluated both those metrics when discussing the results. Note that since covariate terms are only interpreted as having an effect or not, they have marginal relevance to the research question. This Bayesian approach was chosen over a Frequentist approach due to its ability to better deal with very unequal sample sizes across random effect groups, large variation in the amount of weighted sample, and by offering more freedom in the modelling of the response data distribution (McElreath, 2018). This approach was also chosen over direct testing of the effect of predictors on a response variable (which allows testing of only two differences between three groups) as it allows testing all three comparisons between pool types. Pool surface area was used as covariate in all the analyses. Due to the low amount of material in some samples, it was not possible to recover enough animals to build species accumulation curves and thus obtain an unbiased estimate of genera numbers with respect to the sampling effort (Deng etal., 2015). Instead, we account for potential sampling bias in determining the number of genera by using the number of individuals mounted on slides and taxonomically identified as a second covariate in the diversity analysis. For model estimates, the proportion of non-carnivorous individuals, tardigrade density and rotifer abundance were included as covariates, as both tardigrades and rotifers are known food items of carnivorous tardigrades (Kosztyła etal., 2016; Roszkowska et al., 2016; Bryndová et al., 2020), and thus could influence their proportion. Although the sampling in this study was performed on three different occasions, the sampling period was not considered in the analyses as only a few samples were collected in the first and second sampling events (11/56 and 2/56 respectively; SM.01) and in those two sampling trips only clusters A, C, D, E and K were sampled. Moreover, annual fluctuations in tardigrade abundances are known to occur mostly in aquatic or hygrophilous species (Schuster etal., 1977; Kathman & Nelson, 1987; Kinchin, 1996; Schuster & Greven, 2007). For more xeric species (like those of the genus Ramazzottius) annual population fluctuations have not been observed (Kinchin, 1989, 1996) and despite the analysed pools being defined as freshwater aquatic habitats, their tardigrade fauna is mostly composed of xerophilous taxa (Ramazzotti & Maucci, 1983). Geographic distance between samples was not included in the model as a random factor due Hydrobiologia 1 3 Vol:. (1234567890) to the inaccuracy of GPS measurements that is far higher than the distance between most of the samples (SM.01). Cluster identity was used as a random factor because clusters have different slopes, exposure, and/ or mineral composition that can likely influence the local tardigrade communities, and by having different geographic locations they can also partially account for the spatial distribution. Prevalence and abundance were estimated in the same model, whereas diversity was estimated in a separate one. Analysis workflow and R script are presented in SM.03. Analyses were performed with “R v. 4.0.2” software (R Core Team, 2020) and associated packages “bayestestR v. 0.3.0” (Makowski etal., 2019b), “patchwork v. 1.1.0” (Pedersen, 2020), “R2Jags v. 0.6.1” (Su & Yajima, 2015), “tidyverse v. 1.3.0” (Wickham et al., 2019) and “vegan v. 2.5.6” (Oksanen et al., 2020); BGLMMs were fitted with “JAGS v. 4.1.0” (Plummer, 2003). DNA sequencing and species delimitation To test if multiple species of the same genera were present and to provide reference sequences for future studies, we sequenced a fragment of the Cytochrome C Oxidase subunit I (COI) barcode gene. Detailed methods of DNA sequencing and species delimitation, as well as the resulting GenBank accession numbers and species delimitation results are provided in SM.04. Results In total, 14 tardigrade genera were found in the analysed samples, with densities from 0 to 889 tardigrades per gram of dry substrate (SM.01). Tardigrades were found in 78.6% of the samples (44/56). In three genera, multiple species were present, as revealed by COI barcoding (2 Ramazzottius spp., 2 Macrobiotus spp. and 2 Milnesium spp.; SM.04). The most common taxon, Ramazzottius, was found in 37.5% (21/56) of all examined rock pools with densities up to 860 individuals per gram of dry substrate (see SM.01). Mites and nematodes were absent or rare in most of the pools, whereas rotifers were always present and generally common (SM.01). Prevalence was significantly higher in Pans than in Intermediates and Holes (Fig. 3a, Table 1), but no difference between Intermediates and Holes was detected. In addition, no significant difference in tardigrade density was found across pool types (Fig.3b, Table1), and pool area had no effect on either prevalence or density (Fig.3c, Table1). Although diversity was much higher in Pans than in Intermediates and Holes (although P values are slightly above 0.05, the effect size was very large; Fig.3d, Table1), no difference in diversity was found between Intermediates and Holes (Fig.3d, Table1). Neither sampling effort (how many individuals were classified) nor pool area had a significant effect on diversity (Fig.3e, Table1). We found significant differences in the trophic structure across rock pool types: the proportion of noncarnivorous tardigrades was significantly higher in Holes than in Pans and Intermediates, but no difference was found between the latter two types (Fig.3f, Table1). Finally, the proportion of non-carnivorous tardigrades was positively associated with density and pool area, but not significantly affected by rotifer abundance (Fig.3g, Table1). Discussion Prevalence, abundance and diversity We show that tardigrades are abundant members of rock pool communities. In particular, we found prevalence and generic diversity are highest in the shallowest and thus driest pool category (Pans), when compared to the other two rock pool types (Intermediate and Holes). Interestingly, the tardigrade density was not significantly affected by pool type. This result contrasts with a previous study by Jocqué etal. (2007), which reported that high densities of tardigrades, rotifers and mites characterize shallower pools with short wet phases. Similar results (lower tardigrades density) were also obtained by Jönsson (2007) in mosses with experimentally increased hydration. It is interesting to note that the tardigrade community composition at the generic level matches the one previously recorded in the same region (Bertolani & Rebecchi, 1988–1996; Guidetti & Bertolani, 2001) for mosses and lichens (Echiniscus, Macrobiotus, Milnesium, Minibiotus, and Ramazzottius) and for freshwater habitats (Grevenius, Mixibius). Whereas genera more typical of leaf litter (Guidetti & Bertolani 2001; Hypsibius, Mesobiotus, Paramacrobiotus) Hydrobiologia 1 3 Vol.: (0123456789) were absent or extremely rare. However, this comparison is only based on a classification to genus level. Therefore, it is not possible to exclude or confirm that the species found in the rock pools are the same as those recorded in the other types of habitat, nor if mosses and lichens are the source of the taxa present in rock pools. The negative relationship between tardigrade taxonomic richness and water permanence is opposite to that reported in many other animal groups. Many studies from across the world have found a positive correlation between taxonomic richness in rock pools and water permanence (measured either as reduced desiccation frequency or as longer inundation periods) (King etal., 1996; Spencer etal., 1999; Therriault & Kolasa, 2001; Eitam etal., 2004a; Vanschoenwinkel etal., 2009; Kulkarni etal., 2019). However, in one case, no relationship was found (Eitam etal., 2004b). Longer hydroperiods can allow species with slower developmental rates to thrive and benefit from temporal resource partitioning, thus favouring an increase in taxonomic richness (King et al., 1996). Those studies, however, mostly focused on larger microinvertebrates (crustaceans and flatworms), macroinvertebrates (insects) and vertebrates (amphibians); meiofaunal taxa such as tardigrades, rotifers, 0.00 0.25 0.50 0.75 1.00 Pan Intermediate Hole % of samples with Tardigrades a - Tardigrades prevalence (%) 1 10 100 1000 Pan Intermediate Hole Individuals/g of dry sample b - Tardigrades density (T/gr) area -1.0 -0.5 0.0 0.5 1.0 c - Effect of covariate on prevalence and abundance 0 1 2 3 4 5 Pan Intermediate Hole Number of taxa d - Tardigrade diversity (# of taxa) area sampling -0.50-0.25 0.00 0.25 0.50 e - Effect of covariates on diversity 0.00 0.25 0.50 0.75 1.00 Pan Intermediate Hole Proportion of non-carnivorous f - Proportion of non-carnivorous area density rotifers -1 01 g - Effect of covariates on proportion of non-carnivorous Fig. 3 Distribution of posterior estimates from BGLMM. a Tardigrade prevalence (percentage of samples with tardigrades); b Tardigrade density (number of individuals per gram of dry sample); c Effect of covariate (pool area) on tardigrade prevalence and abundance; d Tardigrade diversity (number of taxa); e Effect of covariates (sampling effort and pool area) on tardigrade diversity; f Proportion of non-carnivorous tardigrade individuals; g Effect of covariates (rotifer abundance, tardigrade density and pool area) on the proportion of non-carnivorous tardigrade individuals. Median (filled circle) with 80% (thick line) and 95% (thin line) Credible Intervals. “.” P value < 0.1, “*” P value < 0.05, “**” P value < 0.01 Hydrobiologia 1 3 Vol:. (1234567890) and nematodes were rarely considered or classified below Phylum level. A possible explanation for the higher tardigrade prevalence and diversity in Pans is perhaps the inability of predators to survive the frequent desiccation events. Larger predatory animals that could prey on tardigrades (for example, flatworms and insects that are known predators in rock pools; Brendonck etal., 2002) can resist desiccation only as eggs (Cáceres, 1997; Watanabe, 2006) and might not be able to complete their life cycle and produce new resistant eggs during the short inundation periods. In contrast, tardigrades are not constrained to completing their life cycle between desiccation events, as they can undergo anhydrobiosis in all life stages (Kaczmarek etal., 2019). This positive effect of frequent desiccation through lower predation pressure could explain the observed pattern in tardigrade prevalence and diversity, as already observed in other habitats (Herrmann et al., 2012). It is important to note that evidence of predation on tardigrades by other animals is very scarce. However, according to Schmid-Araya & Schmid (2000), chironomids (Diptera: Chironomidae) from six out of the 17 genera for which data is available are known to feed on tardigrades. Chironomids have been commonly reported from freshwater rock pools (Brendonck etal., 2016), despite being unable to undergo anhydrobiosis (except Polypedilum vanderplanki Hinton, 1951 found only in the African continent; Cornette & Kikawada, 2011). Thus, it is realistic to hypothesize that predation pressure on tardigrades from chironomids is reduced in rock pools that desiccate very fast. To test the latter, one would need to develop a sampling protocol tailored to consistently and quantitatively extract not only tardigrades but also chironomids (and flatworms). Nevertheless, we have qualitative data (see SM.01) that chironomids (pool E9) and flatworms (Gieysztoria sp. aequales group; frozen sample of pool E19) are present in the studied system. Also, one Gieysztoria individual was found with the traces of four tardigrades in its gut (SM.05), providing the first documented case of predation on tardigrades by a flatworm. Table 1 Indices of existence and significance of the comparisons between the three different pool types and of the covariates Calculation of standard effect size of effect of Area on Prevalence was not possible due to the model structure; “.” P value < 0.1, “*” P value < 0.05, “**” P value < 0.01 Bayesian P value Std. effect size (e.s.) Prevalence Pan vs. Intermediate 0.0193* 0.7992 Pan vs. Hole 0.0073** 1.0061 Intermediate vs. Hole 0.5187 0.1959 Abundance Pan vs. Intermediate 0.1527 −0.2424 Pan vs. Hole 0.5107 0.3470 Intermediate vs. Hole 0.1647 0.3097 Area 0.6960 −0.0006 Diversity Pan vs. Intermediate 0.0560 1.7560 Pan vs. Hole 0.0627 2.1730 Intermediate vs. Hole 0.6527 0.4981 Area 0.3820 −0.0983 Sampling 0.2093 0.1075 Non-carnivorous proportion Pan vs. Intermediate 0.5187 −0.0063 Pan vs. Hole 0.0073** −0.6317 Intermediate vs. Hole 0.0193* −0.4131 Area 0.0487* 0.3086 Density < 0.0001** 0.3148 Rotifers 0.4387 0.1173