Short and dynamic: succession of invertebrate communities over a hydroperiod in ephemeral wetlands on arable land

ABSTRACT

The temporal dynamics of aquatic invertebrate communities are known to be influenced by the length of hydroperiods, but only temporary wetlands with relatively long hydroperiods have been well studied. By contrast, few studies have focused on ephemeral wetlands, primarily represented by extremely ephemeral rock pools. In Central Europe, many ephemeral wetlands develop spontaneously directly on arable land, often on the sites of former natural wetlands that have been drained and converted to agricultural land. Here, we focused on aquatic invertebrates in 10 ephemeral pools over periods of inundation to desiccation on arable land in southern Moravia. Despite relatively short hydroperiods of 5–11 weeks, we observed significant changes in community composition, including species replacement. Dynamics differed between macroinvertebrates and microcrustaceans and between different macroinvertebrate feeding and dispersal groups. Predation pressure increased over time and was highest during drying. Passive dispersers were most abundant during the middle phase, whereas the abundance of active dispersers increased throughout the hydroperiod. Because no significant effect of any environmental factor was detected on community change, we hypothesized that community dynamics were driven by differences in species traits and biotic interactions rather than by the changing environment. This study fills a knowledge gap on the temporal dynamics of aquatic invertebrate communities in temporary wetlands by investigating wetlands with short, but not extremely ephemeral, hydroperiods.

KEYWORDS:

• ephemeral pools
• macroinvertebrates
• microcrustaceans
• Pannonian ecoregion
• temporal dynamics
• temporary wetlands

Introduction

Temporary wetlands are intermittently inundated aquatic habitats that differ in the length and predictability of the hydroperiod (i.e., the duration, frequency, and timing of the aquatic phase; Calhoun et al. 2017). They are among the most endangered wetland types (Calhoun et al. 2017), especially in lowlands, because they have disappeared or deteriorated due to ecosystem conversion, mostly from intensive agriculture and river regulation (Brown 1998, Boix et al. 2016, Merta et al. 2016). The composition of the aquatic invertebrate community in temporarily inundated habitats changes during the hydroperiod. Some species are present throughout the aquatic phase while others are present only during a specific period (Williams 1983). Community dynamics are determined by environmental conditions, resource availability, and biotic interactions (Boix et al. 2016). During a hydroperiod, 3 phases of temporal dynamics can be identified based on community composition in most temporary wetlands (Sahuquillo and Miracle 2010, Boix et al. 2016). The filling phase is characterized by allogenic succession when resources are abundant after inundation and biotic interactions are less important. In the middle phase, biotic interactions have a significant effect on community composition, and succession is mainly autogenic. During the drying phase, environmental conditions largely change and fluctuate, and therefore succession is again rather allogenic and mainly influenced by environmental factors (Boix et al. 2016). In temporary pools with relatively short hydroperiods, only 2 allogenic phases are distinguished, with the exception of the autogenic middle phase (Boix et al. 2004, Jocque et al. 2007, Vanschoenwinkel et al. 2010). By contrast, in pools with extremely short hydroperiods, no phases can be distinguished based on species composition changes because time is insufficient for species replacement through colonization, biotic interaction, or changing environmental conditions (Jocque et al. 2007).

The temporal dynamics of communities are related to the exchange of species with different life histories (Williams 1983, Moorhead et al. 1998) and feeding strategies (Williams 1983, Moorhead et al. 1998, Culioli et al. 2006) and the increasing biomass spectrum (Boix et al. 2004, Brucet et al. 2005). Communities in the initial stage of the hydroperiod are usually dominated by passive dispersers (Moorhead et al. 1998) that survive the dry period in the sediment in the various forms of resting stages (i.e., eggs, cysts, diapausing immatures, or adults; Williams 1983). Initial stages are often characterized by higher abundances of filter-feeders and detritivores and lower predation pressure (Moorhead et al. 1998, Culioli et al. 2006). Later stages are generally dominated by actively dispersing invertebrates (Moorhead et al. 1998), and predation pressure increases as predator abundance and body size increase (Kenk 1949, Moorhead et al. 1998, Culioli et al. 2006).

The temporal dynamics of aquatic invertebrates have been studied in temporary wetlands in North America (Kenk 1949, Williams 1983, Moorhead et al. 1998, Hall et al. 2004), in the Mediterranean region in Europe (Bazzanti et al. 1996, Boix et al. 2004, Culioli et al. 2006, Gilbert et al. 2021), and in West Africa (Lahr et al. 1999), as well as in ephemeral pools with extremely short hydroperiods in southern Africa (Meintjes 1996, Jocque et al. 2007). In Central Europe, the few studies focus on soda pans in Hungary (Lengyel et al. 2019, Cozma et al. 2020).

In Central Europe, natural floodplain habitats have been heavily modified and converted to agricultural landscapes, resulting in the development of many temporary wetlands directly on arable land (Lukács et al. 2013). These wetlands are the hotspots of biodiversity in a homogeneous agricultural landscape and host a high diversity of plants (Lukács et al. 2013, Němec et al. 2014), amphibians (Němec and Sychra 2017), birds (Sychra et al. 2021), and some terrestrial (Němec et al. 2012, Heneberg et al. 2018) and aquatic groups of invertebrates (Eder and Hödl 2002, Merta et al. 2016, Němec and Sychra 2017). Ephemeral wetlands on arable land emerge de novo in terrain depressions after snow melting, heavy rains, or floods (Lukács et al. 2013, Schernhammer et al. 2020). They typically develop primarily at sites of former natural wetlands, swamps, marshes, salt marshes, wet meadows and pastures, oxbows, lakes, and stream channels (Lukács et al. 2013, Němec et al. 2014, Merta et al. 2016, Heneberg et al. 2018) and are characterized by early successional stages maintained by regular plowing (Němec et al. 2012, Merta et al. 2016). Their size varies from a few square meters to several hectares, depending on the current hydrological situation, weather, bedrock, and landscape relief (Němec et al. 2012, 2014, Lukács et al. 2013, Merta et al. 2016). Ephemeral wetlands are intermittently flooded (i.e., standing water is periodically present but without a seasonal pattern; van der Valk 2012), mainly influenced by weather (precipitation and evaporation), making their occurrence and the length of hydroperiods highly unpredictable. Moreover, they are patchily distributed (Schernhammer et al. 2020) and because of plowing are only visible in their aquatic phase within croplands, making their detection and maintenance difficult (Merta et al. 2016, Calhoun et al. 2017).

Although the length of hydroperiods can significantly affect temporal dynamics (Boix et al. 2004, Jocque et al. 2007), few studies focus on pools with short hydroperiods. The hydroperiods of ephemeral wetlands on arable land usually vary from 3 weeks to 3 months (Lukács et al. 2013, Merta et al. 2016), and unknown is whether communities can change temporally during such a short aquatic phase.

Here, we focused on the temporal dynamics of aquatic invertebrates in 10 ephemeral pools on arable land in southern Moravia (northern Pannonia). Our aims were to investigate (1) the variability of environmental conditions over the hydroperiod; (2) the dynamics of aquatic invertebrate assemblages, particularly in relation to their trophic structure and dispersal strategies; and (3) the main drivers of short-term dynamics in ephemeral wetlands on arable land. We hypothesized that (a) community dynamics are primarily determined by the length of the hydroperiod rather than environmental characteristics because time since inundation significantly affects colonization success and species replacement (Jocque et al. 2007); (b) the dynamics of active and passive colonizers differ based on their recolonization mechanisms (Boix et al. 2016); and (c) the relative proportion of predators in the community increases during the hydroperiod because of colonization by flying insects.

Material and methods

Sampling sites

The sampling sites were located in southern Moravia (southeastern part of the Czech Republic; Fig. 1). Each of the 10 study pools was developed on arable land at an elevation of 150–220 m a.s.l. The distance between individual pools ranged between 1.5 and 55 km. All pools were re-flooded in spring 2016, with the initial surface area ranging from 0.21 to 25.90 ha and initial water depth 10–40 cm (Supplemental Table S1). The length of their hydroperiod was 5–11 weeks (Supplemental Table S1).

Figure 1. The 10 temporary study pools in southern Moravia, Czech Republic.

Sampling and sample processing

The first sampling was conducted ∼2 weeks after filling and was repeated at 2–3 week intervals until the pools dried out. All 10 pools were filled and therefore sampled during the first (22–31 Mar) and second (6–12 Apr) periods. Only 9 pools were sampled in the third period (26 Apr to 5 May), and 4 pools were sampled in the fourth period (17 May). The remaining sites had dried up before the third or fourth sampling. The variable “time” was calculated for each sample as the number of days since the first sample was collected (22 Mar).

Macroinvertebrates were sampled semiquantitatively. Three people collected macroinvertebrates for 15 min using a metal strainer (20 cm diameter, 0.8 mm mesh size). Samples were pooled and fixed in 75% ethanol. Adult coleopterans and heteropterans were fixed in ethyl acetate. Microcrustaceans were quantitatively sampled using a relatively coarser mesh (250 µm) hand net to reduce clogging of the highly turbid water in the pools. Using ten 1 m long trawls covering different microhabitats within each pool, ∼177 L of pool water sample was filtered through the net and preserved in 4% formalin.

Physical and chemical properties were measured at each sampling time. Water conductivity and pH were measured by a portable Multi Meter Hach-Lange HQ40D (Hach, Loveland, CO, USA). Turbidity was estimated using an ordinal scale: 0 (clear, transparent), 1 (slight turbidity visible), 2 (high turbidity, but the bottom of the pool still visible), and 3 (highly turbid water, transparency up to 10 cm). Vascular plant and macroscopic algal covers were estimated for the entire pool on an ordinal scale from 0 (no plants/algae) to 3 (high plant/algae cover). Pool surface area was measured based on pool photographs and aerial photographs using the measurement tool in www.mapy.cz.

Aquatic invertebrates were identified to the lowest possible taxonomic level. Macroinvertebrates were categorized according to their feeding type (predator, gatherer, shredder, filter-feeder, or grazer) and dispersal mode (passive or active). Feeding types were based on the https://www.freshwaterecology.info/ (Schmidt-Kloiber and Hering 2015) and http://hydro.chmi.cz/isarrow/ (IS ARROW 2009) databases, and dispersal types were obtained from the DISPERSE database (Sarremejane et al. 2020). Microcrustacean abundance was expressed as the number of individuals per liter.

Statistical analyses

All analyses and graphics were performed in R software 4.1.0 (R Core Team 2021) using the vegan (Oksanen et al. 2020), geepack (Hojsgaard et al. 2006), beanplot (Kampstra 2008), and indicspecies (De Caceres and Legendre 2009) packages. Principal component analysis (PCA; rda function from vegan package) was used to describe the variation in environmental variables (water turbidity, cover of vascular plants, cover of algae, pH, water conductivity, water depth, surface area of the pools) in the pools and to examine possible relationships among environmental variables. All variables were used because no strong collinearity was detected by Spearman rank correlation coefficient (ρ < 0.7). Spearman correlation was also used to identify association between environmental variables and time (the number of days since first sampling).

Macroinvertebrate and microcrustacean data were analysed separately because different sampling methods had to be applied for macroinvertebrates (semiquantitative) and microcrustaceans (quantitative). Principal coordinate analysis using Bray–Curtis distances (PCoA; cmdscale function) was used to explore the variability in aquatic invertebrate assemblages. Significantly fitted environmental variables were passively projected into the ordination diagrams. Significance was assessed using a permutation test with 9999 runs. Distance-based redundancy analysis using Bray–Curtis distances (db-RDA; capscale function from vegan package) was used to evaluate the influence of environmental variables and time on the species composition of the aquatic invertebrates inhabiting the pools. A balanced design was required to test the significance of the explanatory variables in this analysis, so a reduced dataset was used. Because only 4 of the 10 sites remained filled through the fourth sampling period, we only analyzed the first 3 periods. One site that dried out before the third sampling period was removed, leaving 9 sites in the dataset, all sampled in the first 3 periods. Significance was tested using a permutation test with 9999 runs, constrained by the split-plot design (sampling sites set as blocks; permutation type set to series) to control for possible bias due to pseudo-replications as each site was sampled repeatedly over time. In addition, the sampling site was included in the model as a covariate to remove the variability explained by the association of each sample with the sampling site. The final models were built by stepwise selection (ordiR2step function from vegan package).

Generalized estimating equations (GEE; geeglm function from geepack package) with a Poisson error structure (GEE-p) were used to examine the temporal dynamics in abundance, species richness, and cumulative species richness. The same model with a gamma error structure (GEE-g) was used to examine the temporal dynamics of beta diversity. Beta diversity was calculated using Sorensen dissimilarity between subsequent samples to estimate the change in community composition over time. The association of each sample with a sampling site was set as a random-effect (grouping) variable to control for possible bias from pseudoreplication. First, a quadratic model was built for each dependent variable. The models were then simplified by backward selection to obtain the minimal adequate model. The significance of the model components was evaluated at a 95% confidence level. GEE-p were also used to separately examine the temporal dynamics of the abundances of 3 groups of microcrustaceans (cladocerans, ostracods, and copepods), feeding types, and dispersal modes. The explained variation of GEE-p and GEE-g was approximated using the marginal R² value.

Partitioning of beta diversity into species turnover (Simpson dissimilarity between subsequent samples) and nestedness (resultant fraction of Sorensen dissimilarity between subsequent samples; i.e., beta diversity without species turnover) was visualised by beanplots (beanplot function from beanplot package).

Indicator species analysis (multipatt function from the indicspecies package) was used to identify species characteristic of each of the 3 successional phases in the studied pools: filling phase (first samples); middle phase (second samples from the sampling sites sampled during the 3 periods, or the second and third samples from sampling sites sampled during the 4 sampling periods); and drying phase (last samples). Only the most frequent (found in >5 samples) and abundant (>50 and 100 individuals for macroinvertebrates and microcrustaceans, respectively) were selected for this analysis.

Results

We observed 2 main gradients of environmental conditions in the studied pools (Fig. 2). The first was defined by increasing vascular plant and macroscopic algae cover, increasing water conductivity, and decreasing turbidity. The second was represented by decreasing surface area and pH (Fig. 2, Table 1). In general, variability in environmental conditions between sites was greater than within sites (Fig. 2b). Only surface area notably varied, even within sites, as indicated by the significant correlation with time (Table 2).

Figure 2. Variability and relationships among measured environmental conditions of all collected samples explored by PCA. Dots refer to sites; samples in the convex hulls were collected at the same site. The first 2 axes of PCA explained 51.65% of the total variance, with axes 1 and 2 explaining 31.21% and 20.45%, respectively.

Table 1. Multiple regression of each explanatory variable and site score on the first 2 PCA axes of environmental variables. Regression coefficients, the fit of each factor into the ordination space, variation in the factor explained by site scores on the first 2 PCA axes in multiple linear regression (R²), and significance (p) of the result based on 9999 permutations are shown. Nonsignificant variable is in italics.

Variable	PC1	PC2	R^2 (%)	p
Turbidity	−0.900	0.436	45.71	<0.001
Vascular plants	0.889	0.457	84.02	<0.001
Algae	0.989	−0.148	68.00	<0.001
pH	0.515	−0.857	43.20	<0.001
Conductivity	1.000	0.001	34.59	0.004
Water depth	−0.086	−0.996	7.79	0.288
Surface	−0.174	−0.985	78.72	<0.001

Table 2. Spearman correlation values and their significance between time and environmental variables. Nonsignificant relationships are in italics; p-values were adjusted by Holm’s correction.

Variable	rho	p
Surface area	−0.514	0.016
pH	−0.310	0.560
Algae	−0.138	1.000
Water depth	−0.029	1.000
Conductivity	−0.026	1.000
Vascular plants	0.020	1.000
Turbidity	−0.001	1.000

We recorded 100 taxa of macroinvertebrates and 25 taxa of microcrustaceans in the pools (Supplemental Table S2). Compositional variance in both macroinvertebrate (Fig. 3) and microcrustacean (Fig. 4) assemblages was associated with time (Fig. 3–4). In addition, variance in the macroinvertebrate assemblages was also affected by water depth and conductivity (Fig. 3), and variance in the microcrustacean assemblages was affected by algae and surface area (Fig. 4). However, based on the results of db-RDA analysis, only time explained a significant portion of the total variability (i.e., 8.6% and 14.5% in macroinvertebrate and microcrustacean assemblages, respectively; Table 3).

Figure 3. Compositional variation among 33 samples of aquatic macroinvertebrate assemblages explored by PCoA and sorted by sampling period. Significant variables were passively projected into the ordination diagram. Abundances of each species occurring at >3 sites were fitted using GAMs (p < 0.05) into the ordination space, and only species with significant fit to the first 2 PCoA axes are shown. First and second axes explained 13.0% and 11.6% of the total variability, respectively. Species abbreviations: Aed.vex = Aedes vexans; AgaJ = Agabus sp. juv.; Ber.sig = Berosus signaticollis; Chi = Chironomus spp.; ColJ = Colymbetes sp. juv.; Eub.gru = Eubranchipus grubii; Hel.min = Helophorus gr. minutus; HydJ = Hydroporus sp. juv.; Hyd.lug = Hydrobaenus lugubris Gr.; Lep.apu = Lepidurus apus; Sig.lat. = Sigara lateralis; Sig.str = Sigara striata; Tri.can = Triops cancriformis; Tub.tub = Tubifex tubifex.

Figure 4. Compositional variation among 33 samples of microcrustaceans explored by PCoA and sorted by sampling period. Significant variables were passively projected into the ordination diagram. Abundances of each species occurring at >3 sites were fitted using GAMs (p < 0.05) into the ordination space, and only species with significant fit to the first 2 PCoA axes are shown. First and second axes explained 23.0% and 15.3% of the total variability, respectively. Species abbreviations: CalK = calanoid copepodites; Chy.sph = Chydorus sphaericus; CycK = cyclopoid copepodites; Dap.cur = Daphnia curvirostris; DapJ = Daphnia juvenile; DapM = Daphnia males; Dap.mag = Daphnia magna; Met.min = Metacyclops minutus; Moi.bra = Moina brachiata s. l.; Sim.vet = Simocephalus vetulus; Ton.lut = Tonnacypris lutaria.

Table 3. Results of distance-based redundancy analysis (db-RDA) for macroinvertebrates and microcrustacean assemblages: gross effects are total effects of variables examined for each variable in a separate model; pure effects are marginal effects of variables added into the final model build by stepwise selection; Adj R² = adjusted coefficient of determination; p = p-value. Note that both gross and pure effects of all other explanatory variables used were insignificant (p > 0.05) (results in Supplemental Table S3).

	Gross effects		Pure effects
	Adj R^2 (%)	p	Adj R^2 (%)	p	Order in model
Macroinvertebrates
Time	8.6	<0.001	8.6	<0.001	1
Microcrustaceans
Time	14.5	<<0.001	14.5	<<0.001	1
Surface area	4.7	<<0.001	—	—	—

We observed different temporal dynamics between the macroinvertebrate and microcrustacean assemblages. Macroinvertebrate species richness did not change significantly over time while microcrustacean species richness increased over the hydroperiod; however, cumulative species richness increased significantly in both macroinvertebrates and microcrustaceans (Table 4). Changes in macroinvertebrate assemblage composition did not differ between subsequent sampling periods, whereas they decreased for microcrustaceans over the hydroperiod. Both macroinvertebrate and microcrustacean abundance increased monotonically over the hydroperiod, reaching the highest average values in the final sampling periods. While ostracod abundance did not change significantly over time, copepod abundance reached the highest values in the second sampling period, and cladoceran abundance was highest in the third and fourth sampling periods (Table 4). For macroinvertebrates, the abundance of predators, gatherers, and shredders increased. Passively dispersing macroinvertebrates were most abundant during the second sampling period, whereas the abundance of active dispersers increased linearly over the hydroperiod and was highest during the final sampling periods (Table 4).

Table 4. Temporal dynamics of macroinvertebrate and microcrustacean abundance, species richness (diversity), cumulative species richness (cum. sp. r.), and beta diversity; response curve types: Mon(0) = constant function; Mon(-) = monotonically decreasing; Mon(+) = monotonically increasing; Uni(0) = unimodal with maximum around the center; and Uni(+) = unimodal with maximum skewed to the higher values; R² is an approximation of coefficient of determination; p-values of coefficients of GEE: p₀ = intercept, p₁ = linear coefficient, and p₂ = quadratic coefficient. Nonsignificant p-values are in italics (for more detailed results of GEE see Supplemental Fig. S1, and Supplemental Table S4).

		Response	R^2 (%)	p0	p1	p2	Types of response
Macroinvertebrates	diversity	Mon(0)	0.0	<<0.001	0.175	0.042
Filter-feeders	abundance	Mon(0)	0.0	<<0.001	0.139	0.068
Grazers	abundance	Mon(0)	0.0	<<0.001	0.102	0.470
Macroinvertebrates	β diversity	Mon(0)	0.0	<<0.001	0.130	0.390
Ostracoda	abundance	Mon(0)	0.0	<<0.001	0.430	0.570
Microcrustaceans	β diversity	Mon(-)	29.2	<<0.001	0.008	0.833
Macroinvertebrates	abundance	Mon(+)	51.1	<<0.001	<<0.001	0.074
Microcrustaceans	abundance	Mon(+)	14.2	<<0.001	0.045	0.410
Predators	abundance	Mon(+)	30.9	<<0.001	<0.001	0.240
Gatherers	abundance	Mon(+)	32.9	<<0.001	<<0.001	0.068
Active dispersers	abundance	Mon(+)	51.3	<<0.001	<<0.001	0.370
Copepoda	abundance	Uni(0)	17.8	<<0.001	0.245	0.036
Passive dispersers	abundance	Uni(0)	5.0	<<0.001	0.997	<0.001
Microcrustaceans	diversity	Uni(+)	34.5	<<0.001	0.002	<0.001
Macroinvertebrates	cum. sp. r.	Uni(+)	40.7	<<0.001	<<0.001	0.002
Microcrustaceans	cum. sp. r.	Uni(+)	72.0	<<0.001	<<0.001	<<0.001
Cladocera	abundance	Uni(+)	25.6	<<0.001	<0.001	0.029
Shredders	abundance	Uni(+)	7.6	<<0.001	0.038	0.008

Species turnover was an important component of beta diversity in both macroinvertebrate and microcrustacean data in most of the studied pools. The relative contribution of species turnover in comparison to nestedness increased slightly for the macroinvertebrates over the hydroperiod. No clear trend in the relative importance of beta diversity components was observed for microcrustaceans (Fig. 5). This finding was also indicated by the composition of species identified as characteristic of the sampling periods (Fig. 6), almost exclusively associated with drying phase (i.e., final sampling periods).

Figure 5. Partitioning of beta diversity between subsequent samples into species turnover and nestedness components for (a) macroinvertebrates and (b) microcrustaceans. White lines represent individual values; black lines show averages.

Figure 6. Temporal dynamics in most common taxa of aquatic invertebrates (found in >5 samples and total abundance >50 individuals for macroinvertebrates or >100 individuals for microcrustaceans). Size of the circle represents relative abundance (as a % of total abundance indicated in the last column). Taxa are increasingly ordered by average time of occurrence. Species characteristic for hydroperiod phases are in bold: M = middle phase; D = drying phase. The indicator value is given after each characteristic species name, and significance is indicated by asterisks; *<0.05; **<0.01; ***<0.001. Sampling periods: 1 = filling phase; 2 = middle phase; 3 = middle or drying phase based on the hydroperiod length; 4 = drying phase.

Discussion

Temporal dynamics during hydroperiod

In the studied pools we clearly observed a significant change in the composition of aquatic invertebrates over time. The hydroperiod length in these pools ranged from 5 to 11 weeks, considerably shorter than hydroperiods in pools commonly used to study temporal dynamics (Kenk 1949, Bazzanti et al. 1996, Culioli et al. 2006). We found a significant increase in cumulative species richness for macroinvertebrates and microcrustaceans, but the increase in species richness was slower or not significant, indicating that species replacement occurred in the study pools, a finding also supported by beta diversity partitioning.

We found that the hydroperiod was too short for environmental conditions to change over time, except for the decreasing surface area of the pools. In general, conditions varied notably more among the sites than over time. Thus, we suspect that community dynamics were not primarily driven by changing environmental conditions, but rather by other processes that promote short-term dynamics in assemblages. Such processes can be associated with differences in species survival strategies and colonization abilities, temporal resource partitioning, predator avoidance, facilitation, and chance in colonization, hatching, and survival (Morin 1999, Boix et al. 2016).

During the filling phase, immature stages of copepods and large branchiopods were abundant in the study pools. However, no species characteristic of the filling phase was identified, most likely because of the large variability in species composition among pools shortly after inundation. During the middle phase, many active colonizers, including their juveniles, were more abundant, but only a single beetle species was identified as a species characteristic of this phase. During the drying phase, cladocerans and both adult and juvenile active colonizers dominated the assemblages. Most taxa (5) were characteristic of this phase, suggesting that similarity of assemblages in temporary wetlands increases over time (Moorhead et al. 1998, Ganguly and Smock 2010, Vanschoenwinkel et al. 2010).

We observed that the abundance of predators increased significantly over time, as in other temporary wetlands (Kenk 1949, Wiggins et al. 1980, Moorhead et al. 1998, Boix et al. 2004, Culioli et al. 2006). Apart from passively dispersing notostracans, we observed that predator densities were mainly related to the establishment of flying insects, which was also reported by Boix et al. (2016) and Epele and Miserendino (2016). Similarly, the increase in shredder and gatherer abundances may be due to flying insect colonization (e.g., some coleopterans and heteropterans, respectively). As total macroinvertebrate and microcrustacean abundance increased over time, the increase in predator abundance also coincided with the increase in prey density. Unlike cladocerans, which increased in abundance over time, the abundance of macroinvertebrate filter-feeders did not change significantly. By contrast, Moorhead et al. (1998) observed a significant decrease in abundance of filter-feeders in Texas playas, primarily due to a decline in abundance of large branchiopods. In the studied pools, large branchiopods were replaced by filter-feeding culicid larvae and some partially filter-feeding chironomid larvae in later periods.

Succession changes related to colonization dynamics

Passively and actively dispersing invertebrates employ different strategies to cope with droughts (Wiggins et al. 1980), which may lead to different temporal dynamics (Vanschoenwinkel et al. 2010). In our study pools, the total abundance of passively dispersing macroinvertebrates was highest during the middle phase of the hydroperiod, likely increasing their chance of producing resting stages before the pools dried out. For example, the notostracan Lepidurus apus, which has a slow growth and maturation rate (Kuller and Gasith 1996, Boven et al. 2008), appeared shortly after the pools were flooded. By comparison, the notostracan Triops cancriformis, which matures within a month after hatching (Petrov and Cvetković 1996, Brtek 2005), appeared in the middle phase.

While copepods reached maximum abundance in the first half of the hydroperiod, cladoceran abundance did not peak until the second half of the hydroperiod, probably because of their preference for higher temperatures. For example, the 2 most common species, Daphnia magna and Moina brachiata, prefer temperatures >22 °C (Hudec 2010). In pools with high abundances of the anostracan Eubranchipus grubii, the later appearance of cladocerans may also be a strategy to reduce competition with these early occurring anostracans, which are highly effective filter-feeders (Celewicz et al. 2018).

By contrast, the density of active dispersers (aquatic insects) increased monotonically during the hydroperiod. A similar pattern was observed in both temporary pools with long hydroperiods (Moorhead et al. 1998, Lahr et al. 1999) and ephemeral pools with short hydroperiods (Meintjes 1996). Colonization of ephemeral residents takes more time than hatching of permanent residents (Jocque et al. 2007). In addition, many are predators, so they may preferentially colonize the habitat later, when prey populations are established (Moorhead et al. 1998). For example, cladocerans and chironomid larvae, which were abundant during the late phases, are suitable prey for Hydroglyphus geminus (Kehl and Dettner 2003), a species characteristic of the drying phase in the study pools.

During the drying phase, juveniles of some active colonizers, such as Sigara sp., Agabus sp., and Hydroporus sp. reached the highest abundance. Some may not have been able to mature and emigrate before the pools dried out, which is consistent with the observations of Wiggins et al. (1980), who noted that ephemeral inhabitants are little affected by seasonal cues, so that the offspring of late colonizers are commonly present during drying. However, a short hydroperiod may increase the probability of their unsuccessful reproduction.

Different colonization strategies of macroinvertebrates are also likely the reason for the constant rate at which their community changes (i.e., no change in Sorensen dissimilarity between subsequent samples). Shortly after inundation, their assemblages consist mainly of passive dispersers. Later, the relative importance of flying colonists in community change should increase, offsetting the decreasing rate of community change by passive dispersers. By contrast, a significant difference in the rate of community change in microcrustaceans was observed among subsequent samples, where it was highest after inundation and decreased monotonically over time.

Characteristics of ephemeral pools on arable land

We found that the studied ephemeral pools on arable land have some similar characteristics to those of highly ephemeral pools with extremely short hydroperiods. These systems are characterized by higher interpool variability (Meintjes 1996) and relatively high density and species richness just before drying (Boix et al. 2004, Jocque et al. 2007). Nevertheless, some of their characteristics are more similar to temporary wetlands with longer hydroperiod, most notably the presence of all 4 types of survival strategies according to Wiggins et al. (1980); significant changes in assemblage composition, including species replacement (Bazzanti et al. 1996, Lahr et al. 1999, Sahuquillo and Miracle 2010); and increasing predation pressure over time (Kenk 1949, Culioli et al. 2006). Finally, low abundance of grazers is one of the unique characteristics of ephemeral pools on arable land.

The main drivers of the temporal dynamics of aquatic invertebrates in the ephemeral pools we studied were differences in their survival and colonization strategies. Facilitation (establishment of prey populations followed by increasing abundance of predators) and temporal resource partitioning (e.g., replacement of filter-feeders) may have additional effects. Regarding environmental conditions, we only observed a significant effect of decreasing pool size on microcrustacean assemblages, which may lead to increasing densities in later periods, for example. Because no other measured environmental variable was significantly correlated with time, we describe successional processes independent from environmental variability. However, given the high variability of environmental conditions in the study pools, the results described in this study seem to reflect general processes and patterns of temporal dynamics in aquatic invertebrate communities in ephemeral pools on arable land.

Our results show that ephemeral wetlands on arable land are specific types of temporary wetlands in terms of various habitat characteristics and the temporal dynamics of their invertebrate assemblages. They are a useful model system for studying temporal dynamics because they allow us to observe changing assemblages, unlike pools with extremely short hydroperiods. Our results indicate that the observed processes are more independent of the changing environment compared to temporary pools with long hydroperiods.

Declaration of interest statement

The authors have no competing interests to declare that are relevant to the content of this article.

Acknowledgements

We are very grateful to Dorota Gvozdjáková, Jana Schenková, Alexandra Černá, Martina Poláková, Vendula Polášková, and Jana Petruželová for their help identifying invertebrate samples.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was primarily supported by the Czech Science Foundation (Grant Number P505/20-17305S); PhD students were also supported by the institutional support of Masaryk University (MUNI/A/1488/2021).