We performed a field experiment to assess the role of conceptually different measures of shredder diversity in the process of leaf litter breakdown in eight fishless ponds located in Slovakia, Central Europe ( Figure 1 ). The ponds are mesotrophic to eutrophic and cover a relatively broad altitudinal gradient from lowland to mountain areas (117-870 m). The surface area ranges from 18 to 2,315 m², and the maximum depth ranges from 0.3 to 1.5 m. For further details on the study site characteristics, see Table 1 .

Senescent leaves of black alder, one of the most frequently used litter species in breakdown studies and species frequently found in the riparian zones of Central European ponds, were collected just after abscission in November 2016. The leaves were air-dried in the laboratory (~25°C) until they reached a constant dry mass (~7 days). Portions of 3.0 g (± 0.09 g) of dry leaves were enclosed in litter bags (20 × 20 cm, 5.0 mm mesh size), which were used to measure the breakdown rate (Benfield et al., 2017). In total, 60 litter bags were prepared. The litter bags were placed randomly on the bottom of each pond at the end of June 2017. To prevent their loss, the bags were secured with nylon cords attached to riparian vegetation. Originally, we intended to perform the study in ten ponds. Because two sites experienced complete drying shortly before the start of the experiment, we increased the number of litter bags in two ponds (Dvory nad Žitavou, Kunerád1) to utilize all prepared bags. Data loggers (Minilog II-T; Vemco, Bedford, Nova Scotia, Canada) recording water temperature at hourly intervals were placed in each pond near the litter bags. One or two bags (in ponds with doubled numbers) were randomly selected and retrieved over six sampling dates. Since dry leaves are easily broken during transportation and handling (Benfield et al., 2017) and initial leaching often represents a substantial loss of leaf mass (Gessner et al., 1999), we accounted for these losses by harvesting the first bags 24 hours after installation. These bags were used to correct the initial weights of the leaf litter for mass loss unrelated to the biological decomposition processes. Subsequent collection of the bags was conducted after approximately 25, 70, 110, 145 and 170 days of incubation. The litter bags were placed to plastic zip-lock bags and transported in the fridge to the laboratory for processing. The leaves were removed from the bags, gently rinsed under tap water, oven-dried at 45°C for 24 hours and weighed to the nearest 0.01 mg. To account for possible bias caused by mineral deposits on the leaves, we combusted the weighed leaves in a muffle furnace (550°C for 1 hour) and estimated the ash-free dry mass (AFDM). We corrected the initial weights of leaf litter in the bags using the AFDM estimates of the bags retrieved after 24 hours.

The breakdown rates in each pond were estimated using the breakdown coefficient (k) based on the exponential decay model (Petersen and Cummins, 1974). The coefficients were quantified as the mass loss per degree-day (the cumulative sum of daily mean water temperatures) to correct for temperature differences among sites. We also report the breakdown coefficients calculated as mass loss per day to facilitate comparisons among studies. The breakdown rate coefficients were calculated as slopes of the log-linear models regressing the proportion of AFDM remaining (%) on the degree-days and days (Benfield et al., 2017).

Each pond was sampled for invertebrates in summer using a standardized semiquantitative PLOCH method (Oertli et al., 2005). As the first step of the PLOCH sampling, mesohabitats were identified, and their coverage was estimated. Five to eight samples were taken at each site depending on the pond area and mesohabitat diversity. The samples were distributed among the mesohabitats in proportion to their relative coverage, with a minimum of one sample per mesohabitat. The samples were collected by sweeping and kicking the mesohabitats using a net with a rectangular frame of size 15 × 11 cm and 0.5 mm mesh size for a fixed time of 30 seconds. The samples were subsequently pooled into one composite sample per site. All collected material was preserved in 4% formaldehyde in the field. Invertebrates were hand-sorted in the laboratory, preserved in 70% alcohol and identified to the lowest possible taxonomic level. The taxa were classified into functional feeding groups following Šporka (2003) and Schmidt-Kloiber and Hering (2015). Only the data on shredding species were used in the subsequent analyses.

The abundance, measured as the average number of shredders per sample, was determined for each pond. The total number of shredder species per pond (species richness) was used as a measure of taxonomic diversity. Functional diversity was calculated as the total branch length of a dendrogram based on functional traits (Petchey and Gaston, 2006). The functional dendrogram was constructed using the Euclidean distance matrix of z score standardized functional traits and the unweighted pair group method with arithmetic means (UPGMA) ( Figure S2A ). Since the selection of functionally important traits is a critical step in the calculation of functional diversity (Petchey and Gaston, 2006), we used maximum body size and voltinism ( Table S1 ), as these features are essential in the metabolism, energy allocation and food processing of invertebrates (Gillooly et al., 2001; Woodward et al., 2005; Menezes et al., 2010). Information on species body size was obtained from Freude et al. (1971); Freude et al. (1979); Rozkošný (1980); İncekara et al. (2005); Straka and Sychra (2007); Skuhrovec et al. (2011) and Kriska (2013), and data on voltinism were based on Šporka (2003); Grabowski et al. (2007) and Janz (2017).

We used ordinary least squares models (OLS) to test the influence of pond environmental characteristics ( Table 1 ) and shredder communities (abundance, species richness, functional diversity) on the breakdown rates of black alder leaf litter. The breakdown rates, pond environmental variables (area, depth, conductivity), and shredder characteristics were log-transformed to satisfy the assumption of normality. However, some of the OLS models showed heteroscedastic residuals which may lead to biased standard error estimates and erroneous conclusions of statistical inference (Zar, 2014). Therefore, we used generalized least squares (GLS) models that are designed to provide unbiased estimates when the assumption of constant error variance is violated (Pinheiro and Bates, 2006). We employed GLS incorporating the varying degrees of heteroscedasticity by assigning different weights to individual observations based on an exponential variance structure. GLS considerably improved the overall model performance, as evidenced by diagnostic plots of residuals. We fitted separate models for each predictor and tested their significance using Wald F tests. The relative importance of the predictors was assessed through (pseudo)determination coefficients (R²) (Nagelkerke, 1991) and standardized (β) coefficients (Schielzeth, 2010).

The analyses were performed in R (R Core Team, 2019) using the packages ggplot2 (Wickham, 2016), nlme (Pinheiro et al., 2019), rcompanion (Mangiafico, 2019) and vegan (Oksanen et al., 2022).

We specifically showed that the alder leaf litter decomposed faster in communities with broader spectra of shredder body size than in communities consisting of similar-sized species. The plausible mechanisms behind the observed positive functional diversity effect on decomposition include resource partitioning and facilitation (Loreau, 1998; Cardinale et al., 2002). Animals of different sizes may exhibit distinct feeding strategies, and such resource partitioning within the same trophic guild can reduce competition while enhancing the overall ecosystem process rate (Kleynhans et al., 2011). In freshwater detritivores, small chironomids can mine inside the leaves, mid-sized isopods and amphipods – such as Asellus and Echinogammarus – preferentially consume the more palatable parts of the leaves, and they may also scrub the leaf surfaces. On the other hand, large shredders such as Sericostoma and Gammarus are capable of biting through the leaf material, including the tougher parts such as veins (Graça et al., 1993; Friberg and Jacobsen, 1994; Wantzen and Wagner, 2006; Tonin et al., 2018). Apart from field observations, Tonin et al. (2018) conducted a controlled experiment to show how this feeding niche complementarity could enhance litter decomposition rates in aquatic ecosystems. These findings are consistent with theoretical predictions that complementarity among species has a systematic effect on ecosystem processes (Loreau, 1998).

In addition, facilitation between species is also thought to be an important mechanism by which biodiversity affects the rates of resource use that govern ecosystem processes (Cardinale et al., 2002). For example, the feeding activity of larger shredders that break down leaves into smaller fragments may facilitate litter consumption by smaller species, leading to elevated rates of leaf litter decomposition (Tonin et al., 2018).

Body size integrates numerous species traits that are relevant to ecosystem functioning (Woodward et al., 2005), and a loss of heterogeneity in body size within communities may have serious consequences for ecosystem processes (Handa et al., 2014). Our study corroborates these findings and highlights the importance of considering the distribution of detritivore body size as a critical community characteristic when examining the relationships between biodiversity and leaf litter decomposition rates in ponds.

The rate of organic matter decomposition observed in our study was more tightly linked with functional diversity than with species richness. Likewise, Reiss et al. (2011) found that variation in detritivore body size may have a stronger effect on detrital processing rates than species diversity. These results align with the expectation of Gessner et al. (2010) who assumed that direct functional characterization of communities, such as functional diversity, should be a more compelling predictor of aquatic biodiversity effects on litter decomposition than measures of taxonomic diversity. From an ecological niche perspective, when species exhibit trade-offs in resource utilization, a higher diversity of coexisting species can lead to more efficient resource utilization and enhanced overall ecosystem functioning (Chase and Leibold, 2003). Although functional diversity was positively correlated with species richness in the studied ponds (r = 0.57) and such positive relationships were also observed in other aquatic ecosystems (Schmera et al., 2017), species richness is apparently not a reliable proxy for ecosystem functioning. Díaz and Cabido (2001) argue that species richness might be an adequate surrogate for functional diversity under the assumption of improbable scenarios involving random or uniform occupation of niche space. However, aquatic species usually show non-random assembly (Urban, 2004; Brown and Swan, 2010) and clumped occupation of niche space, resulting in the aggregation of species into strong functional types (Petchey and Gaston, 2002). For example, species recorded in our study were clustered in five homogeneous groups with convergent traits ( Figure S2A ). Consequently, a unit increase in species richness may affect functional diversity disproportionally if a new species belongs to a complementary group. On the other hand, the addition of a functionally redundant species with similar traits may cause only negligible changes in functional diversity. The strong clumping of pond-dwelling shredders in the trait space and high functional redundancy are evident in horizontal stratification and the saturating relationship between functional diversity and species richness ( Figure S2B ). Due to the high functional redundancy within our dataset, functional diversity can efficiently explain variation in the rate of litter decomposition even when species richness does not.

Black alder litter decomposition observed in the studied ponds proceeded at an average rate of k = 0.013 days⁻¹, which can be classified as fast breakdown (Petersen and Cummins, 1974). Due to a relatively strong connection between riparian vegetation and aquatic systems in ponds and low-order streams, we expected that leaf breakdown rates in ponds would resemble those observed in upper stream reaches. However, the observed breakdown rates seem to be lower than usually reported for headwater streams of 1st to 3rd order ( Figure S4 ). The significance of shredding detritivores for organic matter breakdown was convincingly demonstrated in streams (Cuffney et al., 1990; Wallace et al., 1995; Jonsson et al., 2001) and in pond-like mesocosms (Auber et al., 2011) and natural ponds (Oertli, 1993). Nevertheless, the role and importance of shredders differ among aquatic systems.

Low-order streams are typically characterized by cool, well-oxygenated water, and these habitats often host a high abundance and diversity of shredders (Vannote et al., 1980; Minshall et al., 1983). In contrast, ponds, owing to their morphometric characteristics such as shallow depths and limited volume, might offer less favorable conditions characterized by fluctuating temperatures and often low oxygen contents. In our study, for instance, all ponds experienced maximum recorded temperatures exceeding 14°C, and oxygen concentrations dropped below 2 mg/L at several sites. These conditions have the potential to eliminate important shredder groups, such as gammarids and stoneflies (Toman and Dall, 1998; Pardo and García, 2016). Indeed, we observed low species numbers and abundances of shredders in all studied ponds. Depauperated communities of pond-dwelling shredders have already been reported elsewhere (Oertli, 1993; Bottollier-Curtet et al., 2011). Indeed, studies of litter decomposition in ponds are too scarce to allow for wider generalization, but it seems that slower decomposition of alder leaves in ponds may reflect low diversity and abundance of shredders in these systems.