Accepted Article

Received Date : 03-Jan-2013 Revised Date : 29-Aug-2013 Accepted Date : 02-Sep-2013 Article type : Standard Paper Editor : Guy Woodward Section : Community Ecology

When does diversity matter? Species functional diversity and ecosystem functioning across habitats and seasons in a field experiment

André Frainer1*, Brendan G. McKie2 and Björn Malmqvist1

1. Department of Ecology and Environmental Science, Umeå University, Umeå, Sweden, SE 901 87 2. Department of Aquatic Sciences and Assessment, Swedish University of Agricultural Sciences, Uppsala, Sweden, SE 750 07 * Current address: Department of Arctic and Marine Biology, University of Tromsø, Tromsø, Norway, 9037

Corresponding author: André Frainer, Department of Arctic and Marine Biology, Faculty of BioSciences, Fisheries and Economics, University of Tromsø, Tromsø, Norway, 9037. [email protected]

Running headline: Functional diversity and ecosystem functioning

This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/1365-2656.12142 This article is protected by copyright. All rights reserved.

Accepted Article

Summary 1.

Despite ample experimental evidence indicating that biodiversity might be an

important driver of ecosystem processes, its role in the functioning of real ecosystems remains unclear. In particular, the understanding of which aspects of biodiversity are most important for ecosystem functioning, their importance relative to other biotic and abiotic drivers, and the circumstances under which biodiversity is most likely to influence functioning in nature, is limited. 2.

We conducted a field study that focussed on a guild of insect detritivores in

streams, in which we quantified variation in the process of leaf decomposition across two habitats (riffles and pools) and two seasons (autumn and spring). The study was conducted in six streams, and the same locations were sampled in the two seasons. 3.

With the aid of structural equations modelling, we assessed spatio-temporal

variation in the roles of three key biotic drivers in this process: functional diversity, quantified based on a species trait matrix, consumer density and biomass. Our models also accounted for variability related to different litter resources, and other sources of biotic and abiotic variability among streams. 4.

All three of our focal biotic drivers influenced leaf decomposition, but none was

important in all habitats and seasons. Functional diversity had contrasting effects on decomposition between habitats and seasons. A positive relationship was observed in pool habitats in spring, associated with high trait dispersion, whereas a negative relationship was observed in riffle habitats during autumn. 5.

Our results demonstrate that functional biodiversity can be as significant for

functioning in natural ecosystems as other important biotic drivers. In particular, variation in the role of functional diversity between seasons highlights the importance

This article is protected by copyright. All rights reserved.

Accepted Article

of fluctuations in the relative abundances of traits for ecosystem process rates in real ecosystems.

Key-words: stream ecosystems, path analyses, spatial-temporal variability, species evenness, species traits, litter decomposition

Introduction Despite evidence from controlled experiments demonstrating the potential for species diversity to influence ecosystem processes (Balvanera et al. 2006; Cardinale et al. 2006), there is still much uncertainty about the relevance of these results for understanding the functioning of real ecosystems (Reiss et al. 2009). In particular, the understanding of when and where changes in species diversity are most likely to influence functioning in real world ecosystems, relative to other biotic variables, remains limited (Reiss et al. 2009; Duffy 2009; Hooper et al. 2012). In part, this is a consequence of the dominance of short-term, small scale manipulative settings used in testing biodiversity-ecosystem functioning (B-EF) theory, where the spatial and temporal variability characteristic of real ecosystems, and its implications for ecosystem functioning, may have been overlooked (Stachowicz et al. 2008; Reiss et al. 2009). Spatio-temporal variation in species composition and diversity can influence ecosystem functioning by affecting the distribution of functional traits present in local communities (Mouillot et al. 2013). These are traits that directly influence habitat use and organismal performance, particularly in relation to resource use and processing, and biomass production. Increases in functional diversity may often be a consequence of increases in species richness (Petchey & Gaston 2002), but the two components are not always linearly correlated (Botta-Dukát 2005; Ricotta 2005; Laliberté & Legendre

This article is protected by copyright. All rights reserved.

Accepted Article

2010). An increase in species richness may have little impact on ecosystem functioning if the new species are functionally very similar to those already present in a community (Fonseca & Ganade 2001; Joner et al. 2011). In contrast, the addition of specific, novel traits which have direct positive or negative effects on key ecosystem processes (Burkepile & Hay 2008; Jousset et al. 2011) may have stronger functional consequences. Traits which affect species interactions within functional guilds can also alter ecosystem processes. This is seen both when the activities of a new species facilitates feeding by other consumers, enhancing ecosystem functioning (Jonsson & Malmqvist 2003), and when particular species increase antagonistic interactions (Polley, Wilsey & Derner 2003; McKie et al. 2009; Jousset et al. 2011), driving down resource consumption overall.

Not only the occurrence of specific functional traits, but also the relative distribution of those traits, can vary across habitats and seasons (Beche, Mcelravy & Resh 2006; McGill, Sutton-Grier & Wright 2009). Decreasing trait evenness may be associated with enhanced ecosystem functioning in cases where the dominant trait (or group of traits) are also the most productive (Hector et al. 2002; Dangles & Malmqvist 2004; McKie et al. 2008). Alternatively, if the effects of the traits on a process are complementary to one another, then a more even distribution of traits may favour enhanced ecosystem functioning (Kirwan et al. 2007). A further key biological determinant of ecosystem functioning is the biomass of individuals within functional guilds (Reiss et al. 2011). Well-known relationships between biomass and consumer metabolic requirements have strong effects on consumer-resource interactions (Gruner et al. 2008), and rates of energy flow at an ecosystem level are positively related to bulk producer and consumer biomass (Brown

This article is protected by copyright. All rights reserved.

Accepted Article

et al. 2004). However, these basic relationships between resource consumption and biomass can be altered by density-dependent variation in inter- and intraspecific interactions and resource use efficiency. In systems characterised by pulses of allocthonous resource, progressively decreasing resource availability may increase consumer densities relative to that resource, increasing the potential for strong species interactions (Presa Abos et al. 2006; Tiegs et al. 2008). At high densities, negative density-dependent interactions can result in impaired per capita resource use due to resource competition, causing process rates to decline (Amundsen, Knudsen & Klemetsen 2007). Therefore, partitioning of species effects on ecosystem functioning related to variation in assemblage resource requirements, associated with consumer biomass, versus variation in interaction potential, associated with consumer density and functional diversity, requires simultaneous evaluation of all three biotic drivers.

Spatio-temporal variability in the importance of diversity for ecosystem functioning is likely to be particularly high in communities characterized by mobile, short-lived species where species trait composition can shift rapidly over short to medium spatio-temporal scales, such as terrestrial (Albrecht & Gotelli 2001; Dangles, Carpio & Woodward 2012) or aquatic invertebrate communities (Dangles & Malmqvist 2004; Göthe, Angeler & Sandin 2012) and algal communities (Matthiessen & Hillebrand 2006). In this study, we focussed on leaf litter decomposition as a key component of ecosystem functioning in detritus-based foodwebs, and assessed the importance of detritivore functional diversity for this process in stream ecosystems, relative to other dominant biotic drivers and background environmental variation. We investigated variation in decomposition rates across habitats (riffles and pools) and seasons (autumn and spring), encompassing marked variability in environmental

This article is protected by copyright. All rights reserved.

Accepted Article

conditions (high vs low water flow velocity, cool vs warm temperatures), and resource quality and quantity. We used both structural equations modelling, which allows testing of causal pathways in complex data (Grace et al. 2010) and ANOVA to assess the key questions of when (which seasons) and where (which habitats) functional diversity is important for ecosystem functioning, relative to detritivore density and biomass, and other sources of environmental variation. We hypothesized that (1) biomass is the biotic driver most consistently associated with decomposition, reflecting the strong association between consumer resource requirements and biomass, but that (2) decomposition rates may decline at higher detritivore densities, if density-dependent interactions negatively affect resource processing. Finally, we expect (3) positive relationships between decomposition rates and functional trait diversity, with the highest processing rates expected in assemblages characterised by an even distribution of functionally contrasting traits, reflecting a higher potential for complementary niche differentiation and resource use.

Methods Site description Our study was replicated spatially across six streams located within the province of Västerbotten, northern Sweden (proximate coordinates 63°50’ N, 20°15’ E), all flowing through intact boreal forest, with minimal impact from human disturbances. Streams were highly comparable in basic physico-chemical characteristics, including width (4-6 m), substrate composition (rocky riffles and sandy pools) and riparian vegetation (predominantly the conifers Picea abies (L.) and Pinus sylvestris (L.), and deciduous trees Betula spp and Alnus spp). Within each stream, two adjacent 20 m stretches of riffle and pool habitat were delineated, differing primarily in current velocity,

This article is protected by copyright. All rights reserved.

Accepted Article

substrates, and the composition of detritivore assemblages. Flow measurements (Model 801, Valeport, Totnes, UK) taken at 60% water column depth confirmed the marked flow differences between the two habitats (F1,15 = 104.31, p < 0.001), with higher velocities in riffles than pools (Table 1). Boreal streams are characterized by strong seasonal variation in environmental characteristics and resources, and community composition. We captured this variation by conducting our study across two seasons: autumn 2008 and spring 2009. Leaf litter is more abundant in the autumn, when the most palatable species are still present, whereas in spring it is less abundant, and only less palatable species remain (Staelens et al. 2011). Water temperature and litter standing stocks differed between seasons (both F1,15 > 10.73, p < 0.001), but not habitats (both F1,15 < 2.41, p > 0.14). Higher water temperatures were found in spring, while higher standing stocks of leaf litter were found in autumn (Table 1). Important water chemistry variables, including pH (6.85 ± 0.29) and total nitrogen (0.312 ± 0.11 mg/l) varied little among sites and seasons (all F1,20 < 0.95 , p > 0.34), while total phosphorus was always below the detection limit (< 0.04 mg/l) (Table 1).

Ecosystem functioning Leaves of the two most abundant deciduous trees in the riparian zone, grey alder (Alnus incana (L.) Moench) and silver birch (Betula pendula Roth), were collected just after abscission directly from the ground, and air-dried. Replicate amounts (4 ± 0.05 g) of each species were enclosed separately in mesh bags (mesh opening = 10 mm), which allowed stream invertebrates to colonize the leaf litter. In total, five mesh bags were exposed per habitat in autumn and spring, and retrieved when decomposition of the most rapidly decomposing species reached approximately 50% (42 days in Autumn, 20

This article is protected by copyright. All rights reserved.

Accepted Article

days in spring). This procedure ensured that a similar decomposition stage was sampled in both seasons, facilitating a meaningful assessment of the relative importance of the different biotic drivers at the same point in the decomposition process. After retrieval, leaf litter was rinsed under tap water, with colonizing insects preserved in 70% ethanol. All remaining leaf material was analyzed as ash-free dry-mass (AFDM), following combustion at 550°C in a muffle furnace for 5 h. Leaf decomposition was calculated as a daily rate (kd), by applying the negative exponential model (Petersen & Cummins 1974): kd = ln(Wt/W0)/t, where W0 = initial dry mass (g); Wt = AFDM at time t; and t = exposure time in days. As different thermal regimes across seasons might influence the decomposition process (Boulton & Boon 1991), we further calculated decomposition rates per degree-day (kdd) using the same exponential model, but where t = sum of daily average temperatures, in °C, across the exposure period. W0 was corrected for losses of soluble compounds by leaching, estimated following immersion of replicate litter packs for 24h under slow running tap water (correction factor = 0.75 and 0.83 for alder and birch, respectively).

Differences in leaf litter decomposition were analysed with linear mixed effect models (LMEM) in the R package nlme (Pinheiro et al. 2012). In these analyses, season, habitat, leaf species, and 2 and 3–way interaction terms were set as fixed factors. Stream identity was fitted as a random variable, accounting for background environmental variation among streams not explicitly included in the model. As our samples were collected during one autumn and spring only, we did not have true temporal replication of season. However, our spatial replication of each period, represented by multiple streams, was extensive. As we aimed at directly contrasting habitats and leaf species between the two sampling periods, we fitted season as a fixed

This article is protected by copyright. All rights reserved.

Accepted Article

rather than random factor, while acknowledging that details of these responses may have differed in other years with different seasonal trajectories. Overall, environmental conditions during the two periods were entirely typical of a Swedish boreal autumn (progressively decreasing temperatures and increasing rainfall, with snow falling towards the end of the period) and spring (progressively increasing temperatures and decreasing rainfall following the spring flood). Data was natural log transformed to satisfy assumptions of parametric analysis.

Invertebrate detritivore community composition Invertebrate leaf-shredding detritivores colonising the litter-bags were sorted, identified to the lowest possible taxonomic level and counted. In most cases, identification was to species level, but the genera Amphinemura and Leuctra (both Plecoptera) and Halesus and Limnephilus (both Trichoptera) were identified to genus. Densities were quantified as detritivore number per litter-bag. After identification, detritivores were placed in preweighed aluminium pans, oven-dried at 60°C for 24h, and weighed to the nearest 0.1 mg. Differences in richness, density, and biomass were tested using the LMEM model structure described above, after natural log transformation.

Functional diversity To characterise detritivore functional diversity, we compiled a set of traits (Appendix S1) representing the extent to which detritivores share habitats and resources, and their capacity to influence functioning: (1) feeding strategy, (2) mean per capita species biomass, (3) emergence period, (4) substrate preference, and (5) current velocity preference. Feeding strategy characterizes the breadth of feeding mode for each species (leaf shredding, biofilm grazing, particle gathering, and/or predacious), allowing for a

This article is protected by copyright. All rights reserved.

Accepted Article

comparison between more obligatory and generalist detritivores. Mean per-capita biomass for a species correlates with the metabolic needs of the individuals (Brown et al. 2004), and therefore with their potential impact on ecosystem functioning. The period over which populations emerge from the aquatic to terrestrial environment relates to differences in phenology among the distinct species and reflects energy requirements and allocation across seasons (all our detritivores were insects). Finally, substrate and flow preference reflect microhabitat preferences of each species, which are likely to dictate the conditions under which they operate most efficiently. Mean detritivore biomass data for each species was obtained from a database compiled by Frainer et al. (unpublished data), while all other traits were scored with the aid of information from the Freshwater Ecology database (Schmidt-Kloiber & Hering 2011) (Appendix S1).

As a measurement of species functional diversity, we calculated functional dispersion (FDispersion) (Laliberté & Legendre 2010), which is similar in concept and highly correlated to Rao’s Q functional diversity index (Pearson Correlation = 0.93, p < 0.01 in this study), and simultaneously quantifies both trait dissimilarity and evenness within communities (Botta-Dukát 2005; Laliberté & Legendre 2010). Accordingly, the most functionally dispersed assemblage (high FDispersion) is composed of evenly distributed, dissimilar traits. Using the R package FD (Laliberté & Shipley 2011), we firstly calculated the Gower distance-matrix from the species functional traits. This matrix was then used to calculate a centroid for each detritivore assemblage (i.e. from each litter bag), after accounting for variation in species abundances. FDisperison is the sum of the distances of each species to that centroid, weighted again by species relative abundances (Laliberté & Legendre 2010). Differences in FDisperison between seasons,

This article is protected by copyright. All rights reserved.

Accepted Article

habitats, and litter species were tested using the LMEM model structure as described above, after natural log transformation. Twenty litterbags colonised by only a single detritivore species were excluded from this analysis, since FDispersion cannot be calculated for single species assemblages (Laliberté & Legendre 2010). Finally, the relationship between FDispersion and taxonomic diversity measures (species richness and evenness) was assessed by first standardizing the predictor variables, and then using a multiple regression with species taxonomic richness and species taxonomic evenness – Hurlbert’s PIE (Hurlbert 1971) – as predictors.

Biotic predictors of ecosystem functioning We used Structural Equation Modelling (SEM) to assess influences of the biotic predictors on the decomposition process, relative to background environmental variation among the streams. SEM is based on variance-covariance matrices between predictor and response variables, and uses information on the correlations between these variables, and their statistical significance, to evaluate both direct and indirect causal pathways (Grace 2006).

To assess the importance of the biotic predictors across space and time, we constructed individual SEM’s for each season and habitat, yielding four models in total, each containing 60 sampling units (6 streams x 2 litter species x 5 litter-bags), and 6 explanatory/response variables (stream and litter identity, detritivore biomass, density, FDispersion, and leaf decomposition rate). Stream identity was fitted to account for sources of abiotic and biotic variation between streams not accounted for by the other fitted variables. Stream identity and litter species were fitted as exogenous random variables (i.e.: influencing other variables in the model but not themselves influenced by other

This article is protected by copyright. All rights reserved.

Accepted Article

variables), with means, variances and covariances set as free parameters (Rosseel 2012). The biotic variables detritivore density, biomass, and FDispersion were included as endogenous variables (influenced by the two exogenous variables and free to influence other endogenous variables in the model). Potential correlations between density and biomass, and density and FDispersion were also modelled. Finally, decomposition was set as a response-only endogenous variable, influenced by all other endogenous and exogenous variables. As in the Fdispersion ANOVA, samples containing only one species were removed from the SEMs, since FDispersion cannot be calculated for single-species assemblages. All SEMs were run using the R package lavaan (Rosseel 2012), and all continuous variables were square-root transformed prior to analyses. The robustness of SEM is based on an assessment of overall model fit (e.g., Chisquare), which indicates if a model structure fits the data structure, rather than on the statistical significance of individual relationships within the model (Grace et al. 2010). In cases where a pairwise relationship between variables within a model was only marginally significant at the 5% level, we constructed an alternative model where that relationship was formally set to equal zero and compared the fit of the two models using the function anova (stats package, R Core Team 2012). Variables were retained in the final model when their inclusion significantly improved overall model fit relative to the alternative model. We present results for standardized path coefficients, which allow for comparing relationship strengths within a model (Grace et al. 2010).

Results Litter species decomposition Decomposition rates per day were approximately 7 times faster in spring (0.066 ± 0.007 day-1) than autumn (0.009 ± 0.001 day-1; F1,220 = 443.96; p < 0.001), and differed

This article is protected by copyright. All rights reserved.

Accepted Article

between habitats (F1,220 = 12.85; p < 0.001) and litter species (F1,220 = 7.59; p < 0.001) (Fig. 1). A two-way interaction term between season and habitat (F1,220 = 11.39; p < 0.001) reflected faster decomposition rates in riffles (0.012 ± 0.001 day-1) than in pools (0.007 ± 0.001 day-1) during autumn, whereas there was no difference between habitats in spring (riffle: 0.063 ± 0.010 day-1, pool: 0.070 ± 0.010 day-1; Fig. 1a). Decomposition rates of the litter species differed between seasons (interaction F1,220 = 47.73; p < 0.001), with birch decomposing faster (0.010 ± 0.001 day-1) than alder (0.008 ± 0.001 day-1) in autumn, but alder (0.098 ± 0.013 day-1) decomposing faster than birch (0.037 ± 0.004 day-1) in spring. Finally, a three-way interaction between season, habitat, and litter species arose because faster decomposition of birch in autumn was mainly associated with riffles, while faster decomposition of alder in spring was mainly associated with pools (F1,220 = 5.12; p = 0.025). All these effects remain significant even when decomposition is standardized for temperature, though with less marked differences between seasons (Appendix S2).

Detritivore assemblage characteristics and functional diversity Detritivore density was higher during autumn (F1,220 = 61.09; p < 0.001. Fig. 2a) than in spring, and riffles were characterized by higher densities than pools in both seasons (F1,220 = 92.57; p < 0.001. Fig 2a). Densities on birch litter were higher than on alder (F1,220 > 12.52; p < 0.001. Fig. 2a). No further interactions between season, habitat and/or litter species were significant (all F1,220 < 1.45; p > 0.23). In contrast to density, detritivore biomass was higher in spring (F1,220 = 34.58; p < 0.001. Fig. 2b), with no difference between habitats (F1,220 = 1.01; p = 0.32. Fig. 2b). Biomass was also higher on birch than alder litter (F1,220 = 9.13; p = 0.003. Fig. 2b). No further interactions were significant (all F1,220 < 3.54; p > 0.06).

This article is protected by copyright. All rights reserved.

Accepted Article

Across both seasons and habitats, functional dispersion (FDispersion) was positively related to the taxonomic diversity of detritivores (multiple regression r2 = 0.87, F2,208 = 578.6, p < 0.001), but the relationship was stronger for species evenness (slope = 8.31 ± 0.41, p < 0.001) than richness (slope = 0.47 ± 0.12, p < 0.001). Detritvore species richness was higher in autumn than spring (F1,220 = 123.42, p < 0.001), higher in riffles than in pools (F1,220 = 13.04, p < 0.001), and higher on birch than alder (F1,220 = 5.57, p = 0.019) (Fig. 2d). There were no further significant interactions (p > 0.05). FDispersion was higher in autumn than spring (F1,204 = 54.06, p < 0.001. Fig. 2c), and higher on birch than alder (F1,204 = 6.90, p = 0.009. Fig. 2c), but there was no difference in FDispersion between habitats (F1,204 = 0.01, p = 0.97. Fig. 2c). A significant 2way interaction was found between habitat and litter species (F1,204 = 4.68, p = 0.032), meaning that birch and alder were characterized by similar FDispersion in riffles, but not in pools (Fig. 2c). No further interactions were significant (all F1,204 < 3.39; p > 0.07). Further differences in the composition of detritivore assemblages between habitats and seasons are detailed in Appendices S3-4.

Structural Equation Model Our SEMs had a good overall fit, with mostly non-significant chi-square values, which denote a good fit of the data given the model structure (Fig. 3 and 4). The two autumn models yielded marginally significant relationships between two of the explanatory variables and leaf decomposition (FDispersion in riffles and density in pools). The autumn riffle model fit improved (Chi-square = 4.13; d.f. = 1; p = 0.042) when the relationship between FDispersion and leaf decomposition was maintained in the structure of the model, compared to an alternative model where this relationship was set to zero. However,

This article is protected by copyright. All rights reserved.

Accepted Article

maintenance of the relationship between density and leaf decomposition in the autumn pool model did not improve overall model fit (Chi-square = 2.91; d.f. = 1; p = 0.088). Accordingly, FDispersion was maintained in the riffle model, whereas density was excluded from the pool model. The relationship between the biotic variables and leaf decomposition differed between habitats and seasons in our SEMs. FDispersion and leaf decomposition rates were positively associated in spring pools (Fig. 4b) and negatively associated in autumn riffles (Fig. 3a). There was no association between density and leaf decomposition in autumn (Fig. 3b), but we observed a positive relationship between the two variables in spring pools (Fig. 4b). We also found a positive relationship between biomass and decomposition rates in both habitats during autumn (Fig. 3a and 3b respectively). Density and biomass were positively correlated in autumn riffles and pools, and in spring riffles (Fig. 3a, 3b, and 4a, respectively), and density and FDispersion were negatively correlated in spring riffles (Fig. 4a). The effects of litter species on leaf decomposition differed across seasons. In both habitats, birch decomposed faster than alder during autumn (Fig. 3a,b), whereas the opposite occurred in spring (Fig. 4a,b). Litter identity did not affect detritivore density, biomass or FDispersion in autumn, but during spring birch was associated with higher detritivore biomass in riffles (Fig. 4a) and higher detritivore density in pools (Fig. 4b).

Discussion By simultaneously investigating variability in an important ecosystem process and its major biotic drivers across seasons and habitats, we were able to assess the role of functional diversity in the context of changing environmental and community dynamics.

This article is protected by copyright. All rights reserved.

Accepted Article

We found that detritivore functional diversity was associated with variability in leaf decomposition, but its effects contrasted among habitats and seasons, with both positive and negative relationships observed. In contrast, the detected relationships between functioning and the other two biotic drivers were positive, with biomass most important in the autumn and density in the spring. The variation in importance of these predictors reflects known dynamics in the ecology of detrital food webs in temperate and boreal streams, while the results for functional diversity highlight the importance of fluctuations in the relative abundances of traits for ecosystem process rates in real ecosystems.

In spring, leaf litter standing stocks are highly reduced and other potential food sources, such as algae, are still rare (Lepori et al. 2006). The scarcity of resources may concentrate species together, potentially favouring both an increased co-occurrence of contrasting traits at local scales, and stronger species interactions, and hence stronger biodiversity-functioning relationships. In line with this, we observed a positive relationship between litter decomposition and functional dispersion during the spring, in pool habitats. This result is particularly significant given the decreased richness of insect detritivores in spring, following the early emergence of some stonefly species (Appendix S4). A higher functional dispersion indicates a more even distribution of dissimilar traits, and in our study functional dispersion was more strongly associated with variation in species evenness than richness. Accordingly, the relationship found in spring indicates that higher process rates occurred in pools when functionally distinct traits were present with similar abundances. This points towards a positive effect of niche differentiation (Polley et al., 2003; McKie et al. 2009), though other types of mechanism may also have been important (including facilitative interactions, and

This article is protected by copyright. All rights reserved.

Accepted Article

selection effects related to the presence of particular species). Controlled experiments are needed to fully identify the underlying mechanisms, though facilitation has been observed previously in boreal detritivore assemblages (Jonsson & Malmqvist 2003). Functional dispersion was additionally associated with decomposition in autumn riffles, but in contrast with the spring pools, the relationship was negative, and the correlation relatively weak. The negative relationship indicates that process rates were higher when assemblages were dominated by a few, functionally similar, traits. This is suggestive of positive species identity effects on functioning (Dangles & Malmqvist 2004), particularly given that lower functional dispersion was associated with higher species dominance (lower evenness) in our study.

Detritivore biomass was expected to be a more consistent biotic driver of leaf decomposition across habitats and seasons than either density or functional diversity, reflecting the strong association between consumer biomass, their metabolic requirements, and resource consumption (Vaughn, Spooner & Galbraith 2007; Reiss et al. 2011). Supporting this, we saw overall higher process rates when biomass was greatest in the spring, particularly in pools, apparent even after decomposition rates were corrected for temperature. However, within seasons, biomass was only related to leaf decomposition in the autumn. Autumn is early in the main growth period for the majority of detritivores, which continue feeding under the winter ice and emerge in the spring (Malmqvist, Nilsson & Svensson 1978). It is not surprising that biomass would be a dominant driver of resource consumption when the consumers are actively feeding, reflecting their underlying metabolic needs. Biomass in spring habitats was outweighed in importance by other biotic and abiotic drivers, with stream and litter identity in riffles and detritivore functional

This article is protected by copyright. All rights reserved.

Accepted Article

diversity and density in pools explaining decomposition rates. During the spring, detritivore biomass is dominated by large-bodied, late-instar insects, with case-bearing caddisflies (especially Limnephilidae) particularly prominent in pools in our study (Appendices S3-S4). Many of these caddisflies, while continuing to consume leaf material, adopt an increasingly predacious diet in spring as they mature reproductive structures prior to emergence (Wissinger et al. 2004). The extent of late-instar predacious behaviour among detritivorous caddisflies, including Swedish species (B.G: McKie, unpublished data), can be great (Wissinger et al. 2004), and the acquisition of a substantial portion of energy and nutrients from animal prey by the largest detritivores in our spring assemblages is likely to have weakened the linkage between biomass and leaf decomposition. If confirmed as an explanation for our results, then this may illustrate the potential for phenological shifts in feeding behaviour, and therefore trait expression, to weaken the relationship between resource consumption rates and the biomass of consumer guilds.

The detritivore densities observed in our study during the autumn have been associated with negative effects on per capita consumption rates in previous microcosm studies, attributed to increased rates of interference competition, especially within species (McKie et al. 2008, Jonsson & Malmqvist 2003). However, there was no evidence for negative density-dependent variation in bulk process rates in our field study. Unlike previous microcosm experiments, individuals could freely move among litter patches in our study, which might have allowed individuals to avoid the highest density litter-bags. Even if some species did experience negative density effects on percapita resource consumption, these evidently were not strong enough to alter bulk process rates. Given that our densities were comparable to the highest densities

This article is protected by copyright. All rights reserved.

Accepted Article

observed previously in the region (McKie et al. 2008), these results suggest that negative density-dependent variation in processing rates is unlikely to overcome positive effects due to increased biomass (McKie et al. 2008; Klemmer et al. 2012). Indeed, the only significant density effect on decomposition in our study was a positive relationship in spring pools, which suggests that even if predacious behaviour by late instar caddisflies weakened the linkage between consumer biomass and leaf decomposition, the addition of more individuals nevertheless increased process rates. Stream identity overall explained a significant amount of variation in both leaf decomposition and its biotic predictors, and it was the dominant driver of decomposition in riffle habitats during the spring. In this season, boreal streams are typically subjected to high environmental variability following spring snowmelt and the associated flood peak (Petrin et al. 2007), when water chemistry and flows are particularly variable among streams, and sediment loads are elevated (McKie, Petrin & Malmqvist 2006). Under normal low-moderate flow conditions, current velocity per se has no detectable influence on decomposition (Ferreira & Graça 2006). But under high flow conditions, and particularly in combination with suspended sediments, physical fragmentation associated with abrading currents may overwhelm the influences of detritivores on leaf decomposition (Ferreira & Graça 2006; Niu & Dudgeon 2011). Given our spring study was conducted during this period, it is possible that these factors overwhelmed the influences of biotic factors in the fast-flowing riffles, which were characterized by higher flow variability in the spring than autumn. Stream identity also encompasses background variation in other biological factors, including microbial community composition or the presence of different predators, and it is possible that such factors explain some of the variation in decomposition, either through direct

This article is protected by copyright. All rights reserved.

Accepted Article

effects on the process, and/or by altering the responses of detritivores (Jabiol et al. 2013). The effects of leaf litter identity on ecosystem functioning differed between seasons, with birch more rapidly consumed in autumn, and alder in spring. The reasons for this remain to be understood, but it may reflect differences in carbon and nutrient content in each leaf species (Schindler & Gessner 2009), which coupled with different developmental nutrient requirements of the invertebrate detritivores (Wissinger et al. 2004) in the two seasons could help explain the distinct patterns of leaf decomposition seen here (Hladyz et al. 2009). Alternatively, the contrasting results for the two litter species might reflect differences in microbial responses, which we did not quantify.

It has become increasingly apparent that the effects of biodiversity on ecosystem functioning, whether positive or negative, are likely to be context dependent for most ecosystems (Vaughn, Spooner & Galbraith 2007; McKie et al. 2009; Dangles et al. 2011). This, together with a lack of data from real ecosystems, has generated understandable scepticism, and has hindered the general acceptance of the importance of B-EF research in conservation and management (Srivastava & Vellend 2005; Thompson & Starzomski 2006; Dangles et al. 2011). In this field experiment, we have gathered evidence that functional diversity sometimes explained as much variation in our focal ecosystem process as other major biotic drivers of functioning. By relating the variation in diversity-functioning relationships to known dynamics in boreal stream ecology, we were able to suggest several hypotheses to explain this context dependency, related to dynamics in resource densities, animal phenologies and behaviours, and seasonal disturbance regimes. Significantly, the species traits underpinning the calculation of the functional diversity metrics provide a degree of independence from

This article is protected by copyright. All rights reserved.

Accepted Article

taxonomic measures of community diversity and structure (Mouillot et al. 2013). As such, a stronger focus on documenting systematic variation in not only the occurrence but also the expression of species traits – related to shifts in phenology, competition, resource availability, and other key environmental factors – and how they are distributed within communities offers prospects for developing a more predictive approach, less encumbered by the vagaries of individual species identities, to both basic and applied B-EF spheres.

Acknowledgments We are grateful to A. Bos, M. Duarte, P. Esberg, E. Fältström, J. Gustafsson, and A. Sandling for invaluable assistance in several field and laboratory tasks of this project, and to Jarrett Byrnes for his generous guidance on the construction of structural equation models. Comments on earlier drafts by Olivier Dangles, Per-Arne Amundsen, Rune Knudsen and three anonymous referees are very much appreciated. This research was funded by a grant from the Swedish Research Council (VR 621-2006-375) to B. Malmqvist.

References Albrecht, M. & Gotelli, N.J. (2001) Spatial and temporal niche partitioning in grassland ants. Oecologia, 126, 134–141. Amundsen, P.-A., Knudsen, R.R. & Klemetsen, A.A. (2007) Intraspecific competition and density dependence of food consumption and growth in Arctic charr. Journal of Animal Ecology, 76, 149–158. Balvanera, P., Pfisterer, A.B., Buchmann, N., He, J.-S., Nakashizuka, T., Raffaelli, D. & Schmid, B. (2006) Quantifying the evidence for biodiversity effects on This article is protected by copyright. All rights reserved.

Accepted Article

ecosystem functioning and services. Ecology Letters, 9, 1146–1156. Beche, L.A., Mcelravy, E.P. & Resh, V.H. (2006) Long-term seasonal variation in the biological traits of benthic-macroinvertebrates in two Mediterranean-climate streams in California, U.S.A. Freshwater Biology, 51, 56–75. Botta-Dukát, Z. (2005) Rao's quadratic entropy as a measure of functional diversity based on multiple traits. Journal of vegetation science, 16, 533–540. Boulton, A.J. & Boon, P. (1991) A review of methodology used to measure leaf litter decomposition in lotic environments: Time to turn over an old leaf? Marine and Freshwater Research, 42, 1–43. Brown, J.H., Gillooly, J.F., Allen, A.P., Savage, V.M. & West, G.B. (2004) Toward a metabolic theory of ecology. Ecology, 85, 1771–1789. Burkepile, D.E. & Hay, M.E. (2008) Herbivore species richness and feeding complementarity affect community structure and function on a coral reef. Proceedings of the National Academy of Sciences, 105, 16201–16206. Cardinale, B.J., Srivastava, D.S., Emmett Duffy, J., Wright, J.P., Downing, A.L., Sankaran, M. & Jouseau, C. (2006) Effects of biodiversity on the functioning of trophic groups and ecosystems. Nature, 443, 989–992. Dangles, O. & Malmqvist, B. (2004) Species richness-decomposition relationships depend on species dominance. Ecology Letters, 7, 395–402. Dangles, O.O., Crespo-Pérez, V.V., Andino, P.P., Espinosa, R.R., Calvez, R.R. & Jacobsen, D.D. (2011) Predicting richness effects on ecosystem function in natural communities: insights from high-elevation streams. Ecology, 92, 733–743.

This article is protected by copyright. All rights reserved.

Accepted Article

Dangles, O., Carpio, C. & Woodward, G. (2012) Size-dependent species removal impairs ecosystem functioning in a large-scale tropical field experiment. Ecology, 93, 2615–2625. Duffy, J.E. (2009) Why biodiversity is important to the functioning of real-world ecosystems. Frontiers in Ecology and the Environment, 7, 437–444. Ferreira, V. & Graça, M.A.S. (2006) Do invertebrate activity and current velocity affect fungal assemblage structure in leaves? International Review of Hydrobiology, 91, 1–14. Fonseca, C.R. & Ganade, G. (2001) Species functional redundancy, random extinctions and the stability of ecosystems. Journal of Ecology, 89, 118–125. Göthe, E., Angeler, D.G. & Sandin, L. (2012) Metacommunity structure in a small boreal stream network. The Journal of animal ecology, 82, 449-458. Grace, J. (2006) Structural equation modeling and natural systems. Cambridge University Press, Cambridge. Grace, J., Anderson, T., Olff, H. & Scheiner, S. (2010) On the specification of structural equation models for ecological systems. Ecological Monographs, 80, 67–87. Gruner, D.S., Smith, J.E., Seabloom, E.W., Sandin, S.A., Ngai, J.T., Hillebrand, H., Harpole, W.S., Elser, J.J., Cleland, E.E., Bracken, M.E.S., Borer, E.T. & Bolker, B.M. (2008) A cross-system synthesis of consumer and nutrient resource control on producer biomass. Ecology Letters, 11, 740–755. Hector, A., Bazeley-White, E., Loreau, M., Otway, S. & Schmid, B. (2002) Overyielding in grassland communities: testing the sampling effect hypothesis with

This article is protected by copyright. All rights reserved.

Accepted Article

replicated biodiversity experiments. Ecology Letters, 5, 502–511. Hladyz, S., Gessner, M.O., Giller, P.S., Pozo, J. & Woodward, G. (2009) Resource quality and stoichiometric constraints on stream ecosystem functioning. Freshwater Biology, 54, 957–970. Hooper, D.U., Adair, E.C., Cardinale, B.J., Byrnes, J.E.K., Hungate, B.A., Matulich, K.L., Gonzalez, A., Duffy, J.E., Gamfeldt, L. & O'Connor, M.I. (2012) A global synthesis reveals biodiversity loss as a major driver of ecosystem change. Nature, 486, 105–108. Hurlbert, S.H. (1971) The nonconcept of species diversity: a critique and alternative parameters. Ecology, 52, 577–586. Jabiol, J., McKie, B.G., Bruder, A., Bernadet, C., Gessner, M.O. & Chauvet, E. (2013) Trophic complexity enhances ecosystem functioning in an aquatic detritus-based model system. Journal of Animal Ecology, 82, 1042-1051. Joner, F., Specht, G., Müller, S.C. & Pillar, V.D. (2011) Functional redundancy in a clipping experiment on grassland plant communities. Oikos, 120, 1420–1426. Jonsson, M. & Malmqvist, B. (2003) Mechanisms behind positive diversity effects on ecosystem functioning: testing the facilitation and interference hypotheses. Oecologia, 134, 554–559. Jousset, A., Schmid, B., Scheu, S. & Eisenhauer, N. (2011) Genotypic richness and dissimilarity opposingly affect ecosystem functioning. Ecology Letters, 14, 537– 545. Kirwan, L., Lüscher, A., Sebastià, M.T., Finn, J.A., Collins, R.P., Porqueddu, C.,

This article is protected by copyright. All rights reserved.

Accepted Article

Helgadottir, A., Baadshaug, O.H., Brophy, C., Coran, C., Dalmannsdóttir, S., Delgado, I., Elgersma, A., Fothergill, M., Frankow-Lindberg, B.E., Golinski, P., Grieu, P., Gustavsson, A.M., Höglind, M., Huguenin-Elie, O., Iliadis, C., Jørgensen, M., Kadziuliene, Z., Karyotis, T., Lunnan, T., Malengier, M., Maltoni, S., Meyer, V., Nyfeler, D., Nykanen-Kurki, P., Parente, J., Smit, H.J., Thumm, U. & Connolly, J. (2007) Evenness drives consistent diversity effects in intensive grassland systems across 28 European sites. Journal of Ecology, 95, 530–539. Klemmer, A.J., Wissinger, S.A., Greig, H.S. & Ostrofsky, M.L. (2012) Nonlinear effects of consumer density on multiple ecosystem processes. Journal of Animal Ecology, 81, 770–780. Laliberté, E. & Legendre, P. (2010) A distance-based framework for measuring functional diversity from multiple traits. Ecology, 91, 299–305. Laliberté, E. & Shipley, B. (2011) FD: measuring functional diversity from multiple traits, and other tools for functional ecology. R package version 10-11. Lepori, F., Gaul, D., Palm, D. & Malmqvist, B. (2006) Food-web responses to restoration of channel heterogeneity in boreal streams. Canadian Journal of Fisheries and Aquatic Sciences, 63, 2478–2486. Malmqvist, B., Nilsson, L.M. & Svensson, B.S. (1978) Dynamics of detritus in a small stream in southern sweden and its influence on the distribution of the bottom animal communities. Oikos, 31, 3–16. Matthiessen, B. & Hillebrand, H. (2006) Dispersal frequency affects local biomass production by controlling local diversity. Ecology Letters, 9, 652–662.

This article is protected by copyright. All rights reserved.

Accepted Article

McGill, B.M.B., Sutton-Grier, A.E.A. & Wright, J.P.J. (2009) Plant trait diversity buffers variability in denitrification potential over changes in season and soil conditions. PloS one, 5, e11618–e11618. McKie, B.G., Petrin, Z. & Malmqvist, B. (2006) Mitigation or disturbance? Effects of liming on macroinvertebrate assemblage structure and leaf-litter decomposition in the humic streams of northern Sweden. Journal of Applied Ecology, 43, 780–791. McKie, B.G., Schindler, M., Gessner, M.O. & Malmqvist, B. (2009) Placing biodiversity and ecosystem functioning in context: environmental perturbations and the effects of species richness in a stream field experiment. Oecologia, 160, 757– 770. McKie, B.G., Woodward, G., Hladyz, S., Nistorescu, M., Presa Abos, C., Popescu, C., Giller, P.S. & Malmqvist, B. (2008) Ecosystem functioning in stream assemblages from different regions: contrasting responses to variation in detritivore richness, evenness and density. Journal of Animal Ecology, 77, 495–504. Mouillot, D., Graham, N.A.J., Villéger, S., Mason, N.W.H. & Bellwood, D.R. (2013) A functional approach reveals community responses to disturbances. Trends in Ecology & Evolution, 28, 167–177. Niu, S.Q. & Dudgeon, D. (2011) Environmental flow allocations in monsoonal Hong Kong. Freshwater Biology, 56, 1209–1230. Petchey, O.L. & Gaston, K.J. (2002) Functional diversity (FD), species richness and community composition. Ecology Letters, 5, 402–411. Petersen, R.C. & Cummins, K.W. (1974) Leaf processing in a woodland stream.

This article is protected by copyright. All rights reserved.

Accepted Article

Freshwater Biology, 4, 343–368. Petrin, Z., McKie, B., Buffam, I., Laudon, H. & Malmqvist, B. (2007) Landscapecontrolled chemistry variation affects communities and ecosystem function in headwater streams. Canadian Journal of Fisheries and Aquatic Sciences, 64, 1563– 1572. Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D. & R Core Team. (2012) nlme: Linear and Nonlinear Mixed Effects Models. R package version 31-104. Polley, H.W., Wilsey, B.J. & Derner, J.D. (2003) Do species evenness and plant density influence the magnitude of selection and complementarity effects in annual plant species mixtures? Ecology Letters, 6, 248–256. Presa Abos, C., Lepori, F., McKie, B.G. & Malmqvist, B. (2006) Aggregation among resource patches can promote coexistence in stream-living shredders. Freshwater Biology, 51, 545–553. R Core Team. (2012) R: A language and environment for statistical computing. www.Rproject.org. Reiss, J., Bailey, R.A., Perkins, D.M., Pluchinotta, A. & Woodward, G. (2011) Testing effects of consumer richness, evenness and body size on ecosystem functioning. Journal of Animal Ecology, 80, 1145–1154. Reiss, J., Bridle, J.R., Montoya, J.M. & Woodward, G. (2009) Emerging horizons in biodiversity and ecosystem functioning research. Trends in Ecology & Evolution, 24, 505–514. Ricotta, C. (2005) A note on functional diversity measures. Basic and Applied Ecology,

This article is protected by copyright. All rights reserved.

Accepted Article

6, 479–486. Rosseel, Y. (2012) lavaan: an R package for structural equation modeling. Journal of Statistical Software, 48, 1–36. Schindler, M.H. & Gessner, M.O. (2009) Functional leaf traits and biodiversity effects on litter decomposition in a stream. Ecology, 90, 1641–1649. Schmidt-Kloiber, A. & Hering, D. (2011) The taxa and autecology database for freshwater organisms. URL http://www.freshwaterecology.info [accessed September 2011] Srivastava, D.S. & Vellend, M. (2005) Biodiversity-ecosystem function research: is it relevant to conservation? Annual Review of Ecology, Evolution, and Systematics, 36, 267–294. Stachowicz, J.J., Graham, M., Bracken, M.E.S. & Szoboszlai, A.I. (2008) Diversity enhances cover and stability of seaweed assemblages: the role of heterogeneity and time. Ecology, 89, 3008–3019. Staelens, J., Nachtergale, L., De Schrijver, A., Vanhellemont, M., Wuyts, K. & Verheyen, K. (2011) Spatio-temporal litterfall dynamics in a 60-year-old mixed deciduous forest. Annals of Forest Science, 68, 89–98. Thompson, R. & Starzomski, B.M. (2006) What does biodiversity actually do? A review for managers and policy makers. Biodiversity and Conservation, 16, 1359– 1378. Tiegs, S.D., Peter, F.D., Robinson, C.T., Uehlinger, U. & Gessner, M.O. (2008) Leaf decomposition and invertebrate colonization responses to manipulated litter

This article is protected by copyright. All rights reserved.

Accepted Article

quantity in streams. Journal of the North American Benthological Society, 27, 321– 331. Vaughn, C.C., Spooner, D.E. & Galbraith, H.S. (2007) Context-dependent species identity effects within a functional group of filter-feeding bivalves. Ecology, 88, 1654–1662. Wissinger, S., Steinmetz, J., Alexander, J.S. & Brown, W. (2004) Larval cannibalism, time constraints, and adult fitness in caddisflies that inhabit temporary wetlands. Oecologia, 138, 39–47.

Supporting information The following Supporting Information is available for this article online:

Appendix S1. Functional traits used to calculate functional dispersion. Appendix S2. Figure 1. Decomposition rates (mean ± SE) of A. incana (white bars) and B. pendula (grey bars), expressed as rates per degree-day, in autumn and spring, and pools and riffles. Table 1. Mixed effect model of the decomposition rates per degree-day. Appendix S3. Total biomass (mg DM) of (a) stoneflies and (b) caddisflies in litter-bags containing A. incana (white bars) and B. pendula (grey bars) in two habitats (pools and riffles), and two seasons (autumn and spring) (mean ± SE). Appendix S4. Canonical Correspondence Analysis (CCA) of the detritivore species in pools and riffles, and in autumn and spring. Habitats and seasons are differentiated in this analysis using the continuous variables mean water flow velocity (m/s) (lower in

This article is protected by copyright. All rights reserved.

higher in spring).

Table 1. Habitat- and seasonal differences in water flow velocity (Flow), temperature (Temp), and standing stocks of leaf litter, pooling across the six studied streams in northern Sweden. Values denote means ± standard error. Total phosphorus concentrations were always below detection limit (< 0.04 mg/l). Total Leaf litter Season

Flow (m2/s)

Habitat

pH

Temp (°C)

Nitrogen

(g m2) (mg/l) Pool

0.03 ± 0.01

3.28 ± 0.30

23.5 ± 14.4

6.8 ±

0.37 ±

Riffle

0.33 ± 0.01

3.28 ± 0.27

5.3 ± 3.0

0.1

0.05

Pool

0.03 ±0.01

12.30 ± 0.60

0.4 ± 0.2

6.9 ±

0.32 ±

Riffle

0.42 ± 0.07

12.40 ± 0.61

0.1 ± 0.1

0.1

0.03

Autumn

Spring

Figure 1. Decomposition rates (mean ± SE) of A. incana (white bars) and B. pendula (grey bars), expressed as rates per day, in autumn and spring, and in pools and riffles.

Decomposition rates per day (kd)

Accepted Article

pools and higher in riffles), and mean water temperature (°C) (lower in autumn and

0.12 0.09 0.06 0.03 0

Pool

Riffle

Autumn

Pool

Riffle

Spring

This article is protected by copyright. All rights reserved.

DM), (c) functional dispersion and (d) species richness in two habitats (pools and riffles), and two seasons (autumn and spring) (mean ± SE). White bars represent A. incana and grey bars represent B. pendula. High functional dispersion indicates

40

(a)

(b)

28

Biomass (mg DM)

Density (ind. per litter-bag)

assemblages composed by evenly distributed dissimilar traits.

30

20

10

0

21

14

7

0

4

5

(c) Species richness

Functional Dispersion

Accepted Article

Figure 2. Detritivore (a) density (number of individuals per litter bag), (b) biomass (mg

3

2

1

0

(d)

4 3 2 1 0

Pool

Riffle

Autumn

Pool

Riffle

Spring

Pool

Riffle

Autumn

Pool

Riffle

Spring

Figure 3. Autumn structural equation models for riffles (a) and pools (b) showing the relationships between abiotic and biotic predictor variables and their effects on leaf litter decomposition. One-headed arrows represent causal pathways, while doubleheaded arrows denote correlations between variables. Grey dashed lines represent nonsignificant relationships within the model, while relationships close to significance (0.05 ≤ p ≤ 0.06) and which improve model fit are highlighted as black dashed lines. Numbers in boxes denote non-standardized coefficients. R2 is shown for each

This article is protected by copyright. All rights reserved.

Accepted Article

endogenous variable. Riffle model Chi-square = 0.04, d.f. = 2, p = 0.98. Pool model Chi-square = 3.28, d.f. = 2, p = 0.19. Autumn/Riffle

(a)

Stream identity

0.64

Litter species

0.39

0.29

Density 0.44

R2 = 0.15

Biomass

FDispersion

R2 = 0.43

R2 = 0.09

0.49

-0.16

0.49

0.19

R2 = 0.63

Autumn/Pool

(b)

Stream identity

0.45

Litter species

0.36

0.34

Density 0.38

R2 = 0.16

Biomass

FDispersion

R2 = 0.25

R2 = 0.12

0.41

0.27

0.34

R2 = 0.73

This article is protected by copyright. All rights reserved.

Accepted Article

Figure 4. Spring structural equation models for riffles (a) and pools (b) showing the relationships between abiotic and biotic predictor variables and their effects on leaf litter decomposition. One-headed arrows represent causal pathways, while doubleheaded arrows denote correlations between variables. Grey dashed lines represent nonsignificant relationships within the model. Numbers in boxes denote non-standardized coefficients. R2 is shown for each endogenous variable. Riffle model Chi-square = 6.79, d.f. = 2, p = 0.03. Pool model Chi-square = 1.29, d.f. = 2, p = 0.52. Spring/Riffle

(a)

Stream identity

Litter species

0.40

Density 0.26

R2 = 0.04

-0.54

Biomass

FDispersion

R2 = 0.19

R2 = 0.06

0.45

-0.31

R2 = 0.32

Spring/Pool

(b)

Stream identity

0.43

Litter species

0.49

0.39

Density R2 = 0.37

Biomass

FDispersion

R2 = 0.21

R2 = 0.04

0.38

0.32

-0.82

R2 = 0.67

This article is protected by copyright. All rights reserved.

When does diversity matter? Species functional diversity and ecosystem functioning across habitats and seasons in a field experiment.

Despite ample experimental evidence indicating that biodiversity might be an important driver of ecosystem processes, its role in the functioning of r...
232KB Sizes 0 Downloads 7 Views