Introduction

Plant species richness can profoundly influence the rate of primary production, nutrient cycling, and, once senesced, decay of organic matter (Tilman et al. 1996; Hooper and Vitousek 1997; Gartner and Cardon 2004; Swan and Palmer 2004). Since the majority of primary production from deciduous vegetation is delivered to the detrital pool (McNaughton et al. 1989; Cebrian 1999), loss of plant species from natural communities may have important implications for how this material becomes available to detritivores and adjacent trophic levels. The effect of litter quality on rates of decomposition can depend on the contribution of the component leaf species, and these rates have been reported to deviate considerably from predictions based on single-species decay estimates (Wardle et al. 1997; Swan and Palmer 2004; Hättenschwiler and Gasser 2005; Swan and Palmer 2006). Given continued interest in how loss of biodiversity can alter rates of ecosystem processes and services, understanding the mechanisms explaining such departures is the focus of much ecological inquiry.

Research on why decay of speciose litter proceeds non-additively has progressed much faster in terrestrial versus aquatic ecosystems (as reviewed by Gartner and Cardon 2004; but see Swan and Palmer 2004, 2006). This is despite knowledge that aquatic food webs in shaded temperate streams rely almost exclusively on riparian-derived leaf litter for energy (Wallace et al. 1997; Wallace et al. 1999; Hall et al. 2001). The majority of research has approached assessing non-additive decay of speciose litter from the bottom-up by looking at whether interspecific variation in leaf litter quality interacts synergistically/antagonistically to change decay rate (Wardle et al. 1997; Hoorens et al. 2003; Gartner and Cardon 2004; Swan and Palmer 2004; but see Swan and Palmer 2006). However, both terrestrial and aquatic ecologists know that detritivorous consumers can contribute substantially to leaf litter decay (Wallace et al. 1982; Sponseller and Benfield 2001; Hieber and Gessner 2002), and exhibit striking changes in feeding behavior and colonization dynamics in response to leaf litter nutrient content and secondary compounds (Herbst 1982; Golladay et al. 1983; Irons et al. 1988; Cuffney et al. 1990; Sweeney 1993; Swan and Palmer 2005). Evidence from terrestrial ecosystems suggests that these consumers may mediate how speciose leaf litter degrades via preferential and/or compensatory feeding, but this mechanism has not been tested in aquatic ecosystems (Hättenschwiler and Gasser 2005).

Detritivores inhabiting shaded streams rely on inputs of leaf litter to meet their energy demands (Fisher and Likens 1973; Cummins et al. 1989; Hall et al. 2001). To a large extent, the amount and species distribution of leaf litter on the streambed available to leaf-chewing consumers (shredders; sensu Cummins and Klug 1979) reflect the input rates and species composition of the deciduous tree species in the riparian zone (Johnson and Covich 1997; Swan and Palmer 2004). Known differences in leaf litter chemistry exist between leaf species (Webster and Benfield 1986; Ostrofsky 1997), allowing ecologists to relate the quality of leaf litter as a resource for detritivores to leaf species identity (Golladay et al. 1983). In general, leaf species with lower levels of essential nutrients (e.g., N) and higher concentrations of secondary and structural compounds (e.g., tannins, phenolics, lignin), such as oaks, alter shredder growth/feeding and slow overall decomposition rate (Campbell and Fuchshuber 1995; Tuchman et al. 2002; Swan and Palmer 2004). Conversely, leaf species with higher nutrient levels and lower concentrations of secondary compounds, such as maples and alders, can support higher growth rates of shredders and have been shown to exhibit faster decomposition rates in the field (Iversen 1974; Herbst 1982; Irons et al. 1988; Smock and MacGregor 1988; Swan and Palmer 2004). Given the reported high variability in resource quality that is related to leaf species identity, loss of riparian tree species could impact how in-stream shredders perform, as well as how leaf litter degrades overall (Swan and Palmer 2004, 2006).

Generalist consumers often forego subsisting on a single resource to exploit mixed resources. The advantage of a mixed diet can be to reduce or dilute the intake of toxic substances or to acquire essential nutrients from complementary resources (Freeland and Janzen 1974; Pulliam 1975; Rapport 1980; Bernays et al. 1994). Work on the relationship between consumer performance and resource heterogeneity has focused mostly on herbivores (e.g., Hagele and Rowell-Rahier 1999), omnivores (e.g., Bjorndal 1991), and predators (e.g., Evans et al. 1999), and less on detritivores (but see Swan and Palmer 2006). Leaf litter comprises the food base of many detritivores in both aquatic and terrestrial ecosystems where the majority of primary production enters the detrital pool (McNaughton et al. 1989; Cebrian 1999). Thus, leaf litter can be an inherently heterogeneous resource, requiring detritivores to exploit a mixed diet.

We have previously shown that decomposition rates of mixed-litter assemblages in the field cannot always be predicted by averaging the decomposition rates of each species individually (Swan and Palmer 2004) and, furthermore, that mixed-litter can substantially alter patterns in detritivore growth and feeding (Swan and Palmer 2006). We hypothesized that this non-additive effect was due to preferential feeding by shredders fed mixtures containing low quality detritus. For example, if a consumer is offered two resources, A and B, where A is more palatable than B, the consumer should exhibit higher feeding rate on A. However, when the consumer has no choice, its feeding rate on B may increase in order to compensate for poor resource quality. Therefore, the total mass loss rate of the mixture of A and B should be slower compared with the average of A and B, individually. This shift in feeding rate on a mixed-diet could therefore explain why decay rate of speciose litter can be slow compared to what would be predicted based on single species diets alone (Swan and Palmer 2006). To address this experimentally, we asked: (1) do shredders exhibit preferential feeding on high quality resources when alternative resources are available, and (2) if preferential feeding occurs, is this consistent with non-additive patterns of leaf decomposition of litter mixtures?

Materials and methods

The isopod crustacean Caecidotea communis is a common leaf-shredding detritivore that inhabits small streams across the eastern half of North America (Williams 1972). In early January 2003, gravid female isopods were collected from a spring-fed tributary of the Middle Patuxent River, a 3rd-order Piedmont stream located west of the Chesapeake Bay in Howard County, Maryland, USA (39°15′N, 76°55′E; elevation 100 m). These individuals were kept in the laboratory in aerated stream water and allowed to feed on leaf material collected at the time of sampling. Shortly after hatching, juveniles were collected and the adult isopods separated until enough juveniles had been collected for the experiment (a period of ∼10 days).

Invertebrates and leaf treatments were housed in 500-ml plastic chambers (75 mm in height × 112 mm in diameter). Chamber lids were perforated with two holes (6 mm in diameter), one to accommodate air supply via standard aquarium airline and the other to relieve air pressure. Each chamber was filled to ∼1 cm depth (capacity 100 ml) with air-dried stream sediment (size-distribution, by volume: 4.5% <0.25 mm, 30.6% 0.25–0.50 mm, 39.4% 0.50–1.0 mm, 20.6% 1.0–2.0 mm, 4.8% >2.0 mm) collected from the Middle Patuxent at the time the isopods were sampled.

The study began on 15 January 2003. The design was a two-way factorial design [3 shredder treatments (0, 3, 6 isopods), 8 litter treatments (Table 1)] with a total of n=10 chambers per litter × shredder combination. Isopod density was chosen to approximate the range in isopod densities occurring in the field (975 m−2, SE=675, n=4). Leaf litter of the common riparian trees boxelder (Acer negundo L., C:N=21.1), American sycamore (Platanus occidentalis L., C:N=47.5), black willow (Salix nigra Marsh, C:N=47.2) and slippery elm (Ulmus rubra Muhl., C:N=41.1) were collected during the autumn of 2002 (just after leaf abscission) from the riparian zone of the Middle Patuxent River, and kept in plastic bins in the laboratory until needed the following January. These species represent the dominant species in local riparian habitats and exhibit a realistic range in C:N content, a commonly used index of palatability (analyzed with an automated CHN analyzer, University of Maryland Soils Testing Laboratory, College Park, Maryland, USA). Eight litter treatments were established: four single-species treatments and four three-species treatments (hereafter, “mixed” treatments; Table 1). Each chamber received approximately 300 mg of leaf material (fresh weight) as coarsely broken fragments (Table 1). We attempted to hold mass of the leaf treatments constant; thus, mixed-species treatments were comprised of ∼100 mg of each species.

Table 1 Summary of leaf litter treatments
Full size table

The chambers were randomly assigned locations on a laboratory workbench, leaf litter was added and then the chambers were filled with ∼300 ml of deionized water. Each chamber was inoculated with 5 ml of stream water that had been collected from the Middle Patuxent, aerated since the time of sampling, and filtered twice at 45 μm to remove invertebrates, while still adding microbes. Chambers were arranged in 10 groups of 24 chambers, with each group constituting a sample date. Each group of 24 chambers was supplied with air from a standard aquarium air pump, with air-flow controlled by a separate air valve for each chamber. Water temperature for the duration of the experiment averaged 23.2°C (SD=0.6°C, n=7; range=6.6°C). Light regime was set to a long day cycle of 16 h:8 h light:dark with standard fluorescent lighting. Water was added throughout the study to correct for evaporation.

After 8 days of pre-conditioning and leaching, a complete water change was made and newly hatched isopods (0, 3, or 6) were added to chambers designated for shredders using a pipette (23 January 2003 = day 0). Isopod juveniles averaged 0.039 mg dry mass {SE=0.0020, n=70; ln(mg dry mass)=−4.241+2.424 [ln(length, mm)], r 2=0.97, n=22}. Estimates of leaf ash-free dry mass (AFDM) available at day 0, after leaching, were made by conducting a companion study. Each leaf treatment (Table 1) was allocated to containers (n=5 per treatment, 40 containers total) in the same manner as the main study. After 8 days of leaching/conditioning, the leaf litter was removed, dried to a constant weight at 60°C, combusted for 90 min at 550°C then re-weighed to determine AFDM. The fraction of AFDM estimated from this parallel study was used to estimate leaf AFDM available on day 0 when the shredders were added.

Samples were taken on days 4, 10, 15, 22, 29, 36, 43, 51, 59 and 75. Sampling involved first removing the isopods and preserving them in 70% ethanol, including any newly hatched individuals due to reproduction during the study. Individual isopod dry mass was estimated for each sample using the allometric equation above; length measurements were made with a stereomicroscope fitted with an ocular micrometer. If a chamber assigned to an isopod treatment was found to have none at sampling, that replicate was dropped from the study (this was the case in 22 of the 240 chambers). Leaf litter remaining was removed and sorted by leaf species into separate, pre-weighed tins. The leaves were then dried at 60°C, combusted for 90 min at 550°C, and re-weighed to determine AFDM remaining (Benfield 1996).

Data analysis

Leaf decomposition rate

Due to reproduction during the experiment, the distinction between the three and six individual isopod treatments was no longer clear, and thus all treatments with isopods were combined into a “shredder present” treatment, making the final analysis a two-way factorial [(2 shredder treatments (present, absent), 8 litter treatments]. Decomposition rate was assessed by estimating k in the exponential decay model W t =W o e−kt, where W o is initial mass and W t is the mass of the litter at time t (Petersen and Cummins 1974). Mass at day t (ln AFDMt; g) was analyzed as a function of day, shredder treatment, litter treatment and all interactions. The estimate of the interaction term (day × leaf treatment × shredder treatment) was taken as the k-value for each shredder × leaf treatment combination.

Output from the ANCOVA was used to make three sets of comparisons within each shredder treatment: (1) comparisons between single- and mixed-species treatments, (2) comparisons among single-species treatments, and (3) comparisons between observed decay rate of each mixed-treatment and the predicted decay rate of each mixture (i.e., the average of the decay rates of each leaf species separately). Significance for comparisons among single-species leaf treatments were adjusted using Hommel’s correction for multiple-comparisons (Westfall et al. 1999). For the third set of comparisons, a linear contrast was used to test for differences between the average decay rate across single-species litter treatments and that of the corresponding mixed treatment. For example, a significant difference between the mean of the decay rates estimated for boxelder (B), slippery elm (E), and black willow (W) treatments and that of the mixed treatment of these three species (i.e., B + E + W) would indicate that decay rate on the mixed treatment was non-additive. To maintain Type I error rate and orthogonality, a multiple-linear contrast procedure was first conducted for all four mixed-treatments and, upon obtaining a significant F-test, individual contrasts were then calculated (Sokal and Rohlf 1981).

Shredder feeding rate

Feeding rate was calculated as:

$$ {\text{Feeding}}\;{\text{Rate}} = \frac{{{\text{Leaf}}\;{\text{Mass}}\;{\text{Consumed}}}} {{{\text{Shredder}}\;{\text{Mass}}\;{\text{Gained}} \cdot {\text{days}}}}, $$

where leaf mass consumed is the difference in leaf AFDM at sampling and initial leaf AFDM in the treatments containing the isopods. Shredder mass gained is the difference between total isopod dry mass at sampling and initial isopod dry mass, and days are the number of days since the study began. To ascertain the presence of preferential feeding, analysis of feeding rates of each leaf species—alone, and in mixture—were carried out. To account for the altered availability of leaf species between single and mixed-species diets (i.e., each leaf species was ∼3 times more available in the single-species diet versus any of the three mixed diets), feeding rates in single-species treatments were corrected. On each sample date, the average AFDM of a particular leaf species in mixture across all three shredder treatments (i.e., 0, 3, 6 isopods) was divided by the average AFDM of that same litter species alone. For each species, this correction factor averaged: boxelder: 0.3392, slippery elm: 0.3321, sycamore: 0.3437, and Willow: 0.3452. For example, 33.92% of boxelder AFDM present in the single-species treatment was available initially in each of the three mixtures in which boxelder was present. Then, for each species, a single ANOVA was performed, and comparisons done between feeding rate on the single-species treatment and that species within each of the three mixtures. If the feeding rate was higher in the mixed-species treatment compared to feeding on that species individually, this was taken as evidence for preferential feeding.

All analyses were performed using SAS version 9.1 and evaluated at α=0.05. Denominator degrees of freedom were adjusted using the Satterthwaite approximation.

Results

Leaf decomposition rate

Strong interactions between leaf treatment and the presence of the isopod were evident with respect to total leaf decomposition rate (significant day × leaf × shredder treatment effect; Table 2). Comparing single- and mixed-litter treatments, no difference was evident in the absence of the shredder, but decomposition rate of mix-litter treatments was significantly slower than single-species litter (Fig. 1a). Mixtures of three leaf species were on average 60% slower to lose mass than single-species treatments. This was accentuated by the strong differences seen between single-species leaf treatments in the presence of the isopod. In the presence of the shredder, boxelder was the fastest decomposer, degrading nearly 3.7 × faster than elm, sycamore and willow combined (Fig. 1b). This was not the case in single-species treatments lacking the isopod where decomposition rates quite similar (Fig. 1c).

Table 2 Statistical results for leaf decomposition and feeding rate
Full size table
Fig. 1
figure 1

Total leaf decomposition rate as a function of a number of leaf species, b leaf treatments with shredders present, and c leaf treatments with shredders absent. For the top panel, bars with the same letter are not significantly different. Predicted means (hatched bars) are the average of the single-species estimates for each mixture. All bars are the mean+1SE. Connected bars for single-species leaf treatments indicate no significant difference, otherwise observed decomposition rate differs (P<0.05). For mixtures, comparisons are limited to observed and predicted for each mixture; connected bars indicate no non-additive mixing effect (P > 0.05). Species abbreviations are as follows: B boxelder, E slippery elm, S American sycamore and W black willow

Full size image

Three of the four mixed-treatments decomposed significantly slower than predicted when the isopod was present (F 4,120=4.45, P=0.0022), but were completely additive in the absence of the shredder (F 4,115=0.39, P=0.8170). Focusing on the mixtures containing the shredder (Fig. 1b), the B + E + S, B + E + W and B + S + W mixtures were significantly slower than predicted, whereas decomposition of the mixture that lacked the fastest decomposer, E + S + W, was found to be additive. The magnitude of the reduction in decomposition rate due to mixing averaged k=0.0099, which is greater than the decomposition rate of willow alone (k=0.0092), suggesting that the effect of mixing on slowing decomposition rate was substantial.

Feeding rate

With respect to whether this shredder exhibits preferential feeding, analysis of species-specific feeding rates revealed that Caecidotea communis fed equally among three of the four leaf species used in this study. Feeding rates on boxelder (ANOVA F 3,61=0.19), elm (ANOVA F 3,57=0.96) and willow (ANOVA F 3,52=2.11) were not significantly different when comparing feeding on the leaf species alone, or when it was in mixture (all P values > 0.05; Fig. 2). However, feeding rate on sycamore alone was significantly faster than on this same species in mixture with boxelder and willow (ANOVA F 3,51=3.93, P<0.05; shaded bars Fig. 2c). Isopods consumed sycamore 1.88 times faster when it had no other leaf species to feed upon compared to the B + S + W treatment. However, this was the only mixture demonstrating a significant pattern; feeding rates on sycamore alone compared to sycamore in either the B + E + S or E + S + W treatment did not differ.

Fig. 2
figure 2

Shredder feeding rate (mg per mg per day) on each leaf species individually and for that species in mixture, a boxelder, b slippery elm, c sycamore, d black willow. Shaded bars in (c) highlight the significant difference between feeding rate of sycamore alone versus the mixture treatment B + S + W. Bars are the back-transformed means and error bars are the back-transformed 95% confidence interval. All mixtures are indicated in each plot for clarity. Bars with the same letter are not significantly different (P > 0.05); comparisons are made for each species separately. Note differences in scale of Y-axes. Species abbreviations are as in Fig. 1

Full size image

Discussion

Plant species richness can alter ecosystem processes in many ways, including decay of senesced leaf tissue (Gartner and Cardon 2004; Swan and Palmer 2004, 2006). In temperate terrestrial and aquatic ecosystems, leaf litter enters the detrital pool in a large pulse during autumnal leaf fall and then decomposes, both releasing energy to the food web and recycling nutrients back into the soil or downstream. Extensive work in terrestrial ecosystems and, to a lesser extent, aquatic ecosystems has demonstrated that interspecific variation in leaf litter chemistry can lead to non-additivity in the decay of speciose leaf litter (Gartner and Cardon 2004; Swan and Palmer 2004, 2006). Here, we tested an alternative mechanism for how multi-species leaf litter should degrade by incorporating foraging behavior of detritivorous consumers. We found that preferential feeding by an aquatic detritivore, the isopod Caecidotea communis, can change the decomposition of litter mixtures when compared to the average decay rate of leaf species in the mixture estimated alone.

Shaded stream ecosystems harbor a diverse assemblage of detritivores that rely on terrestrially-borne detritus (Wallace et al. 1997, 1999). Leaf-shredding invertebrates occupy a critical link between terrestrial energy sources (i.e., leaf litter) and adjacent trophic levels by transforming this detritus into biomass thereby creating food for predators and producing particles as a by-product of feeding for downstream filter-feeding consumers (Shepard and Minshall 1984; Cuffney et al. 1990; Dieterich et al. 1997). Given the extent to which quality of leaf litter can alter both shredder growth and feeding rates (Smock and MacGregor 1988; Sweeney 1993; Swan and Palmer 2006), the goal of this study was to determine if shredders alter feeding rates when provided mixed-species leaf litter, and if any such effects explain patterns in mixed-species decomposition rate. Our results revealed that feeding rate by the consumer was elevated on American sycamore, a lesser quality resource, when the animal was not given a choice. When the consumer was given the choice to feed on more palatable leaf species (e.g., boxelder), feeding rate on American sycamore slowed, thus slowing the rate of multi-species litter as a whole. However, this phenomenon was not revealed for all mixtures studied, suggesting other factors need to be considered in future studies. Taken together, this work suggests that consumer foraging may be critical to understanding how diverse leaf litter decomposes in aquatic ecosystems.

What then could be causing the strong, non-additive slowing of total decomposition rates of mixtures in cases where preferential feeding was not evident (i.e., mixtures B + E + S, B + E + W)? Effects of mixing resources, in addition to any interaction with the presence of the shredder, were not assessed with respect to the microbial community. If we chose to correct for microbial degradation, however, we assume that there is no interaction between shredder feeding activity and the degradative ability of the microbes. This may or may not be the case. A recent paper by Ribblet et al. (2005) demonstrates that bacterivory by protists can alter decay rate of leaf litter in aquatic microcosms. Shredders, when consuming leaf material, consume both leaf and microbial biomass. In doing so, they both reduce microbial biomass (via direct consumption) on the substrate and open up fresh leaf material to be colonized by bacteria and fungi. Both activities could alter the fraction of decay rate contributed by the microbial community, but not predictably. Complicating this assumption is the fact that leaf quality varies, and thus feeding activity could induce both positive and negative (or no) changes in activity by the microbial community.

Bacteria and, to a larger extent, fungal communities can contribute to leaf decomposition, but this contribution is estimated to be small compared to loss due to invertebrates and the physical environment (Webster and Benfield 1986; Hieber and Gessner 2002). This occurs via direct digestion of leaf tissue with exo-enzymes, which soften the leaf tissue for shredder consumption (Bärlocher 1985; Webster and Benfield 1986). In the present study, however, differences between single-species treatments and between observed and predicted decomposition rates of mixtures without the shredder did not mirror the pattern seen when the isopod was present. Lack of such consistency suggests that, among mixtures lacking evidence of preferential feeding, there may have been an interaction between shredder presence and microbial degradative ability. While we know of no study to date documenting any such interaction between arthropods and microbes, evidence does exist that bacteria-feeding protists alter the degradative ability of the bacteria, enhancing decomposition rate (Ribblet et al. 2005). Therefore, exploring the interaction between resource quality, shredder presence and microbial activity (e.g., production, enzyme activity) could be vital to completely understanding the complex role mixing leaf species can have on total leaf decomposition rate.

We acknowledge the evidence we present here is from a laboratory study, and necessarily so given the difficulty in isolating detritivore impact on decay in the field (but see Wallace et al. 1982). Leaf decomposition in streams is a function of a number of factors besides detritivore feeding; indeed, leaves do decompose in aquatic systems where shredders are rare or absent (Sponseller and Benfield 2001). Mechanical forces such as substrate abrasion and flow are responsible for a substantial fraction of mass loss (Paul and Meyer 2001; Hieber and Gessner 2002). Also, burial can slow decay by creating anoxic conditions that are less hospitable to invertebrate consumers (Tillman et al. 2003). These factors, combined with water chemistry variables like nutrient concentration and pH, may be quite important to how multi-species litter decays (Grattan and Suberkropp 2001; Dangles and Chauvet 2003). However, in shaded headwater streams where detritivores are abundant, we expect a substantial contribution of feeding to mass loss, even in the context of the aforementioned environmental conditions (Hieber and Gessner 2002). More work under these circumstances is required, and will be essential to completely understanding the mechanisms by which speciose leaf litter decays in aquatic ecosystems.

One impetus for this work was the growing realization that the loss of deciduous riparian tree species is a real phenomenon (Smock and MacGregor 1988; Sweeney 1993; Snyder et al. 2002; Ellison et al. 2005). Since stream ecosystems rely on the delivery of leaf material to the stream during annual leaf-fall, such losses could impact in-stream detritivores (Wallace et al. 1997, 1999; Swan and Palmer 2004, 2005, 2006). The strong, non-additive effects of mixing leaf species suggests that consumer-resource interactions are complicated, and can change if riparian tree species richness is reduced (Swan and Palmer 2006). Evidence from this study supports previous findings that detritivore feeding rates will change, and the subsequent effects on both organic matter dynamics (i.e., leaf decomposition) and detritivore growth can be substantial (Swan and Palmer 2006). Further work is required to broaden these results to other consumers from streams comprised of different deciduous riparian communities.

Decomposition of organic matter is a key ecosystem process, defining the rate at which detritus becomes available to consumers, and is recycled into nutrient pools. The results from this study suggest that the loss of species richness of deciduous vegetation can lead to alterations in the magnitude of leaf decomposition in aquatic ecosystems. The role leaf shredders play in the decomposition process can be further extended to how they mediate the role played by species diversity of leaf litter. By showing that Caecidotea communis exhibits preferential feeding when a diverse diet is present, this work suggests loss of particularly refractory species (e.g., American sycamore) could impact how natural assemblages of leaves degrade in the stream. At elevated levels of leaf species richness, if refractory leaf material is present, it may decompose slower than if alone, increasing the retentiveness of that material in the system. Therefore, preferential feeding by detritivores could be a critical ecological mechanism mediating how species richness of riparian deciduous vegetation can alter in-stream ecosystem processes.