Input of litter in deforested and forested areas of a tropical

Riparian vegetation is the main source of leaves and the main energy source for low order streams. Therefore, changes in its composition affect the structure and processes in streams. We studied the contribution of the riparian vegetation by analysing the annual input of litter in deforested and forested areas of a tropical stream. The lateral, vertical (aerial) and horizontal (drift) litter inputs were analysed separately. The lateral input differed significantly between the two areas and included mostly fallen dry leaves. The vertical input, represented mainly by fallen dry leaves, occurred only in the forested area. The drift transport of litter was not significantly different between the deforested and forested areas and the input was composed mostly by CPOM. The removal of the native forest was clearly reflected in the low contribution of leaf litter in the deforested area.


A TROPICAL HEADSTREAM
Input of litter in deforested and forested areas of a tropical headstream ABSTRACT Input of litter in deforested and forested areas of a tropical headstream. Riparian vegetation is the main source of leaves and the main energy source for low order streams.
Therefore, changes in its composition affect the structure and processes in streams. We studied the contribution of the riparian vegetation by analysing the annual input of litter in deforested and forested areas of a tropical stream. The lateral, vertical (aerial) and horizontal (drift) litter inputs were analyzed separately. The lateral input differed significantly between the two areas and included mostly fallen dry leaves. The vertical input, represented mainly by fallen dry leaves, occurred only in the forested area. The drift transport of litter was not significantly different between the deforested and forested areas and the input was composed mostly by CPOM. The removal of the native forest was clearly reflected in the low contribution of leaf litter in the deforested area.

Introduction
Every year streams receive a large amount of litter dry mass per square meter . A large part of these detritus input consists of leaves from riparian vegetation . Headwater streams are particularly influenced by the riparian vegetation which reduces autotrophic production by shading and supplies energy in the form of vegetal matter of allochthonous origin .
A qualitative and quantitative change in the riparian forest affects the litter input to the streams and thus can modify the structure of all biotic community (Afonso et al., 2000;. In Brazil, as well as in many other countries, the degradation of riparian vegetation results from the disordered expansion of agricultural borders (Rodrigues & Gandolfi, 2001;Heartsill-Scalley & Aide, 2003), contributing to the formation of immense open areas, characterized by grassy and herbaceous vegetation.
The type of vegetation cover and rainfall can also determine conditions of autotrophy or heterotrophy in streams, having a strong effect upon the stream food web structure and dynamics, as showed by  for some tropical streams.
Those changes in land cover can also modify the physical conditions of streams (Heartsill-Scalley & Aide, 2003), with a forest landscape certainly contributing with more biomass to the stream than herbaceous vegetation (Afonso et al., 2000;Heartsill-Scalley & Aide, 2003). However, it is important to determine the actual contribution of the litter input for different conditions of riparian vegetation .
Leaf litter enters streams mainly in a large burst during the period of leaf abscission, and can be either trapped in the reach, or transported downstream (Elosegi, 2005). The entrance ways of this allochthonous matter in the stream include transport by drift from upstream, by direct or vertical input, and by lateral input of material deposited on the forest floor and mobilized by wind or other agents . Thus, the regional climate, the characteristics of the stream and of the semi-deciduous forest could be the main factors that determine this seasonal leaf-litter pattern .
The aim of the present study was to assess and compare the annual input and transport of litter in two different areas of a tropical stream: one area shaded by the presence of a dense gallery forest and other, located downstream the first, deforested and with scattered herbaceous vegetation.

Material and methods
The study was performed at Ribeirão da Quinta stream (23º06'47"S, 48º29'46"W), located at the municipality of Itatinga, São Paulo State, southeast Brazil, at an elevation of 743m a.s.l. This is a third order stream, located on a cattle raising farm, distant from urban areas.
Two areas of this stream with approximately 120 m 2 were chosen for the study. The first, called "Forested area", is shaded by a well-preserved gallery forest. The riparian trees Three types of traps were used in both areas to analyze the litter input: 1) aerial or vertical-input traps, fixed to a metallic frame hanging from nearby trees or stakes ( Figure   1a); 2) lateral-input traps, installed in the right and left banks (Figure 1b and 1c, respectively), fixed to a metallic frame and held by two stakes; 3) drift or horizontal-input trap, fixed to a metallic frame and with two metal rods used to secure the trap in the bed ( Figure 1d). All traps were constructed tying a 1-mm double mesh to a rectangular metallic frame (0.02 m 2 ; Figure 1e) and eight replicates were used in all cases.
The traps were placed in the two areas bimonthly, from May 2004 to June 2005.
Each month the traps stayed in the stream for a 48 hours period. After that, the litter captured in the traps was quantified as dry mass separately into three categories: fallen green leaf, fallen dry leaf and coarse particulate organic matter (CPOM, composed mainly by plant fragments not identifiable). The vertical and horizontal input dry mass were expressed in mg.m -2 .day -1 , considering the trap surface area, and the lateral input dry mass was expressed in mg.m -1 .day -1 , considering the trap length . The During the experimental period the following variables were analyzed: rainfall (obtained from the meteorological station of a farm located about 2 km from the study location), water temperature, current velocity and discharge (measured as described in Leopoldo & Souza, 1979).
Comparison of dry mass among treatments (areas, months and traps) were done by one-way ANOVA, followed by comparisons with the Tukey HSD test. The relationship between dry mass and environmental parameters was tested by the Spearman Rank Correlation.

Results
The lateral input of litter differed significantly between the two areas (F 1,110 = 19.187, p < 0.001 for the left bank and F 1,110 = 15.594, p < 0.001 for the right bank), with the higher value of dry mass in the forested area (Table I) The vertical-input in the stream occurred only in the forested area (Table I) Figure 4).
The horizontal-input of litter was not significantly different between the two areas (F 1,110 = 1.658, p = 0.20; Table I), occurred at all months, and was composed mostly of CPOM ( Figure 5). In the deforested area, the month of high horizontal input was June (179,814.5 mg.m -2 .day -1 ) and in the forested area it was July (157,395 mg.m -2 .day -1 ).
The months of greatest discharge (May 2004 and January 2005) also represented the months of high values of current and rainfall, with the current velocity twice faster in the deforested area ( Figure 6).
The correlation between environmental parameters and drift transport of organic matter was different between areas (Table II). Rainfall incidence showed no correlation with litter transport for both areas. However, discharge was positively correlated to drift transport in the deforested area, while current velocity was negatively correlated in the forested area.

Discussion
In temperate deciduous forests, autumn litter inputs may be as high as 73% of annual amounts (e.g. Abelho & Graça, 1998). Litterfall in tropical forests may be either seasonal, especially when a marked dry season occurs, or non-synchronous, with litter entering at a relatively constant rate over the entire year Gonçalves et al., 2006).
The rainfall registered in the Ribeirão da Quinta stream showed two well-defined seasons, with the highest values of input during the end of the dry season (September), coincident with the lowest value of discharge. In the same region, Uieda & Kikuchi (1995) and Afonso et al. (2000) found that the litterfall peak occurred in the early wet season (September and October), while Henry et al. (1994) observed it in the late dry season (August). According to Pozo et al. (1997), the phenology of litter input and discharge regime may be responsible for the temporal litterfall variation.
The significant correlation between water discharge and drift transport of litterfall in the deforested area for pasture utilization may be related to its localization, nearby the forested area. It is expected that with the increase of water discharge a great quantity of leaf litter and accumulated benthic organic matter from the gallery forest may be transported downstream. The negative correlation of litter transport with current velocity in the forested area corroborates the previous argument.
Other authors (Canhoto & Graça, 1998;Afonso & Henry, 2002;Pretty & Dobson, 2004) also found that during rainfall events the leaf retention also decreases greatly. In the same way, when retention structures are submerged or carried downstream, it will result in greater loss of sediments (Díez et al., 2000;Afonso & Henry, 2002). Therefore, when litter input peaks coincide with low flows, CPOM tends to accumulate in the stream bed . On the contrary, if litter inputs coincide with high flows, transport processes are favorable.
The input of litter from the lateral banks was about ten times larger in the forested than in the deforested area, result similar to the one found by Henry et al. (1994) and Afonso et al. (2000), both working in a stream of the same basin. The highest lateral input of litter at the end of the dry season (July) and the beginning of the wet season (September) probably was influenced by two climatic agents: the scouring of this material by the wind during drought and by the rain during flooding. Moreover, the first flood after a long dry period may scour a much larger amount of litter than similar floods later in the season . It is probable that the regional climate associated with the characteristics of semi-deciduous forest could be the main factor that determined this seasonal leaf-litter pattern (e.g. . The transport of litter downstream is a very important parameter for analysis of the energy distribution along the stream, mainly for areas with altered riparian vegetation. In the short stretch of the Ribeirão da Quinta stream studied, one may assume that drift inputs equaled drift outputs, as argued by . This presupposition is further justified since lateral input occurred rarely and aerial input did not occur in the deforested area, while the drift transport was similar between both areas. The amount of litter transported by drift surpassed the other inputs and the greatest part of the former was made up of CPOM, probably originated from organic benthic stocks. Iversen et al. (1982) also reported a large contribution of CPOM (92%) in the output and of leaves in the inputs (71%). Clearly, the origin and quality of this organic matter is of great importance to the understanding of stream dynamics.
The difference in the quality of litter as a function of input origin was very evident.
First, a large part of the litter entering a stream channel consists of gallery forest leaves, particularly fallen dry ones. Second, a part of these leaves are directly transported downstream, deposited in the stream channel and on the stream banks. Third, the mobilization and distribution of stored litter depend on climatic agents, such as rainfall and wind. A large portion of this material imported upstream consists of unidentifiable fragments, originated probably from the breakdown process (e.g. .
Such patterns are very characteristic, as much in tropical (Henry et al., 1994;Afonso et al., 2000;Gonçalves et al., 2006) as in tempered streams (e.g. . Given that riparian vegetation is the main source of leaves and, therefore the main energy source , changes in its composition affect structure and process within streams (Heartsill-Scalley & Aide, 2003). This alteration was clearly shown by the low contribution of leaf litter in the deforested area. In a way to maximize conservation, to establish connectivity and to increase the minimum riparian forest width, priority should be given to the stream ecosystem (Heartsill-Scalley & Aide, 2003).

Seasonal leaf mass loss in two contrasting stretches of a tropical headstream ABSTRACT
The seasonal leaf mass loss of two leaf species, one herbaceous (Picreus decumbens) and other arboreal (Cabralea canjerana), was analyzed in a forested and a deforested area of a tropical stream, during the dry (June-July 2004) and wet (November-December 2004) seasons. Leaf bags containing 5.00 ± 0.05 g of fresh and pre-conditioned leaves were installed in the stream bottom in each area and season and removed after 3 and 13 days of immersion. Our results showed an interaction of factors determining the leaf mass loss, which apparently is dependent on several factors, like plant species, season, water temperature, and current. The tree common in the gallery forest area, Cabralea canjerana, presented a leaf loss dry mass more influenced by the differences in temperature between the dry and wet seasons, with leaf loss higher during the wet season. Otherwise, for the herbaceous plant common in the deforested area, Pycreus decumbens, the leaf loss dry mass was more influenced by the differences in light incidence and current found when compared the forested and deforested areas, with leaf loss higher in the forested area. In the beginning of the decomposition process (3 rd day) the significant differences found on dry mass loss emphasized the differences between plant species and the effect of microorganisms accelerating the process on the previously conditioned leaves. Later (13 th day), the spatial and seasonal differences also played an important role in the decomposition process, through the differences on light incidence, water current and temperature found when compared the forested and deforested areas and the wet and dry seasons. Apparently, the fast leaf breakdown found in this tropical stream may have been caused by differences in leaf quality related to the initial chemical composition and texture of the leaves, which can also improve the action of microbes and invertebrates.

INTRODUCTION
Litter breakdown is traditionally analyzed as a continuous process with three overlap stages: leaching, conditioning and fragmentation . The functional unit comprised by shedders, microbes, and terrestrial litter plays a key role in breaking down CPOM of terrestrial origin into FPOM that becomes distributed in the aquatic system (Cummins et al., 1989;. The leaching of soluble compounds is generally rapid and may account for a substantial decrease in initial litter mass Bärlocher, 2005a).
But on a global scale, one may expect considerable variation in leaching related to leaf species composition, climate , and a variety of other factors such as stream temperature, current and litter quality . When leaves enter the streams, they are rapidly colonized by microorganisms  thus initiating the leaf conditioning process. In this process the microbial assemblages are important not only by enhance breakdown directly by macerating leaves, metabolizing the leaf tissue, and incorporating into secondary production, but also indirectly increasing the palatability of detritus to invertebrate shredders . Shredders have been shown to feed preferentially on conditioned leaves (e.g. . By feeding on leaves, shredders incorporate some nutrients in secondary production, accelerate leaf fragmentation, and produce abundant FPOM, which are ecologically important for populations of collectors inhabiting lower reaches of streams . The abundant literature on leaf litter breakdown in temperate streams contrasts with the scarce information available from tropical streams Mathuriau and Chauvet, 2002;. In a stream located in the southeast of Brazil we intended to evaluate the leaf mass loss through a manipulative experiment in which the variables analyzed were: (i) type of plant, by using two common riparian species, an invader herbaceous plant (Picreus decumbens) and a native arboreal plant (Cabralea canjerana), (ii) importance of microorganisms colonization, by using leaves previously colonized and not, (iii) characteristics of the riparian vegetation, by conducting the experiment in a forested and a deforested stretch of the same stream, and (iv) time of the year, by investigating the leaf mass loss in a wet and a dry period. We hypothesized that different leaf species, conditioned or not, incubated in contrasting areas and periods of the year with differences in temperature and rainfall can modify breakage time of the leaf litter imported to the stream.

Study site
The study was carried out in the Ribeirão da Quinta (23º06'47"S, 48º29'46"W), located in the municipality of Itatinga, State of São Paulo, southeastern Brazil, at an elevation of 743m a.s.l. This is a third order stream, located on a cattle ranch, with the riparian vegetation partially preserved.
The study was conducted in two consecutive stretches of this stream. The stretch shaded by a well-preserved gallery forest was called "forested area" and the stretch located about 300m downstream the forested area, with the marginal vegetation composed only by herbaceous vegetation, was called "deforested area". The composition of the marginal vegetation at the two areas was presented in Table I.

Experimental setup
Two types of freshly leaves were used in the experiment, Cabralea canjerana a common tree in the gallery forest area and Pycreus decumbens an herbaceous plant common in the deforested area. The amount of 5.00 ± 0.05g of freshly collected leaves was enclosed into a mesh bag (20 x 15 cm, with 10 mm mesh opening). Forty (twenty of each plant) of these mesh bags were individually enclosed in another nylon bag (25 x 20 cm, with 0.25mm mesh opening) to prevent macroinvertebrates colonization and placed in the stream bottom for 14 days of conditioning (colonization by microorganisms), attained to fishing weights to keep the bags below the water column. After this conditioning period, the external nylon bag (0.25mm mesh openings) was carefully removed and the same amount of fresh leaf mesh bags (twenty unconditioned of each plant) was installed. This was considered the day 0 of the experiment (representing 14 days of immersion for half of the samples, called conditioned samples, and 0 day of immersion for the other half, called unconditioned samples).
Twenty replicates of each plant species, ten conditioned and ten unconditioned, were randomly removed after 3 and after 13 days of experiment. Each leaf pack sample was placed into a plastic bag and transported to the laboratory in an icebox. The bags were then rinsed with distilled water and each leaf was carefully washed to remove the macroinvertebrates colonizers. The leaf material was dried at 70ºC for 48h and then weighed.
The manipulative experiment was performed during the dry (June-July 2004) and wet (November-December 2004) seasons. The same experimental design was repeated in both areas and seasons. The water characteristics measured during the experimental period were: conductivity (OAKTON-TDSTestr3 TM ), light incidence (PANLUX Electronic -GOSSEN), current velocity and discharge (Leopoldo and Souza, 1979), pH and temperature (Horiba model D-14).

Leaf mass loss and data analysis
The initial leaf dry mass was obtained by linear regression of data from forty leaf samples of each plant species, C. canjerana (R 2 = 0.9939) and P. decumbens (R 2 = 0.9943) during the dry season and C. canjerana (R 2 = 0.9934) and P. decumbens (R 2 = 0.9909) during the wet season. The percentage of leaf dry mass remaining was computed by the difference between initial and final dry mass, multiplied by 100.
The percentage of leaf dry mass remaining was log-transformed and these data were submitted to an analysis of variance (One-Way ANOVA) to test differences between seasons, stream area, plant species and leaf treatment (conditioned and unconditioned; 3 and 13 days). The differences between slopes (comparison of leaf mass loss over time) were tested with an analysis of covariance (ANCOVA), followed by Tukey´s test (Zar, 1999). The percentage of leaf dry mass remaining and the environmental parameters were analyzed by the Principal Component's analysis (PCA) to determine the relative importance of the environmental parameters in the leaf decomposition (Statistica 5.1, 1996).

RESULTS
Seasonal differences were observed for temperature, discharge and light incidence data, with high values measured during the wet season (Table II). Spatial differences were more characteristic for current velocity and light incidence (Table II). In the deforested area the current velocity was twice higher than in the forested area. The light incidence was higher in the deforested than in the forested area, forty-eight times during the dry season and thirty times during the wet season.
The analysis of leaf mass loss ( Figure 1) showed a pronounced difference between unconditioned and conditioned leaves for both plant species, with less dry mass remaining for the conditioned leaves. The variance analyses confirmed these differences and showed also a difference between plant species and over time (Table III). After 3 days, the dry mass loss was significantly different among plant species and leaf treatments (in general, higher for P. decumbens and for conditioned leaves), but not significant for areas and seasons. After 13 days all comparisons showed significant results, with more leaf mass loss for conditioned leaves of P. decumbens and at the wet season experiment. The leaf mass loss on the last experimental day of the wet season was significantly higher than that of the dry season ( Figure 1).
The analysis of slope comparison (ANCOVA) showed significant difference for conditioned leaves of P. decumbens incubated in the forested area at both seasons (Table   IV).
The analysis of the relative importance of the environmental parameters in the leaf decomposition showed that the two Principal Component axes explained a high percentage of data variance (76% of cumulative variance; Table V). The variables that showed significant positive contribution to the total variance of the first axis were water current, light incidence and experimental area. These variables determined the position of P.
decumbens, with a high dry mass remaining in the deforested area with high current and light incidence ( Figure 2). For the second axis, temperature and season showed positive significant contribution to the total variance (Table V), determining the position of C.
canjerana, with a high dry mass remaining related to the dry season with low temperature ( Figure 2).

DISCUSSION
The leaf mass loss of P. decumbens and C. canjerana was fast when compared to a wide variety of plant species studied by Petersen and Cummins (1974) in a temperate stream. The two studied plants reached about 50% of mass loss between 13 (for unconditioned) and 17 (for conditioned) days of immersion, compared to 40 days observed by Petersen and Cummins (1974). Other works in tropical streams also showed fast leaf mass loss (Mathuriau and Chauvet, 2002;Dobson et al., 2003;Abelho et al., 2005;Gonçalves et al., 2006) which are comparable with the highest breakdown rates found in the literature (e.g. Bärlocher, 2005b). In addition, Gonçalves et al. (2006) noted that the leaf mass loss was faster in a deforested area with fast current velocity.
Our results showed an interaction of factors determining the leaf mass loss, which apparently is dependent on several factors, like plant species, season, water temperature, and current. The tree common in the gallery forest area, Cabralea canjerana, presented a leaf loss dry mass more influenced by the differences in temperature between the dry and wet seasons, with leaf loss higher during the wet season. Otherwise, for the herbaceous plant common in the deforested area, Pycreus decumbens, the leaf loss dry mass was more influenced by the differences in light incidence and current found when compared the forested and deforested areas, with leaf loss higher in the forested area.
The literature attributed the rapid breakdown to several other factors such as high activity of shredders (Petersen, 1984;Pearson and Tobin, 1989;Kobayashi and Kagaya, 2005;Tanaka et al., 2006) or microbes (Dudgeon, 1989;Dudgeon and Wu, 1999;Gessner, 2001;Mathuriau and Chauvet, 2002;Dobson et al., 2003;Abelho et al., 2005), and variations in stream water chemistry (Díez et al., 2002;Abelho and Graça, 2006); however, in most cases, the roles of these factors are inconclusive. Although all of these works used the same daily decay coefficient, there are a number of environmental different variables  such as leaf species, climate, latitudes and, consequently, distinct physical and biological effects, which can determine different breakdown rates. To allow comparisons with other studies, it is important pay careful attention to the measurement of these variables during the course of the experiment and tries to assess their influence on decomposition rates .
Nevertheless, as in the present study, it is difficult to determine which of these factors contribute to convert this coarse leaf-litter (CPOM) into fine particulate organic matter (FPOM), and when they may do so, probably because they act synergistically.
According to , the optimum sampling strategy is to collect packs frequently during the early stages of the study in a way to detect the rapid initial changes.
This procedure was used in our work and showed important differences over time. In the beginning of the decomposition process the significant differences found on dry mass loss emphasized the differences between plant species and the effect of microorganisms accelerating the process on the previously conditioned leaves. Later, the spatial and seasonal differences also played an important role in the decomposition process, through the differences on light incidence, water current and temperature found when compared the forested and deforested areas and the wet and dry seasons.
The leaf texture could have been the main factor that determined different initial mass loss. Similarly to Mathuriau and Chauvet (2002) we used two contrasting fresh-leaf species, one herbaceous with thin soft leaves (P. decumbens) and one arboreal with thick tough leaves (C. canjerana), with less dry mass remaining for soft leaves than for though leaves for both works. Moreover, Mathuriau and Chauvet (2002) also observed rapid initial mass loss in leaves similar to the present work.
However, conditioned leaves lost about 20-40% more mass than unconditioned leaves, independent of species. It is highly probable that this high percentage of mass loss is partially related to the previous immersion of 14 days for microorganisms colonization added to the 3 days common to the unconditioned leaves, with leaching of soluble compounds and microbial degradation corresponding to the processes acting simultaneously during this initial leaf decomposition. Abelho et al. (2005) noted that dry leaves lost 19% of initial mass during the first 24 hours. The leaching of soluble compounds is generally rapid and may account for a substantial decrease in initial mass Bärlocher, 2005a).
Previous studies in tropical and temperate streams often used pre-dried instead of fresh leaves, which greatly enhances in leaching (e.g. Mathuriau and Chauvet, 2002), but this procedure complicates the comparison between fresh and dry leaves. The drying of leaves fractures membranes and alters the cuticle, rendering the leaf more susceptible to attack by microbes and invertebrates and also enhancing the loss of solute compounds , and thus causing an overestimation of breakdown. A similar overestimation effect was observed in the present work where conditioned leaves lost much more mass than fresh leaves. But in natural conditions the colonization by microorganisms may start before the leaves reach the stream and probably accounts for a substantial decrease in initial mass .
After two weeks of immersion the leaf mass loss was higher in the wet than in the dry season. The increase of water discharge during the wet season probably increased the leaching of the leaves and, consequently, played a strong effect on leaf structure disintegration. The effects of physical abrasion can also change temporally according to discharge fluctuations (Kobayashi and Kagaya, 2005). Consequently, the soft P.
decumbens leaves lost more mass than the tough C. canjerana leaves at this time of the year, despite an isolated case in the deforested area during the wet season when C.
canjerana lost more mass than P. decumbens. It is probable that in this case the mass loss was overestimated since the force of water abrasion may have disintegrated entire leaves into some finer particulate fraction, difficult to measure .
The current velocity can also have an indirect effect on the leaf mass loss, acting as a controlling agent of the biotic factors, like the remove of macroinvertebrate, that, in turn, are directly involved in the decomposition process. This effect was evidenced by the fact that the stream deforested area, where a high current velocity occurs, was the stretch that presented a significant leaf mass remaining (or less decomposition) of P. decumbens. The results also demonstrated for C. canjerana a negative correlation between percentage of dry mass remaining and temperature and, consequently, season. The increase of the temperature during the wet season optimized the decomposition process of this arboreal plant species.
The differences in the decomposition rates between tropical and temperate streams may be related to the velocity at which microbes colonize and decompose leaf material (Abelho et al., 2005). Although the degradation by microbes and fragmentation by invertebrates are considered the main mechanisms determining the mass loss, in some systems invertebrates can be considered as unimportant in energy transference in detritus based systems, while in other cases they may be the key elements . According to Mathuriau and Chauvet (2002), the patterns in leaf litter breakdown at low latitude is characterized by low shredder involvement and high microbial processing.
Apparently, the fast leaf breakdown found in this study may have been caused by differences in leaf quality, related to the initial chemical composition and texture of the leaves, and by differences on physical environmental characteristics, which also can improve the action of microbes and invertebrates. In addition, the current velocity can be     including mollusks, crustaceans and, predominantly, insects. The differences on the surrounded vegetation of the two areas defined a strong spatial variation in the abundance of macroinvertebrates, which was higher in the deforested area. Otherwise, a temporal variation on the total abundance was not so evident, although a temporal pattern was significant when analyzed the taxonomic groups separately, some taxa more abundant during the dry season and other during the wet season. Abundance differences in function of the type and quality of the leaves were more evident for the second treatment, with a significant preference for pre-conditioned leaves. The high abundance and diversity of macroinvertebrates in the leaf-bags reinforce the importance of the allochthonous matter input, that can be used as food or refuge against predators and environmental perturbations.    Tabela VI. Resultados da análise de variância (One-way ANOVA) usada para comparar a abundância dos 15 grupos taxonômicos mais representativos, separadamente para as estações seca e chuvosa. Para as comparações significativamente diferentes (p< 0.05) foi indicado em qual das situações comparadas a abundância foi maior, considerando as variáveis: tipo de entorno (Aa-área aberta e Af-área fechada), tipo de folha (Arb-Cabralea canjerana e Herb-Picreus decumbens) e condições da folha (Ff-folha fresca e Fc-folha condicionada). Grupos taxonômicos: Mollusca-Ancylidae (Anc), Coleoptera-Heterelmis (Het), Diptera-Chironomidae (Chi), Diptera-Empididae (Emp), Diptera-Simuliidae (Sim), Ephemeroptera-Americabaetis (Ame), Ephemeroptera-Traverhyphes (Tra), Ephemeroptera-Farrodes (Far), Ephemeroptera estádios iniciais (Eph), Trichoptera-Phylloicus (Phy), Trichoptera-Smicridea (Smi), Trichoptera-Neotrichia (Neo), Trichoptera-Ochrotrichia (Orc), Trichoptera-Hydroptila (Hyd), Trichoptera estádios iniciais (Tri).

DIET OF INVERTEBRATES SAMPLED ON LEAF-BAGS INCUBATED IN A TROPICAL HEADSTREAM
Trabalho a ser submetido à publicação na Acta Limnologica Brasiliensia.

headstream ABSTRACT
The diet of macroinvertebrates sampled in leaf-bags immerged in a tropical stream was analyzed in a spatial (a forested and a deforested area) and temporal scale (dry and wet season). The macroinvertebrates were represented mostly by detritivores specialized in fine detritus (69%), followed by generalist detritivores (10% with diet based on fine and coarse detritus), carnivores (10%), omnivores (8%), and one detritivore genera specialized on coarse detritus (3%). The detritivores also presented a broad spatial and temporal distribution and

INTRODUCTION
Headstreams are particularly influenced by riparian vegetation because the ratio of shoreline to the area of stream bottom is high and because vegetation provides shade and organic input (Cummins, 1977;. The intimate relationship between the stream and its riparian zone forms the basis for a significant portion of the annual energy input , Cummins & Klug, 1979. Also, the study of the aquatic fauna interaction during the utilization of the food resources is an important knowledge in the understanding of the trophic structure and organization of an ecosystem . Every year streams receive great quantities of litter, a large part of which consisted of leaves from the riparian vegetation , an important source of allochthonous organic matter assimilated by the aquatic fauna . After entering low order streams, leaves are subject to physical abrasion, microbial degradation and invertebrate fragmentation , with shredders very well-known for reducing detritus to fecal particles . These particles become part of the fine particulate organic matter , which is ecologically important to populations of collectors inhabiting lower reaches of streams . The shredders and collectors are thus the major primary consumers in streams, proving the main link between the organic input and the predatory invertebrates and vertebrates (Cheshire et al., 2005).
Because of the importance of detritus in stream ecosystems, major emphasis has been directed toward the compartmentalization of this food source. A considerable amount of information on the processing of detritus in low-order forested streams is available in the literature (e.g. . The comprehension of the relationship between this basal resource and the aquatic invertebrates is based mainly on gut content studies (Basaguren et al., 2002;Rosi-Marshall & Wallace, 2002;Motta & Uieda, 2004;Cheshire et al., 2005;Lancaster et al., 2005;Albariño & Villanueva, 2006).

Study area
The study was carried out in the Ribeirão da Quinta stream (23º06'47"S, 48º29'46"W), located in the municipality of Itatinga, São Paulo State, southeast Brazil, at an elevation of 743m a.s.l.. This is a third-order stream, located on a cattle raising farm, distant from urban areas. In the investigated region, the climate is tropically warm and wet, with only one dry month (August) and two distinguishable seasons: a dry season between April and September and a wet season between October and March (E. M. Carvalho & V. S. Uieda, unpublished data).

Macroinvertebrates sampling
Two areas of this stream, each with surface area of about 120m 2 , were chosen for the study. The first is shaded by a well-preserved gallery forest and, in this way, is called "Forested area". The second is located about 300m downstream and is surrounded only by herbaceous vegetation, being called "Deforested area".
The study was performed during In the laboratory the leaves were rinsed with distilled water on three granulometric sieves (GRANUTEST, mesh of 1.00; 0.50 and 0.25 mm). The sieves were inspected under stereomicroscope for sorting the macroinvertebrates. The animals were identified to the genera level when possible (Pennak, 1978;Lopretto & Tell, 1995;Merritt & Cummins, 1996;Fernández & Domínguez, 2001;Costa et al., 2004;Olifiers, 2004;Costa et al., 2006) and counted for the determination of abundance. In a way to estimate the relative space used by each macroinvertebrate group when colonizing the leaves, the area occupied (number of grid squares) by each taxa was determined using a graduate slide.

Diet analysis
The diet was determined for all taxa that achieved values of abundance and occupied area higher than 1% on each season and area, except for Chironomidae larvae. As has been pointed out by some authors (Nessimian & Sanseverino, 1998;Nessimian et al., 1999;Henriques-Oliveira et al., 2003, Motta & Uieda, 2004, this family shows great diet variation with genera belonging to different trophic groups. In function of this genera flexibility on diet it is necessary the identification of Chironomidae until genera level for a correct diet analysis, what was not possible to do in this study. This material will be identified and analyzed in a future work. When possible, ten animals of each taxa and of each treatment (season and area) were dissected. Their guts were extracted and analyzed under a stereomicroscope and a microscope. Gut contents were classified into four categories: coarse detritus (coarse particulate organic matter -CPOM), fine detritus (fine particulate organic matter -FPOM), filamentous and unicellular algae, and animals. The area occupied (number of grid squares) by each food type was measured using a graduate slide. The relative area (%) covered by each food type was determined and the taxa were assigned to one of the three trophic groups: (1) detritivores consuming CPOM and FPOM, (2) carnivores preying on animals, (3) omnivores consuming resources from two or more food categories. The herbivorous trophic group was not considered due to the origin of the material consumed (only decomposing leaf litter).
Although the macroinvertebrates were sampled in leaf-bags of two plant species, the diet analysis showed for the same taxa no differences in function of the species of leaves colonized. Thus, this variable was not considered and the diet differences were analyzed only in relation to the area and season that the taxa was sampled.
To visualize the seasonal and spatial organization of the macroinvertebrates in relation to the diet we used the Bray-Curtis measure of dissimilarity and cluster analysis performed with the program Biodiversity Professional version 2 (McAleece, 2004).

RESULTS
The macroinvertebrates sampled in leaf-bags resulted in a total of 5,826 specimens belonging to 22 families and nine orders, most represented by aquatic insects (Table I).
During the dry season Chironomidae was the dominant taxa, representing more than 50% of the abundance (Table II). During the wet season Trichoptera was the dominant family in the deforested area, while Ephemeroptera was the dominant family in the forested area (Table   III). The number of taxa represented by more than 1% of abundance and occupied area, and then used for diet analysis, varied between seasons and areas: 11 in the dry season-deforested area, 10 in the dry season-forested area, 17 in the wet season-deforested area, 11 in the wet season-forested area (Tables II and III). Except for Chironomidae larvae, not analyzed, the taxa analyzed but with the diet not represented in the Figure 1 had all individuals with the gut empty.
Fine detritus was the main food category consumed by the macroinvertebrates of Ribeirão da Quinta stream, with few exceptions (Figure 1). This food resource was the only food ingested for most taxa or was, in some cases, consumed in association with coarse detritus, animal material, and algae. The four taxa found on both seasons and areas, Ancylidae, Heterelmis, Traverhyphes and Farrodes, showed feeding specialization for fine detritus. The Trichoptera genus Phylloicus fed mainly on coarse detritus and was sampled only in the forested area. Animal material was consumed mainly by two Odonata genera and by the Plecoptera genera Ancroneuria. This last taxon was the only one that showed temporal variation on diet ( Figure 1). Algae was not consumed isolated and was found in the gut content always in low percentage.
Grouping the species by similarity in the diet (Figure 2) indicated the presence of five groups. The trophic group of detritivores that consumed predominantly or exclusively fine detritus (Group I) was the most representative in the two areas and seasons and comprised 69% of the analyzed taxa. The last 31% were grouped in carnivores that consumed mainly or exclusively animal material (Group II -10%), omnivores consuming two or more resources, including animal material (Group III -8%), detritivores that consumed fine and of coarse detritus in a similar amount (Group IV -10%), and one detritivore genus specialized on coarse detritus (Group V -3%).

DISCUSSION
Allochthonous litter is the dominant energy resource for organisms in low-order shaded streams (Dobson, 2005;. A considerable amount of litter entering streams is retained within the channel (Pozo, 2005) and is available as food to the vast majority of detritivores and microbial decomposers (Dobson, 2005). This litter can be colonized by a great biomass of macroinvertebrates (Uieda & Gajardo, 1996), which can use it not only as food but also as shelter.
The macroinvertebrates community sampled in leaf-bags was composed mostly by Insecta, represented by seven orders, usually cited as representatives of the benthic community associated with litter (Crisci- Bispo et al., 2007). Of the diverse taxonomic groups that comprise the stream macroinvertebrates community, none has been more studied than the aquatic insects. The aquatic insects are not only diverse taxonomically and functionally, but they are frequently the most abundant large macroinvertebrates collected in stream benthic samples (Hauer & Resh, 1996).
Other authors analyzing the benthic macroinvertebrates associated to the rocky substratum of the Ribeirão da Quinta stream found a benthic community similar to the present work Ribeiro & Uieda, 2005;. Once that the community sampled in the leaf-bags is similar to the one usually observed in rocky substratum, then which is the intimate trophic relationship between this community and the litter? It may be over-simplistic to assume that all invertebrates collected from litter are feeding specifically on the leaves .
Fine particulate organic matter (FPOM) or fine detritus of unidentified origin was the most abundant food resource found in the gut content of the macroinvertebrates from the Ribeirão da Quinta stream, classified as detritivores, specialists or generalists. The taxa with large occurrence and abundance also used fine detritus as the main food resource. Motta & Uieda (2004) also found particulate organic matter in the diet of most aquatic insects of a tropical stream. It has been suggested that detritus has the potential to support systems with a great diversity of species (e.g. Rosemond et al., 1998;Moore et al., 2004). High species richness, a great availability of detritus as a food resource and a high number of detritivore species was also found in the epiphytic compartment of a tropical stream of the same basin of Ribeirão da Quinta (Motta & Uieda, 2005).
In some cases, the fine detritus was also found associated with other resources, as coarse particulate organic matter and algae. When the main ingested food is coarse detritus, the detritivorous consumer is usually classified in the functional group of shredder (Merritt & Cummins, 1996). In this sense, the Trichoptera genus Phylloicus was the only one that can be classified as shredder in this study. Some works suggest that shredders are scarce in tropical streams (Dobson et al., 2002), mostly because most of the common shredder taxa from temperate systems are lacking in the tropics (Rosemond et al., 1998, Cheshire et al., 2005. Moreover, shredding invertebrates may be less important in tropical streams because there are alternative decomposition pathways for leaves, such as faster microbial processing due to higher temperatures Mathuriau & Chauvet, 2002).
Some macroinvertebrates consumed algae associated to fine detritus, or also to coarse detritus and animal material, but always in small amount and never alone. The animals that fed on a combination of algae, organic matter and microbiota by scraping submerged rocks or macrophytes are classified by some authors as periphyton feeders (Uieda et al., 1997;Motta & Uieda, 2004). The trophic group of periphyton feeders used for fishes by Uieda et al. (1997), for macroinvertebrates by Motta & Uieda (2004), and correspondent to herbivorous and detritivorous of the functional feeding group of scrapers by Merritt & Cummins (1996), was not used here due to the low proportion of algae consumed, choosing to keep the definition of detritivores for these consumers.
The trophic group of carnivores was represented by one genus of Plecoptera (Anacroneuria) and two of Odonata (Hetaerina and Heteragrion). Motta & Uieda (2004) also observed carnivory between genera of Odonata, which reduced the feeding overlap by consuming different insect orders, like found for the carnivorous genera of the Ribeirão da Quinta stream. Hetaerina consumed Ephemeroptera (Americabaetis and juveniles of early stages) and Trichoptera. Heteragrion consumed mainly Diptera (Chironomidae) and fewer amounts of early developmental stage of Ephemerotera. The diet of Anacroneuria was based on aquatic insects, mainly Chironomidae, and on fewer amounts of Trichoptera (Smicridea, Glossosomatidae) and of Ephemeroptera (juveniles of early developmental stage). All three carnivores genera were polyphagous with respect to their animal prey, what, according to Lancaster et al. (2005), is common for many other predatory insects.
True omnivory (mixing plant and animal food) is common among terrestrial and marine arthropods, but poorly documented in freshwater systems (Lancaster et al., 2005). The only two genera showing true omnivory in the Ribeirão da Quinta stream, Smicridea and Anacroneuria, were the most representative genera of Trichoptera and Plecoptera orders, respectively.
A minor seasonal change in functional feeding groups was also found by Motta & Uieda (2004) and related by those authors with constant food resource availability. The seasonal dietary changes involved mostly modification in the proportion of food items consumed, like also found by Motta & Uieda (2004). The spatial analysis showed similar results, reinforcing the trophic structure stability of the macroinvertebrate community studied.  hypothesized that endemic macroinvertebrates population respond mainly to changes in the relative availability of food sources. According to the authors, several lines of evidence point toward relatively greater use of biofilm as a food resource in agriculture streams. Nevertheless, the specialist shredder Phylloicus have limited distribution and appear to be restricted to the forested area. For Albariño & Villanueva (2006), the hydraulic conditions and food availability were responsible for a higher density of a specialist shredder. In the same way, our results suggest that food availability (high leaf litter availability in forested area) and stream retentiveness (high leaf litter standing stock during the dry season) are suitable habitat traits to this species.
Like argued by , some of the negative impacts of deforestation on sensitive taxa could be reduced by maintenance of vegetated riparian zones.
Shredders and collectors are considered the major primary consumers in forest streams, providing the main link between the organic inputs and the predatory invertebrates (Cheshire et al., 2005). However, these forest specialists are often micro-endemic and particularly vulnerable to deforestation (Cummins & Klug, 1979;Baxter et al., 2005).

INTRODUCTION
Many headwater streams are strongly influenced by riparian vegetation which reduces autotrophic production by shading and supplies energy in the form of leaves . The decomposition of litter in streams is a biological process involving both microorganisms and invertebrate consumers  resulting in the incorporation of energy from the litter into secondary production and in the release of large amounts of fine particulate organic matter (FPOM -Gessner et al., 1999). The production of FPOM can be ecologically important for populations of collector living further downstream .
Given the natural variability of rain and litter input in streams, which dictate retention and transport of organic matter (e.g. Pardo & Álvarez, 2006), it is plausible that many stream invertebrates exhibit some variation in their diet. This has been demonstrated specifically in shredders. For instance, Friberg & Jacobsen (1994) showed that conditioned alder leaves and fresh filamentous green algae were equally palatable for two shredder species. Mihuc & Mihuc (1995) also showed that four shredder species exhibited similarly high growth rates when fed periphyton and coarse particulate organic matter (CPOM) resources. Franken et al. (2005) showed that biofilm on leaf surfaces can be an important component of the nutritional for two shredder species. Feeding plasticity or generalist diet may allow invertebrates to cope with the variability of food sources in streams.
hirtum were calculated to consume between 2.3 and 8.6 times the mean annual CPOM standing stock in the stream (González & Graça, 2003;Azevedo-Pereira et al., 2006). Those high values and the observed variability of organic matter in the stream bed throughout the year (González & Graça, 2003) suggest that the detritivore guild, at times, may have limited food in the S. João stream and that they may be capable of modifying their behavior and diet according to food availability. Certainly the same situation occurs in many other systems .
In the present study we hypothesize that shredders are able to maintain viable population in streams because of their capability of feeding and growing on alternative food sources. As a test organism we used Sericostoma vittatum Rambur. The study was carried out in the laboratory and, besides alder leaves, we used as alternative food sources FPOM, a macrophyte, and biofilm. Seven food items were used in the experiments. CPOM was provided in the form of pieces of leaves of Alnus glutinosa. Leaf powder came from Alnus glutinosa and Acacia dealbata, which were obtained with a homogenizer (Poly Tron ® PT 2100). The material was sieved and only the fraction passing through a sieve of 1.00 mm and retrieved by a sieve of 0.18 mm was used. All the above leaves were conditioned for 2 weeks in nylon bags (0.5mm mesh size) submerged in the stream. FPOM from the S. João and Ceira streams (tributaries of the Mondego River) were collected directly from the stream bed and sieved as described above. As a macrophyte food source we used Miriophyllum aquaticum (Vell.) Verdc. This species is common in the lower sections of the Mondego River basin and was available during the experimental period. Biofilm was obtained by exposing stream cobbles (≈ 40 mm diameter) in a shallow in a tank at the University Botany Garden for 4 weeks. By that time, a visible apparently uniform algal grow was evident over all substrates.

MATERIALS AND METHODS
The ash free dry mass (AFDM) of each food item was determined as the difference between dry mass (oven dried at 60ºC, 2 days) and the ash mass (550º C, 4 h). Total nitrogen and phosphorous content in all food items were determined according to Flindt & Lillebø (2005). Chlorophyll a content was determined by spectrophotometry (Eaton et al., 1995).
Three replicates of each food item were used for the above analyses and expressed per unit of A total of 280 specimens of S. vittatum with sizes ranging from 3.4 to 6.8 mg were used in the experiments. Specimens were allocated individually into plastic cups containing 200 ml of filtered stream water (GF/ C Whatman) and the bottom covered by stream sand (ignited for 4 h at 550ºC). Aeration was provided with pipette plastic tips connected to an air pump. The invertebrates were randomly assigned to food categories, being 40 individuals for each food items (seven treatments). Food was provided "ad libitum" -leaves: several pieces of approximately 2 X 2 cm each; FPOM: two spoons of material; macrophyte: 2 leaves; biofilm: 3 cobbles. The amount of available food was daily checked and replenish if necessary. All food items and 50% of the water were renewed weekly. The individuals who pupated were not replaced.
There were no statistical differences in the initial mass of individuals in the seven treatments (ANOVA p >0.86). The experiments were carried out for 14 days under 15 ± 1ºC and a photoperiod of 12:12 h (light:dark). The temperature 15ºC was previously shown to be near the optimum for growth in this species (González & Graça, 2003).
The initial and final sizes (in mg) of each individual were estimated from the diameter (mm) of the anterior opening of the caddis case according to Canhoto (1994; r 2 = 90.8%; n = 27). Daily growth rates (DGR) were calculated as the difference between the final and initial dry mass, divided by the elapsed time in days (14 days). The percentage of daily growth was computed by dividing the DGR by the initial mass, multiplied by 100 (Feio & Graça, 2000).
Comparison of growth rates and physicochemical properties of food among treatments were done by one-way ANOVA performed on arcsine-transformed data (Zar, 1999), followed by comparisons with the Tukey HSD test for unequal N (replicates). The relationship between growth and food properties was tested by the Spearman Rank Correlation.

RESULTS
The 7 food items differed in all physical and chemical parameters (ANOVA p < 0.001). Nitrogen content from leaves or leaf powder was similar to natural FPOM from S.
João stream, but lower than the natural FPOM from the larger Ceira River (Table 1). The percentage of organic material in the natural FPOM was very low and different between rivers: approximately 18% in the S. João and 4% in the Ceira (Table 1). FPOM from streams had some algae, although much lower than the biofilm recovered from stones (Table 1). The mean caloric content was highest in leaves either entire or reduced to FPOM (Table 1). The two natural FPOM differed significantly in terms of calorie content. It was lower in the Ceira than in the S. João stream ( Table 1).
Survival of S. vittatum larvae during the experiment (two week) was 100%. However, some animals pupated and therefore growth of these specimens was not computed. The highest pupation occurred in specimens fed FPOM A. glutinosa (17.5%) followed by FPOM S. João (15%). The size of specimens entering pupation in the set fed with A. glutinosa (in the form of FPOM and CPOM) was about double that for the other food items (Fig. 1). No pupation occurred in larvae feeding on macrophyte and biofilm.
Growth rate of S. vittatum was significantly influenced by the food item (ANOVA, p = 0.0082). The food item promoting the highest growth was A. glutinosa, in the form of FPOM (6.48 % day -1 ) and CPOM (4.24 % day -1 , Fig. 2); all other forms of FPOM and biofilm provided relatively low growth rates (0.77 -1.77 % day -1 , Fig. 2). Growth on FPOM from A.
dealbata promoted 4 times lower growth rate than FPOM from A. glutinosa. Neither nitrogen, phosphorus nor energy content was significantly correlated with grow (Fig. 3).

DISCUSSION
Growth values reported in the present work for sets of shredders feeding on alder (4.24 % day -1 ) were in the upper range of those reported by Graça et al. (2001; 2.9% day -1 , for sets of specimens with initial size of 1.7 mg). The aim of the experiments here reported was to determine whether typical shredders can present growth using alternative food resources. The main result was that A. glutinosa, both in the form of whole leaves and in reduced FPOM form, promoted similar growth for S. vittatum. Therefore, the chemical composition of these leaves, rather than their physical form, was the important factor for growth in our laboratory experiments. This is consistent with the fact that all other forms of FPOM provided relatively low growth. FPOM from A. dealbata and A. glutinosa, were physically similar to each other, but the latter food item promoted 4 times more growth. The high value of A. glutinosa leaves as a food resource was also reported for other shredders (e.g. Friberg & Jacobsen, 1994, 1999, Jacobsen & Friberg, 1995, González & Graça, 2003.
The macrophyte M. aquaticum was also used as food source by S. vittatum and promoted some growth. However, the value was 64% lower, compared with leaves of A.
glutinosa. Friberg & Jacobsen (1994) also showed that the second most consumed food item was also fresh macrophyte, presenting values ranging from 73 to 37% when compared with leaves of A. glutinosa; however, these authors did not measure growth.
When specimens were fed biofilm, their growth was 58% lower than that obtained with leaves of A. glutinosa. The capability of shredders for growth when fed with periphyton was also demonstrated with five shredder species by Mihuc & Mihuc (1995). Franken et al. (2005) also showed that biofilm on leaf surfaces had a significant positive effect on the growth of two shredder species. In acidic streams where grazers were absent, Ledger & Hildrew (2005) found that typical shredders such as Leuctridae were important grazers, regulating benthic algae. It seems therefore that some shredders may, in some instances, use benthic algae as energy source.
The FPOM collected directly from the stream was high in nutrient content. FPOM is supposedly constituted by very refractive particles such as pieces of leaves not consumed by fungi or invertebrates, and very high quality resources such as fecal pellets, algae and bacteria. The quality of FPOM is therefore expected to vary seasonally and across streams and therefore its capability to promote growth is variable. In terms of nutrients, Graça et al. (2001) and Friberg & Jacobsen (1999) reported in A. glutinosa nitrogen values of 3.5 and 3.1% respectively, and phosphorus levels of 0.09% in both, which is comparable to the values found in the present study, 3.8 and 0.05%, respectively. The biofilm we measured on the incubated stones was 3.8 mg.m -2 of chlorophyll a, which is similar to the 2.5 mg.m -2 reported by Fisher Wold & Hershey (1999). We calculated biomass at 10 g.m -2 AFDM, which was higher than the 4 g.m -2 cited by the same authors. However, those differences are expected given the large differences in water chemistry, temperature and other conditions in rivers.
None of the factors − nitrogen, phosphorus or caloric content − was correlated with growth, as also reported by Friberg & Jacobsen (1999) for nitrogen content. This lack of correlation suggests that other food properties, such as the presence of plant chemical compounds and chemical defenses, micro-nutrients or others are important for shredder consumers. It is relevant to note that leaves of A. dealbata and A. glutinosa in the form of FPOM promoted the minimum and maximum animal growth rates, respectively, though they presented relatively equal energy content values. In many cases it is difficult, if not impossible, to distinguish between selective feeding and physical or chemical restriction that prevent the intake of all apparently available food material . Since the texture of FPOM samples provided to S. vittatum was similar, it is clear the chemical composition of the food is important for shredders.
Pupation may also be an indicator of food quality. Diverse life-history patterns have evolved to enable species to exploit foods that are seasonably available, to time emergence for appropriate environmental conditions, to evade unfavorable physical conditions, and to minimize repressive biotic interaction (Merritt & Cummins, 1996). Whereas individuals fed on alder entered into pupation at a large size, sets of animals feeding on other food sources diverted energy from growth into pupation. Smaller pupae may result in smaller adults and presumably, smaller-sized reproductive output (Begon et al., 1990).
In a broad ecological context, if the results shown here can be generalized for other shredder species, then the Functional Feeding Groups should be taken from a very flexible perspective, in which shredders feed on CPOM when available, but are capable of surviving, growing and reproducing using other resources. According with  shredders can function as a generalist or specialist at any point in time. This flexible feeding strategy may explain densities of shredders higher than the expected from the available resources (e.g., González & Graça, 2003;Azevedo-Pereira et al., 2006;. In an ecological context, a generalist strategy makes sense if consumers inhabit environments with high variability of food resources. That is the case when litter input is seasonal in temperate stream systems where hydrological events may wash away leaves  and where spring and summer litter input is minimal. Therefore, according with , the use of Functional Feeding Groups to describe resource assimilation at the population or community level may be inappropriate, once that generalist resource partitioning seems to be predominance among lotic macroinvertebrates.