Riparian vegetation represents a transition zone between terrestrial and aquatic ecosystems (Naiman et al., 2005; Tank et al., 2010). Riparian zones display ecological functions such as bank stabilization, water control (quality and quantity), organic matter (OM) input, and conservation of biological diversity (Allan, 2004; Allan and Castillo, 2007). The vegetation in these areas also serves as a source of energy for aquatic ecosystems by providing allochthonous OM (Webster and Meyer, 1997; Tank et al., 2010). The input of coarse particulate organic matter (CPOM) accounts for approximately 90% of the allochthonous OM in headwater streams (Campbell et al., 1992), mainly in closed canopy areas, by decreasing photosynthetic production (Tank et al., 2010; Graca et al., 2015). Evaluation of CPOM throughout the year is thus important for understanding OM cycling in riparian zones (Abelho, 2001), mainly in tropical streams (Rezende et al., 2016). Studies addressing ecological processes in tropical streams have emerged only in the last decade (Goncalves et al., 2006a; Franca et al., 2009; Lisboa et al., 2014; Rezende et al., 2016).
CPOM fractions (leaves, flowers, fruits, and branches) are transformed to fine particulate organic matter (FPOM) and later into dissolved organic matter (DOM) by decomposing organisms (Gessner et al., 1999; Graca et al., 2015). The mineralization of OM by decomposition is a continuous process in trophic webs, which can be influenced by biotic and abiotic factors (Gessner et al., 1999; Graca et al., 2015). For example, higher water nutrient concentrations can increase the rate of leaf litter breakdown by increasing the activity of decomposers (Goncalves et al., 2007; Rezende et al., 2014). Chemical characteristics of the leaf litter also impact organic matter dynamics and nutrient cycling in these environments (Wantzen et al., 2008; Rezende et al., 2014; Graca et al., 2015). Leaf litter with higher concentrations of secondary compounds (e.g., polyphenols), structural compounds (e.g., lignin and cellulose), and hardness can decrease litter quality and breakdown rates (Wantzen et al., 2008; Rezende et al., 2014; Graca et al., 2015).
Fungi, bacteria, and benthic invertebrates are the main decomposer organisms in the transformation of CPOM to FPOM and DOM (Gessner et al., 1999; Graca, 2001; Graca et al., 2015). Aquatic fungi (e.g., hyphomycetes) can mineralize litter through enzymatic action, driving the breakdown of structural and recalcitrant compounds (e.g., lignin and cellulose). On the other hand, bacteria decompose labile molecules (e.g., secondary metabolites), manly in the initial stage of this process (Gessner et al., 1999; Graca, 2001; Graca et al., 2016). Microbial communities are also responsible for the nutritional enrichment of litter, increasing palatability and facilitating litter use by benthic macroinvertebrate shredders and scrapers (Graca et al., 2015; Sales et al., 2015). Although some tropical streams have high density and richness of shredders (Cheshire et al., 2005; Allan et al., 2009), most studies report low numbers (Boyero et al., 2011, 2012; Rezende et al., 2015) compared scrapers, which function in leaf fragmentation by consuming periphyton in these systems (Wantzen and Wagner, 2006; Goncalves et al., 2012a).
In South American savanna streams, litterfall occurs mainly in the dry season due to hydrological stress (Goncalves and Callisto, 2013; Goncalves et al., 2014a). Leaf litter breakdown may vary throughout the year, stimulated by increases in temperature (higher metabolism in the system), density of decomposer communities (microorganisms and invertebrates), nutrient concentrations (manly nitrogen and phosphorus), and water flow (fiscal abrasion) (Goncalves et al., 2014b; Graca et al., 2016). However, the majority of tropical studies evaluate only i) specific species of litter; ii) specific sample periods (60 to 90 days, manly in dry season); and iii) leaves collected outside of the riparian vegetation of the study stream. This methodological approach may negligence the high species number in tropical riparian zones and the mixing effects on the leaf litter input and breakdown in this system (Gessner et al., 2010).
Only two studies have been carried out in tropical streams, which measure aggregate leaf litter breakdown and litter fall input over the course of one year (Sales et al., 2015; Rezende et al., 2016). According to Rezende et al. (2016), this approach produces results that are more realistic in terms of measurement of litter breakdown, as they reflect seasonal variation in litter quality throughout the year. Another interesting aspect that has not been previously investigated is riparian zone litter dynamics in transition biomes (e.g. savannah and rainforest). Our hypotheses were constructed based on tropical studies of leaf litter breakdown and litter fall(Sales et al., 2015; Rezende et al., 2016), and in the transition characteristic of studied riparian zone, as follows: i) the total CPOM input will be higher than savanna riparian zones (less productive by lower rainfall), but lower than Atlantic rainforest (more productive due to higher rainfall); ii) the CPOM will peak during the transition from dry to rainy season (due to hydric stress and mechanical removal by rain); and iii) seasonal changes will accelerate leaf breakdown in rainy season due to increase in physical abrasion and decomposer community activity (higher metabolisms). This study aims to describe the litter input and leaf breakdown of allochthonous CPOM over one year in a savanna stream.
The study was conducted in the Boleiras stream (first order) in the Rio Preto State Park, Minas Gerais, Brazil (18°07’04” S - 43°20’42” W). The stream is located in the upper region of Jequitinhonha Valley in the Espinhaco Mountain complex, which has an average altitude of 800 meters. The climate regime is a tropical AW (with a dry winter), with a rainy season from November to March (average rainfall 223 mm) and a dry season from June to August (average rainfall 8 mm; http://hidroweb.ana.gov.br/).
Input of organic matter
Litterfall was measured monthly from January to December 2012. We selected 5 sampling points spaced 20 meters apart, for a total length of 100 meters along the stream (for details, see Sales et al., 2015; Rezende et al., 2016). The litter directly entering the stream (vertical input, VI) was measured at each sampling point using 3 rows of 6 buckets each (0.53 m2) (i.e., a total of 18 buckets per point at 5 points = 90 replicates) suspended 2 m above the stream with ropes, which were transversely displayed in 5 rows (points were used as replicates: 5 points x 11 months = 55 replicates). The bucket bottoms were perforated to allow rainwater to evacuate. At monthly intervals, accumulated litter in the buckets was retrieved and weighed in situ (wet weight), and the contents of the bucket with the highest leaf litter mass in each row was used for the leaf litter breakdown experiments (see below). The contents of the remaining buckets (5 buckets in each row, which were not used in the leaf litter breakdown experiment) were oven dried to a constant mass (60oC, 72 h), and the resulting weight was used as a correction factor to estimate the moisture of the litter used in the leaf litter breakdown experiment. The dry material was separated into the following categories: leaves, branches, reproductive material (flowers and fruits), and ‘miscellaneous’ (unidentified particulate organic matter).
Leaf litter breakdown and aquatic invertebrates
For this analysis, we used the leaves from the vertical input that were collected in 15 buckets above the stream; the leaves were weighed and placed in 15 litter bags (30 × 30-cm, with a 10-mm mesh), and submerged (leaf litter from one bucket for one litter bag) in the stream for 30 days (until the next sampling period). The amount of material placed in the bags ranged from 1 to 3 g (litter bags having different sample weight due to monthly changes in litterfall throughout the year), and after one month, the leaf litter samples were removed and replaced with new ones (for additional information, see Sales et al., 2015; Rezende et al., 2016). The collected litter bags were transported to the laboratory in a cooler. According to Rezende et al. (2016), the use of bulk litter instead of leaves from known species better reflects decomposition at a site, representing a more realistic measurement of litter breakdown by accounting for seasonal variation in litter quality.
The remains of the leaf litter were washed with distilled water over a 250 μm sieve in the laboratory, and the collected invertebrates were fixed in 70% ethanol. Invertebrates were later identified to family and genus, and classified into functional trophic groups as follows: gathering- collectors, filtering-collectors, shredders, scrapers, and predators, according to Hamada et al. (2014), Perez (1988), Merritt and Cummins (1996), and (Cummins et al., 2005). Leaf discs (12 mm diam.) were removed with a cork borer from five randomly selected leaves, and the 5 discs were used to determine ash-free dry mass (AFDM; for correction of inorganic compounds). We calculated the AFDM by subtracting the ash weight (muffle in 750°C for 4h) from dry weight. The remaining leaf litter was placed in aluminum trays and dried in an oven at 60°C for 72 h to determine dry weight. Due to the high water flow and safety concerns during litterbag installation, the leaf breakdown experiment could not be performed from January to March.
Physical and chemical parameters of water and leaf litter
A multianalyzer (model 85, YSI Inc., Yellow Springs, OH, USA) was used to measure in situ temperature (°C), electrical conductivity (μS cm–1), pH (measured using YSI’s EcoSense pH100A), and dissolved oxygen (mg L–1). On each sampling occasion, we measured current velocity (m s–1) using a flow meter (Sigma Sports model FP101, Global Water Instrumentation, Inc., Gold River, CA, USA), and the depth and width of the stream with measuring tape. Rainfall and air temperature data (mean values for the entire month) were obtained from a meteorological station (number 1843015) of the National Water Agency of Brazil (http://hidroweb.ana.gov.br/).
Total polyphenol and tannin concentrations were estimated per 100 mg of leaf litter as proposed by Barlocher and Graca (2005) and Graca and Barlocher (2005), respectively. Phenolic concentrations were estimated per leaf after 30 days of incubation. The initial concentration per leaf was not measured for the breakdown experiment.
We are describing a case study (in a single stream) and therefore, the results and inference only relate to this stream. All analyses were performed in R version 3.3.0. Data normality was assessed with a Kolmogorov-Smirnov test, and homogeneity of variance was assessed with Levene’s test; values were ln (+1) transformed if necessary. Differences in litterfall between the vertical inputs (points as replicates) and the inputs (response variables) over time (explanatory variable) were analyzed by repeated-measures ANOVA (RM-ANOVA; Crawley, 2007), and a contrast analysis was used to assess differences among months (Crawley, 2007). In this contrast analysis (orthogonal), the months were ordered by increasing input values and tested in a pairwise fashion (with the months with the closest values). Stepwise model simplification was performed by sequentially adding monthly values that did not affect the model, and testing against the next month in the sequence (for more details, see chapter 9 in Crawley, 2007).
Monthly leaf litter breakdown rates (k) were obtained from a model assuming negative exponential mass loss during the 30 days of incubation (Wt=W0e-kt: Wt = remaining weight; W0 = initial weight; -k = decay rate; t = time). RM-ANOVAs and contrast analyses were used to test for significant differences among months (explanatory variable) in the remaining mass in the litter bags, and the abundances of scrapers and shredders (response variables). The average values of 5 points along the stream were used as repeated measures (over the months). A oneway ANOVA was used to test phenol and tannin concentrations (dependent variable) over the months (explanatory variables) (Crawley, 2007), and associations between variables were assessed using a Spearman’s correlation test (Crawley, 2007). A multiple linear regression was used to evaluate relationships between remaining mass and climate variables (air temperature and precipitation), physical and chemical water properties (dissolved oxygen, conductivity, pH, current speed, and water flow), the decomposer community (scrapers and shredders), and chemical characteristics of leaves (initial tannins, and initial and final phenols) (Crawley, 2007). In multiple linear regression, the beta coefficients represent the independent contributions of each independent variable to the prediction of the dependent variable. This type of correlation is also referred to as a partial correlation.
Water and leaf litter physical and chemical Parameters
Rainfall during the study period was low (mean of 10}37 mm; Fig. 1) with peaks in January and November. The average annual flow rate was 0.25 m3 s (}0.1), and water temperature ranged from 18 to 22°C, with air temperature ranging from 16 to 23°C. The water stream was circumneutral, with low electrical conductivity but high levels of dissolved oxygen (Tab. 1). The average concentrations of secondary compounds were 5% tannins and 11% phenols for total litter mass throughout the year. Higher leaf litter tannin concentrations (12%) were observed in August (ANOVA; F(8, 85)= 3.56; P=0.001), but the concentration of phenols did not change over time (ANOVA; F(8, 85)= 1.85, P=0.078).
Organic matter input
Total OM input (litterfall) over the 11-month study period was 335 g.m–2 year –1 or 30}5 SE g.m–2 month–1, indicating an annual average of 360}5 g.m–2 year –1. Leaf litter was the OM category with the highest annual contribution (50%) and two input peaks, one in April and another in September and October (Tab. 2; Fig. 2). The lowest values occurred in the dry season from June to August. Branches had the second largest contribution (31}3%), with an input peak in November that coincided with higher rainfall (Tab. 2; Fig. 2). Flowers and fruits contributed 18% (}1) of the total OM, with an input peak in May (Tab. 2; Fig. 2). Miscellaneous OM made the smallest contribution with 2% (}1), which did not change throughout the year. Branches positively correlated with rainfall (r=0.63, P=0.03), but there were no significant correlations among any other categories of OM input with either rainfall or air temperature (P>0.05).
Leaf litter breakdown rates
The average remaining mass was 65% (}3) over the year, with lower values in May (49%) and higher values in August (79%). The mean decomposition coefficient (“k”) was -0.016, ranging from -0.008 (August) to -0.029 (May). Similar to that observed for vertical input litterfall, the remaining leaf mass showed a bimodal distribution with two peaks, the first in August and the second in November and December (Tab. 2; Fig. 3). The remaining mass (adjusted R2 = 0.76, F(3,5)= 7.44; P=0.03) showed a positive relationship with water flow (Beta= 0.70) and water temperature (Beta= 0.38), but a negative relationship with scraper abundance (Beta= -0.42). The remaining mass was not affected by the other variables tested.
Aquatic invertebrate communities
The density of invertebrates that colonized the leaf litter ranged from 260 (October) to 9 individuals g–1 (November), averaging 117 (}9 SE) ind g–1. The average richness over the year was 10, and was lowest in November (2 taxa) and highest in October (15 taxa). During the study period, 10,038 invertebrates were collected that were associated with leaf litter, with the most abundant taxa being Tanypodinae (25%), Chironominae (19%), and Orthocladiinae (16%) in the family Chironomidae (Diptera), which represented 60% of the collected individuals.
Only 6% of all invertebrates were classified as shredders, with the highest values observed in August (12%) and November (13%) (Tab. 2; Fig 3). The shredder trophic functional group was represented by Triplectides sp., Notalina sp., Marilia sp., Phylloicus sp. (Trichoptera), Paragripopteryx sp. (Plecoptera) and Hyalella sp. (Crustacea; Tab. 3). The total relative abundance of scrapers was twice more (12%) than that of shredders, with significantly higher values in July (18%) compared to other months (Tab. 2; Fig. 3). The ‘scraper’ trophic functional group was represented by Oxyethira sp., Neotrichia sp., Metrichia sp., Helichopsyche sp. (Trichoptera), Miroculis sp., Hagenulopsis sp., Farrodes sp. and Askola sp. (Ephemeroptera) (Tab. 3).
Organic matter input
The total litter fall input in this transition system (360 to 365 g m–2 year–1) was higher than other savanna systems (288 to 336 g m–2 year–1 in Goncalves et al., 2006a; Franca et al., 2009; Goncalves and Callisto, 2013; Rezende et al., 2016), but lower than typically found in tropical rainforests (113 to 2812 g m–2 year–1 in Abelho, 2001; Chara et al., 2007; Gregorio et al., 2007; Zhou et al., 2007; Alvarez et al., 2009; Cogo and Santos, 2013; Lisboa et al., 2014). Savanna systems (e.g., the Cerrado) and tropical rainforests (e.g., Atlantic forest and the Amazon forest) have a different precipitation range, which explains the differences in productivity (Goncalves and Callisto, 2013; Goncalves et al., 2014a). Therefore, the transition zone between Cerrado and Atlantic forest, as in this study area, may have intermediate patterns in plant productivity in riparian zones.
Total CPOM in the litterfall was seasonal, with peaks in dry and rainy season. Leaves were the highest fraction in of CPOM, as shown in other tropical studies (Wantzen et al., 2008; Alvarez et al., 2009; Chave et al., 2010; Tank et al., 2010; Zhang et al., 2014; Rezende et al., 2016). The higher leaf biomass and total CPOM in the transition period is also consistent with other studies of savanna (Goncalves et al., 2006a; Goncalves and Callisto, 2013; Rezende et al., 2016) and other Brazilian tropical systems (Gregorio et al., 2007; Franca et al., 2009; Goncalves et al., 2014a). This leaf senescence is triggered by desiccation due to water stress, and explains the higher CPOM input from August to November (Campanella and Bertiller, 2008). Leaf renewal can also be stimulated by mechanical removal by the first rains (Larned, 2000; Rezende et al., 2016). Input of branches (the second highest CPOM) in November also indicates greater mechanical removal by rainfall and associated winds (Vanschaik et al., 1993; Goncalves et al., 2014a).
The novelty in litterfall refers to high leaf litter input also in April, which has not observed in other savanna systems (Goncalves et al., 2006a; Goncalves and Callisto, 2013; Rezende et al., 2016) or other Brazilian biomes (Gregorio et al., 2007; Franca et al., 2009; Goncalves et al., 2014a). Riparian vegetation is an ecotone that serves as an ecological corridor among biomes, and the species composition includes that of the adjacent systems (Wantzen et al., 2008; Goncalves and Callisto, 2013). Thus, the ecotone characteristics of the riparian zones are highlighted due to geographical proximity (transition system) of this savanna stream to the Atlantic forest biome. The input peak in April may also be a phenological response of one specific species that was not observed in previous studies (Goncalves and Callisto, 2013). This phenological response may also explain the increase in flower and fruit biomass in May, due to lower rainfall and less physical damage to plant parts associated with germination. We can also conclude that phenological and seasonal factors drive CPOM input from transition riparian vegetation.
Leaf litter breakdown rates
The peak in leaf litter breakdown in May (beginning of dry season) contrasts that observed in others savanna streams, which typically peak in summer season (December to February) (Sales et al., 2015; Rezende et al., 2016). This result may be explained by the peak in high energy resources such as flowers and fruits (Goncalves et al., 2006a; Franca et al., 2009; Goncalves and Callisto, 2013), as an increase in flower and fruit input can promote environmental enrichment and accelerate leaf litter breakdown (Goncalves et al., 2014b; Alvim et al., 2015a, 2015b). The input of higher-quality organic matter (e.g., flowers and fruits, leaf litter species with high concentrations of nutrients such as nitrogen and phosphorus) can increase the density of decomposers (Ferreira et al., 2014; Cornut et al., 2015; Graca et al., 2015), which consequently consume the lower-quality organic matter (e.g., leaf litter with high concentrations of secondary and structural compounds). These results are the first to indicate flowers and fruits as potential influence in nutrient dynamics tropical streams.
Leaf litter breakdown rates (k ~ 0.016 day–1) show an upper range comparable to other savanna streams (0.007 to 0.016 day–1 in Goncalves et al., 2006b; Goncalves et al., 2007; Moretti et al., 2007a; Goncalves et al., 2012b), and were classified as ‘fast’ for tropical systems according to Goncalves et al. (2014b). However, shredders were not associated with leaf litter breakdown rates. On the other hand, the remaining mass showed a positive relationship with water flow (due to rainfall increase), possibly through physical abrasion (Rezende et al., 2014). The lower abundance of shredders with increased water flow and litter breakdown can be explained by the washing force, which decreases colonization by invertebrates (Ferreira and Graca, 2006; Rezende et al., 2014). Finally, the higher activity of scrapers compared to shredders in the current study generally agrees with results of other studies in savanna streams (Goncalves et al., 2006b; Moretti et al., 2007a). Scrapers can promote the fragmentation of plant tissue by scraping the periphyton growing on the substrate (Goncalves et al., 2006b).
The density (2 to 780 individuals.g–1 in Moretti et al., 2007a; Ligeiro et al., 2010; Goncalves et al., 2012a; Goncalves et al., 2012b; Rezende et al., 2016) and average richness of invertebrates (1 to 22 taxa in Moretti et al., 2007b; Alvim et al., 2015b; Rezende et al., 2016) at our site was low compared to other savanna streams. The most abundant taxa were Tanypodinae, Chironominae, and Orthocladiinae, belonging to the family Chironomidae (Diptera). The dominance of Chironomidae is typical in tropical and temperate streams (Milošević et al., 2012; Biasi et al., 2013; Uieda and Carvalho, 2015) due to their high adaptive capacity for various environmental conditions (Anderson and Ferrington, 2012; Milošević et al., 2012). The low abundance of shredders (higher in August and November) and scrapers is also typical of tropical streams (Boyero et al., 2011, 2012).
The total CPOM (productivity) input in the transition riparian zone was higher than that of other savanna systems, but lower than typical for Atlantic forest; this supports our first hypothesis. The CPOM also shows peaks in the transition period (from dry to rainy season) in response to water stress, supporting our second hypothesis. The higher leaf litter breakdown in May can be explained by the peak in input of higher-quality organic matter (flowers and fruits). The positive relationship between the remaining mass and water flow corroborates our third hypothesis (by increase of physical abrasion). Therefore, assuming that climate factors (e.g., rainfall) directly affect the input and quality of CPOM and leaf litter breakdown rates, changes in the climate can alter functioning of savanna streams. The transition riparian zone (e.g. savanna and rainforest) show intermediate patterns of plant productivity and leaf processing in tropical zones. Leaf litter breakdown was driven by environmental enrichment, indicating that changes in the savanna riparian composition may also alter ecosystem function.