Effects of deforestation on macroinvertebrate diversity ...

2 downloads 0 Views 289KB Size Report
Nov 4, 2002 - 1994, Thorpe & Lloyd 1999, Pierre et al. 2000, Forti et ...... Pierre, J., Ometo, H. B., Martinelli, L. A., Ballester, M. V., Gessner, A.,. Krusche, A. V. ...
Arch. Hydrobiol.

158

3

317–342

Stuttgart, November 2003

Effects of deforestation on macroinvertebrate diversity and assemblage structure in Ecuadorian Amazon streams Berit H. Bojsen1 * and Dean Jacobsen1 With 5 figures, 6 tables and 1 appendix

Abstract: The effects of deforestation on stream macroinvertebrate faunas were studied at twelve sites located in an area of fragmented rainforest in the Ecuadorian Amazon. The most pronounced changes in habitat characteristics with reduced riparian canopy cover were a reduced amount of litter detritus on the streambed and an increased periphyton biomass. Alpha diversity decreased with the degree of deforestation. Beta diversity was also lower in deforested than forested sites, indicating that macroinvertebrate composition among the forested sites were more heterogeneous than among the deforested sites. Total macroinvertebrate density increased with decreasing canopy cover, and with increasing periphyton biomass. The relative density of Ptilodactylidae, Tanypodinae, Euthyplociidae, Libellulidae and Megapodagrionidae were positively related with either canopy cover or litter detritus. A marked effect on the trophic structure of the macroinvertebrate fauna was found. The relative density of collectors decreased with canopy cover and the amount of litter detritus, while the relative density of predators increased. Shredder abundance was low and unrelated with canopy cover. Temporal variability in the macroinvertebrate data were greater in the deforested sites than in the forested sites. Using two-way indicator species analysis (TWINSPAN) and principal components analysis (PCA) riparian canopy cover was found important structuring the macroinvertebrate assemblages. Litter detritus associated with particulate organic material was the main variable related with the PCA ordination axes. Key words: Tropical streams, Ecuador, macroinvertebrates, deforestation.

Introduction Increased human activities during recent history have lead to widespread deforestation in temperate areas. Stream ecologists have studied physical, chem1

Authors’ address: Freshwater Biological Laboratory, University of Copenhagen, Helsingørsgade 51, 3400 Hillerød, Denmark. * Author for correspondence; [email protected] DOI: 10.1127/0003-9136/2003/0158-0317

0003-9136/ 03/0158-0317 $ 6.50

ã 2003 E. Schweizerbart’sche Verlagsbuchhandlung, D-70176 Stuttgart

318

Berit H. Bojsen and Dean Jacobsen

ical and biological effects of removing forest along stream banks, and the importance of riparian forest for stream ecosystems is well established in temperate areas (Sweeney 1993). In temperate regions, changes in stream macroinvertebrate assemblages associated with deforestation have been attributed to changing light levels, temperature regime, nutrient levels, substrate particle size distribution and available food sources (Gurtz & Wallace 1984, Noel et al. 1986, Richards et al. 1993, Sabater et al. 1998). Parallel with increased population levels, industrial and farming activities in tropical countries, tropical deforestation and concomitant disturbance of stream ecosystems has enhanced. Some attention has been paid to this subject (Dudgeon 1994, Henry et al. 1994, Thorpe & Lloyd 1999, Pierre et al. 2000, Forti et al. 2000), but still there is a lack in our knowledge on effects of deforestation on stream habitats and on the structure and function of stream macroinvertebrate faunas. Disturbance through variable flow regimes is a prime factor controlling macroinvertebrate communities in running waters (e.g. Stanford & Ward 1983, Reice et al. 1990). In contrast to most temperate streams, tropical streams are generally highly influenced by regular spates (Covich 1988). Jacobsen & Encalada (1998) suggested that flow related instability of tropical Andes streams was the main feature structuring the invertebrate fauna, while other site characteristics were less important. This leads to the expectation that tropical stream macroinvertebrates might be less affected by changes in habitat characteristics due to deforestation than observed for macroinvertebrates in temperate streams. The macroinvertebrate fauna in streams with highly variable flow regimes is assumed to survive by seeking refugee in sheltered places or burrowing in the hyporheic zone (Lancaster & Hildrew 1993). However, if shelter (e.g. tree roots penetrating the stream banks) is reduced as a consequence of deforestation, tropical macroinvertebrates might be highly sensitive to deforestation. In almost all groups of animals and plants, species richness peaks in the humid tropics near the equator. It has been estimated that 10 – 20 % of the world’s species of many groups of plants and animals inhabit Ecuador (Coloma 1999), although Ecuador covers an area of only 272 km2. Tropical lowland streams usually have a very species rich fauna. This is apparent not only for fish (Lowe-McConnell 1987) but also for the aquatic insects (Jacobsen et al. 1997). Deforestation may therefore potentially lead to an exceptionally high loss of taxonomic and/or functional diversity in Amazonian streams compared to most temperate ones. In Ecuador, large areas of the Amazonian rainforest are still undisturbed by humans, but here like many in other places in Latin America, Africa and Asia, the forest is threatened by increasing human activities. Ecuador has one of the highest rates of deforestation in the world (World Resources Institute 1985, Mecham 2001).

Effects of deforestation on macroinvertebrate diversity

319

The purpose of this study was to investigate the effects of deforestation on diversity, density patterns, trophic structure, temporal variability, and assemblage structure of the macroinvertebrate faunas in small, tropical streams in the Ecuadorian Amazon. The study was conducted in an area of fragmented forest, with streams that had not suffered dramatic changes in habitat characteristics due to deforestation or other human activities. A series of 12 study sites were chosen so that stream reaches both disturbed and undisturbed by deforestation were represented within a gradient of riparian canopy cover. The relationships between macroinvertebrate data and the gradient in canopy cover were measured along with a number of other habitat characteristics.

Methods Study area and sampling sites The study was conducted in the upper Rio Napo drainage basin in the Ecuadorian Amazon. The studied streams were located close to each other within a maximum distance of 12 km, between the coordinates 1ƒ 00¢ N, 77ƒ 40¢ W and 1ƒ 05¢ N, 77ƒ 30¢ W, at an elevation of approximately 400 m at the foothills of the Andes. The climate is equatorial with a mean annual temperature above 24 ƒC and a mean annual rainfall of approximately 5000 mm. A mixture of primary forest, secondary forest and small deforested areas used for cultivation and pastures characterises the area. The study included 12 sites (50-m reaches) in nine headwater streams (1. – 3. order). The study sites were chosen so that half of them represented forested reaches (undisturbed by deforestation at the study site) and the other half deforested reaches (disturbed by deforestation at the study site), resulting in a gradient of canopy cover from 6 % to 82 % among the 12 study sites.

Environmental variables A number of environmental variables, including habitat and catchment characteristics, were measured at each study site (Table 1). Canopy cover was measured by a spherical densiometer (Model-C from Robert E. Lemmons, Forest Densiometers) at six points along the centre of each stream reach and expressed as mean percentages. Each stream reach was categorised in relation to the degree of deforestation of the catchment by use of maps and personal observations. Stream sites characterised as deforested (riparian canopy cover < 50 %) were assigned category 1 if the total catchment were deforested, category 1.5 if half of the catchment was deforested and category 2 the total catchment was forested. Stream sites characterised as forested (riparian canopy cover > 50 %) were assigned category 3 if the total catchment was deforested, category 3.5 if half of the catchment was deforested and category 4 if all of the catchment were forested. Also, each study site were categorised according to distance to the main river, given the numbers 1 (0 –1km), 2 (1– 2 km), 3 (2 – 3 km) and 4 (3 – 4 km).

320

Berit H. Bojsen and Dean Jacobsen

Table 1. Environmental characteristics (means and ranges) measured at deforested stream sites (n = 6) and forested stream sites (n = 6) in the upper Rio Napo drainage basin, Ecuador. Deforested sites

Forested sites

Canopy cover (%) Catchment type (category 1 – 4) Distance to main stream (category 1 – 4) Depth CV Mean depth (cm) Stream width (cm) Mean current velocity (m s – 1) Water conductivity (mS cm – 1) Water temperature (ƒC) pH PO4-P (mg l – 1) NH3-N (mg l – 1) NO3-N (mg l – 1) Particulate organic material (mg cm – 2) Suspended solids (mg l – 1) Periphyton biomass (mg m – 2) Litter detritus on the streambed (%)

23 (6 – 48) 1–2 1–4 0.71 (0.59 – 0.92) 20 (10 – 31) 265 (150 – 435) 0.15 (0.05 – 0.29) 46 (15 – 115) 25 (23 – 26) 6.7 (6.5 – 7.1) 34 (12 – 89) 47 (34 – 51) 54 (8 – 121) 0.7 (0.3 – 1.5) 8.5 (3.0 – 19.2) 42 (17 – 72) 5 (0 – 11)

77 (68 – 82) 3–4 1–4 0.80 (0.44 – 0.99) 13 (4 – 23) 330(135 – 640) 0.12 (0.03 – 0.21) 43 (16 – 111) 23 (21 – 25) 6.2 (5.4 – 6.9) 43 (11 – 73) 50 (44 – 66) 76 (18 – 112) 1.2 (1.0 – 2.5) 6.7 (3.7 – 14.1) 10 (4 – 23) 19 (5 – 35)

Substrate types Sand (%) Gravel/Pebble (%) Cobble/Boulder (%) Coefficient of substrate heterogeneity

20 (3 – 55) 37 (24 – 46) 29 (3 – 43) 0.59 (0.54 – 0.68)

15 (1 – 46) 35 (11 – 54) 25 (0 – 52) 0.58 (0.41 – 0.71)

Habitat types Pool (%) Run (%) Riffle (%)

33 (12 – 48) 34 (22 – 47) 32 (19 – 63)

21 (10 – 38) 41 (15 – 66) 38 (17 – 75)

Mean depth and substrate types were measured by cross-stream transects located at 2 – 4 m intervals (depending of the size of the stream), with sampling points 0.25 m apart. At each sampling point, depth (measured to nearest cm) and substrate type (categorised as leaves (litter detritus), sand, gravel, pebble, cobble and boulder) were measured. Each substrate type was calculated as percent cover of the total stream site (of the total number of sample points). The relative distribution of the habitats, pool, run or riffle was measured at each site. Coefficients of variation in water depth (depth CV) and substrate heterogeneity (Baattrup-Pedersen & Riis 1999) were calculated from the cross-stream transect measurements. Mean current velocity along reaches was determined by applying a salt solution with known conductivity and volume upstream of a 30 – 50 m reach and then recording the conductivity every 5 or 10 second downstream of the reach. Mean current velocity was then calculated as the time elapsed for half of the salt to pass the stream reach divided by the length of the reach.

Effects of deforestation on macroinvertebrate diversity

321

Water conductivity and temperature were measured with a conductivity meter (YSI model 30) and pH with a pH-meter (Radiometer model 203) two or three times during the sampling period. To measure concentrations of phosphate, ammonium and nitrate, six water samples were collected at each site. Phosphate and ammonium were measured spectophotometrically according to Solórzano (1969) and Murphy & Riley (1962), while nitrate was measured on an Alpkem Autoanalyser. Total suspended solid loads were measured by filtering known volumes of stream water through pre-dried Whatman GF/C filters. The filters and associated load were oven-dried to constant weight at 80 ƒC. Coarse and fine particulate organic material (POM) on the streambed was measured from six stones (diameter 8 –10 cm) collected at random from riffles, runs and pools at each study site. A 200- mm mesh net was held downstream each stone while it was lifted quickly into the net. Each stone was brushed carefully. The organic content was determined by drying (80 ƒC, 12 hours) and combusting the material (500 ƒC, 1 hour). POM was calculated as the mass loss upon ignition. Periphytic algal biomass on stone surface was measured as the amount of chlorophyll-a. Twelve stones (diameter 8 –10 cm) were collected randomly at each stream site and frozen within 8 hours of collection before extraction. Pigments were extracted in the dark in 98 % ethanol, and chlorophyll-a concentrations were measured spectrophotometrically. Stone surface area was estimated following the procedure given by Dall (1979): 1.2 ´ (LW + LH + WH) where L, W and H are length, width and height of the stones. Half of the total stone surface area was used in the calculation of area-specific periphyton biomass.

Macroinvert ebrate sampling Benthic macroinvertebrates were sampled during the wet season from May to August and the dry season from November to December in 1999. In each season four surber samples (area 500 cm2, mesh size 200 mm) were taken from moderate to fast flowing riffles and runs, and two core samples (area 200 cm2) were taken from pools at each site. Core samples were collected at 40 – 80 cm water depth on soft substrate and 6 to 8 cm of the upper sediment was sampled. In pools containing coarse substrate, six stones were collected by rapidly lifting them into a 200- mm hand net. The animals caught in the net were combined with those caught by brushing the stone surfaces. In the laboratory samples were hand sorted without magnification. All insects were identified to family level, except for the Chironomidae and the Limonidae that were identified to subfamily level. Macroinvertebrates were quantified as individuals m – 2, according to the relative distribution of habitats. Macroinvertebrate data separated between seasons (wet and dry season) were used for some of the analysis. The macroinvertebrates were assigned to four major functional feeding groups, shredders, predators, filterers and collectors (deposit feeders and grazers) according to Dudgeon (1989, 1994) and Merritt & Cummins (1996).

322

Berit H. Bojsen and Dean Jacobsen

Dat a analysis In the data analysis, both continuous data (gradient analysis) and categorical data (comparing forested and deforested sites) were used. For practical reasons the overall data are presented using a categorical design. The study design allows both types of analysis. In order to estimate taxon richness independent of sample size (number of individuals), Fishers Alpha diversity index was used (Rosenzweig 1997). The relationship between Fishers Alpha and environmental variables were assessed by Spearman rank correlation. Beta diversity was calculated and compared for forested and deforested streams by two indices. Whittakers Beta index described the relation between total number of species and the mean species richness of the stream sites, and Routledges Beta index related the total number of species and the number of species pairs with overlapping distribution (Wilson & Schmida 1984). Alpha diversity is used to describe local diversity (point diversity) while beta diversity describes the change in species composition from site to site. All calculations of diversity are made at family level. However, family richness of insects at individual stream sites is highly correlated to species richness (Bournaud et al. 1996, Wright et al. 1997). When identification to species is not consistently, family richness is assumed to reflect species richness. Relationships between relative density of each taxa and environmental variables (canopy cover, catchment type, leaf cover on streambed and periphyton biomass) were tested using Pearson Product Moment correlation. Data were tested for normality and homogeneity of variance prior to application of the parametric tests. Inter-correlations among the environmental data were assessed similarly. Stream sites were classified by two-way-indicator-species-analysis (TWINSPAN; Hill 1979). Pseudo-species cut level were defined as 0, 2, 5, 10 and 20 on reduced relative density (%) data (square root transformed). Taxa that were found with only one individual sample –1 and/or in only one or two sites were excluded from the data. Significant differences in species relative density and environmental variables between groups were tested using ANOVA. All ordination analyses were performed on square root transformed data. Gradient length and pattern of variation in the taxa data were measured by detrended correspondence analysis (DCA). Ordination methods were chosen according to gradient length and amount of variance captured. Principal components analysis (PCA, linear method) was used to perform the ordination of macroinvertebrate assemblage. Relationships between site scores and environmental variables were assessed by Pearson Product-Moment correlation. The strength of the environmental variables was tested using Redundancy analysis (RDA) with and without selecting a counterpart as covariable. The significance the variables were tested by 199 unrestricted Monte Carlo Permutations.

Effects of deforestation on macroinvertebrate diversity

323

Results Environmental stream charac teristics

All forested stream sites had a high canopy cover ranging from 68 to 82 % (mean 77%), while canopy cover was low at the deforested sites (6 % to 48 %, mean 23 %) (Table 1). Cover of litter detritus on the streambed ranged from 0 % at a deforested site (mean 5 %) to 35 % at a forested site (mean 19 %). Periphyton biomass on stones varied considerably between stream groups but was higher at the deforested sites (mean 42 mg chl m – 2) than at the forested sites (mean 10 mg chl m – 2). Streambeds dominated by either sand, gravel/ pebble or cobble/boulders were represented in both stream groups, but dominance of gravel/pebbles was most frequent in both groups. Mean depth varied from 4 to 31 cm and mean stream width from 135 to 640 cm in all twelve stream sites. Water temperature ranged from 21 to 26 ƒC and was highest in the deforested sites (mean 25 vs. 23 ƒC). No pronounced difference in substrate heterogeneity, depth CV, habitat composition, conductivity or mean current velocity was found between forested and deforested sites. However, a greater range (higher variance) in depth CV, pH, gravel/pebbles and substrate heterogeneity was obvious for the forested sites (Table 1). Spates occurred regularly (weekly) in all sites (personal observation). Aquatic plants were absent, while filamentous algae were common in some of the deforested sites. We observed at high degree of inter-correlation between environmental variables and canopy cover and catchment type were clearly reflected in the amount of litter detritus and periphyton biomass. Correlation coefficients between all environmental variables are listed in Appendix A. Among all environmental variables, canopy cover was positively correlated with catchment type, litter detritus and negatively with periphyton biomass (p < 0.05).

Diversity and abundance

A total of 42 macroinvertebrate taxa were collected (Table 2). Macroinvertebrate density and number of taxa at a site ranged from 343 to 4982 individuals m – 2 and from 18 to 32 taxa site –1. Number of taxa in the forested stream sites ranged from 18 to 28 taxa site –1 and the deforested sites ranged from 21 to 32 taxa site –1. Mean macroinvertebrate density was more that three times greater in the deforested sites (mean = 2598; range = 1498 – 4982) compared with the forested sites (mean = 692; range = 343 –1206) (t-test, p < 0.001). Also, total macroinvertebrate density was strongly correlated with catchment type and canopy cover and modestly with litter detritus and periphyton biomass (p < 0.05) (Fig. 1 & Table 3).

324

Berit H. Bojsen and Dean Jacobsen

Table 2. Densities (individuals m – 2) of macroinvertebrate taxa found in the forested and the deforested stream sites in the upper Rio Napo Drainage basin, Ecuador. Numbers in parentheses are the number of sites where each taxon was found in either forested (n = 6) or deforested (n = 6) sites. Means and ranges (in parentheses) of the total number of taxa and individuals m –2 are shown at the bottom of the table. Functional feeding group is indicated: col = collectors, fil = filterers, pre = predators and shr = shredders. Taxonomic groups Platyhelminthes Turbellaria Nematomorpha Annelida Oligochaeta Hirundinea Mollusca Gastropoda Arthropoda Aracnida Acari Insecta Ephemeoptera

Trichoptera

Plecoptera Odonata

Coleoptera

Heteroptera Megaloptera Lepidoptera Diptera

Forested sites

Deforested sites

Functional feeding group

2.8 (2) 0.8 (2)

41.2 (4) –

pre pre

0.7 (2) 0.2 (1)

6.5 (3) –

col pre

Ancylidae



0.3 (1)

col

Hydracarina

0.6 (1)

1.9 (3)

pre

Baetidae Euthyplociidae Leptophlebiidae Tricorythidae Glossosomatidae Hydropsychidae Hydroptilidae Leptoceridae Odontoceridae Philopotamidae Polycentropodidae Perlidae Gomphidae Libellulidae Calopterygidae Coenagrionidae Megapodagrionidae Elmidae Gyrinidae Hydrophilidae Ptilodactylidae Psephenidae Gerridae Naucoridae Veliidae Corydalidae Pyralidae Chironomidae Chironominae Orthochadiinae Tanypodinae Empididae Psychodidae Simuliidae Ceratopogonidae Limonidae Hexatominae Pediciinae

24.5 (6) 29.8 (3) 41.5 (6) 11.6 (4) – 27.6 (5) 0.9 (3) – 0.5 (1) 1.6 (5) 1.3 (2) 7.2 (5) 1.6 (2) 5.2 (4) 0.5 (2) 4.3 (5) 3.5 (5) 34.2 (6) 1.3 (2) 0.4 (3) 25.4 (6) 7.8 (4) 2.5 (3) 5.9 (3) 2.4 (3) 3.1 (2) 2.2 (6)

158.4 (6) 31.4 (4) 173.7 (6) 226.1 (6) 14.5 (3) 126.1 (6) 52.7 (5) 1.7 (3) 39.4 (4) 10.8 (4) 0.6 (2) 19.0 (6) 4.2 (2) 8.6 (5) 0.3 (1) 11.7 (5) 11.4 (5) 308.4 (6) – 5.6 (2) 9.5 (4) 44.9 (5) 27.3 (4) 7.7 (4) – 5.6 (5) 68.9 (6)

col pre col col col fil col col shr fil pre pre pre pre pre pre pre col pre pre shr col pre pre pre pre col

122.2 (6) 156.1 (6) 59.7 (6) 0.1 (1) 1.9 (2) 29.1 (5) 0.8 (2)

343.2 (6) 576.8 (6) 150.6 (6) 0.3 (1) – 72.9 (6) 5.4 (4)

col col pre pre col fil pre

20.2 (6) 4.0 (4) 39 (18 – 28) 692 (343 – 1206)

23.5 (5) pre 8.7 (4) pre 37 (21 – 32) 2598 (1498 – 4982)

Total number of taxa Mean number of individuals m – 2

Effects of deforestation on macroinvertebrate diversity

325

Fig. 1. Relationship between macroinvertebrate density (individuals m – 2) and Fishers Alpha diversity index, and the four environmental variables canopy cover (%), catchment type, cover of litter detritus on the streambed (%), and periphyton biomass (mg m – 2). See: Table 3 for correlations coefficients. Table 3. Correlation coefficients between total macroinvertebrate density, total number of taxa, and Fishers Alpha diversity index, and environmental variables measured in the upper Rio Napo drainage basin, Ecuador. Significance levels of Pearson Product Moment correlation: * p < 0.05, ** p < 0.01, *** p < 0.001. Canopy cover Catchment type Litter detritus on the streambed Periphyton biomass Particulate organic material Suspended solids Phosphate % Run

Total density

Number of taxa

Fishers Alpha

– 0.76** – 0.83** – 0.62* 0.55* – 0.49 – 0.11 – 0.29 – 0.38

– 0.35 – 0.38 – 0.06 0.33 – 0.68* – 0.64* – 0.62* – 0.76**

0.66* 0.69* 0.42 – 0.61* 0.29 – 0.33 0.02 0.06

The alpha diversity was positively related with canopy cover and catchment type, and negatively with periphyton biomass (p < 0.05) (Fig. 1 & Table 3). Total number of taxa was not significantly related with any of the four “land use” variables (canopy cover, catchment type, litter detritus and periphyton biomass), although it was negatively correlated with particulate organic material (Table 3). Number of taxa was positively related with the number of individuals (r = 0.75, p < 0.05). The beta diversity indices indicated that beta diversity was greater in the forested sites (Whittakers Beta: 0.66, Routledges Beta: 0.37) compared with the deforested sites (Whittakers Beta: 0.13, Routledges Beta: 0.05). This indi-

326

Berit H. Bojsen and Dean Jacobsen

cated lower heterogeneity among macroinvertebrate faunas in the deforested sites. Specific patterns in macroinvertebrat e data

Diptera, Ephemeroptera and Coleoptera dominated macroinvertebrate densities in both forested and deforested sites. The families Chironomidae and Elmidae accounted for 60 % and 53 % of the total number of individuals m – 2 in the forested and the deforested sites, respectively. The relative densities (%) of sixteen taxa were correlated with one or more of the “land use” variables canopy cover, catchment type, litter detritus and periphyton biomass (Table 4). The relative densities of nine taxa were positively correlated with canopy cover, catchment type and/or litter detritus or negatively correlated with periphyton biomass. These taxa were Ptilodactylidae, Hexatominae, Tanypodinae, Euthyplociidae, Veliidae, Pyralidae, Libellulidae, Megapodagrionidae and Philopotamidae. The taxa Elmidae, Psephenidae, Tricorythidae, Gerridae, Glossosomatidae, Hydroptilidae and Oligochaeta increased their relative density with decreasing canopy cover or with increasing periphyton biomass, their relative abundances were thus positively affected by deforestation. No single taxon clearly separated the two stream groups, i.e. no single taxon common in one group was completely missing in the other. Table 4. Correlation coefficients between canopy cover, catchment type, litter detritus and periphyton biomass and the relative density of several macroinvertebrate taxa from the upper Rio Napo drainage basin, Ecuador. Only taxa that were significantly related to one of the four environmental variables are shown. Significance levels of correlation: * p < 0.05, ** p < 0.01 and *** p < 0.001. Canopy cover Elmidae Ptilodactylidae Psephenidae Hexatominae Tanypodinae Euthyplociidae Tricorythidae Gerridae Veliidae Pyralidae Libellulidae Megapodagrionidae Glossosomatidae Hydroptilidae Philopotamidae Oligochaeta

– 0.84*** – – – – – – 0.70* – – – 0.58* – 0.59* – 0.68* – – 0.60*

Catchment type – 0.83*** – – – – – 0.70* – – – 0.76** – – – 0.64* – – –

Litter detritus – 0.64* – 0.56* – 0.71** 0.68* – – – – 0.61* 0.80** – 0.66* – – –

Periphyton biomass 0.66* – 0.68* – 0.59* – – – – – 0.59* – – – – – – 0.70* –

Effects of deforestation on macroinvertebrate diversity

327

Functional feeding groups

Absolute densities (individuals m – 2) and relative densities (%) of the functional feeding groups were related with catchment type (Fig. 2). Marked decreases in absolute densities with catchment type were apparent for collectors (r = – 0.82, p < 0.001), filterers (r = – 0.72, p < 0.01) and predators (r = – 0.73, p < 0.01). The absolute densities of collectors were furthermore negatively correlated with both canopy cover (r = – 0.75) and litter detritus (r = – 0.66) (p < 0.05) and positively with periphyton biomass (but not significantly, r = 0.53, p = 0.075). The absolute density of filterers and predators were modestly negatively correlated with canopy cover (r = – 0.66 and – 0.62, respectively), and predators were also negatively correlated with particulate organic material

Fig. 2. Relationship between mean absolute and relative density of four functional feeding groups and catchment type (type 1 (n = 2), type 1.5 (n = 4), type 3 (n = 1), type 3.5 (n = 3), type 4 (n = 2)). col = collectors, fil = filterers, pre = predators and shr = shredders.

328

Berit H. Bojsen and Dean Jacobsen

Fig. 3. A: Mean macroinvertebrate density in wet and dry season for both forested and deforested sites. The differences in mean density between rain and dry season were significant for the deforested sites (t-test, p < 0.05). B: Relationship between total density in wet and dry season for each site. Points above the 1 : 1 line show that density was higher in dry than rain season.

(r = – 0.59) (p < 0.05). The absolute density of shredders was not correlated with any of the four “land use”. The relative densities of the functional feeding groups were apparent between the catchment types. The relative density of collectors were negatively correlated with catchment type (r = 0.63, p < 0.05) and decreased from approximately 70 % to 40 % from catchment type 1 to 4.

Effects of deforestation on macroinvertebrate diversity

329

The relative densities of collectors were furthermore significantly correlated with litter detritus (r = – 0.78), but non-significantly correlated with periphyton biomass (r = 0.46, p = 0.13). The relative density of predators was lower at catchment type 1 and 1.5 compared with type 3 – 4, but the difference was not significant. Instead, the relative density of predators was very strongly correlated with litter detritus (r = 0.91, p < 0.001). Similar to the absolute density of shredders, the relative density of shredders unrelated with any of the four “land use”.

Temporal dissimilarity

The total macroinvertebrate density differed more between wet and dry seasons in deforested than in forested sites (Fig. 3 A). While there were a small, not significantly difference in density between seasons in the forested sites (´ 1.4), there was a marked effect in the deforested sites; mean density was 1.8 times greater in the dry season compared with the wet season (t-test, p < 0.05). The more pronounced effect of season in the deforested sites is also obvious in Fig. 3 B, where the relationship between seasons is shown for each site. In both forested and deforested sites, four out of six sites had a higher density in the dry season than in the wet season (points above the 1: 1 line).

Multivariate patterns – TWINSPAN

Classification of macroinvertebrate assemblages by TWINSPAN identified by first division two groups of sites that differed significantly with respect to canopy cover and catchment type (t-test, p < 0.001) (Fig. 4). The two groups did not differ significantly with respect to any other environmental variables (p > 0.05). Further division of sites, gave no meaning as groups containing one site were identified. TWINSPAN group 1 contained seven sites, included were all sites classified as forested and one classified as deforested. The deforested site included in group 1 had a relatively high canopy cover (35 %) and was classified in catchment type 1.5. Only deforested sites were included in TWINSPAN group 2. Taxa responsible for the first division were primarily Tricorythidae, Odontoceridae and Corydalidae. These three indicator species had either higher relative density or higher relative occurrence in number of sites in group 2. In accordance with TWINSPAN analysis, a number of taxa had preference for group 1; Hydrophilidae, Veliidae, Orthochladinae, Simuliidae and Hexatoninae, while Baetidae, Glossosomatidae, Leptoceridae, Oligochaeta and Turbellaria were biased towards group 2. Furthermore, the relative density of Ptilodactylidae was significantly higher in group 1 and that of Pyralidae in group 2.

330

Berit H. Bojsen and Dean Jacobsen

Fig. 4. Mean (with SE) environmental variables between TWINSPAN groups classified on the basis of macroinvertebrate assemblage structure, and significance level of t-tests. Only variables showing significant variance between groups are shown.

Multivariate patterns – Ordination

Ordination by PCA resulted in a gradient length of 1.2 SD, indicating that taxa turn-over was at a range where linear response models are most suitable. An unconstrained ordination method was used as the number of environmental variables highly exceeded the number of samples. The eigenvalues of the first two PCA ordination axes (l1 = 0.26, l2 = 0.18) explained 43 % of the cumulative variance in the taxa data (Table 5). The relationship between ordination axes and environmental variables is shown in Table 5. Three environmental variables were negatively correlated with PCA axis 1 (particulate organic material, water temperature and conductivity) and two were positively correlated (stream width and nitrate concentration). Gravel/pebble and pH were negatively correlated with PCA axis 2, and litter detritus and substrate heterogeneity were positively correlated with this axis. The macroinvertebrate distribution along the environmental gradients is shown in Figure 5. At the left side of the diagram are taxa associated with stream sites with high water temperature,

Effects of deforestation on macroinvertebrate diversity

331

Table 5. Summary of the results of PCA on the macroinvertebrate assemblage data from stream in the upper Rio Napo in Ecuador, and correlation coefficients between PCA axis 1 and 2 and the environmental variables. Significance levels of correlation: * p < 0.05, ** p < 0.01 and *** p < 0.001. Eigenvalues Cumulative percentage variance of species data Correlation coefficients with axes Litter detritus on the streambed Particulate organic matter Stream width Water temperature Conductivity pH Nitrate concentration Substrate heterogeneity Gravel/Pebble

Axis 1

Axis 2

0.26 25.6

0.18 43.2

– 0.39 – 0.59* 0.67* – 0.61* – 0.78** – 0.12 0.64* 0.08 0.08

0.61* – 0.26 – 0.13 0.00 – 0.37 – 0.74** 0.15 0.82** – 0.66*

conductivity, particulate organic material and low stream width and nitrate concentration positioned, those are e.g. Philopotamidae, Leptophlebiidae, Oligochaeta and Hydropsycidae. Positioned to the right side of the diagram and associated with high nitrate concentration and high stream width are e.g. Psephenidae, Perlidae, and Elmidae. At the upper part on the diagram are Euthyplocidae, Megapodagrionidae and Simuliidae positioned, indicating association with sites with high litter detritus and substrate heterogeneity and low pH and gravel/pebble. The strength of the environmental variables significantly correlated with PCA axes were tested (Table 6). When taxa data were constrained only to one variable, the variables with the largest significantly explanatory power were conductivity (19.7 %), particulate organic material (15.9 %), stream width (15.2 %), litter detritus (14.3 %) and water temperature (13.8 %). The variance explained by substrate heterogeneity, nitrate concentration, pH and gravel/ pebble were not significant (p > 0.10). In partial RDA, the effects of all significant variables were partialled out from litter detritus. Those analyses showed that litter detritus lost explanatory power (11.6 % – 12.8 %, p > 0.10 – 0.185) when particulate organic material, substrate heterogeneity, gravel/pebble and nitrate concentration was partialled out. Also, substrate heterogeneity, gravel/pebble and nitrate concentration lost explanatory power (9.0 % – 12.1 %, p > 0.125 – 0.330) when litter detritus was partialled out. The partial RDA showed that the variance caused by difference in litter detritus could not be attributed to effects of stream width, water temperature, pH and conductivity, while it was not possible to decide whether litter detritus, particulate organic material, substrate heterogeneity, gravel/pebble or nitrate concentration were

332

Berit H. Bojsen and Dean Jacobsen

Fig. 5. Principal components analysis (PCA) biplot of macroinvertebrate assemblages from streams in the upper Rio Napo drainage basin, Ecuador. Stream sites were indicated by circles of different size increasing according to riparian canopy cover. Environmental gradients are indicated. Taxa indicated by numbers are 1: Corydalidae, 2: Ceratopogonidae, 3: Naucoridae, 4: Hexatominae, 5: Glossosomatidae, 6: Gerridae, 7: Pyralidae, 8: Hydroptilidae, 9: Coenagrionidae, 10: Leptoceridae, 11: Libellulidae, 12: Tanypodinae, 13: Veliidae, 14: Hydrophilidae, 15: Megapodagrionidae. Ubiquitous taxa are not shown; those were Odontoceridae, Gomphidae and Turbellaria.

the most important variable. Litter detritus and particulate organic material, substrate heterogeneity, gravel/pebble and nitrate concentration together explained a significant portion of the total variance in the taxa data. However, substrate heterogeneity, gravel/pebble and nitrate concentration was non-significant in the RDA; therefore, we assume that litter detritus and particulate organic material together had the highest strength.

Discussion Environmental variables

The most pronounced changes in the stream environment due to reduced riparian canopy cover were a higher biomass of periphyton and a lower amount of litter detritus on the streambed. The negative relationship between canopy cover and periphyton biomass caused by higher insolation agree with our ex-

Effects of deforestation on macroinvertebrate diversity

333

Table 6. Summary of partial RDA of macroinvertebrate assemblages from streams in the upper Rio Napo drainage basin, Ecuador. Percentage variance explained by selected environmental variables before and after fitting covariables and the ratio of the first constrained eigenvalue (l1) to the unconstrained eigenvalue (l2). The selected environmental variables are those that were significantly correlated with PCA axis 1 and axis 2 (Pearson Product-Moment, p < 0.05). Gradient lengths of the environmental variables in partial DCCA fell in the range from 0.93 to 1.53, why a linear model (RDA) was used. p = significance level of 199 unrestricted Monte Carlo permutations. Variable Conductivity POM Stream width Litter detritus Water temperature Substrate heterogeneity Nitrate concentration pH Gravel/pebble Litter detritus Litter detritus Litter detritus Litter detritus Litter detritus Litter detritus Litter detritus Litter detritus POM Substrate heterogeneity Gravel/pebble Nitrate concentration Water temperature pH Stream width

Covariable

Stream width Water temperature pH Conductivity POM Substrate heterogeneity Gravel/pebble Nitrate concentration Litter detritus Litter detritus Litter detritus Litter detritus Litter detritus Litter detritus Litter detritus

RDA Variance (%)

RDA l1/l2

RDA p

19.7 15.9 15.2 14.3 13.8 13.7 12.8 12.4 11.7 12.5 14.2 16.1 13.6 14.1 12.8 11.6 12.0 14.8 12.1 9.0 10.5 13.7 12.4 13.4

1.02 0.76 0.82 0.61 0.73 0.54 0.68 0.49 0.46 0.68 0.77 0.82 0.84 0.67 0.57 0.49 0.65 0.75 0.42 0.38 0.57 0.74 0.49 0.72

0.005 0.025 0.030 0.050 0.045 0.100 0.105 0.105 0.150 0.040 0.045 0.010 0.040 0.065 0.100 0.185 0.130 0.055 0.125 0.330 0.235 0.075 0.105 0.075

pectations and has been described for other tropical streams as well (Dudgeon & Chan 1992). Litter detritus was, besides being directly correlated with canopy cover and catchment type, also correlated with mean depth and depth CV. These depth variables probably reflected hydraulic characters that determined the retention of leaves on the streambed. This was evident in one of the forested site that had a high canopy cover (68 %) but only 5 % cover of litter detritus. That site had a low mean depth, low depth CV and low cover of cobble/boulders on the streambed, but a high current velocity and therefore a low retention of leaves. In this case, low retention capacity was related to low substrate stability, and litter detritus might not only reflect canopy cover and

334

Berit H. Bojsen and Dean Jacobsen

retention capacity, but also to some degree substrate stability. Lack of substantial changes in habitat characteristics in our study was probably due to local climatic and geological factors and to the absence of massive deforestation in the area.

Abundance and diversity

Significant increases in total macroinvertebrate density in streams affected by reduced riparian forest is well known and reported for temperate streams (Hawkins et al. 1982, Brehmer & Hawkins 1986, Friberg et al. 1997, Noel et al. 1986, Harding et al. 1998). In our study high factorial increases in mean density were observed for several taxa [e.g. Baetidae (´ 6), Psephenidae (´ 6), Elmidae (´ 9) and Tricorythidae (´ 20)]. A positive and food related association between macroinvertebrate abundance and periphyton biomass has earlier been described for tropical streams (Dudgeon 1988, 1989, Dudgeon & Chan 1992). We found that the relationships between macroinvertebrate density and periphyton biomass and litter detritus were clearly associated with canopy cover and catchment type. Clear relationship between riparian canopy cover and periphyton biomass and/or detritus is not always evident (e.g., Dudgeon 1994). Alpha diversity was clearly reduced by deforestation. Lower alpha diversity at the deforested sites was not associated with lower number of taxa but was due to high densities pulling values for alpha diversity down. Lower macroinvertebrate diversity in streams affected by deforestation has been reported for temperate (Clenaghan et al. 1998, Harding et al. 1998) as well as for tropical streams (Giller & Twomey 1993). The lower beta diversity found among the deforested relative to the forested sites was due to taxa that were more abundant in deforested sites and therefore caused lower variance in taxa richness and lower heterogeneity in taxa composition. Even though we did not find pronounced differences in mean values of environmental variables with canopy cover (except for litter detritus, periphyton biomass and water temperature), we believe that decreased heterogeneity in habitats with reduced canopy cover explained the greater homogeneity of the fauna observed for the deforested sites. We are not aware of other studies showing loose of heterogeneity among streams in deforested areas and find the result important.

Temporal variability

The greater temporal variability in macroinvertebrate density observed in the deforested sites than in the forested sites suggests that the macroinvertebrate

Effects of deforestation on macroinvertebrate diversity

335

fauna in the deforested sites were more unstable, and probably had low resistance to disturbance in comparison to the fauna in the forested sites. We believe that the spates occurring regularly in these streams might have a more pronounced effect in the deforested sites, either directly because variation in discharge might be greater, or indirectly because fewer flow refuges might be available, or the competition for them higher. This should be seen in accordance with the “catastrophe avoidance” model within the “flow-refugia” hypothesis (Robert son et al. 1995). According to this model, a disturbance will result in an active or passive relocation of refuges into habitat patches that afford protection and a subsequent redistribution. However, we found no evidence of refuges being less available in the deforested sites, e.g. lower substrate heterogeneity or lower depth CV. We did not make any direct measurements of refuges, but it seems reasonable to assume that at least along the banks of the forested sites the availability of refuges was higher than along deforested banks where no tree roots were penetrating the streambed. Nevertheless, substrate stability and composition probably was important understanding the observed temporal variability in the macroinvertebrate faunas. Trophic structure

As described for tropical streams in South America (Villalobos et al. 1997), Africa (Tumwesigye et al. 2000) and Asia (Dudgeon 1988, Dudgeon 1994, Yule 1996) we found that shredders accounted only for a small part for the macroinvertebrate fauna compared to the predictions of the River Continuum Concept (Vannote et al. 1980), and to what has been described as typical for small temperate streams (e.g. Haefner & Wallace 1981, Cummins et al. 1989, Reed et al. 1994). The lack of relationship between shredder abundance and riparian canopy cover, catchment type and litter detritus suggest that insect shredders play a minor role in decomposing leaf litter in these streams. Poor representation of shredders in stream faunas has also been found in New Zealand, where low retention of detritus has been proposed as an explanation for this (Linklater 1995, Friberg et al. 1997). The high frequency of spates typical of many tropical streams makes it tempting to turn to the same explanation for the lack of relationship between riparian vegetation and shredders observed in tropical streams. However, we found low shredder abundance even in streams with high litter detritus, as did Dudgeon (1988, 1994) in Hong Kong headwater streams with relatively high standing crop of detritus. Dudgeon (1982) showed that leaf litter exported by spates in a tropical stream was replaced by transport from upstream reaches or from the banks. This was confirmed by our personal observations. Inhibition of shredders by higher levels of toxic condensed tannins in tropical plant leaves has been proposed as a another possible explanation for the poor shredder fauna usually found in trop-

336

Berit H. Bojsen and Dean Jacobsen

ical streams (Stout 1989). Yule (1996) suggested that most of the shredding in tropical rainforest might occur in the canopy and that most of the leaf litter that enters the stream is already broken down. This does not correspond with our finding of a high number of intact leaves in the forested stream, which potentially could support a rich shredder fauna. Graça et al. (2001) showed that tropical shredders exhibit the same basic patterns of food exploitation as temperate shredders. Consequently, the role of shredders in stream detritus dynamics may well be applicable to tropical streams. This suggests that the leaves were degraded mainly by microbes or that non-insects, e.g. shrimps and crabs that are very common in these streams, have taken over the role of the insect shredders. Leaf decomposition by microbes seems relatively more important in tropical streams because the microbes are physiologically more able to maintain optimal metabolic rates in warmer water (Irons et al. 1994). The increase in the density of collectors with decreasing canopy cover seemed to be associated with the increase in periphyton biomass (as expected), even though the relationship was not statistically significant. However, both absolute and relative density of collectors were strongly correlated with canopy cover. According to the River Continuum Concept, predatory invertebrates change little in relative dominance with stream order or with the importance of riparian vegetation. In contrast, we found that the relative density of predators increased with litter detritus. However, hydraulic conditions rather than canopy cover may explain this, e.g. it seems possible that substrate and prey abundance were more stabile in streams with high litter detritus. This could account for the observed relationship between predator abundance and leaf cover. Reduced riparian canopy cover associated with deforestation markedly changed the functional organization of the macroinvertebrate fauna. Assemblage structure

Our results suggest that major patterns in the macroinvertebrate assemblages were regulated by variables that either were directly or indirectly related with “land use”, while water conductivity was also important. We think that litter detritus and particulate organic material together reflected both direct effects by serving as food and substrate for the macroinvertebrates and indirect effects that probably were related to hydraulic characters of the streams. This was supported by the fact that litter detritus was correlated with mean depth and depth CV. This indicated that retention was explicit in the explanatory power of particulate organic material and litter detritus. It is well documented that retention of organic material in streams depends on discharges and substrate roughness (Scarsbrook & Townsend 1994, Kishi et al. 1999). Riparian vegetation plays an important double role by its direct input of organic material

Effects of deforestation on macroinvertebrate diversity

337

and by providing retention structures such as roots, submerges branches and dead zones (Ehrman & Lamberti 1992, Maridet et al. 1995). Environmental gradients related to macroinvertebrate assemblage structure in this study were similar to those related to assemblage structure in temperate regions (e.g. Tate & Heiny 1995, Williams et al. 1997, Reece & Richardson 2000, Brown & May 2000) showing that the chemical environment, the physical environment and land use are of importance.

Conclusion Patterns in macroinvertebrate density, diversity, functional organisation, and community structure were mainly related to riparian canopy cover, catchment type, litter detritus, conductivity and hydraulic conditions. Deforestation influenced the macroinvertebrate fauna by leading to (1) higher overall density, (2) lower alpha and beta diversity (3), changed functional organisations to higher dominance of collectors and (4) higher temporal variability. Acknow ledgements The research described in this paper was funded by the Danish Council for Development Research, Danish Ministry of Foreign Affairs, DANIDA (grant number 90 880). We gratefully acknowledge the Fundación Jatun Sacha for providing accommodation in the rainforest and assistance in the field, the Departamento de Ciencias Biologicas de Pontificia Católica del Ecuador for providing laboratory facilities and several students from there for assistance with laboratory and fieldwork. We are indebted to Klaus Brodersen for assistance with data analysis.

Referenc es Baattrup-Pedersen, A. & Riis, T. (1999): Macrophyte diversity and composition in relation to substratum charactistics in regulated and unregulated Danish streams. – Freshwat. Biol. 42: 375 – 385. Bournaud, M., Cellot , B., Richoux, P. & Berrahou, A. (1996): Macroinvertebrate community structure and environmental characteristics along a large river: congruity of patterns for identification to species or family. – J. N. Amer. Benthol. Soc. 15: 232 – 253. Brehmer, D. J. & Hawkins, C. (1986): Effects of overhead canopy on macroinvertebrate production in a Utah stream. – Freshwat. Biol. 16: 287– 300. Brown, L. R. & May, J. T. (2000): Macroinvertebrate assemblage on woody debris and their relations with environmental variables in the lower Sacramento and San Joaquin river drainage, California. – Environmental Monitoring and Assessment 64: 311– 329. Clenaghan, C., Giller, P. S, O’Halloran, J. & Hernan, R. (1998): Stream macroinvertebrate communities in a conifer-afforested catchment in Ireland: relationships to physico-chemical and biotic factors. – Freshwat. Biol. 40: 175 –193.

338

Berit H. Bojsen and Dean Jacobsen

Coloma, F. (1999): Ecuador – Pais de la Megadiversidad. Imprenta Mariscal. Covich, A. P. (1988): Geographical and historical comparisons of neotropical streams: biotic diversity and detrital processing in highly variable habitats. – J. N. Amer. Benthol. Soc. 7: 361– 386. Cummins, K. W., Wilzbach, M. A., Gates, D. M., Perry, J. B. & Taliaferro, W. B. (1989): Shredders and riparian vegetation. – BioScience 39: 24 – 30. Dali, P. C. (1979): A sampling technique for littoral stone dwelling organisms. – Oikos 33: 106 –112. Dudgeon, D. (1982): Sparial and seasonal variation in the standing crop of periphyton and allochthonous detritus in a forest stream in Hong Kong with notes on the magnitude and fate of riparian leaf fall. – Arch. Hydrobiol. Suppl. 64: 189 – 220. – (1988): The influence of riparian vegetation on macroinvertebrate community structure in four Hong Kong Streams. – J. Zool. 216: 609 – 627. – (1989): The influence of riparian vegetation on the functional organisation of four Hong Kong stream communities. – Hydrobiologia 179: 183 –194. – (1994): The influence of riparian vegetation on macroinvertebrate community structure and functional organisation in six New Guinea streams. – Hydrobiologia 291: 65 – 85. Dudgeon, D. & Chan, I. K. K. (1992): An experimental study of the influence of periphytic algae on invertebrate abundance in a Hong Kong stream. – Freshwat. Biol. 27: 53 –73. Ehrman, P. T. & Lambert i, G. A. (1992): Hydraulic and particulate matter retention in a 3rd-order Indiana stream. – J. N. Amer. Benthol. Soc. 11: 341– 349. Forti, M. C., Boulet , R., Melfi, A. J. & Neal, C. (2000): Hydrogeochemistry of small catchment in northeastern Amazonia: A comparison between natural with deforestation parts of the catchment (Serra do Navio, Amapá State, Brazil). – Water Air Soil Poll. 118: 263 – 279. Friberg, N., Winterbourn, M. J., Shearer, K. A. & Larsen, S. E. (1997): Benthic communities of forest streams in the South Island, New Zealand: effects of forest type and location. – Arch. Hydrobiol. 138: 289 – 306. Giller, P. S. & Twomey, H. (1993): Benthic macroinvertebrate community organisation in two contrasting rivers – between-site differences and seasonal patterns. – Biol. Environ. 93 B: 115 –126. Graça, M. A. S., Cressa, C., Gessner, M. O., Feio, M. J., Callies, K. A. & Barrios, C. (2001): Food quality, feeding preferences, survival and growth of shredders from temperate and tropical streams. – Freshwat. Biol. 46: 947– 957. Gurtz, M. E. & Wallace, J. B. (1984): Substrate mediated response of stream invertebrates to disturbance. – Ecology 65: 1556 –1569. Haefner, J. D. & Wallace, J. B. (1981): Shifts in aquatic insect populations in a first order southern Appalachian stream following a decade of old field succession. – Can. J. Fish. Aquat. Sci. 38: 353 – 359. Harding, J. S., Benfield, E. F., Bolstad, P. V., Helfman, G. S. & Jones, E. B. D. (1998): Stream biodiversity: The ghost of land use past. – Proc. Natl. Acad. Sci. 95: 14843 –14847. Hawkins, C. P., Murphy, M. L. & Anderson, N. H. (1982): Effects of canopy, substrate composition, and gradient on the structure of macroinvertebrates communities in cascade range streams in Oregon. – Ecology 63: 1840 –1856.

Effects of deforestation on macroinvertebrate diversity

339

Henry, R., Uiedo, V. S., Alfonso, A. A. de O. & Kikuchi, R. M. (1994): Input of allochthonous matter and structure of fauna in a Brazilian headstream. – Verh. Internat. Verein. Limnol. 25: 1866 –1870. Hill, M. O. (1979): TWINSPAN – A FORTRAN program for arranging multivariate data in an ordered two-way table by classification of the individuals and attributes. – Section of Ecology and Systematic, Cornell University, Ithaca, New York. Irons, J. G., Oswood, M. W., Stout, R. J. & Pringle, C. M. (1994): Latitudinal patterns in leaf litter breakdown: is temperature really important? – Freshwat. Biol. 32: 401– 411. Jacobsen, D. & Encalada, A. (1998): The macroninvertebrate fauna of Ecuadorian highland streams in the wet and dry season. – Arch. Hydrobiol. 142: 53 –70. Jacobsen, D., Schult z, R. & Encalada, A. (1997): Structure and diversity of stream invertebrate assemblages: the influence of temperature with altitude and latitude. – Freshwat. Biol. 38: 247– 261. Kishi, C., Nakamura, F. & Inoue, M. (1999): Budget and retention of leaf litter in Horonai stream, southwestern Hokkaido, Japan. – Jpn. J. Ecol. Tokyo 49: 11– 20. Lancaster, J. & Hildrew, A. G. (1993): Flow refugia and the microdistribution of lotic macroinvertebrates. – J. N. Amer. Benthol. Soc. 12: 385 – 393. Linklat er, W. (1995): Breakdown and detritivore colonisation of leaves in three New Zealand streams. – Hydrobiologia 306: 241– 250. Lowe-McConnel, R. H. (1987): Ecological studies in tropical fish communities. – Cambridge University Press, Cambridge. Maridet, L., Wasson, J. G., Philippe, M. & Amoros, C. (1995): Benthic organic matter dynamics in three streams: Riparian vegetation or bed morphology control. – Arch. Hydrobiol. 132: 415 – 425. Mecham, J. (2001): Causes and consequences of deforestation in Ecuador. – Centro de Investigacion de los Bosques Tropicales – CIBT. Merrit t, R. W. & Cummins, K. W. (1996): An introduction to the aquatic insects of North America, 3rd ed. – Kendall/Hunt Publishing Co., Dubuque. Murphy, J. & Riley, J. P. (1962): A modified single solution method for the determination of phosphate in natural waters. – Analytica Chemira Acta 27: 21– 26. Noel, D. S., Martin, C. W & Federer, C. A. (1986): Effects of forest clearcutting in New England on stream macroinvertebrates and periphyton. – Environmental Management 10: 661– 670. Pierre, J., Ometo, H. B., Martinelli, L. A., Ballest er, M. V., Gessner, A., Krusche, A. V., Victoria, R. L. & Williams, M. (2000): Effects of land use on water chemistry and macroinvertebrates in two streams of the Piracicaba river basin, south-east Brazil. – Freshwat. Biol. 44: 327– 337. Reece, P. F. & Richardson, J. S. (2000): Benthic macroinvertebrate assemblage of coastal and continental streams and large rivers of southwestern British Columbia, Canada. – Hydrobiologia 439: 77– 89. Reed, J. L., Campbell, I. C. & Bailey, P. C. E. (1994): The relationship between invertebrate assemblage and available food at forest and pasture sites in three southeastern Australian streams. – Freshwat. Biol. 32: 641– 650. Reice, S. R., Wissmar, R. C. & Naiman, R. J. (1990): Disturbance regimes, resilience, and recovery of animals communities and habitats in lotic ecosystems. – Environmental Management 14: 647– 659.

340

Berit H. Bojsen and Dean Jacobsen

Richards, C., Host, G. E. & Arthur, J. W. (1993): Identification of predominant environmental factors structuring stream macroinvertebrate communities within a large agricultural catchment. – Freshwat. Biol. 29: 285 – 294. Robert sen, A. L., Lancaster, J. & Hildrew, A. G. (1995): Stream hydraulics and the distribution of microcrustacea: a role for refugia? – Freshwat. Biol. 33: 469 – 484. Rosenzweig, M. L. (1997): Species diversity in space and time. – Cambridge University Press. Sabat er, S., Butturini, A., Muñoz, I., Ronani, Wray, J. & Sabater, F. (1998): Effects of removal of riparian vegetation on algae and heterotropha in a Mediterranean stream. – J. Aquat. Ecosystem Stress and Recovery 6: 129 –140. Scarsbrook, M. R. & Townsend, C. R. (1994): The role of grass leaf litter in streams draining tussock grassland in New Zealand: retention, food supply and substrate stability. – Freshwat. Biol. 32: 429 – 443. Solórzano, L. (1969): Determination of ammonia in natural waters by the phenolhypochlorite method. – Limnol. Oceanogr. 14: 799 – 801. Stanford, J. A. & Ward, J. V. (1983): Insect diversity as a function of environmental variability and disturbance in stream systems. – In: Barnes, J. R. & Minshall, G. W. (eds.): Stream Ecology, Application and Testing of General Ecological Theory. – Plenum Press, New York, pp. 265 – 278. Stout, R. T. (1989): Effects of condensed tannins on leaf processing in mid-latitude tropical streams: a theoretical approach. – Can. J. Fish. Aquat. Sci. 46: 1097–1105. Sweeney, B. W. (1993): Effects of streamside vegetation on macroinvertebrate communities of White Clay Creek in Eastern North America. – Proc. The Academy of Natural Science of Philadelphia 144: 291– 340. Tate, C. M. & Heiny, J. S. (1995): The ordination of benthic invertebrate communities in the south Platte River Basin in relation to environmental factors. – Freshwat. Biol. 33: 439 – 454. Thorpe, T. & Lloyd, B. (1999): The macroinvertebrate fauna of St. Lucia elucidated by canonical correspondence analysis. – Hydrobiologia 400: 195 – 203. Tumwesigye, C., Yusuf, S. K. & Makanga, B. (2000): Structure and composition of benthic macroinvertebrates of a tropical forest stream, River Nyamweru, western Uganda. – Afr. J. Ecol. 38: 72 –77. Vannote, R. L., Minshall, G. W, Cummins, K. W., Sedell, J. R. & Cushing, C. E. (1980): The river continuum concept. – Can. J. Fish. Aquat. Sci. 37: 130 –137. Villalobes, M., Guzman-Danilo, De J., Urribarri, P., Lopez, C. & Rincon, J. E. (1997): Leaf litter decomposition in a tropical intermittent stream (Zulia State, Brazil). – Boletin del Centro Investigaciones Biologicas Universidad del Zulia 31: 121–134. Williams, D. D., Williams, N. E. & Cao, Y. (1997): Spatial differences in macroinvertebrate community structure in spring in southeastern Ontario in relation to their chemical and physical environments. – Can. J. Zool. 75: 1404 –1414. Wilson, M. V. & Schmida, A. (1984): Measuring beta diversity with presenceabsence data. – J. Ecol. 72: 1055 –1064. World Resources Institute (1985): Tropical forest: A call to Action. – Washington, D.C. Wright, J. F., Moss, D. & Furse, M. T. (1997): Macroinvertebrate richness at running-water sites in Great Britain: a comparison of species and family richness. – Verh. Internat. Verein. Limnol. 26: 1174.

Effects of deforestation on macroinvertebrate diversity

341

Yule, C. M. (1996): Trophic relationships and food webs of the benthic invertebrate fauna of two aseasonal tropical streams on Bougainville Island, Papua New Guinea. – J. Tropic. Ecol. 12: 517– 534. Submitted: 4 November 2002; accepted: 13 July 2003.

Catchment type Litter detritus (%) Periphyton (mg m – 2 ) POM (mg cm – 2) Distance Vmean (m s – 1) Stream width (cm) Mean depth (cm) Depth CV Temperature (ƒC) Conductivity (mS cm – 1) PH SS (mg l – 2) PO4P (mg l – 1) NH3 N (mg l – 1) NO3 N (mg l – 1) Substrate heterogeneity Gravel/Pebbles (%) Cobble/Boulder (%) Sand (%) Pool (%) Run (%) Riffle (%)

Canopy cover

–0.91*** 0.68* –0.57* 0.21 0.36 –0.19 0.24 –0.36 0.17 –0.48 –0.19 –0.50 –0.32 0.02 0.25 0.19 0.07 –0.12 0.07 –0.43 –0.29 0.32 0.00

Catchment type

0.70* –0.67* 0.28 0.25 –0.25 0.25 –0.46 0.32 –0.57 –0.09 –0.53 –0.18 0.16 0.05 0.17 –0.02 –0.38 –0.02 –0.26 –0.38 0.19 0.17

Litter Detritus

–0.70* 0.25 0.55 –0.31 –0.37 –0.61* 0.68* 0.01 0.21 –0.62* –0.24 0.12 0.30 –0.25 0.31 –0.48 –0.12 –0.26 –0.33 0.09 0.19

Periphyton –0.52 –0.40 0.49 0.25 0.67** –0.51 0.12 –0.51 0.32 0.02 –0.58* –0.33 –0.11 0.30 0.34 0.41 –0.06 0.45 0.16 –0.19

POM –0.16 –0.17 –0.32 –0.54 –0.10 0.16 0.60* 0.22 0.66* 0.43 –0.37 –0.47 –0.49 –0.08 –0.50 0.24 –0.62* 0.48 0.12

Distance –0.40 –0.33 –0.51 0.43 0.18 0.00 –0.33 –0.40 0.15 0.49 –0.14 0.40 –0.16 0.00 –0.34 –0.12 0.06 0.04

Vmean 0.20 0.32 – 0.03 0.14 – 0.04 0.31 0.20 – 0.19 0.21 0.06 0.23 0.46 – 0.16 0.07 – 0.03 0.25 – 0.15

Stream width 0.47 –0.30 –0.64 –0.50 0.13 –010 –0.10 –0.17 0.45 –0.04 0.31 0.44 –0.06 0.27 0.04 –0.23

Mean depth – 0.49 – 0.15 – 0.59* 0.28 – 0.19 – 0.41 – 0.11 0.31 0.05 0.51 0.21 0.21 0.86*** – 0.16 – 0.55

Depth CV 0.13 0.39 – 0.30 – 0.12 0.38 0.28 – 0.20 0.33 – 0.49 – 0.06 – 0.21 – 0.17 – 0.04 0.17

Temp. 0.59* 0.33 0.29 0.08 0.09 0.80** 0.12 –0.02 –0.08 –0.10 –0.02 0.14 –0.11

Cond. 0.47 0.64* 0.76** – 0.09 – 0.64* – 0.40 – 0.08 – 0.42 0.24 – 0.32 0.51 – 0.12

pH 0.65* 0.49 –0.24 –0.29 –0.44 0.70* –0.28 0.54 0.19 0.57 –0.57

SS 0.55 –0.25 –0.41 –0.31 0.22 –0.44 0.54 –0.30 0.49 –0.13

Gravel/Pebble

Substrate heterogeneity NO3N

NH3N

PO4 P

– 0.02 – 0.20 0.28 – 0.47 0.55 – 0.03 0.19 0.12 0.22 –0.34 – 0.53 –0.13 0.06 0.44 – 0.43 0.50 –0.13 0.09 –0.58 0.51 – 0.74** – 0.25 0.03 0.12 0.15 – 0.16 0.22 –0.07 0.59* 0.27 – 0.21 –0.28 0.58* – 0.61* 0.27 – 0.23 –0.22 0.09 0.09 – 0.29 0.27 –0.15

Cobble/Boulder

Appendix 1. Correlation coefficients between environmental variables. Significance levels of Pearson Product moment correlations: * p < 0.05, ** p < 0.01 and *** p < 0.001. POM = particulate organic matter, SS = suspended solids, Vmean = mean current velocity, distance = distance to main river.

Sand

342 Berit H. Bojsen and Dean Jacobsen