Revealing the pathways by which agricultural land‐use affects stream fish communities in South Brazilian grasslands

相似文献   

Depauperate freshwater fish communities in Sabah: the role of barriers to movement and habitat quality

Up to 30 species of fish were generally found in medium-sized rainforest streams in Sabah, Malaysia. However, depauperate fish communities were found in two streams above large (>8 m) waterfalls (five and nine species) and in two very small streams located in deep forest (five and six species). A fifth stream with small cascades and waterfalls had a faunal list of 12 species. Species resident above waterfalls were predominantly herbivores while fishes in the two forest streams were from a variety of trophic groups. Abundance and biomass of fishes above waterfalls were significantly lower than all other sites. A translocation experiment was performed in one stream to distinguish between the hypotheses that fish communities above waterfalls were determined solely by colonization ability and were unsaturated with species or that they were limited by habitat quality or food. Seven species (775 individuals) were trans located into one isolated section and four species (570 individuals) into another. Species that were trans located included trophic groups that were not represented by resident species. Twelve months after translocation, only four and one trans located species were collected in the two sections, all at greatly reduced densities. After 20 months, the number of species were three and two, respectively. One trans located species, Rasbora sumatrana , had increased in abundance from 12 months and juveniles were present in the population. Individuals of other trans located species appeared to be remnants of stocked populations. Abundance and biomass of resident species fluctuated widely between years. Interpretation of results was complicated by a large flood which substantially changed habitat conditions about a month after translocations were performed. It is suggested that two different factors were responsible for depauperate communities: movement barriers for waterfall sites and physicochemical conditions and/or habitat availability at other sites.     

Watershed urbanization alters the composition and function of stream bacterial communities

Watershed urbanization leads to dramatic changes in draining streams, with urban streams receiving a high frequency of scouring flows, together with the nutrient, contaminant, and thermal pollution associated with urbanization. These changes are known to cause significant losses of sensitive insect and fish species from urban streams, yet little is known about how these changes affect the composition and function of stream microbial communities. Over the course of two years, we repeatedly sampled sediments from eight central North Carolina streams affected to varying degrees by watershed urbanization. For each stream and sampling date, we characterized both overall and denitrifying bacterial communities and measured denitrification potentials. Denitrification is an ecologically important process, mediated by denitrifying bacteria that use nitrate and organic carbon as substrates. Differences in overall and denitrifying bacterial community composition were strongly associated with the gradient in urbanization. Denitrification potentials, which varied widely, were not significantly associated with substrate supply. By incorporating information on the community composition of denitrifying bacteria together with substrate supply in a linear mixed-effects model, we explained 45% of the variation in denitrification potential (p-value<0.001). Our results suggest that (1) the composition of stream bacterial communities change in response to watershed urbanization and (2) such changes may have important consequences for critical ecosystem functions such as denitrification.     

Patterns of assemblage change in prairie stream fishes in relation to urban stormwater impoundments

The effects of stream urbanization on fishes have been well studied in general. Yet despite the wealth of knowledge available for streams in many different ecoregions, relatively little is known of the effects of urbanization on prairie stream fishes. Management of urban stormwater through impoundment has the potential to fragment streams, and habitat fragmentation on nonurban streams has been documented to relate to declines in small-bodied mobile minnow species. We asked whether urban habitat fragmentation through stormwater impoundment would relate to a similar decline in small-bodied fishes in Cottonwood Creek, a stream system partially managed by stormwater impoundment in central Oklahoma. Analyses with basic metrics of ecological tolerance, richness, community structure, and multivariate ordination found negative relationships between cyprinid richness and abundance and a metric of urban habitat fragmentation, as well as between Lepomis humilis, a small-bodied sunfish, and the metric of urban habitat fragmentation. We review potential hypotheses for these biological patterns in fragmented urban streams, including predation, lack of successful reproduction, and lack of ability to recolonize above barriers.     

Convergence of fish communities from the littoral zone of reservoirs

1. Understanding factors that regulate the assembly of communities is a main focus of ecology. Human‐engineered habitats, such as reservoirs, may provide insight into these assembly processes because they represent novel habitats that are subjected to colonization by fishes from the surrounding river basin or transported by humans. By contrasting community similarity within and among reservoirs from different drainage basins to nearby stream communities, we can test the relative constraints of reservoir habitats and regional species pools in determining species composition of reservoirs. 2. We used a large spatial database that included intensive collections from 143 stream and 28 reservoir sites within three major river basins in the Great Plains, U.S.A., to compare patterns of species diversity and community structure between streams and reservoirs and to characterize variation in fish community structure within and among major drainage basins. We expected reservoir fish faunas to reflect the regional species pool, but would be more homogeneous that stream communities because similar species are stocked and thrive in reservoirs (e.g. planktivores and piscivores), and they lack obligate stream organisms that are not shared among regional species pools. 3. We found that fish communities from reservoirs were a subset of fishes collected from streams and dominant taxa had ecological traits that would be favoured in lentic environments. Although there were regional differences in reservoir fish communities, species richness, patterns of rank abundance and community structure in reservoir communities were more homogonous across three major drainage basins than for stream communities. 4. The general pattern of convergence of reservoir fish community structure suggests their assembly is constrained by local factors such as habitat and biotic interactions, and facilitated by the introduction of species among basins. Because there is a reciprocal transfer of biota between reservoirs and streams, understanding factors structuring both habitats is necessary to evaluate the long‐term dynamics of impounded river networks.     

Landscape Determinants of Nonindigenous Fish Invasions

Much has been written about the influence of exotic or nonindigenous species on natural habitats and communities of organisms, but little is known of the physical or biological conditions that lead to successful invasion of native habitats and communities by exotics. We studied invasivity factors in headwater streams of the Susquehanna River West Branch, which drains portions of the northern Appalachian Plateau. A replicated (two major tributaries) 3 × 3 factorial design was used to determine landscape effects of size (stream order) and quality (land use) on abiotic (physical and chemical) and biotic (fish community structure and function) stream attributes. Seven (21%) of thirty-four fish species (brown trout, common carp, mimic shiner, bluegill, smallmouth bass, fantail darter, and banded darter) collected in the eighteen streams sampled were nonindigenous to the basin. Watershed size (stream orders 1, 3, and 5) significantly affected stream geomorphologic and habitat variables (gradient, width, depth, current velocity, diel water temperature, bank overhang, canopy cover, and woody debris density) but not water-quality variables, while land use in watersheds (conservation, mining, and agriculture) significantly affected measured water-quality variables (alkalinity and concentrations of manganese, calcium, chloride, nitrate, and total dissolved solids) but not stream physical or habitat quality. Both watershed size and land use affected fish-community variables such as presence of particular species, species density, species diversity, tolerance diversity, and mean fish size, but in both cases the effect was transparent to native-origin status of fish species. No relationships were found between occurrence of nonindigenous species in watersheds and trophic structure or functional diversity. Therefore, the hypothesis that reduced species diversity increases vulnerability to nonindigenous species was not supported. However, the spatial variation associated with both water-quality and habitat-quality factors was greater in streams with mixed (those with nonindigenous species) than with exclusively native assemblages. These findings suggest that the mechanism for successful invasion by nonindigenous or exotic species is through change in water or habitat quality associated with human or natural disturbances, such as agriculture and mining activities in watersheds. Biotic factors appear to play no or a lesser role in the invasibility of northern Appalachian lotic systems.     

三峡水库175 m试验性蓄水期库区及其上游江段鱼类群落结构时空分布格局

根据2010—2012年在三峡水库及其上游江段5个江段的商业性渔获物调查结果,对该区域鱼类群落结构的时空分布格局进行了分析。调查中共收集到鱼类87种,隶属于8目18科63属。沿坝前江段溯河而上至库尾以上流水江段,鱼类种类数逐渐增加。采用聚类分析(CA)和非度量多维标度(NMDS)方法对三峡水库175 m试验性蓄水期间各江段的鱼类种类组成进行分析,发现这5个调查江段的鱼类种类组成可以分为两个类群:类群Ⅰ包括秭归、巫山、云阳3个库区的江段,其鱼类种类组成在各年间的差异程度较大;而类群Ⅱ包括库尾的涪陵江段以及上游的江津江段,其鱼类种类组成在各年间差异程度较小。采用CA和NMDS方法对鱼类群落结构时空分布格局的分析结果表明,5个调查江段的鱼类群落结构在各年间可以分为不同类群;云阳与涪陵江段的鱼类群落结构的相似性逐年增加,而与秭归、巫山江段的相似性逐年减少;涪陵江段与江津江段的鱼类群落结构相似性也呈逐年减少趋势。总体而言,175m试验性蓄水对库中和库尾江段的鱼类群落结构的影响较大,但对库首以及库尾以上流水江段的影响均较小。     

Hierarchical spatial structure of stream fish colonization and extinction

Spatial variation in extinction and colonization is expected to influence community composition over time. In stream fish communities, local species richness (alpha diversity) and species turnover (beta diversity) are thought to be regulated by high extinction rates in headwater streams and high colonization rates in downstream areas. We evaluated the spatiotemporal structure of fish communities in streams originally surveyed by Burton and Odum 1945 (Ecology 26: 182–194) in Virginia, USA and explored the effects of species traits on extinction and colonization dynamics. We documented dramatic changes in fish community structure at both the site and stream scales. Of the 34 fish species observed, 20 (59%) were present in both time periods, but 11 (32%) colonized the study area and three (9%) were extirpated over time. Within streams, alpha diversity increased in two of three streams but beta diversity decreased dramatically in all streams due to fish community homogenization caused by colonization of common species and extirpation of rare species. Among streams, however, fish communities differentiated over time. Regression trees indicated that reproductive life‐history traits such as spawning mound construction, associations with mound‐building species, and high fecundity were important predictors of species persistence or colonization. Conversely, native fishes not associated with mound‐building exhibited the highest rates of extirpation from streams. Our results demonstrate that stream fish colonization and extinction dynamics exhibit hierarchical spatial structure and suggest that mound‐building fishes serve as keystone species for colonization of headwater streams.     

Spatial patterns of the biological traits of freshwater fish communities in south‐west France

Spatial patterns in the combinations of biological traits of fish communities were studied in the Garonne River system (57 000 km², south‐west France). Fish species assemblages were recorded at 554 sampling sites, and the biological traits of species were described using a fuzzy‐coding method. A co‐inertia analysis of species distributions and biological traits identified some spatial patterns of species trait combinations. Fish species richness progressively increased from up‐ to downstream sections, and the longitudinal patterns of fish assemblages partitioned the river into clear biogeographic areas, such as the brown trout Salmo trutta (headwater streams), the grayling Thymallus thymallus , the barbel Barbus barbus and the bream Abramis brama zones (most downstream sections), which fitted with Huet's well‐known zonation for western European rivers. Only a few biological traits, chiefly related to life‐history attributes, significantly influenced the observed fish distributions. Fecundity, potential size, maximum age and reproductive factor increased from headwater to plain reaches. As a theoretical framework for assessing and predicting the functional organization of stream fish communities, spatial variations in species traits can be related to habitat conditions, thus providing explicit spatial schemes that may be useful to the design of both scientific studies and river management.     

Effects of headwater impoundment and channelization on invertebrate drift

The construction of a flood control impoundment on Twitty's Creek added large numbers of organisms of limnetic origin to the stream ecosystem. However, the number of limnetic organisms per unit volume of water decreased rapidly as the distance downstream from the reservoir increased and, during most sampling periods, made up an insignificant portion of the total drift biomass at 7.2 km downstream. Factors favoring the extended downstream drift of limnetic organisms were high stream discharge and low water temperature.Several taxa of benthic organisms had much lower drift rates in the station immediately below the dam than at other stations and several taxa commonly taken at other stations were not captured immediately below the reservoir outfall. One possible explanation is that these organisms may have longer drift recruitment distances than the distance from the reservoir outfall to the sample location.A comparison of drift densities of organisms of benthic origin and benthic standing crop densities in channeled and unchanneled streams revealed that drift densities were higher in channeled streams than in unchanneled streams for most taxa of invertebrates. In addition, channeled streams appeared to have lower benthic standing crops than unchanneled streams for most taxa of invertebrates.In stream sections impacted by either channelization or the Twitty Lake outfall, the energy dynamics of the stream ecosystems were altered by increased density of drifting invertebrates. From the standpoint of increasing food availability to the fish fauna of the stream, these changes would appear to benefit drift feeding species and negatively impact bottom feeding species.     

Variation in local abundance and species richness of stream fishes in relation to dispersal barriers: implications for management and conservation

1. Barriers to immigration, all else being equal, should in principle depress local abundance and reduce local species richness. These issues are particularly relevant to stream‐dwelling species when improperly designed road crossings act as barriers to migration with potential impacts on the viability of upstream populations. However, because abundance and richness are highly spatially and temporally heterogeneous and the relative importance of immigration on demography is uncertain, population‐ and community‐level effects can be difficult to detect. 2. In this study, we tested the effects of potential barriers to upstream movements on the local abundance and species richness of a diverse assemblage of resident stream fishes in the Monongahela National Forest, West Virginia, U.S.A. Fishes were sampled using simple standard techniques above‐ and below road crossings that were either likely or unlikely to be barriers to upstream fish movements (based on physical dimensions of the crossing). We predicted that abundance of resident fishes would be lower in the upstream sections of streams with predicted impassable barriers, that the strength of the effect would vary among species and that variable effects on abundance would translate into lower species richness. 3. Supporting these predictions, the statistical model that best accounted for variation in abundance and species richness included a significant interaction between location (upstream or downstream of crossing) and type (passable or impassable crossing). Stream sections located above predicated impassable culverts had fewer than half the number of species and less than half the total fish abundance, while stream sections above and below passable culverts had essentially equivalent richness and abundance. 4. Our results are consistent with the importance of immigration and population connectivity to local abundance and species richness of stream fishes. In turn, these results suggest that when measured at appropriate scales (multiple streams within catchments), with simple protocols amenable to use by management agencies, differences in local abundance and species richness may serve as indicators of the extent to which road crossings are barriers to fish movement and help determine whether road‐crossing improvements have restored connectivity to stream fish populations and communities.     

Stream fish communities and their associations to habitat variables in a rain forest reserve in southeastern Brazil

This paper describes the spatial variability of fish communities and identifies patterns of association between fish communities and habitat variables, including anthropogenic factors. We sampled streams inside and in the surroundings of a rain forest reserve in the southeast of Brazil in the rainy season. We could distinguish three main groups of streams: upland streams (draining the upland, flat portions of the mountain ridge of Serra de Paranapiacaba), adventitious streams of clearwater mountain torrents (small streams draining confined valleys in the slopes of this sierra), and large streams of clearwater mountain torrents (relatively unconstrained large streams close to the foots of the mountains). Despite the high variability of fish communities associated with these streams we identified some patterns using exploratory statistical analyses. These patterns were corroborated by additional field observations and information from the scientific literature. The main differences in fish community composition and diversity among the three groups of streams are probably related to large-scale factors such as elevation and position of the stream in the watershed. However, differences within these three groups seem to be mostly due to site-specific factors. Differences of instream characteristics are likely to be caused by natural variability of the ecosystems but also, in some cases, by human disturbances like pollution from human settlements, agriculture and mining.     

Comparison of fish communities between above‐ and below‐dam sections of small streams; barrier effect to diadromous fishes

To analyse the effects of small‐scale dams on fish communities, species diversity, population density, biomass, migration over dams and trophic relationships were investigated in the above‐ and below‐dam sections of three Japanese streams. The dams were 1·5–3·9 m high, constructed within 400 m of the sea, and had no impoundment or fishway. Fish communities were mainly composed of diadromous fishes. Of the 17 species sampled, only a few species of gobies, Rhinogobius spp. and Gymnogobius spp. were able to climb and migrate upstream over the dams. Consequently the number of species, and total density and biomass of fishes below the dams was high, but low above the dams. One of the most dominant fish below dams, Plecoglossus altivelis altivelis , predominantly fed on benthic algae, whereas other fishes preyed on benthic invertebrates. Trophic relationships above dams were very simple, the few species of gobies preying on chironomid larvae and other aquatic invertebrates.     

Influences of low-head dams on the fish assemblages in the headwater streams of the Qingyi watershed, China

The influences of low-head dams on the fish assemblages were examined in this study, using fish data collected in six treatment and five reference sites at three low-head dams in the headwater streams of the Qingyi watershed, China. Comparing with those in the reference sites, local habitat variables were significantly altered by low-head dams in the treatment sites, involving wider channel (only in the impoundment area), deeper water and slower flow. Fish species richness varied significantly across seasons, not across site categories, suggesting that these low-head dams did not alter species richness. However, significant decreases in fish abundance and density were observed in the impoundment areas immediately upstream of dams, but not in the plunge areas downstream. Fish assemblage structures kept relative stability across seasons, and their significant difference between-site was only observed between the impoundment areas and the sites far from dams upstream. This variation in assemblage structures was due to the differing relative abundance of some co-occurring species; more lentic but less lotic fish was observed in the impoundment areas. The spatial and temporal patterns of fish assemblages were correlated with local habitat in this study area. Wetted width had negative correlation with fish species richness, abundance and density, respectively. Water temperature also positively affected species richness. In addition, wetted width, water depth, current velocity and substrate were the important habitat variables influencing assemblage structures. Our results suggested that, by modifying local habitat characteristics, low-head dams altered fish abundance and density in the impoundment areas immediately upstream of dam, not in the plunge areas immediately downstream, and thereby influenced fish assemblage structures in these stream segments.     

不同类型人类活动干扰对河岸带外来植物群落的影响——以北京永定河为例

强烈的人类活动对自然生境的扰动促进了外来植物侵入河岸带,但不同类型的人类活动干扰对外来植物的影响有所不同。按照不同人类活动干扰类型设置了山峡段、平原段和城市段,通过比较河段间河岸带外来植物群落特征的差异,探讨不同人类活动干扰类型对河岸带外来植物群落的影响。结果表明:(1)永定河河岸带共有维管束植物27科72属101种,外来植物13科28属29种。外来种比例高达28.7%,其中82.4%的国外外来种来自美洲和亚洲。(2)不同河段间的外来植物群落的物种构成和优势度呈现出显著性差异。外来种比例由山峡段的20.9%,上升至平原段和城市段的30.2%和25.5%;而优势度由12.1%,分别上升至13.4%和17.5%。(3)不同河段间的外来植物群落生活型结构有显著性差异。多年生草本植物在山峡段比例最高,达到66.7%,在平原段和城市段较低,分别为46.2%和30.8%;而一年生草本植物在城市段比例最高,达到69.2%,在平原段和山峡段较低,分别为53.8%和33.3%。(4)平原段属于农业干扰类型,其河岸带外来植物主要以农业类杂草为主,如禾本科的假稻(Leersia japonica)和菊科的钻叶紫菀(Aster subulatu);城市段是城市干扰类型,外来植物中入侵种比例和优势度较高,典型入侵种为反枝苋(Amaranthus retroflexus)和牛筋草(Eleusine indica)等。北京永定河不同河段河岸带外来植物的种类构成特点,反映了河岸带外来植物受快速城市化、农业活动等不同人类活动干扰的影响呈逐渐扩大的趋势。     

Effects of disturbance caused by selective timber extraction on fish communities in Sabah, Malaysia

To examine the effects of selective timber extraction on fish communities in Sabah, Malaysia, quantitative samples of fishes were taken from thirteen streams running through undisturbed rainforest or through forest that had been selectively logged 3–18 years previously. Multivariate analysis (canonical discriminant analysis and cluster analysis) indicated that mesohabitats within streams (riffles and pools) and differences in stream size were more important in determining community structure than logging history. Riffles in streams running through logged or undisturbed forest were indistinguishable using relative biomass or abundance data, as were pools from small streams (approximate order 2). Fish communities from pools in larger streams showed some separation in multivariate space corresponding to a complex set of relative changes in abundance and/or biomass between species. However it was difficult to unambiguously assign such changes to logging regime alone. There appeared to be some differences in fish communities between streams in recently-logged (3–7 years) and old-logged (17–18 years) areas related to abundance or biomass of three cyprinids (Garra borneensis, Lobocheilos bo and Osteochilus chini). Only one species, Pangio mariarum, was not found in streams in logged forest, but it was only found at one location in undisturbed forest. A number of other species showed significant differences in abundance or biomass between sites but most of these were only present at some sites and in low abundance. Principal components analysis of habitat data showed that riffle sites were homogeneous whatever their logging history as were pools in unlogged large streams. Pools in logged large streams were significantly more heterogeneous but in a random rather than systematic manner. It is concluded that the type of selective logging practices used locally have low impact on fish communities through mechanisms of persistence and/or rapid recolonisation.     

Zooplankton community changes between forest and meadow sections in small headwater streams, NW Poland

Headwater streams are important resources in production of organic matter, but zooplankton of headwater streams has rarely been studied. In the present study spatial changes in zooplankton communities between upper headwater forest section and downstream meadow section of four small streams were examined (NW Poland). The environmental conditions of stream riparian zone, e.g., the presence of leaf litter, pools, slackwaters and wetlands had a great influence on the spatial changes in zooplankton. Low gradient of stream bed, longer water residence time and larger surfaces of wetlands and slackwaters also positively affected zooplankton communities, especially in the meadow section. Along the streams, from the stream headwater to the downstream-meadow section, significant spatial changes in zooplankton communities were observed. Significantly higher number of taxa and density of zooplankton were observed in meadow sections than in headwater forest sections. In the forest sections, there was a definite domination of benthic and littoral rotifers in the zooplankton composition, while in the meadow sections, planktonic rotifers dominated. Cladocerans were observed only in downstream meadow section; two littoral species Coronatella rectangula and Peracantha truncata and one planktonic Daphnia magna which probably were washed from close pools of small stagnant water bodies. Copepods were noted along the entire length of streams. The occasional presence of planktonic rotifers in the forest section was probably caused by their washout from a few small stagnant water bodies.     

Macrozoobenthos communities and biocoenotic assessment of ecological status in relation to degree of human impact in small streams in southwest Germany

One of the major objectives of the VALIMAR project is to determine the ecological significance of various fish biomarker studies as indicators of chronic pollution in small streams in southwest Germany. Results of these fish biomarker investigations were compared to information from complementary studies on the meiobenthos, macrobenthos, and fish community studies to assess the ecological significance of these biomarker investigations. The main objective of this study was to provide biological assessments of the biomarker sites on the basis of the macrozoobenthos communities. Since no validated framework for the assessment of the biological integrity existed in Germany, two multimetric approaches were adapted to the whole stream system by investigating benthos and fish communities of 46 sites of varying degrees of human disturbance. Assessment of the communities was conducted in accordance with the European Community Water Framework Directive. Species distribution of benthos depended upon stream type and pollution status of streams. Biological attributes and bioindices of benthos communities, however, did not correlate with typological parameters like stream size or dominant substrate but correlated better with pollution parameters like conductivity or chloride concentration. Using a set of 18 measures, such as portion of sessile individuals, Rheoindex, oxygen availability index, and portion of pool dwellers, the benthos communities were characterized and evaluated. The composition of the fish communities was mainly determined by stream type, pollution and migration barriers. The influence of chemical parameters could be assessed by developing a“fish chemistry index”, which calculatesthe similarity of the present fish community with the potential natural community, but excludes those species strongly effected by deficits in stream channel morphology. Both fish chemistry index and benthic indices strongly correlated with pollution index parameters, clearly distinguishing between the more polluted Körsch sites and the less pollutedKrähenbach and Aich sites. Most of the single bioindices as well as overall assessment by multimetric indices indicated a gradient of decreasing quality from the reference stream Krähenbach to theslightly polluted Aich and Körsch upstream site (KE, upstream of all sewage treatment plants) and finally to the most polluted Körsch site directlybelow the most upstream sewage treatment plant (KD). According to the Water Framework Directive, the classification of ecological status of the benthos communities ranges from “high” (best of 5 classes) forthe reference stream to “bad” (5th class) for KD. Assessment of the fish community tends to score somewhat worse than the benthos due to deficits in morphological quality of the stream reaches. The benthos assessment and the newlydeveloped “fish chemistry index” correlated well with chemical water quality and hence with biomarkers, whereas ecological status of fish and overall ecological status was also influenced by river morphology. In conclusion all tested assessment methods on biocoenotic level are reliable indicators for the degree of human disturbance on small streams, whereas biomarkers are more suited for risk assessment and the investigation ofcause-effect-relationships.     

Deforestation alters the resource base and biomass of endemic stream insects in eastern Madagascar

1. Rainforest streams in eastern Madagascar have species‐rich and diverse endemic insect communities, while streams in deforested areas have relatively depauperate assemblages dominated by collector‐gatherer taxa. We sampled a suite of benthic insects and their food resources in three primary rainforest streams within Ranomafana National Park in eastern Madagascar and three agriculture streams in the park's deforested peripheral zone. We analysed gut contents and combined biomass and stable isotope data to examine stream community responses to deforestation in the region, which is a threatened and globally important hotspot for freshwater biodiversity. 2. Gut analyses showed that most taxa depended largely on amorphous detritus, obtained either from biofilms (collector‐gatherers) or from seston (microfilterers). Despite different resource availability in forest versus agriculture streams, diets of each taxon did not differ between stream types, suggesting inflexible feeding modes. Carbon sources for forest stream insects were difficult to discern using δ¹³C. However, in agriculture streams dependence on terrestrial carbon sources was low relative to algal sources. Most insect taxa with δ¹³C similar to terrestrial carbon sources (e.g. the stonefly Madenemura, the caddisfly Chimarra sp. and Simulium blackflies) were absent or present at lower biomass in agriculture streams relative to forest streams. Conversely, collector‐gatherers (Afroptilum mayflies) relied on algal carbon sources and had much higher biomass in agriculture streams. 3. Our analyses indicate that a few collector‐gatherer species (mostly Ephemeroptera) can take advantage of increased primary production in biofilms and consequently dominate biomass in streams affected by deforestation. In contrast, many forest stream insects (especially those in the orders Plecoptera, Trichoptera and Diptera) depend on terrestrial carbon sources (i.e. seston and leaf litter), are unable to track resource availability and consequently decline in streams draining deforested landscapes. These forest‐specialists are often micro‐endemic and particularly vulnerable to deforestation. 4. The use of consumer biomass data in stable isotope research can help detect population‐level responses to shifts in basal resources caused by anthropogenic change. We also suggest that restoration of vegetated riparian zones in eastern Madagascar and elsewhere could mitigate the deleterious effects of deforestation on sensitive, endemic stream taxa that are dependent on terrestrial carbon sources.