<Emphasis Type="Italic">β</Emphasis>-Diversity and vegetation structure as influenced by slope aspect and altitude in a seasonally dry tropical landscape

Topography strongly affects the distribution of insolation in the terrain. Patterns of incoming solar radiation affect energy and water balances within a landscape, resulting in changes in vegetation attributes. Unlike other regions, in seasonally dry tropical forest areas the potential contribution of topography-related environmental heterogeneity to β-diversity is unclear. In Mt. Cerro Verde (Oaxaca), S. Mexico, we: (1) modelled potential energy income for N- and S-facing slopes based on a digital elevation model, (2) examined the response of vegetation structure to slope aspect and altitude and (3) related variations in plant diversity to topography-related heterogeneity. Vegetation survey and modelling of potential energy income (SOLEI-32 model) were based on 30 plots equally distributed among three altitudinal belts defined for each slope of the mountain; combining the three altitudinal belts and the two slopes produced six environmental groups, represented by five vegetation plots each. Potential energy income was about 20% larger on the S than on the N slope (9,735 versus 8,138 MJ/m²), but it did not vary with altitude. In addition, the temporal behaviour of potential energy income throughout the year differed greatly between slopes. Vegetation structure did not show significant changes linked to the environmental gradients analysed, but altitude and aspect did affect β-diversity. We argue that the classic model of slope aspect effect on vegetation needs reconsideration for tropical landscapes. Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorised users.     

Riparian influences on stream fish assemblage structure in urbanizing streams

We assessed the influence of land cover at multiple spatial extents on fish assemblage integrity, and the degree to which riparian forests can mitigate the negative effects of catchment urbanization on stream fish assemblages. Riparian cover (urban, forest, and agriculture) was determined within 30 m buffers at longitudinal distances of 200 m, 1 km, and the entire network upstream of 59 non-nested fish sampling locations. Catchment and riparian land cover within the upstream network were highly correlated, so we were unable to distinguish between those variables. Most fish assemblage variables were related to % forest and % urban land cover, with the strongest relations at the largest spatial extent of land cover (catchment), followed by riparian land cover in the 1-km and 200-m reach, respectively. For fish variables related to urban land cover in the catchment, we asked whether the influence of riparian land cover on fish assemblages was dependent on the amount of urban development in the catchment. Several fish assemblage metrics (endemic richness, endemic:cosmopolitan abundance, insectivorous cyprinid richness and abundance, and fluvial specialist richness) were all best predicted by single variable models with % urban land cover. However, endemic:cosmopolitan richness, cosmopolitan abundance, and lentic tolerant abundance were related to % forest cover in the 1-km stream reach, but only in streams that had <15% catchment urban land cover. In these cases, catchment urbanization overwhelmed the potential mitigating effects of riparian forests on stream fishes. Together, these results suggest that catchment land cover is an important driver of fish assemblages in urbanizing catchments, and riparian forests are important but not sufficient for protecting stream ecosystems from the impacts of high levels of urbanization.     

Error assessment in decision-tree models applied to vegetation analysis

Methods were developed to evaluate the performance of a decision-tree model used to predict landscape-level patterns of potential forest vegetation in central New York State. The model integrated environmental databases and knowledge on distribution of vegetation. Soil and terrain decision-tree variables were derived by processing state-wide soil geographic databases and digital terrain data. Variables used as model inputs were soil parent material, soil drainage, soil acidity, slope position, slope gradient, and slope azimuth. Landscapescale maps of potential vegetation were derived through sequential map overlay operations using a geographic information system (GIS). A verification sample of 276 field plots was analyzed to determine: (1) agreement between GIS-derived estimates of decision-tree variables and direct field measurements, (2) agreement between vegetation distributions predicted using GIS-derived estimates and using field observations, (3) effect of misclassification costs on prediction agreement, (4) influence of particular environmental variables on model predictions, and (5) misclassification rates of the decision-tree model. Results indicate that the prediction model was most sensitive to drainage and slope gradient, and that the imprecision of the input data led to a high frequency of incorrect predictions of vegetation. However, in many cases of misclassification the predicted vegetation was similar to that of the field plots so that the cost of errors was less than expected from the misclassification rate alone. Moreover, since common vegetation types were more accurately predicted than rare types, the model appears to be reasonably good at predicting vegetation for a randomly selected plot in the landscape. The error assessment methodology developed for this study provides a useful approach for determining the accuracy and sensitivity of landscape-scale environmental models, and indicates the need to develop appropriate field sampling procedures for verifying the predictions of such models.     

The importance of land-use legacies for modeling present-day species distributions

Context

Land-use legacies play an important role in shaping contemporary species distributions. However, land-use legacies are rarely considered in species distribution models (SDMs) that aim to model present-day species distributions across the landscape, even though they can lead to a species absence in suitable areas. SDMs that do not account for land-use legacies will likely result in biased predictions of species distributions.

Objective

We examine the importance of land-use legacies for modeling present-day distributions of tree species at a regional scale, assessing how the addition of land-use legacy variables improves predictive power of SDMs.

Methods

We generated land-use legacy variables using raster layers of reconstructed historical agricultural land use and 3310 inventory plots. SDMs were developed for six forest tree species based on climatic, edaphic, and topographic variables, and with (SDM_LU) and without (SDM_Base) land-use legacy variables. We compared the predictive power between SDM_LU and SDM_Base models and then quantified the local importance of land-use legacy variables relative to other abiotic variables.

Results

Our results show that the importance of land-use legacy variables for present-day species distributions and the improvement on the predictive power of SDMs is species-specific. The inclusion of land-use legacy variables improved SDMs primarily by lowering errors of commission and increasing the overall accuracy of prediction.

Conclusion

The influence of land-use legacies on SDMs suggests that, for some tree species, incorporating land-use legacies can accurately identify suitable areas that are not occupied by the species due to land-use legacies, and advance our understanding of their present-day distributions.

     

Influences of habitat and land cover on fish distributions along a tributary to Lake Ontario,New York

Influences of habitat and land cover on fish distributions were determined along a lentic–lotic gradient along a tributary to Lake Ontario, New York. Nonmetric multidimensional scaling, cluster analysis, and specific characterization methods were used to classify the fish species into five groups based on their similar patterns of distribution, species-specific habitat relationship, and relative abundance observed along the gradient. A stepwise regression approach was used to select the best habitat and land cover variables to explain variations in the distribution pattern of each fish group. Distribution patterns of the five fish groups were significantly explained by either a set of the selected habitat or land cover predictor variables or a combination of both. Of the 10 habitat variables, water depth, current velocity, aquatic plants, algae, woody debris, sand, and rock-bedrock were selected to explain the variations in distribution patterns of one or more fish groups. Of the 16 land cover types, evergreen wetlands, evergreen plantations, successional shrubs, shrub swamps, roads, and urban areas were selected to explain the variations in distribution patterns of at least one fish group.     

Environmental,historical, and contextual determinants of vegetation cover: a landscape perspective

We formulated and tested models of relationships among determinants of vegetation cover in two agroforested landscapes of eastern North America (Haut Saint-Laurent, Quebec, Canada) that differed by the spatial arrangement of their geomorphic features and intensity of agricultural activities. Our landscape model compared the woody plots of each landscape in terms of the relative influence of environmental attributes, land use history (1958 – 1997), and spatial context (i.e., proximity of similar or contrasting land cover). Our vegetation model evaluated the relative contribution of the same sets of variables to the distributions of herbs, trees, and shrubs. Relationships were assessed using partial Mantel tests and path analyses. Significant environmental and contextual differences were found between the vegetation plots of the two landscapes, but disturbance history was similar. Our vegetation model confirms the dominant effect of historical factors on vegetation patterns. Whereas land-use history overrides environmental and contextual control for trees, herbaceous and shrub species are more sensitive to environmental conditions. Context is determinant only for understory species in older, less-disturbed plots. Results are discussed in relevance to vegetation dynamics in a landscape perspective that integrates interactions between environmental and human influences.     

Predicting land cover changes and their impact on the sediment influx in the Lake Balaton catchment

The land cover pattern in the Lake Balaton catchment (Hungary) has been changing since decollectivization in the 1990s. These land cover changes significantly impact the landscape connectivity, controlling the influx of sediments into the lake. A comparison of high resolution land cover maps from 1981, 2000 and 2005 showed a significant extensification of the agriculture with land cover conversions from arable land and vineyards to grassland and forest. For each land unit transition probabilities were assessed using logistic regression techniques to evaluate to which extent land cover changes are controlled by physical or socio-economic parameters. A stochastic land cover allocation algorithm was applied to generate future land cover patterns. The landscape connectivity for each of the simulated land cover patterns was assessed by means of a distributed routing algorithm. The simulations suggest that further land abandonment in the upslope parts of the catchment will cause a non-linear reduction of average soil erosion rates. The changes, however, have a relatively low impact on the sediment volume entering the lake because of the land unit’s poor connectivity with permanent river channels. The major contributors to the lakes sediment load are the vineyards near the lakeshore. They are likely to be maintained because of their touristic value. A significant reduction of the total sediment input in the lake can be expected only if soil conservation measures in the vineyards near the shorelines are undertaken.     

Linking Land-use, Water Body Type and Water Quality in Southern New Zealand

Land-use and vegetation cover have been linked to the nutrient levels (nitrogen, phosphorus) of surface waters in several countries. However, the links generally relate to streams and rivers, or to specific types of standing water, for example shallow lakes in a geologically defined region. We measured physical variables and nutrient chemistry of 45 water bodies representative of the wide range of lentic wetland environments (swamps, riverine wetlands, estuaries, reservoirs, shallow lakes, deep lakes) in Otago, New Zealand, and related these to catchment variables and land-use in order to assess the potential influence of catchment modification on water quality of these diverse wetlands. Catchment boundaries and land cover were derived from maps using ArcView GIS software. Our predictions that concentrations of nutrients and other components of water quality would correlate positively with the nature and intensity of catchment modification were confirmed in multivariate analyses. Physical and chemical measures were positively related to the extent of modification in the catchment (percentage of the catchment in pasture, planted forest, scrub and urban areas), and negatively related to lack of catchment modification (more of the catchment in bare ground, tussock grassland and indigenous forest). The strong negative correlations between nutrient concentrations, suspended sediment, water colour and the percentage of tussock cover in the catchment imply that increased conversion of the␣native tussock grassland to pastoral farming in␣Otago will increase nutrient concentrations and␣reduce water quality of the diverse lentic ecosystems.     

Forest harvest patterns and landscape disturbance processes

A physically-based model of the topographic influence on debris flow initiation and a rule-based model for wind damage were used to assess the influence of forest clearcutting patterns (i.e., location, size, shape and distribution of cut units) on the potential for landscape disturbance by these processes in Charley Creek watershed, Washington State, USA. Simulated clearcutting patterns consisted of 7, 9 or 26 ha square or rectangular harvest units distributed in either an aggregated or dispersed pattern under three stream-buffering scenarios. The slope-stability model predicted that potentially unstable ground is concentrated along steep headwater streams and inner-gorge side-slopes. Areas susceptible to wind damage were determined from the combination of slope, aspect, elevation, soil drainage and primary tree species. Among the variables examined here, the location of harvest units constitutes the most important factor influencing the potential for shallow landsliding. In contrast, the location, size, and shape of clear cuts and the interactions among these three factors significantly influenced the potential for wind damage. Minimal correspondence between areas predicted to be potentially unstable and areas susceptible to wind damage implies that harvest patterns designed to mitigate the potential for shallow landsliding may not necessarily reduce the potential for wind damage. Our results demonstrate that: (1) the location of timber harvesting is more important than the geometry of harvest activity in influencing shallow landsliding; (2) forest harvest patterns strongly influence the potential for disturbance processes; and (3) a single cutting pattern will often fail to meet all landscape management goals. This revised version was published online in July 2006 with corrections to the Cover Date.     

Assessing the utility of topographic variables in predicting structural complexity of tree stands in a reforested urban landscape

The transformation of natural landscapes into impervious built-up surfaces through urbanization is known to significantly interfere with the ecological integrity of urban landscapes and accelerate climate change and associated impacts. Although urban reforestation is widely recognised as an ideal mitigation practice against these impacts, it often has to compete with other lucrative land uses within an urban area. The often limited urban space provided for reforestation therefore necessitates the optimization of the ecological benefits, which demands spatially explicit information. The recent proliferation of tree stands structural complexity (SSC) and topographic data offer great potential for determining the ecological performance of reforested areas across an urban landscape. This study explores the potential of using topographic datasets to predict SSC in a reforested urban landscape and ranks the value of these topographic variables in determining SSC. Tree structural data from a reforested urban area was collected and fed into a tree stand structural complexity index, which was used to indicate ecological performance. Topographic variables (Topographic Wetness Index, slope, Area Solar Radiation and elevation)- were derived from a Digital Elevation Model (DEM) and used to predict SSC using the Partial Least Squares (PLS) regression technique. Results show that SSC varied significantly between the topographic variables. Results also show that the topographic variables could be used to reliably predict SSC. As expected, the Topographic Wetness Index and slope were the most important topographic determinants of SSC while elevation was the least valuable. These results provide valuable spatially explicit information about the ecological performance of the reforested areas within an urban landscape. Specifically, the study demonstrates the value of topographic data as aids to urban reforestation planning.     

Modeling Landscape Vegetation Pattern in Response to Historic Land-use: A Hypothesis-driven Approach for the North Carolina Piedmont,USA

Current methods of vegetation analysis often assume species response to environmental gradients is homogeneously monotonic and unimodal. Such an approach can lead to unsatisfactory results, particularly when vegetation pattern is governed by compensatory relationships that yield similar outcomes for various environmental settings. In this paper we investigate the advantages of using classification tree models (CART) to test specific hypotheses of environmental variables effecting dominant vegetation pattern in the Piedmont. This method is free of distributional assumptions and is useful for data structures that contain non-linear relationships and higher-order interactions. We also compare the predictive accuracy of CART models with a parametric generalized linear model (GLM) to determine the relative strength of each predictive approach. For each method, hardwood and pine vegetation is modeled using explanatory topographic and edaphic variables selected based on historic reconstructions of patterns of land use. These include soil quality, potential soil moisture, topographic position, and slope angle. Predictive accuracy was assessed on independent validation data sets. The CART models produced the more accurate predictions, while also emphasizing alternative environmental settings for each vegetation type. For example, relic hardwood stands were found on steep slopes, highly plastic soils, or hydric bottomlands – alternatives not well captured by the homogeneous GLM. Our results illustrate the potential utility of this flexible modeling approach in capturing the heterogeneous patterns typical of many ecological datasets.     

Local,landscape and regional factors structuring benthic macroinvertebrate assemblages in Swedish streams

A total of 694 streams were sampled for benthic macroinvertebrates in the autumn of 1995 as part of the Swedish national stream survey. After removal of sites considered as impacted, data from 428 streams as well as a large number of environmental variables were used to determine the relative importance of local, landscape, and large scale factors in explaining the variability in species composition of benthic stream macroinvertebrates. The environmental variables were divided into seven explanatory variable groups: local physical, local chemical, catchment land use/cover, catchment bedrock geology, Quaternary geology in catchment, regional factors (such as ecoregion) and spatial position. Partial Canonical Correspondence Analysis was used to partition the total explained variance in the species data into these variable groups. The pure (or unique) effects of the seven variable groups accounted for 69.1%, and combinations of variable groups (interaction terms) the remaining 30.9% of the total explained variability. Local scale variables such as in-stream substratum, vegetation in and near the stream (riparian zone), and some chemical variables were most strongly associated with the among-site variability. Local physical (24.4%) and local chemical (20.4%) variables explained the largest part of the among-site variability of community assemblages. These results are of importance when planning conservation and management measurements, implementing large-scale biomonitoring programs, and predicting how human alterations will affect running water ecosystems.This revised version was published online in May 2005 with corrections to the Cover Date.     

Influence of forest cover on in-stream large wood in an agricultural landscape of southeastern Brazil: a multi-scale analysis

Large wood (LW) is critical to the structure and function of streams and forests are the main LW source to stream channels. To assess the influence of forest cover changes at different spatial scales on in-stream LW quantity, we selected eighteen catchments (2nd–4th order) in Southeastern Brazil with forests at different levels of alterations. In each catchment we quantified the pattern of forest cover (% cover and relative catchment position), the physical characteristics of catchments (elevation and slope), the characteristics of channels (wetted channel width and depth), the abundance and volume of in-stream LW, and the frequency of LW pools. We used simple and multiple linear regression to assess the response of LW variables to landscape and stream reach variables. Most of the LW was relatively small; 72 % had a diameter <20 cm, and 66 % had a length <5 m. Although percent forest cover at reach scale had substantial support to explain LW variables, the best predictors of LW variables were forest cover at broader scales (LW abundance and LW pool frequency were best predicted by forest at intermediate distance at the catchment scale and LW volume was best predicted by forest cover at the drainage network scale), suggesting that downstream transport is an important process in addition to local processes in our study area. These findings have important management implications because although low forested reaches receive less LW from local forests (or no LW in the case of deforested stream reaches), they are receiving LW from upstream forested reaches. However, the material is generally small, unstable and likely to be easily flushed. This suggests that not only should riparian forest conservation encompass the full drainage network, but forests should also be allowed to regenerate to later successional stages to provide larger, higher quality LW for natural structuring of streams.     

The Importance of Spatial Scale for Conservation and Assessment of Anuran Populations in Coastal Wetlands of the Western Great Lakes,USA

Distributions of pond-breeding amphibians may be influenced by habitat factors at different spatial scales. We used anuran calling surveys to investigate the association between 5 anuran species and habitat variables measured within 100, 500, 1000, and 3000 m of sampling points at 63 coastal wetlands along the US shores of Lake Michigan and Lake Huron. Stepwise logistic regression was used to create predictive models for each species at each spatial scale. Our results confirm the view that habitat variables at multiple scales influence frog distributions, but the strength of predictive models was significantly affected by the spatial scale at which habitat variables were derived. Remotely sensed habitat variables within a 3000 m radius were among the most effective predictors of occurrence for American toad (Bufo americanus), eastern gray treefrog (Hyla versicolor), spring peeper (Pseudacris crucifer), and green frog (Rana clamitans melanota). The western chorus frog (Pseudacris triseriata) was predicted most effectively by variables derived within a 500 m radius. For the most part, these anurans exhibited species-specific responses to habitat variables; however the suite of landscape-scale variables associated with urban land use appeared in all species’ regression models. Associations with landscape-scale variables coupled with well-documented habitat needs at local breeding sites suggest that conservation and assessment of frogs and toads in coastal wetlands should consider the influence of habitat variables at multiple spatial scales.     

Does the scale of our observational window affect our conclusions about correlations between endangered salmon populations and their habitat?

Differences in the strength of species-habitat relationships across scales provide insights into the mechanisms that drive these relationships and guidance for designing in situ monitoring programs, conservation efforts and mechanistic studies. The scale of our observation can also impact the strength of perceived relationships between animals and habitat conditions. We examined the relationship between geographic information system (GIS)-based landscape data and Endangered Species Act-listed anadromous Pacific salmon (Oncorhynchus spp.) populations in three subbasins of the Columbia River basin, USA. We characterized the landscape data and ran our models at three spatial scales: local (stream reach), intermediate (6th field hydrologic units directly in contact with a given reach) and catchment (entire drainage basin). We addressed three questions about the effect of scale on relationships between salmon and GIS representations of landscape conditions: (1) at which scale does each predictor best correlate with salmon redd density, (2) at which scale is overall model fit maximized, and (3) how does a mixed-scale model compare with single scale models (mixed-scale meaning models that contain variables characterized at different spatial scales)? We developed mixed models to identify relationships between redd density and candidate explanatory variables at each of these spatial scales. Predictor variables had the strongest relationships with redd density when they were summarized over the catchment scale. Meanwhile strong models could be developed using landscape variables summarized at only the local scale. Model performance did not improve when we used suites of potential predictors summarized over multiple scales. Relationships between species abundance and land use or intrinsic habitat suitability detected at one scale cannot necessarily be extrapolated to other scales. Therefore, habitat restoration efforts should take place in the context of conditions found in the associated watershed or landscape.     

Land-use changes in Illinois,ASA: The influence of landscape attributes on current and historic land use

The Illinois Geographic Information System was used to compare the soil and landscape attributes of the State with its historic vegetation, current land use, and patterns of land-use change over the past 160 years. Patch structural characteristics among land types in four geographic zones were also compared. The assessment of patch characteristics revealed a highly modified State with most land patches controlled by human influences and relatively few by topographic and hydrologic features. Correlation and regression analyses determined the relationships of land type and abundance within each of 50 general soil associations to properties of the soil associations - typically slope, texture, organic matter, productivity index, and available waterholding capacity. The distribution of the historic vegetation of the State and its current deciduous forests and nonforested wetlands related moderately (r² 0.44) to various landscape attributes. Urban and other highly modified land types were less closely related.     

The change of a satoyama landscape and its causality in Kamiseya,Kyoto Prefecture,Japan between 1970 and 1995

We focused on patterns of land use in a particular satoyama landscape (Japanese traditional rural landscape, comprised of an integral social and ecological network of a village and its surroundings, such as agricultural lands, open forestlands and forests), and the effects of human activities upon them during Japan’s economic growth of the last few decades. Changes of landscape patterns and their probable causes were traced since the beginning of the 1900s to the present, and clarified. Societal, economic and technological changes, especially those that occurred after 1970, were considered the focal points from which major landscape changes developed. We compared the spatial features, patterns of land use and landscape diversities of each land unit, defined in terms of both their natural and man-made conditions for the year 1970, to those of 1995. We found land-use diversity to be strongly related to changes in the patterns of land use, with a decrease in diversity for all land units after 1970. Diversity of forest-age distribution on the other hand, increased. These changes, with the complex, changing patterns of each land unit, could be explained by differences in accessibility from the village and variations in the topography, as well as land ownership of the land units. We selected those land units found to have responded to these factors between 1970 and 1995, and classified them into four types of pattern changes, determined mainly by accessibility and topography.This revised version was published online in May 2005 with corrections to the Cover Date.     

Integrating local breeding pond, landcover, and climate factors in predicting amphibian distributions

Species distributions are influenced by many processes operating over varying spatial scales. The development of species distribution models (SDMs), also known as ecological niche models, has afforded the opportunity to predict the distributions of diverse taxa across broad geographic areas and identify variables that are potentially important in regulating these distributions. However, the integration of site-specific habitat data with broad scale climate and landcover data has received limited attention in an SDM framework. We investigate whether SDMs developed with breeding pond, landcover, and climate variables can accurately predict the distributions of nine pond-breeding amphibians in eastern Missouri, USA. Additionally we investigate the relative influences of each environmental variable on the distribution predictions for each study species, and whether the most influential variables are shared among multiple taxa. Boosted regression tree (BRT) SDMs were developed for each species with 38 abiotic and biotic environmental variables, including data from the breeding ponds, surrounding landcover, and climate. To test the models, field surveys were performed in 2007 and 2008 at 103 ponds for nine amphibian species. BRT models developed with breeding pond, landcover, and climate data accurately predicted the occurrences of six of nine species across the study area. Furthermore, the presence of each species was best predicted by a unique combination of environmental variables. Results also suggest that landcover and climate factors may be more influential for species near the edge of their geographic ranges, while local breeding pond factors may be more important for species nearer to the center of their ranges.     

Scale-dependent importance of environment, land use and landscape structure for species richness and composition of SE Norwegian modern agricultural landscapes

Knowledge of variation in vascular plant species richness and species composition in modern agricultural landscapes is important for appropriate biodiversity management. From species lists for 2201 land-type patches in 16 1-km² plots five data sets differing in sampling-unit size from patch to plot were prepared. Variation in each data set was partitioned into seven sources: patch geometry, patch type, geographic location, plot affiliation, habitat diversity, ecological factors, and land-use intensity. Patch species richness was highly predictable (75% of variance explained) by patch area, within-patch heterogeneity and patch type. Plot species richness was, however, not predictable by any explanatory variable, most likely because all studied landscapes contained all main patch types – ploughed land, woodland, grassland and other open land – and hence had a large core of common species. Patch species composition was explained by variation along major environmental complex gradients but appeared nested to lower degrees in modern than in traditional agricultural landscapes because species-poor parts of the landscape do not contain well-defined subsets of the species pool of species-rich parts. Variation in species composition was scale dependent because the relative importance of specific complex gradients changed with increasing sampling-unit size, and because the amount of randomness in data sets decreased with increasing sampling-unit size. Our results indicate that broad landscape structural changes will have consequences for landscape-scale species richness that are hard or impossible to predict by simple surrogate variables.     

Spatial patterns of macroinvertebrate functional feeding groups in streams in relation to physical variables and land-cover in Southwestern France

Artificial neural networks were used to quantify the distribution of macroinvertebrate functional feeding groups (FFGs) in relation to physical variables and to land-cover in the Adour–Garonne stream system (SW France; 116,000 km²). The relative abundances of 5 FFGs were calculated from macroinvertebrate data recorded at 165 sampling sites. Each site was characterized using 5 physical variables (elevation, stream order, stream width, distance from the source, slope) and 3 land-cover variables (% forested, % urban areas, % agricultural areas). The sites were first classified using the Self-Organizing Map algorithm (SOM), according to the physical and land-cover variables. Two major clusters of sites corresponded to anthropogenically modified and natural areas, respectively. Anthropogenically modified areas were clearly divided into agricultural and urban landscapes. Each major cluster was divided into 3–4 subsets of sites according to a topographic gradient of physical variables. To examine the variability of the communities, FFG proportions at the 165 sites were examined on the SOM trained with physical and land-cover variables. When the riverine landscape was natural, FFG patterns responded to the upstream–downstream gradient in physical variables. When the landscape was altered by agriculture or urbanization, the effects of land-cover on FFGs overcame the influence of the physical variables. The categorization of the landscape into forested, agricultural, and urban areas was relevant to detect changes in FFG patterns. In light of increasing development along riparian zones, the use of SOMs to detect responses of FFGs to landscape alterations at regional scales exemplifies an effective technique for assessing river health based on ecological indicator groups.