A GIS-based index for relating landscape characteristics to potential nitrogen leaching to wetlands

We developed a spatially-explicit, quantitative Nitrogen Leaching Index to assess the potential for non-point source subsurface nitrogen pollution to wetlands. The index was based on the leaching potential of the watershed soils, the amount of nitrogen available for leaching, and the spatial position of nitrogen sources in the watershed. A raster or cell-based geographic information system (GIS) was used to estimate the necessary data inputs for calculating the index, such as soil hydrologic group, land use/soil type combination, groundwater residence time, and location of septic systems. The Total and Average Watershed Nitrogen Leaching Index (TWNLI and AWNLI) were calculated by summing and averaging, respectively, individual cell contributions over a watershed.Analysis of nine wetland watersheds in central New York state, USA, with mixed forest and agricultural land uses illustrated the use of the index for identifying and ranking wetlands with potential nitrogen pollution. Results showed that the spatial characteristics of a watershed potentially can effect subsurface nitrogen delivery to groundwater-dominated wetlands. The use of an index based on watershed soils, topography, and land use may be useful for assessing potential nitrogen pollution to wetlands at a regional scale.     

Models of presence–absence estimate abundance as well as (or even better than) models of abundance: the case of the butterfly Parnassius apollo

Models relating species distribution records to environmental variables are increasingly applied to biodiversity conservation. Such techniques could be valuable to predict the distribution, abundance or habitat requirements of species that are rare or otherwise difficult to survey. However, despite widely-documented positive intraspecific relationships between occupancy and abundance, few studies have demonstrated convincing associations between models of habitat suitability based on species occurrence, and observed measures of habitat quality such as abundance. Here we compared models based on field-derived abundance and distribution (presence–absence) data for a rare mountain butterfly in 2006–2008. Both model types selected consistent effects of environmental variables, which corresponded to known ecological associations of the species, suggesting that abundance and distribution may be a function of similar factors. However, the models based on occurrence data identified stronger effects of a smaller number of environmental variables, indicating less uncertainty in the factors controlling distribution. Furthermore, cross-validation of the models using observed abundance data from different years, or averaged across years, suggested a marginally stronger ability of models based on occurrence data to predict observed abundance. The results suggest that, for some species, distribution models could be efficient tools for estimating habitat quality in conservation planning or management, when information on abundance or habitat requirements is costly or impractical to obtain.     

Exploring the impact of green space health on runoff reduction using NDVI

This study examines the impact of green space health on local flooding based on the analysis of eighty-two watersheds in four Texas metropolitan statistical areas: Dallas, Houston, San Antonio, and Austin. The runoff records in October 2007 and October 2012 were selected for the assessment. The study met the methodological challenge posed by comparison by using the Normalized Difference Vegetation Index (NDVI) datasets produced based on the Moderate Resolution Imaging Spectroradiometer (MODIS) imagery of the 250-m resolution as a proxy to represent the health of green space. Two linear regression models were employed to explain the variation in mean daily runoff depth in 2007 and 2012, while controlling multiple contextual variables.Results indicate that watersheds containing healthier green spaces were likely to generate lower amounts of runoff in both periods. Standardized coefficients of green space health also show that the NDVI is a powerful and significant predictor to explain variation in runoff. These findings illustrate the important role of urban green spaces in attenuating local flooding and may provide planners and decision-makers with a method to consider, using this kind of objective greenery index in further developing local and regional green infrastructure and land-use plans.     

Uncertainty analysis of a spatial habitat suitability model and implications for ecological management of water bodies

Habitat suitability index (HSI) models have been generally accepted in ecological management as a means to predict effects of pressures and restoration measures on habitats and populations. HSI-models estimate habitat suitability from relevant habitat variables. Because outcomes of HSI-studies may have significant consequences, it is crucial to have insight into the uncertainties of the predictions. In this study a method for uncertainty analysis, using Monte Carlo simulations, was developed and applied for a HSI-model of pondweed (Potamogeton pectinatus) in Lake IJsselmeer, The Netherlands. Uncertainties in both habitat model functions and in input data were considered. The magnitude of the uncertainties in model functions were estimated by a panel of experts, and the uncertainty was highest at intermediate values of the suitability index (0.4–0.6). Uncertainty in the predicted habitat suitability is spatially correlated with variations in environmental habitat variables such as water quality and substrate. The estimated uncertainty may be considered acceptable for the purposes of water management, namely directing ecological rehabilitation and conservation activities. However, the uncertainties may be too high to meet the accuracy requirements of legislation such as the EU Bird and Habitat directive.     

Changes in large-scale patterns of plant biodiversity predicted from environmental economic scenarios

In the industrialized world large sums of money are spent on measures to preserve biodiversity by improving environmental quality. This creates a need to evaluate the effectiveness of such measures. In response we devel oped a model, NTM, that links plant biodiversity to abiotic variables that are under human control. These vari ables are: vegetation management, and the soil variables groundwater level, and nitrogen availability. We used species richness and the criteria of the Red Lists, i.e., the rarity and decline per species as measure for potential changes in biodiversity. NTM uses a statistical approach, and models potential plant biodiversity based on the above criteria as a non-linear function of the three soil variables. The regression model is calibrated on a data set consisting of 33,706 vegetation relevés. Because field data of vegetation combined with measurements of soil variables are insufficiently available, we used the average of Ellenbergs indicator values of the species in each relevé as a proxy. NTM was subjected to both validation and uncertainty analysis. The validation was car ried out by comparison with an independent data set. The uncertainty analysis showed that uncertainty in abso lute biodiversity values is large, but that comparative scenario studies can be carried out with an acceptable uncertainty. As an example we show the evaluation of the impact of three European economic scenarios on po tential plant biodiversity in the Netherlands. Although there were differences per vegetation type and per region, potential plant biodiversity had a tendency to increase, with the highest increase for the scenario with the highest reduction in atmospheric deposition of nitrogen and acidity.This revised version was published online in May 2005 with corrections to the Cover Date.     

Characterizing watershed-delineated landscapes in Pennsylvania using conditional entropy profiles

When the objective is to characterize landscapes with respect to relative degree and type of forest (or other critical habitat) fragmentation, it is difficult to decide which variables to measure and what type of discriminatory analysis to apply. It is also desirable to incorporate multiple measurement scales. In response, a new method has been developed that responds to changes in both the marginal and spatial distributions of land cover in a raster map. Multiscale features of the map are captured in a sequence of successively coarsened resolutions based on the random filter for degrading raster map resolutions. Basically, the entropy of spatial pattern associated with a particular pixel resolution is calculated, conditional on the pattern of the next coarser parent resolution. When the entropy is plotted as a function of changing resolution, we obtain a simple two-dimensional graph called a conditional entropy profile, thus providing a graphical visualization of multi-scale fragmentation patterns.Using eight-category raster maps derived from 30-meter resolution LANDSAT Thematic Mapper images, the conditional entropy profile was obtained for each of 102 watersheds covering the state of Pennsylvania (USA). A suite of more conventional single-resolution landscape measurements was also obtained for each watershed using the FRAGSTATS program. After dividing the watersheds into three major physiographic provinces, cluster analysis was performed within each province using various combinations of the FRAGSTATS variables, land cover proportions and variables describing the conditional entropy profiles. Measurements of both spatial pattern and marginal land cover proportions were necessary to clearly discriminate the watersheds into distinct clusters for most of the state; however, the Piedmont province essentially only required the land cover proportions. In addition to land cover proportions, only the variables describing a conditional entropy profile appeared to be necessary for the Ridge and Valley province, whereas only the FRAGSTATS variables appeared to be necessary for the Appalachian Plateaus province. Meanwhile, the graphical representation of conditional entropy profiles provided a visualization of multi-scale fragmentation that was quite sensitive to changing pattern.     

Wetland effects on lake water quality in the Minneapolis/St. Paul metropolitan area

A method developed to evaluate the cumulative effect of wetland mosaics on water quality was applied to 33 lake watersheds in the seven-county region surrounding Minneapolis-St. Paul, Minnesota. A geographic information system (GIS) was used to record and measure landscape variables derived from aerial photos. Twenty-seven watershed land-use and land-cover variables were reduced to eight principal components which described 85% of the variance among watersheds. Relationships between lake water quality variables and the first six principal components plus an index of lake mixis were analyzed through stepwise multiple regression analysis. A combination of three landscape components (wetland/watershed area, agriculture/wetlands, and forest/soils components) explained 49% of the variance in a trophic state index, even though most of the lakes examined were already highly eutrophic, and thus were influenced by internal loading. The regression equations explained a range of 14 to 76% of the variation in individual water quality variables. Forested land-use was associated with lower lake trophic state, chloride, and lead. High lake trophic state was associated with agricultural land-use and with wetland distance from the lake of interest. The extent of wetlands was associated with low total lead and high color in lakes downstream. Wet meadows or herbaceous, seasonally-flooded wetlands contributed more to lake water color than did cattail marshes.     

Land-use history and topographic gradients as driving factors of subalpine Larix decidua forests

European larch (Larix decidua Mill.) forests in the Alps are cultural landscapes that have been shaped by humans for centuries through traditional management. Biological and historical data sources were employed, and a multi-scale approach was adopted to capture the influence of factors affecting the structure of these forests. Landscape and stand scale dynamics were analyzed in four watersheds (c. 13,000 ha) of the western and central Italian Alps that have experienced different land-use intensities. Observed landscape changes were generalized using path analyses developed from a common conceptual model. Stand structure and a range of environmental variables were sampled in 203 circular plots, and land use and anthropogenic variables were derived from thematic maps and aerial photographs. We used multivariate statistical analyses (ordination and SEM models) to relate forest structure, anthropogenic influences, land uses, and topography. The most commonly observed land cover transition was an expansion of forests at the expense of open areas. All studied watersheds were dominated by larch forests, but their structure and spatial pattern differed greatly. Anthropogenic variables were less important at Ventina, the least accessible site, but emerged as fundamental to explain stand structure in the other study sites. Complexity of topography and proximity to roads had influenced past human activities mainly in the most accessible sites. Regeneration density was greatest at lower elevations and closer to human settlements. Quantification of the role played by forest harvesting and cattle grazing in past centuries is critical for understanding how global change factors may influence future dynamics of mountain forests in the European Alps and similar cultural landscapes worldwide.     

Nonparametric machine learning for mapping forest cover and exploring influential factors

Context

The contribution of forest ecosystem services to human well-being varies over space following the dynamics in forest cover. Use of machine learning models is increasing in projecting forest cover changes and investigating the drivers, yet references are still lacking for selecting machine learning models for spatial projection of forest cover patterns.

Objectives

We assessed the ability of nonparametric machine learning techniques to project the spatial distribution of forest cover and identify its drivers using a case study of Tasmania, Australia.

Methods

We developed, evaluated, and compared the performance of four nonparametric machine learning models: support vector regression (SVR), artificial neural networks (ANN), random forest (RF), and gradient boosted regression trees (GBRT).

Results

The results demonstrated that RF far outperformed the other three models in both fitting and projection accuracy, and required less computional costs. GBRT outperformed SVR and ANN in projection accuracy. However, RF exhibited serious overfitting due to the full growth of its decision trees. The influence rankings of explanatory variables on spatial patterns of forest cover were different under the four models. Land tenure type and rainfall were identified among the top four most influential variables by all four models. The ranking produced by the RF model was significantly different with topographic factors associated with land clearing and production costs (elevation and distance to timber facilities) being the two most influential variables.

Conclusions

We encourage practitioners to consider nonparametric machine learning methods, especially RF, when facing problems of complex environmental data modelling.

     

Landscape context matters: local habitat and landscape effects on the abundance and patch occupancy of collared lizards in managed grasslands

The distribution and abundance of a species may be simultaneously influenced by both local-scale habitat features and the broader patch and landscape contexts in which these populations occur. Different factors may influence patch occupancy (presence–absence) versus local abundance (number of individuals within patches), and at different scales, and thus ideally both occupancy and abundance should be investigated, especially in studies that seek to understand the consequences of land management on species persistence. Our study evaluated the relative influences of variables associated with the local habitat patch, hillside (patch context), and landscape context on patch occupancy and abundance of the collared lizard (Crotaphytus collaris) within tallgrass prairie managed under different fire and grazing regimes in the northern Flint Hills of Kansas, USA. Using a multi-model information-theoretic approach that accounted for detection bias, we found that collared lizard abundance and occupancy was influenced by factors measured at both the local habitat and landscape scales. At a local scale, collared lizard abundance was greatest on large rock ledges that had lots of crevices, high vegetation complexity, and were located higher up on the hillslope. At the landscape scale, collared lizard abundance and occupancy were both higher in watersheds that were burned frequently (1–2 year intervals). Interestingly, grazing only had a significant effect on occupancy and abundance within less frequently burned (4-year burn interval) watersheds. Our results suggest that, in addition to the obvious habitat needs of this species (availability of suitable rock habitat), land-management practices have the potential to influence collared lizard presence and abundance in the grasslands of the Flint Hills. Thus, mapping the availability of suitable habitat is unlikely to be sufficient for evaluating species distributions and persistence in such cases without consideration of landscape management and disturbance history.     

Utilizing GIS Technologies in Selection of Suitable Vineyard Sites

Abstract

Geographic information system (GIS) technologies and a weighted linear indexing model were used for suitability analysis of potential vineyard sites in Illinois. The model included a macroscale climate variables layer (40 points), a mesoscale climate variables layer (40 points), a soil properties layer (10 points) and a current land use variables layer (10 points) for a possible 100 points. Macroscale climate variables, growing degree day summation for a 33-year period (1969-2002) and occurrence of -26°C were interpolated using thin plate smoothing splines over the Illinois terrain using 100 m2 resolution digital elevation models (DEM). Using the same DEMs, absolute elevation, slope, and aspect were reclassified using surface analysis of the terrain to model the effects of mesoscale climate variables in Jackson and Union Counties in Illinois (study area). Locations in the study area above 259 m in absolute elevation (above sea level), with gently rolling slopes (5 to 10%), and facing North, East, or Northeast received the most points within the mesoscale climate layer. Soils that drained well or moderately well with moderate organic matter content (2 to 3%) received the most points in the soil properties layer. Lands that promoted ease of conversion to vineyards were assigned the most points in the current land use layer. The resulting four layers were arithmetically summed and suitability maps with 760 m2 resolution were developed. In the study area, over 18,155 hectares of land highly suitable or suitable for viticulture were identified. A portion of the existing orchard and vineyard acreage (223 ha) in the study area were surveyed with a global positioning system. Of the surveyed vineyard acreage, over 81% lies within the suitable or better ranking according to the model. Of the surveyed orchard acreage, over 50% lies within the suitable or better ranking.     

Effects of stream map resolution on measures of riparian buffer distribution and nutrient retention potential

Riparian ecosystems are interfaces between aquatic and terrestrial environments recognized for their nutrient interception potential in agricultural landscapes. Stream network maps from a broad range of map resolutions have been employed in watershed studies of riparian areas. However, map resolution may affect important attributes of riparian buffers, such as the connectivity between source lands and small stream channels missing in coarse resolution maps. We sought to understand the influence of changing stream map resolution on measures of the river network, near-stream land cover, and riparian metrics. Our objectives were: (1) to evaluate the influence of stream map resolution on measures of the stream network, the character and extent of near-stream zones, and riparian metrics; (2) to compare patterns of variation among different physiographic provinces; and (3) to explore how predictions of nutrient retention potential might be affected by the resolution of a stream map. We found that using fine resolution stream maps significantly increased our estimates of stream order, drainage density, and the proportion of watershed area occurring near a stream. Increasing stream map resolution reduced the mean distance to source areas as well as mean buffer width and increased the frequency of buffer gaps. Measures of percent land cover within 100 m of streams were less sensitive to stream map resolution. Overall, increasing stream map resolution led to reduced estimates of nutrient retention potential in riparian buffers. In some watersheds, switching from a coarse resolution to a fine resolution stream map completely changed our perception of a stream network from well buffered to largely unbuffered. Because previous, broad-scale analyses of riparian buffers used coarse-resolution stream maps, those studies may have overestimated landscape-level buffer prevalence and effectiveness. We present a case study of three watersheds to demonstrate that interactions among stream map resolution and land cover patterns make a dramatic difference in the perceived ability of riparian buffers to ameliorate effects of agricultural activities across whole watersheds. Moreover, stream map resolution affects inferences about whether retention occurs in streams or riparian zones.     

Watershed analysis of forest fragmentation by clearcuts and roads in a Wyoming forest

Remotely sensed data and a Geographic Information System were used to compare the effects of clearcutting and road-building on the landscape pattern of the Bighorn National Forest, in north-central Wyoming. Landscape patterns were quantified for each of 12 watersheds on a series of four maps that differed only in the degree of clearcutting and road density. We analyzed several landscape pattern metrics for the landscape as a whole and for the lodgepole pine and spruce/fir cover classes across these maps, and determined the relative effects of clearcutting and road building on the pattern of each watershed. At both the landscape- and cover class-scales, clearcutting and road building resulted in increased fragmentation as represented by a distinct suite of landscape structural changes. Patch core area and mean patch size decreased, and edge density and patch density increased as a result of clearcuts and roads. Clearcuts and roads simplified patch shapes at the landscape scale, but increased the complexity of lodgepole pine patches. Roads appeared to be a more significant agent of change than clearcuts, and roads which were more evenly distributed across a watershed had a greater effect on landscape pattern than did those which were densely clustered. Examining individual watersheds allows for the comparison of fragmentation among watersheds, as well as across the landscape as a whole. Similar studies of landscape structure in other National Forests and on other public lands may help to identify and prevent further fragmentation of these areas.     

Transitions in forest fragmentation: implications for restoration opportunities at regional scales

Where the potential natural vegetation is continuous forest (e.g., eastern US), a region can be divided into smaller units (e.g., counties, watersheds), and a graph of the proportion of forest in the largest patch versus the proportion in anthropogenic cover can be used as an index of forest fragmentation. If forests are not fragmented beyond that converted to anthropogenic cover, there would be only one patch in the unit and its proportional size would equal 1 minus the percentage of anthropogenic cover. For a set of 130 watersheds in the mid-Atlantic region, there was a transition in forest fragmentation between 15 and 20% anthropogenic cover. The potential for mitigating fragmentation by connecting two or more disjunct forest patches was low when percent anthropogenic cover was low, highest at moderate proportions of anthropogenic cover, and again low as the proportion of anthropogenic cover increased toward 100%. This fragmentation index could be used to prioritize locations for restoration by targeting watersheds where there would be the greatest increase in the size of the largest forest patch.     

A method for the use of landscape metrics in freshwater research and management

Freshwater research and management efforts could be greatly enhanced by a better understanding of the relationship between landscape-scale factors and water quality indicators. This is particularly true in urban areas, where land transformation impacts stream systems at a variety of scales. Despite advances in landscape quantification methods, several studies attempting to elucidate the relationship between land use/land cover (LULC) and water quality have resulted in mixed conclusions. However, these studies have largely relied on compositional landscape metrics. For urban and urbanizing watersheds in particular, the use of metrics that capture spatial pattern may further aid in distinguishing the effects of various urban growth patterns, as well as exploring the interplay between environmental and socioeconomic variables. However, to be truly useful for freshwater applications, pattern metrics must be optimized based on characteristic watershed properties and common water quality point sampling methods. Using a freely available LULC data set for the Santa Clara Basin, California, USA, we quantified landscape composition and configuration for subwatershed areas upstream of individual sampling sites, reducing the number of metrics based on: (1) sensitivity to changes in extent and (2) redundancy, as determined by a multivariate factor analysis. The first two factors, interpreted as (1) patch density and distribution and (2) patch shape and landscape subdivision, explained approximately 85% of the variation in the data set, and are highly reflective of the heterogeneous urban development pattern found in the study area. Although offering slightly less explanatory power, compositional metrics can provide important contextual information.     

The historic influence of dams on diadromous fish habitat with a focus on river herring and hydrologic longitudinal connectivity

The erection of dams alters habitat and longitudinal stream connectivity for migratory diadromous and potamodromous fish species and interrupts much of organismal exchange between freshwater and marine ecosystems. In the US, this disruption began with colonial settlement in the seventeenth century but little quantitative assessment of historical impact on accessible habitat and population size has been conducted. We used published surveys, GIS layers and historical documents to create a database of 1356 dams, which was then analyzed to determine the historical timeline of construction, use and resultant fragmentation of watersheds in Maine, US. Historical information on the anadromous river herring was used to determine natural upstream boundaries to migration and establish total potential alewife spawning habitat in nine watersheds with historic populations. Dams in Maine were constructed beginning in 1634 and by 1850 had reduced accessible lake area to less than 5% of the virgin 892 km² habitat and 20% of virgin stream habitat. There is a near total loss of accessible habitat by 1860 that followed a west-east pattern of European migration and settlement. Understanding historic trends allows current restoration targets to be assessed and prioritized within an ecosystem-based perspective and may inform expectations for future management of oceanic and freshwater living resources.     

Land ownership and landscape structure: a spatial analysis of sixty-six Oregon (USA) Coast Range watersheds

Patterns of land ownership and forest cover are related in complex and ecologically significant ways. Using a Geographic Information System and regression analysis, we tested for spatial relationships between the structure of land ownership and forest cover across 66 watersheds in the state of Oregon (USA), Coast Range mountains. We found that in these watersheds (1) forest cover diversity increased with land ownership diversity, (2) size of forest patches increased with size of land ownership patches, and (3) connectivity of forest cover increased with connectivity of land ownership. Land ownership structure explained between 29% and 40% of the variability of forest cover structure across these watersheds. Driving this relationship are unique associations among particular ownership classes and various forest cover classes. The USDA Forest Service and the USDI Bureau of Land Management were associated with mature forest cover; private industry was associated with young forest cover; nonindustrial private forest owners were associated with a wide diversity of cover classes. Watersheds with mixed ownership appear to provide greater forest cover diversity, whereas watersheds with concentrated ownership provide less diverse but more connected forest cover. Results suggest that land ownership patterns are strongly correlated with forest cover patterns. Therefore, understanding landscape structure requires consideration of land ownership institutions, dynamics, and patterns.This revised version was published online in May 2005 with corrections to the Cover Date.     

Using indicators of land-use development intensity to assess the condition of coastal wetlands in Hawai‘i

Although wetland condition assessment procedures have been developed, validated, and calibrated in the continental United States, they have not yet been fully developed or field-tested for wetlands in Hawai‘i. In order to address the need for comprehensive assessment methods for Hawaiian coastal wetlands, our research compared three indicators of landscape condition (landscape development intensity, road density, and forest cover) with wetland condition as measured by rapid assessment methods (RAM) and detailed field data collected on soil and water quality. We predicted that wetlands located in the least developed landscapes would have more nutrient rich soils, yet lower nutrient levels in the surface water, and would receive the highest rapid assessment scores. The hypotheses of our study were generally supported. However, while the correlations between landscape variables and δ¹⁵N isotopes and CRAM scores were relatively strong, the correlations between the landscape indicators and the other Level II and III field indicators were not very strong. These results suggest that further calibration and refinement of metrics is needed in order to more accurately assess the condition of Hawaiian coastal wetlands. A more detailed land use map, in addition to more comprehensive assessments of wetland water quality and biotic integrity would likely improve the relationships between indicators of landscape condition and wetland condition. Nonetheless, our research demonstrated that landscape analysis at larger scales (1,000 m buffers and watersheds) could provide managers with valuable information on how regional stressors may be affecting wetland water quality (measured as δ¹⁵N in plant tissue) as well as overall wetland condition (RAM scores).     

Land use and host community characteristics as predictors of disease risk

Land use has been implicated as a major causal factor in the recent surge of emerging and reemerging zoonotic diseases worldwide. Whirling disease, a parasitic infection caused by the myxozoan, Myxobolus cerebralis, has led to major declines in wild trout populations within the Intermountain West of the USA and is suspected to be exacerbated by land and stream management practices that create favorable habitat for the oligochaete host, Tubifex tubifex. Our objectives were to quantify relationships between whirling disease risk and (1) land use and (2) characteristics of the oligochaete host community within four major watersheds in western Montana. Risk was quantified by the severity and prevalence of infection in caged sentinel rainbow trout. A geographic information system (GIS) was used to model land use (e.g., agriculture, mines) within watersheds. Importance of predictor variables was assessed using regression tree and random forest analyses. A low proportion of riparian forest, high road density, high oligochaete density, and high Limnodrilus hoffmeisteri density were identified as important predictors of high risk. Although we did not examine the mechanisms underlying these correlations, a low proportion of riparian forest and a high road density likely increase stream sedimentation, thus, increasing habitat for oligochaetes. Although relationships identified are not necessarily causal, the ability to predict areas most at risk of M. cerebralis establishment and proliferation using broad scale predictors should serve as a useful management tool within Montana and elsewhere.     

Small-scale effects of historical land use and topography on post-cultural tree species composition in an Alpine valley in southern Switzerland

Investigations of spatial patterns in forest tree species composition are essential in the understanding of landscape dynamics, especially in areas of land-use change. The specific environmental factors controlling the present patterns, however, vary with the scale of observation. In this study we estimated abundance of adult trees and tree regeneration in a Southern Alpine valley in Ticino, Switzerland. We hypothesized that, at the present scale, spatial pattern of post-cultural tree species does not primarily depend on topographic features but responds instead to small-scale variation in historical land use. We used multivariate regression trees to relate species abundances to environmental variables. Species matrices were comprised of single tree species abundance as well as species groups. Groups were formed according to common ecological species requirements with respect to shade tolerance, soil moisture and soil nutrients. Though species variance could only be partially explained, a clear ranking in the relative importance of environmental variables emerged. Tree basal area of formerly cultivated Castanea sativa (Mill.) was the most important factor accounting for up to 50% of species’ variation. Influence of topographic attributes was minor, restricted to profile curvature, and partly contradictory in response. Our results suggest the importance of biotic factors and soil properties for small-scale variation in tree species composition and need for further investigations in the study area on the ecological requirements of tree species in the early growing stage.