Predicting Southern Appalachian overstory vegetation with digital terrain data

Vegetation in mountainous regions responds to small-scale variation in terrain, largely due to effects on both temperature and soil moisture. However there are few studies of quantitative, terrain-based methods for predicting vegetation composition. This study investigated relationships between forest composition, elevation, and a derived index of terrain shape, and evaluates methods for predicting forest composition. Trees were measured on 406 permanent plots within the boundaries of the Coweeta Hydrologic Lab, located in the Southern Appalachian Mountains of western North Carolina, USA. All plots were in control watersheds, without human or major natural disturbance since 1923. Plots were 0.08 ha and arrayed on transects, with approximately 380 meters between parallel transects. Breast-height diameters were measured on all trees. Elevation and terrain shape (cove, ridge, sideslope) were estimated for each plot. Density (trees/ha) and basal area were summarized by species and by forest type (cove, xeric oak-pine, northern hardwoods, and mixed deciduous). Plot data were combined with a digital elevation data (DEM), and a derived index of terrain shape at two sampling resolutions: 30 m (US Geological Survey), and 80 m (Defense Mapping Agency) sources. Vegetation maps were produced using each of four different methods: 1) linear regression with and without log transformations against elevation and terrain variables combined with cartographic overlay, 2) kriging, 3) co-kriging, and 4) a mosaic diagram. Predicted vegetation was compared to known vegetation at each of 77 independent, withheld data points, and an error matrix was determined for each mapping method.We observed strong relationships between some species and elevation and/or terrain shape. Cove and xeric oak/pine species basal areas were positively and negatively related to concave landscape locations, respectively, while species typically found in the mixed deciduous and northern hardwood types were not. Most northern hardwood species occurred more frequently and at higher basal areas as elevation increased, while most other species did not respond to elevation. The regression and mosaic diagram mapping approaches had significantly higher mapping accuracies than kriging and co-kriging. There were significant effects of DEM resolution on map accuracy, with maps based on 30 m DEM data significantly more accurate than those based on 80 m data. Taken together, these results indicate that both the mapping method and terrain data resolution significantly affect the resultant vegetation maps, even when using relatively high resolution data. Landscape or regional models based on 100 m or lower resolution terrain data may significantly under-represent terrain-related variation in vegetation composition.     

Spatial patterns of fire occurrence in Catalonia,NE, Spain

In this paper, we analyse spatial patterns of fire occurrence in Catalonia (NE Spain) during 1975–98. Fire scar maps, discriminated by means of 30–60 m resolution remote sensing imagery, have been used as a source of fire occurrence. We employ several visual or analytical approaches to interpret fire occurrence in this region, such as those of Minnich and Chou (1997), Ricotta et al. (2001) or Krummel et al. (1987). Crucial spatial patterns such as fire size distribution, fire frequency distribution, spots and residual vegetation islands are documented. In addition, several geographical layers were overlaid with burned area maps in order to determine interactions between fire occurrence and environmental parameters such as altitude, slope, solar radiation, and burned land cover. Assuming that fire occurrence is well determined by such a posteriori empirical factors we detect areas most prone to fire in this region and aim to enhance the local forest management and conservation plans.     

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.     

Landscape composition is the strongest determinant of bird occupancy patterns in tropical forest patches

Context

Biodiversity in tropical region has declined in the last decades, mainly due to forest conversion into agricultural areas. Consequently, species occupancy in these landscapes is strongly governed by environmental changes acting at multiple spatial scales.

Objectives

We investigated which environmental predictors best determines the occupancy probability of 68 bird species exhibiting different ecological traits in forest patches.

Methods.

We conducted point-count bird surveys in 40 forest sites of the Brazilian Atlantic forest. Using six variables related to landscape composition and configuration and local vegetation structure, we predicted the occupancy probability of each species accounting for imperfect detections.

Results

Landscape composition, especially forest cover, best predicted bird occupancy probability. Specifically, most bird species showed greater occupancy probability in sites inserted in more forested landscapes, while some species presented higher occurrence in patches surrounded by low-quality matrices. Conversely, only three species showed greater occupancy in landscapes with higher number of patches and dominated by forest edges. Also, several species exhibited greater occupancy in sites harbouring either larger trees or lower number of understory plants. Of uttermost importance, our study revealed that a minimum of 54% of forest cover is required to ensure high (> 60%) occupancy probability of forest species.

Conclusions

We highlighted that maintaining only 20% of native vegetation in private property according to Brazilian environmental law is insufficient to guarantee a greater occupancy for most bird species. We recommend that policy actions should safeguard existing forest remnants, expand restoration projects, and curb human-induced disturbances to minimise degradation within forest patches.

     

Wildfires and landscape patterns in the Eastern Iberian Peninsula

The relations between disturbance regime and landscape patterns have been developed from a theoretical perspective, but few studies have tested these relations when forces promoting opposing heterogeneity patterns are simultaneously operating on a landscape. This work provides quantitative evidence of these relations in areas dominated by human activity, showing that landscape heterogeneity decreases disturbance spread. In turn, disturbance introduces a source of landscape heterogeneity, but it is not enough to counterbalance the homogeneity trend due to agricultural abandonment. Land cover changes and wildfire occurrence (fires larger than 0.3 km²) have been monitored in the Tivissa municipality (208.4 km²) (Catalonia, NE Spain) from 1956 to 1993. Land cover maps were obtained from 1956, 1978 and 1993 and they were overlaid with fire occurrence maps obtained for the 1975–1995 period from 60 m resolution remote sensing images, which allow the identification of burned areas by sudden drops in Normalized Difference Vegetation Index (NDVI). Changes in landscape patterns in relation to fire regime have been analyzed considering several parameters: patch density, mean patch size, mean distance to the nearest neighbour of the same category, edge density, and the Shannon diversity index. In the 1956–1993 period there is a trend to increasing landscape homogenization due to the expansion of shrub­lands linked to a decrease in forest surface, and to the abandonment of agricultural lands. This trend, however, is not constant along all the period. Fires are more likely to occur in woody, homogenous areas, increasing landscape heterogeneity, as observed in the 1978–1993 period. This increase in heterogeneity does not counterbalance the general trend to landscape homogenization as a consequence of agricultural abandonment and the coalescence of natural vegetation patches.This revised version was published online in May 2005 with corrections to the Cover Date.     

<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.     

Persistent changes in forest vegetation and seed bank 1,600 years after human occupation

Past land use is an important factor determining vegetation in temperate deciduous forests. Little is known about the long-term persistence of these impacts on vegetation but especially on the seed bank. This study assessed whether soil characteristics remain altered 1,600 years after human occupation and if this yielded persistent differences in forest plant communities and their seed bank in particular. Compiègne forest is located in northern-France and has a history of continuous forest cover since the end of Roman times. Twenty-four Gallo-Roman and 24 unoccupied sites were sampled and data were analysed using paired sample tests to investigate whether soil, vegetation and seed bank still differed significantly. The soil was persistently altered on the Gallo-Roman sites resulting in elevated phosphorus levels and pH (dependent on initial soil conditions) which translated into increased vegetation and seed bank species richness. Though spatially isolated, Gallo-Roman sites supported both a homogenized vegetation and seed bank. Vegetation differences were not the only driver behind seed bank differences. Similarity between vegetation and seed bank was low and the possibility existed that agricultural ruderals were introduced via the former land use. Ancient human occupation leaves a persistent trace on forest soil, vegetation and seed bank and appears to do so at least 1,600 years after the former occupation. The geochemical alterations created an entirely different habitat causing not only vegetation but also the seed bank to have altered and homogenized composition and characteristics. Seed bank differences likely persisted by the traditional forest management and altered forest environment.     

Landscape effects mediate breeding bird abundance in midwestern forests

We examine the influence of both local habitat and landscape variables on avian species abundance at forested study sites situated within fragmented and contiguous landscapes. The study was conducted over a six year period (1991–1996) at 10 study sites equally divided between the heavily forested Missouri Ozarks and forest fragments in central Missouri. We found greater species richness and diversity in the fragments, but there was a higher percentage of Neotropical migrants in the Ozarks. We found significant differences in the mean number of birds detected between the central Missouri fragments and the unfragmented Ozarks for 15 (63%) of 24 focal species. We used stepwise regression to determine which of 12 local vegetation variables and 4 landscape variables (forest cover, core area, edge density, and mean patch size) accounted for the greatest amount of variation in abundance for 24 bird species. Seven species (29%) were most sensitive to local vegetation variables, while 16 species (67%) responded most strongly to one of four landscape variables. Landscape variables are significant predictors of abundance for many bird species; resource managers should consider multiple measures of landscape sensitivity when making bird population management decisions.Order of first two authors decided by coin toss     

High resolution species distribution models of two nesting water bird species: a study of transferability and predictive performance

Species distribution modelling is increasingly used in ecological studies and is particularly useful in conservation planning. Models are, however, typically created with a coarse resolution, although conservation planning often requires a high resolution. In this study we created high resolution models and explored central aspects of the modelling procedure; transferability and predictive performance of the models. We created models for two breeding water bird species, common eider Somateria mollissima and herring gull Larus argentatus, based on data from two regions in the Finnish archipelago (234 islands). We used seven variables which we considered as potential predictors of nest site location: distance to forest, distance to rock and distance to low vegetation, exposure, elevation, slope and curvature of the land surface. We tested the predictive ability of the models crosswise between the areas by using area under the receiver operating characteristic curve. The models were transferable between our study areas and the predictive performance varied from fair to excellent. The most important predictors overall were exposure and distance to forest. More general models, with higher regularization values in the Maxent software, had better transferability regarding predictive performance. However, when we fitted a model based on 60% of the data from both regions and evaluated the model on the remaining 40%, the most complex model had the highest accuracy. Extrapolation of SDMs, evaluated on data from the same region, should therefore always be done with caution as the most accurate model might not have the best transferability if it is not general enough.     

Incorporating spatial non-stationarity of regression coefficients into predictive vegetation models

The results of predictive vegetation models are often presented spatially as GIS-derived surfaces of vegetation attributes across a landscape or region, but spatial information is rarely included in the model itself. Geographically weighted regression (GWR), which extends the traditional regression framework by allowing regression coefficients to vary for individual locations (‘spatial non-stationarity’), is one method of utilizing spatial information to improve the predictive power of such models. In this paper, we compare the ability of GWR, a local model, with that of ordinary least-squares (OLS) regression, a global model, to predict patterns of montane ponderosa pine (Pinus ponderosa) basal area in Saguaro National Park, AZ, USA on the basis of variables related to topography (elevation, slope steepness, aspect) and fire history (fire frequency, time since fire). The localized regression coefficients exhibited significant non-stationarity for four of the five environmental variables, and the GWR model consequently described the vegetation-environment data significantly better, even after accounting for differences in model complexity. GWR also reduced observed spatial autocorrelation of the model residuals. When applied to independent data locations not used in model development, basal areas predicted by GWR had a closer fit to observed values with lower residuals than those from the optimal OLS regression model. GWR also provided insights into fine-scale controls of ponderosa pine pattern that were missed by the global model. For example, the relationship between ponderosa pine basal area and aspect, which was obscured in the OLS regression model due to non-stationarity, was clearly demonstrated by the GWR model. We thus see GWR as a valuable complement to the many other global methods currently in use for predictive vegetation modeling.     

Response of high mountain landscape to topographic variables: Central pyrenees

An objective method for inductively modelling the distribution of mountain land units using GIS managed topographic variables is presented. The landscape of a small high mountain catchment in the Spanish Pyrenees, covered with grassland, was classified in ten land units by hierarchical agglomerative clustering, using a sample of 194 random plots, in which classes of vegetation, soils and landforms were defined. Additionally, seven layers of topographic variables (altitude, slope angle, aspect, solar radiation, topographic wetness index, specific catchment area, and regolith thickness) were created from a Digital Elevation Model. The affinity of each land unit to the topographic variables was calculated using Binary Discriminant Analysis (BDA), after dichotomising the latter around their mean values. Then, the distribution of each land unit was predicted by boolean operations combining step by step distributions for the seven topographic variables ordered, for each unit, after the absolute values of the Haberman’s residuals in BDA. The predicted distributions were tested (χ²) against that of the observed sampling plots. From the original ten land units, the distributions of eight of them were successfully predicted (four are related to the slope sequence, two reflect the water accumulation in the soil, and two respond to geomorphic processes) while the remaining two had to be rejected. Part of the catchment (39%) was not assigned to any land unit, probably because more distributed variables accounting for snow distribution are necessary.     

Modeling grain-size dependent bias in estimating forest area: a regional application

A better understanding of scaling-up effects on estimating important landscape characteristics (e.g. forest percentage) is critical for improving ecological applications over large areas. This study illustrated effects of changing grain sizes on regional forest estimates in Minnesota, Wisconsin, and Michigan of the USA using 30-m land-cover maps (1992 and 2001) produced by the National Land Cover Datasets. The maps were aggregated to two broad cover types (forest vs. non-forest) and scaled up to 1-km and 10-km resolutions. Empirical models were established from county-level observations using regression analysis to estimate scaling effects on area estimation. Forest percentages observed at 30-m and 1-km land-cover maps were highly correlated. This intrinsic relationship was tested spatially, temporally, and was shown to be invariant. Our models provide a practical way to calibrate forest percentages observed from coarse-resolution land-cover data. The models predicted mean scaling effects of 7.0 and 12.0% (in absolute value with standard deviations of 2.2 and 5.3%) on regional forest cover estimation (ranging from 2.3 and 2.5% to 11.1 and 23.7% at the county level) with standard errors of model estimation 3.1 and 7.1% between 30 m and 1 km, and 30 m and 10 km, respectively, within a 95% confidence interval. Our models improved accuracy of forest cover estimates (in terms of percent) by 63% (at 1-km resolution) and 57% (at 10-km resolution) at the county level relative to those without model adjustment and by 87 and 84% at the regional level in 2001. The model improved 1992 and 2001 regional forest estimation in terms of area for 1-km maps by 15,141 and 7,412 km² (after area weighting of all counties) respectively, compared to the corresponding estimates without calibration using 30 m-based regional forest areas as reference.     

Relationships between land cover and the surface urban heat island: seasonal variability and effects of spatial and thematic resolution of land cover data on predicting land surface temperatures

We investigated the seasonal variability of the relationships between land surface temperature (LST) and land use/land cover (LULC) variables, and how the spatial and thematic resolutions of LULC variables affect these relationships. We derived LST data from Landsat-7 Enhanced Thematic Mapper (ETM+) images acquired from four different seasons. We used three LULC datasets: (1) 0.6 m resolution land cover data; (2) 30 m resolution land cover data (NLCD 2001); and (3) 30 m resolution Normalized Difference Vegetation Index data derived from the same ETM+ images (though from different bands) used for LST calculation. We developed ten models to evaluate effects of spatial and thematic resolution of LULC data on the observed relationships between LST and LULC variables for each season. We found that the directions of the effects of LULC variables on predicting LST were consistent across seasons, but the magnitude of effects, varied by season, providing the strongest predictive capacity during summer and the weakest during winter. Percent of imperviousness was the best predictor on LST with relatively consistent explanatory power across seasons, which alone explained approximately 50 % of the total variation in LST in winter, and up to 77.9 % for summer. Vegetation related variables, particularly tree canopy, were good predictor of LST during summer and fall. Vegetation, particularly tree canopy, can significantly reduce LST. The spatial resolution of LULC data appeared not to substantially affect relationships between LST and LULC variables. In contrast, increasing thematic resolution generally enhanced the explanatory power of LULC on LST, but not to a substantial degree.     

A GIS-based method for the reconstruction of the late eighteenth century forest vegetation in the Prignitz region (NE Germany)

Our goal was to reconstruct the late eighteenth century forest vegetation of the Prignitz region (NE Germany) at a scale of 1:50,000. We also wanted to relate the historical forest vegetation to the actual and potential natural vegetation. For these purposes, we selected 15 woody species and transferred relevant data found in historical records from various sources together with the recent localities of (very) old individuals belonging to these woody species into ArcView GIS. Following multi-step data processing including the generation of a point density layer using a moving window with kernel estimation and derivation of vegetation units applying Boolean algebra rules together with information on site conditions, we derived 17 forest communities corresponding to the potential natural vegetation. We were able to reconstruct the historical forest vegetation for 90% of the forest area ca. 1780. Only two of the 17 forest communities covered large parts of the forested area. The oak forest with Agrostis capillaris covered about 44% of the total forest area, and alder forests on fenland made up about 37%. Oak-hornbeam forests with Stellaria holostea comprised slightly less than 6% of the forest area, while all other forest communities comprised less than 1%. The historical forest vegetation is more similar to the potential forest vegetation and quite different from the actual forest vegetation because coniferous tree species currently cover approximately two-thirds of the actual forest area. The most beneficial result of this study is the map of high-resolution historical vegetation units that may serve as the basis for various further studies, e.g., modelling long-term changes in biodiversity at the landscape scale.     

Landscape factors and regional differences in recovery rates of herb layer richness in Flanders (Belgium)

The recovery of understory plants in recent forests is critical for evaluating the overall capacity of landscapes to maintain plant biodiversity. Here we used a large data set of vegetation plots from the Flemish Forest Inventory in combination with maps of forest history and soil-based Potential Natural Vegetation to evaluate regional differences in the rate of recovery of understory plant diversity in three regions of Flanders, Belgium. We expressed the degree of recovery in recent forests using the species richness of ancient forests as a reference point, and found strong differences among regions in the average level of recovery. These differences appeared to be due to regional variation in average patch connectivity and age (ultimately stemming from differences in land use history) and – to a lesser extent – environmental conditions. We also found an increase in the proportional representation of vertebrate dispersed species and species with short-distance dispersal with increasing levels of recovery. Our results highlight the potential drivers of inter-regional variation in the process of recovery of plant diversity during restoration, and they emphasize the importance of historical and spatial context in the recovery process.     

Scaling Local Species-habitat Relations to the Larger Landscape with a Hierarchical Spatial Count Model

Much of what is known about avian species-habitat relations has been derived from studies of birds at local scales. It is entirely unclear whether the relations observed at these scales translate to the larger landscape in a predictable linear fashion. We derived habitat models and mapped predicted abundances for three forest bird species of eastern North America using bird counts, environmental variables, and hierarchical models applied at three spatial scales. Our purpose was to understand habitat associations at multiple spatial scales and create predictive abundance maps for purposes of conservation planning at a landscape scale given the constraint that the variables used in this exercise were derived from local-level studies. Our models indicated a substantial influence of landscape context for all species, many of which were counter to reported associations at finer spatial extents. We found land cover composition provided the greatest contribution to the relative explained variance in counts for all three species; spatial structure was second in importance. No single spatial scale dominated any model, indicating that these species are responding to factors at multiple spatial scales. For purposes of conservation planning, areas of predicted high abundance should be investigated to evaluate the conservation potential of the landscape in their general vicinity. In addition, the models and spatial patterns of abundance among species suggest locations where conservation actions may benefit more than one species.     

Multiple landscape scales: An intersite comparison

Vegetation transect data from three locations were analyzed to determine if multiple scales of pattern could be detected. The sites included a semiarid grassland in New Mexico, a series of calcareous openings in a deciduous forest in Tennessee, and a shrub-steppe system in Washington. The data were explored with four statistical techniques. A scale of pattern was accepted if detected by more than one analytical method or located by a single method in multiple taxa. The analyses indicated 3–5 scales of pattern on all three sites, as predicted by Hierarchy Theory.     

Predicting altered connectivity of patchy forests under group selection silviculture

Group selection silviculture creates canopy openings that can alter connectivity in patchy forests, thereby affecting wildlife movement and fire behavior. We examined effects of group selection silviculture on percolation (presence of continuously forested routes across a landscape) in Sierra Nevada East-side pine forest in northern California, USA. Four ~ 250 ha project areas were analyzed at three map resolutions in three ways: analyzing forest cover maps for percolation before and after group-selection treatment, placing simulated group openings in forest cover maps until fragmentation occurred, and comparing project areas to neutral maps that varied in forest cover and self-adjacency. Two project areas were fragmented (i.e., did not percolate) prior to treatment, one resisted fragmentation, and the other became fragmented by treatment when analyzed at 30 m cell resolution. Median simulated openings required to create fragmentation agreed well with the actual number. There was a well-defined transition between percolating and non-percolating neutral maps; increased aggregation of forest lowered the critical value at which forests percolated. A logistic model based on these maps predicted percolation behavior of the project areas effectively, but alternative generating algorithms gave slightly different predictions. A graph of this model provides a straightforward way to visualize how close a landscape is to fragmentation based on its forest cover and aggregation. In East-side Sierran landscape, fragmentation from group-selection openings may make the landscape less hospitable to the American marten but more resistant to crown fire.     

Vegetation productivity consequences of human settlement growth in the eastern United States

In this study, we investigated the impact of human settlement growth on vegetation carbon uptake in the eastern United States between 1992/1993 and 2001. Human settlement growth was measured by changes in the density of housing units. Vegetation carbon uptake was estimated with gross primary production (GPP) based on the light-use efficiency approach applied to satellite imagery. Annual GPP was found to increase by approximately 140?g?C?m^?2 on average for the entire study area in 2001 compared to 1992/1993, accompanied by region-wide increases in downward shortwave radiation and minimum daily temperature. Changes in GPP, however, varied significantly by different types of settlement growth. Exurbanized areas, where the rural settlement (less than 0.025?units per acre) converted to exurbs (0.025?C0.6?units per acre), were associated with approximately 157?g?C?m^?2 increase in GPP due to high vegetation proportions. Suburbanization, the conversion from exurban settlement to suburbs (0.6?C4?units per acre), was related with a decline of GPP by 152?g?C?m^?2 due to progressive development of built-up land cover. Results help to understand the potential of carbon mitigation in the human-dominated landscapes using vegetation as a natural store of carbon dioxide. This in turn has implications for the low-carbon development planning along the gradient of human settlement densities.     

Rural housing is related to plant invasions in forests of southern Wisconsin,USA

Forests throughout the US are invaded by non-native invasive plants. Rural housing may contribute to non-native plant invasions by introducing plants via landscaping, and by creating habitat conditions favorable for invaders. The objective of this paper was to test the hypothesis that rural housing is a significant factor explaining the distribution of invasive non-native plants in temperate forests of the Midwestern US. In the Baraboo Hills, Wisconsin, we sampled 105 plots in forest interiors. We recorded richness and abundance of the most common invasive non-native plants and measured rural housing, human-caused landscape fragmentation (e.g. roads and forest edges), forest structure and topography. We used regression analysis to identify the variables more related to the distribution of non-native invasive plants (best subset and hierarchical partitioning analyses for richness and abundance and logistic regression for presence/absence of individual species). Housing variables had the strongest association with richness of non-native invasive plants along with distance to forest edge and elevation, while the number of houses in a 1 km buffer around each plot was the variable most strongly associated with abundance of non-native invasive plants. Rhamnus cathartica and Lonicera spp. were most strongly associated with rural housing and fragmentation. Berberis thumbergii and Rosa multiflora were associated with gentle slopes and low elevation, while Alliaria petiolata was associated with higher cover of native vegetation and stands with no recent logging history. Housing development inside or adjacent to forests of high conservation value and the use of non-native invasive plants for landscaping should be discouraged.