Strong Influence of Variable Treatment on the Performance of Numerically Defined Ecological Regions

Numerical clustering has frequently been used to define hierarchically organized ecological regionalizations, but there has been little robust evaluation of their performance (i.e., the degree to which regions discriminate areas with similar ecological character). In this study we investigated the effect of the weighting and treatment of input variables on the performance of regionalizations defined by agglomerative clustering across a range of hierarchical levels. For this purpose, we developed three ecological regionalizations of Switzerland of increasing complexity using agglomerative clustering. Environmental data for our analysis were drawn from a 400 m grid and consisted of estimates of 11 environmental variables for each grid cell describing climate, topography and lithology. Regionalization 1 was defined from the environmental variables which were given equal weights. We used the same variables in Regionalization 2 but weighted and transformed them on the basis of a dissimilarity model that was fitted to land cover composition data derived for a random sample of cells from interpretation of aerial photographs. Regionalization 3 was a further two-stage development of Regionalization 2 where specific classifications, also weighted and transformed using dissimilarity models, were applied to 25 small scale “sub-domains” defined by Regionalization 2. Performance was assessed in terms of the discrimination of land cover composition for an independent set of sites using classification strength (CS), which measured the similarity of land cover composition within classes and the dissimilarity between classes. Regionalization 2 performed significantly better than Regionalization 1, but the largest gains in performance, compared to Regionalization 1, occurred at coarse hierarchical levels (i.e., CS did not increase significantly beyond the 25-region level). Regionalization 3 performed better than Regionalization 2 beyond the 25-region level and CS values continued to increase to the 95-region level. The results show that the performance of regionalizations defined by agglomerative clustering are sensitive to variable weighting and transformation. We conclude that large gains in performance can be achieved by training classifications using dissimilarity models. However, these gains are restricted to a narrow range of hierarchical levels because agglomerative clustering is unable to represent the variation in importance of variables at different spatial scales. We suggest that further advances in the numerical definition of hierarchically organized ecological regionalizations will be possible with techniques developed in the field of statistical modeling of the distribution of community composition.     

Comparison of Stream Invertebrate Response Models for Bioassessment Metrics1

Waite, Ian R., Jonathan G. Kennen, Jason T. May, Larry R. Brown, Thomas F. Cuffney, Kimberly A. Jones, and James L. Orlando, 2012. Comparison of Stream Invertebrate Response Models for Bioassessment Metrics. Journal of the American Water Resources Association (JAWRA) 48(3): 570-583. DOI: 10.1111/j.1752-1688.2011.00632.x Abstract: We aggregated invertebrate data from various sources to assemble data for modeling in two ecoregions in Oregon and one in California. Our goal was to compare the performance of models developed using multiple linear regression (MLR) techniques with models developed using three relatively new techniques: classification and regression trees (CART), random forest (RF), and boosted regression trees (BRT). We used tolerance of taxa based on richness (RICHTOL) and ratio of observed to expected taxa (O/E) as response variables and land use/land cover as explanatory variables. Responses were generally linear; therefore, there was little improvement to the MLR models when compared to models using CART and RF. In general, the four modeling techniques (MLR, CART, RF, and BRT) consistently selected the same primary explanatory variables for each region. However, results from the BRT models showed significant improvement over the MLR models for each region; increases in R² from 0.09 to 0.20. The O/E metric that was derived from models specifically calibrated for Oregon consistently had lower R² values than RICHTOL for the two regions tested. Modeled O/E R² values were between 0.06 and 0.10 lower for each of the four modeling methods applied in the Willamette Valley and were between 0.19 and 0.36 points lower for the Blue Mountains. As a result, BRT models may indeed represent a good alternative to MLR for modeling species distribution relative to environmental variables.     

Regionalization of Landscape Pattern Indices Using Multivariate Cluster Analysis

Regionalization, or the grouping of objects in space, is a useful tool for organizing, visualizing, and synthesizing the information contained in multivariate spatial data. Landscape pattern indices can be used to quantify the spatial pattern (composition and configuration) of land cover features. Observable patterns can be linked to underlying processes affecting the generation of landscape patterns (e.g., forest harvesting). The objective of this research is to develop an approach for investigating the spatial distribution of forest pattern across a study area where forest harvesting, other anthropogenic activities, and topography, are all influencing forest pattern. We generate spatial pattern regions (SPR) that describe forest pattern with a regionalization approach. Analysis is performed using a 2006 land cover dataset covering the Prince George and Quesnel Forest Districts, 5.5 million ha of primarily forested land base situated within the interior plateau of British Columbia, Canada. Multivariate cluster analysis (with the CLARA algorithm) is used to group landscape objects containing forest pattern information into SPR. Of the six generated SPR, the second cluster (SPR2) is the most prevalent covering 22% of the study area. On average, landscapes in SPR2 are comprised of 55.5% forest cover, and contain the highest number of patches, and forest/non-forest joins, indicating highly fragmented landscapes. Regionalization of landscape pattern metrics provides a useful approach for examining the spatial distribution of forest pattern. Where forest patterns are associated with positive or negative environmental conditions, SPR can be used to identify similar regions for conservation or management activities.     

Comparing Hydrogeomorphic Approaches to Lake Classification

A classification system is often used to reduce the number of different ecosystem types that governmental agencies are charged with monitoring and managing. We compare the ability of several different hydrogeomorphic (HGM)—based classifications to group lakes for water chemistry/clarity. We ask: (1) Which approach to lake classification is most successful at classifying lakes for similar water chemistry/clarity? (2) Which HGM features are most strongly related to the lake classes? and, (3) Can a single classification successfully classify lakes for all of the water chemistry/clarity variables examined? We use univariate and multivariate classification and regression tree (CART and MvCART) analysis of HGM features to classify alkalinity, water color, Secchi, total nitrogen, total phosphorus, and chlorophyll a from 151 minimally disturbed lakes in Michigan USA. We developed two MvCART models overall and two CART models for each water chemistry/clarity variable, in each case comparing: local HGM characteristics alone and local HGM characteristics combined with regionalizations and landscape position. The combined CART models had the highest strength of evidence (ω_i range 0.92–1.00) and maximized within class homogeneity (ICC range 36–66%) for all water chemistry/clarity variables except water color and chlorophyll a. Because the most successful single classification was on average 20% less successful in classifying other water chemistry/clarity variables, we found that no single classification captures variability for all lake responses tested. Therefore, we suggest that the most successful classification (1) is specific to individual response variables, and (2) incorporates information from multiple spatial scales (regionalization and local HGM variables).     

Reduced complexity models for regional aquatic habitat suitability assessment

Generalizable methods that identify suitable aquatic habitat across large river basins and regions are needed to inform resource management. Habitat suitability models intersect environmental variables to predict species occurrence, but are often data intensive and thus are typically developed at small spatial scales. This study estimated mean monthly aquatic habitat suitability throughout Utah (USA) for Bonneville Cutthroat Trout (Oncorhynchus clarkii utah) and Bluehead Sucker (Catostomus discobolus) with publicly available, geospatial datasets. We evaluated 15 habitat suitability models using unique combinations of percent of mean annual discharge, velocity, gradient, and stream temperature. Environmental variables were validated with observed conditions and species presence observations to verify habitat suitability estimates. Stream temperature, gradient, and discharge best predicted Bonneville Cutthroat Trout presence, and gradient and discharge best predicted Bluehead Sucker presence. Simple aquatic habitat suitability models outperformed models that used only streamflow to estimate habitat for both species, and are useful for conservation planning and water resources decision-making. This modeling approach could enable resource managers to prioritize stream restoration across vast regions within their management domain, and is potentially compatible with water management modeling to improve ecological objectives in management models.     

Identifying Biotic Integrity and Water Chemistry Relations in Nonwadeable Rivers of Wisconsin: Toward the Development of Nutrient Criteria

We sampled 41 sites on 34 nonwadeable rivers that represent the types of rivers in Wisconsin, and the kinds and intensities of nutrient and other anthropogenic stressors upon each river type. Sites covered much of United States Environmental Protection Agency national nutrient ecoregions VII—Mostly Glaciated Dairy Region, and VIII—Nutrient Poor, Largely Glaciated upper Midwest. Fish, macroinvertebrates, and three categories of environmental variables including nutrients, other water chemistry, and watershed features were collected using standard protocols. We summarized fish assemblages by index of biotic integrity (IBI) and its 10 component measures, and macroinvertebrates by 2 organic pollution tolerance and 12 proportional richness measures. All biotic and environmental variables represented a wide range of conditions, with biotic measures ranging from poor to excellent status, despite nutrient concentrations being consistently higher than reference concentrations reported for the regions. Regression tree analyses of nutrients on a suite of biotic measures identified breakpoints in total phosphorus (~0.06 mg/l) and total nitrogen (~0.64 mg/l) concentrations at which biotic assemblages were consistently impaired. Redundancy analyses (RDA) were used to identify the most important variables within each of the three environmental variable categories, which were then used to determine the relative influence of each variable category on the biota. Nutrient measures, suspended chlorophyll a, water clarity, and watershed land cover type (forest or row-crop agriculture) were the most important variables and they explained significant amounts of variation within the macroinvertebrate (R ² = 60.6%) and fish (R ² = 43.6%) assemblages. The environmental variables selected in the macroinvertebrate model were correlated to such an extent that partial RDA analyses could not attribute variation explained to individual environmental categories, assigning 89% of the explained variation to interactions among the categories. In contrast, partial RDA attributed much of the explained variation to the nutrient (25%) and other water chemistry (38%) categories for the fish model. Our analyses suggest that it would be beneficial to develop criteria based upon a suite of biotic and nutrient variables simultaneously to deem waters as not meeting their designated uses.     

Some thoughts on using a landscape framework to address cumulative impacts on wetland food chain support

Problems of using food chain support as a functional attribute of a wetland are discussed. It is suggested that primary production may not be the metric that best evaluates food chain support. Environmental constructs of the wetland and resultant habitat variables appear to yield more information on life-support functions. A landscape-oriented approach is derived to separate hierarchically the wet-lands into ecological regions and landscape elements. This classification scheme allows for predetermination of environmental constraints and the possible natural limits of wetland food chain support. It is proposed that models derived from spatial location theory be used to determine the movement of animals from wetland patches experiencing impacts on food chain support. Patch size, distance between patches, habitat diversity, and environmental constraints are incorporated in these models.     

Downscaling climate change scenarios in an urban land use change model

The objective of this paper is to describe the process through which climate change scenarios were downscaled in an urban land use model and the results of this experimentation. The land use models (Urban Growth Model [UGM] and the Land Cover Deltatron Model [LCDM]) utilized in the project are part of the SLEUTH program which uses a probabilistic cellular automata protocol. The land use change scenario experiments were developed for the 31-county New York Metropolitan Region (NYMR) of the US Mid-Atlantic Region. The Intergovernmental Panel on Climate Change (IPCC), regional greenhouse gas (GHG) emissions scenarios (Special Report on Emissions Scenarios (SRES) A2 and B2 scenarios) were used to define the narrative scenario conditions of future land use change. The specific research objectives of the land use modeling work involving the SLEUTH program were threefold: (1) Define the projected conversion probabilities and the amount of rural-to-urban land use change for the NYMR as derived by the UGM and LCDM for the years 2020 and 2050, as defined by the pattern of growth for the years 1960-1990; (2) Down-scale the IPCC SRES A2 and B2 scenarios as a narrative that could be translated into alternative growth projections; and, (3) Create two alternative future growth scenarios: A2 scenario which will be associated with more rapid land conversion than found in initial projections, and a B2 scenario which will be associated with a slower level of land conversion. The results of the modeling experiments successfully illustrate the spectrum of possible land use/land cover change scenarios for the years 2020 and 2050. The application of these results into the broader scale climate and health impact study is discussed, as is the general role of land use/land cover change models in climate change studies and associated environmental management strategies.     

Modelling land use change with generalized linear models--a multi-model analysis of change between 1860 and 2000 in Gallatin Valley, Montana

This paper develops an approach to modelling land use change that links model selection and multi-model inference with empirical models and GIS. Land use change is frequently studied, and understanding gained, through a process of modelling that is an empirical analysis of documented changes in land cover or land use patterns. The approach here is based on analysis and comparison of multiple models of land use patterns using model selection and multi-model inference. The approach is illustrated with a case study of rural housing as it has developed for part of Gallatin County, Montana, USA. A GIS contains the location of rural housing on a yearly basis from 1860 to 2000. The database also documents a variety of environmental and socio-economic conditions. A general model of settlement development describes the evolution of drivers of land use change and their impacts in the region. This model is used to develop a series of different models reflecting drivers of change at different periods in the history of the study area. These period specific models represent a series of multiple working hypotheses describing (a) the effects of spatial variables as a representation of social, economic and environmental drivers of land use change, and (b) temporal changes in the effects of the spatial variables as the drivers of change evolve over time. Logistic regression is used to calibrate and interpret these models and the models are then compared and evaluated with model selection techniques. Results show that different models are 'best' for the different periods. The different models for different periods demonstrate that models are not invariant over time which presents challenges for validation and testing of empirical models. The research demonstrates (i) model selection as a mechanism for rating among many plausible models that describe land cover or land use patterns, (ii) inference from a set of models rather than from a single model, (iii) that models can be developed based on hypothesised relationships based on consideration of underlying and proximate causes of change, and (iv) that models are not invariant over time.     

Integrating Landscape Ecology and Geoinformatics to Decipher Landscape Dynamics for Regional Planning

We used remote sensing and GIS in conjunction with multivariate statistical methods to: (i) quantify landscape composition (land cover types) and configuration (patch density, diversity, fractal dimension, contagion) for five coastal watersheds of Kalloni gulf, Lesvos Island, Greece, in 1945, 1960, 1971, 1990 and 2002/2003, (ii) evaluate the relative importance of physical (slope, geologic substrate, stream order) and human (road network, population density) variables on landscape composition and configuration, and (iii) characterize processes that led to land cover changes through land cover transitions between these five successive periods in time. Distributions of land cover types did not differ among the five time periods at the five watersheds studied because the largest cumulative changes between 1945 and 2002/2003 did not take place at dominant land cover types. Landscape composition related primarily to the physical attributes of the landscape. Nevertheless, increase in population density and the road network were found to increase heterogeneity of the landscape mosaic (patchiness), complexity of patch shape (fractal dimension), and patch disaggregation (contagion). Increase in road network was also found to increase landscape diversity due to the creation of new patches. The main processes involved in land cover changes were plough-land abandonment and ecological succession. Landscape dynamics during the last 50 years corroborate the ecotouristic-agrotouristic model for regional development to reverse trends in agricultural land abandonment and human population decline and when combined with hypothetical regulatory approaches could predict how this landscape could develop in the future, thus, providing a valuable tool to regional planning.     

Modelling the conversion of Colombian lowland ecosystems since 1940: drivers, patterns and rates

In biologically mega-diverse countries that are undergoing rapid human landscape transformation, it is important to understand and model the patterns of land cover change. This problem is particularly acute in Colombia, where lowland forests are being rapidly cleared for cropping and ranching. We apply a conceptual model with a nested set of a priori predictions to analyse the spatial and temporal patterns of land cover change for six 50-100 km(2) case study areas in lowland ecosystems of Colombia. Our analysis included soil fertility, a cost-distance function, and neighbourhood of forest and secondary vegetation cover as independent variables. Deforestation and forest regrowth are tested using logistic regression analysis and an information criterion approach to rank the models and predictor variables. The results show that: (a) overall the process of deforestation is better predicted by the full model containing all variables, while for regrowth the model containing only the auto-correlated neighbourhood terms is a better predictor; (b) overall consistent patterns emerge, although there are variations across regions and time; and (c) during the transformation process, both the order of importance and significance of the drivers change. Forest cover follows a consistent logistic decline pattern across regions, with introduced pastures being the major replacement land cover type. Forest stabilizes at 2-10% of the original cover, with an average patch size of 15.4 (+/-9.2)ha. We discuss the implications of the observed patterns and rates of land cover change for conservation planning in countries with high rates of deforestation.     

An Ecoregional Context for Forest Management on National Wildlife Refuges of the Upper Midwest,USA

To facilitate forest planning and management on National Wildlife Refuges, we synthesized multiple data sources to describe land ownership patterns, land cover, landscape pattern, and changes in forest composition for four ecoregions and their associated refuges of the Upper Midwest. We related observed patterns to ecological processes important for forest conservation and restoration, with specific attention to refuge patterns of importance for forest landbirds of conservation priority. The large amount of public land within the ecoregions (31–80%) suggests that opportunities exist for coarse and meso-scale approaches to conserving and restoring ecological processes affecting the refuges, particularly historical fire regimes. Forests dominate both ecoregions and refuges, but refuge forest patches are generally larger and more aggregated than in associated ecoregions. Broadleaf taxa have increased in dominance in the ecoregions and displaced fire-dependent taxa such as pine (Pinus spp.) and other coniferous species; these changes in forest composition have likely also affected refuge forests. Despite compositional changes, larger forest patches on refuges suggests that they may provide better habitat for area-sensitive forest landbirds of mature, compositionally diverse forests than surrounding lands if management continues to promote increased patch size. We reason that although fine-scale research and monitoring for species of conservation priority is important, broad scale (ecoregional) assessments provide crucial context for effective forest and wildlife management in protected areas.     

Predicting estuarine sediment metal concentrations and inferred ecological conditions: an information theoretic approach

Empirically derived relationships associating sediment metal concentrations with degraded ecological conditions provide important information to assess estuarine condition. Resources limit the number, magnitude, and frequency of monitoring activities to acquire these data. Models that use available information and simple statistical relationships to predict sediment metal concentrations could provide an important tool for environmental assessment. We developed 45 predictive models for the total concentrations of copper, lead, mercury, and cadmium in estuarine sediments along the Southern New England and Mid-Atlantic regions of the United States. Using information theoretic model-averaging approaches, we found total developed land and percent silt/clay of estuarine sediment were the most important variables for predicting the presence of all four metals. Estuary area, river flow, tidal range, and total agricultural land varied in their importance. The model-averaged predictions explained 78.4, 70.5, 56.4, and 50.3% of the variation for copper, lead, mercury, and cadmium, respectively. Overall prediction accuracies of selected sediment benchmark values (i.e., effects ranges) were 83.9, 84.8, 78.6, and 92.0% for copper, lead, mercury, and cadmium, respectively. Our results further support the generally accepted conclusion that sediment metal concentrations are best described by the physical characteristics of the estuarine sediment and the total amount of urban land in the contributing watershed. We demonstrated that broad-scale predictive models built from existing monitoring data with information theoretic model-averaging approaches provide valuable predictions of estuarine sediment metal concentrations and show promise for future environmental modeling efforts in other regions.     

Hydrologic Modeling Uncertainty Resulting From Land Cover Misclassification1

Abstract: A stochastic, spatially explicit method for assessing the impact of land cover classification error on distributed hydrologic modeling is presented. One‐hundred land cover realizations were created by systematically altering the North American Landscape Characterization land cover data according to the dataset’s misclassification matrix. The matrix indicates the probability of errors of omission in land cover classes and is used to assess the uncertainty in hydrologic runoff simulation resulting from parameter estimation based on land cover. These land cover realizations were used in the GIS‐based Automated Geospatial Watershed Assessment tool in conjunction with topography and soils data to generate input to the physically‐based Kinematic Runoff and Erosion model. Uncertainties in modeled runoff volumes resulting from these land cover realizations were evaluated in the Upper San Pedro River basin for 40 watersheds ranging in size from 10 to 100 km² under two rainfall events of differing magnitudes and intensities. Simulation results show that model sensitivity to classification error varies directly with respect to watershed scale, inversely to rainfall magnitude and are mitigated or magnified by landscape variability depending on landscape composition.     

Eco-Environmental Evolution, Control, and Adjustment for Aibi Lake Catchment

Aibi Lake in north Xinjiang is a typical lake of the arid area, but with a peculiar wetland–arid area ecosystem. Due to the climate becoming drier and the disturbance of human activities, the eco-environment of Aibi Lake catchment has degraded. It was found in our study that there were spatial–temporal changes of vegetation cover, plant species, and soil physical and chemical properties in the catchment. In the upper section of alluvial–fluvial plains, the desertified steppe of Stipa and Artemisia spp. is developed with vegetation cover of some 50%. Haloxylon ammodendron desert occupies the lower section with vegetation cover of some 60%. In these regions with an intensive human disturbance, vegetation has degraded into herb vegetation of annual plant complexes. On the margins of the alluvial–fluvial fans, the lakeshore, and the surrounding regions where the river mouths join the lake, different azonal vegetation—Phragmites communis marsh, Phragmites communis meadow, and Tamarix shrubs—have developed with a vegetation cover of some 80%. On heavier, salinized land, succulent halophyte desert vegetation dominated by Halocnemum strobilaceum has formed with a fractional canopy cover of 10–15%. Haloxylon persicum, Aristida pennata, and other species with a vegetation cover of 30–50% grow in the sand desert zone on the periphery in the lake. In contrast with the 1950s, the vegetation cover around the lakebed and at the river deltas has slightly increased; however, the vegetation cover around the periphery of the lake has decreased and the plant species have still degraded. The surface soils on the windward area and the dried lakebed that have lost vegetation protection have become coarser, whereas the land on the leeward side of the lake has accumulated fine particles. In contrast with the 1980s, soil organic matter has declined markedly. The analyses of climatic data show that the number of days of drifting dust in Jinghe County and Bole City increased in the last 20 years. In the investigation, we found that intensively developed land, the bare lakebed, and abandoned cultivated land provided a great deal of material for drifting dust. In conclusion, we consider the eco-environmental degradation resulting from the inappropriate human activities and put forward recommendations for land-use adjustment and dust control.     

The Ecological Modelling System of NELUP

Within the NERC/ESRC Land Use Programme ecological change is assumed to be driven principally by agronomic activity. Land use change is defined as being of two types depending on whether there is a complete change of cover or a modification of an existing cover. Some agronomic activities lead to quantitative changes between land cover types and others to more subtle qualitative changes within land cover types. A modelling system based on three distinct models is described for predicting the effects of the two types of change in land cover on the distribution of plants, invertebrates and birds. The potential use of the system in analysing problems of land use change is illustrated using a simple scenario based on changing nitrogenous fertilizer regimes in lowland agriculture.     

Representative Landscapes in the Forested Area of Canada

Canada is a large nation with forested ecosystems that occupy over 60% of the national land base, and knowledge of the patterns of Canada’s land cover is important to proper environmental management of this vast resource. To this end, a circa 2000 Landsat-derived land cover map of the forested ecosystems of Canada has created a new window into understanding the composition and configuration of land cover patterns in forested Canada. Strategies for summarizing such large expanses of land cover are increasingly important, as land managers work to study and preserve distinctive areas, as well as to identify representative examples of current land-cover and land-use assemblages. Meanwhile, the development of extremely efficient clustering algorithms has become increasingly important in the world of computer science, in which billions of pieces of information on the internet are continually sifted for meaning for a vast variety of applications. One recently developed clustering algorithm quickly groups large numbers of items of any type in a given data set while simultaneously selecting a representative—or “exemplar”—from each cluster. In this context, the availability of both advanced data processing methods and a nationally available set of landscape metrics presents an opportunity to identify sets of representative landscapes to better understand landscape pattern, variation, and distribution across the forested area of Canada. In this research, we first identify and provide context for a small, interpretable set of exemplar landscapes that objectively represent land cover in each of Canada’s ten forested ecozones. Then, we demonstrate how this approach can be used to identify flagship and satellite long-term study areas inside and outside protected areas in the province of Ontario. These applications aid our understanding of Canada’s forest while augmenting its management toolbox, and may signal a broad range of applications for this versatile approach.     

Agriculture adjustment, land-use transition and protected areas in Northwestern Argentina

Land-use change is the main component of regional environmental change, while protected areas represent a direct land use policy to prevent its potentially negative effects on biodiversity and environmental services. We combined an analysis of trends in land use and human demography with trends in creation of protected areas during the last three decades in northwestern Argentina, a subtropical region including a wide range of environments. The eighty nine administrative analysis units of the region were classified into four ecological groups based on their percentage of cover by the six eco-regions of the study area: (1) "Dry valleys"; dominated by Middle-elevation deserts; (2) "Highlands", dominated by High-elevation alpine zones and plateaus; (3) "Humid ecosystems", dominated by Foggy grasslands and Humid forests, and (4) "Dry forests". Between 1970 and 2002, human population became concentrated in urban areas and land use trends varied greatly among the four ecological groups. Agricultural area decreased in the Highlands and increased in the other regions, particularly in the Dry forests. Domestic animals decreased in Humid ecosystems, Highlands and the Dry valleys; and remained constant in the Dry forests. Several protected areas were created, but most of them were established in regions undergoing a decreasing intensity of land use. Overall, the analysis shows that agricultural production is becoming concentrated in the areas more suitable for modern agriculture while marginal agriculture areas and, particularly, extensive grazing are decreasing. The creation of protected areas reflects the decreasing opportunity costs of marginal areas and is failing to protect the eco-regions most threatened by current land-use trends.     

Regional Representativeness of Swedish Reference Lakes

/ Recent focus has been placed on ecoregion delineations for providing an appropriate framework for monitoring and assessment of region-specific attainable water/habitat quality. Using an ecoregion approach to stratify variance, this study was conducted to determine whether earlier (subjectively) selected Swedish reference lakes may be considered as regionally representative reference sites when compared with a randomly selected lake population. Predictive modeling by discriminant function analysis with lakes classified by ecoregion and lake surface area and six physicochemical variables showed that the greater majority of reference lakes may be considered as regionally representative. The highest proportion of lake "misclassifications" occurred in the boreonemoral ecoregion, a relatively diverse ecoregion of southern Sweden. This apparent bias may be in part be due to the criteria used in selecting regional reference lakes. In the earlier selection of reference lakes emphasis was placed on lakes not being adversely affected by land usage or pollutant discharges, consequently forest lakes were often overrepresented and sites in agricultural areas underrepresented in the selected reference sites. As a complement to predictive modeling, PCA ordination showed the placement of reference lakes within the random lake population and indicated where reference sites might be missing along potentially important ecological gradients. KEY WORDS: Regionalization; Ecoregion; Representativeness; Reference; Ordination; Modeling; Temperate lakes