Deforestation Patterns Research Articles

Myths, memories and metaphors: recollecting landscape change in the Eritrean highlands

Using the case of Eritrea, this article investigates why people's visual recollections and oral accounts of past landscapes can be an unreliable source of accurate information on environmental change. In Eritrea, a longstanding narrative exists that claims that forest cover throughout the country has decreased from 30% of land cover in the late nineteenth century to less than 1% today. Yet popular recollections are contradicted by available archival and photographic evidence, all of which indicates that the landscape has changed far less dramatically than generally imagined. This article explores why a disjunction between the historical evidence regarding the extent and pattern of deforestation in the central Eritrean highlands and people's memories and beliefs about this process should have evolved. Drawing on recent findings in social psychology, and placing this within prevailing debates on landscape and memory, it seeks to explain why visual recollections may be strongly influenced by the metaphoric and symbolic meaning that landscapes hold for individuals and communities. In particular it looks at how memories about the process of deforestation may be shaped less by actual observation of a physical event than by the nature of people's interaction with past authorities, and by individual experiences of economic hardship and scarcity. The article also looks at how the popular myth of a supposedly once prosperous landscape in Eritrea has been employed and encouraged by those in power – either as a justification for various colonial policies or, in the case of the present government, as a powerful impetus to nation building.

How deforestation pattern in the Amazon influences vertebrate richness and community composition

The effects of habitat configuration on species persistence are predicted to be most apparent when remaining habitat cover is below 30%. We tested this prediction by comparing vertebrate communities in 21 landscapes located in the southern Amazonia, includ- ing 7 control landscapes (*100% of forest cover) and 14 fragmented landscapes (4 9 4 km). The fragmented landscapes retained similar proportions of forest (*25%), but had contrasting configurations, resulting from two different deforestation patterns: the ''fish-bone pattern'' common in small properties, and the large- property pattern generally used by large ranchers. Vertebrates were surveyed in all landscapes in Febru- ary-July 2009 with interviews (n = 150). We found a significant difference in reported species richness among the fish-bone, large-property, and control areas (mean = 29.3, 38.8 and 43.5 respectively). Control areas and large-properties tended to have a higher number of specialist species (mean = 13.7, and 11.7, respectively), when compared with the fish-bone pattern (5.1). Verte- brate community composition in the control and large- properties was more similar to one another than to those of the fish-bone landscapes. The number of fragments was the main factor affecting the persistence of species, being negatively associated with specialist species rich- ness. Species richness was also positively related with the size of the largest fragment structurally connected to the studied landscapes (i.e., a regional scale effect). Our results demonstrated that the large-property pattern, which results in less fragmented landscapes, can maintain a more diverse community of large vertebrates, including top predators, which are considered fundamental for maintaining ecosystem integrity. These results support the hypothesis that landscape configuration contributes to the persistence and/or extirpation of species.

Rainfall and Deforestation Dilemma for Cereal Production in the Sudano-Sahel of Cameroon

Time series data reveals that the Sahel of Cameroon has experienced several years of deficits in cereal production. The debate attributes the observed trends to low rainfall. Uncertainties in the debate on the role of rainfall as a principal causal factor are evident and need verification. Both field and desk studies have been used; this involved the administration of 200 questionnaires and focused group discussions. The desk studies included detailed literature review and analysis of Normalized Difference Vegetation Index (NDVI) satellite images of vegetation and rainfall from the Global Precipitation Climatology Project (GPCP). The results show that cereal production has declined while rainfall has increased by 30-40% in the last decade in the study area. With an increase in rainfall, the observed decline in cereal production cannot be explained by climate only. Land use changes such as deforestation patterns are significant in explaining the trends in cereal output as seen in population perceptions.

Typology and Characterization of Amazon Colonists: A Case Study Along the Transamazon Highway

This study develops criteria for classifying the actor groups shaping frontier development along the Transamazon Highway colonization project in the Brazilian Amazon, as a basis to improve understanding of their specific contributions to environmental degradation and socio-economic development. Based on an analysis of responses to questionnaires by 93 colonists representing different migration trajectories, production systems, socio-economic strategies and deforestation patterns, actor groups could be statistically classified according to their type of production and level of capitalization. A property size threshold discriminating small and largeholders in the study area is presented and compared with previous attempts to establish such a threshold. Largeholders practicing large-scale cattle ranching and smallholders practicing diversified family agriculture were found to be the two predominant colonist types. Smallholder farming practices were found to be more appropriate to the local environmental conditions than those implemented by cattle ranchers.

Assessing landscape changes and dynamics using patch analysis and GIS modeling

Landscape ecology has brought spatial analysis and geospatial modeling to the forefront of ecological research. In this study, we investigated changes in landscape structure and the ecological consequences of urbanization in the Twin Cities Metropolitan Area using spatial indices, change analysis, and geospatial modeling. Major land transformation patterns from 1975 to 2006 were mapped using multi-temporal classification maps. Landscape metrics for 1975, 1986, 1998, and 2006 were calculated and analyzed using Patch Analyst. The deforestation process due to urbanization in the future was modeled using GEOMOD within the framework of 2030 regional land use planning. Results reveal urban area expanded 82% from 1975 to 2006, mainly by displacing forest and agricultural lands by in-filling gaps within the urban area or near its periphery. Leap-frog development was noticeably absent. Many small patches of forest and agriculture located within or near the urban and suburban areas were replaced directly by urban land uses. Urbanization has become ecologically complicated but geometrically simpler whereas the rural landscape became less abundant and diverse. We predict: (1) forest cover will continue to decrease consistently due to on-going urbanization process; and (2) the pattern of deforestation will occur within the urban area due to in-filling and along its periphery as the urban area expands outward into undeveloped areas.

Global Climate Indices - A Problem or a Solution to the Geospatial Analysis of Infectious Diseases

Many climatic and geological variables can modulate the association of infectious disease exposure and disease temporally. Temperature, precipitation, wind velocity and directionality, air humidity, barometric pressure, and seismic activity can be included to name a few. However, as we think about these covariates and how they interact to facilitate the development of geo-climatically influenced infectious diseases, several other parameters also come to mind such as diurnal cyclical fluctuations, deforestation, and global migratory patterns. Indeed, very quickly, our efforts to minimize residual confounding become laborious, with every newly discovered interaction opening up “a new can of worms.”

Modeling spatial pattern of deforestation using GIS and logistic regression: A case study of northern Ilam forests, Ilam province, Iran

This study aimed to predict spatial distribution of deforestation and detects factors influencing forest degradation of Northern forests of Ilam province. For this purpose, effects of six factors including distance from road and settlement areas, forest fragmentation index, elevation, slope and distance from the forest edge on the forest deforestation were studied. In order to evaluate the changes in forest, images related to TM1988, ETM + 2001 and ETM + 2007 were processed and classified. There are two classes as, forest and non-forest in order to assess deforestation factors. The logistic regression method is used for modeling and estimating the spatial distribution of deforestation. The results show that about 19,294 ha from forest areas are deforested in the 19 years. Modeling results also indicate that more deforestation occurred in the fragmented forest cover and in the areas of proximity to forest/non forest edge. Furthermore, slope and distance from road and settlement areas had negative relationships with deforestation rates. Meanwhile, deforestation rate is decreased with increase in elevation. Finally, a simple spatial model is presented that is able to predict the location of deforestation by using logistic regression. The validation was also tested using ROC approach which was found to be 0.96.

Modeling spatial pattern of deforestation using GIS and logistic regression: A case study of northern Ilam forests, Ilam province, Iran

This study aimed to predict spatial distribution of deforestation and detects factors influencing forest degradation of Northern forests of Ilam province. For this purpose, effects of six factors including distance from road and settlement areas, forest fragmentation index, elevation, slope and distance from the forest edge on the forest deforestation were studied. In order to evaluate the changes in forest, images related to TM1988, ETM + 2001 and ETM + 2007 were processed and classified. There are two classes as, forest and non-forest in order to assess deforestation factors. The logistic regression method is used for modeling and estimating the spatial distribution of deforestation. The results show that about 19,294 ha from forest areas are deforested in the 19 years. Modeling results also indicate that more deforestation occurred in the fragmented forest cover and in the areas of proximity to forest/non forest edge. Furthermore, slope and distance from road and settlement areas had negative relationships with deforestation rates. Meanwhile, deforestation rate is decreased with increase in elevation. Finally, a simple spatial model is presented that is able to predict the location of deforestation by using logistic regression. The validation was also tested using ROC approach which was found to be 0.96. Key words : Deforestation modeling, remote sensing, logistic regression, Zagros forests, Ilam province, Ilam.

Behaviour and space in agent-based modelling: Poverty patterns in East Kalimantan, Indonesia

Applied models of policy interventions are increasingly expected to consider households’ responses to these interventions, which makes agent-based modelling popular in applied policy situations. Implementing an adequate level of agent heterogeneity and mapping it into a spatial environment are critical factors of such applied modelling. However, policy applications demand the characterisation and parameterisation of behavioural response functions of heterogeneous agents and the spatial distribution of heterogeneous agents, which are neither highly transparent nor greatly tested steps in implementing agent-based models. This paper describes an agent-based model of fuel price changes for a case study in East Kalimantan, Indonesia, and specifically: (a) the characterisation and parameterisation approach, (b) resulting agent types for approximating behavioural heterogeneity, and (c) emerging spatial poverty and deforestation patterns. The model highlights the spatial dynamics of poverty dynamics, indicating that the direct impact of deforestation on poverty among forest-dwelling communities is to trigger their migration into peri-urban areas. Overall, the model suggests that poverty increases in response to fuel price reductions.

Gold Mining in the Peruvian Amazon: Global Prices, Deforestation, and Mercury Imports

Many factors such as poverty, ineffective institutions and environmental regulations may prevent developing countries from managing how natural resources are extracted to meet a strong market demand. Extraction for some resources has reached such proportions that evidence is measurable from space. We present recent evidence of the global demand for a single commodity and the ecosystem destruction resulting from commodity extraction, recorded by satellites for one of the most biodiverse areas of the world. We find that since 2003, recent mining deforestation in Madre de Dios, Peru is increasing nonlinearly alongside a constant annual rate of increase in international gold price (∼18%/yr). We detect that the new pattern of mining deforestation (1915 ha/year, 2006–2009) is outpacing that of nearby settlement deforestation. We show that gold price is linked with exponential increases in Peruvian national mercury imports over time (R2 = 0.93, p = 0.04, 2003–2009). Given the past rates of increase we predict that mercury imports may more than double for 2011 (∼500 t/year). Virtually all of Peru's mercury imports are used in artisanal gold mining. Much of the mining increase is unregulated/artisanal in nature, lacking environmental impact analysis or miner education. As a result, large quantities of mercury are being released into the atmosphere, sediments and waterways. Other developing countries endowed with gold deposits are likely experiencing similar environmental destruction in response to recent record high gold prices. The increasing availability of satellite imagery ought to evoke further studies linking economic variables with land use and cover changes on the ground.

An Assessment of Potential Benefits to Smallholders of REDD+ Components in the Philippines

Many sectors in the Philippines are looking at the potential of Reducing Emissions from Deforestation and Forest Degradation-plus (REDD+) under the UN Framework Convention on Climate Change (UNFCCC) to help finance forest protection and rehabilitation in the country. However, one major problem is that there is little information on the potential benefits the country can expect under REDD+. Specifically, it is not known how each component activities of REDD+ can benefit smallholder farmers. Thus, this paper assesses the potential benefits of activities under REDD+ to smallholder farmers in the country. The key question is what the potential of REDD+ in the Philippines is for improving the sequestration potential of the forest sector and to serve as a form of supplemental livelihood for rural forest dwellers? The main approach of the paper is to summarize what is known about: the historical pattern of deforestation and degradation, the driving forces behind them, community-based forest management (CBFM), tenure and rights, and to analyze the implications of Copenhagen and Cancun meetings for the Philippines. The main finding of the study is that depending on which REDD+ activity is implemented, smallholder farmers under CBFM areas would have varying roles and potential benefits. Smallholder farmers will benefit the most from avoiding forest degradation and enhancing of forest stocks activities because these activities pose the highest potential carbon credits. Due to the rising total forest cover of the country, very few carbon credits are expected from avoiding deforestation. This implies that government policies and programs could focus on preparing local communities and institutions for activities that decrease forest degradation and enhance carbon stocks. In addition, there are many uncertainties and information gaps remaining. For example, the rate of biomass degradation in Philippines forests and the drivers of forest degradation are still unknown. The ability of government agencies to implement REDD+ is still inadequate. A strong capacity building program is therefore essential.

Patterns and Trends of Forest Loss in the Colombian Guyana

ABSTRACTSpatial patterns of tropical deforestation and fragmentation are conditional upon human settlement characteristics. We analyze four different human occupation models (indigenous, colonist frontier, transition and established settlement) in the Colombian Guyana Shield at three different times: 1985, 1992 and 2002, and compared them for: (1) deforestation rates; (2) the amount of forest as classified according to a fragmentation pattern (interior forest, edge forest, perforated forest and forest patch); (3) various fragmentation metrics using repeated measures analysis of variance; and (4) potential future deforestation trends though the implementation of a spatially explicit simulation model. The indigenous and colonist frontier occupation models had low rates of deforestation (0.04%/yr), while the well‐established settlement occupation model had the highest rate (3.68%/yr). Our results indicate that the four occupation models generate three deforestation patterns: diffuse, which can be subdivided into two subpatterns (indigenous and colonist), geometric (transition) and patchy (established settlement). The area with the established settlement model was highly fragmented, while in the transition occupation area, forest loss was gradual and linked to economic activities associated with the expansion of the agricultural frontier. The simulation of future trends revealed that indigenous and colonist areas had a constant, albeit small, loss of forest covers. The other models had a deforestation probability of 0.8 or more. Overall, our results highlight the need for new and urgent policies for reducing forest conversion that consider intraregional variability in human occupation linked to differences in land‐use patterns.Abstract in Spanish is available in the online version of this article.

A GIS Cost-Benefit Analysis-Based Methodology to Establish a Payment for Environmental Services System in Watersheds: Application to the Calan River in Honduras

As a potential mechanism for halting patterns of agriculture-driven deforestation in Central America, Payment for Environmental Services (PES) programs have generated widespread interest. These programs involve transferring payments for a positive externality (in this case, the provision of clean drinking water) from those who use it to those whose activities provide it. In this article, we report on the research underpinning the development of a PES mechanism for the medium-sized city of Siguatepeque, Honduras. We used a combination of cost-benefit analysis, hydrological, and GIS-based methods to identify key areas where a change in land use or other sanitary practices would significantly improve water quality; to estimate the costs of a conversion in land use patterns; to estimate local residents' willingness to pay for water; and to analyze the legal structures under which such a project might take form in Honduras. Our results suggested that local residents were willing to pay for water provision both using money and community labor. These payments are sufficient to cover part of the costs of implementing the system. Several scenarios with different levels of externalities corresponding to different provision levels of land use change are proposed in order to study the most viable ones. On a methodological level, these results attest to the value of including GIS-based landscape planning and cost-benefit analysis in a methodology for designing and implementing PES systems.

Spatiotemporal modeling of the expansion of mechanized agriculture in the Bolivian lowland forests

Since the 1980s, mechanized soybean production in Bolivia has caused extensive deforestation in the northeast of Santa Cruz de la Sierra in the eastern Bolivian lowlands. We analyze the spatial and temporal dynamics of deforestation due to mechanized agriculture with spatially explicit logistic regression models. Deforestation patterns are derived from the classification of Landsat imagery by Killeen et al. (2007) and include five time steps (1976, 1986, 1992, 2001 and 2005). We associate deforestation with geophysical and socioeconomic determinants. Our model controls for spatial autocorrelation and temporal dependencies, and we assess the robustness of the results for several model formulations.The expansion of mechanized agriculture is concentrated in areas with favorable environmental conditions, good market access and close proximity to prior deforestation. While overall dynamics remained relatively stable over time, the expansion of mechanized agriculture between 2001 and 2005 became more tolerant to excessive rainfall and less dependent on fertile soils. This mirrors the increasing penetration of mechanized agriculture into humid and less fertile Amazonian rainforests in the northern portion of the study area. The map of deforestation probability substantiates these patterns and shows the highest propensities for future deforestation in the north. Our study demonstrates the value of spatial regression models to better understand the development of deforestation dynamics over a 30-year time span, and contributes to the formulation of policies that aim to reduce deforestation. Yet the results are sensitive to hidden correlations between independent variables, and we therefore advocate a careful evaluation of regression results for different model formulations.

Habitat fragmentation and the desiccation of forest canopies: A case study from eastern Amazonia

Across the deforestation frontier in eastern Amazonia, we examined the relationship between edge-related forest desiccation and deforestation patterns using remote-sensing techniques. Canopy-moisture levels were estimated over an eight-year period in three study sites that encompassed ∼100,000 km 2 in area. We found four main effects of deforestation on dry-season canopy desiccation. First, intact forests showed no detectable change in canopy water content whereas forests adjacent to clearings showed significant water loss. Second, the distance to which edge-related desiccation penetrated into forest interiors varied among landscapes with differing forest loss and fragmentation. In moderately fragmented landscapes (with 65% and 51% remaining forest cover), canopy desiccation extended 1–1.5 km into forest interiors, whereas in heavily fragmented landscapes (20% forest cover) desiccation penetrated up to 2.7 km into forests. Third, the magnitude of edge-related desiccation varied among landscapes with differing fragmentation. Moderately fragmented landscapes exhibited a greater magnitude of change in canopy-water loss over the first 1 km from an edge than did heavily fragmented landscapes. Finally, forest desiccation penetrated further into forests over time in the moderately fragmented landscapes, but not in the heavily fragmented landscape, where edge-related desiccation had evidently ‘saturated’ remaining forests. We conclude that protracted dry seasons will have far more serious effects on fragmented than intact rainforests, with the former becoming highly vulnerable to destructive fires. With ∼30,000 km of new forest edge being created annually in Brazilian Amazonia, these finding have serious implications for forest conservation.

Rural property size drives patterns of upland and riparian forest retention in a tropical deforestation frontier

Deforestation in Brazilian Amazonia accounts for a disproportionate global scale fraction of both carbon emissions from biomass burning and biodiversity erosion through habitat loss. Here we use field- and remote-sensing data to examine the effects of private landholding size on the amount and type of forest cover retained within economically active rural properties in an aging southern Amazonian deforestation frontier. Data on both upland and riparian forest cover from a survey of 300 rural properties indicated that 49.4% (SD = 29.0%) of the total forest cover was maintained as of 2007, and that property size is a key regional-scale determinant of patterns of deforestation and land-use change. Small properties (≤150 ha) retained a lower proportion of forest (20.7%, SD = 17.6) than did large properties (>150 ha; 55.6%, SD = 27.2). Generalized linear models showed that property size had a positive effect on remaining areas of both upland and total forest cover. Using a Landsat time-series, the age of first clear-cutting that could be mapped within the boundaries of each property had a negative effect on the proportion of upland, riparian, and total forest cover retained. Based on these data, we show contrasts in land-use strategies between smallholders and largeholders, as well as differences in compliance with legal requirements in relation to minimum forest cover set-asides within private landholdings. This suggests that property size structure must be explicitly considered in landscape-scale conservation planning initiatives guiding agro-pastoral frontier expansion into remaining areas of tropical forest.

Understanding deforestation in the southern Yucatán: insights from a sub-regional, multi-temporal analysis

The southern Yucatan has been identified as a deforestation hot spot. Land-change studies document the extent of forest loss at a regional scale, and case studies provide insights into the drivers of deforestation at the household level. Those studies have paid minimal attention to sub-regional analysis, especially to discrete land-management units above the household level. This analysis of upland forest change addresses the range of variation in deforestation among 96 ejidos (communal lands) and the Calakmul Biosphere Reserve, the two dominant land-tenure and land-management units in the region. Satellite imagery, census, and land-tenure data are used to establish the extent and location of deforestation patterns, and multivariate techniques are employed to identify biophysical and socioeconomic variables that explain such patterns. Results show that, for the 1984–1993 period, deforestation in the southern Yucatan was not as prevalent as implied by its hot spot designation. Three clusters of deforestation are identified. A logistic regression analysis establishes that size of forest holdings, population growth, and location in the precipitation gradient correlate with ejidos that experienced higher deforestation rates than the rest of the land-tenure units. For the 1993–2000 period, conservation programs and changes in the economic context of this “hollow frontier” contributed to reduce deforestation rates and extent. This analysis illustrates the spatio-temporal heterogeneity of much tropical forest change and caution that it should bring to simple formulations of modeling this change and prescribing policies to control it.

Assessment of deforestation, biodiversity loss and the associated factors: case study of Ijesa-Ekiti region of Southwestern Nigeria

Deforestation is driven by a variety of factors, and has resulted in land use changes that threaten biodiversity, water and energy resources. However, lack of reliable data and survey information in Nigeria has made the estimation of the effect of deforestation difficult to establish. Consequently, the extent and rate of deforestation are less well known. The study therefore, examined and analyzed the spatial and temporal patterns of deforestation over the period of 25 years (1978–2003); measured the rates, trends and explained the factors that determined deforestation in Ijesa-Ekiti region of southwestern Nigeria. The major sources of data for the study were satellites images. These were Landsat MSS 1978, with spatial resolution of 80 m, SPOT XS 1986, SPOT XS 1994, with 20 m spatial resolution and NigeriaSat_1 2003, with 32 m spatial resolution. To make them comparable, they were georeferenced to the same coordinates system, filtered, resampled and enhanced for visualization in a GIS environment. Furthermore, Ilesa, Ijebu-Ijesa, Efon-Alaaye, Iloko-Ijesa, Erin-Oke and Erin-Ijesa were identified and selected for ground truthing to validate the tonal values recorded in the images with the features on the ground. The result of ground truthing was combined with visual image interpretation as training sites for supervised classification. Focus Group Discussions were held with people who had lived in the area for over 20 years as a means of eliciting factors of deforestation and the effects on forest biodiversity. The results indicated forest loss of 53,469.23 ha over the period of 25 years at an annual deforestation rate of 7.21, 2.47, and 5.40% per year for 1978–1986, 1986–1994 and 1994–2003, respectively. FGDs with various categories of people in the bigger towns confirmed deforestation in the area and were due to illegal lumbering, intensive agricultural practices and growth of settlements resulted from increase in human population. FGDs also revealed extinction of many forest species in their communities. In conclusion, the study advanced our understanding on techniques of analyzing deforestation using geo-spatial technology. It also generated a synthesis of information on the rates of deforestation and its driving forces, which are a complex mix of anthropogenic factors, the chief of which has been the conversion of forest resources to agricultural land use.

Integrating species life‐history traits and patterns of deforestation in amphibian conservation planning

AbstractAim To identify priority areas for amphibian conservation in southeastern Brazil, by integrating species life‐history traits and patterns of deforestation.Location State of São Paulo, Brazil.Methods We used the software Marxan to evaluate different scenarios of amphibian conservation planning. Our approach differs from previous methods by explicitly including two different landscape metrics; habitat split for species with aquatic larvae, and habitat loss for species with terrestrial development. We evaluated the effect of habitat requirements by classifying species breeding habitats in five categories (flowing water, still water permanent, still water temporary, bromeliad or bamboo, and terrestrial). We performed analyses using two scales, grid cells and watersheds and also considered nature preserves as protected areas.Results We found contrasting patterns of deforestation between coastal and inland regions. Seventy‐six grid cells and 14 watersheds are capable of representing each species at least once. When accounting for grid cells already protected in state and national parks and considering species habitat requirements we found 16 high‐priority grid cells for species with one or two reproductive habitats, and only one cell representing species with four habitat requirements. Key areas for the conservation of species breeding in flowing and permanent still waters are concentrated in southern state, while those for amphibians breeding in temporary ponds are concentrated in central to eastern zones. Eastern highland zones are key areas for preserving species breeding terrestrially by direct or indirect development. Species breeding in bromeliads and bamboos are already well represented in protected areas.Main conclusions Our results emphasize the need to integrate information on landscape configuration and species life‐history traits to produce more ecologically relevant conservation strategies.

Using Ecological Land Units for Conservation Planning in a Southwestern Ohio Watershed

A Geographic Information System was used to assess the representation of ecological land units (ELUs) both within and external to the current preserve network of the Lower Twin Creek Watershed (LTCW), southwestern Ohio. ELUs were classified using multivariate and cluster analyses on forest canopy tree species and seven physiographic and soil variables derived from digital elevation models and a soil series map. Cluster analysis of the five most significant variables (landform, soil drainage, hillshade, curvature, and percent slope) influencing vegetation distribution resulted in nine discrete ELUs. They included uplands dominated by Fagus grandifolia-Acer saccharum, dry slopes dominated by Quercus spp.-Carya ovata, mesic slopes dominated by a mixed mesophytic community, and wet floodplains dominated by Platanus occidentalis-Populus deltoldes. The area of forest in each ELU in the preserve network was compared to the area of forest in each ELU in the watershed as a whole and the potential natural area of forest cover determined by topographic position. Well-drained upland and poorly-drained upland ELUs were most underrepresented in the preserve network followed by wet broad floodplains. Results of the gap analysis based on current vegetation distribution were influenced by the pattern of deforestation in the LTCW, which was, in turn, related to topographic position: slopes were more likely to be forested and incorporated into a preserve system than more level lands suitable for agriculture. Protecting and restoring forest in upland ELUs will be necessary to capture a full range of ecosystem diversity in the LTCW.