An investigation of uncertainty in field habitat mapping and the implications for detecting land cover change

Land cover data for landscape ecological studies are frequently obtained by field survey. In the United Kingdom, temporally separated field surveys have been used to identify the locations and magnitudes of recent changes in land cover. However, such map data contain errors which may seriously hinder the identification of land cover change and the extent and locations of rare landscape features. This paper investigates the extent of the differences between two sets of maps derived from field surveys within the Northumberland National Park in 1991 and 1992. The method used in each survey was the Phase 1 approach of the Nature Conservancy Council of Great Britain. Differences between maps were greatest for the land cover types with the smallest areas. Overall spatial correspondence between maps was found to be only 44.4%. A maximum of 14.4% of the total area surveyed was found to have undergone genuine land cover change. The remaining discrepancies, equivalent to 41.2% of the total survey area, were attributed primarily to differences of land cover interpretation between surveyors (classification error). Differences in boundary locations (positional error) were also noted, but were found to be a relatively minor source of error. The implications for the detection of land cover change and habitat mapping are discussed.     

Effects of changing spatial scale on the analysis of landscape pattern

The purpose of this study was to observe the effects of changing the grain (the first level of spatial resolution possible with a given data set) and extent (the total area of the study) of landscape data on observed spatial patterns and to identify some general rules for comparing measures obtained at different scales. Simple random maps, maps with contagion (i.e., clusters of the same land cover type), and actual landscape data from USGS land use (LUDA) data maps were used in the analyses. Landscape patterns were compared using indices measuring diversity (H), dominance (D) and contagion (C). Rare land cover types were lost as grain became coarser. This loss could be predicted analytically for random maps with two land cover types, and it was observed in actual landscapes as grain was increased experimentally. However, the rate of loss was influenced by the spatial pattern. Land cover types that were clumped disappeared slowly or were retained with increasing grain, whereas cover types that were dispersed were lost rapidly. The diversity index decreased linearly with increasing grain size, but dominance and contagion did not show a linear relationship. The indices D and C increased with increasing extent, but H exhibited a variable response. The indices were sensitive to the number (m) of cover types observed in the data set and the fraction of the landscape occupied by each cover type (P _k); both m and P _kvaried with grain and extent. Qualitative and quantitative changes in measurements across spatial scales will differ depending on how scale is defined. Characterizing the relationships between ecological measurements and the grain or extent of the data may make it possible to predict or correct for the loss of information with changes in spatial scale.     

The use of remotely-sensed satellite imagery for landscape classification in Wales (U.K.)

Remotely-sensed satellite data from Landsat TM and MSS were processed digitally to make landscape classifications of three study areas of south east Wales. The classifications were constructed by classifying major variations in land cover mosaics within the areas, and using these data to group the 1 km × 1 km cells of the National Grid into landscape classes according to the combination of cover types found within them. The TWINSPAN algorithm, which is a polythetic, divisive classification method, was used as the basis of the study.The results showed that while satellite imagery could only be used to extract information about land cover, the close association betwen landscape, land cover and terrain meant that the major physical divisions in the study area could also be detected in the landscape classification. The landscape types recognised in the study were found to be consistent with those indicated in other independent data which relate to the areas. These data included the ITE Land Classes for Great Britain, and the Agricultural (June) Census statistics for England and Wales.The approach to landscape classification described allows landscape classifications to be made rapidly. These classifications can provide a sampling frameworks for landscape survey in areas where basic map data are lacking or resources for field survey are limited. The landscape classifications can also assist in making landscape evaluations since they allow different landscape types to be compared in respect of such properties such as their typicalness, rarity, naturalness and position on a geographical or ecological gradient.     

Temporal and spatial changes in an area of the New Jersey Pine Barrens landscape

In order to document the extent of landscape fragmentation for a section of the New Jersey Pine Barrens region, we have used satellite image and spatial analysis to monitor landscape change between 1972 and 1988. Land-cover patterns were quantified by mean, number, and size of patches; and amount of edges between land cover types. During the intervening sixteen year period, fractal dimension, diversity, and contagion generally decreased while dominance, disturbance and edges increased, indicating a trend to a more dissected and disturbed landscape. There was an increase in the number of forest patches and a significant decrease in the average size of forest patches. In contrast, the mean patch size for the non-forest category has increased as a result of a coalescence of patches. The landscape fragmentation is shown by a downward shift in the distribution of forest patches by size class. These changes in landscape pattern have implications for many ecological processes and resources. Management practices need to consider landscape fragmentation in the Pinelands National Reserve in order to preserve the essential character of the Pine Barrens landscape.     

Stochastic generating models for simulating hierarchically structured multi-cover landscapes

For simulating hierarchically structured raster maps of landscapes that consist of multiple land cover types, we extend the concept of neutral landscape models to provide a general Markovian model. A stochastic transition matrix provides the probability rules that govern landscape fragmentation processes by assigning finer resolution land cover categories, given coarser resolution categories. This matrix can either be changed or remain the same at different resolutions. The probability rules may be defined for simulating properties of an actual landscape or they may be specified in a truly neutral manner to evaluate the effects of particular transition probability rules.For illustration, model parameters are defined heuristically to simulate properites of actual watershed-delineated landscapes in Pennsylvania. Three landscapes were chosen; one is mostly forested, one is in a transitional state between mostly forested and a mixture of agriculture, urban and suburban land, while the third is fully developed with only remnant forest patches that are small and disconnected. For each landscape type, a small sample of raster maps are simulated in a Monte Carlo fashion to illustrate how an empirical distribution of landscape measurements can be obtained.     

A new approach for rescaling land cover data

The resolution of satellite imagery must often be increased or decreased to fill data gaps or match preexisting project requirements. It is well known that a change in resolution introduces systematic errors of size, shape, location and amount of contiguous land cover types. Nevertheless, robust methods for rescaling landscape data are frequently required to assess patterns of landscape change through time and over large areas. We developed a new method for rescaling spatial data that allows map resolution (grain size) to be either increased or decreased while holding the total proportion of land cover types constant. The method uses a weighted sampling net of variable resolution to sample an existing map and then randomly selects from the frequency of cover types derived from this sample to assign the cover type for the corresponding location in the rescaled map. The properties of the sampling net had a variable effect on measures of landscape pattern with the characteristic patch size (S) the most robust metric and the number of clusters (A) the most variable. A comparison of up-scaled and down-scaled maps showed that this process is not symmetrical, producing different errors for increases versus decreases in grain size. Rescaling Landsat (30 m) imagery to the 10 m resolution of SPOT imagery for four National Park units within Maryland and Virginia resulted in errors due to rescaling that were small (1–2%) relative to the total error (∼11%) associated with these images. The new rescaling method is general because it provides a single method for increasing or decreasing resolution, can be applied to maps with multiple land cover types, allows grid geometry to be transformed (i.e., square to hexagonal grids), and provide a more consistent basis for landscape comparisons when maps must be derived from multiple sources of classified imagery.     

Modeling habitat dynamics accounting for possible misclassification

Land cover data are widely used in ecology as land cover change is a major component of changes affecting ecological systems. Landscape change estimates are characterized by classification errors. Researchers have used error matrices to adjust estimates of areal extent, but estimation of land cover change is more difficult and more challenging, with error in classification being confused with change. We modeled land cover dynamics for a discrete set of habitat states. The approach accounts for state uncertainty to produce unbiased estimates of habitat transition probabilities using ground information to inform error rates. We consider the case when true and observed habitat states are available for the same geographic unit (pixel) and when true and observed states are obtained at one level of resolution, but transition probabilities estimated at a different level of resolution (aggregations of pixels). Simulation results showed a strong bias when estimating transition probabilities if misclassification was not accounted for. Scaling-up does not necessarily decrease the bias and can even increase it. Analyses of land cover data in the Southeast region of the USA showed that land change patterns appeared distorted if misclassification was not accounted for: rate of habitat turnover was artificially increased and habitat composition appeared more homogeneous. Not properly accounting for land cover misclassification can produce misleading inferences about habitat state and dynamics and also misleading predictions about species distributions based on habitat. Our models that explicitly account for state uncertainty should be useful in obtaining more accurate inferences about change from data that include errors.     

Uncertainty analysis as a tool for refining land dynamics modelling on changing landscapes: a case study in a Spanish Natural Park

In this study we developed a methodology aimed at improving the assessment of inter-annual land cover dynamics from hard classified remotely sensed data in heterogeneous and resilient landscapes. The methodology is implemented for the Spanish Natural Park of Sierra de Ancares, where human interference during the last century has resulted in the destruction and fragmentation of the original land cover. We ran supervised classifications, with a maximum likelihood algorithm (Maxlike), on a temporal series of Landsat images (1991–2005), followed by an uncertainty assessment using fuzzy classifications and confusion indices (CIs). This allowed us to show how much (and where) of the resulting maps contained a substantial amount of error, distinguishing data that might be useful to measure land change from data that are not particularly useful when applying a post-classification comparison methodology. In this way, we can detect true changes not skewed by the effects of uncertainty. Even if patterns of change were always coherent amongst years, they were more realistic after reducing uncertainty, in spite of a substantial decrease in the number of available pixels (i.e. unmasked by the method). We then computed land cover dynamics by means of a model specifically designed to determine the frequency of disturbances (mainly fire events) and the vegetation recovery time during the study period. Model outputs showed correlated landscape patterns at a broad scale and provided useful results to explore land cover change from pattern to process.     

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

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

Predicting land cover change and avian community responses in rapidly urbanizing environments

We used an integrated modeling approach to simulate future land cover and predict the effects of future urban development and land cover on avian diversity in the Central Puget Sound region of Washington State, USA. We parameterized and applied a land cover change model (LCCM) that used output from a microsimulation model of urban development, UrbanSim, and biophysical site and landscape characteristics to simulate land cover 28 years into the future. We used 1991, 1995, and 1999 Landsat TM-derived land cover data and three different spatial partitions of our study area to develop six different estimations of the LCCM. We validated model simulations with 2002 land cover. We combined UrbanSim land use outputs and LCCM simulations to predict changes in avian species richness. Results indicate that landscape composition and configuration were important in explaining land cover change as well as avian species response to landscape change. Over the next 28 years, urban land cover was predicted to increase at the expense of agriculture and deciduous and mixed lowland forests. Land cover changes were predicted to reduce the total number of avian species, with losses primarily in native forest specialists and gains in common synanthropic species such as the American Crow (Corvus brachyrhynchos). The integrated modeling framework we present has potential applications in urban and natural resource planning and management and in assessing of the effects of policies on land development, land cover, and avian biodiversity.     

Effects of changing scale on landscape pattern analysis: scaling relations

Landscape pattern is spatially correlated and scale-dependent. Thus, understanding landscape structure and functioning requires multiscale information, and scaling functions are the most precise and concise way of quantifying multiscale characteristics explicitly. The major objective of this study was to explore if there are any scaling relations for landscape pattern when it is measured over a range of scales (grain size and extent). The results showed that the responses of landscape metrics to changing scale fell into two categories when computed at the class level (i.e., for individual land cover types): simple scaling functions and unpredictable behavior. Similarly, three categories were found at the landscape level, with the third being staircase pattern, in a previous study when all land cover types were combined together. In general, scaling relations were more variable at the class level than at the landscape level, and more consistent and predictable with changing grain size than with changing extent at both levels. Considering that the landscapes under study were quite diverse in terms of both composition and configuration, these results seem robust. This study highlights the need for multiscale analysis in order to adequately characterize and monitor landscape heterogeneity, and provides insights into the scaling of landscape patterns. This revised version was published online in May 2005 with corrections to the Cover Date. This revised version was published online in July 2006 with corrections to the Cover Date.     

The accuracy of land cover-based wetland assessments is influenced by landscape extent

Widespread degradation of wetlands has motivated the development of tools to evaluate wetland condition. The application of field-based tools over large regions can be prohibitively expensive; however, land cover data may provide a surrogate for intensive assessments, enabling rapid and cost-effective evaluation of wetlands throughout whole regions. Our goal was to determine if land cover data could be used to estimate the biotic integrity of wetlands in Alberta??s Beaverhills watershed. Biotic integrity was measured using both plant- and bird-based indices of biotic integrity (IBIs) in 45 wetlands. Land cover data were extracted from seven nested landscape extents (100?C3,000?m radii) and used to model IBI scores. Strong, significant predictions of IBI scores were achieved using land cover data from every spatial extent, even after factoring out the influence of location to address the spatial autocorrelation of land cover classes. Plant-based IBI scores were best predicted using data from 100?m buffers and bird-based IBI scores were best predicted using data extracted from 500?m buffers. Road cover or density and measures of the proportion of disturbed land were consistent predictors of IBI score, suggesting their universal importance to plant and bird communities. Simplified models using the proportion of undisturbed land were less accurate than more detailed models (reductions in r ² of 0.31?C0.32). Regardless of the level of detail in land cover classification, our results emphasize the need to optimize landscape extent for the taxonomic group of interest: an issue that is typically poorly articulated in studies reporting on the development of GIS-based assessment methods. Our results also highlight the need to calibrate models in test areas before scaling up, to ensure predictive accuracy.     

Optimizing landscape selection for estimating relative effects of landscape variables on ecological responses

Empirical studies of the relative effects of landscape variables may compromise inferential strength with common approaches to landscape selection. We propose a methodology for landscape sample selection that is designed to overcome some common statistical pitfalls that may hamper estimates of relative effects of landscape variables on ecological responses. We illustrate our proposed methodology through an application aimed at quantifying the relationships between farmland heterogeneity and biodiversity. For this project, we required 100 study landscapes that represented the widest possible ranges of compositional and configurational farmland heterogeneity, where these two aspects of heterogeneity were quantified as crop cover diversity (Shannon diversity index) and mean crop field size, respectively. These were calculated at multiple spatial extents from a detailed map of the region derived through satellite image segmentation and classification. Potential study landscapes were then selected in a structured approach such that: (1) they represented the widest possible range of both heterogeneity variables, (2) they were not spatially autocorrelated, and (3) there was independence (no correlation) between the two heterogeneity variables, allowing for more precise estimates of the regression coefficients that reflect their independent effects. All selection criteria were satisfied at multiple extents surrounding the study landscapes, to allow for multi-scale analysis. Our approach to landscape selection should improve the inferential strength of studies estimating the relative effects of landscape variables, particularly those with a view to developing land management guidelines.     

Probabilistic patterns of inundation and biogeomorphic changes due to sea-level rise along the northeastern U.S. Atlantic coast

Context

Coastal landscapes evolve in response to sea-level rise (SLR) through a variety of geologic processes and ecological feedbacks. When the SLR rate surpasses the rate at which these processes build elevation and drive lateral migration, inundation is likely.

Objectives

To examine the role of land cover diversity and composition in landscape response to SLR across the northeastern United States.

Methods

Using an existing probabilistic framework, we quantify the probability of inundation, a measure of vulnerability, under different SLR scenarios on the coastal landscape. Resistant areas—wherein a dynamic response is anticipated—are defined as unlikely (p < 0.33) to inundate. Results are assessed regionally for different land cover types and at 26 sites representing varying levels of land cover diversity.

Results

Modeling results suggest that by the 2050s, 44% of low-lying, habitable land in the region is unlikely to inundate, further declining to 36% by the 2080s. In addition to a decrease in SLR resistance with time, these results show an increasing uncertainty that the coastal landscape will continue to evolve in response to SLR as it has in the past. We also find that resistance to SLR is correlated with land cover composition, wherein sites containing land cover types adaptable to SLR impacts show greater potential to undergo biogeomorphic state shifts rather than inundating with time.

Conclusions

Our findings support other studies that have highlighted the importance of ecological composition and diversity in stabilizing the physical landscape and suggest that flexible planning strategies, such as adaptive management, are particularly well suited for SLR preparation in diverse coastal settings.

     

Modelling Habitat Suitability for Deciduous Forest Focal Species – A Sensitivity Analysis using Different Satellite Land Cover Data

We explored the usefulness of three satellite land cover data sets available to land managers in south-central Sweden for conservation planning using four deciduous forest focal resident bird species with different habitat requirements. Habitat suitability models using empirical species-specific habitat parameters and a Geographic Information System were applied to evaluate and compare the degree of consistency among three different land cover data sets. The study area encompassed 10,000 km² in a landscape mosaic of managed boreal forests and is within the distribution range of all four focal species. Although the three land cover data sets indicated similar total amounts of deciduous forest, the habitat suitability models showed that different land cover data yielded inconsistent results regarding the amount and distribution of suitable habitat within 5×5 km grid cells. Given this sensitivity to the choice of land cover data sets, the habitat suitability models showed positive relationships among the selected focal species for each land cover data set separately. As expected, decreasing amounts of suitable habitat were identified for species with higher specialisation. Thus, because habitat suitability models are an appropriate way to gain insight into the functionality and connectivity of habitat networks, land cover data must be carefully evaluated and if necessary combined with other landscape information for effective conservation planning.     

The role of landscape structure in species richness distribution of birds,amphibians, reptiles and lepidopterans in Mediterranean landscapes

The parameters referring to landscape structure are essential in any evaluation for conservation because of the relationship that exists between the landscape structure and the ecological processes. This paper presents a study of the relationships between landscape structure and species diversity distribution (estimated in terms of richness of birds, amphibians, reptiles and butterflies) in the region of Madrid, Spain. The results show that the response of species richness to landscape heterogeneity varies depending on the group of species considered. For birds and lepidopterans, the most important factor affecting the distribution of richness of species is landscape heterogeneity, while other factors, such as the specific composition of land use, play a secondary role at this scale. On the other hand, richness of amphibians and reptiles is more closely related to the abundance of certain land-use types. The study highlights the importance of heterogeneity in Mediterranean landscapes as a criterion for landscape planning and for definition of management directives in order to maintain biodiversity.     

CORINE-based landscape indices weakly correlate with plant species richness in a northern European landscape transect

We present the results of one of the few available tests of how CORINE (CLC2000) is likely to perform as a basis for the calculation of landscape indices, for environmental monitoring over large areas. This paper investigates to what extent landscape structural indices based on this widely used European land cover database can be used to predict plant species richness in a 2,000 km² transect in the northeast of Scotland. We investigate both statistical and map resolution issues by comparing the performance of CORINE-based common landscape indices with the same indices derived from a much more detailed geographic data set. In our case study, only shape-related indices show correlation with species richness, but effect size, important for monitoring, is small. The results highlight the area-specific and map specific nature of the performance of landscape indices for protecting plant diversity. Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users.     

Modeling scale-dependent landscape pattern,dispersal, and connectivity from the perspective of the organism

Understanding the impacts of habitat fragmentation on dispersal is an important issue in landscape and conservation ecology. Here I examine the effects of fine- to broad-scale patterns in landscape structure on dispersal success of organisms with differing life-history traits. An individual-based model was used to simulate dispersal of amphibian-like species whose movements were driven by land cover and moisture conditions. To systematically control spatial pattern, a landscape model was created by merging simulated land cover maps with synthetic topographic surfaces. Landscapes varied in topographic roughness and spatial contagion in agriculture and urban land cover. Simulations included three different species types that varied in their maximum potential dispersal distances by 1-, 2-, or 4-fold. Two sets of simulations addressed effects of varying aspects of landscape structure on dispersal success. In the first set of simulations, which incorporated variable distances between breeding patches, dispersal success was lowest for all species types when anthropogenic cover was patchily distributed. In the second set, with interpatch distances held constant as landscape composition varied, dispersal success decreased as anthropogenic cover became spatially contagious. Both sets revealed strong main effects of species characteristics, interpatch distances and landscape composition on dispersal success; furthermore, scale-dependent patterns in land cover and moisture gradients had a stronger effect on longer- than shorter-ranging species types. Taken together, these simulations suggest that heuristic conservation strategies could potentially be developed based on important but limited life history information.     

The Impact of Misclassification in Land Use Maps in the Prediction of Landscape Dynamics

Land use maps are widely used in modeling land use change, urban sprawl, and for other landscape related studies. A misclassification confusion matrix for land use maps is usually provided as a measure of their quality and uncertainty. However, this very important information is rarely considered in land use map based studies, especially in modeling landscape dynamics. Ignoring uncertainty of land use maps may cause models to provide unreliable predictions. This study is an attempt to investigate the impact of the accuracy of land use maps used as input for an urban sprawl model. In the study area, the regional confusion matrix has been localized using a topographical map. Based on the regional and local confusion matrices, several error levels have been defined. The results showed that a localized confusion matrix that reflected the characteristics of the study area had error rates that were much different than the regional confusion matrix. The predictions of the probability of urban sprawl based on the land use maps and defined error levels were quite different.     

Changes in landscape patterns in Georgia,USA

The objectives of this study were to determine how landscape patterns in Georgia, USA have changed through time and whether the spatial patterns varied by physiographic region. Historical aerial photography was used to analyze spatial patterns of land use from the 1930's to the 1980's. Land use patterns were quantified by: (1) mean number and size of patches; (2) fractal dimension of patches; (3) amount of edge between land uses; and (4) indices of diversity, dominance, and contagion. Forest cover increased in aerial extent and in mean patch size. The mean size of agricultural patches increased in the coastal plain and decreased in the mountains and piedmont. Edges between land uses decreased through time, indicating less dissection of the landscape. Fractal dimensions also decreased, indicating simpler patch shapes. Indices of diversity and dominance differed through time but not among regions; the contagion index differed among regions but not through time. A geographic trend of decreasing diversity and increasing dominance and contagion was observed from the mountains to the lower coastal plain. Landscape patterns exhibited the greatest changes in the piedmont region. Overall, the Georgia landscape has become less fragmented and more connected during the past 50 years. Changing patterns in the landscape may have implications for many ecological processes and resources.