The modifiable areal unit problem and implications for landscape ecology

Landscape ecologists often deal with aggregated data and multiscaled spatial phenomena. Recognizing the sensitivity of the results of spatial analyses to the definition of units for which data are collected is critical to characterizing landscapes with minimal bias and avoidance of spurious relationships. We introduce and examine the effect of data aggregation on analysis of landscape structure as exemplified through what has become known, in the statistical and geographical literature, as theModifiable Areal Unit Problem (MAUP). The MAUP applies to two separate, but interrelated, problems with spatial data analysis. The first is the “scale problem”, where the same set of areal data is aggregated into several sets of larger areal units, with each combination leading to different data values and inferences. The second aspect of the MAUP is the “zoning problem”, where a given set of areal units is recombined into zones that are of the same size but located differently, again resulting in variation in data values and, consequently, different conclusions. We conduct a series of spatial autocorrelation analyses based on NDVI (Normalized Difference Vegetation Index) to demonstrate how the MAUP may affect the results of landscape analysis. We conclude with a discussion of the broader-scale implications for the MAUP in landscape ecology and suggest approaches for dealing with this issue.     

Detecting dominant landscape objects through multiple scales: An integration of object-specific methods and watershed segmentation

Complex systems, such as landscapes, are composed of different critical levels of organization where interactions are stronger within levels than among levels, and where each level operates at relatively distinct time and spatial scales. To detect significant features occurring at specific levels of organization in a landscape, two steps are required. First, a multiscale dataset must be generated from which these features can emerge. Second, a procedure must be developed to delineate individual image-objects and identify them as they change through scale. In this paper, we introduce a framework for the automatic definition of multiscale landscape features using object-specific techniques and marker-controlled watershed segmentation. By applying this framework to a high-resolution satellite scene, image-objects of varying size and shape can be delineated and studied individually at their characteristic scale of expression. This framework involves three main steps: 1) multiscale dataset generation using an object-specific analysis and upscaling technique, 2) marker-controlled watershed transformation to automatically delineate individual image-objects as they evolve through scale, and 3) landscape feature identification to assess the significance of these image-objects in terms of meaningful landscape features. This study was conducted on an agro-forested region in southwest Quebec, Canada, using IKONOS satellite data. Results show that image-objects tend to persist within one or two scale domains, and then suddenly disappear at the next, while new image-objects emerge at coarser scale domains. We suggest that these patterns are associated to sudden shifts in the entire image structure at certain scale domains, which may correspond to critical landscape thresholds.This revised version was published online in May 2005 with corrections to the Cover Date.     

The effect of spatial scale on Konza landscape classification using textural analysis

Spatial scale is inherent in the definition of landscape heterogeneity and diversity. For example, a landscape may appear heterogeneous at one scale but quite homogeneous at another scale. In assessing the impact of burning and grazing on the Konza Prairie Research Natural Area (a tallgrass prairie), spatial scale is extremely important. Textural contrast algorithms were applied to various scales of remote sensing data and related to landscape units for assessment of heterogeneity under a variety of burning treatments. Acquired data sets included Landsat multispectral scanner (MSS), with 80 m resolution, Landsat thematic mapper (TM), with 30 m resolution, and high resolution density sliced aerial photography (with a 5 m resolution). Results suggest that heterogeneous areas of dense patchiness (e.g., unburned areas) must be analyzed at a finer scale than more homogeneous areas which are burned at least every four years.     

A practical toolbox for design and analysis of landscape genetics studies

Landscape genetics integrates theory and analytical methods of population genetics and landscape ecology. Research in this area has increased in recent decades, creating a plethora of options for study design and analysis. Here we present a practical toolbox for the design and analysis of landscape genetics studies following a seven-step framework: (1) define the study objectives, (2) consider the spatial and temporal scale of the study, (3) design a sampling regime, (4) select a genetic marker, (5) generate genetic input data, (6) generate spatial input data, and (7) choose an analytical method that integrates genetic and spatial data. Study design considerations discussed include choices of spatial and temporal scale, sample size and spatial distribution, and genetic marker selection. We present analytical methods suitable for achieving different study objectives. As emerging technologies generate genetic and spatial data sets of increasing size, complexity, and resolution, landscape geneticists are challenged to execute hypothesis-driven research that combines empirical data and simulation modeling. The landscape genetics framework presented here can accommodate new design considerations and analyses, and facilitate integration of genetic and spatial data by guiding new landscape geneticists through study design, implementation, and analysis.     

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.     

Spatial and temporal analysis of landscape patterns

A variety of ecological questions now require the study of large regions and the understanding of spatial heterogeneity. Methods for spatial-temporal analyses are becoming increasingly important for ecological studies. A grid cell based spatial analysis program (SPAN) is described and results of landscape pattern analysis using SPAN are presentedd. Several ecological topics in which geographic information systems (GIS) can play an important role (landscape pattern analysis, neutral models of pattern and process, and extrapolation across spatial scales) are reviewed. To study the relationship between observed landscape patterns and ecological processes, a neutral model approach is recommended. For example, the expected pattern (i.e., neutral model) of the spread of disturbance across a landscape can be generated and then tested using actual landscape data that are stored in a GIS. Observed spatial or temporal patterns in ecological data may also be influenced by scale. Creating a spatial data base frequently requires integrating data at different scales. Spatial is shown to influence landscape pattern analyses, but extrapolation of data across spatial scales may be possible if the grain and extent of the data are specified. The continued development and testing of new methods for spatial-temporal analysis will contribute to a general understanding of landscape dynamics.     

On the accuracy of landscape pattern analysis using remote sensing data

Advances in remote sensing technologies have provided practical means for land use and land cover mapping which is critically important for landscape ecological studies. However, all classifications of remote sensing data are subject to different kinds of errors, and these errors can be carried over or propagated in subsequent landscape pattern analysis. When these uncertainties go unreported, as they do commonly in the literature, they become hidden errors. While this is apparently an important issue in the study of landscapes from either a biophysical or socioeconomic perspective, limited progress has been made in resolving this problem. Here we discuss how errors of mapped data can affect landscape metrics and possible strategies which can help improve the reliability of landscape pattern analysis.     

Thermal infrared remote sensing for analysis of landscape ecological processes: methods and applications

Thermal infrared (TIR) remote sensing data can provide important measurements of surface energy fluxes and temperatures, which are integral to understanding landscape processes and responses. One example of this is the successful application of TIR remote sensing data to estimate evapotranspiration and soil moisture, where results from a number of studies suggest that satellite-based measurements from TIR remote sensing data can lead to more accurate regional-scale estimates of daily evapotranspiration. With further refinement in analytical techniques and models, the use of TIR data from airborne and satellite sensors could be very useful for parameterizing surface moisture conditions and developing better simulations of landscape energy exchange over a variety of conditions and space and time scales. Thus, TIR remote sensing data can significantly contribute to the observation, measurement, and analysis of energy balance characteristics (i.e., the fluxes and redistribution of thermal energy within and across the land surface) as an implicit and important aspect of landscape dynamics and landscape functioning.The application of TIR remote sensing data in landscape ecological studies has been limited, however, for several fundamental reasons that relate primarily to the perceived difficulty in use and availability of these data by the landscape ecology community, and from the fragmentation of references on TIR remote sensing throughout the scientific literature. It is our purpose here to provide evidence from work that has employed TIR remote sensing for analysis of landscape characteristics to illustrate how these data can provide important data for the improved measurement of landscape energy response and energy flux relationships. We examine the direct or indirect use of TIR remote sensing data to analyze landscape biophysical characteristics, thereby offering some insight on how these data can be used more robustly to further the understanding and modeling of landscape ecological processes.     

Scale detection in real and artificial landscapes using semivariance analysis

Semivariance analysis is potentially useful to landscape ecologists for detecting scales of variability in spatial data. We used semivariance analysis to compare spatial patterns of winter foraging by large ungulates with those of environmental variables that influence forage availability in northern Yellowstone National Park, Wyoming. In addition, we evaluated (1) the ability of semivariograms to detect known scales of variability in artificial maps with one or more distinct scales of pattern, and (2) the influence of the amount and spatial distribution of absent data on semivariogram results and interpretation. Semivariograms of environmental data sets (aspect, elevation, habitat type, and slope) for the entire northern Yellowstone landscape clearly identified the dominant scale of variability in each map layer, while semivariograms of ungulate foraging data from discontinuous study areas were difficult to interpret. Semivariograms of binary maps composed of a single scale of pattern showed clear and interpretable results: the range accurately reflected the size of the blocks of which the maps were constructed. Semivariograms of multiple scale maps and hierarchical maps exhibited pronounced inflections which could be used to distinguish two or three distinct scales of pattern. To assess the sensitivity of semivariance analysis to absent data, often the product of cloud interference or incomplete data collection, we deliberately masked (deleted) portions of continuous northern Yellowstone map layers, using single scale artificial maps as masks. The sensitivity of semivariance analysis to random deletions from the data was related to both the size of the deleted blocks, and the total proportion of the original data set that was removed. Small blocks could be deleted in very high proportions without degrading the semivariogram results. When the size of deleted blocks was large relative to the size of the map, the corresponding variograms became sensitive to the total proportion of data removed: variograms were difficult or impossible to interpret when the proportion of data deleted was high. Despite success with artificial maps, standard semivariance analysis is unlikely to detect multiple scales of pattern in real ecological data. Semivariance analysis is recommended as an effective technique for quantifying some spatial characteristics of ecological data, and may provide insight into the scales of processes that structure landscapes.     

Effects of changing spatial resolution on the results of landscape pattern analysis using spatial autocorrelation indices

Understanding the relationship between pattern and scale is a central issue in landscape ecology. Pattern analysis is necessarily a critical step to achieve this understanding. Pattern and scale are inseparable in theory and in reality. Pattern occurs on different scales, and scale affects pattern to be observed. The objective of our study is to investigate how changing scale might affect the results of landscape pattern analysis using three commonly adopted spatial autocorrelation indices,i.e., Moran Coefficient, Geary Ratio, and Cliff-Ord statistic. The data sets used in this study are spatially referenced digital data sets of topography and biomass in 1972 of Peninsular Malaysia. Our results show that all three autocorrelation indices were scale-dependent. In other words, the degree of spatial autocorrelation measured by these indices vary with the spatial scale on which analysis was performed. While all the data sets show a positive spatial autocorrelation across a range of scales, Moran coefficient and Cliff-Ord statistic decrease and Geary Ratio increases with increasing grain size, indicating an overall decline in the degree of spatial autocorrelation with scale. The effect of changing scale varies in their magnitude and rate of change when different types of landscape data are used. We have also explored why this could happen by examining the formulation of the Moran coefficient. The pattern of change in spatial autocorrelation with scale exhibits threshold behavior,i.e., scale effects fade away after certain spatial scales are reached (for elevation). We recommend that multiple methods be used for pattern analysis whenever feasible, and that scale effects must be taken into account in all spatial analysis.     

A comparison of quantitative methods for examining landscape pattern and scale

Ecologists have long recognized the importance of spatial and temporal patterns that characterize heterogeneity in landscapes. However, despite the realization that inferences about ecological phenomena are scale dependent, little attention has been paid to determining appropriate scales of measurement (e.g., plot or grain size) in studies of landscape dynamics or ecosystem change. This paper compares the results from three data sets using several quantitative methods available for characterizing landscape heterogeneity and/or for determining scale of measurement. Methods evaluated include tests of non-randomness, estimation of patch size, spectral analysis, fractals, variance ratio analysis, and correlation analysis. The results showed that no one method provides consistently good estimates of scale. Thus, sampling strategies for landscape studies should be derived from estimates of patch size and/or scale of pattern obtained from more than one of these methods.     

The r.le programs for multiscale analysis of landscape structure using the GRASS geographical information system

Geographical information systems (GIS) are well suited to the spatial analysis of landscape data, but generally lack programs for calculating traditional measures of landscape structure (e.g., fractal dimension). Standalone programs for calculating landscape structure measures do exist, but these programs do not enable the user to take advantage of GIS facilities for manipulating and analyzing landscape data. Moreover, these programs lack capabilities for analysis with sampling areas of different size (multiscale analysis) and also lack some needed measures of landscape structure (e.g., texture).We have developed the r.le programs for analyzing landscape structure using the GRASS GIS. The programs can be used to calculate over sixty measures of landscape structure (e.g., distance, size, shape, fractal dimension, perimeters, diversity, texture, juxtaposition, edges) within sampling areas of several sizes simultaneously. Also possible are moving window analyses, which enable the production of new maps of the landscape structure within windows of a particular size. These new maps can then be used in other analyses with the GIS.     

Hierarchical relationships between landscape structure and temperature in a managed forest landscape

Management may influence abiotic environments differently across time and spatial scale, greatly influencing perceptions of fragmentation of the landscape. It is vital to consider a priori the spatial scales that are most relevant to an investigation, and to reflect on the influence that scale may have on conclusions. While the importance of scale in understanding ecological patterns and processes has been widely recognized, few researchers have investigated how the relationships between pattern and process change across spatial and temporal scales. We used wavelet analysis to examine the multiscale structure of surface and soil temperature, measured every 5 m across a 3820 m transect within a national forest in northern Wisconsin. Temperature functioned as an indicator – or end product – of processes associated with energy budget dynamics, such as radiative inputs, evapotranspiration and convective losses across the landscape. We hoped to determine whether functional relationships between landscape structure and temperature could be generalized, by examining patterns and relationships at multiple spatial scales and time periods during the day. The pattern of temperature varied between surface and soil temperature and among daily time periods. Wavelet variances indicated that no single scale dominated the pattern in temperature at any time, though values were highest at finest scales and at midday. Using general linear models, we explained 38% to 60% of the variation in temperature along the transect. Broad categorical variables describing the vegetation patch in which a point was located and the closest vegetation patch of a different type (landscape context) were important in models of both surface and soil temperature across time periods. Variables associated with slope and microtopography were more commonly incorporated into models explaining variation in soil temperature, whereas variables associated with vegetation or ground cover explained more variation in surface temperature. We examined correlations between wavelet transforms of temperature and vegetation (i.e., structural) pattern to determine whether these associations occurred at predictable scales or were consistent across time. Correlations between transforms characteristically had two peaks; one at finer scales of 100 to 150 m and one at broader scales of ^>300 m. These scales differed among times of day and between surface and soil temperatures. Our results indicate that temperature structure is distinct from vegetation structure and is spatially and temporally dynamic. There did not appear to be any single scale at which it was more relevant to study temperature or this pattern-process relationship, although the strongest relationships between vegetation structure and temperature occurred within a predictable range of scales. Forest managers and conservation biologists must recognize the dynamic relationship between temperature and structure across landscapes and incorporate the landscape elements created by temperature-structure interactions into management decisions.     

Disturbance propagation by bark beetles as an episodic landscape phenomenon

Landscapes are the resultant of ecological processes and events operating on many different space-time scales. Large scale disturbance is recognized as a major influence on landscape patterns, but the impact of small scale events is often overlooked. We develop an hierarchical framework to relate lightning and bark beetle population dynamics to the southern pine forest landscape using the concepts of disturbance propagation and amplification. The low level lightning disturbance can be propagated to the landscape level when weather and forest stand structure facilitate bark beetle epidemics. We identify epidemics as biotically-driven episodes that alter landscape structure. The concept of the landscape as the spatial dimension of these episodes is represented in a conceptual model linking insect-host and landscape mosaic interactions.     

Effects of landscape structure on nest predation in roadsides of a midwestern agroecosystem: a multiscale analysis

Nest predation is an important cause of mortality for many bird species, especially in grassland ecosystems where generalist predators have responded positively to human disturbance and landscape fragmentation. Our study evaluated the influence of the composition and configuration of the surrounding landscape on nest predation. Transects consisting of 10 artificial ground nests each were set up in 136 roadsides in six watersheds in south-central Iowa. Nest predation on individual roadside transects ranged from 0 to 100% and averaged 23%. The relationship of landscape structure within spatially-nested landscapes surrounding each roadside transect (within 200, 400, 800, 1200, and 1600 m of the transect line) to nest predation was evaluated by using multiple regression and canonical correlation analyses. The results of this multiscale landscape analysis demonstrated that predation on ground nests was affected by the surrounding landscape mosaic and that nest predators with different-sized home ranges and habitat affinities responded to landscapes in different ways. In general, wooded habitats were associated with greater nest predation, whereas herbaceous habitats (except alfalfa/pasture) either were associated with less nest predation or were not important. Different landscape variables were important at different spatial scales. Whereas some block-cover habitats such as woodland were important at all scales, others such as rowcrops and alfalfa/pasture were important at large scales. Some strip-cover habitats such as gravel roads and paved roads were important at small scales, but others such as wooded roadsides were important at all all scales. Most landscape metrics (e.g., mean patch size and edge density) were important at large scales. Our study demonstrated that the relationships between landscape structure and predator assemblages are complex, thus making efforts to enhance avian productivity in agricultural landscapes a difficult management goal.     

Neutral models for the analysis of broad-scale landscape pattern

The relationship between a landscape process and observed patterns can be rigorously tested only if the expected pattern in the absence of the process is known. We used methods derived from percolation theory to construct neutral landscape models,i.e., models lacking effects due to topography, contagion, disturbance history, and related ecological processes. This paper analyzes the patterns generated by these models, and compares the results with observed landscape patterns. The analysis shows that number, size, and shape of patches changes as a function of p, the fraction of the landscape occupied by the habitat type of interest, and m, the linear dimension of the map. The adaptation of percolation theory to finite scales provides a baseline for statistical comparison with landscape data. When USGS land use data (LUDA) maps are compared to random maps produced by percolation models, significant differences in the number, size distribution, and the area/perimeter (fractal dimension) indices of patches were found. These results make it possible to define the appropriate scales at which disturbance and landscape processes interact to affect landscape patterns.     

Assessing landscape health by scaling with remote sensing: when is it not enough?

Assessment of the health of landscapes, by monitoring their condition over space and time, is needed to better understand the processes for sustaining or, in many cases, repairing them. Remote sensing is a tool that can efficiently identify and assess areas of landscape damage at different scales and help land managers solve specific problems. Remote sensing may appear to be a panacea for all monitoring situations but sometimes the information it provides is not enough by itself. In this paper we give examples of both scenarios—when remote sensing alone is adequate and when it is not. When remotely sensed data alone is not sufficient, monitoring problems can be solved by incorporating additional finer scale data. We use a five-step procedure based on scaling to help land managers answer the question: when is remote sensing data alone not sufficient to underpin the information needs required to achieve a specific management goal?     

The influence of scale and the spatial characteristics of landscapes on land-cover mapping using remote sensing

Statistical analyses provide a means for assessing relationships between landscape spatial pattern and errors in the estimates of cover-type proportions as land-cover data are aggregated to coarser scales. Results from a multiple-linear regression model suggest that as patch sizes, variance/mean ratio, and initial proportions of cover types increase, the proportion error moves in a positive direction and is governed by the interaction of the spatial characteristics and the scale of aggregation. However, the standard linear model does not account for the different directions of scale-dependent proportion error since some classes become larger and others become smaller as the scene is aggregated. Addition of indicator variables representing class-type significantly improves the performance by allowing the model to respond differently to different classes. A regression tree model provides a much simpler fit to the complex scaling behavior through an interaction between patch size and aggregation scale. An understanding of the relationships between landscape pattern, scale, and proportion error may advance methods for correcting land-cover area estimates. Such methods could also facilitate high-resolution calibration and validation of coarse-scale remote-sensing-based land-cover mapping algorithms. Ongoing initiatives to produce global land-cover datasets from remote sensing, such as efforts within the IGBP and the EOS MODIS Land-Team, include significant emphasis on high level calibration and validation activities of this nature.     

Additive partitioning of plant species diversity in an agricultural mosaic landscape

In this paper, we quantify the effects of habitat variability and habitat heterogeneity based on the partitioning of landscape species diversity into additive components and link them to patch-specific diversity. The approach is illustrated with a case study from central Switzerland, where we recorded the presence of vascular plant species in a stratified random sample of 1'280 quadrats of 1 m² within a total area of 0.23 km². We derived components of within- and between-community diversity at four scale levels (quadrat, patch, habitat type, and landscape) for three diversity measures (species richness, Shannon index, and Simpson diversity). The model implies that what we measure as within-community diversity at a higher scale level is the combined effect of heterogeneity at various lower levels. The results suggest that the proportions of the individual diversity components depend on the habitat type and on the chosen diversity aspect. One habitat type may be more diverse than another at patch level, but less diverse at the level of habitat type. Landscape composition apparently is a key factor for explaining landscape species richness, but affects evenness only little. Before we can test the effect of landscape structure on landscape species richness, several problems will have to be solved. These include the incorporation of neighbourhood effects, the unbiased estimation of species richness components, and the quantification of the contribution of a landscape element to landscape species richness.