Abstract
In this review paper, we highlight some of the trends and issues that have dominated ecotone research over the decade of 1996–2006. The terms and definitions of ecotone research in vegetation ecology are reviewed. We summarize the most important techniques and highlight the discrepancies between the definitions and their scientific application in vegetation ecology. We see a need for semantic uniformity with regard to the term and the definition of “the ecotone”, as the variable and the non-exclusive use of terms and definitions can be a source of confusion when interpreting and comparing different studies. To avoid further confusion, a unique definition of the term “ecotone” should be agreed upon, based upon a set of general characteristics. We therefore adapted and extended the definition from Holland et al. (Ecotones: the role of landscape boundaries in the management and restoration of changing environments, 1991) to “A multi-dimensional environmentally stochastic interaction zone between ecological systems with characteristics defined in space and time, and by the strength of the interaction”. We also advocate that (1) a shift in focus from one-dimensional to two-dimensional techniques in ecotone characterization is desirable and (2) more research into novel techniques, including multi-dimensional data and time series, is needed in view of local and global ecotone changes.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
Introduction
At the beginning of the 20th century, Livingston (1903) defined an ecotone as: “A stress line connecting points of accumulated or abrupt change is an ecotone.” Hereby, this definition of the ecotone even predates the well-known ecosystem concept as conceived by Tansley (1935). In less formal words, an ecotone is described in Webster’s Dictionary and Thesaurus of the English Language as “a transitional area between two ecological communities”. Ecotones can be found in many forms and across a range of scales, from a few centimeters to several kilometers. They are boundaries between biomes or ecosystems (Hansen and di Castri 1992), and can be compared to the semi-permeable membranes of cells (Naiman and Decamps 1997). Ecotones have a profound influence on adjacent ecosystems (Saunders et al. 1999; Cadenasso and Pickett 2000; Laurance et al. 2001). The fast rate of change in ecotones is reflected by ecotone dynamics such as fluxes between neighboring ecological systems, and can affect even remote locations within these systems (Naiman and Décamps 1990; Saunders et al. 1999; Cadenasso et al. 2003). Furthermore, as an ecotone is a dynamic entity with both spatial and temporal properties, its width and position evolve with time during succession or environmental changes on both a local and a global scale (Wiens et al. 1985; Forman 1995).
Aside from their role in material flow between ecosystems, ecotones have other properties. Ecotones play an important role in speciation, and harbor more species than core areas of ecosystems (Schilthuizen 2000). Ecotones are also indicators of local (Risser 1993; Enserink 1997; Smith et al. 1997; Allen and Breshears 1998; Loehle 2000; Goldblum and Rigg 2005) and global changes (Churkina and Svirezhev 1995). Most ecotones are described as either physical or functional units, or as a combination of both (Kent et al. 2006). Furthermore, ecotone research can be linked to landscape ecology topics such as edge penetration, edge effects, interior habitat, and ecological gradients (Ewers and Didham 2006) as patches in fragmented landscapes are spatially defined by their physical limits, or borders (Forman 1995). Both biotic and abiotic properties of the patches are influenced by and subjected to spatial dynamics across these borders (Fernandez et al. 2002). Increasing fragmentation of the landscape will also increase the importance of these dynamics within the created patches (Li 2000). Typical examples of ecotones can be found in the arctic forest-tundra transition (Hofgaard and Wilmann 2002; Ranson et al. 2004; Stow et al. 2004; Jia et al. 2006), the forest-woodland transition (Allen and Breshears 1998), forest-savannah transitions (Mariotti and Peterschmitt 1994; Puyravaud et al. 1994; Favier et al. 2004) and elevational gradients in mountainous areas (Camarero et al. 2006).
In this paper, we provide an overview of the terms and definitions of the ecotone that have been proposed since the first definition of Livingston (1903). We discuss the trends and characteristics of these definitions as they have changed over time. Since the introduction of the ecotone concept, and particularly over the last 25 years, the number of publications on this subject has grown steadily. Considering the steady growth in ecotone research, we review the relation between the techniques and definitions as used during a decade (1996–2006) of ecotone research of terrestrial vegetation ecology. Furthermore, a short overview of the available techniques used to identify and describe ecotones, including both those originating in the early days of ecotone research and the more recent techniques, is provided. Each methodology is described with key references to provide the reader with a comprehensive reference list.
Various definitions
Due to the diverse nature of ecotone research, we limited our review analysis to ecotones in terrestrial vegetation on all different scales. To avoid confusion about terms and definitions within the framework of this paper, we use the definition of ecotone as proposed by Holland et al. (1991), i.e., “ecotones are zones of transition between adjacent ecological systems, having a set of characteristics uniquely defined by space and time scales and by the strength of interactions between adjacent ecological systems”. This definition explicitly includes the spatial as well as the temporal aspects of the ecotones that are governed by the interaction between adjacent ecological systems. We use this definition because it is the most inclusive compared to other definitions (Table 1). Ecosystems have, strictly speaking, no given units and are subjected to the normative rationale of the observer. However, the term is often used as a fixed and given unit. To avoid confusion, we opted for the use of the term “ecological systems” at all different scales, instead of the term "ecosystem".
Hansen and di Castri (1992) considered the above-mentioned definition of Holland et al. (1991) as applicable to any hierarchical level from a few centimeters to a few kilometers, or from the population level to the biosphere level depending on the perspective of the researcher. Ecotones are often referred to as edges (Orlóci and Orlóci 1990; Ries and Sisk 2004) or boundaries (Fortin et al. 2000; Fagan et al. 2003; Kent et al. 2006). They are sometimes described as broad areas of transition (Smith et al. 1997; Holland et al. 1991), whereas edges are often considered at more local scales, such as the boundary between two patch types (Ries and Sisk 2004). Therefore, the term “edge” is frequently used when interpreting effects of anthropogenic change, whereas the term “ecotone” is commonly used in studies of more natural transitions between ecological systems. The term “boundary” is often used as a synonym for an edge, but implies relative (or absolute) impermeability, which may or may not be the case for an edge (Fagan et al. 2003). Although different definitions exist, we use these definitions synonymously because ecotones all share common characteristics. These common characteristics are the origin, function, spatial structure, and changes throughout time (Strayer et al. 2003).
Table 1 summarizes the most common terms, their definitions, and their associated dimensionality, which is strongly linked to the definition. The table shows that over the years, a range of quite different definitions and terms has emerged. Furthermore, the same terminology can have different definitions, and comparable definitions can have different terminologies used by different authors. The variety in these definitions often originates from different interpretations because of a researcher’s point of view, his or her scientific background, or the experimental setup that has been chosen (e.g., the grain and extent of the study). Interesting reflections such as, for example, those by van der Maarel (1990) on the differences between ecotones and ecoclines, did not result in a clear definition or terminology. This variable and non-exclusive use of terms and definitions can be a source of confusion when interpreting and comparing different studies. However, it can be argued that until today, the plethora of terms and definitions did not hamper research as ecotone research continues and increases. Although the various terms may not have led to far-reaching confusion or chaos, it does not encourage a more transparent view on ecotone research in vegetation studies or otherwise. Furthermore, confusing terminology does not encourage interdisciplinary work nor comparative studies. Ecologists who like to compare empirical studies should be careful and ensure that the different studies are truly comparable (Strayer et al. 2003). A concise use of terms would be beneficial to all scientists including those that are just starting on ecotone research. To avoid further confusion, a set of clearly defined terms and guidelines for defining the ecotone is encouraged.
In reviewing the current definitions, we developed a set of guidelines or characteristics for the term ecotone based upon a synthesis of key components proposed in various studies (Table 1) and guided by the classification of Strayer et al. (2003). As ecotone processes can be characterized in space and time, the ecotone is considered to be multi-dimensional. Consequently, the ecotone is not confined to one or two dimensions, and more often than not, has a temporal component (Wiens et al. 1985; Forman 1995). However, the representation of the ecotone is often limited by the dimensionality of the technique used to characterize its (multi-dimensional) properties. An example of the temporal component of ecotones is illustrated by the rapid shift of a forest woodland ecotone under a variable climate as described by Allen and Breshears (1998). Furthermore, ecotones are created by processes driving an ecological response. Often multiple processes are driving this response and the final ecotone characteristics. Therefore a definition should highlight the importance of a multivariate approach. Based upon the previously mentioned characteristics and the definition of Holland et al. (1991), an ecotone can be defined as: “An environmentally stochastic interaction zone between ecological systems with characteristics defined in space and time, by the strength of the interaction and their driving processes”.
Techniques and methods: an overview
The range of techniques used to quantify or delineate ecotones is as diverse as the plethora of definitions. To investigate the occurrence of various techniques in recent terrestrial vegetation studies and their relation to the various definitions we searched the ISI Web of Knowledge Science Citation index on the topics of ecotone, boundary, transition, edge, ecocline, and gradient between 1996 and 2006. We further limited the search query to the following subject areas: multidisciplinary sciences, plant sciences, biology, environmental sciences, ecology, forestry, physical geography, soil sciences, and biodiversity conservation. A final refinement of the search query on document type, article, and reviews, produced 6,560 peer-reviewed articles. Articles were deemed relevant to our analysis if they used one of the techniques described below as part of a research paper on terrestrial vegetations. Furthermore, all reviews concerning ecotones were included. We analyzed the resulting 332 peer-reviewed articles spanning 10 years by grouping them according to their topic or the technique used in the analysis. Details on the various techniques are described below and are further summarized in Table 2. We chose only to include terrestrial vegetation studies as including all scientific disciplines would generate an unmanageable volume of data.
We make a clear distinction between one-dimensional and two-dimensional techniques, as the ecotone has an inherently spatial character with width and has complementary information such as variation in width and shape (Strayer et al. 2003; Fortin 1994; Johnston et al. 1992; Table 2 column 2). Furthermore, ecotones are often defined by more than one ecological or geophysical property or process (Cadenasso et al. 1997), making the ecotone width very location-specific because of the interplay of various properties and processes. The statistical testability in Table 2 indicates that certain techniques allow statistical permutation tests, which can determine the probability of the delineation of the ecotone (Fortin and Drapeau 1995). The scale-dependence refers to the sensitivity of the techniques to the scale (grain as well as extent) at which the analysis is performed (Strayer et al. 2003). As is common in landscape ecology, ecotones can be defined using different grain sizes (Turner 2005). Depending on grain size, ecotones can appear in various configurations or disappear completely (Strayer et al. 2003). For example, a grain size tailored to detect and characterize a forest grassland ecotone will not likely pick up small ecotones within the grassland community itself. Consequently, acknowledging the scale component in any definition or technique is of paramount importance.
Moving split window
The moving split window (MSW) technique (Whittaker 1967) was one of the first techniques used to describe ecotones. The MSW technique identifies ecotones along a one-dimensional gradient-oriented transect by comparing the variance between two adjacent sampling windows. The dissimilarity is calculated with a distance metric depending on the variable under investigation. A comprehensive review on the use of the MSW technique has been given by Ludwig and Cornelius (1987). More recently, the MSW technique has been implemented in a study of treeline ecotones in the Pyrenees (Camarero et al. 2006).
Wavelets
In addition to the MSW technique, a recent development is the wavelet analysis of transect data (Camarero et al. 2006). The wavelet technique covers multiple scales and allows for estimations of the scale at which the ecotone is the most prominent (Dale and Mah 1998). The wavelet approach is less sensitive to noise than the MSW technique. Although popular, the MSW approach relies heavily on arbitrary choices made by the researcher. Both the position and orientation of the transect and the spacing of the sampling are often chosen arbitrary. The scale-dependence of the algorithm together with an arbitrary choice in sample spacing can introduce a systematic bias.
Ordination techniques
Choesin and Boerner (2002) compared the MSW technique with an ordination technique, namely Detrended Correspondence Analysis (DCA, Hill and Gauch 1980), to compare their boundary detection capabilities. An overview of classic ordination techniques is given by Gauch et al. (1977). However, ordination is limited by strong distortions due to non-linear species abundance relations. To overcome the distortions such as the horseshoe effect in principal component analysis (PCA) and the arc effect in correspondence analysis (CA), new techniques were created. These techniques include DCA, non-metric multi-dimensional scaling (NMDS, Anderson 1971), artificial neural networks (ANN, Lek and Guegan 2000) and self-organizing maps (SOM, Kohonen 1982). All these ordination techniques share the same basic idea. They are multivariate techniques that aim to reduce multi-dimensional datasets to a lower dimension for exploratory analysis.
Sigmoid wave curve fitting
In addition to the above-mentioned techniques used to evaluate transect data, the sigmoid wave technique, as introduced by Timoney et al. (1993), was successfully applied by Cairns and Waldron (2003). The technique fits a sigmoid curve to transect data to estimate parameters such as steepness and the width of the ecotone. Because of their one-dimensional nature, transect-based techniques are not appropriate for two-dimensional spatial fields (Jacquez et al. 2000). However, an effort has been made to extend the sigmoid wave technique into two dimensions (Hufkens et al. 2008). Although this technique has shown to be successful, it is limited by extensive preprocessing and calculation times.
Two-dimensional boundary detection methods
To overcome the limitations with regard to the dimensionality of the data, two-dimensional boundary detection methods have been developed. The work of Fortin (1994, 1999) and Kent et al. (2006) emphasized the importance of using two-dimensional data across an ecotone. Most of these techniques are directly related to image analysis and operate on two-dimensional rasterized digital satellite or photogrammetric images. The simplest operators on these images are edge detection filters, and calculate first-order derivatives between adjacent pixels in a 3 × 3 moving window (Pitas 2000). For example, ecotones of live green vegetation can be detected by scanning an image for the maximum contrast in vegetation indices using edge detection filters (Johnston and Bonde 1989). These first-order-based filters only indicate the position of the boundary. Edge detection filters based upon second-order derivatives can be used to detect the extent of a transition as well as its position. These filters are called Laplacian filters. However, Laplacian filters are sensitive to noise and need smoothening, which results in a loss of detail present in the original dataset (Pitas 2000). The straightforward implementation and interpretation of edge detection filters and their results make them a convenient tool for characterizing ecotones. However, these advantages are countered by some disadvantages as the high sensitivity to noise and the calculation intensity of the algorithms.
Clustering techniques
Clustering techniques also originated within the field of image analysis. Spatially constrained clustering, as used in ecotone research (Camarero and Guti 2002; McIntire and Fortin 2006), is a multivariate technique that identifies edges of homogeneous areas. Similarity within the homogeneous areas can be quantified using various similarity measures (Anderberg 1973). This results in closed and sharp boundaries between homogeneous areas. This technique only provides information about the position of the boundary, but not about the extent. A recent adaptation of this technique includes the multiscale wavelet-based approach (Hay et al. 2001). Clustering techniques have many advantages, as the easy implementation and quite straightforward interpretation of the results. However, their use is limited as clustering only provides data about the position of the ecotone and not about the ecotone width or intensity.
Fuzzy logic
Many processes in nature are inherently fuzzy or imprecise, which is reflected in their attribute ambiguity (Brown 1998). This attribute ambiguity is neglected in spatially constrained clustering. Therefore, fuzzy set modeling aims to identify the fuzzy areas or fuzzy sets (Burrough and Frank 1996). A fuzzy set is created by location uncertainty or by the inherent fuzzy nature of the process. Fuzzy set modeling is a clustering technique where a membership function is calculated for sample locations, such as the classification entropy (Bezdek et al. 1984) or the confusion index (Burrough and Frank 1996). Both indices vary between zero and one. They are zero if the membership is in one class only, and one if spread evenly across classes. Thus, the ecotone and its extent can be represented as an uncertainty value. Ecotone properties, as the rate of change and the position of the ecotone, can be extracted and represented by spatially constrained clustering (Fangju and Hall 1996) or posterior threshold values to the calculated probabilities (Arnot et al. 2004; Fisher et al. 2006; Hill et al. 2007). Moreover, as with some other techniques, it is uncertain if the (observed) fuzzy boundaries and derived maps using alpha-cuts correspond with true ecotone regions, and validation is therefore needed (Arnot et al. 2004; Hill et al. 2007). Additionally, change detection using fuzzy methods is difficult as Boolean reasoning does not apply (Fisher et al. 2006). Notwithstanding, fuzzy set modeling has been successfully applied in studies of a forest-savannah ecotone in Ghana (Foody and Boyd 1999) and of an alpine treeline ecotone (Hill et al. 2007).
Wombling techniques
Another two-dimensional technique that is older than the previously described techniques is the wombling technique (Womble 1951). Although this technique was developed in 1951, it has only recently become commonly used (Oden et al. 1993; Fortin 1994, 1997, 1999; Fortin and Drapeau 1995; Fortin et al. 2000). The wombling technique computes the first-order partial derivatives of each adjacent sampling point. Input can be a rectangular lattice, as in the case of remote sensing imagery, from an irregular lattice, or even from categorical data. In lattice-wombling partial derivatives are calculated over a four-cell window of the regular (lattice) data. If irregular data are used, the partial derivatives of three nearest-neighbor points are used, effectively implementing Delaunay triangles in this irregular point distribution. Categorical wombling uses categorical data points or areas in regular or irregular distributions. When the adjacent points or surfaces are similar, the rate of change will approach zero; when the rate of change is high, the calculated slope values will also be higher. Boundary elements (BE) are those locations with a high (average) rate of change and slope values. Defining “high” is however arbitrary (Jacquez et al. 2000). Generally, slope values in the upper 5–10 percentile are considered boundary elements. This arbitrary choice of what is considered a BE and the difficult implementation is one of the few drawbacks of wombling. More detailed discussions of ecotone detection by the wombling technique are provided by Fortin et al. (2000), Jacquez et al. (2000) and Kent et al. (2006).
Spatial statistics
Finally, two-dimensional point patterns and possible ecotone characteristics can be described using spatial statistics and other statistical techniques. Examples of these techniques can be found in spatial point patterns (Pélissier and Goreaud 2001) and Getis-type statistics (Wulder and Boots 1998; Boots 2001).
Towards a standard conceptual framework?
Although most definitions rely on the two-dimensional nature of the ecotone (see Table 1, dimensionality column), our review of the 332 papers published between 1996 and 2006 on terrestrial ecotone research showed a large discrepancy between the definitions and the applied methods (Table 2). Even though various studies highlight the need for more two-dimensional research because of the inherent spatial nature of the ecotone (Fortin 1994; Johnston et al. 1992) more than 50% of the reviewed publications apply one-dimensional techniques. Within this group, ~40% used MSW based methods to identify the ecotone. However, the adoption of two-dimensional ecotone detection techniques seems to be limited. Less than 24% of the reviewed papers used two-dimensional techniques. Advantages of the MSW transect-based techniques are their easy implementation and evaluation as well as their ability to handle univariate as well as multivariate data (Ludwig and Cornelius 1987), which explain their popularity compared to the two-dimensional techniques. Nevertheless, these one-dimensional techniques can be extended into two dimensions by using multiple transects to describe the ecotone (Ludwig and Cornelius 1987; Johnston et al. 1992). However, there are few studies that use multiple transects and integrate the results into a clear two-dimensional evaluation of the ecotone (Fortin 1994; Camarero et al. 2006). Some key issues inherent to the MSW analysis remain problematic. The analysis is scale-dependent, making comparisons between studies difficult. The position of the transect(s) determines, to a great extent, the outcome of the analysis when used in a more limited one-dimensional setup (Cairns and Waldron 2003). The one-dimensional techniques also comprise ~10% of all ecotone studies based upon vegetation analysis. When the vegetation analysis technique DCA was compared to the MSW technique, DCA revealed fine-scaled ecotone details, but the results were often difficult to extrapolate (Choesin and Boerner 2002). Therefore, DCA should be considered as a supplement to MSW, and should not be used as an independent technique. Yet, the MSW analysis has been employed in almost twice as many studies as the two-dimensional techniques. The remaining studies used a modeling approach in 12.9% of the cases, some 8% were revision of techniques and finally 4.3% were theoretical papers.
When Table 1 and the statistical results are compared, the distribution of publications over one- and two-dimensional techniques are the reverse of the definitions (Table 1). This comparison shows a lack of attention to the two-dimensional techniques and aspects of the ecotone. In recent years, two-dimensional data have become more abundant as satellite remote sensing data can be obtained from various providers. For small-scale studies, alternative techniques of data acquisition can be used. Airplane-mounted cameras, for example, can provide high-resolution remote sensing data (Huete et al. 1999). Conventional digital cameras mounted on various other low-cost platforms, such as a cherry picker (White et al. 2000) or a tethered balloon (Chen and Vierling 2006) can also provide scientifically sound data.
Ideally, a reliable technique should adhere to the following guidelines: (1) be multivariate to elucidate the pattern-process interaction of various biotic and abiotic variables within the ecotone; (2) be scale-independent to allow comparisons across datasets with a different spatial grain and extent; (3) be multi-dimensional to allow an analysis of both the spatial and temporal factors that influence the ecotone; and (4) be statistically testable to test the ecotone transition against a random neutral situation. Only the wombling technique meets all these requirements. Wombling has been proven to be multivariate, scale-independent (Fortin 1999) and statistically testable (Fortin and Drapeau 1995). The use of remote sensing imagery also allows for the evaluation of time series. Within this perspective, time series should be considered in ecotone research, as ecotones are well-known indicators of global and local changes (Risser 1993; Churkina and Svirezhev 1995; Enserink 1997; Smith et al. 1997; Allen and Breshears 1998; Loehle 2000; Schilthuizen 2000; Goldblum and Rigg 2005). Repeated measurements can be used to reveal temporal shifts of the ecotone (Ludwig and Cornelius 1987). A two-dimensional approach should be favored when evaluating ecotone time series since one-dimensional analyses are sensitive to location-based variability (Foody and Boyd 1999). We should therefore consider the use of two-dimensional time series analyses of ecotones more readily. From the summary of techniques (Table 1), we conclude that at present wombling is the most complete technique to describe ecotones and their related properties.
However, the gathered statistics also emphasize the need for more theoretical and conceptual studies (4.3%). In the overview of techniques (Table 2), we observe that most techniques were developed before the period under review (1996–2006), and recent innovations have been slow. Even though the wombling technique could be regarded as the “one-size-fits-all” answer within ecotone research, innovation and the search for alternative methodologies for ecotone research remain important. Since ecotones can be identified on various scales, the resolution of measurements could influence the outcome of a study. Although wombling is scale-independent, a multiscale approach could benefit ecotone research in characterizing ecotone-specific scales (Hay et al. 2001). Moreover, innovative research on alternative techniques, such as the statistical approaches of Pélissier and Goreaud (2001) and the curve-fitting approach (Cairns and Waldron 2003), should be further encouraged.
Conclusions
Although comprehensive frameworks for ecotone research are present (e.g., Cadenasso et al. 2003; Ries and Sisk 2004; Strayer et al. 2003), there is an urgent need to implement these frameworks. The present review illustrates the need for uniform terminology and definition for an ecotone that spans multiple scientific domains. We created a set of guidelines to be used in a new definition. We extended the original definition of Holland et al. (1991) to make it more inclusive and adhere to these guidelines. However, we feel that only the multivariate component in our definition is significantly different from the one proposed by Holland et al. (1991), the adoption of this definition as most suitable should be considered. Although emphasizing the multivariate nature of the ecotone could enhance the definition even further. In general, any definition should be based upon general ecotone attributes and not on specific functional or structural characteristics.
Furthermore, the non-exclusive and variable use of terms and definitions can be a source of confusion when interpreting and comparing different studies. Therefore, a standard framework of attribute types such as the one of Strayer et al. (2003) should be adopted to refine the general definition when needed. Wombling could be adopted as the best technique for ecotone detection, but the exploration of new techniques should be encouraged.
Although the analysis of ecotone research was limited to vegetation sciences the results and conclusions of this study are of a more general nature and can be extrapolated to every ecotone study alike. Scientists should shift their focus from the frequently used one-dimensional techniques to two-dimensional techniques, as they provide better insight into ecotone processes and structures. An effort should be made to develop and extend the use of multi-dimensional and multi-scale data in ecotone research. The data, technology and techniques are available and should be put to best use.
References
Allen CD, Breshears DD (1998) Drought-induced shift of a forest-woodland ecotone: rapid landscape response to climate variation. Proc Natl Acad Sci USA 95:14839–14842. doi:10.1073/pnas.95.25.14839
Allen TFH, Starr TB (1982) Perspectives for ecological complexity. University of Chicago Press, Chicago
Anderberg MR (1973) Cluster analysis for applications. Academic Press, New York
Anderson AJB (1971) Ordination methods in ecology. J Ecol 59:713–726. doi:10.2307/2258135
Arnot C, Fisher PF, Wadsworth R, Wellens J (2004) Landscape metrics with ecotones: pattern under uncertainty. Landsc Ecol 19:181–195. doi:10.1023/B:LAND.0000021723.24247.ee
Bezdek JC, Ehrlich R, Full W (1984) FCM: the fuzzy c-means clustering algorithm. Comput Geosci 10:191–203. doi:10.1016/0098-3004(84)90020-7
Boots B (2001) Using local statistics for boundary characterization. GeoJournal 53:339–345. doi:10.1023/A:1020106528639
Brown DG (1998) Classification and boundary vagueness in mapping presettlement forest types. Int J Geogr Inf Sci 12:105–129. doi:10.1080/136588198241914
Burrough P, Frank A (1996) Geographic objects with indetermined boundaries. Taylor & Francis, London
Cadenasso ML, Pickett STA (2000) Linking forest edge structure to edge function: mediation of herbivore damage. J Ecol 88:31–44. doi:10.1046/j.1365-2745.2000.00423.x
Cadenasso ML, Traynor MM, Pickett STA (1997) Functional location of forest edges: gradients of multiple physical factors. Can J For Res 27:774–782. doi:10.1139/cjfr-27-5-774
Cadenasso ML et al (2003) An interdisciplinary and synthetic approach to ecological boundaries. Bioscience 53:717–722. doi:10.1641/0006-3568(2003)053[0717:AIASAT]2.0.CO;2
Cairns DM, Waldron JD (2003) Sigmoid wave transitions at alpine treeline. Geogr Ann A 85:115–126. doi:10.1111/1468-0459.00193
Camarero JJ, Guti E (2002) A multivariate approach to the study of the spatial structure of treeline ecotones. Commun Ecol 3:9–18. doi:10.1556/ComEc.3.2002.1.2
Camarero JJ, Gutierrez E, Fortin MJ (2006) Spatial patterns of plant richness across treeline ecotones in the Pyrenees reveal different locations for richness and tree cover boundaries. Glob Ecol Biogeogr 15:182–191. doi:10.1111/j.1466-822X.2006.00211.x
Chen XX, Vierling L (2006) Spectral mixture analyses of hyperspectral data acquired using a tethered balloon. Rem Sens Environ 103:338–350. doi:10.1016/j.rse.2005.05.023
Choesin D, Boerner REJ (2002) Vegetation boundary detection: a comparison of two approaches applied to field data. Plant Ecol 158:85–95. doi:10.1023/A:1014720508155
Churkina G, Svirezhev Y (1995) Dynamics and forms of ecotone of under the impact of climatic change: mathematical approach. J Biogeogr 22:565–569. doi:10.2307/2845954
Clements FE (1905) Research methods in ecology. University Publishing Company, Lincoln
Dale MRT, Mah M (1998) The use of wavelets for spatial pattern analysis in ecology. J Veg Sci 9:805–814. doi:10.2307/3237046
Di Castri F, Hansen A, Holland MM (1988) A new look at ecotones: emerging international projects on landscape boundaries. Biol Int (Spec Iss) 17:1–163
Enserink M (1997) Life on the edge: rainforest margins may spawn species. Science 276:1791–1792. doi:10.1126/science.276.5320.1791
Ewers RM, Didham RK (2006) Continuous response functions for quantifying the strength of edge effects. J Appl Ecol 43:527–536. doi:10.1111/j.1365-2664.2006.01151.x
Fagan WF, Fortin M-J, Soykan C (2003) Integrating edge detection and dynamic modeling in quantitative analyses of ecological boundaries. Bioscience 53:730–738. doi:10.1641/0006-3568(2003)053[0730:IEDADM]2.0.CO;2
Fangju W, Hall GB (1996) Fuzzy representation of geographical boundaries in GIS. Geogr Inf Syst 10:573–590
Favier C, Chave J, Fabing A, Schwartz D, Dubois MA (2004) Modelling forest-savanna mosaic dynamics in man-influenced environments: effects of fire, climate and soil heterogeneity. Ecol Modell 171:85–102. doi:10.1016/j.ecolmodel.2003.07.003
Fernandez C, Acosta FJ, Abella G, Lopez F, Daz M (2002) Complex edge effect fields as additive processes in patches of ecological systems. Ecol Modell 149:273–283. doi:10.1016/S0304-3800(01)00464-1
Fisher P, Arnot C, Wadsworth R, Wellens J (2006) Detecting change in vague interpretations of landscapes. Ecol Inf 1:163–178. doi:10.1016/j.ecoinf.2006.02.002
Foody GM, Boyd DS (1999) Fuzzy mapping of tropical land cover along an environmental gradient from remotely sensed data with an artificial neural network. J Geogr Syst 1:23–35. doi:10.1007/s101090050003
Forman RTT (1995) Some general-principles of landscape and regional ecology. Landsc Ecol 10:133–142. doi:10.1007/BF00133027
Fortin MJ (1994) Edge detection algorithms for two-dimensional ecological data. Ecology 75:956–965. doi:10.2307/1939419
Fortin MJ (1997) Effects of data type on vegetation boundary delineation. Can J For Res 27:1851–1858. doi:10.1139/cjfr-27-11-1851
Fortin MJ (1999) Effects of quadrat size and data measurement on the detection of boundaries. J Veg Sci 10:43–50. doi:10.2307/3237159
Fortin MJ, Drapeau P (1995) Delineation of ecological boundaries—comparison of approaches and significance tests. Oikos 72:323–332. doi:10.2307/3546117
Fortin MJ, Olson RJ, Ferson S, Iverson L, Hunsaker C, Edwards G et al (2000) Issues related to the detection of boundaries. Landsc Ecol 15:453–466. doi:10.1023/A:1008194205292
Frochot B, Lobreau JP (1987) Etude quantitative de l’effet de la lisière sur les populations d’oiseaux: définitions et principes méthodologiques. Rev Ecol (Terre Vie) 4:7–15
Gauch HG, Whittaker RH, Wentworth TR (1977) A comparative study of reciprocal averaging and other ordination techniques. J Ecol 65:157–174. doi:10.2307/2259071
Goldblum D, Rigg LS (2005) Tree growth response to climate change at the deciduous-boreal forest ecotone, Ontario, Canada. Can J For Res 35:2709–2718. doi:10.1139/x05-185
Hansen A, di Castri F (1992) Landscape boundaries: consequences for biotic diversity and ecological flows. Springer, Berlin Heidelberg New York
Hargrove WW, Hoffman FM (1999) Using multivariate clustering to characterize ecoregion borders. Comput Sci Eng 1:18–25. doi:10.1109/5992.774837
Hay GJ, Marceau DJ, Dube P, Bouchard A (2001) A multiscale framework for landscape analysis: object-specific analysis and upscaling. Landsc Ecol 16:471–490. doi:10.1023/A:1013101931793
Hill MO, Gauch HGJ (1980) Detrended correspondence analysis: an improved ordination technique. Vegetatio 42:47–58. doi:10.1007/BF00048870
Hill RA, Granica K, Smith GM, Schardt M (2007) Representation of an alpine treeline ecotone in SPOT 5 HRG data. Remote Sens Environ 110:458–467. doi:10.1016/j.rse.2006.11.031
Hofgaard A, Wilmann B (2002) Plant distribution pattern across the forest-tundra ecotone: the importance of treeline position. Ecoscience 9:375–385
Holland MM, Risser PG, Naiman RJ (1991) Ecotones: the role of landscape boundaries in the management and restoration of changing environments. Chapman and Hall, New York
Huete A, Keita F, Thome K, Privette J, van Leeuwen W, Justice C et al (1999) A light aircraft radiometric package for MODLAND quick airborne looks (MQUALS). EOS Earth Obs 11:22–25
Hufkens K, Scheunders P, Ceulemans R (2008) Estimating the ecotone width in patchy ecotones using a sigmoid wave approach. Ecol Inf 3:97–104
Jacquez GM, Maruca S, Fortin MJ (2000) From fields to objects: a review of geographic boundary analysis. J Geogr Syst 2:221–241. doi:10.1007/PL00011456
Jia GJ, Epstein HE, Walker DA (2006) Spatial heterogeneity of tundra vegetation response to recent temperature changes. Glob Chang Biol 12:42–55. doi:10.1111/j.1365-2486.2005.01079.x
Johnston CA, Bonde JP (1989) Quantitative analysis of ecotones using a geographic information system. Photogramm Eng Remote Sensing 55:1643–1647
Johnston CA, Pastor J, Pinaym G (1992) Landscape boundaries: consequences for biotic diversity and ecological flows. In: Hansen A, Di Castri F (eds) Quantitative methods for studying landscape boundaries. Springer, Berlin Heidelberg New York
Kent M, Moyeed RA, Reid CL, Pakeman R, Weaver R (2006) Geostatistics, spatial rate of change analysis and boundary detection in plant ecology and biogeography. Prog Phys Geogr 30:201–231. doi:10.1191/0309133306pp477ra
Kohonen T (1982) Self-organized formation of topologically correct feature maps. Biol Cybern 43:59–69. doi:10.1007/BF00337288
Laurance WF, Didham RK, Power ME (2001) Ecological boundaries: a search for synthesis. Trends Ecol Evol 16:70–71. doi:10.1016/S0169-5347(00)02070-X
Lek S, Guegan J-F (2000) Artificial neural networks: application to ecology and evolution. Springer, Berlin Heidelberg New York
Li BL (2000) Fractal geometry applications in description and analysis of patch patterns and patch dynamics. Ecol Modell 132:33–50. doi:10.1016/S0304-3800(00)00303-3
Lillesand TM, Kiefer RW (1994) Remote sensing and image interpretation. Wiley, New York
Livingston BE (1903) The distribution of the upland societies of Kent country, Michigan. Bot Gaz 35:36–55. doi:10.1086/328315
Loehle C (2000) Forest ecotone response to climate change: sensitivity to temperature response functional forms. Can J For Res 30:1632–1645. doi:10.1139/cjfr-30-10-1632
Ludwig JA, Cornelius JM (1987) Locating discontinuities along ecological gradients. Ecology 68:448–450. doi:10.2307/1939277
Mariotti A, Peterschmitt E (1994) Forest savanna ecotone dynamics in India as revealed by carbon isotope ratios of soil organic matter. Oecologia 97:475–480. doi:10.1007/BF00325885
McIntire EJB, Fortin M-J (2006) Structure and function of wildfire and mountain pine beetle forest boundaries. Ecography 29:309–318. doi:10.1111/j.2006.0906-7590.04278.x
Metzger JP, Muller E (1996) Characterizing the complexity of landscape boundaries by remote sensing. Landsc Ecol 11:65–77. doi:10.1007/BF02093740
Naiman RJ, Decamps H (1997) The ecology of interfaces: riparian zones. Annu Rev Ecol Syst 28:621–658. doi:10.1146/annurev.ecolsys.28.1.621
Naiman RJ, Décamps H (1990) The ecology and management of aquatic-terrestrial ecotones. The Parthenon Publishing Group, Paris
Oden NL, Sokal RR, Fortin MJ, Goebl H (1993) Categorical wobbling—detecting regions of significant change in spatially located categorical variables. Geogr Anal 25:315–336
Odum EP (1971) Fundamentals of ecology. WB Saunders, Philadelphia
Orlóci L, Orlóci M (1990) Edge detection in vegetation: Jornada revisited. J Veg Sci 1:311–324. doi:10.2307/3235706
Pélissier R, Goreaud F (2001) A practical approach to the study of spatial structure in simple cases of heterogeneous vegetation. J Veg Sci 12:99–108. doi:10.2307/3236678
Pitas I (2000) Digital image processing algorithms and applications. Wiley, New York
Puyravaud J-P, Pascal J-P, Dufour C (1994) Ecotone structure as an indicator of changing forest-savanna boundaries (Linganamakki region, Southern India). J Biogeogr 21:581–593. doi:10.2307/2846033
Ranson KJ, Sun G, Kharuk VI, Kovacs K (2004) Assessing tundra-taiga boundary with multi-sensor satellite data. Remote Sens Environ 93:283–295. doi:10.1016/j.rse.2004.06.019
Ries L, Sisk TD (2004) A predictive model of edge effects. Ecology 85:2917–2926. doi:10.1890/03-8021
Risser PG (1993) Ecotones at local to regional scales from around the world. Ecol Appl 3:367–368. doi:10.2307/1941904
Saunders SC, Chen J, Drummer TD, Crow TR (1999) Modelling temperature gradients across edges over time in managed landscape. For Ecol Manage 117:17–31. doi:10.1016/S0378-1127(98)00468-X
Schilthuizen M (2000) Ecotone: speciation-prone. Trends Ecol Evol 15:130–131. doi:10.1016/S0169-5347(00)01839-5
Smith TB, Wayne RK, Girman DJ, Bruford MW (1997) A role for ecotones in generating rainforest biodiversity. Science 276:1855–1857. doi:10.1126/science.276.5320.1855
Stow DA, Hope A, McGuire D, Verbyla D, Gamon J, Huemmrich F et al (2004) Remote sensing of vegetation and land-cover change in Arctic Tundra Ecosystems. Remote Sens Environ 89:281–308. doi:10.1016/j.rse.2003.10.018
Strayer DL, Power ME, Fagan WF, Pickett STA, Belnap J (2003) A classification of ecological boundaries. Bioscience 53:723–729. doi:10.1641/0006-3568(2003)053[0723:ACOEB]2.0.CO;2
Tansley AG (1935) The use and abuse of vegetational concepts and terms. Ecology 16:284–307. doi:10.2307/1930070
Timoney KP, Laroi GH, Dale MRT (1993) Sub-arctic forest-tundra vegetation gradients—the sigmoid wave hypothesis. J Veg Sci 4:387–394. doi:10.2307/3235597
Turner MG (2005) Landscape ecology: what is the state of the science? Ann Rev Ecol Syst 36:319–344. doi:10.1146/annurev.ecolsys.36.102003.152614
van der Maarel E (1990) Ecotones and ecoclines are different. J Veg Sci 1:135–138. doi:10.2307/3236065
White MA, Asner GP, Nemani RR, Privette JL, Running SW (2000) Measuring fractional cover and leaf area index in arid ecosystems: digital camera, radiation transmittance, and laser altimetry methods. Remote Sens Environ 74:45–57. doi:10.1016/S0034-4257(00)00119-X
Whittaker RH (1967) Gradient analysis of vegetation. Biol Rev Camb Philos Soc 442:207–264. doi:10.1111/j.1469-185X.1967.tb01419.x
Wiens JA, Crawford CS, Gosz JR (1985) Boundary dynamics—a conceptual-framework for studying landscape ecosystems. Oikos 45:421–427. doi:10.2307/3565577
Womble WH (1951) Differential systematics. Science 114:315–322. doi:10.1126/science.114.2961.315
Wulder M, Boots B (1998) Local spatial autocorrelation characteristics of remotely sensed imagery assessed with the Getis statistic. Int J Remote Sens 19:2223–2231. doi:10.1080/014311698214983
Acknowledgments
Koen Hufkens holds an interdisciplinary PhD grant from the Research Council of the University of Antwerp (UA-BOF-2005).
Author information
Authors and Affiliations
Corresponding author
About this article
Cite this article
Hufkens, K., Scheunders, P. & Ceulemans, R. Ecotones in vegetation ecology: methodologies and definitions revisited. Ecol Res 24, 977–986 (2009). https://doi.org/10.1007/s11284-009-0584-7
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11284-009-0584-7