• No se han encontrado resultados

Diagnostico de la calidad de agua para cultivar tilapia

3.1.2 Importancia ........................................ ¡Error! Marcador no definido.1

3.1.10.2 Diagnostico de la calidad de agua para cultivar tilapia

Landscapes are complex and heterogeneous land areas containing patterns formed by different forms of disturbances (Forman and Godron 1986; Linke et al. 2007). Specifically, a forested landscape often changes in response to different elements, including fire disturbance, insect infestation, global changes in climate, and human activity (Perera and Euler 2000). The most apparent process and change in a forested landscape is disturbances from wildfire. This has been responsible for the formation of heterogeneous elements within a landscape. Spatial heterogeneity in a landscape has a close relation with stability and biodiversity where high heterogeneous landscape encourages interactions (Duning and Xiezhen 1999).

Wildfire is a major natural disturbance and an important factor that shapes the landscape structure in the boreal forests. Fires in boreal forests are often intense and frequent (Johnson 1995; Cui et al. 2009) and consume substantial forest cover (Perera et al. 2009b), but do not burn the entire landscape (Whelan 1995; Johnson et al. 1998; Leduc et al. 2007). Owing to the variations in weather and site conditions (e.g., vegetation, topography, and natural firebreaks) (Rochadi et al. 1999; Perera et al. 2007), forest fire shapes the patterns of forest structure (Agee 1998; Linke et al. 2007; Hely et al. 2010) and creates a complex and heterogeneous landscape mosaic comprising patches of different size, age, shape, and tree species compositions (Turner

34

1989; Diaz-Delgado et al. 2004; van Wagtendonk 2004; Mermoz et al. 2005; Madoui et al. 2010; Vinatier et al. 2010). The spatial patterns of post-fire landscape structure (e.g., forest land cover) are useful to understand various ecological processes such as species dynamics and fire

disturbances. The patterns also have direct implications for various aspects of forested landscapes: economic values (i.e., selection of sites for harvesting), social concerns (i.e., conservation of wilderness) (Thompson 2000) and ecological values (i.e., habitat for various organisms). Understanding the patterns of post-fire residual structure helps forest managers to determine the structural elements that should be retained to emulate fire disturbances and preserve the biological diversity of the ecosystems.

One particular way of understanding fire disturbances and their effects is assessing the patterns of landscape structure following a fire. Wildfires affect the physical landscape structure, age class distributions, ecotones, and positions of forest boundaries (Weber and Flannigan 1997). This study focused only on forest landscape structure, referring to the pattern of a landscape that is determined by its type of use and its structure (i.e., size, shape, arrangement, and distribution of landscape elements (patches, corridors, and matrix) (Walz 2011). In this study, the term landscape structure refers to the patterns of post-fire residual patches, specifically the composition, arrangement, and the resulting spatial relationships among individual patches.

2.1.1. Post-fire residual patches

Wildfire is one of the main natural disturbances consuming substantial forest cover, influencing and reshaping the landscape mosaic of boreal forests (Madoui et al. 2010). One of the characteristic features of wildfires is the existence of unburned areas within a fire-disturbed landscape, which are referred to as post-fire residuals. The presence of a residual patch is due to different geo-environmental factors that interactively affect fire behaviour and the resulting

patterns of post-fire landscapes. The term residual patch is broadly defined as remnants of the pre-fire forest ecosystems that have retained their structure and were not entirely reduced to ash or charcoal during the fire. In this study, the term residual patch is used to describe remnants of the pre-fire forest ecosystems (i.e., live undisturbed vegetation patches) that are not physical connected to the footprint perimeter; this is regardless of size, age, and species composition of the patches. For detailed description on the types and meanings of different patches and fire footprint, please refer to (§1.3).

35

2.1.2. Landscape pattern metrics (LPM) for spatial pattern analysis

Understanding the patterns of residual patches plays an important role in inferring ecological processes such as fire disturbances and species dynamics (Griffith 2004; Mermoz et al. 2005; Vinatier et al. 2010); this has been central to the study of landscape ecology (Diaz- Delgado et al. 2004). In dealing with landscape ecology, the basic characteristics of a landscape (i.e., structure, function, and change) should be understood (Forman and Godron 1986; Turner 1989). Landscape structure has been used extensively in the landscape ecological literature, primarily to describe both landscape composition and configuration (Gustafson 1998; Linke et al. 2007). A landscape’s composition is described by the number of categories and amount of different spatial elements within a landscape but without being spatially explicit (McGarigal et al. 2002; Remmel and Csillag 2003; Linke et al. 2007). Landscape configuration refers to the physical distribution of patches within the landscape (McGarigal and Marks 1995; Remmel and Csillag 2003; Griffith 2004; Cifaldi et al. 2004; Lin et al. 2010). In order to understand the interaction between spatial patterns and process, the spatial heterogeneity of a landscape must be identified and quantified in meaningful ways (Turner 1989; Wu et al. 2000; Blaschke et al. 2002).

One of the characteristic features of a wildfire is the tendency to generate important biological diversity, which is used to describe the degree of heterogeneity in ecosystem structure and composition (Burton et al. 2008). One particular way of addressing the spatial heterogeneity in a landscape is by computing series of landscape pattern metrics (LPM) (Turner 1989; Corry and Lafortezza 2007); hence, an emphasis has been placed on developing methods to quantify landscape structure. LPM refer to indices obtained from categorical maps, and are focused on the characterization of the geometric and spatial properties of landscape patterns (McGarigal et al. 2002). The metrics have been widely used to characterize spatial heterogeneity, infer ecological processes (e.g., forest disturbances and species dynamics) (Riitters et al. 1995; Forman and Godron 1986; Griffith 2004; Lin et al. 2010).

Boreal forest fires involve factors and processes operating at different scales (King and Perera 2006) and thus the resulting patterns and variabilities can be studied on a wide range of scales (Pickett et al. 1999). However, there is considerable uncertainty regarding the appropriate scale at which measurements and analyses are undertaken (Griffith 2004; Cifaldi et al. 2007; Linke et al. 2007). LPM used to measure landscape structure relies on digital spatial data; yet the characteristics of the data are constantly changing depending on how scale is defined (Turner 1989; Corry and Lafortezza 2007) and grain sizes are aggregated (He et al. 2002). Additionally, spatial patterns manifest as processes operate over multiple spatial scales (Turner 1989;

36

not apply to another (Perveen and James 2010). Because of this multiplicity, scale holds the key to understanding pattern-process interactions; this has led to the hierarchical perspective in landscape ecology. The choice of an appropriate scale depends on the analysis method used to extract information about the phenomena and specific research objectives investigated

(Woodcock and Strahler 1987). The relationship between patterns and scale has also been an integral component landscape ecology (Wu and Li 2006), and as a result the definition of scale has to be well established.

2.1.3. Scale and its importance for pattern analyses

Various researchers have approached the issue of scale and scaling from related but different perspectives. In landscape ecology, for example, the scaling of patterns and processes is often addressed by considering multiple scales at which spatial pattern analyses are

undertaken (e.g., Benson and MacKenzie 1995; Moody and Woodcock 1995; Wu et al. 2002; Zhu et al. 2006). Scale in landscape ecology involves both grain and extent, which are related to the spatial resolution of a given study area and area of coverage respectively. To understand the scale effect, the spatial patterns over multiple scales should be studied and hierarchical linkages among them should be established using scaling approaches. There are two approaches to multi-scale analyses: 1) the direct method that uses inherently multiple scale approaches, and 2) the indirect multi-scale method that uses single-scale methods repeatedly at different scales (Wu et al. 2000). The direct methods contain multiple-scale components in their mathematical formulation or procedures, and thus are either hierarchical or multiple-scaled (Wu et al. 2000). Some of the direct methods used in landscape ecology include wavelet analysis, lacunarity analysis, and spectral analysis. The indirect approach on the other hand can use methods that are designed for single-scale analysis, such as the wide variety of landscape metrics (e.g., shape and area related metrics) as well as statistical measures (mean, variance, correlation, and regression coefficient). The most common approach to study the scale effect issue of scale, and implemented in this study, is the indirect approach, (Wiens 1989). The indirect methods was applied because it allows one to compute the various aspects of spatial patterns (composition, configuration, and fragmentation), and compare the LPM over multiple scales.

2.1.4. Research framework

Several studies have described the spatial patterns of natural fires (Diaz-Delgado et al. 2004; van Wagtendonk 2004; Mermoz et al. 2005; Collins et al. 2007; Meddens et al. 2008; Hely et al. 2010; Dragotescu and Kneeshaw 2012). Despite their importance for understanding fire

37

disturbances and species dynamics, there are relatively few studies undertaken to characterize the spatial patterns of residual patches and their spatial distribution within a disturbed landscape (Schmiegelow et al. 2006; Madoui et al. 2010). Moreover, previous studies examined the effect of scale change on measures of spatial structure (e.g., Turner 1989; Wiens 1989; Benson and MacKenzie 1995; Gustafson 1998; Nikora et al. 1999; Kok and Veldkamp 2001; Wu et al. 2002; Zhu et al. 2006; Haire and McGarigal 2009). The studies revealed that the LPM are sensitive to the changes in grain size, but the response of LPM to changing grain size varies depending on the dataset and aggregation techniques applied. In spite of this, our understanding of the interactions between scale and landscape pattern is limited; hence landscape pattern quantification remains an important issue for investigation and more analyses are needed to characterize patterns over multiple scales (Leduc et al. 2007). Moreover, little is known about the effects of scale on parameters that characterize the spatial structure for fine spatial resolution data (i.e., less than 30 m); yet this spatial resolution remains useful for detecting spatial structure (Corry and Lafortezza 2007).

The objectives of this chapter are to: 1) characterize the spatial patterns of post-fire residual vegetation patches, 2) examine the sensitivity of the metrics to changing grain size, 3) identify some general rules for comparing LPM obtained at different scales (i.e., establishing scaling rules), 4) assess the impact of land cover types on residual patch occurrence (i.e., are particular land cover types more likely to generate residual patches?), and 5) evaluate the spatial association of natural firebreaks (surface water) and fire perimeter with the occurrence of residual patches. To achieve these objectives, eleven spatial metrics (derived at landscape and class level) were analysed, across five spatial resolutions R4, R8, R16, R32, and R64.

In this chapter, I hypothesized that 1) the metrics that characterize the patterns of post- fire landscape structure would not be consistent across the fire events as the fire size and fire intensity would vary across the fire events; 2) the pattern and characteristics of residual patches, with respect to size, shape, and composition is sensitive to scale change; 3) the occurrence of residual patches is explained with respect to certain non-burnable areas, and 4) a tendency exists for residual patches to be concentrated near surface water. By addressing these, the study serves as a proxy for understanding the range of variability for fire in forest ecosystems (Collins et al. 2007). This study also provides useful information concerning the status and dynamics of boreal forests, and elements for long-term benchmark monitoring and conservation related to disturbances.

38