|
|
| Bildiri Özetleri | |
| Ana Sayfaya Dönüş |
| ISD Ana Sayfası |
|
Hakan Mete Doğan General Directorate of Agricultural Research, GIS & Remote Sensing Research Center, Ankara, Turkey ABSTRACT One of the most important problems in arid regions of world is desertification, which is exemplified by shrub invasion and grassland fragmentation. Identification of the spatio-temporal dynamics of such mechanisms is among the most difficult tasks related to range management decision making. In this study, statistical analyses and GIS visualization procedures are employed to display changing conditions on the Chihuahuan Desert Rangeland Research Center, New Mexico from 1982 to 1993. Results denote a proof of the notion that visualization is a substantial tool for monitoring range condition. They also indicated the dramatic mesquite (Prosopis glandulosa) change between the years (1982 and 1993) and some statistically significant relationships between the dependent and independent natural variables. INTRODUCTION Desertification is a critical environmental problem. According to the United Nations Environment Program (1987); about 3500 million hectares of land are under pressure of desertification. Every year about 6 million hectares of land are irretrievably lost to desertification, and an additional 21 million hectares are so degraded that agricultural facilities become uneconomic. Serious desertification affected 57 million people in rural populations in 1977, and this amount reached 135 million people in 1984. The conditions became crucial in the rainfed croplands in the year 2000. In the northern Chihuahuan desert, serious changes in vegetation have been observed during the last l00 years, where vegetation dynamics are determined by the interactions of topographic position, soil development, and human impact (Buffington and Herbel 1965). Consequently, large areas of former black grama (Bouteloúa eriopoda) grassland have been replaced by shrubland communities and dominated by creosotebush (Larrea tridentata), mesquite (Prosopis glandulosa) and tarbush (Flourensia cernua DC.). When arid lands are invaded by mesquite, desertification come into subject, and management of these kinds of areas become critical. Recently, geographic data that contain the measurements of three-dimensional space and time have gained great importance for the studies of spatial and temporal relationships in landscapes. Scientific visualization supports the analysis and communication of these kinds of data (Brown et al. 1995). Geographic Information Systems (GIS), the application of computer graphic visualization in 3-dimensions, is a tool that expands our understanding on a variety of spatial and temporal relationships by graphically visualizing spatial data (Watson 1992; Habb 1995). MATERIAL and METHOD Chihuahuan Desert lies south of the international border and extends into parts of New Mexico, Texas and even sections of southeastern Arizona. It is the largest desert in North America. The research area, located in 20 km north of Las Cruces-New Mexico, is called the Chihuahuan Desert Rangeland Research Center (CDRRC), and it was established by New Mexico University in 1927. CDRRC has a typical semidesert grassland climate with an abundance of sunshine and a wide range between day and night temperatures. Low relative humidity and extremely variable precipitation are the important characteristics of the climate. In winters, precipitation is mostly low intensity rain, and snow is rare. On the contrary, localised high intensity thunderstorms occur in summers. Although the temperature might be favorable for plant growth about 200 days, normal growth occurs only from 90 to 100 days because of available moisture conditions. It is clear that moisture is an important determinant of plant growth in the area. The average annual precipitation from 1915 to 1962 was determined as 231.14 mm at the Range Headquarters, and 52% of that amount, 120.19 mm, recorded between July 1 and September 30 (Buffington and Herbel 1965). The precipitation data, collected from 10 rain gauges between 1979 and 1996 (18-year period) in the CDRRC, were converted to a rain map and used for spatial analysis in this study. According to the Soil Maps of New Mexico, specifically for the Dona Ana County Area (Bulloch and Neher 1980), the study area basically consists of two soil groups, namely Berino-Bucklebar association (BJ) and Wink-Harrisburg association (WH). The soil map that shows soil groups and experimental pastures of the study area is given in Figure 1.  Black grama, mesa dropseed (Sporobolus flexuosus (Thurb.) Rydb.) and Aristida spp. growing on the sandy upland sites are the main grass species while tobosa (Hilaria mutica (Buckl.) Benth.) and burrograss constitute the main forage species in the lower areas on CDRRC. The main browse plant is fourwing saltbush (Atriplex canescens (Pursh) Nutt.), which normally occurs in association with mesquite (Prosopis juliflora (Swartz) DC. var. glandulosa (Torr.) Cockerell), creosotebush, and tarbush. Yucca and broom snakeweed are defined as other shrubby plants in the area (Buffington and Herbel 1965). A grazing study, applied to the rangeland in 1967, divided the pasture area into four different grazing management pastures (Figure 1) that were named: pasture 15 (1266.82 ha), 3W (670.13 ha), 3N (501.86 ha ), and 3S (499.59 ha). A year-long grazing management study was conducted in pasture 15 (control pasture), while late winter-early summer, summer, and fall grazing management schemes were applied in pasture 3N, 3S, and 3W, respectively. All of the pastures were formed according to vegetation types and condition classes (from excellent to poor condition). The studies were based on the design of a simple two-group experiment that was detailed by Spatz and Johnson (1989). According to the 7.5-Minute Series Orthophotomap of U.S. Geological Survey (1982) maps (Selden Canyon and Summerford Mountain) the elevation of the study area changes between 1317 and 1332 meters. The highest elevations (1326-1332 m) are observed in the east (Pasture 3N and Pasture 3S) and north sides (Pasture 3W) of the area. On the other hand, the lowest elevations (1323-1317 m) are observed in the northwest and southwest (Pasture 15) of the area. Mesquite dunes around the big plants are observed in some of the nothern parts. These formations that constituted some undulating topography among the big mesquite plants were not reflected in this topography map. The data that contain canopy, density and volume values of mesquite were collected from 220 fixed transects on the pastures in 1982 and 1993 (unpublished data, CDRRC). Each transect`s location was determined according to Universal Transverse Mercator (UTM) coordinates by using Global Positioning System (GPS). Each transect represents an area approximately 806 m2. Pasture 15 has the most transects (92), while pasture 3W has 48 transects. Pasture 3S and 3N have the same number of transects (40 transects for each). Volume values indicate total volume of all mesquite (m3) on the transects and density values represent the number of mesquite plants per hectare (10 000 m2). Canopy values explain how much land area was covered by mesquite plants, and they are shown as percentage (%) values. Canopy, density and volume variables were combined with the UTM coordinate data. These coordinate data were transformed into DBase data format as X and Y values so that Surfer software (Golden Software, 1990) can use them. Mesquite variables (canopy, density and volume) were defined as the Z values of this study. Consequently, the worksheet data files that consist of XYZ values were formed to develop grid files that are necessary for producing filled contour maps. Difference maps for canopy, density and volume were produced by subtracting the map values of 1982 from the map values of 1993, and they were put together in the same figure (Figure 2). The difference maps showed the pastures where most mesquite variables` change occurred. Statistical analyses formed an important part in this research. At the starting point, data characteristics were determined graphically and statistically. Histograms and boxplots were chosen to create the graphical representation of data. In addition, Levene and Shapiro-Wilks tests were employed to explore the data in a statistical way (Norušis 1993). SPSS (1993, 1997) and SAS (1985) softwares were used in all statistical analyses. Graphical and statistical results showed that mesquite data set is not normally distributed. Therefore, some appropriate transformations were needed for the data set to conduct further statistical analyses in a more dependable way. A ln(variable +1) transformation formula was applied to mesquite data, because the data set showed negative binomial data characteristics depending on the distribution, mean and variance patterns (Norušis 1993). After exploring data and making transformations, Two-Way-Anova (Analysis of Variance) statistical analysis was conducted to distinguish if the differences were statistically significant or not in the four pastures for the years 1982 and 1993. After conducting Two-Way-Anova, the Least Significant Difference (LSD) test was applied to establish confidence intervals and group the pastures. Moreover, a series of regression analyses were applied to understand the nature of the relationships between the dependent (canopy, density, volume) and independent (seasonal rain, elevation, soil) variables. Statistical analyses gave the chance to establish logical connections between the mesquite data and the maps that were produced by using exploratory data analyse (EDA) methods. The residuals from regressions were investigated and mapped to check the validity of regression equations in this study. DISCUSSION The filled contour maps delineated the mesquite variables` change between the years (1982 and 1993) and among the pastures. Consequently, a lot of differences that are impossible to be shown in otherwise were detected and displayed for each mesquite variable (canopy, density and volume). By using these visual tools, the nature and characteristics of mesquite establishment and invasion were understood very well in the study area. This research emphasized the importance of GIS tools to understand the spatio-temporal characteristics of geographical events. All of the maps were interpreted and the mesquite invasion characteristics were determined. The first important mesquite invasion characteristic was detected for mesquite density increases (Figure 2). Maps indicated that mesquite establishment, specifically mesquite density increases, took place around the areas where bigger volume and canopy values were observed. Bigger and mature plants play a nucleus role for mesquite establishment in this area. This finding showed parallelism with Archer and Scifres` (1988) study. Moreover, the areas with the most changes were determined. The biggest changes were observed for the north and south side of the area where again bigger and maturer plants are located (Figure 2).  The two-way-Anova (Analysis of Variance) indicated that the differences between the years and among the pastures were statistically significant (P<0.05) for all of the mesquite variables in question. Depending on the visual displays and Two-Way-Anova statistical analysis, it might be said that mesquite density, canopy and volume variables increased significantly from 1982 to 1993. The pastures where different grazing methods were applied, have different mesquite variable characteristics. On the other hand, it is not so easy to answer why there was a dramatic mesquite increase in the north (pasture 3W and 15) and south (pasture 15 and 3S) parts of the study area, because in an ecosystem there are many factors interacting with each other, and different grazing methods might be one of these factors. Moreover, maps implied that most mesquite invasion takes place around core areas where mesquite plants are already established and big volume values were observed regardless of the grazing methods (Figure 2). Therefore, the different grazing methods may not be a factor to explain this kind of mesquite invasion. Seasonal rain is one of the key factors in Chihuahuan Desert Environment. Although the effect of seasonal rain on the mesquite invasion is still widely debated in the literature, the mesquite density variable increased with the higher seasonal rain amounts (positive relationship) in pasture 3N, 3W and 15. Regression lines that have different slopes delineated the direction and strength of relationships between the seasonal rain and density. Pasture 3S showed a distinct character compared to the other three pastures because no significant relationship between the mesquite density and seasonal rain was detected within this pasture. The question as why there is not a significant relationship between the density and seasonal rain only in pasture 3S is worth of attention. Pasture 3S has smaller plants in both years, and it receives less rain relative to the other pastures. This could support the assumption that most mesquite density increases take place around core areas where mesquite plants are already established, and larger plants were observed. The results also might indicate the importance of the effect of seasonal rain on mesquite density increases. Mesquite canopy and volume variables accelerated with increasing values of seasonal rain in all of the pastures with the exception of pasture 3S. This situation might be related to the size and age of plants, infiltration and nutrient characteristics of the soil, and some drought periods that are hidden in the mean seasonal rain values. Pasture 3S has relatively smaller plants compared to the other three pastures, and receives less seasonal rain. This characteristic of the pasture 3S answers the question partially but it is not enough to explain the whole process. More data that provide detailed information about rain, soil and infiltration characteristics are necessary. The relationships between the elevation and other dependent mesquite variables showed positive characters in pasture 3W that receives higher seasonal rain compared to the other pastures. When the elevation increases, mesquite canopy, density and volume variables also increase with different regression line slopes in pasture 3W. On the other hand, the relationships between the elevation and same dependent mesquite variables displayed negative (inverse) characters in pasture 3N and 3S that receive less seasonal rainfall within the study area. When the elevation increases, mesquite variables decrease with different slopes in pasture 3N and pasture 3S. Results imply that the elevation effect on the mesquite variables might be changed with the amount of seasonal rain. Any other variables such as some soil and rain characteristics might be important in this process. More detailed data about the rain are needed for further research. Soil was one of the most important independent variables in this study. Maps from the Soil Survey of Dona Ana County Area (Bulloch and Neher 1980) were used to investigate the soil effects on the mesquite variables. Although these soil maps were helpful to catch some important relationships for pasture 15, they did not supply enough information about pasture 3N, 3S and 3W because all of the pastures located on one type of soil except pasture 15 (Figure 1). The effects of soil and seasonal rain on mesquite variables were investigated by combining their effects for pasture 15. Some interesting relationships were determined in pasture 15. Regression lines showed the different soil responses of mesquite variables to the seasonal rain. Two relationships were observed ; (1) the relationship between the seasonal rain and canopy (1982) and (2) the relationship between the seasonal rain and density (1993). The regression graphs displayed differences between the two soil types for both relationships (Figure 3). In both graphs the slopes of regression lines that belong to Berino-Bucklebar (BJ) soil association were found higher comparing to the lines of Wink-Harrisburg (WH) soil association.  According to these findings it might be suggested that the response to seasonal rain is higher in BJ soil association. Why the response of BJ soil found higher in both graphs was investigated by using the available soil information. The first important difference between two soil associations was identified in the soil texture classes. The BJ soil association contains finer soil particles such as clay whereas no clay particles were described in the texture classes of WH soil association in any depth. As a result, higher available water capacity values were observed for BJ soil association, and this is an extremely important soil variable that is related to plant growth and its establishment in arid and semiarid environments. The higher response of BJ might come from its higher clay contents and available water capacity. These findings highlighted the importance of soil types in mesquite establishment and invasion of the study area. Results showed that, soil might be one of the most effective factors on the mesquite invasion, and the soil map of Soil Survey of Dona Ana County did not supply enough information to understand the whole process in this area. A more detailed soil map that delineates some important physical and chemical soil characteristics may be very useful for future research. The most important variables for the future soil maps might be soil texture, water holding capacity, infiltration characteristics and nutrient contents (Nitrogen, Phosphorus, and Potassium). Although rain data produced some reasonable results, the information about the rain intensity and duration might be necessary to understand the small and younger mesquite plants` response to drought. Results also indicate that topography might be effective on the mesquite invasion in the study area. A detailed contour map (1 or 2 feet intervals) might be very useful to display micro-topography and the relationships between the elevation and other dependent mesquite variables. In addition, It might be important to know the other vegetation types and their distributions in this area. A vegetation map may be useful to search the reasons of differences between the dependent and independent variables. The key factor for the further research activities about the mesquite invasion might be hidden in the geomorphology. REFERENCES Archer S. and C.J. Scifres. 1988. Autogenic Succession and the Physiognomic Conversion of a Grassland to a Woodland. In La Copita Research Area Consolidated Progress Report-1988, J.W. Stuth. Texas Agricultural Experiment Station, CPR?4592, pp 2. Brown, W. M., M. Astley, T. Baker, H. Mitasova. 1995. GRASS as an Integrated GIS and Visualization System for Spatio?Temporal Modeling. Twelfth International Symposium on Computer? Assisted Cartography, Charlotte, North Carolina Volume: 4 pp. 89?99. Buffington, L. C. and C. H. Herbel. 1965. Vegetational Changes on a Semidesert Grassland from 1858 to 1963. Ecological Monographs. 35:139-164. Bulloch, H. E. Jr. and R. E. Neher. 1980. Soil Survey of Dona Ana County Area New Mexico, United States Department of Agriculture, Soil Conservation Service, in cooperation with United States Department of Interior, Bureau of Management, and New Mexico Agricultural Experiment Station. United States Department of Agriculture, Washington D.C. Golden Software. 1990. Surfer Users` Guide. Golden, Co. Habb, F. J. 1995. Computer Generation and Rendering of Terrain. Master`s of Science Thesis, University of Neveda, Las Vegas, Nv. Norušis, M. J. 1993. SPSS for Windows Base System User`s Guide, Release 6.0. SPSS Inc. Chicago. SAS. 1985. SAS User`s Guide: Basics, Version 5 Edition. SAS Institute Inc., North Carolina. SPSS. 1993. SPSS 6.0 for Windows. SPSS Inc., Chicago. SPSS. 1997. SYSTAT 7.0 for Windows: Unparalleled Research Quality Statistics and Graphics. SPSS Inc., Chicago. Spatz, C. and J. O. Johnson. 1989. Basic Statistics. Tales of Distributions. 4ed. Brooks/Cole. Pasific Grove, Ca. UNEP, 1987. Sands of Change. UNEP Environmental Brief No 2. Watson, D.F. 1992. Contouring. A Guide to the Analysis and Display of Spatial Data. Pergamon Press. Oxford, England. |