159
Landscape Ecology 18: 159–171, 2003. © 2003 Kluwer Academic Publishers. Printed in the Netherlands.
Research article
Land use, land cover changes and coastal lagoon surface reduction associated with urban growth in northwest Mexico Arturo Ruiz-Luna* and César A. Berlanga-Robles Research Centre for Food and Development (CIAD), A.C. Unidad Mazatlán, México; *Author for correspondence (e-mail:
[email protected]) Received 18 April 2001; accepted in revised form 14 January 2003
Key words: Coastal lagoons, Cumulative impact, LULC, Remote sensing, Urban development Abstract Coastal land use and land cover changes, emphasizing the alterations of coastal lagoons, were assessed in northwest Mexico using satellite imagery processing. Supervised classifications of a Landsat series (1973–1997) and the coefficients Kappa (K) and Tau (), were used to assess the area and verify the accuracy of the classification of six informational classes (urban area, aquatic systems, mangrove, agriculture, natural vegetation, and aquaculture). Pixel-by-pixel change detection among dates was evaluated using the Kappa Index of Agreement (KIA). Besides the overall estimation of the aquatic systems class, variations in the three lagoons present in the study area were analyzed individually. Measures of agreement between the classification and reference data indicate that the accuracy for the classification ranked from moderate to high (K = 0.76 ± 0.07; = 0.77 ± 0.06). From 1973 to 1997 urban area has doubled, growing to the north and the northeast, extending mainly over natural vegetation and agricultural land. La Escopama and El Sabalo, two of the lagoons studied, reduced their size to less than half that estimated in 1973, but the main estuarine system in the study area, Estero de Urias - El Infiernillo, has maintained its area without noticeable changes. However, the surrounding landscape in Estero de Urias - Infiernillo is changing from natural vegetation and agriculture to urban land use. Consequently, to limit as much as possible changes in the area to natural causes, some management measures must be considered to design urban development plans and to recover and preserve the natural areas, on a broad scale rather than a local spatial scale. Introduction Population increases, together with rural and urban activities, directly affect the availability and quality of natural resources, but also induce secondary effects that must be evaluated from a regional view. The human activities involve changes in the environment, sometimes expressed as modification at landscape levels. Agriculture, forestry, dams, harbors, industry, almost any activity developed on a significant scale, modifies the natural environment. The magnitude of these activities and their effects are related to urban growth, and therefore urban development must be seen as part of the ecological systems. The study of
urban areas as ecological systems is relatively new, and because it includes social and environmental interactions on different time and spatial scales, it could be considered a different approach for the evaluation of human influence on natural systems. The development of ecological models integrating the human dimension could lead to a better understanding and the needed mechanisms to solve environmental problems. Though this approach is desirable, it is not easy to achieve because the magnitude of change from urban growth is variable and appears to be a function of the local economy and the characteristics of the landscape (Grimm et al. 2000). Also, the extension of some urban areas is sometimes large enough to limit
160 the scope of field studies and consequently it is necessary to incorporate new tools and methods to study this. Remote sensing seems to be the most practical tool to detect quickly and at relatively low cost the transformations of the main landscape elements by means of multitemporal analysis. With remote sensing image datasets, it is possible to assess the major transformations and to evaluate the trends of land-use changes and handle them to outline adaptive management strategies (Bailly and Nowell 1996; O’Regan 1996; Green et al. 1996). Determination of this kind of criteria is essential for sustainable management of the landscape, to assess the regional carrying capacity, and to avoid irreversible degradation caused by misuse or abuse of the natural resources (van Mansvelt and van der Lubbe 1999). The characteristics and intensity of the economic activities associated with urban growth in the coastal zone can induce alterations not only on the hydrologic patterns, soil structure, and natural vegetation, but also cause cumulative effects on the coastline and coastal ecosystems and force the intensification of erosive and siltation processes and changes in the landscape (McCreary et al. 1992; Bailly and Nowell 1996; Bedford 1999; Ji et al. 2001; Jackson et al. 2001). The environmental stress normally increases when a series of activities as industry, fishing, tourism, and harbor are present at the same time, as in Mazatlan, northwest Mexico. This city is in a floodplain zone, almost at sea level (< 10 m), and is associated with estuarine water bodies that have been continually modified (INEGI 1993). During the last few decades, the estuarine systems Estero de Urias, Estero del Infiernillo, El Sabalo, and La Escopama, all of them linked to Mazatlan’s urban area, have been rapidly impacted by eutrophication from wastewater drainage and altered by natural drying, but the impact has been accelerated by deforestation and filling to gain land for urban use. Furthermore, El Camaron lagoon, located in the middle of the urban area, has practically disappeared and it is now occupied by residential and tourist infrastructure. These changes are inherent in the development of urban areas that modify natural covers, modify land uses, and alter the nature of the biogeochemical cycles and consequently the natural community structures (McCreary et al. 1992; Zalidis et al. 1997; Ji et al. 2001). These changes occur where urban development is present and are common in the coastal zone (where most of the world population is concentrated), but here, at the end of the basin drainage system,
these alterations can be magnified and they could be irreversible because the coastal zone receives the cumulative impact from higher land (Johnston 1994; McCold and Saulsbury 1996; Bedford 1999). Local authorities and social researchers have documented the major historical changes in the physiognomy of Mazatlan (Beraud 1996; Vega 1998), but these reports are merely descriptive, non-quantitative, and do not include updated records of these modifications. In addition, the available information is related to the urban history but does not include the interactions with the surrounding landscape, because depending on the area and nature of the analysis, evaluations in situ can be expensive. Considering this, the aims for the present study were: 1) to estimate, by means of standard satellite imagery analysis, how Mazatlan’s urban area has expanded from 1973 to 1997, and 2) to identify land cover and land-use changes in the adjacent landscape as consequence of this growth, with special attention on the estuarine systems surface.
Study area and remote sensing image dataset Mazatlan is a city on the Pacific coast of Mexico (23°12⬘ N; 106°25⬘ W) founded during the 16 th century. The natural conditions, characterized by the presence of wetlands such as the La Escopama, El Sabalo, and Urias system, limited a fast colonization and population was only 966 in 1765. This figure doubled by 1804 (Beraud 1996) and the population has now grown to reach more than 300,000 in 1995 (INEGI 1999). The port was one of the most important activities of Mazatlan, but the nature of the surrounding landscape and the physiography of the region (floodplains, lagoons, beaches) and the use of its natural resources (water bodies, wetlands, beaches, dry forest) have changed the city into a center of diverse economic activities. Now the economy of Mazatlan depends on fishing, aquaculture, agriculture, and tourism, probably the present major economic activity. All of these factors have the potential to transform the environment. Despite the diverse economic activities, Mazatlan is mainly urban with most of the employed population (63%) working in the commerce and services sector and just 12% dedicated to other activities (INEGI 1999). A considerable area of the original bay has been filled and land has been gained from the sea and wet-
161
Figure 1. Map contour of Mazatlan downtown in 1903 (black dashed line) overlaid on a 1997 Landsat TM scene, showing the areas gained from the sea for urban use, including connections to some islands. The 1903 map exhibits inaccuracies, but it is useful to show the magnitude of changes.
lands for urban uses. Road building has caused transformations in the coastal profile of Mazatlan, but most of these changes were done before the dates included in the present study (Figure 1). The area is located at the end of the Rio Presidio basin, with the river located 15 km southeast. It is mostly characterized by a warm-subhumid climate with an extensive dry season (December to June) and 80 to 100 mm average rainfall during the summer. In summer and the early fall, the coast is exposed to strong tropical
storms and hurricanes. The Rio Presidio basin covers around 3900 km 2 in the state, but the study area was isolated by masking some parts of the scenes, including the ocean, to give a final surface area of about 180 km 2, where the urban development has influence (Figure 2). Because of the climate of the region and to reduce the error caused by seasonal differences, the four Landsat scenes used here were acquired almost at the same time of the year, between February and May,
162
Figure 2. Study area. Landsat TM scene (path 31; row 44) acquired May 1997. Enclosed in squares are shown the aquatic systems analyzed in this study. Coordinates are in UTM units (zone UTM 13).
before the rainy season started. Three of these images were obtained with the Multispectral sensor (MSS) on 3 March 1973, 10 March 1986, and 19 April 1992. The fourth is a Thematic Mapper (TM) image from 27 May 1997. All the images are on path 31 and row
44 in the world reference system and are projected using the Universal Transverse Mercator system (zone UTM 13) on the Clarke 1866 ellipsoid.
163 Methods The image processing was achieved using MultiSpec v. 1.2 for Windows (Landgrebe and Biehl 1995) and Idrisi v. 2.0 for Windows (Eastman 1995). Because the MSS and TM images have substantial differences in radiometric and spectral resolution, the evaluation was done once the four images were classified (postclassificatory analysis). To make the map suitable by date overlay, the outputs were geometrically and geographically corrected to obtain the same pixel size and position in all the scenes. A linear mapping function and the nearest neighbor algorithm were used for the resampling (Eastman 1995). The images achieved a final size of 441 columns and 561 rows, with a pixel resolution of 60 m. To achieve the supervised classification, using maximum likelihood as the clustering algorithm, sets of training fields were digitized on-screen for each of the classes to be evaluated to define spectral signatures (Brondizio et al. 1994; Grignetti et al. 1997; Mas 1997; Ringrose et al. 1997). At least five clusters of pixels by class were digitized, with sample sizes of 30–50 pixels. All bands in the MSS and TM were used for the supervised classification, with exception of channel TM-6 (thermal infrared), whose spatial resolution (120 m) and spectral range (10.4 to 12 μm) make it useless for the purposes of this work. Six general informational classes were defined for the study; urban area (UA), aquatic system surface (AS), mangrove (MN), agriculture (AG), and natural vegetation (NV). Aquaculture (AQ), represented by shrimp farming, was only assessed for the 1997 TM image, because of the best spatial resolution (30 m × 30 m pixel) in that image. The class UA included bare soils inside the urban area and beach because their spectral responses are similar and it was not possible to discriminate between them. To avoid confusion with agricultural bare soils, a mask limiting the urban area was used for the evaluation. Similarly, AG includes soils used as grassland and NV comprised the dry forest and secondary succession, covers that frequently are under- or overestimated when they are assessed separately (Helmer et al. 2000). The coverage for these classes was estimated in km 2 for all the scenes after their classification, and rounded off to the nearest ten km 2. The results of the classification were validated with field and ancillary data consisting of stereoscopic aerial photography (1985, 1991, 1995), land use and vegetation cartography at a scale of 1:50,000,
videography (1997), ground truthing data from different studies done in the area (Ramírez 1998; Ruiz and Berlanga 1998), and several visits to the field during 1999–2000. The accuracy of the 1997 classification was estimated using three methods, all based on the error matrix or confusion matrix. This is a square array with the same classes in rows and columns, that compares results from the classification against reference data obtained from field work, maps, or other georeferenced data. The main diagonal values represent the agreements between the classification output and the reference data and they are the same as the total number of test data, when the agreement is total (Congalton and Green 1999). The first of the three methods, overall accuracy, is the simplest one, and it is obtained as the sum of the main diagonal number divided by the total number of test points in the matrix. Using the same matrix, the producer’s and user’s accuracy can be estimated for each category. Producer’s accuracy is calculated as the ratio between the number of sampling units correctly classified in a given class by the total sampling units assigned to the same class in the reference data. User’s accuracy is calculated similarly, but the correctly classified units are divided by the total number of units classified in the same category. Both values are ways of representing individual accuracies instead the overall accuracy (Congalton and Green 1999). Seen as a percentage, the overall accuracy is an approach to determine the accuracy level of the classification, but because it does not take into account the positive values caused by chance, this measure tends to overestimate the agreement of the classification. The accuracy of the classification was also assessed evaluating the Kappa (K) and Tau () coefficients, that are flexible indices used when agreement because of chance is included in the assessment. These indices are also evaluated from the confusion matrix, but by contrast with overall accuracy, they include in the calculations the off-diagonal elements (omission and commission errors) that are ignored by the overall accuracy index. Both indices give statistical elements to ensure the classification is not a product of chance and are easy to interpret (Ma and Redmon 1995; Congalton and Green 1999). It is also possible to obtain their variances and to produce confidence intervals, giving one the ability to compare differences among two classification outputs or different dates for the same scene. Once the validation was done, the resultant classifications were transformed from cells to area (km 2),
164 Table 1. Error matrix for a supervised classification of a scene from Landsat TM imagery (path 31; row 44). Classified data
Aquatic systems (AS) Urban area (UA) Mangrove (MN) Agriculture (AG) Natural vegetation (NV) Aquaculture (AQ) Column totals
Reference data AS
UA
MN
AG
NV
AQ
Row totals
Producer’s accuracy
User’s accuracy
27 3 2 0 0 0 32
1 53 6 0 4 0 64
1 1 32 1 0 1 36
0 3 6 24 1 1 35
1 2 2 0 22 0 27
0 0 2 0 1 8 11
30 62 50 25 28 10 205
0.84 0.83 0.89 0.69 0.81 0.73
0.90 0.85 0.64 0.96 0.79 0.80
Overall accuracy = 0.81 ± .05 (95% confidence interval)
and results were arranged as time-series (area by year) for each class. These data were used for successive analyses to evaluate the trends followed by each class and to detect possible associations among changes in the different covers. Also output maps were analyzed pixel by pixel, using a cross tabulation method, to evaluate changes among classes. Values for the Kappa Index of Agreement (KIA) were estimated using matrices of change between dates, to detect how different a classification was with respect to the previous date, as well as to quantify the proportion of change from one class to other (Congalton and Green 1999). Population data from official sources were used to connect these figures with the expansion of the urban area by using simple linear regression model (population vs. area). The main estuarine systems associated with Mazatlan, Laguna La Escopama, and Laguna El Sabalo were isolated to evaluate their area and changes of the associated landscape over time. The estuarine system Estero de Urias, including the Estero el Infiernillo, was also evaluated, but because it is open to the Gulf of California an artificial boundary was delineated between the ends of the two breakwaters in the main channel of the estuary. The area of these water bodies was determined assuming that differences in water levels are mostly caused by drying and siltation and not as consequence of differences in rainfall, because images were acquired during the dry season. There were some interannual differences, especially for the 1996–1997 rainy season that was higher than that in the previous years (Berlanga 1999).
Results A total of 205 test points were verified in the area and from these only 166 are displayed in the main diagonal of the error matrix for the 1997 scene (Table 1). As expected, the values obtained in the accuracy assessment for K and coefficients were lower than that obtained for overall accuracy, considering that the latter overestimates accuracy because it ignores off-diagonal elements. The value obtained for overall accuracy was 0.81 (81% agreement), while individually the best producer’s accuracy was obtained for the mangrove category (89%), and the worse for agriculture (69%). By contrast, those classes were the best and the worse classified when user’s accuracy was evaluated (agriculture 96%, mangrove 64%), however there is no relationship between those estimates. For the K and coefficients, both gave similar values, slightly higher for Tau, with a difference narrow enough (0.01) to be neglected. The 95% confidence intervals for both coefficients were K = 0.69– 0.83 (0.76 ± 0.07) and = 0.71–0.83 (0.77 ± 0.06). The values obtained represent agreements from moderate (0.40–0.80) to strong (> 0.80) with the reference data (Congalton and Green 1999). Using accuracy assessments, and also because not all the pixels agreed with one of the six informational classes (around 10% of the total), the results from the assisted classification show that for the selected fringe the dominant landscape at 1997 was agriculture (Table 2), though this class had an important decrease after an expansion period during 1973–1986. The UA class is the second largest in the recent thematic map, almost the same as AG, with the urban area now more than twice what it was in 1973. The natural vegetation is the most modified land cover, decreasing more than 50% during the studied period. The mangrove
165 Table 2. Estimation of land use coverage of the urban area of Mazatlan using Landsat image (MSS and TM) analyses from 1973 to 1997. (UA: Urban area, AS: Aquatic systems, MN: Mangrove, AG: Agriculture, NV: Natural vegetation, AQ: Aquaculture). Year
1973
1986
1992
1997
1973–86
1986–92
1992–97
1973–97
Class
km 2
km 2
km 2
km 2
(%)
(%)
(%)
(%)
UA AS MN AG NV AQ unclassified Total area
25.1 16.9 9.1 69.1 44.3 n.a. 13.2 177.6
36.7 15.5 7.0 72.5 37.3 n.a. 9.1 177.9
38.3 17.1 8.0 66.0 28.1 n.a. 20.5 178.0
54.5 17.6 7.7 59.9 21.0 1.7 15.1 177.6
45.9 −8.3 −23.0 4.9 −15.8
4.4 10.8 14.0 −8.9 −24.6
42.6 3.0 −3.7 −9.3 −25.1
117.3 4.6 −15.5 −13.3 −52.5
n.a. = non available data
coverage showed a small decline of slightly more than 15%. For the pixels that were not assigned a category, the maximum amount was seen in the 1992 classification (12% of the total area), followed by the 1997 (8.5%) and the 1973 classification (7.4%). Visual analysis of the output maps showed that pixels without assigned category were located close to interfaces between the defined classes, especially between the water-land interface. For the agriculture and aquatic systems, the initial period (1973–1986) did not have substantial variations in extent (less than 10% each). Mangrove and natural vegetation classes reduced their coverage 23% and 16% during the same period. After 1992, only the urban area was still growing whereas natural vegetation and agriculture decreased to their lowest levels in 1997. At the end of the study period (1992–1997), the urban area class gained almost 1.23 km 2 year −1, whereas aquatic systems class was relatively stable and mangrove, agriculture, and natural vegetation decreased about 0.06, 0.4, and 1.0 km 2 year −1. During 1973–1986, the urban area and mangrove classes had their largest relative changes among dates analyzed, and natural vegetation had the lowest (−16%). Urban areas grew around 46%, and there is a reduction of around 23% for the mangrove areas estimated in 1986, compared to the previous figures (Table 2). Natural vegetation showed a constant reduction, with the highest loss occurring during 1992–1997, when agriculture also displayed a large reduction in its area. For the possible relationships among temporal figures for expansion and reduction of the studied classes, the correlation coefficients were significant at ␣ = 0.05 (d.f. = 3) for natural vegetation with urban area (−0.94) and agriculture (0.83), and for agricul-
ture with aquatic systems (−0.90). The last has a high correlation value, but this relationship does not make sense because agriculture decrease does not promote aquatic surface enlargement. Changes in both covers could be related indirectly, but are independent, just showing inverse trends during the period of study. Cross tabulation analyses among classification outputs for 1973 and 1997 indicate a moderate agreement (Overall Kappa = 0.68), and values of KIA for the comparisons between one date and the next, ranged from 0.71 to 0.73. Also, the associated Kramer’s V correlation coefficient varied from 0.54 to 0.60 (Table 3), This means that major changes in the landscape occurred during 1973 to 1986, though disagreement rates were similar between dates (around 30%) during the 24-year period. The comparison between 1973 and 1997 outputs had an overall KIA value around 0.68, maintaining the aquatic systems class as the most stable the whole time (KIA = 0.71 ± 0.06), when the earliest of both images in data sets was the reference. By contrast, dry forest and secondary succession that are included in the natural vegetation class was the least constant class (KIA = 0.31 ± 0.06). When analyzing the matrices of change detection between dates, there is a consistent pattern. The aquatic systems surface evaluated in 1973 partially changed to urban areas, mangrove, and agriculture in further assessments, gaining together in 1997 around 25% from the aquatic systems surface originally evaluated for 1973 (Table 4). The classes natural vegetation, urban areas, and agriculture were closely related, showing constant changes among them, especially with losses of natural vegetation to agriculture and urban area classes. By 1986, almost 50% of the pixels classified as natural vegetation in 1973 changed
166 Table 3. Kappa Index of Agreement values for pairwise comparisons from Landsat imagery time series of the urban area of Mazatlan using Landsat images (MSS and TM) analyses from 1973 to 1997. (UA: Urban area, AS: Aquatic systems, MN: Mangrove, AG: Agriculture, NV: Natural vegetation, AQ: Aquaculture). Period Class
1973–1986
1986–1992
1992–1997
1973–1997
Mean
AS UA MN AG NV Overall Kappa Kramer’s V
0.69 0.58 0.42 0.53 0.39 0.71 0.57
0.8 0.6 0.5 0.5 0.3 0.7 0.6
0.69 0.74 0.51 0.52 0.29 0.73 0.58
0.6 0.6 0.4 0.4 0.2 0.6 0.5
0.71 0.64 0.49 0.50 0.31
Table 4. Cross tabulation matrix to assess Mazatlan’s landscape change between 1973 (reference data) and 1997, using classified LANDSAT scenes. In cells, number of pixels by cover category. Pixels belonging to background and ocean are not considered. 1973
1997 unclassified Aquatic systems (AS) Urban area (UA) Mangrove (MN) Agriculture (AG) Natural vegetation (NV) Aquaculture (AQ) TOTAL 1973
unclassified
AS
UA
MN
AG
NV
AQ
TOTAL 1997
372 571 1616 212 659 137 45 3612
308 3054 773 123 308 116 4 4686
584 395 4506 87 1082 215 81 6950
326 340 195 1129 433 79 28 2530
1643 258 5814 252 9003 2028 155 19153
940 276 2185 335 5106 3266 165 12273
0 0 0 0 0 0 0 0
4198 4900 15156 2138 16625 5843 478 49338
into agriculture (40%) and urban areas (8%). Similarly, agriculture changed to urban area around 17% of the estimated figure in 1973 and natural vegetation recovered 21% from this category, probably as secondary succession. At the end of the study period, urban area grew more than 100%, taking from agriculture and natural vegetation an equivalent area of 83% and 31% of the estimated urban surface in 1973. The expansion of the urban area (UA) has moved towards the north and northeast of the area, completely enclosing the lagoons El Sabalo and El Infiernillo, and there is an increase along the north axis of the Urias estuary. Figure 3 displays the overlapping of four thematic maps, derived from the satellite imagery analysis, representing the coverage of urban area to date. A polygon with the limits of Mazatlan’s urban area updated to 1999, which include some areas for future urban developments, is included in this figure to compare with the results of the classification. The urban expansion versus time displayed an average rate around 1.2 km 2 per year, with the high-
est rate observed between 1992 and 1997 (3.3 km 2 per year). When the urban expansion was associated with the population number (N), analyzing data extrapolated from the national censuses done by the Instituto Nacional de Estadística, Geografía, e Informática (INEGI) for 1970, 1980, 1990, and 1995, the data were fitted to a linear model by simple regression analysis (N = 6280UA + 56340; R 2 = 0.87). Even though expectations for population and urban growth are not the same as here (linear), trends are included for modeling or GIS inputs. The results indicate that population growth and the expansion of the urban area are highly correlated, despite population density not being homogeneous in the area. The most densely populated areas (150–300 hab/ha) are downtown and to the northeast of Mazatlan. The main expansion is projected to the north, which has been classified for tourist- residential and tourist-commercial land uses with lower population densities.
167
Figure 3. In gray scale, the growth of the Mazatlan urban area (1973–1997) obtained by Landsat imagery supervised classification. The overlaid black line is a 1999 digitizing of Mazatlan from an updated chart.
168 Table 5. Variation of the estimated surface for three aquatic systems associated with Mazatlan’s urban growth, northwest Mexico, from 1973 to 1997. Estimated aquatic system surface (km 2)
La Escopama El Sabalo Estero de Urias-Infiernillo
1973
1986
1992
1997
2.4 2.8 11.7
1.6 1.2 11.8
1.3 1.1 13.5
0.5 0.9 14.4
The numeric analysis does not reveal effects on the aquatic systems in the study area, and it seems that aquatic surface area has been stable or even grew around 5% over the 1973 estimated area. However, the visual analysis of the thematic maps allowed us to detect that some systems, especially the smaller and shallower, are heavily altered. Changes on the surface area of Urias-El Infiernillo system were relatively small compared with the other systems in the area that lost more than half their original area estimated in 1973 (Table 5). La Escopama decreased to about 1/5 of that in 1973; the 1973 estimate indicated a surface around 250 ha, in 1997 the surface is around 50 ha (Figure 4a). A similar process took place at Laguna El Sabalo, which reduced its surface from 275 ha to less than 95 ha in the study period. The water surface area decreased rapidly during 1973–1986, and after that the surface has stayed at about the same level because of the building of a marina that has a permanent opening to the sea (Figure 4b). The systems Estero de Urias and Estero el Infiernillo, considered as one system for the evaluation, had variations in the water surface area, slightly increasing the estimated area close to 5% (Figure 4c). More than an increase in the Urias system surface, the increase in the assessed area is a consequence of the improvement in technical characteristics of the sensor TM (spatial and spectral resolution), which allows a better classification.
Discussion and Conclusions The aims of the present study were focused on the estimation of Mazatlan’s urban area growth during a 24-year period using remote sensing techniques, and to relate this expansion with possible effects on the landscape, particularly with changes in land uses and loss of estuarine areas. It was possible to characterize Mazatlan’s associated landscape with an accuracy level that is categorized as moderate to high (Landis
and Koch 1977). User’s and Producer’s accuracies were mostly above 80% agreement, with some exemptions. Most of the confusion is associated with the classification processes that eliminate a portion (1%–5%) of pixels that do not match with any of the selected classes. Also, there is an error induced by the spectral likeness between the target classes, by misclassification of the interfacing pixels (i.e., pixels located between the edge of two or more different classes), and by the dominance of a land use or coverage (i.e., bare soils against low density mangrove shrubs). Finally, an unknown error results as consequence of land-use changes occurring between the last image acquisition date (1997) and the field work for test points (1999–2000). Despite that, the 1997 classification accuracy assessment is high enough to conclude that the output map is a reasonable description of the present landscape characteristics for the Mazatlan urban area and countryside, and validated the procedures for the subsequent classifications in this analysis. The variations detected in this study indicate the study area has changed from a landscape dominated by agriculture, with around 40% of the study area in 1973, to an agriculture-urbanized area. Natural vegetation appeared as the second largest coverage in 1973, but it is also the most rapidly transformed, especially for agriculture uses and later to urban areas. This is a common sequence where urban development occurs, with deforestation of the natural covers for agriculture uses, and then urban expansion and consumption of the arable land at rates that are areadependent (Ji et al. 2001). Natural coverages (aquatic systems, mangrove, and natural vegetation) represented around 40% of the total in 1973 and they decreased to around 25% of the total classified areas in 1997. Such a decrease has effect on the habitat availability for wild species, particularly for the terrestrial species, considering that mangrove and natural vegetation had a reduction from 30 to 16%, whereas
169
Figure 4. Changes in area of aquatic systems associated with Mazatlan’s urban area, assessed by remote sensing techniques for a time series of Landsat (MSS, TM), from 1973 to 1997. a) La Escopama. b) Laguna Sabalo. c) Estero de Urias-Infiernillo.
aquatic surface maintained its level without noticeable changes during the period of study. For the aquatic ecosystems included in this study, the main changes in the Urias-El Infiernillo estuarine system were made prior to the period covered by this study, and most of these were associated with growth of the Mazatlan urban area (Beraud 1996). There is no significant change in the extent of this system since 1973, and most of the transformations are related to changes in land use, especially increasing the adjacent urban area, and transforming natural areas to
agriculture and shrimp aquaculture. By contrast, the estuaries La Escopama and El Sabalo have been seriously modified and dried, but most of these changes were undetectable when the total water surface was considered because the area of the Estero de Urias dominates. Even assuming part of the difference is an effect of tides, rainfall, and water retention rates, the landscape around La Escopama is highly disturbed, mainly by the decrease of the natural vegetation around it and the increase of bare and eroded soils. A
170 similar process is observed in El Sabalo, with a critical reduction of the aquatic surface that finally has been stopped with the construction of a marina for residential-tourist use. Nevertheless, the transformation in this system is extreme because it is now surrounded by the urban area that limits the connection between some terrestrial and aquatic ecological processes that naturally occur there. De la Garza et al. (1985), reported an extension around 128 ha for this system, close to the extension found here, and they characterized pollution, obstruction of freshwater interchange and alterations of the lagoon profile as the main problems in this system. For the Estero de Urias – Infiernillo system, the analysis reveals that El Infiernillo has been severely transformed but tends to recover its area after some restoration promoted by local governmental and nongovernmental agencies. However its health and functionality, as well as that of the Urias system, have been seriously modified because of their close contact with human developments, as noted by several authors. Urban waste, runoff from a thermoelectrical power plant, organic waste from fishing industry and shrimp aquaculture, eutrophication, and anomalous heavy metal (Cd, Zn, Cu, Pb) concentrations have been reported for this system (Soto-Jimenez and Paez-Osuna 2001; OchoaIzaguirre et al. 2002), and several studies with biomonitors showed that the Estero de Urias – Infiernillo system has areas highly polluted, related to anthropogenic activities (Ruelas-Inzunza and PáezOsuna 2000; Mendez 2002). Present results on land use and land cover assessments around Mazatlan are indicative of reductions in habitat availability, but are not sufficient to evaluate damage or impact on other systems, even when there is evidence that opportunistic and human-commensal species benefit, but some others have a notable reduction in their population size, especially large mammals (Cuaron 2000). Clearcutting for future urbanization or for grassland and agricultural uses without development plans including buffer areas will cause landscape fragmentation, as has happened south of the study area (RuizLuna and Berlanga-Robles 1999). Because the Mazatlan urban area has been increasing, doubling its size in a 24-year period, it is probable that La Escopama could be dried and filled and finally eliminated, as has happened to other aquatic systems such as the lost Laguna Gaviotas now occupied for urban use (Vega 1998).
Consequently, to limit as much as possible the extinction of the aquatic systems in the area to natural causes, some management measures must be implemented, such as the creation of buffer areas, or as suggested by Luque (2000) for a national reserve in the USA, to consider ways to recover and preserve the natural vegetation because maintaining biodiversity, water quality, and aesthetic values are as important as providing services for human development. However, considering local aesthetic, self-interest or utilitarian standards, to design urban development plans can be enough to preserve part of natural areas, but not to protect the structure and functionality required to sustain and maintain the natural biodiversity. These areas should be promoted considering a regional, wide scale instead a local or limited scale, and taking into account ecological criteria, such as landscape fragmentation and other landscape indicators.
Acknowledgements This research has been funded by CONACYT (Ref. 28347B). We thank the North American Landscape Characterization program (NALC-NASA) that provided us with some of the Landsat images. Thanks to Dr. Ellis Glazier for editing the English-language text. Mention of trade names or commercial products does not constitute endorsement or recommendation for use by these agencies.
References Bailly B. and Nowell D. 1996. Techniques for monitoring coastal changes: a review and case study. Ocean and Coastal Management 32: 85–95. Bedford B.L. 1999. Cumulative effects on wetland landscapes: links to wetland restoration in the United States and southern Canada. Wetlands 19: 775–788. Beraud L.J.L. 1996. Historical actors of the urbanization of Mazatlan (in Spanish). DIFOCUR. Government of Sinaloa, Culiacan, Sinaloa, Mexico, 218 p. Berlanga R.C.A. 1999. Evaluation of the current conditions and landscape changes in the wetlands of the south coast of Sinaloa, Mexico: an approach with remote sensing data (in Spanish). MSc Thesis, National Autonomous University of Mexico (UNAM), México, 111 p. Brondizio E.S., Moran E.F., Mausel P. and Wu Y. 1994. Land use change in the Amazon estuary: Patterns of Caboclo settlement and landscape management. Human Ecology 22: 249–278.
171 Congalton R.G. and Green K. 1999. Assessing the Accuracy of Remotely Sensed Data: Principles and Practices. Lewis Publishers, Boca Raton, Florida, USA, 180 p. Cuaron A.D. 2000. Effects of land-cover changes on mammals in a neotropical region: a modeling approach. Conservation Biology 14: 1676–1692. De la Garza S.R., Macias M.A., Pérez P.R. and Sanchez M.M.A. 1985. Algunos aspectos ecológicos de la ictiofauna del estero ⬙El Sabalo⬙ con un estudio sobre la alimentación de Centropomus robalito Jordan y Gilbert. Esc. Ciencias del Mar. Professional Thesis, Universidad Autónoma de Sinaloa, México, 110 p. Eastman J.R. 1995. IDRISI for WINDOWS. User’s guide. IDRISI Production. Clark University, Worcester, Massachusetts, USA, 120 p. Green E.P., Mumby P.J., Edwards A.J. and Clark C.D. 1996. A review of remote sensing for the assessment and management of tropical coastal resources. Coastal Management 24: 1–40. Grignetti A., Salvatori R., Casacchia R. and Manes F. 1997. Mediterranean vegetation analysis by multi-temporal satellite sensor data. International Journal of Remote Sensing 18: 1307–1318. Grimm N.B., Grove J.M., Picket S.T.A. and Redman C.L. 2000. Integrated approaches to long-term studies of urban ecological systems. BioScience 50: 571–584. Helmer E.H., Brown S. and Cohen W.B. 2000. Mapping montane tropical forest successional stage and land use with multi-date Landsat imagery. International Journal of Remote Sensing 21: 2163–2183. INEGI 1993. Mazatlan, Sinaloa Estate. Municipality statistical book (in Spanish). INEGI, Mexico, 123 p. INEGI 1999. Statistical yearbook of the Estate of Sinaloa (in Spanish). INEGI, Mexico, 187 p. Jackson J.B.C., Kirby M.X., Berger W.H., Bjorndal K.A., Botsford L.W., Bourque B.J. et al. 2001. Historical overfishing and the recent collapse of coastal ecosystems. Science 293: 629–638. Ji C.Y., Liu Q., Sun D., Wang S., Lin P. and Li X. 2001. Monitoring urban expansion with remote sensing in China. International Journal of Remote Sensing 22: 1441–1455. Johnston C.A. 1994. Cumulative impacts to wetlands. Wetlands 14: 49–55. Landgrebe D. and Biehl L. 1995. An introduction to MultiSpec. Purdue University, West Lafayette, Indiana, USA, 85 p. Landis J. and Koch G. 1977. The measurement of observed agreement for categorical data. Biometrics 33: 159–174. Luque S.S. 2000. Evaluating temporal changes using Multi-Spectral Scanner and Thematic Mapper data on the landscape of a natural reserve: the New Jersey Pine Barrens, a case study. International Journal of Remote Sensing 21: 2589–2611. Ma Z. and Redmon R.L. 1995. Tau coefficients for accuracy assessment of classification of remote sensing data. Photogrammetric Engineering and Remote Sensing 61: 435–439. Mas J.F. 1997. Monitoring land-cover changes in the Terminos Lagoon region, Mexico: A comparison of changes detection techniques. In: Proceeding of IV International Conference on Remote Sensing for Marine and Coastal Environments, Orlando, Florida, 17–19 March 1997. Vol. 1. Environmental Research Institute of Michigan, Ann Arbor, Michigan, USA, pp. 159– 167.
McCold L.N. and Saulsbury J.W. 1996. Including past and present impacts in cumulative impact assessments. Environmental Management 20: 767–776. McCreary S., Twiss R., Warren B., White C., Huse S., Gardels K. et al. 1992. Land use change and impacts on the San Francisco estuary: A regional assessment with national policy implications. Coastal Management 20: 219–253. Mendez N. 2002. Annelid assemblages in soft bottoms subjected to human impact in the Urias Estuary (Sinaloa, Mexico). Oceanologica Acta 25: 139–147. Ochoa-Izaguirre M.J., Carballo J.L. and Páez-Osuna F. 2002. Qualitative chmnges in macroalgal assemblages under two contrasting climatic conditions in a subtropical estuary. Botanica Marina 45: 130–138. O’Regan P.T. 1996. The use of contemporary information technologies for coastal research and management – A review. Journal of Coastal Research 12: 192–204. Ramírez Z.J.R. 1998. Modeling environmental change trends in the estuary Urias, Mazatlan, Sin., through a multitemporal study with Landsat images (in Spanish). MSc Thesis, CIAD-Mazatlan Unit., Mexico, 85 p. Ringrose S., Vanderpost C. and Matheson W. 1997. Use of image processing and GIS techniques to determine the extent and possible causes of land management/fenceline induced degradation probles in the Okavango area, northern Botswana. International Journal of Remote Sensing 18: 2337–2364. Ruelas-Inzunza J.R. and Páez-Osuna F. 2000. Comparative bioavailability of trace metals using three filter-feeder organisms in a subtropical coastal environment (Southeast Gulf of California). Environmental Pollution 107: 437–444. Ruiz L.A. and Berlanga R.C.A. 1998. Coastal landscape change detection of Mazatlan, Sinaloa, associated with urban growth (in Spanish). In: Proceedings of the V Interamerican Conference on Environment. La Habana, Cuba, Digital version 5-040.doc. Ruiz-Luna A. and Berlanga-Robles C.A. 1999. Modifications in coverage patters and land use around the Huizache-Caimanero Lagoon System, sinaloa, México: A multi-temporal analysis using Landsat images. Estuarine Coastal and Shelf Science 49: 37–44. Soto-Jimenez M.F. and Paez-Osuna F. 2001. Dsitribution and normalization of heavy metal concentration in mangrove and lagoonal sediments from Mazatlan harbor (SE Gulf of California). Estuarine, Coastal and Shelf Science 53: 259–274. van Mansvelt J.D. and van der Lubbe M.J. 1999. Checklist for Sustainable Landscape Management. Elsevier, Amsterdam, The Netherlands, 181 p. Vega A.E. 1998. Un norte (in Spanish). In: Carrillo R.A. and Ibarra E.G. (eds), History of Mazatlan. Municipality of Mazatlan/Autonomous University of Sinaloa, Mexico, pp. 21–53. Zalidis G.C., Mantzavelas A.L. and Gourvelou E. 1997. Environmental impacts on Greek wetlands. Wetlands 17: 339–345.
