Recent changes in landscape-dynamics trends in tropical highlands, Central Mexico.

Carlos Arredondo-León, Julio Muñoz-Jiménez and Arturo García-Romero

Carlos Arredondo-León. Architect, Universidad de Michoacana, Mexico. M.S. in Landscape Architecture, Universidad de Baja California, Mexico. Doctoral candidate in Geography, Universidad Nacional Autónoma de México (UNAM). Mexico.

Julio Muñoz-Jiménez. Doctor in Geography, Universidad Complutense de Madrid (UCM), Spain. Professor, UCM, Spain.

Arturo García-Romero. Doctor in Geography, UCM, Spain. Researcher, UNAM, Mexico. Address: Instituto de Geografía. Circuito Exterior s/n, Ciudad Universitaria, México, C.P. 04510. e-mail: agromero@igg.unam.mx 

SUMMARY

Recent trends in landscape dynamics were investigated in various forest ecosystems along the Tuxpan river basin (1887.5km²; 620-3640masl) in the State of Michoacan, Mexico. Landscape maps were developed from the interpretation, in GIS-ILWIS ver. 3.0, of land covers corresponding to four different dates (1976-2000), and the typology was based on origin vegetation, physiognomic development and permanence of the land-use disturbance. Maps were cross-related and calculations were derived on surfaces, percent change and mean annual transformation rate, and transition matrices were obtained so as to identify the main change processes. In order to explain the distribution of landscape-dynamics patterns, a landscape geography method was applied using relief units and potential vegetation as the bases for determining landscape systems. The study area comprises three landscape-dynamics patterns: i. fir forest in high volcanic peaks (>3100masl), highly preserved but showing a trend towards an increase in traditional agriculture and deforestation; ii. pine-oak forests in volcanic slopes and peaks (1900-3100masl) and tropical dry forest on sedimentary basement (<1700masl) showing a high deforestation and heavy use, with a trend towards abandonment of crops, expansion of shrublands/grasslands and scarce forest regeneration; and iii. pine-oak forest in low volcanic slopes (1700-2600) with the highest inherited modifications and a trend towards land-use intensification in human settlements and forest plantations.

Cambios recientes de la dinámica del paisaje en tierras altas del trópico, Centro de México.

RESUMEN

Se estudiaron las tendencias recientes de la dinámica del paisaje en distintos ecosistemas forestales de la cuenca del río Tuxpan (1887,5km² y 620-3640msnm), en el estado de Michoacán, México. Los mapas de paisajes se obtuvieron a partir de la interpretación, en SIG-ILWIS ver. 3.0, de las coberturas del suelo en cuatro fechas (1976-2000). Para la tipología se consideró el origen, desarrollo fisonómico de la vegetación y permanencia del disturbio asociado al uso del suelo. Los mapas se cruzaron entre sí, se calcularon superficies, porcentajes de cambio e índices de transformación media anual y se elaboraron matrices de transición para identificar los principales procesos de cambio. Para explicar la distribución de la dinámica del paisaje se aplicó un método de la geografía del paisaje que utiliza las unidades del relieve y la vegetación potencial para determinar sistemas de paisajes. El área comprende 3 patrones de dinámica del paisaje: i. abetal de altas cumbres volcánicas (>3100msnm) con alta conservación, pero tendencia al incremento del cultivo tradicional y la deforestación; ii. bosques de pino-encino de cumbres y de laderas volcánicas altas (1900-3100) y selva baja caducifolia del basamento sedimentario (<1700) con alta deforestación e intensificación del uso, pero tendencia al abandono del cultivo y expansión de la vegetación secundaria; y iii. bosque de pino-encino de laderas volcánicas bajas (1700-2600) con la mayor transformación heredada y tendencia a la intensificación del uso del suelo en asentamientos humanos y plantaciones forestales.

Mudanças recentes nas tendências da dinâmica de paisagens do México Central.

RESUMO

Estudaram-se as tendências recentes da dinâmica da paisagem em distintos ecossistemas florestais da bacia do rio Tuxpan (1887,5km² e 620-3640msnm), no estado de Michoacán, México. Os mapas de paisagens se obtiveram a partir da interpretação, em SIG-ILWIS ver. 3.0, das coberturas do solo em quatro datas (1976-2000). Para a tipologia se considerou a origem, o desenvolvimento fisionômico da vegetação e a permanência do distúrbio associado ao uso do solo. Os mapas se cruzaram entre si, se calcularam superfícies, porcentagens de mudança e índices de transformação média anual; foram elaboradas matrizes de transição para identificar os principais processos de mudança. Para explicar a distribuição da dinâmica da paisagem se aplicou um método da geografia da paisagem que utiliza as unidades de relevo e a vegetação potencial para determinar sistemas de paisagens. A área compreende três padrões de dinâmica da paisagem: i. abetal de altos cumes vulcânicos (>3100msnm) com alta conservação, mas tendência ao incremento do cultivo tradicional e ao desflorestamento; ii. bosques de pinho-encino de cumes e de ladeiras vulcânicas altas (1900-3100) e selva baixa caducifólia do basamento sedimentário (<1700) com alto desflorestamento e intensificação do uso, mas tendência ao abandono do cultivo e expansão da vegetação secundária; e iii. bosque de pinho-encino de ladeiras vulcânicas baixas (1700-2600) com a maior transformação herdada e tendência à intensificação do uso do solo em assentamentos humanos e plantações florestais.

KEYWORDS / Land Use / Land-Cover Dynamics / Landscape / Mexico / Temperate Forests/

Received: 08/13/2007. Modified: 07/04/2008. Accepted: 07/07/2008.

Landscape-dynamics do not take place homogeneously across a given area, but there are changes in the balance between negative and positive processes (Forman and Godron, 1986; Serrão et al., 1996; Bastian and Röder, 1998; Bocco et al., 2001: Ramírez, 2001). This situation is particularly relevant in mountain areas, where the broad topographic and altitudinal gradients result in a wide biophysical diversity (Smethurst, 2000; Jansky et al., 2002). In central Mexico, diversity increases because a high proportion of the population lives in mountain areas (>1800masl; Chávez et al., 2001) and their recent history have led to the development of different land-use and deforestation patterns (García-Romero, 2002).

In order to address problems such as the nexus of biophysical diversity with the landscape-dynamics, landscape geography approaches are particularly useful (Arler, 2000; Works and Hadley, 2004) since landscape morpho-structural and mesoclimatic variables are not considered homogeneous, but vary according to a certain order that is reflected in the formation of patterns, or landscape systems (Vitousek et al., 1981; Ortiz and Toledo, 1998; Pimm, 1999; Kristensen et al., 2003; Burgos and Maass, 2004). Within a landscape system, the dynamics take place in hierarchically lower and visually homogeneous landscapes contained in it, and each of them corresponds to a specific system’s response status to disturbance and/or regeneration processes (Bertrand, 1968; Forman and Godron, 1986; Farina, 1998; Gragson, 1998).

Except for mature vegetation that characterizes the landscape with the highest development and stability in the system, all other landscape types are regarded as secondary, resulting either from natural and/or human disturbances, or from land-use abandonment and regeneration processes (Bastian and Röder, 1998; Muñoz, 1998). Under this approach, the positive or negative direction of the dynamics of a landscape system results from the balance of changes between landscapes with different functional significances (Bertrand, 1968; Serrão et al., 1996; Farina, 1998; Veldkamp and Lambin, 2001).

On the other hand, in the past decades several authors have pointed out that land-use change is the main mechanism leading the landscape-dynamics processes (Gragson, 1998; Smethurst, 2000; Bocco et al., 2001; Chávez et al., 2001; Veldkamp and Lambin, 2001; Jansky et al., 2002). This is of importance because when a type of landscape is replaced over time by another one, the process involves a transformation that may be either positive or negative in terms of the intensity and permanence of the damage associated to land use. According to Nepstad et al. (1991) and Lambin (1997) landscape-dynamics processes may be either negative when they are associated to disturbance and intensification of damage, or positive when related to landscape conservation and the (natural) landscape regeneration capacity.

Study Area

The Tuxpan river basin (1887.5km²) is located in the center-south side of the Trans-Mexican Volcanic Belt, which runs across the country from east to west along parallel 19°N. Its wide altitudinal gradient (3000m) allows the formation of three bioclimatic belts that respond to the distribution of geological structures: tropical dry forest on hills modeled upon a Cretaceous sedimentary basement (<1700masl), pine-oak forests on Plio-Quaternary volcanic slopes and the Tuxpan river alluvial plain (1700-3100masl), and pure fir forests in the peaks of the main mountain ranges (>3100masl).

At present the population (194143) is distributed in 167 villages and towns devoted to cultivation and cattle raising. Agricultural expansion affects the sedimentary hills and volcanic piedmonts, so that the forest area has retreated to the slopes and peaks of volcanic complexes (Figure 1). However, there are no previous studies clarifying the direction and rhythm of land-use changes, their consequences on forest-landscape regeneration or the ecological features of recovered areas.

This study aims at assessing the magnitude and direction of the recent (1976-2000) landscape dynamics and determining the processes that control these changes. The working hypothesis stems from the consideration that, in mid-sized mountain areas, the formation of different landscape systems affected by several biophysical and socioeconomic characteristics, suggests the existence of different patterns of landscape dynamics led by different stocks of change processes.

Methods

Landscape typology and multi-temporal cartography

Landscape maps were obtained from the interpretation in GIS (ILWIS ver. 3.0) of land covers in 1995 digital orthophotos (scale 1:75000) and 1976 and 1986 Landsat MSS, 1992 Landsat TM and 2000 Landsat ETM satellite images. Covers were interpreted through a "visual" method that uses direct, associative and deductive techniques to differentiate landscape "features" (Enciso, 1990; Mas and Ramírez, 1996; Arnold, 1997; Slaymaker, 2003; Chuvieco, 2002). The size of the minimum cartographical area was 4ha (Campbell, 1996), and color compounds (red, green and blue) 2,3,4 in Landsat MSS, and 3,2,1 (natural color) and 4,5,7 (false color) in Landsat TM were used for a better differentiation of covers. In order to check the information and determine the landscape typology, we used the classification of land use and vegetation of INEGI (1983), which provided the most suitable tool to allow updating and eventual assessments.

To verify the units and identify new classes, diverse indicators were reviewed (Slaymaker, 2003) comprising digital maps of slopes, altimetry and exposition of hillsides. Additionally, field inspections and interviews were conducted, resulting in 12 landscape classes which differ in terms of origin (mature vegetation, disturbance or regeneration) vegetation physiognomic development, type and permanence of the disturbance associated to land use.

Images were corrected geometrically and georeferenced to topographic maps scale 1:50000, through the Tie-Points method (Maus, 1996; ITC, 2001). Control points were taken from the land-cover map (1:50000; INEGI, 1983), and the RMSE index, or Sigma £2 was used to check the precision (Mas and Ramírez, 1996; ITC, 2001). In order to verify the image superposition, it was necessary to trace segments, over the orthophotos, which must not exceed the 30m (1 pixel) of the 2000 satellite image. The first interpretation was made over more detailed 1995 orthophotos of the area, and the resulting map was used as the basis for interpreting satellite images. The TM images of 1992 and 2000 have better spatial, spectral and radiometric resolution, so the interpretation was easier with support in the field work and sources of edited information. In the case of the MSS images of 1976 y 1986, the interpretation was supported with land use and vegetation maps (1:50000) from INEGI (1983).

In order to validate the classifications, besides the previous verifications, criteria include the interpretation of crown structures of plant communities, knowledge of the environment of the study area and field work, and the support of bibliography and cartography in the decision making. In addition, 60 sites were set up for verification of georeferenced fields with GPS technology.

Landscape system cartography

In order to determine and explain the landscape-dynamics patterns in the study area, a landscape classification system was used consisting of two taxonomic units: landscape system and landscape subsystem. The former is defined by morpho-lithology and climate features, that is, variables having spatial and temporal widespread range (1:50000-1:100000) and considering, at this scale, a kind of dynamically stable landscape (Bertrand, 1968).

The subsystem enables to distinguish landscapes belonging to the same bioclimatic environment but differing in morpho-lithology contents, with consequences that affect the distribution of the natural resources, slope sensibility and land use patterns. Both landscape system and subsystem maps were obtained from the integration (in GIS) of two information layers (1:50000):

Relief units. The map depicts a synthesis of the morpho-lithological and climatic aspects conditioning landscape stability and sensitivity (Lugo, 1988; Guerrard, 1993). To elaborate such units, the topographic patterns associated to different lithology types were interpreted in aerial photographs (1:75000; INEGI, 1978; Silva, 1979; Palacio, 1985) and the resulting units were analyzed and finally regrouped according to altitude, slopes and density drainage (Lugo, 1988; Guerrard, 1993).

Potential vegetation. The map shows the potential distribution of mature vegetation types that, at this scale, were integrated to the analysis as attributes of the relief units. The interpretation was conducted on the land-cover map, and in the case of areas devoid of mature vegetation, morpho-lithological, biogeographical and climatic aspects were used, as well as land-use patterns of the area (Rzedowski, 1988; INEGI, 1983).

Landscape-dynamics trends and processes

In order to assess landscape-dynamics magnitude and trends between the four dates of analysis, cross-tabulations, in GIS, were conducted between landscape maps and landscape systems, and the databases were exported to a statistics software to calculate areas, percent changes and mean annual transformation rates (MATR) as percentages (modified from Nascimento 1991).

K= 100×[(X₁/X₀)^1/n]-1

where K: mean annual transformation rate, X₀: landscape cover at the beginning of the period, X₁: landscape cover at the end of the period, and n: period duration.

Transition matrices were developed (Ramírez, 2001) resulting in 198 types of changes that were then grouped (depending the origin, physiognomic development and permanence of the disturbance of the landscape classes involved in the changing process) into two main groups plus 4 variants (Bastian and Röder, 1998). Positive processes were conservation (permanence of mature forests with scattered crops and forest use) and regeneration (conversion of a landscape class into another with a higher development and stability, in relation to the mature vegetation). Negative processes were disturbance (conversion of a landscape class into another less developed one, in relation to the mature vegetation, assuming damage of higher intensity and permanence) and intensification (permanence or turnover between landscapes with poor natural vegetation development and with land use other than forest, leading to negative consequences on landscape regeneration capability and stability).

Results

Landscape dynamics and processes

The Tuxpan river basin comprises 12 landscape classes and 4 groups (Figure 2 and Table I):

Forests with scattered farming and forest use. These comprise an area of 873.6km² (46% of the study area), with 675.3km² corresponding to pine-oak forest, followed by tropical dry forest (138.1km²), fir forest (50.7km²) and cloud mountain forest (9.6km²).

Secondary vegetation under extensive use. This comprises an area of 564.7km² (30% of the study area), of which secondary shrublands and induced grasslands comprise 381.5km², followed by highly fragmented (100.2km²) and fragmented forests (82.9km²).

Cultivated vegetation under intensive/extensive agro-forest use. The area devoted to agriculture is extensive (404.4km², 21% of the study area) and includes traditional crops (corn, bean, gourd) cultivated in small active and resting (<2 years) plots (298.7km²), as well as agro-forest plantations (105.7km²) under irrigation production systems.

Areas devoid of vegetation under extensive/intensive use. These comprise 46.2km², including human settlements (13.7km²), water bodies (6.5km²), and soil/rock areas (26.1km²) corresponding to rock-extraction quarries, active alluvial fans and terraces and eroded areas.

The mean annual transformation rate (MATR) during 1976-2000 (Table I), indicates a significant expansion of the forest area, from 781.1 to 873.6km² (41-46% of the study area), mainly due to the expansion of pine-oak forests (MATR= 0.4, 0.0 and 2.5 in the three periods) from 572.8 to 675.3km². In contrast, the secondary vegetation showed an evident retreat, from 674.6 to 564.7km² (36-30% of the area), mainly due to reduction of fragmented and highly fragmented forests (MATR= -2.4 and -1.7).

Agricultural landscapes also showed a slight increase, from 387.8 to 404.4km² (21% of the study area). However, this data resulted from different trends which led to a reduction in crops area (MATR= -0.1 in the latest periods) from 303.4 to 298.7km², while agro-forest plantations underwent ups (MATR= 1.0, -0.4 and 3.3 in the three periods), from 84.4 to 105.7km². Landscapes devoid of vegetation displayed a slight increase, from 44.0 to 46.2km² (2% of the study area), due to the expansion of human settlements (MATR= 2.3, 5.8 and 7.3 in the three periods), while soil/rock areas were reduced (MATR= -0.7, 0.0 and -3.2), and water bodies remained unchanged.

Table I shows the transition matrix (1976-2000) for 12 landscape classes. Intensification was the most important process (44% of total changes), with water bodies and human settlements maintaining 99 and 97% of their original area, followed by agro-forest plantations and crops (86 and 85%), shrublands/grasslands (82%) and highly fragmented forests (60%).

As regards conservation processes (710.9km² and 38% of total changes), the cloud mountain forest kept 98% of its original area, followed by the tropical dry forest (97%), pine-oak forest (90%) and fir forest (87%). Mature pine-oak forest regeneration (199.7km² and 11% of total changes) took place from fragmented forests (50% of their surface), highly fragmented forests (30%) and shrublands/grasslands (7%). Likewise, shrubland/grassland regeneration resulted from agro-forest plantations (5% of their surface area) and crops (5%) abandonment. Disturbance was less significant (140.9km² and 8% of total changes), and was due to the expansion of shrublands/grasslands over fragmented forests (8% of their surface), fir forests (8%), highly fragmented forests (7%) and pine-oak forests (5%). Furthermore, cultivated land expansion affected fir forests (6%) and shrublands/grasslands (4%).

Landscape systems

The relief units and potential vegetation maps enabled to determine the distribution area of 3 landscape systems and 3 subsystems (Figure 2):

Landscape system of fir forest in high volcanic peaks (S1). In the basin highlands (3% of study area) altitude exceeds 3100masl and is associated to a moderately cold (0-18°C) and sub-humid (800-900mm) climate with summer rainfall –C(E)(w₂)(w) (García, 2004). The potential vegetation is fir forest (Abies religiosa) with scattered Pinus, Quercus and Cupressus trees. The slope’s abrupt morphology (slopes of up to 30°), modeled upon volcanic tuffs and brecchias of Plio-Quaternary andesites, accentuate the risk from runoff and soil erosion (Price, 1998).

Landscape system of pine-oak forest in volcanic slopes and peaks (S2). Some low volcanic peaks and slopes (83% of total) are distributed between 1700 and 3100masl, leading to a temperate and sub-humid climate with summer rainfall –C(w₂)(w) (García, 2004). The pine-oak forest is the potential vegetation, formed by "casimbo" pine communities (Pinus pseudostrobus) along with Juniperus and Quercus trees. The morphological diversity of slopes influences the resource distribution and the system’s sensitivity to disturbance, resulting in three landscape subsystems.

Landscape subsystem of pine-oak forest in volcanic peaks (S2a). The S2 peaks (2300-2900masl and 10% of the study area) correspond to a volcanic complex delimited by Oligo-Miocene ignimbrite domes. The heavily eroded dome slopes (30°) continue with extensive and stable hills and cineritic plains in the central portion (3° slope).

Landscape subsystem of pine-oak forest in high volcanic slopes (S2b). Plio-Quaternary vulcanoclastic piedmonts bordering the main S1 and S2a peaks are located between 1900 and 2900masl, although they reach 3100m in the slopes connecting with S1. The differences in height and the terrain slope (>45°) increase the lithological basement sensitivity (volcanic tuffs and brecchias), evidenced as an intense soil erosion, a high drainage density and gravitational events.

Landscape subsystem of pine-oak forest in low volcanic slopes (S2c). The bottom of the north half of the basin (1700-2600masl) comprises the base of volcanic piedmonts (26% of the study area), partially covered by basaltic spills and Quaternary cineritic cones, and by the Tuxpan river’s narrow alluvial plain. The morphology consists in plains and gently sloping hills (0-15°), although some cineritic cones reach up to 3100masl, being affected by intense soil erosion and gullying.

Landscape system of tropical dry forest in sedimentary basement (S3). In the basin’s southern sector (15% of the study area) there are limestone, conglomerate and schist outcrops plicated and faulted since the Tertiary. The major relief element is the deep V-shaped Tuxpan river valley (up to 500m level drop and 45° slope), with hills and some basaltic mesas upon S2 lava flows. Altitude (620-1700masl) allows a moderately warm (12-33°C) and sub-humid climate, with an irregular rainfall distribution (800-900mm in 3 to 4 summer months) -(A)C(w₁)(w) (García, 2004). The tropical dry forest forms communities with high plant diversity and endemisms (60% of the rainforest species; Trejo and Dirzo, 2000), where low trees (8-12m) predominate, along with some temperate-forest elements and other life forms (Rzedowski, 1988).

Landscape-

dynamics variability

between landscape

systems

Figure 2 and Table II show the structure and dynamics (1976-2000) in landscape systems. As expected, the fir forest in volcanic peaks (S1, >3100masl) has the less diverse landscape pattern (five landscape classes), comprising extensive areas of mature forest (85% of study area), small farming plots (7%) and scattered shrublands/grasslands (3%).

In contrast, the pine-oak forest in volcanic slopes and peaks (S2) has a highly diverse landscape pattern (12 landscape classes). Although mature forests form the landscape matrix (45% of study area), this is considerably fragmented by crops (19%) and shrublands/grasslands (17%). Of the three landscape subsystems forming S2, the high volcanic slopes (S2b) displayed the highest similarity with the general pattern. There, the instability derived from topography and altitude fosters extensive mature forests (53%), scarce distribution of crops (11%), high abandonment and shrubland/grassland regeneration (17%) and highly fragmented forests (11%). In volcanic peaks (S2a) the landscape pattern includes a lower diversity (9 landscape classes) and with prevalence of grasslands (32%) distributed in large areas of plains and low piedmonts. Mature forests (57%) and fragmented and highly fragmented forests (7%) are noticeably restricted inside volcanic domes. In contrast, in low volcanic slopes (S2c) a highly diverse farming pattern prevails (11 landscape classes), with predominance of crops (41%) and agro-forest plantations (16%) at the expense of the reduced mature forest area (25%).

For its part, the landscape pattern of tropical dry forest in sedimentary basement (S3) is diverse (9 landscape classes) but with scarce forest surface (48%). The biophysical limitations derived from drought (9 months per year) plus heavy rainfall lead to alternating erosion, reflected as scarce development of crops (1%) and agro-forest plantations (6%), contrasting with the expansion of shrublands/grasslands (41%).

During 1976-2000 the higher increase in forest surface benefited S2 (MATR= 0.4, 1.0 and 2.6 in the three periods; Figure 3). This advance was due to different trends between the three S2 landscape subsystems. While S2a attained important benefits (MATR= 1.3, 0.2 and 4.6), expanding from 76.5 to 110.0km² of the forest area, S2b underwent lower growth rates but that nevertheless affected a more extensive area, from 380.3 to 437.4km². S2c displayed a discrete recovery, which rose since the third period (MATR= 2.4).

The secondary vegetation showed various trends. While shrublands/grasslands displayed a stable behavior (MATR= 0.1), fragmented and highly fragmented forests retreated (MATR= -2.5 and -1.8), mostly S2a and S2c fragmented forests (MATR= -5.3 and -2.8) and the S2b highly fragmented forest (MATR= -1.8). To note, in contrast with the decline in crops (MATR for S2c= -0.1, 0.2 and -0.2), forest plantations showed a positive trend (MATR= 1.1). Likewise, the system underwent an expansion of human settlements (MATR= 0.5, 5.9 and 7.4) along with a reduction in S2a and S2c areas with soil/rocks (MATR= -4.7 and -3.0).

Unexpectedly, S1 showed a retreat in mature and fragmented forests (MATR= -0.1 and -1.2). Besides, the small cultivated land was very dynamic (MATR= 2.2) by having a 68% increase, similarly to shrublands and grasslands (MATR= 1.0), which underwent a 26% growth. S3 also showed a stable behavior, with slight retreats in both the forest area (MATR= -0.4 in the latest period) and highly fragmented forests (MATR= -0.2 in the first period). Despite the reduction of crops (MATR= -0.4), agro-forest plantations showed a positive trend (MATR= 0.8), rising from 14.8 to 17.8km². Landscapes related with intensive land uses were affected by the reduction of human settlements (MATR= -0.7).

Figure 4 shows the main change processes of landscape dynamics (1976-2000) for the six landscape systems. The results reveal that S2c showed the most unfavorable situation due to a low conservation (20% of total changes) plus high intensification (60%) and disturbance (10%). The situation of S3 is similar, as conservation was lower than intensification processes (48 and 50%, respectively), while regeneration and disturbance were non-significant (0.2 and 1.5%). In S2a and S2b, conservation processes prevailed (36 and 42%, respectively) over intensification (32 and 38%), as did the regeneration process (23 and 12%) over disturbance (10 and 8%). In contrast, S1 displayed the most favorable situation, with an evident predominance of conservation (81%) over intensification (10%) and disturbance (5%).

Discussion

Contrasting with other authors that report high deforestation areas in the country (Carabias, 1990; Masera et al., 1997; SEMARNAP, 1997; Ochoa and González, 2000; Bocco et al., 2001; Galicia and García-Romero, 2007), the study area displayed a positive overall landscape dynamics and allowed the expansion of the forest area (162.58km²) at a rate (MATR value) of 0.09.

The explanation of this finding is complex, due to the balance between negative (51% of total) and positive (49%) processes, as well a non-significant agricultural landscape growth (from 20.5 to 21.4%) and stability of landscapes associated to intensive uses (2%). However, the data variability between the six landscape systems showed clear differences in the direction and rhythm of landscape dynamics, enabling to determine the existence of three behavioral patterns:

Fir forest in high volcanic peaks under conservation status. The instability caused by the high topographic and altitudinal gradients at >3100masl (Guerrard, 1993; Smethurst, 2000; Bocco et al., 2001; Jansky et al., 2002), added to the legal restrictions imposed by the Monarca Butterfly Biosphere Reserve (Giménez de Azcárate et al., 2003) restrain accessibility and farming development, while favoring the expansion of the forest area (85% of its surface).

This is important since, according to Farina (1998) and Galicia and García-Romero (2007), the landscape matrix exerts a significant control on the system’s dynamics, leading to high conservation (81% of total changes in this system) and low intensification (10%). Despite this favorable situation, the recent landscape dynamics showed a trend towards cultivated land expansion (MATR= 2.2) and clandestine fir (Abies religiosa) clearing (Rzedowski, 1988; Ramírez, 2001; Works and Hadley, 2004). The disturbance surface area was higher than the regeneration area (5 and 4%, respectively) even affecting fragmented forests (MATR= -1.2) that up to now have been key for the system’s stability and resilience, reflecting, as in other areas of Mexico, the lack of government programs interested in reducing the effects of corruption on forest land use (Works and Hadley, 2004).

Landscape systems under intensification-conservation status. Pine-oak forests in volcanic peaks (S2a) and in high volcanic slopes (S2b), and the tropical dry forest in sedimentary basement (S3), representing 71% of the study area, are characterized by a long history of land-use expansion and diversification (Bocco et al., 2001). Thus, by the early 1970s the inherited deforestation had reduced the forest area of the three systems to less than half their total surfaces (39.7, 6.4 and 49.5% for S2a, S2b and S3, respectively), resulting in a reduction of conservation (35.7, 42.1 and 48.4%) similar to the intensification area (31.7, 38.2 and 49.9%).

However, the three systems displayed a positive landscape dynamics. For example, regeneration surpassed disturbance in S2a (23 and 10% of total changes) and S2b (12 and 8%), enabling the expansion of mature forests (MATR= 1.6 and 0.6). The explanation is complex and is related to the drop of corn prices and the lack of farming development and subsidies (Toledo et al., 1989; Ochoa and González, 2000) that resulted in increased poverty, social margination and vulnerability (Rzedowski, 1988; Serrão et al., 1996; Byers, 2000; Giménez de Azcárate et al., 2003), and also affected the frequency and duration of abandonment periods (Bocco et al., 2001).

In contrast to other areas (Vitousek et al., 1981; Pimm, 1999; Tasser and Tappeiner, 2002; Kristensen et al., 2003), the expansion of shrublands/grasslands (S2b MATR= 0.3) does not represent a higher regeneration and resilience of disturbed systems. For example, in S2b 24% of forest-plantation land shifted to shrubland/grassland, but only 2% led to the regeneration of fragmented forests, affecting mostly steep slopes that, in theory, should be dedicated to forest management and conservation (Bocco et al., 2001). In these areas, the active exchange between cultivated land and shrubland/grassland has led to a higher landscape heterogeneity and instability (Burgos and Maass, 2004). Even in S3, where disturbance and regeneration exerted only a slight control on landscape dynamics (0.2 and 1.5% of total changes in these systems, respectively), this may suggest that the levels of heterogeneity and damage have remained unchanged for a long time.

This situation suggests grave perspectives, since during the study period the regeneration of mature forests occurred mainly in fragmented and highly fragmented forests (68% and 46% of their surface in S2a), which have not benefited from farming abandonment (MATR of fragmented forests in S2a= -5.3 and in S2b= -1.3). It is worth noting that, in general, forest management comprises clearing and/or selective cultivation of a few species, restraining an integral resource management. For example, in S2a resin production and timber in sawmills (heartwood) favors "lacio" pine (Pinus michoacana) and "chinese" pine (P. leiophylla) plantations (Works and Hadley, 2004), affecting other native species, such as the "casimbo" pine (P. pseudostrobus).

Pine-oak forest in low volcanic slopes under intensification-disturbance status (S2c). The landscape pattern is similar to the flat bottoms of the large alluvial basins of central Mexico, characterized (Rzedowski, 1988; Bocco et al., 2001; Ramírez, 2001) by a noticeable historical development of human settlements, cultivated land (40% of its surface) and agro-forest plantations (16%). Pine-oak deforestation has been so intense (23% of its surface in 2000) that the remaining communities appear restricted to the interior of cineritic cones and are dominated by non-commercial oak species (Works and Hadley, 2004).

As expected, intensification was the most important process (60.4% of total changes in this system), while conservation had the lowest figure in the basin (20.0%). However, the recent landscape dynamics shows a slight decline in cultivation (MATR= -0.2) along with an expansion of the forest area (MATR= 0.4) at the expense of ancient fragmented forests (54% of its surface) and shrublands/grasslands (11%). The cultivated field abandonment is explained on similar grounds as above for other systems, adding in this case the effect of heavy environmental imbalances due to gullying upon weathered volcanic ashes. This situation contrasts with the expansion of agro-forest plantations (MATR= 1.1) dedicated to production for regional commerce (Chávez et al., 2001), which accounts for the predominance of disturbance upon forest regeneration (10 and 9%).

Conclusions

The recent landscape dynamics shows that the six landscape systems respond to three behavior patterns. The fir forest in high volcanic peaks (>3100masl) is characterized by high conservation, but shows a trend towards increasing traditional agriculture and permanence of deforestation, promoted by high and hard-to-access areas which favor clandestine activities. The pine-oak forests in volcanic slopes and peaks (1700-3100masl) and the tropical dry forest in sedimentary basement (<1700masl) show high deforestation and use intensification, but also a trend towards abandonment of traditional agriculture, expansion of shrublands/grasslands, and scarce forest regeneration. The pine-oak forest in low volcanic slopes (1700–2600masl) shows the highest inherited transformation and displays a trend towards intensification of land use in human settlements and forest plantations.

The agricultural expansion that affected mostly the landscape systems of pine-oak forest and tropical dry forest (S2 and S3) took place prior to the study period. The trends observed over the past three decades show the intensification of a large area dedicated to agriculture, but with a trend towards predominance of positive change processes (conservation and regeneration) compared with negative processes (disturbance and intensification), particularly in the case of the regeneration of fragmented forests into mature forests.

However, today the regeneration of secondary and mature forests from shrublands/grasslands is a hard-to-complete process, due to the high risk of reactivation of land use for agriculture and cattle raising, which jeopardizes the current positive trends. Finally, it is worth noting that, in view of the decline of traditional agriculture, the supply of infrastructure into the region has led to the intensification and expansion of irrigated crops and human settlements, mainly in alluvial plains having soil, water and communication availability.

Acknowledgments

This research was financed by DGAPA-UNAM (PAPIIT- 302505 and 309108), Mexico.

REFERENCES

1. Arler F (2000) Aspects of landscape or nature quality. Landscape Ecol. 15: 291-312.        [ Links ]

2. Arnold RH (1997) Land use and land cover mapping. In Interpretation of airphotos and remotely sensed imagery. Prentice Hall, New Jersey, USA. pp. 36-43.        [ Links ]

3. Bastian O, Röder M (1998) Assessment of landscape change by land evaluation of past and present situation. Landscape Urban Plann. 41: 171-182.        [ Links ]

4. Bertrand G (1968) Paysage et géographie physique globale. Esquisse méthodologique. Rev. Géogr. Pyrénées Sud-Oest 39: 249-272.        [ Links ]

5. Bocco G, Mendoza M, Masera O (2001) La dinámica del cambio del uso del suelo en Michoacán. Una propuesta metodológica para el estudio de los procesos de deforestación. Invest. Geogr. 44: 18-38.        [ Links ]

6. Burgos A, Maass JM (2004) Vegetation change associated with land-use in tropical dry forest areas of Western Mexico. Agriculture, Ecosyst. Env. 104: 475-481.        [ Links ]

7. Byers AC (2000) Contemporary landscape change in the Huascarán National Park and buffer zone, Cordillera Blanca, Perú. Mountain Res. Devel. 20: 52-63.        [ Links ]

8. Campbell JB (1996) Introduction to remote sensing. Guilford. New York, USA. 622 pp.        [ Links ]

9. Carabias J (1990) En búsqueda de alternativas ecológicas para el uso de los recursos. In Rojas R (Coord.) En busca del equilibrio perdido. El uso de los recursos naturales en México. Universidad de Guadalajara. México. pp. 47-62.        [ Links ]

10. Chávez MC, Nava G, Velázquez L, Nava Y, Mondragón J, Carbajal H, Pedraza AM, Reyes BG, Arriaga C (2001) Agricultural research for development in the Mexican highlands: collaboration between a research team and campesinos. Mountain Res. Devel. 21: 113-117.        [ Links ]

11. Chuvieco SE (2002) Teledetección Ambiental. La observación de la tierra desde el espacio. Ariel Ciencia. Barcelona, España. 586 pp.        [ Links ]

12. Enciso JL (1990) La fotointerpretación como instrumento de apoyo a la investigación urbana. Universidad Autónoma Metropolitana, México. 47 pp.        [ Links ]

13. Farina A (1998) Principles and methods in Landscape Ecology. Chapman and Hall. Cambridge, UK. 235 pp.        [ Links ]

14. Forman RTT, Godron M (1986) Landscape ecology. John Wiley & Sons. New York, USA. 619 pp.        [ Links ]

15. Galicia L, García-Romero A (2007) Land use and land cover change in highland temperate forests in the Izta-Popo National Park, central Mexico. Mountain Res. Devel. 27: 48-57.        [ Links ]

16. García E (2004) Modificaciones al sistema de clasificación climática de Köppen. Serie Libros No. 6. Instituto de Geografía, UNAM. México. 90 pp.        [ Links ]

17. García-Romero A (2002) An evaluation of forest deterioration in the disturbed mountains of western México City. Mountain Res. Devel. 22: 270-277.        [ Links ]

18. Giménez de Azcárate J, Ramírez MI, Pinto M (2003) Las comunidades vegetales de la Sierra de Angangueo (estados de Michoacán y México, México): clasificación, composición y distribución. Lazaroa 24: 87-111.        [ Links ]

19. Gragson T (1998) Potential versus actual vegetation: human behavior in a landscape medium. In Balée W (Ed.) Advances in Historical Ecology. Columbia University Press. New York, USA. pp. 213-231.        [ Links ]

20. Guerrard AJ (1993) Landscape sensitivity and change on Dartmoor. In Thomas DSG, Allison RJ (Eds.) Landscape Sensitivity. Wiley. London, UK. pp. 49-63.        [ Links ]

21. INEGI (1978) Carta geológica, hojas E14-A14 E14-A15, E14-A24, E14-A25, E14-A26, E14-A34, E14-A35, escala 1:50000. Instituto Nacional de Estadística, Geografía e Informática, México.        [ Links ]

22. INEGI (1983) Carta de Uso de Suelo y Vegetación, hojas E14-A34, E14-A35, E14-A34, E14-A24, E14-A14, escala 1:50000. Instituto Nacional de Estadística, Geografía e Informática, México        [ Links ]

23. ITC (2001) Ilwis 3.0 Academic User's Guide. ITC. Enschede, Netherland.        [ Links ]

24. Jansky L, Ives JD, Furuyashiki L, Watanabe T (2002) Global mountain research for sustainable development. Global Env. Change 12: 231-239.        [ Links ]

25. Kristensen NP, Gabric A, Braddock R, Cropp R (2003) Is maximizing resilience compatible with established ecological goal functions? Ecol. Model. 169: 61-71.        [ Links ]

26. Lambin EF (1997) Modelling and monitoring land-cover change processes in tropical regions. Progr. Phys. Geogr. 21: 375-393.        [ Links ]

27. Lugo JI (1988) Elementos de Geomorfología Aplicada (métodos cartográficos). Instituto de Geografía, UNAM, México.        [ Links ]

28. Mas JF, Ramírez I (1996) Comparison of land use classifications obtained by visual interpretation and digital processing. ITC J. 3: 278-283.        [ Links ]

29. Masera O, Ordóñez MJ, Dirzo R (1997) Carbon emissions from Mexican forests: current situation and long term scenarios. Climate Change 35: 256-295.        [ Links ]

30. Maus P (Ed.) (1996) Guidelines for the Use of Digital Imagery for Vegetation Mapping. USDA, Washington, DC, USA. 125 pp.        [ Links ]

31. Muñoz J (1998) Paisaje y Geosistema. Una aproximación desde la Geografía Física. En Martínez de Pisón E (Ed.) Paisaje y Medio Ambiente. Fund. Duques de Soria. Universidad de Valladolid, Spain. pp. 45-56.        [ Links ]

32. Nascimento JR (1991) Discutendo números do desmatamento. Interciencia 16: 232-239.        [ Links ]

33. Nepstad D, Uhl C, Serrão E (1991) Recuperation of a degraded Amazonian landscape: forest recovery and agricultural restoration. Ambio 20: 248-255.        [ Links ]

34. Ochoa S, González M (2000) Land-use and deforestation in the highlands of Chiapas. Appl. Geogr. 20: 17-42.        [ Links ]

35. Ortiz B, Toledo V (1998) Tendencias de la deforestación de la selva Lacandona (Chiapas, México): el caso de Las Cañadas. Interciencia 23: 318-327.        [ Links ]

36. Palacio JL (1985) Geomorfología Regional del oriente de Michoacán y occidente del Estado de México. Tesis. Universidad Nacional Autónoma de México.         [ Links ]

37. Pimm SL (1999) The dynamics of the flows of matter and energy. In McGlade J (Ed.) Advanced Ecological Theory, Principles and Applications. Blackwell. London, UK. pp. 172-193.        [ Links ]

38. Price M (1998) Global Change in Mountains. Sbornik Ceske Geograficke Spolectosti. 103: 108-117.        [ Links ]

39. Ramírez I (2001) Cambios en las cubiertas del suelo en la sierra de Angangueo, Michoacán y Estado de México, 1971-1994-2000. Invest. Geogr. 45: 39-55.        [ Links ]

40. Rzedowski J (1988) Vegetación de México. Limusa. Mexico. 432 pp.        [ Links ]

41. SEMARNAP (1997) Estrategia Integral para el Desarrollo Sustentable de la Región de la Mariposa Monarca. Una Propuesta para Discusión. Secretaría de Medio Ambiente, Recursos Naturales y Pesca. México.         [ Links ]

42. Serrão E, Nepstad D, Walker R (1996) Upland agricultural and forestry development in the Amazon: sustainability, criticality and resilience. Ecol. Econ. 18: 3-13.        [ Links ]

43. Silva L (1979) Contribution a la Connaisance de l´Axe Volcanique et Petrologique des Laves du Michoacán Oriental. Tesis. Université d´Aix en Marseille, France. 230 pp.        [ Links ]

44. Slaymaker D (2003) Using georeferenced large-scale aerial videography as a surrogate for ground validation data. In Wulder MA, Franklin SE (Eds.) Remote Sensing for Forest Environments: Concepts and Case Studies. Kluwer. Hardbound, USA. pp. 469-488.        [ Links ]

45. Smethurst D (2000) Mountain Geography. Geogr. Rev. 90: 35-56.        [ Links ]

46. Tasser E, Tappeiner U (2002) Impact of land use changes on mountain vegetation. Appl. Veget. Sci. 5: 173-184.        [ Links ]

47. Toledo VM, Carabias J, Toledo C, González-Pacheco C (1989) La producción Rural en México: Alternativas Ecológicas. Colección Medio Ambiente 6. Fund. Universo XXI, Mexico. 392 pp.        [ Links ]

48. Trejo I, Dirzo R (2000) Deforestation of seasonally tropical dry forest: a national and local analysis in Mexico. Biol. Cons. 94: 133-142.        [ Links ]

49. Veldkamp A, Lambin EF (2001) Predicting land-use change. Agric. Ecosyst. Env. 85: 1-6.        [ Links ]

50. Vitousek PM, Reiners WA, Melillo JM, Grier CC, Gosz JR (1981) Nitrogen cycling and loss following forest perturbation: the components of response. In Barret GW, Rosenberg DR (Eds.) Stress Effects on Natural Ecosystems. Willey. London, UK. pp. 115-128.         [ Links ]

51. Works MA, Hadley K (2004) The cultural context of forest degradation in adjacent Purépechan communities, Michoacán, Mexico. Geogr. J. 17: 22-38.        [ Links ]