MOUNTAIN MEDITERRANEAN LANDSCAPE EVOLUTION CAUSED BY THE ABANDONMENT OF TRADITIONAL PRIMARY ACTIVITIES: A STUDY OF THE SPANISH CENTRAL PYRENEES.

Teodoro Lasanta-Martínez*, Sergio M. Vicente-Serrano** and José Mª Cuadrat-Prats***

* Instituto Pirenaico de Ecología (CSIC). Campus de Aula Dei. Apdo. 202. 50015. Zaragoza. Spain e-mail:

** Centre d'Etudes Spatiales de la Biosphère (CESBIO). 18 avenue. Edouard Belin, bpi 2801, 31401 Toulouse cedex 9, France

*** Departamento de Geografía y Ordenación del Territorio. Universidad de Zaragoza. C/ Pedro Cerbuna 12. Ciudad Universitaria. 50009. Zaragoza. Spain. e-mail: and

Abstract:

This paper analyses the changes inon the plant cover in a valley of the Spanish Central Pyrenees during the second half of the 20thCc. The role played by the extensification of the farming activities in this process has been examinedstudied. The method used was based on the evolutionary mapping of the plant cover, the application of landscape indices and the spatial analysis of livestock pressure. Throughout Along the century the stocking rate in the valley has decreased dramatically and sheep have been replaced by cattle. Most crop areas have been abandoned, and these areas have gone through a revegetation process (involving both natural vegetation and forest plantation). Nowadays the forest is the most abundant element. The study shows that both agricultural set-aside and the stocking rate decrease have led to a more natural landscape. Finally, the positive and negative effects of extensification and the changes in vegetation patterns associated with it are debated.

Key words: Landscape evolution, extensification, land use changes, stocking density, Pyrenees, Spain.

  1. INTRODUCTION

Since the middle of the 20th century the Spanish Mediterranean highlands have gone through profound changes on their demographic structure, social organisation and usage of space by humans. Literature distinguishes clearly between two systems of land management: the traditional system and the recent type (Anglada et al., 1980; Lasanta, 1990; Frutos et al., 1996). The first one is characterised by great complexity in its organisational schemes, in order to use all resources and so, be able to support a large human population and a large herd of livestock. This is framed within a system of very restricted relationships with other space. The recent system, on the other hand, sets up very dynamic interactions with the outside world, but greatly simplifies the relationships within the areas forming its territory. That is, it intensifies the most fertile and better-connected areas, while it leaves aside those less profitable, which are used very extensively (García-Ruiz, 1990; García-Ruiz and Lasanta, 1990).

The change in land management has resulted in very important transformations in the landscape, characterised by the spread of natural vegetation, both forest and bushland (Ubalde et al., 1999; Gallart and Llorens, 2003; Poyatos et al., 2003; Vicente-Serrano et al., 2004). Knowledge of the spatial and temporal patterns of such process has great socio-economic interest; related to potential and the quality patterns of grazing and forest resources. It has also environmental interest; related to location and access to the streambeds of sediment sources, runoff production and quality, risk of fires. And finally, it has interest from the landscape perspective related to; diversity, patchiness and structure of the plant landscape, aesthetic quality and tourist attractiveness (Wickham et al., 2000; Nagendra et al., 2004).

This paper studies the changes on the plant cover during the second half of the 20th century in a valley of the Spanish Central Pyrenees. The Borau valley is a representative example of the twenty-century changes in the Pyrenees: depopulation, abandonment of agriculture and decrease of livestock number. Moreover, the tourist activities are scarce in this valley, and the landscape transformations are only related to internal management changes. It attempts to explain the way in what management affects landscape dynamics when there is a change from an intensive land use, based on farming activities, to the extensive use of larger areas. The plant cover or the plant-land use mosaic is the landscape component most sensitive to changes in the environmental conditions or due to man’s actions; consequently, it is the key variable usually studied on landscape analysis (Bernáldez, 1981; Forman, 1995; Perry, 2002).

2. STUDY AREA

The study was conducted in the Borau valley, Spanish Central Pyrenees (Fig. 1), where a significant process of abandonment of the traditional land use practices has occurred. It is a small valley (4,186.2 ha.) crossed by the river Lubierre. It has a northerly-southerly direction, like the rest of the Pyrenean valleys. The altitudinal gradient is considerable, the lowest point being 840 m. and the highest one 2,566 m. There are two large relief units: the calcareous inner ranges and the flysch ranges (alternating marls and sandstones), rising closely in the south. The latter sector occupies a larger extent, a homogeneous area with rounded hilltops and smooth hillsides with slopes ranging from 20% to 60% (Soler-Sampere and Puigdefábregas, 1972). In the flysch area active geomorphological processes continue, mainly in the forms of overland flow runoff and rills; and landslides develop on steeper slopes (García-Ruiz and Puigdefábregas, 1982).

The area has a Mediterranean mountain climate with certain continental influence. Precipitation reaches 800-1000 mm p.a. on the valley bottoms and immediate hillslopes, and more than 2,000 mm on the highest areas. Rainfall is concentrate in spring and winter, with relatively dry summers. Average annual temperature is 10º C, but differences between months are important. In January the mean temperature is 2.3 ° C and in July is 19.1° C. The 0º C isotherm during the cold season (November – April) is located at 1,650 m. (Rijckborst, 1967; García-Ruiz et al., 1986).

The potential vegetation establishes two communities: the montane community, between 800 and 1,800 m. high, where Quercus pubescens prevails on the sunny slopes and Pinus sylvestris on the shaded ones between 1,200-1,800 m. of altitude. Above 1,800 m. there is a subalpine community with a prevalence of Pinus uncinata (Rivas-Martínez, 1987; Chauvelier, 1987). Both plant communities were very much affected by human actions, that deforested both the sunny hillsides to be put into cultivation and the upper forestland to expand summer pasture lands. Scrublands with Echinospartum horridum and small forests of Pinus sylvestris prevail in the montane community nowadays, while summer pasturelands develop on the subalpine community. The most representative species being: Trifolium alpinum, Festuca rubra, Nardus stricta, Carex montana and Festuca eskia, (Remón, 1997).

3. METHOD

The cartographic analysis of the plant cover was produced for four different dates: 1957 – this was the date when the first aerial photographs was available –, 1978, 1990 and 2000. Analysis was based on fieldwork and aerial photographs interpretation from 1957 (scale 1:32,000), 1978 (scale 1:18,000), 1990 (scale 1:20,000). The year 2000 cartography was based on field work and redrawing 1990 cartography. Twelve categories can be distinguished: 1- pasture above timberline, 2- pastures above timberline with scrub, 3- dense forest, 4- open forest, 5- reafforested areas, 6-open scrub, 7- dense scrub, 8- abandoned fields with open scrub, 9-abandoned fields with dense scrub, 10- cultivated plots (cereal and grassland), 11- bare rock, 12- populated areas.

Map production involved the use of Geographical Information System (ArcInfo, 7.04), which allowed depiction of plant cover evolution between 1957 and 2000. A database for every year was developed to estimate various landscape indices: partial and global stability, dominance and landscape geometry. The landscape indices are used to quantify landscape pattern to reflect the ability of landscape to support certain ecosystems and under such point of view, landscape pattern is an integrative measure of an ecosystems ability to provide habitat, prevent environmental degradation and support other natural processes functions (Geoghegan et al. 1997). The formulation of landscape indices used is:

  • Landscape stability: The index is used to distinguish between landscape categories that have gone through important changes and those other categories with minor transformations. This index has been produced for every category (Rosenfield and Fitzpatrick-Lius, 1986; Duguy, 1998) and for the whole territory. It is estimated by the following equation:

= stability index

= area where category i coincides in 1957 and 2000

= category i area in 1957

= category i area in the year 2000

The estimation of the Landscape Global Stability was based on the stability of each particular category:

= Total stability

This index ranges between 1 and –1, which allows comparing different spaces. If = 1, both covers overlap totally, both for each category and the global. If = -1 there is no overlapping between both covers. If = 0 all changes may be due to chance.

  • Landscape dominance: An index based on Shanon-Weaver’s information theory has been used for each time period (Shanon and Weaver, 1962):

= number of land use categories in the study area.

= proportion of category i in the study area.

= 1n (m) = maximum diversity when all uses are present on the same proportion.

The higher the indicator values are, the greater its role in the landscape.

  • Landscape geometry index (fractal dimension = F), assesses landscape complexity in relation to the complexity of the patch set perimeter (Mandelbrot, 1983; Kienast, 1993). To conduct this analysis the perimeter-area method has been used. The fractal dimension is estimated by the regression of 1n (A) against 1n (P/4), for each landscape type on the map. The fractal dimension is calculated according F = 2m, where m is the slope of the regression adjustment equation between ln (A) against ln (P/4), (Lovejoy 1982). A is the area and P the perimeter of the patches.

Anthropogenic management in the valley during the second half of the 20th century has been based on livestock activities. In order to know the stocking rate in each area, information about livestock censuses (livestock type and number of heads) was collected from state offices, and both cattle and sheep grazing areas and the stocking rate in each area were determined (View details in Vicente-Serrano, 2001).

Figure 2 gives an average annual summary of the stocking rate of cattle and sheep in the period 1957-2000. The values have been estimated by the product of the stocking rate and the number of days the animals have been grazing for: ((cattle equivalent/ha) x days on pasture).

The impact of livestock activities on the plant cover was estimated by a Predictive Discriminant Analysis (PDA), where changes on the plant cover were also related to topographical variables.

PDA is used to explain the value of a dependent categorical variable based on its relationship to one or more predictors (Huberty, 1994). Given a set of independent variables, PDA attempts to find linear combinations of those variables (topographic and livestock variables) that best separate the groups of cases of the dependent variable. These combinations are called discriminant functions (Hair et al. 1998). The procedure automatically chooses a first function that separates the groups as much as possible. It then chooses a second function that is both uncorrelated with the first function and provides as much further separation as possible. The procedure continues adding functions in this way until reaching the maximum number of functions as determined by the number of predictors and categories in the dependent variable. The PDA allows the determination of which predictor variables contribute to most of the inter-category differences of the dependent variable. The SPSS software was used to carry out the PDA. The dependent variables were selected according to temporal evolution of vegetation cover between 1957 and 2000.

  1. RESULTS

4.1 - ABANDONMENT OF TRADITIONAL LAND USE

In this area the traditional organisation of space was based on the maximum use of all resources available in the framework of a management system that was very dependent on local resources. Most of the roads that joined Pyrenean valleys with the local settlement centres were built at the beginning of the century –between 1905 and 1920- (Pujadas and Comas, 1975). The arrival of new products and new ideas from the lowlands was the cause for the initial change of the Pyrenean rural society and the transformation of traditional land use. After the 1950’s and 1960’s the definite incorporation of the highlands in the national market was much more obvious, and this increased the exchange of products and promoted the specialised production in the valleys. It was at this time that it became clear that there were lesser possibilities for highlands to enter a dynamic and wide market. Evidence of this is shown by a sharp population decline, shrinkage of the farming area and a drop in livestock numbers.

In 1910 the population of the Borau valley was 460 people, while in 1991 there were just 73 people, a population loss of 84.1%. Along the 20th century the population density dropped from 11 inhabitants/km2 in 1910 to 1.7 inhabitants/km2 in 1991.

Reduction of the farming area was simultaneous to the population decline. From analysis of 1957 aerial photographs, we estimate that crop fields could have occupied 43% of the land, stretching over most low and medium hillslopes in the valley. This number was estimated from the surface of abandoned lands and cultivated areas in the aerial photographs of 1957. The first statistical sources are of 1944. In this year only the 14.4% of the whole valley area was cultivated (CSIC, 1944). Set-aside continued on later decades: 9.4% of the land was under cultivation in 1957, 4.9% in 1978 and 1.7% in 1990 –the same as in 2000–. Consequently, more than 38% of the initially cultivated area has been abandoned. The farming land still under cultivation, all of it grassland, is concentrated on the valley bottom of the river Lubierre and on the nearby hillslopes, while the abandoned fields are located on steeper hillslopes and far from the centres of population.

Traditional animal husbandry relied upon sheep breeding, based on a trashumance system. The flocks were in the Borau valley in June, July, August and September, while they grazed in the uncultivated and fallow fields in the Ebro Basin, about 150 km. far from the Borau valley (Pallaruelo, 1993; Pinilla, 1995). This system allowed very high sheep numbers, managed in flocks of 2,000-3,000 heads. In the middle of the 19th century, there were around 21,000 sheep in the Borau valley (Remón, 1997). However the crisis of the migratory system started at the end of the 19th century due to decline in the wool trade. It was due also to the promotion of agricultural activities in the EbroBasin (cultivation of bare fields and expansion of irrigated farming), in contrast to animal husbandry, and finally, to the drop of the number of shepherds. The result was that the sheep census in 1911 showed less than 7000 head and this number continued dropping through later years. Figure 3 shows clearly the decreasing trend during the 20th century. Only in the mid-eighties, there was a slight rise in these migrating flocks thanks to the EU subsidies to this activity (Errea, 1995).

The decline in sheep stock was partially offset by the increase of cattle after 1950 (Fig. 4), as these ones require less labour force and adapt better to their housing than sheep. The increase of cattle, however, did not make up for the losses of sheep, so livestock pressure on the land has had large fluctuations over the 20th century from 817.3 livestock units (cattle equivalent) in 1911 to 337.3 in 1950, 183 in 1980 and 219.3 in 2000.

Livestock pressure decline and the replacement of sheep by cattle have occurred together with changes in the grazing system. Initially pastures above timberline were used in summer while intermediate hillsides (abandoned fields, bushlands and forests) were grazed in spring and autumn. The cultivated area provided cattle and sheep with little food. After the sixties, grazing has been concentrating in richer pasture areas or those ones with easier access for cattle, particularly the lower pastures above timberline and in the cultivated meadows. Medium hillslopes and some sectors in the pastures above timberline are barely grazed, following a similar pattern as in all the Pyrenean valleys (García-Ruiz and Lasanta, 1993). Cattle graze on mountain passes and meadows, while sheep are concentrated on mountain passes and intermediate hillsides, though with a very low stocking rate.

4.2 - LANDSCAPE EVOLUTION ANALYSIS

As a result of the decline of man's pressure on the land, remarkable changes have occurred on land use (Fig. 5).

In 1957 the abandoned fields occupied 1402.1 ha. (33.4% of the valley area), and was the most extensive land use. Natural forest occupied 1031.8 ha (24.6%). Pastures above timberline ranked third with 805.9 ha. (19.2%). Cultivated areas and bushlands occupied also extensive areas. In the year 2000 the natural forest was the most extensive use with 1948.5 ha (46.5% of the valley area), besides, 322.9 ha. of Pinus sylvestris reafforested after the 1950's must be added to that area. Forestland increased in areas formerly occupied by dense scrub or abandoned fields, whose areas have been decreasing progressively, and, to a lesser extent, in areas formerly occupied by pasture above timberline. Cultivated areas have also been reduced and replaced by abandoned fields, while pastures above the timberline have not changed much (Lasanta et al., 2001).

Landscape global stability between 1957 and 2000 has been very low, as only 46.3% of the valley area has not changed its coverage between these two dates. Table 1 includes stability values of the different landscape categories. Apart from 'population centre' and 'bare rock' categories, which have not changed their size, dense forest, reafforested forest and pastures above the timberline have been the most stable categories. The rest of the categories have been characterised by their instability: open forest has been a transitional visual category, as the areas occupied by open forest in 1957 generally corresponded to dense forest in 2000, while open forest is found now in abandoned fields. Abandoned fields and bushlands show a high instability level as plant succession tends progressively towards open forest first and to then dense forest.