Navigation – Plan du site
12ème Colloque Européen de Géographie Théorique et Quantitative, St-Valéry-en-Caux, France, 7-11 septembre 2001

Geographic information system and spatial effect

Applied to the study of the relationship between dynamics of ligneous cover and sheep breeders’ practices
Marianne Cohen et Catherine Mering


Un système d’information géographique, élaboré avec une base de données originale, montre un effet spatial dans la dynamique de la couverture ligneuse. Cette organisation spatiale est déterminée par des facteurs biologiques et sociaux, sur un mode continu ou discontinu. Par une démarche itérative, différentes hypothèses explicatives– rôle du pâturage, du feu et de la coupe-, sont testées, en appliquant plusieurs fonctionnalités du SIG. La gestion des paysages ouverts par les éleveurs du Causse Méjan apparaît durable dans quelques configurations.

Haut de page

Texte intégral


1In European rural areas, modernisation of agriculture brings out relative or total abandonment of some lands, where culture cannot be intensified. The maintenance of these areas is therefore given up (Dérioz, 1994). This is the case of some grasslands where grazing is more or less abandoned. Then, social vegetal species, as pioneer ligneous, begin to spread over these lands, and gradually form a more and more dense ligneous cover (Muller, 1996). This process is known as vegetation succession; in this case progressive secondary succession started with the decreasing intensity of human use of these lands (Lepart & Escarré, 1983). The progressive closing of the ligneous cover brings out a loss of aesthetical and biological qualities, as well as a decrease of pasture resources. In fact, on one hand, these grasslands represent a specific environment where some specialised species, as orchids, can find appropriate conditions of life. On the other hand, they represent an economical resource, in regions where breeding is still based on exploration of natural grasslands, as well as an original landscape to be preserved (Cohen dir., 2003). Here, a social dynamics starts off an ecological one, which in its turn, has consequences identified by social actors -farmers, inhabitants, developers, nature protectors, tourists...- as an environmental problem. Far from the old debate upon determinism (Bertrand, 1975, Durand-Dastes 1985, Friedberg 1992), this type of vertical interactions between dynamics of society and of biophysical environment is linked with topical questionings of theoretical geography (Pinchemel & Pinchemel 1988, Mathieu & Robic 2001, Ferrier 2001, Bertrand & Bertrand 2002).

2Our study is located in the Causse Méjan, a karstic highland in South France (Lozère). This highland is divided in two parts, an occidental woody area, and an oriental grassland area. The villages of Rieisse and Rouveret (fig. 1) were chosen on the woody part, because of their characteristic extension of shrubs and trees, in relation to rather ancient land abandonment and supposed influence of dolomite rock lay-down. However, previous multivariate analyses show the lack of influence of dolomite, because of ecological compensation phenomena. Other environmental factors (ex. soils) were correlated either to mature downy oak forest or degraded grassland. On the other hand, an important factor is the influence of grazing, here combined with fire and cutting for landscape management, depending on farmers strategies and nature representations (Cohen & Alexandre, 1997).

3One of our objectives is to measure the part of the different practices of sheep breeders in the ligneous cover dynamics, as well as their interactions, and what teaching it brings out for sustainable management of these regions. Such objective is suitable with a geographic scale - the site, at1:25 000 - which allows to reach the individual social actor, i-e the farmer, and to estimate his capacity of controlling vegetal landscapes, in an unfavourable social and ecological context (low demographic density, poorly dynamic breeding systems, numerous seed-holders in the surroundings).

  • 1  We do not consider here historical practices, which take place in the napoleonian land register, a (...)

4Moreover, ligneous cover dynamics due to sheep-breeders’practices can be described by models with a strong spatial component, with potential horizontal interactions. From a biological point of view : a) the distance from the seed-holders is determinant for spatial expansion of young ligneous, b) these seed-holders are organised in space as a landscape pattern. From a social point of view, practices take place in a spatial pattern linked up to history of property (land register)1 : a) pastoral practices : grazing inside the agricultural plots and proceedings of sheep flock inside or linking these plots, b) practices upon vegetation, i-e cutting and burning, detected inside the landscape unit.

Figure 1 : Location map

5An other objective of this work is to show how these spatial patterns interfere one with the other, and what kind of spatial model raises from their confrontation. We suppose here that spatial structure is not merely social-dependant (Brunet, 2001), but results from interactions between a spatial organisation produced by societies and the one of biophysical environment, both dynamics, as « geographical environment » (Pinchemel, op. cit., Cohen 2001). Hence, several spatial questions are raising :

  • Can we observe spatial and time discontinuities or continuities between those phenomena? What is the effect of « boundaries » between territories (here two villages) ? How does the neighbourhood effect interfere in spatial pattern ?

  • How can we explain spatial pattern of phenomena ? Are farmers, as individual actors, « able to modify or to create spatial structures » (Pumain, 2001) ? Is this spatial organisation more than a local combination ?

6The method involved here to analyse relationship between practices and ligneous cover dynamics is based on systematic study of vertical spatial links between both phenomena. This kind of approach classically requires statistical multivariate tools such as canonical analyses (Cohen, Alexandre et al., op. cit., 1998). Therefore, at this site scale, this analyses, based on a statistical population of random net of points of survey, is not very suitable with spatial analyses. In the present work, we have experimented a GIS approach in order to study vertical relationship between environmental parameters and social ones. The major hindrance of such an approach is that it gives qualitative rather than quantitative answer to the central interrogation which is « is there a link between the closing of vegetal landscape and the farmers’ practices » ? The main advantages is that GIS tools usually provide a large variety of answers to such a question in terms of localisation and that the answers have immediate iconic representations (Brunet, op. cit.), more appropriate with a spatial analyses, even at this rather large scale.


7Our hypotheses are about the biological process itself (1), the relationship between it and the social process (2-4), and social process itself (5), whose consistency has been put to test formerly (Cohen & Alexandre, op. cit.) :

    • 2  This vegetal species is considered as a « nomad » (Rameau et al., 1979), i-e either with a certain (...)

    closing of landscapes is due to pioneer ligneous species scattered by the seed-holders (Muller, op. cit.), as scots pine (Pinus sylvestris), common juniper (Juniperus communis), common box (Buxus sempervirens), aspen (Populus tremula), whitebeam (Sorbus aria) or even downy oak (Quercus humilis)2 in the studied region (karstic highlands in South France),

  • the decrease of pastoral activities favours ligneous scattering. Two processes interact : the lower disturbance due to trampling or grazing young ligneous seedlings ; the incomplete grazing of grass, providing spatial irregularities favourable to the establishment of seedlings (Acherar et al., 1984),

  • cutting and burning practices may either stop or temporarily restrain the ligneous conquest (Metailié 1981, Métailié et al., 1995),

  • when they are associated by the farmer, cutting and burning practices may interact with pastoral ones in the control of ligneous overgrow (Noy-Meier, 1995, Barbaro, 1999),

  • these practices take place in breeding systems, and are linked to a global strategy of farmers on their environment (Girard et al. 2001), to their representation of nature (Friedberg 1997, Cohen, Friedberg & Mathieu 1997), which refer to individual actors level.

GIS, a tool for testing hypotheses : a stepwise method

8GIS is very often used in landscape studies, with very different aims. For instance, landscape ecology (Forman & Godron, 1986) uses GIS to map habitat for rare faunal (Haines-Young et al., 1993, Ripple Eds 1994, Rushton et al. 1997, Raphael et al., 1998, Dettmers & Bart 1999, Gough & Roshton 2000, Franzreb & Llyod, 2000, Osborne et al, 2001) or vegetal species (Sperduto & Congalton, 1996). Therefore, temporal dimension is integrated to assess problems of loss of connectivity (ex. Hedgerows, Burel & Baudry 1990), in relation to change in land-use (Zhou et al., 1999) and their consequences for population dynamics and biodiversity (Ouin et al. 2000, Suarez-Seoane & Baudry 2002).

9GIS is also used to include spatial dimension to prospective management either in rural (Girard et al., op. cit.) or suburban areas (Harms 1999). Theoretical studies which are purposed for studying dynamics of landscape units by way of GIS are based exclusively on land register and on land-use attributes (Lardon et al., 1999), although we take into account such biological features of landscape as ligneous cover and flora.

10Other studies about landscape dynamics are entirely based on vegetation mapping deduced from remote-sensing image analyses (Jakubauskas et al., 1990, Parresol & McCollum, 1997) or on rules proceeding from physical laws (ex. reflectance of types of urban surfaces from remote sensing data : Couret, 1994, 1996, Rhami et al., 1999), contrary to our work, also based on land survey data.

11Our use of GIS is closer to fewer works of ecology or geography which use this tool to compare landscape structure and farming systems (Baudry et al., 2000), in particular those in context of land abandonment, showing relations between land cover change and land-owner (Turner et al., 1996), land cover and landscape patterns dynamics (Mast et al., 1997, Poudevigne et al., 1997, Ernoult et al., 2003). Other experiences, even in a very different context (19th century in southern Quebec, Paquette & Domon 1997) are interesting to discuss the place of natural versus social factors in land use dynamics.

12GIS tools are used here for studying spatial relationship between :

  • social phenomena such as: modernisation of pastoral activities, cutting and burning practices, land property and land use,

  • and natural phenomena such as : bush overgrowing, spontaneous timbering.

13The GIS is a help for testing hypotheses about spatial relationship between social and natural phenomena. These hypotheses are rules or logic assertions deduced from literature and previous multivariate statistic studies (Cohen & Alexandre, op. cit.), although other studies present simultaneously results obtained with GIS and statistics (Poudevigne et al. op. cit.).

14The main functions of GIS which are used here for that purpose are the geographic joint and the spatial masks built up with buffer zones around given spatial structures. The software used is Savane (Souris, 1995). Before making any sophisticated request about spatial relationship between phenomena under study, one has first to display spatial organisation of the basic parameters which are involved in the questioning, as dynamics of ligneous cover, dynamics of grazing, intensity of practices which are used by the farmers to restrain the ligneous overgrowing. The corresponding thematic maps are obtained by classical geographical joint between relations of two dates when necessary, and/or by selecting the relevant attribute to be represented (fig. 2-5).

15The further steps of proceeding the GIS were designed in order to reveal some spatial occurrence of the hypotheses under study (fig. 6, 7). Spatial masks are used to extract residues unexplained by a first hypotheses in order to test the following one. The first hypotheses put to test is about a vertical relationship between ligneous vegetation dynamics and practices such as burning or cutting. Once it has been tested, another one, as influence of decreasing grazing on ligneous overgrow, is studied on the spatial restriction which would be defined by the locations where the previous hypotheses is clearly satisfied (see frame 1).

An original geographic data base

  • 3  Nevertheless, the choice of zonal mapping creates also an artefact, putting emphasis on spatial di (...)

16The geographic database is original, i-e in largest part constituted by the results of proper inquiries, which is possible on this rather large scale (tab.1, Oulebsir, 1997)3. This avoids the artefact of an existing spatial structure (ex. administrative pattern) or nomenclature (types of land use, ex. Corine land cover) in the presentation of data, an inconvenience of external databases.




Type of relation



Vegetation 1994


Numerized vectorial

Spatial zoning of aerial infra-red photos 1989 + field survey 1994

Flora and landscape

Anthropic signs of burning and cutting

Environment conditions

Vegetation 1963


Numerized vectorial

Spatial zoning of aerial infra-red photos 1963

Landscape (ligneous cover)

Agricultural plots


Numerized vectorial

Inquiries + land register

Practices during the sixties

Practices in 1992-1994

Tableau 1 : The database

17From the raw documents (aerial infra-red photos of 1963 and 1989), a spatial zoning of the ligneous cover is produced for both dates. A stereoscopic view allows to distinguish three heights of ligneous (small : shrub, bushes or young trees taller than one meter; medium and high trees); infra-red to differentiate species (resinous, as scots pine or common juniper; leafy species as common box, pubescent oak, whitebeam or aspen). Ligneous cover is estimated in percent by comparison with Folk chart (Godron et al., 1983). It produces very accurate maps: 475 objects in 1963, 374 objects in 1994 (tab.1). Taking into account vertical structure of vegetation, 5 % error is expected from aerial photos analyses.

18For the current situation, the same photo-interpretation is verified by field survey. Boundaries between these “landscape units” are verified. Each spatial entity is described with attributes like flora (exhaustive list of present species), vegetation cover and species contributions, environment conditions (soil, erosion marks, slope, etc...) and signs of human action (Cohen & Alexandre, op. cit.).

19The inquiries among farmers of the villages of Riesse (two active) and Le Rouveret (two active, two retired) provided information about the agricultural plots and about the pastoral activities during the period between the sixties and the nineties. Information as frequency, season and spatial pattern of grazing (Cohen & Petit, 1995) are compared with agronomic inquiries (Brun et al., 1978, Osty & Auricoste 1989, Lardon et al. 1995) in order to verify their consistency. Other data are: type of property, opinion of farmer about natural resource (Cohen, Friedberg & Mathieu, op. cit.). Combination between land register and proceedings of sheep flock produces a zonal map of pastoral activity, with 70 objects (fig. 1). Frequency and farmers’justification of cutting or burning inside vegetal landscape units (relation vegetation 1994) are deduced from inquiries. They are compared with field survey to check their consistency and spatial pattern.

20The interval between the two dates of description of ligneous cover and farmers’ practices, i-e 31 years, is chosen to highlight the interactions between short-time changes. It is too small to catch the whole vegetative succession, for example until mature forest. However, it is interesting to show a phenomenon with a very strong social perception, i-e invasion by pioneer ligneous, and to search its relation to the evolution of agriculture. Nevertheless, in some cases, this interval may be too large to reach an accurate understanding of the phenomenon (see note 4). On the other hand, this scale of time is suitable to farmer inquiries based on comparison between present and former situations, based on their personal records, even if we must be careful in the interpretation of those facts.

Study of ligneous cover dynamics with GIS

21The ligneous overgrow may be studied by GIS according to both retrospective and prospective approaches (frame 1:1st step).

22By retrospective approach, each spatial entity is determined by the geographic joint of the relations of vegetation of 1963 and 1994. Then, the arithmetic difference between the values of the attribute “sum of cover percentages of low, medium and high ligneous layers” described at each date is computed.

Numerical attributes

Qualitative attributes

- 90 to – 50 %

Mean regression

-50 to – 10 %

weak regression  

- 10 to +10 %


+ 10 to +50 %

weak progression

+ 50 to + 90 %

mean progression

+ 90 to + 130 %

strong progression

+ 130 to + 170 %

very strong progression

Tableau 2 : Attributes of ligneous cover evolution

  • 4  For example, on west side of the studied area, the large stable zone is covered on both dates by g (...)

23The map (fig. 2) is obtained by the threshold of such numerical attribute, in equal intervals around a class of stability (insignificant change, in relation to 5 % error, i-e -10 to + 10 percent, tab. 2). Crops are excluded of the analyses. Interpretation must remain careful, for the evolutions may be not linear or stability may be an artefact4.

24First-order spatial pattern (regional effect) can be interpreted with help of anthropic factors, on different scales, and well-known spatial models :

  • difference between the two villages (regression is concentrated in Rieisse lands), as territory effect,

  • neighbourhood to crops or village, a very general spatial structure of human activity, as polarisation effect, while modern breeding leads to concentrating human action on the more productive areas,

  • plots effect which may explain some strong discontinuities between dynamics, linked to social use of land.

252nd order spatial organisation is depending on two interacting factors : social control of the ligneous overgrow and the neighbourhood effect.

  • Small sized patches, in a context of strong increase of ligneous cover, correspond to strong spatial interactions between groups of seed-holders (orange-coloured), and areas surrounding the first ones where ligneous scatter during the period (dark orange and red), typical of edge effect in forest dynamics by nucleation (Forman & Godron op. cit., Rameau 1999).

  • small fragmented areas in large geometric patches, in regressive ligneous areas,correspond to conservation of coppices in burned plots by farmers, in order to provide a shelter for sheep during the hot hours, with no spatial interactions between these remnant woods and burned grasslands.

  • Medium sized patches of stable zones could be interpreted as a half-way case.

Fig. 2 : Detection of ligneous cover dynamics

26Figure 2 shows a first-order spatial organisation. A very high progression of ligneous cover is observed both at the western and eastern extremities of the studied area, although weak dynamics are observed in the centre of the area and around crops, and high regression in the south west.

27Internal spatial pattern, i-e 2nd order spatial organisation, shows significant differences, a micro-mosaics pattern in strongly progressive landscape areas (red to yellow), large blocks with small fragmented areas in regressive areas (blue), meso-mosaic pattern in stable areas (green) with one exception (note 4).

28However, the retrospective map does not take into account small seedlings, lower than one meter, because of the limits of photo-interpretation. Well, these young ligneous are significant for the future evolution of these landscapes. If they are numerous, one may expect a closing of vegetal landscape, with biological and economical negative consequences for social actors. Moreover, to-day practices - pasture, cutting or burning- are supposed to have a strong influence on these young ligneous settlements, whereas they cannot “ explain ” the presence of adult trees, linked with former land use. Now, sustainable landscape management must give a special attention to these nowadays practices and their efficiency.

Tableau 3 : Retrospective and prospective ligneous cover evolution

29Prospective approach of ligneous overgrow brings more information. The risk of ligneous overgrow is estimated by the rate of young ligneous measured on field, thresholded in 4 classes. This attribute has been combined with the previous one, disregarding the intensity of ligneous cover dynamics. By this way, we obtain 9 combinations, showing time continuity or discontinuity between processes of ligneous overgrow (tab.3). As formerly, crops are excluded from analyses.

30In comparison with fig. 2, map 3 shows similarities (regional effect) and dissimilarities. Neighbourhood of wooded Gorges du Tarn (fig. 1) seems to favour downy oak regeneration in Rieisse territory (chorological effect). Time continuities or discontinuities are in a large part explained with spatial arguments :

  • stable zones, located in medium patched areas, have a high risk of future ligneous overgrow, because of the neighbourhood of numerous seed-holders in surrounding progressive areas (edge effect) ;

  • the neighbourhood of Rouveret or Rieisse crops areas is favourable to absence of risk (polarisation effect).

Fig. 3 : Prospective approach: risk of ligneous overgrow

31Future risk of ligeous overgrow is large over all the studied territories. However, the map shows regional effect (regressive areas with low risk on south-west, and highly progressive areas on extreme west and east), and chorological effect (neighbourhood from wooded Gorges du Tarn), on north border of studied area.

32Time continuity in the process of ligneous overgrow is observed in progressive and regressive areas, and in stable areas around Rouveret and Rieisse crops areas. However, in the centre of map, future dynamics (red, fig.3) is more rapid than past one (yellow, orange, fig.2). In the same area, discontinuities are observed on some stable zones (yellow-green, fig.3) with high risk of future ligneous overgrow.

Social practices on ligneous vegetation : an explanation of ligneous dynamics ?

33Located inside vegetal landscape units and revealed by inquiries among residents and by field survey, practices over ligneous vegetation, as cutting or burning, are more or less intensive, regarding to their repetition and association.. Combining these criteria, we obtain 4 classes (tab.4, frame 1:1st step).

No practice

Unique practice

Frequent practices

Intense practices

No cutting or burning

Low frequency and partial cutting or burning

- Low frequency cutting and burning

- High frequency or general cutting or burning

High frequency (15-20 years) and associated practices

Tableau 4 : Intensity of burning and cutting

Situation in 1963

Comparison between 63-94

Type of hidden practice


Decreasing of ligneous cover

Hidden cutting

Leafy ligneous cover <10%

Strong increasing (>20%) of leafy medium and high ligneous

Hidden fire

Tableau 5 : Hidden practices on vegetation deduced from ligneous cover dynamics

34Moreover, GIS enables to create a new information : the “hidden practices”, i-e the ones that did not leave any sign on field or have not been revealed by the inquiry. A comparison between ligneous cover in 1963 and 1994 shows areas with a very rapid progression of leafy ligneous (stump shooting after « hazardous fire » in 1949 in a downy oak forest), and regression of ligneous cover (due to cutting).

35This allows to create two more classes : hidden fire and hidden cutting (tab.5). Joining this relation (ligneous dynamics 1963-1994) with the previous one (vegetation 1994), and combining respective attributes (hidden practices, cutting and burning practices), we obtain 6 classes, excluding crops.

36Similarities between spatial pattern of ligneous cover dynamics (fig. 2) and of these social practices (fig. 4) leads us to think about a link between both phenomena. According to one of our hypotheses, these practices may explain the previously studied phenomena of regression of ligneous cover. On the contrary, hidden fire of 1949 may explain good regeneration of downy oak, previously interpreted as a neighbourhood effect with wooded Gorges du Tarn (fig. 3).

Fig. 4 : Farmers’s practices  on ligneous vegetation

37Some regional effect previously observed are magnified in figure 4 :

  • Territory effect, i-e opposition between Rieisse and Rouveret lands, in relation with land property ; however, on the border between the two villages, we observe a transition area, due to uncertain status of lands,

  • Parcel effect, with large spatial patterns.

Pasture : an other explanation of ligneous overgrow

Evolution of GI index

Qualitative attribute

- 4 à – 2

Increase of grazing intensity

- 1 to +1

Stability of grazing intensity

+2 to + 3

low decrease of grazing intensity

+ 4 to +7

mean decrease of grazing intensity

+ 8 to + 14

high decrease of grazing intensity

Tableau 6 : Evolution of grazing intensity

  • 5  This index has been previously tested in multivariate analyses, showing good correlation with vege (...)

38The evolution of pasture is described, for each season, with a qualitative parameter, i-e an index of grazing intensity, from 0 to 4, combining stocking rate (n.sheep * n.days of 8 hours/n.ha.), impact of herbivores on grassland (from farmers’inquiries and field survey), and proportion of external food provided by the farmer5. During a year, the index may sum 16 points (4 per season). Comparing this index during the sixties and the nineties, we observe either a decreasing of expected impact of sheep on vegetation, or an increasing or a stable one (tab.6, frame 1: 1st step).

39Such a land abandonment, as shown in fig. 5, is not a current situation in the Causse Méjan (Lardon et al., 1995). Once more, we observe a regional effect, between Rieisse and Rouveret territories. The situation of Rouveret is linked with less dynamic and modernised farming than in Rieisse. Moreover, dairy production (in Rouveret) is more dependant on external food than meat production (in one farm of Rieisse), and its expected impact on vegetation lower. The property of land interferes with previous factor, as collective lands of Rouveret are more abandoned than private ones, as well as the neighbourhood of the village or the crops (polarisation effect, also in Rieisse split into two crop areas, fig. 1).

40According to our hypotheses, the decrease of grazing intensity would provide the scattering of ligneous vegetation followed by the closing of vegetal landscape. Such hypotheses is verified by visual comparison between map 2 and map 5 at the extreme western and eastern parts on the map. On the contrary, keeping grazing intensity allows the stability of the ligneous cover, it is the case near Rouveret village in a small plot. Elsewhere, spatial relationship between pastoral activity and ligneous dynamics is less obvious. Their respective spatial patterns are different : the agricultural plots lattice is not well fitted to heterogeneity of the ligneous cover nor to its dynamics. Moreover, it is possible to mistake the pasture effect on vegetation dynamics with the one due to burning and cutting practices.

Fig. 5 : Hypothesis on pasture. Evolution of the grazing intensity between  1960 and 1990

41Grazing intensity increase is observed in very few parcels in this area, typical of a strong land abandonment. In both territories of Rieisse and Rouveret, grazing intensity  has generally decreased since the sixties (map 5). However, Rouveret territory is much more concerned by this phenomenon.  Neighbourhood from the village or the crops favour stability or increase of grazing intensity.

Iterative spatial filtering of hypotheses

42When visually comparing the three following maps: dynamics of ligneous cover (fig. 2), practices of burning and cutting (fig. 4), and evolution of grazing intensity (fig. 5), one may notice some spatial coincidence, but also a risk of confusion in their interpretation. To avoid such a problem, we proceed by an iterative spatial filtering of hypotheses.

  • 6  The sheep load is too much weak (1 sheep/ha.) to have an influence in such strong regression of li (...)

431. We assume that only cutting and burning, but not the pasture6, may be responsible for the regression of the ligneous cover. This hypotheses is tested by creating a new combined attribute [‘cutting & burning’ x ‘ligneous cover dynamics’], and then a mask over the space units where we observe the following configurations (frame 1: 2nd step) :

Regression of ligneous cover

Regression of ligneous cover

Regression of ligneous cover

Unique practice

Frequent practices

Intense practices

442. In order to analyse the remaining situations (residue), an attribute is created by combining grazing and ligneous cover evolutions, aiming at evaluating the part of the decrease of the pastoral activity in the ligneous overgrow, and on the other hand the part of the stability or increase of stocking rate on stability of ligneous cover (frame 1 : 3rd step).

Ligneous cover evolution

Grazing intensity evolution

-1 : -10 à + 10 % : stability

-1 : stable or increasing

1 :+10 à + 50 % : weak progression

1 : weak decrease

2 : +50 à + 90 % : mean progression

2 : mean decrease

3 : +90 à + 130 % : strong progression

3 : strong decrease

4 :+ 130 à + 170 % : very strong progression

4 : very strong decrease

Tableau 7 : Comparison of evolutions of grazing intensity and ligneous cover

45Such link is studied here by qualitative comparison between two processes with five levels of intensity, which could be assimilated to rank statistics (tab.7).

46At last, it may happen that some parameters may have strong spatial relationship considering only low, medium or high values although when studied as quantitative data they may have small linear correlation coefficient (it is the case here). This method enables to go further in the explanation of weak correlation coefficient between two quantitative parameters (for example, considering the case of two processes of same direction and different intensity, or « rank »). Furthermore, such proceedings are well fitted with the level of precision of our data, especially those concerning practices during the sixties, based on farmers’memory. The ratio (R) between both indices shows six different cases, in which the hypotheses of pasture is more or less confirmed (tab 8).

Tableau 8 : Hypotheses of a relation between grazing intensity and ligneous cover evolutions

473. The following step is to combine this attribute with the previous one. At last, we make the hypotheses, that excepted the masked combination, grazing, cutting and burning can explain either together or separately, ligneous overgrow or stability (frame 1 : 4th step). By doing so, we obtain a new spatial attribute with 33 modalities !

Explosion from combining attributes

48Such iterations of combinations of qualitative attributes on spatial units provides very rapidly an explosion of the number of modalities. By doing so, the legend becomes more and more complex, and the iconic representation less and less legible, whereas this property is an important requirement. It becomes then necessary to find out objective criteria to reduce the number of modalities created by the combination of primary attributes, without damaging its reliability, an other important demand of a « model-map » (Brunet, op. cit.).

  • A first criterion may be used after the mapping: it consists in suppressing modalities which are very weakly represented on the map after the crossing operation. By doing so, the aspect of the iconic entities of the map gets smoother aspect.

  • A second criterion consists in reducing the observed situations to some few types, by disregarding the intensity of the various phenomena and by focusing only on their orientation (progression, regression or stability of grazing intensity and ligneous cover) or types (ex. cutting and burning). This results also in a smoother aspect of iconic entities (fig. 6). By this way, this new attribute contains 6 modalities, as shown in tab.9.

Tableau 9 : Relations between ligneous cover dynamics, grazing intensity evolution, cutting and burning

Tableau 9 : Relations between ligneous cover dynamics, grazing intensity evolution, cutting and burning

Fig. 6 . Simplification of attributes and hence of spatial structure

49By doing so, we attempt to answer the following questions (frame 1: 4th step):

  • what is the expected part of decreasing human activity on ligneous overgrow?

  • what is the efficiency (when remaining) of grazing and/or burning-cutting towards ligneous overgrow ?

  • are there cases when these practices are not efficient enough to control ligneous overgrow, in spite of the increase of grazing intensity ?

Two social factors of biological dynamics and their interactions

50The final result of our proceeding is a map showing the respective part of pasture, cutting and burning practices in the evolution of ligneous cover. To the 6 modalities of the combined attribute (tab 9), we add previous masked case, i-e regression of ligneous cover due to cutting or burning. The map becomes a tool to analyse the possibilities of social control on the overgrow of the ligneous vegetation by individual actors in an unfavourable context, i-e woody environment and little intense and modernised farming (fig. 7, frame 1: 4th step).

51Regional effect seems to be linked with individual farmers strategies, dynamics of farmer systems and land property, either between Rouveret and Rieisse territories, or between northern and southern parts of Rieisse territory. We notice transition areas, where these opposites are spatially organised in medium patches pattern, which let us think of spatial interactions between both parts, as could occur in such a socio-ecotone.

52Rouveret accumulates driving factors of uncontrolled ligneous cover dynamics : low dynamic farmer system, low grazing, no perspective of succession, restricted possibilities of cutting and burning in collective lands. Moreover, Rouveret is an example of the low influence of pasture activity in such context, compared with the resultats when combined with cutting or burning, as previously shown by field experiment by Noy Meier (op. cit.). Where grazing intensity is stable, i-e in very few plots close to the farm, ligneous cover remains equal, but in the largest part of the territory, its decrease leads to an unmanagable ligneous dynamics.

53In the southern part of Rieisse, the ligneous overgrow is socially controlled through a project of development by the farmer ; he manages his lands with care, spending much time cutting and burning at the proper moment (february-march), in order to get good pasture for his sheep, and to maintain his farm for his grandson. In the northern part, decrease of pastoral activity, the neighbourhood of wooded Tarn valley and burning jointly favour ligneous dynamics. This northern part is divided between the former farmer and another one, who manages burning and cutting with less care. Well, burning or cutting are not sufficient to maintain open landscape if grazing intensity is decreasing.

Fig. 7 : Respective part of pasture and cutting-burning practices in the dynamics of ligneous cover

54One may notice on fig.7 :

  • a regional effect from west to east. The territory of Riesse is  controlled  by pasture and/or cutting and burning, as the one of Rouveret is largely uncontrolled. One note a transition area on centre-south,

  • a regional effect, between the southern part of Rieisse territory better controlled than the northern one. Once more, we note a transition area between theses two areas,

  • the influence of the neighbourhood of Le Rouveret crops area on control of ligeous overgrow by pasture.

55These farmers have not the same attitude before fire. For Rouveret farmers, it is a theft ; for southern Riesse farmer, fire is a know-how, just as fire is the only solution to ligneous overgrow for the northern Rieisse farmer.

56Some discrepancies are linked with a more distant temporality. In spite of increasing grazing and cutting, some areas are overgrown, which shows the limits of those social practices, not intense enough perhaps in the XIXth century woodlands (ex. south of Rieisse, Doumergans, Sarrades, fig. 7). Other case : in spite of the decrease of grazing, ligneous cover remains stable in south oriented slopes of Rouveret (Adrecht). At last, there is a conflict upon the use of these collective land, and in this context, inquiry about grazing may not be very reliable. Moreover, if we consider former use, these plots were reserved for cattle breeding associated with small plots of culture during the XIXth century, practices more favourable to conservation of dense grasslands (Cohen & Alexandre, op. cit.), and then unfavourable to ligneous scattering (Acherar et al., op. cit.). Ancient land register also shows that plots close to crops areas were often cultivated during the XIXth centhury (Oulebsir, op. cit.).

57Taking account of future risk of ligneous overgrow (map 3), it appears that vegetation management by farmers allows a long-lasting control of ligneous vegetation dynamics in very restricted areas : the south-west corner by repetitive and accurate practices, and near Rouveret village, by regular pasture on ancient crop lands.

58In the central medium patched area, such management seems short-lasting : scattered cuttings are not efficient to control ligneous overgrow in a context of low pasture and weak social appropriation (lend areas).

59Finally, large areas escape from farmer control : practices of cutting and burning are either too old (downy oak forest) or not careful enough (extreme north-west), pasture is decreasing, seed holders are numerous.

Frame 1 : A four-step method


  • 7  Despite a very carefull digitalization (Oulebsir, op. cit.), we could not avoid very small discrep (...)

60In this critic discussion, the first point concerns the improvement of the quality of data. This could be achieved by analysing multispectral high resolution satellite images (ex. Spot 5) with numerical methods instead of manual photo-interpretation and drawing. Such proceeding should improve the georeferencement and hence the accuracy of superposition of diacronic relations7. On the other hand, inquiries among farmers could be improved by commune registers study and ground survey with farmers.

61The second point concerns the use of multivariate analyses to avoid combination of over-numerous qualitative attributes produced by thresholding quantitative attributes, such as ligneous overgrow or grazing intensity evolution. As a matter of fact, the negative aspect of the procedure is the multiplication of spatial structures provides by the combination of qualitative attributes which may lead to an undesired fragmentation of the space. Maps resulting from this kind of processing may the be illegible. We strictly solved this problem by simplifying combined attributes, but so doing, pertinent information about the intensity of phenomena is lost.

62On another hand, spatial entities created by the proceeding may be submitted at each step to statistical analyses, in order to check if the new spatial structures (east/west; north/south, distance to crops) provided by the GIS treatment are statistically relevant. Another perspective would be integrated methods for simulating qualitative spatial processes, for example using multi-agent systems (Camarra et al. 1996, Guérin-Pace et al.,1996, Libourel 1999, Mathian et al., 1999), taking into better account the whole set of available data, in particular social inquiries.

63More generally, proceeding GIS with zonal relations raises the question of continuous or discontinuous spatial organisation of phenomena. With GIS procedures, discontinuities in vegetal landscape are inferred by definition of spatial entities, which are only defined by their boundaries. In reality, ecological factors, which have an influence on vegetation, change progressively (Alexandre et al., 2002, Ernoult et al., 2003). Contact between grasslands and woodlands is also gradual, when woodland is gaining on grasslands (edge effect, Rameau, op. cit.).

64On the contrary, social practices can create discontinuities in landscape, for example between the agricultural plots. However, sheep can follow spatial pattern of vegetation, when they are searching grass inside a plot, or even within a group of plots. Their influence on vegetation is stronger around the paths they follow with the farmer, and can be nil elsewhere. This way, they can be considered as a vector of horizontal interactions between zones (for example, scattering seeds).

65The choice of mapping phenomena, with zonal discrete model, diminishes opportunities to verify these interferences, even if very accurate spatial zoning reduces this bias in the description of reality (Kent et al., 1997).


66The interest of the GIS methods to study spatial relationship between social and natural dynamic phenomena has been confirmed. The GIS is a way to produce such analyses, by intersection of maps with their original attributes, when a multivariate analyses does not manipulate directly the spatial entities derived from maps and images and intersection between them. In this way, GIS has been used here to discuss about a general problem of geography, i-e relationship between social and natural dynamic systems, rather than to aim at a technical feat, as we merely use existent functionalities of software. Our first objective has been completed, as GIS allowed to measure the spatial part of driving factors of landscape dynamics, i-e grazing evolution, cutting and burning frequency, and shows the influence of additional spatial factors, as neighbourhood and edge effect.

67The GIS procedures (spatial masks, geographic joint, half-joint…) enable the creation of new spatial entities with attributes inherited from combination of former ones, and let original spatial structures appear by a stepwise spatial analyses. Frequently, these spatial structures are similar to well known models, as nucleation dynamics, polarisation, or more unexpected ones, as regional effect (village boundaries). Combination between social and natural spatial structures can emphasise these general models or get them confused ; in the second case, spatial pattern is just a local combination.

68We observed, as a general result, that spatial pattern of ligneous overgrow is magnified when crossed with an anthropic relation or attribute, which let us think that this social organisation is underlying the one of dynamics of ligneous cover, highlighting its influence on it, as correlation coefficient showed it formerly (Cohen & Alexandre, 1997) and as pointed out in other regions (Alard & Poudevigne op. cit.). Moreover, the accurate regionalisation obtained by stepwise GIS procedures allows to give more and more emphasis to the influence of individual farmers, their strategy and projects, in function of the location of their lands, and how it influences the control -or uncontrol- of ligneous overgrow. Another important factor, linked with the former one, is the appropriation of land (Turner et al., op. cit.).

69On another hand, taking into account prospective ligneous overgrow gives much influence to biologic and chorologic factors, although a strong part of to-day social practices on ligneous seedlings scattering was expected. Moreover, vegetation management by farmers appears not so efficient for the future as for the past time. That is the main restriction for validating our hypotheses upon social driving factors of ligneous overgrow control. Another discrepancy with general model of analyses calls upon longer temporalities than studied phenomena (ancient land use during the XIXth century).

70Long-lasting control of ligneous overgrow demands some special social practices. Moderate grazing should be sufficient on ancient crop or ancient cattle breeding plots. Elsewhere, cutting and burning, if they are realised with much care, time and spatial consistency, represent an efficient association with grazing. In other cases, these social practices are no more intensive enough, releasing ligneous overgrow and closing of landscape. Those results draw perspectives about the management of these areas, showing a few adequate combinations of social practices to sustainable open landscapes, necessary for maintaining sheep-breeding activity in these regions.

Haut de page


ACHERAR M., LEPART J., DEBUSSCHE M., 1984. La colonisation des friches par le pin d’Alep (Pinus halepensis Miller) en Languedoc méditerranéen. Acta Oecologica. Oecol. Plant., 5 (19) : 179-189.

ALEXANDRE F., COHEN M., GENIN A., LECOMPTE M. et coll., 2002. Essai phytogéographique dans les Alpes occidentales, entre Rhône et Pô, Espace géographique, Tome 31, n° 2, pp. 153-162.

Barbaro L., 1999. Dynamique agro-écologique des communautés de pelouses sèches calcicoles du Vercors méridional : application à la gestion conservatoire de la biodiversité par le pastoralisme, Thèse de Biologie, Univ. J. Fournier, Grenoble, 179 p.

Baudry J.,Le Cœur D., Burel F., Thenail C., 2000. A holistic landscape ecological study ot the interactions between farm-activities and ecological patterns in Brittany, France, Landscape and urban planning, 50, n° 1-3, pp.119-128.

Bertrand C. & G., 2002, Une géographie traversière. L’environnement à travers territoires et temporalités, Paris, Arguments, 311 p.

Bertrand G., 1975. Pour une histoire écologique de la France rurale, in Duby G. (Eds), Histoire de la France rurale, Ed. du Seuil, Paris : 34-113.

Braque R., 1988. Biogéographie des continents, Paris, Masson, 470 p.

BRUN A., CHASSANY J.P., OSTY P.L., PETIT F.E., NAERT B., 1978. Le Causse Méjan, Ed. INRA-ESR, 5 tomes.

BRUNET R., 2001. Models in geography ? A sense to research, XIIth European colloquium on Theoretical and Quantitative Geography, St Valery en Caux, 7-11 septembre 2001, Informative programme and abstracts, pp.4-19.

BUCHE P., KING D., LARDON S. (Eds), 1992. Gestion de l'espace rural et SIG, Ed. INRA. 421 p.

Burel F., Baudry J.1990. Structural dynamics of a hedgerow network landscape in Brittany, France, Landscape ecology, 4, n° 4, pp. 197-210.

Camarra A., FERREIRA F., CASTRO P, 1996. Spatial simulation modelling, in Fischer, Scholten, Unwin (eds.), Spatial analytical perspectives in GIS, GISDATA n° 4, Taylor & Francis, pp.229-244.

Cheylan J.P., Lardon S., Mathian H., Sanders L., 1994. Les problématiques liées au temps dans les SIG, Revue Internationale de géomatique, vol.4, n° 3-4, pp. 287-305.

COHEN M. (Dir.), 2003. La brousse et le berger. Une approche interdisciplinaire de l’embroussaillement des parcours, Paris CNRS Editions, 356 p.

COHEN M., 2001. La nature est-elle vraiment naturelle ? Monde biophysique, nature et société. Géoforum de Saint-Denis, Nature et Ville, Géographes associés n° 25, pp.67-71.

Cohen M., Alexandre F. et coll., 1997. Embroussaillement, pratiques et représentations sociales : une recherche de corrélations, Nature, Sciences, Sociétés, Vol.5, n° 1 : 31-44.

COHEN M., ALEXANDRE F., MATHIEU N., 1998. Modelling interactions between biophysical and social systems : the example of bush progression in Causse Méjan, 6 pages + figures, Cybergeo, rubrique Rostock.

COHEN M., ALEXANDRE F., MERING C., 1999. Boisement naturel et embroussaillement sur le causse Méjan, intérêt d’une approche croisée entre sciences de la nature et sciences sociales, Ingénieries - EAT, pp. 91-103.

COHEN M., FRIEDBERG C., MATHIEU N., 1997. Pratiques et représentations de la forêt chez les éleveurs du Causse Méjan, "La Forêt : perceptions et représentations", textes réunis par A.Corvol, P.Arnoult & M.Hotyat, Ed. L'Harmattan, pp. 289-302.

COHEN M., PETIT F.E., 1995. Les pratiques pastorales sur le causse Méjan et leur évolution récente, in Bonniol J.L., Saussol A. (Ed) Grands Causses : nouveaux enjeux, nouveaux regards, en hommage à Paul Marres, Millau, Ed. Causses & Cévennes, pp.207-219.

COURET D., 1994. Système d’information géographique, inégalités dans le logement et ségrégation spatiale à Quito (Equateur), Ed. Orstom, Coll. Etudes et thèses, Thèse de géographie Université de Rouen, 235 p.

COURET D., 1996. Système d’information urbain ou système de penser la ville ? Réflexions autour d’un projet de mise en perspective de l’information pour l’observation du changement urbain à Abidjan, in Minvielle J.P. (Eds), Systèmes d’information et questions de développement, Cahiers des Sciences humaines, Vol.32, n° 4, pp.787-802.

DÉRIOZ P., 1994. Friches et terres marginales en basse et moyenne montagne, Revers sud-est du Massif Central, Structure et dynamique spatiale n° 1, Univ. Avignon/Lab.SDS, PUF Avignon, 330p.

Dettmers R., Bart J., 1999. A GIS modeling method to predicting forest songbird habitat, Ecological applications, 9, pp. 152-163.

Durand-Dastes F., 1985. A propos du déterminisme, la complexité des options possibles, Espace géographique 2 : 105-108.

Faerber J., 1995. Le feu contre la friche. Dynamique des milieux, maîtrise du feu et gestion de l’environnement dans les Pyrénées centrales et occidentales, Thèse de géographie, Univ. de Toulouse, 363p.

FERRIER J.P., 2001, Man and woman in the XXIth century : concepts; measures and sustainable living, Communication XIIth European colloquium on theoretical and quantitative geography, 7-11 September 2001, St Valery en Caux.

Franzreb K.E, Llyod F.T, 2000. Integration of long-term research into a GIS based landscape habitat model for the red-cockades woodpecker, Studies in avian biology, 21, pp. 65-74.

Friedberg C., 1997. Diversité, ordre et unité du vivant dans les savoirs populaires, Nature, Sciences, Sociétés, Vol.5, n° 1 : 5-17.

Forman R.T.T., Godron M., 1986, Landscape ecology, New York, John Wyley & Sons, 619 p.

Girard N., Moulin H.G., Osty P.L., Bellon S., Hubert B., Lardon S., 2001. Categorising combinations of farmer’s land use practices : an approach based on examples of sheep farms in the South of France, Agronomie, 21, n° 5, pp.453-459.

GODRON M., DAGET P., LONG G., SAUVAGE C., EMBERGER L., LE FLOCH’ E., POISSONET J., WACQUANT J.P., 1983. Code pour le relevé méthodique de la végétation et du milieu, Ed. CNRS, 292 p.

Gough M.C, Roshton S.P., 2000. The application of GIS-modelling to mustelid landscape ecology, Mammal review, 30, n° 3-4, pp. 197-216.

Guérin-Pace F. MATHIAN H., PUMAIN D., SANDERS L., BURA S ., 1996. Les systèmes multi-agents pour modéliser l’émergence des réseaux urbains, in Boquet, Courgeau, Pumain (eds.), Analyse spatiale de données démographiques, INED-John Libbey EUROTEXT, pp.281-309.

Haines-Young R., Green D.R., Cousins H. (Eds),1993. Landscape ecology and GIS, Taylor and Francis London.

Harms W.B.,1999. Landscape fragmentation by urbanization in the Netherlands : options and ecological consequences, J. Of environmental science, 11, n° 2, pp.141-148.

Jakubauskas M.E., Lulla K.P., Mausel W., 1990. Assessment of vegetation change in a fire-altered forest landscape, Photogrammetric engeneering and remote sensing. 56, pp.371-377.

Julve Ph., 1985. Réflexions sur la structure et la dynamique des lisières forestières. Conséquences sur le système, Colloques phytosociologiques XIV : Phytosociologie et foresterie, Nancy : 55-79.

KENT M., GILL W.J., WEAVER R.E., ARMITAGE R.P., 1997. Lanscape and plant community boundaries in biogeography, Progress en physical geography, 21,3,pp. 315-353.

Lardon S., LIBOUREL T., CHEYLA J.P., 1999. Concevoir la dynamique des entités spatio-temporelles, Revue internationale de géomatique, vol.9, n° 1, pp. 45-55.

Lardon S., Osty P.L., Triboulet P., 1995. Elevage et éleveurs du Causse Méjan (Lozère) : dynamique de la mise en valeur et contrôle de l’espace, in Bonniol J.L., Saussol A. (Eds), Grands Causses : nouveaux enjeux, nouveaux regards, en hommage à Paul Marres, pp. 219-242, Millau, Ed. Causses & Cévennes.

LAURINI R., MILLERET-RAFFORT F., 1993. Les bases de données en géomatique, Paris Hermès, 340 p.

Lepart J., Escarré J., 1983. La succession végétale, mécanismes et modèles : analyse bibliographique, Bull. Ecol. T. 14, 3, pp. 133-178.

Libourel T.,1999. How databases perform changes ?, in F.A.Raper & J.P.Cheylan (eds.), Foramlising and representing spatial-temporal change in spatial socio-economic units, London : Taylor & Francis, GISDATA Series.

Martin D.J., 1999. Spatial representation: the social scientists perspective. In Longley P.A., Goodchild, M.F., Maguire D.J. and Rhind DW (Eds) Geographical Information Systems, Volume I, Principles and Technical Issues, Wiley and Sons, Chichester.

Mast J.N., Veblen T.T., Hodgson M.E., 1997. Tree invasion within a pine:grassland ecotone : an approach with historic aerial photography and GIS modeling, Forest ecology and management, 93, pp. 181-194.

Mathian H., SANDERS L., MIKULA D., 1999. Modelling the dynamics of spatial systems within a GIS : problems and perspectives, in Fortheringham Wegener (eds.), Spatial modelling and GIS : new potential for new models ?Taylor & Francis.

MATHIEU N., ROBIC M.C., 2001. Géographie et durabilité : redéployer une expérience et mobiliser de nouveaux savoir-faire, in Jollivet M. (Eds), Le concept de développement durable, Ed. Elzevier, 288 p.

Metailié J.P., 1981. Le feu pastoral dans les Pyrénées centrales (Barousse, Oueil, Larboust), Paris, CNRS, 294 p.

METAILIE J.P., BUFFIERE D., FAERBER J., LE CARO Ph., 1995. Local comitees of prescribed burning and rangeland management in the Central French Pyrenees, Vth International Rangeland Congress « Rangelands in a Sustainable Biosphere », Salt Lake City, 07 : 23-28, poster 3p.,1fig.

Muller S.,1996. Exposé introductif au colloque « Biodiversité et gestion des écosystèmes prairiaux » : déterminisme et évolution de la biodiversité dans les écosystèmes prairiaux, Acta bot. Gallica, 143 : 233-238.

Noy Meier I., 1995. Interactive effects of fire and grazing on structure and diversity of Mediterranean grasslands, J. Vegetation Sci. 6, pp. 701-710.

Osborne P.E., ALONSO J.C., BRYANT R.G., 2001, Modelling landscape-scale habitat using GIS and remote sensing : a case study with rgeat bustards, Journal of applied ecology, 38, pp.458-471.

OSTY P.L., AURICOSTE C., 1989. Une image des élevages du Causse : évolutions récentes (1975-1983) et questions pour l'avenir, Annales du Parc National des Cévennes, Tome 4 : 15-54.

Ouin A., Burel F., Paillat G., 2000. Spatial dynamics of wood mouse (Apodemus sylvaticus) in an agricultural landscape under intensive use in the Mont Saint Michel Bay (France), Agriculture,ecosystems and environment, 78, n° 2, pp. 159-165.

OULEBSIR D., 1997. Apport d’un SIG à l’étude de l ’embroussaillement sur le Causse Méjan, mémoire de maîtrise co-dirigé par Mmes Mering et Cohen, UFR GHSS, 155 pages + annexes.

Paquette S., Domon G., 1997. The transformation of the agroforestry landscape in the 19th century : a case study in southern Quebec (Canada), Landscape and urban planning, 37, n° 3-4, pp. 197-209.

Parresol B.R., McCollum J.,1997. Characterizing and comparing landscape diversity, using GIS and a contagion index, Journal of sustainable forestry, ½, pp.249-261.

PINCHEMEL P. & G., 1988. La face de la terre. Eléments de géographie, Paris, A. Colin, 519 p.

ERNOULT A., BUREAU F., POUDEVIGNE L., 2003. Patterns of organisation in changing landscapes : implications for the management of biodiversity, Landscape ecology, Vol.18, n° 3, pp. 239-251.

Poudevigne L., Alard D., Van Rooij S., Morinp., 1997. Dynamics of rural landscapes and their main driving factors : a case study in the Seine Valley, Normandy, France, Landscape and urban planning, 38, n° 1-2, pp. 97-103.

PUMAIN D., 2001. Communicating about theoretical géography, Conférence inaugurale, XIIth European colloquium on Theoretical and Quantitative Geography, St Valery en Caux, 7-11 septembre 2001.

RAHMI M., RABANT E., CAMBREZY L., ABDI M.M., 1999. Demography, data processing and cartography, in Environment, cartography, demography and geographic information systems in the refugee camps Oadaab, Kakuma-Kenya 3, Ed. IRD Nairobi (Kenya)-Bondy (France), 124 p.

RAMEAU J.C., 1999, Accrus, successions végétales et modèles de dynamique linéaire forestière, Ingénieries EAT, n° spécial Boisement naturel des terres agricoles en déprise : 33-48.

Rameau J.C., Mansion D., Dumé G., 1989. Flore forestière française. Guide écologique illustré. Tome 1 : Plaines et collines, Ed. I.D.F.-Minist. de l’Agriculture, 1785 p.

Raphael M.G., McKelvey K.S., Galleher B.M., 1998. Using geographic information systems and spatially explicit population models for avian conservation : a case study, in Marzluff J.M. & Sallabanks R. (Eds), Avian conservation : research and management, Island Press, Wasington D.C., pp.65-74.

Ripple W.J.(Ed.),1994. The GIS application book : example in natural ressources, a compendium, Bethesda, ASPRS.

Rustin S.P., Lurz P.W.W., Fuller R., Garson P.J., 1997. Modelling the distribution of the red and grey squirrel at the lanscape scale : a combined GIS and population dynamics approach, Journal of applied ecology, 34, pp. 1137-1154.

SOURIS M., 1995. Savane. Manuel de référence, IRD, 231 p.

Sperduto M.B., Congalton R.G., 1996. Predicting rare orchid (small whorled pogonia) habitat using GIS, Photogrammetric Engeneering and Remote sensing, 62, pp. 1269-1279.

Suarez-Seoane S., Baudry J., 2002. Scale dependance of spatial patterns and cartography on the detection of landscape change : relashionship with species’ perception, Ecography 25, n° 4, pp. 499-511.

TOMLIN D., 1990. Geographic Information System and Cartographic modelling, Ed. Prentice Hall.

Turner M.G., Wear D.N., Flamm R.O., 1996. Land ownership and land-cover change in the southern appalachian highlands and the Olympic peninsula, Ecological applications, 6, n° 4, pp. 1150-1172.

Wilson A.G., 1971. A family of spatial interaction models and associated developments, Environment and Planning A., Vol 3, pp1-32

Zhou Z.Z., Cai M.T., XU Y.T.,1999. Land-use and landscape dynamics in a rural area, Forest research, 12, n° 6, pp.599-605.

Haut de page


1  We do not consider here historical practices, which take place in the napoleonian land register, and which influence has been already shown (Cohen et al., 1997, 1998, 1999, 2003, Oulebsir 1997).

2  This vegetal species is considered as a « nomad » (Rameau et al., 1979), i-e either with a certain capacity of colonisation, in special conditions, or more long-lasting and stress tolerant.

3  Nevertheless, the choice of zonal mapping creates also an artefact, putting emphasis on spatial discontinuities rather than on continuities. We try to minimise this inconvenient mapping transitions areas too (cf. final discussion).

4  For example, on west side of the studied area, the large stable zone is covered on both dates by grasslands ; but farmers inquiries teach us that during the period, pine thicket grows, has been burned, and turned back to grasslands.

5  This index has been previously tested in multivariate analyses, showing good correlation with vegetation (Cohen & Alexandre, op. cit.), the mere sheep load is less accurate (Cohen, Alexandre & Mering, 1999).

6  The sheep load is too much weak (<1 sheep/ha.) to have an influence in such strong regression of ligneous cover.

7  Despite a very carefull digitalization (Oulebsir, op. cit.), we could not avoid very small discrepancies in the superposition of the two relations vegetation 1963 and vegetation 1994.

Haut de page

Pour citer cet article

Référence électronique

Marianne Cohen et Catherine Mering, « Geographic information system and spatial effect », Cybergeo : European Journal of Geography [En ligne], Dossiers, document 272, mis en ligne le 25 mai 2004, consulté le 22 août 2018. URL : ; DOI : 10.4000/cybergeo.3616

Haut de page


Marianne Cohen

umr Ladyss, Université Paris 7-Pôle Image

Articles du même auteur

Catherine Mering

umr Prodig, Université Paris 7-Pôle Image

Articles du même auteur

Haut de page

Droits d’auteur

© CNRS-UMR Géographie-cités 8504

Haut de page