Vegetation ecological studies at the lower course of Sabor River (Tras-os-Montes, NE-Portugal)

Contents

1. Introduction page

2. Methodology
2.1 Study area
2.1.1 General considerations
2.1.2 Topography
2.1.3 Climate
2.1.4 Hydrology
2.1.5 Geology
2.1.6 Soils
2.1.7 Bioclimatic description
2.1.8 Biogeography and Vegetation
2.1.9 Land use
2.2 Sampling
2.3 Data analysis
2.3.1 Species richness
2.3.2 Floristic-structural analysis
Arrangement of basic structural matrices
- Diversity BSM
- Abundance BSM
- Stratification BSM
- Conjoint matrix of diversity and abundance
- Conjoint matrix of diversity, abundance
and stratification
Arrangement of contingency matrices
- Life form contingency matrix
- Frequency class contingency matrix
Cluster analysis
Ordination by Principal Component Analysis
- Clusters
- Distribution of clusters over apparent
communities, land use intensity and geographic sectors
- Distribution of life forms
- Distribution of frequency classes
- Maximal Expressive Amplitudes
Discriminant Canonical Analysis
Assessment of intensity of land use

3. Results
3.1 Species richness
3.2 Floristic-structural analysis
Cluster analysis with BSM
Ordination by PCA
Correlation Clusters (Tendencies of Behaviour)
- Diversity
- Abundance
- Stratification
- BSMDA
- BSMDAS
Expressive Amplitudes
Life form, frequency classes and expressivity
in apparent communities
Discriminant analyses
- Frequency classes by community types
- Life forms by community types
- Frequency classes by land use categories
- Life forms by sectors
- Frequency classes by sectors
Assessment of land use intensity

4. Discussion
4.1 Species richness
4.2 Floristic-structural analysis

5. Literature cited

Appendices
Appendix 1: Definition of the geographical sectors
Appendix 2: Sample plots
Appendix 3: Species catalogue

1. Introduction

Considerable effort has been made to valúate the biotic diversity arising on a spatial and temporal scale within areas subjected to anthropized cultivations, and the degree of species richness in those areas, as well as the benefits proceeding from them, are often underestimated (Paoletti et al., 1992; Paoletti & Pimentel, 1992). These ecosystems, commonly referred to as agroecosystems, have been defined as «an interactive group of biotic and abiotic components, some of which are under human control, that forms a unified whole» (Elliot & Cole, 1989 Agroecosytems are also considered as decisive interfaces between the anthropic and the biotic components within the framework of landscape ecology (Fig 1.1). This implies the recognition of the dynamic role of man as a forming element by means of land use, reflecting the spatial intervention on ecosystems with the objective of adapting them to man’s needs; thus quests for more systematic investigations of the ecological implications arose (Bolòs, 1992; Naveh & Lieberman, 1990; Vilàs, 1992).

illustration not visible in this excerpt

Fig. 1.1. Agroecosystems as linkage between the elements that constitute landscapes. Adapted from Bolòs (1992).

Evidence exists that biotic diversity can increase with the complexity of landscape structure resulting from traditional agricultural systems (Altieri et al., 1987; Gliessman, 1990). In most cases, however, agroecosystems and their parental ecosystems are often substantially different in structure and function, as often becomes manifest through reduced species diversity, lower resilience or greater homogeneity, amongst others, in the derived ecosystems (Altieri et al., 1983).

A trait that distinguishes these systems from most other natural systems are the periodic and chronic disturbances intrinsic to agricultural management, leading to changes with different effects, according to their spatial extension; for example, changes at local scale are supposed to affect dominant species, structure and boundaries between different patches of vegetation, whereas at a regional scale alterations in land use or changes in the arrangement of basic landscape-forming units are expected consequences (Elliot & Cole, 1989; Lepart & Debussche, 1992; Okey, 1996).

In the functional context of agroecosystems, riparian vegetation communities are particularly interesting, because they exhibit a high degree of structural and compositional diversity (Crespí et al., 2001b). Riparian vegetation occupies one of the most dynamic areas of landscape, and the distribution as well as the composition of riparian plant communities reflects history of both fluvial disturbance regimes (floods) and nonfluvial disturbance regimes of adjacent upland areas (Gregory et al., 1991; see Fig. 1.2). Riverine landscapes are constituted by extensive interconnected series of biotopes and environmental gradients, and natural disturbance regimes maintain a multiplicity of interactive pathways across the whole system, so that these landscapes need to be envisaged in a holistic geomorphic perspective. The action of disturbance and environmental gradients in concert generates broad-scale patterns and processes responsible for elevated level of biodiversity (Ward, 1998).

illustration not visible in this excerpt

Fig. 1.2. General classification of zones composing the riverine landscape. Zones 2 and 3 make up together the complete riparian zone of influence. Key: (1) active channel, (2) riparian zone, (3) zone of influence, (4) upland area. Adapted from Naiman et al. (1992)

The particular interest of riparian vegetation resides in the fact that on a large scale, alluvial landscapes present a continuum from flooding-prone to particularly flood-protected sites, thus assuring refuges and propagule sources at any given time even under changing situations and limiting species loss to at most local levels (Schnitzler, 1997). Additionally, higher fertility as a result of periodical flooding in many of those habitats contributes to high species richness, productivity and a high potential of coexistence. Because the microrelief particularities and local distribution patterns of those sites are expected to create distinct patches of vegetation units, various stages of succession are supposed to exist, ranging from innovation stages on sites exposed to high kinetic energy of floods (characterized by patches of pioneer herbaceous, shrub and tree species), over equilibrium and climactic stages with a highly structured functioning and fragmentation into mosaic units (caused by different maturity, decay or death of individuals), to elimination stages. For all the aforementioned reasons, riparian zones exercise potentially important roles within landscapes as corridors for plant dispersal and additional importance arises during periods of rapid environmental change because of frequently ameliorated microclimates along the river valleys (Franklin, 1992; Gregory et al., 1991).

The subject of the present study is the Mediterranean agroecosystem located in the valley of the lower course of Sabor river, northeastern Portugal. This agroecosystem is characterized by a profound adaptation of its landscape to a large-scale agriculture that had continued duration up to the beginning of the 1970s (cf. also chapter 2), and is now in a state of apparent recuperation over great parts of its area. Interestingly, the studied area comprises a series of habitats and sites the conservational value of which had been stated on various occasions: parts of the river course had been classified as CORINE biotope (code C11800100 Rio Sabor), comprising an area of 6.400 hectares exclusively of wetlands; the river valley of the lower course includes also a series of habitats of communitarian interest, in the sense of the Directive Habitats 92/42/CE, having inclusively being proposed for integration in the European habitat network NATURA 2000 (Farinha & Trindade, 1994; Koe et al., 1998; Ministerio do Ambiente, 1999; Romäo, 1992).

However, the project of construction of a hydroelectric power plant located at the lower course of this river, dating back to the 1950s (Hidro­eléctrica do Douro, 1961), has recently gained renewed interest on part of energy suppliers and local power. Although an assessment of environmental impact has been elaborated, there exists practically no information of public access concerning the values and potentials of the specific landscape of this river valley.

Among the heavy potential impacts ensuing such a profound anthropogenic flow regulation, the following can be mentioned (Ward, 1998):

- disruption of natural disturbance regimes
- truncation of environmental gradients
- severing of interactive pathways
- elimination of upstream-downstream linkages

These are all supposed to interfere with successional trajectories, habitat diversity and migration pathways, leading ultimately to drastically altered landscape functions and reduced biodiversity. Considering the dimensions of the before mentioned project, the need for studies at a large spatial scale to elucidate the structure and potentials of vegetation seems sufficiently evident, all the more as the ecosystem in concern is exposed to a climate type that promotes erosion, thus being at utmost vulnerability to suffer from irreversible effects of large scale impacts.

With this background, the purpose of the present work is to apply an analytical methodology to describe the floristic-structural behaviour of the vegetation of this agroecosystem. The methodology has already been applied to various situations in several ecosystems in northern Portugal (Crespi et al., 2000, 2001a, 2001b, 2001c; Fernandes, 2001; PIOT-ADV, unpubl.), and consists in a multivariate statistical analysis of structural matrices based on a set of relevés representing the distinct community types present in the study region, in order to enable the characterization of the floristic-structural dynamics of the vegetation in the considered zone. structure is in this context envisaged as a product of population dynamics dependent upon the competitive capacities of each species (i.e. according to an individualistic, Gleasonian viewpoint, e.g. O’Neill et al., 1986), so that the structural dynamics of the communities could be explained by interactions between the different populations within the given context of their metapopulations (Holyoak & Ray, 1999; Husband & Barrett, 1996) on the one hand, and ecological factors on the other hand, resulting eventually in chaotic processes subjacent to the structural behaviour of these communities and a multiplicity of stable states (May, 1976, 1977; Stone, 1993; Stone & Ezrati, 1996; Tilman & Wedin, 1991). Consequently, in addition to the dynamics between distinct communities, also dynamics at an intracommunitarian level could be expected, notable through a higher degree of variability of floristic-structural combinations within a given type of community, providing, simultaneously, an estimation of resilience by analysing this combinatorial variability (Crespi et al., 2001a, 2001b).

Reference to resilience is especially interesting while considering the effects of anthropogenic disturbance on the structure of ecosystems, together with resistance, i.e. the ability to withstand shifts induced by perturbations. Resilience constitutes an important property of ecosystems, which is underlying to a more or less rapid restoration of initial structures and functions after disturbance having ceased (Allen, 2001; Batabyal, 1998; Okey, 1996; Westman, 1986; Whitford et al., 1999).

Namely in Mediterranean regions, the anthropogenic impact has not always led to drastic decrease in resilience of native vegetation, and a large variety of possible landscapes remained, thus maintaining high levels of biodiversity (Lepart & Debussche, 1992). Nevertheless, resilience seems to depend, in a complex way, upon the diversity of landscape, defined as the combination of number of vegetation types and the degree of spatial heterogeneity, with heterogeneity being likely to result in faster recuperation (Bascompte & Rodríguez, 2001; Cumming, 2002; Lavorel, 1999). Consequently any kind of perturbation drastically affecting the diversity of landscape would be expected to reduce significantly the resilience of an ecosystem. Species and functional diversity are also supposed to be important in the context of ecosystems’ performance and resilience to environmental changes, but there is still a great demand for elucidation of the relative roles of these parameters (Lavorel et al., 1998).

These considerations constitute an outline of the purposes of the present study, focussing the impacts of agricultural activities on the diversity of floristic-structural combinations in relation to a more pristine vegetation type, as has been studied in the case of another agroecosystem (Crespi et al., 2001c).

2. Methodology

2.1 Study area

2.1.1 General considerations

The whole area considered in the present study is located in the valley of the lower course of the Sabor river and pertains to the region of Trás-os- Montes e Alto Douro (Fig. 2.1). This region corresponds to the northeastern area of the continental Portuguese territory, and is circumscribed by the district of Bragança. It is part of the Portuguese portion of the Douro river drainage basin, and the Sabor river, traversing the region in a roughly North to South orientation, is the first right-bank tributary of the former on Portuguese territory. The total surface of the Sabor drainage basin amounts to 3 830 km[2] (Hidro-Eléctrica do Douro, 1961).

illustration not visible in this excerpt

Fig. 2.1. Localisation of the study area in northeastern Portugal.

2.1.2.Topography

The region is characterized by the persistence of residual primitive plateaus (peneplains) with average altitudes of 700 - 800 m, and is rich in erosion surfaces dissected by a drainage network. Water courses are enclosed in deep and narrow valleys that present frequently steep slopes, with the valley of Sabor river and its main tributaries standing out, i.e. Maças, Angueira and Azibo rivers, as well as the rivers Ribeira de Zacarías and Ribeira de Vilariça (Cabral, 1985).

The eastern parts of the region can be considered as a prolongation of the Iberian meseta, interrupted north of the central chain by the tectonic accident designated as Manteigas-Vilariça-Bragança, of late variscian origin, following a general NNE-SSW orientation, which after neotectonic reactivations during the late Cenozoic formed a step between the northern Meseta and the higher plains of Trás-os-Montes (Cabral, 1995). Thus the surface of northeastern Trás-os-Montes was elevated to altitudes in the range of 600 to 800 metres above sea level. Unlike the peneplain on the Spanish side, with low relief and Tertiary cover, the plateau on the Portuguese side is deeply incised by its rivers, which have carved canyons of considerable depth into the ancient crystalline rocks; the dissection of the actual drainage network is supposed to date back to times as recent as the beginnings of Quaternary only (Medeiros, 1987; Stanislavski, 1959). To illustrate these erosive forces, the average altitude of the Sabor drainage basin area amounts to 660 m (Hidro-Eléctrica do Douro, 1961), whereas, within the considered area (cf. 2.2), the level of Sabor’s riverbed decreases from 200 m to 140 m.

illustration not visible in this excerpt

Fig. 2.2. Altitudes (A) and slopes (B) in northeastern Trás-os-Montes. The zone comprising the sample sites is outlined. (SNIRH, 2003)

2.1.3 Climate

Portugal as a whole is a region of transition between Atlantic and Mediterranean influence (Ribeiro, 1991). Although the northern parts of the Portuguese territory commonly are thought to be dominated by Atlantic climate, namely the eastern part of the province, protected by several high mountain chains - Gerês, Barroso, Marao, Alvao, Montemuro - has a marked continental character. The profoundly carved valley of Douro river favours the progression, from east to west, of a continental type of climate (Medeiros, 1987). Especially in the lower parts of upstream Douro valley, the protected position enables the existence of a Mediterranean climate (Alcoforado et al., 1993).

Considering various climatic and physiographic characteristics of Trás- os-Montes, climatically homogeneous zones have been established, concerning the intersection of thermic and precipitation regimes (Agroconsultores & COBA, 1991). Three main sub-zones are considered, by analogy to local designation: Terra Fria (cold land), Terra de Transigäo (transition land) and Terra Quente (warm land). The cold land, covering mainly the higher surfaces comprised between 600 and 1300 m, where mean annual rainfall ranges from 600 to 1400 mm, generally tends to have longer winters and snow may fall from early December to March; frosty days may occur even in May (Ferreira et al., 1996). This zone, however, is not represented in the sites considered in this study.

The complete study area is inserted in the Terra Quente zone, which represents areas with altitudes from sea level to around 500 m, thus in general incident with the slopes of river valleys. Winters tend to be shorter and milder and snowfall is rare, and the latest frosts use to occur only in March. According to the precipitation regime, this zone can be subdivided into further sub-zones; the only situation, however, occurring in the study area corresponds to annual rainfalls inferior to 600 mm. The transition zone is located at altitudes ranging from 400-500 to 600-700 m. Temperature and rainfall are similar to that found in the warm land, with the only difference of cooler summers. Table 2.1 resumes some climatic characteristics.

illustration not visible in this excerpt

Table 2.1. Some basic characteristics of climatically homogeneous zones of northeastern Portugal. Compiled after Agroconsultores & COBA (1991) and Ferreira et al. (1996).

In global terms, the mountain areas receive higher rainfalls than lower lands, and, on the other hand, the temporal distribution of rainfall is high, so that high differences occur between the wettest and the driest months; there are also significant differences in mean air temperature between the coldest and hottest months at all locations in the region. Winter rainfalls represent about 70-80% of the total annual precipitations, and a marked hydric deficit is notable during the months June to September (Ferreira et al., 1996; see also Fig. 2.3, and Tables 2.2, 2.3).

illustration not visible in this excerpt

Tab. 2.2. Rainfall and temperature data of some stations in the study region. In bold letters, stations located within or in proximity to the study area. Alt = altitude; T = mean annual temperature; P = mean annual rainfall.

illustration not visible in this excerpt

Fig. 2.3. A) Thermopluviometric diagram of station Torre de Moncorvo, best representing the conditions within the studied zone ( 41°10’ N, 7° 03’ W, 408 m). Rainfall data corresponds to the period 1941-70 (INMG, 1990), whereas the temperature data corresponds to the period 1925-54 (SMN, 1970). The mean total rainfall was 549 mm, the mean annual temperature 15,2 °C. B) Rainfall in northeastern Trás-os-Montes (SNIRH, 2003).

The climate of the region has basically a subcontinental character, with predominant climatic characteristics of the before mentioned transition land, as the major part is constituted by plateau areas; at the slope of the Sabor river valley, climate corresponds to the warm land-type. Table 2.3 shows a synthesis of basic climatic parameters of the zone.

illustration not visible in this excerpt

Tab. 2.3. Climatic parameters of the study area according to orographic position. T = mean annual temperature; Tmax = annual mean of maximal temperatures; Tmin = annual mean of minimal temperatures; P = mean annual rainfall; Hrel = mean annual relative humidity; I = mean annual insolation. Compiled after Agroconsultores & COBA (1991), based on Instituto Nacional de Meteorologia e Geofísica data, referring to the period 1951-80.

2.1.4 Hydrology

The rainfall regime (mean annual rainfall over all the Sabor drainage basin in the period 1941-56 was 730 mm) is reflected in a comparably low mean specific flow rate of Sabor river, which equalled 6,6 l s-[1] km-[2] in the same period (Hidro-Eléctrica do Douro, 1961). However, the low flow rate does not reflect the great variability of flow: during the winter floods, the volume of water flow in the river beds of Douro can increase up to 770 times compared with summer, leading in the narrower valleys to an increase of water levels of some tens of metres (Medeiros, 1987). These conditions are expected to have strong influence upon vegetation in the floodplain and the lower hillslope of the valley.

2.1.5 Geology

Although the whole considered area pertains to the so called Galician- Castilian zone of the Iberian mass of the variscian orogeny, which is basically made up of granites and orthogneisses, the Douro river and his main tributaries follow a belt of Palaeozoic or pre-Cambrian schists, with some inserted layers of quarzite (Lautensach, 1964). Hence it follows that the research area is mainly composed of schists forming the bedrock of the studied sites (Martins, 1985).

Nearly the whole extension of the river valley in consideration is constituted mainly by metamorphic rocks pertaining to the greywacke-schist complex, Ordovician schists and quartzites in the Moncorvo-subregion, metamorphic Silurian schists in the zone where the basin of Ribeira do Zacarías enters, and some spots of granites in diverse localities. Upstream, the study area is limited by the lithological complex corresponding to the Morais massive, with predominance of greenschists, mica-schists, gneisses, and amphibolites, and the Silurian complex, constituted by metamorphic and quartzitic rocks (Agroconsultores & COBA, 1991; DGMSG, 1974; Ministério da Economia, 2000; see Fig. 2.4).

2.1.6 Soils

The soil types prevailing in the study area are basically leptosols and anthrosols, with small occurrences of fluvisols wherever the extension of the river’s floodplain permits. Leptosols are very thin soils presenting generally only an A horizon of roughly 0,1 m depth, and are limited in this area by a hard rock layer which is nearly invariably schist; those soils are typical of sloping lands and drier areas, as the valley considered here (Agroconsultores & COBA, 1991; Ferreira et al., 1996). Two classes of leptosols occur in Sabor valley: orthi-dystric leptosols and orthi-eutric leptosols, the former based on schist and the latter, in some areas adjoining the upstream limit of the study area, on basic rocks. It is by far the soil unit of major representation in the region; after deep ploughing, those soils are suitable for extensive agriculture as for example cereal cropping and permanent tree crops as olive and almond, supporting equally forests (Agroconsultores & COBA, 1991).

Anthrosols are soils that have suffered profound modification of original characteristics provoked by human activities. The only type represented in the study area, namely in the section Picöes - Felgar, are surribi-aric anthrosols, resulting from profound mobilizations and greater movements of material and from mixtures of original soils with fragments of the substrate rock; they are mainly used for the same crops as the leptosols (Agroconsultores & COBA, 1991).

Fluvisols are composed of sedimentary materials, without any expression of horizons, rich in organic matter (Ferreira et al., 1996); although they are fertile and frequently are irrigated for intensive cropping (as in the nearby Vilariça valley), in Sabor valley they do not have any special use due to their remoteness and difficult access.

2.1.7 Bioclimatic description

According to a comparative study of some aridity scales in Portugal, the study area pertains to a pre-Mediterranean domain (Alcoforado et al., 1993). The Mediterranean climate is characterized by scarcity of rainfalls during summer, the occurrence of at least two months where the numerical value of mean precipitation is inferior to twice the value of mean temperature (P < 2T), although there might be excesses of rainfall in winter (Costa et al., 1998). The Mediterranean character of the vegetation and its correlation with the upper Douro, Sabor and Tua rivers was already stated by Rozeira (1944); he outlined the importance of that rivers for the Mediterranean elements, both in terms of distribution of species and in terms of the modification of climate within their valleys.

Considering the Thermicity index It, and following the classification of Rivas-Martínez (1987,1996) and Rivas-Martínez et al. (2002), the studied zone pertains to a mediterranean thermotype, and of the six bioclimatic stages recognized for the Mediterranean area, three are represented in the studied zone, the thermomediterranean, with 350 < It < 450, the mesomediterranean (210 < It < 350) and the supramediterranean ( 80 < It < 210), and the ombroclimate varies from dry (annual precipitation between 350 and 500 mm) to subhumid (500 < P < 900; Ministerio do Ambiente,1999). As a general rule, the Mediterranean character becomes more accentuated the more the locality is situated in the interior and the lower its altitude is (Ribeiro, 2000a).

illustration not visible in this excerpt

Fig. 2.5. Thermoclimatic belts in northeastern Portugal, with the study area outlined (Rivas-Martinez et al., 2002).

The greatest part of the valley of the lower course of Sabor river pertains to a subhum id mesomediterranean^[1] stage, with expression of a thermomediterranean stage with dry ombroclimate at lowest altitudes (riverbed) within the most downstream portions (Costa et al., 1998; Rivas- Martínez & Loidi, 1999; cf. Fig. 2.5).

2.1.8 Biogeography & Vegetation

Biogeography, as a branch of geography, deals with the spatial distribution of species, linking information about the physical environment with the biological, supported by chorological, geological, bioclimatic and phytosociological data, as plants supply the biggest part of terrestrial biomass, and due to their fixed character (Costa et al., 1998).

One of the traditionally used criterions in the recognition and demarcation of biogeographical areas is thus the distinction and cartography of taxa or syntaxa that present a regionally restricted distribution, i.e. endemisms (Rivas-Martínez, 1987).

In accordance with Rivas-Martínez (1987) and Costa et al. (1998), the biogeographical classification of the region is as follows (see Fig. 2.6):

Holarctic Kingdom

Mediterranean Region

Western Mediterranean Subregion

Ibero-Atlantic Mediterranean Superprovince Carpetan Iberic Leones Province Lusitan Duriensean Sector

Terra Quente Superdistrict

illustration not visible in this excerpt

Fig. 2.6. Biogeography of northeastern Portugal to sector level. The study area is outlined. Key: Eurosiberian Region - 4.4: Juresian Sector.

Mediterranean Region - 15.6: Salmanticensean Sector 15.8: Lusitan Duriensean Sector 15.9: Bercian-Sanabriensean Sector (Rivas-Martínez et al., 2002).

The whole Lusitan Duriensean Sector is a highly complex unity, due to the territory’s physiography, composed of a substrate of meso- or even nearly thermomediterranean plateaus and valleys and some inserted supramediterranean elevations.

The river valleys in the study area are believed to be an important landscape element determining plant migrations and a shelter for thermophilic species during cold periods. Hence various climatic disjunctions can be found there, from xerophytic paleoclimatic relicts as Juniperus oxycedrus, mesophytic paleoclimatic relicts like Arbutus unedo and Buxus sempervirens, to pre-Würmian mediterranean climatic disjunctions like Allium roseum, Asparagus albus, Olea europaea var. sylvestris and Rhamnus lycioides subsp. oleoides (Honrado et al., 2001).

In the following, a generic characterization of the vegetation is given, according to Capelo et al. (1998) and Costa et al. (1998). The Terra Quente Superdistrict is characterized by its climatophilous mixed forests of cork oak and juniper (Rusco aculeati-Quercetum suberis subassociation juniperetosum oxycedri ined.), and its subserial communities, mainly dwarf scrubs of Lavandulo-Cytisetum multiflori and broom-dominated communities of Cytiso multiflori-Retametum sphaerocarpa as forest mantle vegetation and also in progressive successional processes, or Cytiso scoparii-Retametum sphaerocarpa in regressive succession. As a first step in the degradation of cork oak woods, shrubby or woody formations of species with lustrous leaves occur (Arbutus unedo accompanied by thermophilous species like Phillyrea angustifolia, Pistacia terebinthus and Viburnum tinus, or in even more degraded stages, Erica arborea). Cork oak is frequently associated with Q. faginea, as already observed by Braun-Blanquet et al. (1956). Oligotrophic therophytic grasslands are frequent, classified as Anthyllido lusitanicae- Tuberarietum guttati. On more acid and exposed soils, communities of Lavandulo sampaioanae-Cistetum populifolii can be identified.

In edapho-xerophilous conditions, mainly occupying steep slopes, holm oak dominated forests, classified as Genisto hystricis-Quercetum rotundifoliae juniperetosum oxycedri, substitute the cork oak woods. In subserial stages, again broom dominated communities, in that case pertaining to Cytiso multiflori-Retametum sphaerocarpa occur, which might be substituted by extremely poor communities dominated by Cistus ladanifer. Under basiphilous conditions, dwarf scrubs of Lavandulo sampaioanae-Cistetum albidi may occur. The underlying intensive anthropozoic dynamics of these scrubby formations has first been described by Braun-Blanquet et al. (1964).

Bordering the margins of permanent watercourses, with marked edapho-hygrophilous conditions, riparian forest of Scrophulario scorodoniae- Alnetum glutinosae are constant, but they are restricted to very narrow strips of gallery forests, due to the topography of the mainly V-shaped river valley, with Clematis campaniflora and Scrophularia scorodonia as notable companion species. Bordering the temporary watercourses, but also the more lotic facies of permanent watercourses with strong currents, willows of Salicion salviifoliae appear. In the floodplain of Sabor river, rupicolous chamaephytic communities of Diantho laricifolii-Petrorhagietum saxifragae are frequent, with notable thermophilous traits, and a peculiar scrub community dominated by Buxus sempervirens and Erica arborea (cf. also Aguiar et al., 1999).

In the semiarid ombrotype of meso- and thermomediterranean stages, coinciding with the deepest parts of the valley in the most downstream section of Sabor river, there is no expression of the climactic forests at all, but instead remnants of the typical shrubby formations of Pistacio-Rhamnetalia alaterni (Asparago-Rhamnion oleoidis), with the species Asparagus albus and Rhamnus oleioides as bioindicators (Rivas-Madrtínez, 1987), can be observed. A simplified illustration of the distribution of communities is shown in Fig. 2.7.

illustration not visible in this excerpt

Fig. 2.7. Distribution of important communities in the valley of Sabor river (Hoelzer & Aguiar, unpubl.)

2.1.9 Land use

Land use is ancient in the region, with first human settlements dating back to Holocene, and most ancient agriculture came about in the transition between the 6th to the 5th millennium BC, a fact that was equally confirmed for the proximities of the study area (Sanches, 1997); cereal cultivation seems to be usual at least from the late 6th millennium BC on, and from the third millennium BC on there is evidence of livestock raising. Around the early second millennium, it appears that an intensification and diversification in subsistence activities plus an increased number of permanent settlements took place.

The evolution of the cultural landscape for the studied area is, consequently, essentially comparable with that described for the Mediterranean as a whole (Naveh & Lieberman, 1990): after the emergence of hunter-gatherer economies in upper Pleistocene, a gradual intensification of anthropogenic factors (including burning), the Neolithic brought more profound agricultural transformations, with the creation of a diversified flora of species adapted to drought, fire and grazing. The evolution of denser pastoral-agrarian populations implied clearance of arable slopes, often terraced or patchcultivated. Gradually, the dense woody natural landscape was transformed into more open cultural landscape.

In present times, different systems of land use in northeastern Portugal have been identified (Moreira, 1984), two of which being represented in the considered area (Agroconsultores & COBA, 1991). The most important at the slopes of Sabor valley has been the cereal and olive exploration system of Terra Quente (Fig. 2.8), where cereal monoculture - wheat and some rye - is predominating, mainly based on the traditional biannual rotation of crop and fallow. The cereal crops are nowadays without any importance at Sabor valley slopes, completely abandoned since the early 1970s (Ginja, A., personal comm.), but were extremely extended following the "Wheat campaign”, consisting in a forced augmentation of wheat cultivated areas in the whole country as a mean to reduce importations, in the decade of 1930 (Martins, 1985).

illustration not visible in this excerpt

Fig. 2.8. Traditional land use in the Terra Quente zone (adapted from Moreira, 1984).

A common pattern of land use organization found in the region resides in a radial structure of utilization intensity, with the villages constituting the centre, surrounded by horticulture and other intensive crops, then in a greater distance the cereal fields, and finally the woodland and scrubland dominated areas (Aguiar & Rodrigues, 2002). Furthermore, it has to be remarked that the whole traditional system of agriculture of Trás-os-Montes was built upon an efficient utilization of the disposable resources, with insignificant resort to external inputs (Vaz & Vaz, 2000). of outstanding importance within the context of local economy are the orchards of olive and almond trees, either as monoculture or mixed. These are often the only viable crops where slopes are too steep for any other kind of cultivation. Besides, some benefit is provided by the cork oak woodlands, both by spontaneous and by cultivated populations. The olive and almond groves, installed by means of cuttings, are generally treated in an extensive regime; two ploughings and one raking were common (Baltasar, 1989). However, as a result of the lack of manpower in present days, the number of ploughings per year may be reduced to one or even less in most orchards. Nowadays, the current treatment consists of two ploughings in the case of olive groves and generally one ploughing in almond groves (Ginja, A., personal communication). It is worth to remark that in the Trás-os-Montes region namely the extension of the cultivation of olive trees became established only since the 18th century (Ribeiro, 1991), after a previous introduction during the 16th century in the Sabor valley and other zones nearby (Pereira, 1997). From historical sources it is known that the major diffusion of olive cultivation started in the beginning of the 19th century (Mendes, 1994).

Other crops had outstanding importance in past times. Up to the 1970s flax was cultivated, but it was around the middle of the 18th century that the related industry reached its height (Oliveira, 2000). It was cultivated preferably in the proximity of watercourses for fibre extraction. Nowadays, there are practically no areas where flax is cultivated. Hemp for fibre production was equally basis of a well-developed industry in this region, until its substitution by silk industry in the middle of the 18th century (Cabral, 1895). The silk industry, based on important groves of mulberry, had a major expression in the whole region until the end of the 19th century (Cepeda, 1999). Another cultivated crop in the region used in leather industry (tannery) was sumac (Rhus coriaria); nowadays there exist some spontaneous remnant formations dominated by this shrub in some points of Sabor valley.

Out of the immediate proximity of settlements, at the slopes of the valley, punctually horticulture is practised, in small plots adjacent to some small temporary watercourses draining into the Sabor floodplain.

Integral part of the mentioned kind of exploration is the natural pasture by sheep and goats, with the multiple function of meat, milk and wool production, based on spontaneous vegetation and also remainders of pruning as leaves and twigs of the olive and almond trees. This occurs over nearly all the area, with some preference to the floodplain areas during summer, because of scarcity of water.

Another important observation is the existence of ancient communal lands, the so-called baldios, that is large areas where intensive agriculture has never been possible (Oliveira et al., 1995; Ribeiro, 1991). As a consequence of the general decline of the region, following the great emigration movements starting at the beginning of 20th century and becoming most accentuated in the 1960s (Cepeda, 1999), these lands often became transformed into completely unused lands. in the 19th century, an estimated three quarters of the area of Trás-os-Montes were baldios (Ribeiro, 1991). Particular decisive is the fact that mainly since 1938 large scale forest plantations were installed in these communal areas, and given the importance of these areas as an additional subsistence source (episodic crops, apiculture, pasture), violent conflicts rose up, involving forest fires as weapon in a political conflict (Baptista, 1978). Although there is no documentation available for this specific region, especially since the second half of the 1960s an increased tendency for fires in baldio areas becomes evident (Devy-Vareta, 1993).

At last, the somehow scattered distribution of cultivated land in the considered area is equally noteworthy; land tenure is predominantly characterized by small sized farms and large number of plots per family (Ferreira et al., 1996; Gonçalves, 1991). The abandonment of a great number of plots as a consequence of the rural exodus has aggravated the access to the productive fields. This has contributed to an increase of uncultivated areas of great dimensions in the entire region, and affected equally the slopes of Sabor valley.

To sum it up, the whole region of valley of Sabor river was occupied by large scale cultivations, leading to a profound modification of landscape which remained patent up to the 1970s, with a nearly total suppression of uncultivated areas. The present extensive areas of seemingly natural vegetation developed in the relative short time span of the last few decades. Fig. 2.9 shows this alteration of landscape documented by aerial photographs taken in 1965 and 1995, respectively.

illustration not visible in this excerpt

Fig. 2.9. Aerial photographs taken in 1965 (left; ©Direcçao Geral de Serviços Florestais e Agrícolas) and 1995 (right; © Centro Nacional de Informaçâo Geográfica) of the river course at sector Vilar Seco - Salgueiro (Mogadouro). The topmost zone corresponds to the sample sites Quinta do Azinhal/Castro Vicente-Portela of sector Remondes of this study, the area located at the bottom corresponds to sample sites Salgueiro/Legoinha of sector Parada (cf. Appendix 1). The extension of areas covered with not cultivated vegetation has significantly increased over the last 40 years.

2.2 Sampling

Fieldwork was conducted during the months May, June and July 2002. The part of the river valley considered here is limited, upstream, by the bridge of national road EN 216 which crosses the river between the cities of Macedo de Cavaleiros and Mogadouro, also known as "Ponte de Remondes”, with the coordinates 6°48’ west of Greenwich / N 41°24’ (UTM 29TPF835854), and the distance from Sabor river’s mouth amounts to some 45 km. Downstream, the limit of the study area is defined by the section Felgar-Picöes, with the coordinates W 6°59’ / N 41°15’ (UTM 29TPF693680), at a distance of approximately 16 km from the river mouth. Thus, the extension of the river course studied here totals approximately 29 km (Fig. 2.10).

illustration not visible in this excerpt

Fig. 2.10. Map of the sampled area in the downstream section of Sabor river. Sample sites are indicated with green dots, villages are highlighted in red.

The sample plots were arranged in a way to allow for representing the various vegetation communities at different height levels, in a roughly perpendicular orientation in relation to the riverbed, at both sides of the river. A particular difficulty resided in an exact delimitation of entirely homogeneous sample plots, as a result of the complex patchy arrangement of vegetation. In order to permit that the sampled vegetation stands were as homogeneous as possible within each relevé, and at the same time could yield a more valid representation of the local diversity, the dimension of sample plots was constantly fixed at 10 x 10 m^[2]. Localization was proceeded using a compass and topographical maps at the scale 1 : 25 000 (IGE, 1995, 1996) to enable determination of UTM grid coordinates. Altogether, 109 relevés were taken (Appendices 1 and 2).

Each relevé was accompanied by a description of basic topographical parameters (coordinates, altitude, exposition, slope), total cover of each stand was estimated, and all species occurring within the sample plot were recorded. Of those species that could not be identified immediately in situ, herbarium specimens were collected (whole plants or distinctive parts of bigger plants), labelled with collection number (date, site code and access number) and stored and carried in plastic bags. The specimens collected during each day were prepared for pressing and drying in a wooden plant press, arranged inside sheets of dried newspapers. Moist papers were substituted daily for dry until complete dryness was achieved. These specimens^[2] were used for taxonomical identification, resorting to specific literature (Castroviejo et al., 1986, 1990, 1993, 1997, 1999, 2000, 2001; Franco, 1971, 1984; Franco & Afonso, 1998; González, 1994; Jahns, 1982; Rollán, 1981; Valdés et al., 1987). visual estimates of cover-abundance were assigned to each species on the five-point Braun-Blanquet scale:

illustration not visible in this excerpt

Stratification was recorded by attributing, in each relevé, a layer to every present species:

(1) herbaceous layer
(2) dwarf shrub layer
(3) shrub layer
(4) tree layer

Note that for a given species only the most representative layer was considered, for example, if within a forest or scrubland community both adult and juvenile individuals of one species occurred, but the former dominated, only adults were specified.

As one important parameter for the structural analysis, to each species the respective life form, as proposed by Raunkiaer (e.g. Kershaw, 1973), was attributed, resorting to the classification proposed by Franco (1971, 1984) and Franco & Afonso (1998). Thus the following life forms are considered:

(1) therophytes (annual plants completing life history from seed to seed within one favourable season
(2) hemicryptophytes (perennating buds at ground level)
(3) chamaephytes (perennating buds or shoot apices borne close to ground, often lignified)
(4) nanophanerophytes ( < 2m)
(5) microphanerophytes ( 2 - 8m)
note: epiphytic phanerophytes were included here
(6) mesophanerophytes ( 8 - 30m)
(7) macrophanerophytes (> 30m)
(8) geophytes (perennating buds below ground level - rhizomes, bulbs or tubers)
(9) hydrophytes (perennating buds under water)

Land use intensity was classified according to an estimated four-point scale:

(1)uncultivated areas (no visible traces of cultivation, or only extensive pasture)
(2)abandoned cultivated areas, but still with marked agrarian traits
(3)presently cultivated areas, apparently low-frequent treatments (ploughings)
(4)presently cultivated areas, frequent treatments (ploughings)

Additionally, to each sampled vegetation stand an a priori community type was attributed, according to the following typification:

(1)Includes all wood formations with stem height greater than two metres, in the region mainly micro- or mesoforests of micro- or mesophanerophytes, not exceeding 24 m (Rivas-Martínez et al., 2002; Tomaselli, 1982). Samples designated as "BO” (bosque)
(2)Forestations, including all planted wood formations. Designation "FO”
(3) All scrubland communities were included in this category, only alluding to structure and physiognomy in the sense of a community of wooden, mostly xerophyllous nanophanerophytes and chamaephytes with an aerial system that cannot be distinguished clearly into stem and branchage, with size and dimension depending on degradation state (Tomaselli, 1982). Thus this community type subsumes both the communities known as “maquias” (high scrubs) and "garrigues” (dwarf scrubs). Designated as “MA” (matagais)
(4) Commercial groves of olive and/or almond trees, but sometimes exhibiting varying degrees of succession, according to treatment frequency. Designation “PO” (pomar)
(5) Riparian communities: all vegetation stands of floodplain directly affected by flooding events. “RI”
(6) Rupicolous communities: vegetation stands on bare rocky
substrate, not considering the communities located in floodplain, which are included in 5. Designation “RU”

2.3 Data analysis 2.3.1 Species richness

A first approach to describe the diversity of the vegetation was to examine if variations in total species numbers could be attributed to some kind of disturbance or environmental gradient, and to this end the following parameters were used:

- Community type
- Altitude
- Distance from riverbed
- Slope
- Exposition
- Land use intensity

To examine eventual correlations of species numbers with the mentioned factors, a comparison of means was carried out, applying one-way ANOVA with the program Minitab for Windows, release 13 (minitab, Inc., 2002) to check for significance.

To check the relation between species numbers and distance, the distance of each sample plot relative to the riverbed was determined from the topographical maps (IGE, 1995, 1996) and the following distance classes were established:

illustration not visible in this excerpt

The altitude was also determined resorting to the topographical maps, and the following classes were defined:

illustration not visible in this excerpt

Exposition and slope were measured in situ with a compass, equipped with a tool to determine inclination, to built the following classes:

2.3.2 Floristic-structural analysis

For the structural characterization of the vegetation the methodological scheme proposed by Crespí et al. (2000, 2001a, 2001b, 2001c) was applied, which focuses on the floristic and structural combinations within a given set of plant communities. It consists in a transformation of the information obtained through the sampling of vegetation, into numerical matrices that allow for comparative multivariate analysis. Objective is the characterization of the structural behaviour of the vegetation in consideration.

These numerical matrices can be distinguished into two categories, in the following designated as basic structural matrices (BSM) and contingency matrices.

Arrangement of basic structural matrices (BSM)

The first step of the mentioned structural analysis is based on the creation of matrices, which contain the basic information relating to the structure of vegetation. Structure of a given plant assemblage is considered, in the present context, to have the following main qualities (e.g. Kershaw, 1973):

- the horizontal arrangement, i.e. the spatial distribution of species within an area
- the vertical arrangement of species or stratification

The spatial distribution can be focussed by two aspects, taking into account only the presence/absence of species, or the frequency of individuals belonging to these species, hereby expressed as the abundance of each species in a given stand. As far as the vertical arrangement is concerned, the floristic-structural behaviour of communities is examined with an emphasis on the life forms constituting the respective plant assemblages, as an indication of the inherent potential to develop different layers, expressing at once the adaptation of communities to the ecological conditions in force (Crespí et al., 2000). The characterization of the actual, and not only the potential, layer for each species constitutes an extension in relation to the original methodological framework, whose applicability and eventual advantages shall be tested in the course of the present work.

- Diversity BSM

The matrix containing the diversity data, as all the other matrices, is organized in a way that rows correspond to the species (382), and columns represent the sample plots (109). The diversity BSM is based on the presence (numerical value 1) and absence (0) of species within each sample. As this kind of nonmetrical data is unsuitable for correlation or distance calculations (e.g. Hair et al., 1995), this matrix must be prepared in a way to allow for disposal of metrical values. The method used for this adaptation consists in using an importance scale, based on the importance of each species within in the whole set of samples. To this end, the presence value of each species is replaced by the respective mean value over the totality of samples, following the method proposed by Crespí et al. (2000, 2001a, 2001b, 2001c). Tab. 2.4 shows an example of data treatment. This matrix, as all the following ones, was standardized, by converting each variable to standard scores (Z scores), to permit comparison of the different variables measured on different scales (Hair et al., 1995) and the construction of conjoint matrices.

[...]

---

^[1] Thermicity index It for Torre de Moncorvo equals 279 (with T=15,2°C, M=9.3°C and m=3.4°C [SMN, 1970]).

^[2] Voucher samples of all collected species are deposited in the herbarium of Escola Superior Agrària in Bragança, Portugal (international code: BRESA)