- Top of page
- Materials and Methods
- Supporting Information
In recent years, the importance of robust systems classifying biogeographical patterns (i.e. regionalizations) has been emphasized (Olson et al., 2001; Whittaker et al., 2005; Mackey et al., 2008; Procheş, 2008; Kreft & Jetz, 2010). Besides their use in providing a taxonomy of distributional areas based on taxa, they serve as the basis for conservation planning (see Mackey et al., 2008 and literature therein). Mackey (2008) noted that such analyses of species distributions provide information concerning prior and predicted future distributional and evolutionary events.
Biogeographical regionalizations provide a basic summary of how life on Earth is patterned as a result of history and current physical and biological determinants (Kreft & Jetz, 2010). Therefore, they provide an indispensable background for answering basic and applied questions in biogeography, evolutionary biology, systematics and conservation (Morrone, 2009; Posadas et al., 2010). Two methods that have been widely applied to define biogeographical regionalizations are cluster analysis (Sneath & Sokal, 1973; Kreft & Jetz, 2010) and parsimony analysis of endemicity (Crisci et al., 2000, 2003; Crisci, 2001; Katinas et al., 2004; Crisci & Katinas, 2009). Both methods provide objective approaches for classifying biotas into a hierarchical system because they use quantitative measures of similarity among areas to reveal natural patterns of distribution (López et al., 2008).
The agglomerative cluster algorithm, unweighted pair-group method using arithmetic averages (UPGMA), is one method that has been used to analyse distributional information of taxa (Márquez et al., 2001; Moreno Saiz & Lobo, 2008). Sneath & Sokal (1973) and, more recently, Kreft & Jetz (2010) have shown that the UPGMA algorithm produces less distortion in relation to the original similarities than complete, single and other average linkages, and performs better for biogeographical regionalization than other algorithms.
Parsimony analysis of endemicity (PAE; Rosen, 1988) is a method that classifies areas (localities or quadrats) according to their shared taxa, resulting in a hierarchical classification of the geographical units. It aids recognition of biotic assemblages and generates biogeographical regionalizations. According to Rosen (1988), PAE cladograms probably reflect geological factors and other prevailing ecological conditions without being able to discriminate among the influences of each factor.
Parsimony analysis of endemicity also allows recognition of areas of endemism, defined as the congruent distributions of two or more species (Platnick, 1991), sometimes resulting from evolution of different taxa that share the same isolated regions (Rosen, 1978) and thus evolve under the same ecological and historical conditions.
Here, we present an example of the application of these techniques for the Iberian Peninsula (Spain, Andorra, Portugal and part of France) and the Balearic Islands. This region is a particularly good model because:
- The Mediterranean area is exceptionally rich in biological diversity (Blondel & Aronson, 1999) and many studies have been carried out in it (Thompson, 2005; Blondel & Médail, 2009). The main centres of Mediterranean plant diversity are located primarily on its islands and peninsulas, that is: Anatolian, Balcan, Italian, and Iberian (Médail & Quézel, 1997). These islands and peninsulas have had an important role not just as cradles of biodiversity (evolution, radiation) but as refugia, especially during Quaternary glaciations (Taberlet et al., 1998; Hewitt, 1999). For example, 25% of the Mediterranean glacial refugia identified by Médail & Diadema (2009) are located in Iberia and the Balearic Islands.
- Biogeographical analyses of the Iberian Peninsula together with the Balearic archipelago have been carried out to establish spatial patterns using different taxonomic groups (i.e. Sainz Ollero & Hernández Bermejo, 1985; García Barros et al., 2002; Rivas-Martínez et al., 2002; Carrascal & Lobo, 2003; Moreno Saiz & Lobo, 2008; Romo & García-Barros, 2010; see Table 1). However, other authors have studied Iberia separately to avoid the complications that inclusion of species of the archipelago may create (Márquez et al., 1997, 2001; Vargas et al., 1998).
- Previous studies of south-western Europe were based on incomplete or insufficiently large datasets, representing fewer taxa and so did not include the full range of plant biodiversity.
Table 1. Primary models of biogeographical regionalizations based on vascular plants for south-western Europe
|Reference||Study group||Area||Geographical grain||Method|
|Sainz Ollero & Hernández Bermejo (1985)||Endemic dicotyledons (n = 1200)||SW Europe||OGUs of varying size||Ward, Lance-Williams|
|Moreno Saiz et al. (1998)||Endemic monocotyledons (n = 182)||SW Europe||100, 50 and 10 km UTM squares||twinspan, decorana|
|Márquez et al. (2001)||Pteridophytes (n = 113)||Iberian Peninsula||OGUs of varying size||UPGMA|
|García Barros et al. (2002)||Endemics (129 plants + 351 animals)||SW Europe||100 km UTM squares||UPGMA, PAE|
|Rivas-Martínez et al. (2002)||A number of taxa and plant communities||SW Europe and Canary Islands||–||Expert knowledge|
|Moreno Saiz & Lobo (2008)||Pteridophytes (n = 123)||SW Europe||50 km UTM squares||Ward|
We compiled the largest vascular plant database used so far in a biogeographical study of this area, including all fern and gymnosperm species and a large number of angiosperm species. The use of such databases that include information on a large portion of a region's biodiversity is potentially critical to avoid biases introduced by limited geographical data (Soria-Auza & Kessler, 2008 and references therein).
Therefore, our objective is to analyse the distributional pattern of the vascular flora of the Iberian Peninsula and Balearic archipelago. Both cluster and parsimony methods are applied to delineate a biogeographical scheme for south-western Europe. Additionally, PAE will allow the definition of areas of endemism, defined as groups of cells that share two or more exclusive species or as individual cells with at least two exclusive taxa (autapomorphies; Morrone, 1994). These areas of endemism would be considered as potentially relevant for biodiversity conservation (Posadas, 1996).
- Top of page
- Materials and Methods
- Supporting Information
Phytogeographical classifications for south-western Europe based on plant taxa and plant communities are summarized in Table 1. Many are exclusively defined by endemic plants, based on the idea that they are bioindicators (Braun-Blanquet, 1923; Hernández Bermejo & Sainz Ollero, 1984; but see Hengeveld, 1990). For example, Sainz Ollero & Hernández Bermejo (1985) recognized a Balearic sector and five homogeneous areas within the Iberian Peninsula, with three transition zones in NE, NW and SW Iberia, respectively. Moreno Saiz et al. (1998) discussed the effect of grid size in regionalizations and proposed up to 14 regions at the scale of 50 km.
Márquez et al. (2001) used the distribution of pteridophyte species within political units and identified three strong boundaries dividing the Iberian Peninsula into four main regions. Later, Moreno Saiz & Lobo (2008) noted that such results are based on inadequate operational geographic units (OGUs) and incomplete data and they proposed a classification of 10 regions for fern species; they also used discriminant analysis to identify the environmental variables that are most highly correlated with such regions.
Rivas-Martínez et al. (1990, 2002) have been publishing a series of proposals for regionalization of the area based on physiography, distribution of indicator plants, and especially potential natural vegetation. Their phytocoenological systems have been widely used, even though their methodology has never been made explicit. Their last system splits south-western Europe into Eurosiberian and Mediterranean regions, and these in turn into eight provinces (Rivas-Martínez et al., 2002).
García Barros et al. (2002) studied both the endemic flora and fauna in south-western Europe, using PAE and a coarse grid (100 km × 100 km) to identify up to 36 endemic areas, many corresponding to mountain chains, with the areas grouped into four main sectors, based on overall similarity.
Some previous studies focused only on Iberian biogeography to avoid ecological or evolutionary concerns related to Balearic insularity. However, our analyses have shown the pertinence of combining the Balearic archipelago together with the Iberian Peninsula in biogeographical analyses of the western Mediterranean, as in some previous studies (e.g. Pichi Sermolli et al., 1988; Carrascal & Lobo, 2003; Moreno Saiz & Lobo, 2008; but see Romo & García-Barros, 2010). Both cluster analysis and PAE retained the Balearic Islands as an ingroup of the eastern Iberian–Balearic territory and only separated the two at higher branching levels. The issue of how much of southern French territory should be included as part of south-western Europe study area is yet to be determined.
Neither UPGMA nor PAE analyses supported the classical division between Eurosiberian (or Circumboreal sensu Takhtajan, 1986) versus Mediterranean regions in south-western Europe, a latitudinal division roughly paralleling the Pyrenean–Cantabrian montane axis. Although several faunal, floral and phytocoenological studies do suggest such a division (i.e. Rivas-Martínez et al., 1990, 2002; Carrascal & Lobo, 2003; Galicia et al., 2010; Rueda et al., 2010), its primacy has not been supported by other studies, including our own (Moreno Saiz et al., 1998; García Barros et al., 2002). Floristic studies of both endemic and non-endemic plants show that this Eurosiberian–Mediterranean boundary has not been impermeable; many taxa at different times have crossed it in both directions for reasons related to ecology and history (Hewitt, 1999; Vargas, 2003; Gómez & Lunt, 2007). Our large database highlights the importance of connections between Atlantic and Mediterranean climatic regions for a larger number of species than for the small subset of strictly Eurosiberian and Mediterranean floristic elements. Again, the first dichotomy in both our analyses reflects the strongly longitudinal division between acidic and basic substrates in Iberia (Fig. 1b).
The cluster analysis (UPGMA) enabled us to recognize nine sectors in south-western Europe, representing the Balearic Islands, different mountain ranges, inner plateaus, and the south-western maritime areas. This emphasizes the Iberian physiography and thus the importance of environmental constraints. Human disturbance gives rise to a 10th cluster, made up of a small number of inland cells, transformed by agriculture centuries ago and having a depauperate flora.
PAE similarly showed two main longitudinal clades, but with the longitudinal division located further east. Unlike the cluster analysis, it splits the Cantabrian from the Pyrenean Mountains and the Baetic and Iberian systems. The basal position of La Mancha and surrounding areas (Clade 1 in Fig. 3) could be related with the use of a hypothetical all-zero outgroup for PAE. These plains on the south plateau have been under cultivation for thousands of years, resulting in the extinction of many species. The paucity of species here is exacerbated by the lack of attention by botanists to this degraded landscape.
The linking of the Balearic Islands to Valencian and Catalonian coastal cells (Fig. 3) also was seen for endemic monocots (Moreno Saiz et al., 1998) and is supported by common phytocoenological patterns (Rivas-Martínez et al., 2002). These coastal areas and the Balearic Islands were influenced by the same Alpine uplift, but their similarity could also have been increased by eustatic Mediterranean sea-level changes. At the end of the Miocene and during the Pleistocene, the distance of the western-most islands from the mainland was periodically decreased, enabling, at the height of the glacial periods, more frequent exchanges between their floras (Sáez & Rosselló, 2001; Rosselló & Castro, 2008); during interglacial periods, sea levels rose and the distance again increased. This cyclical process resulted in periodic isolations of plant populations, allowing them to differentiate.
Clade 8 largely consists of the drainages of the Guadiana and Guadalquivir rivers. It also includes the western end of the Baetic Sierras, i.e. the elevations of Ronda and Grazalema, which stood separated from the rest of the Baetic System by a depression (the present valley of the Guadalhorce River), during the lower Messinian (Medina-Cazorla et al., 2010). This sector somewhat resembles the Andalusian pteridological region of Moreno Saiz & Lobo (2008), a territory characterized by some relictual taxa and a large number of fern species. Environmental factors explain less variability here than in the rest of south-western Europe, so historical constraints may have played an important role (Moreno Saiz & Lobo, 2008).
Finally, PAE identified 19 pluricelled areas of endemism, which can be subsumed into 12 Iberian and Balearic areas due to the spatial overlap of some of them (Fig. 4). As congruent distributions of two or more species are the result of evolution under the same ecological and historical conditions (Rosen, 1978), the integration of ecology and history may not only discern processes of spatial arrangement but also highlight some areas of conservation interest (Crisci et al., 2006).
The areas of endemism determined from PAE analysis are mostly concentrated in montane areas, particularly elevations nearer the coast. Such a correlation was already noted by García Barros et al. (2002) and by Castro Parga et al. (1996) and Lobo et al. (2001) in their studies on richness and endemism. Lobo et al. (2001) found a strong positive correlation between plant diversity and elevational range throughout south-western Europe in general.
The nested pattern shown by some areas of endemism found in this analysis supports establishment of certain areas for conservation purposes based on their species richness (Posadas, 1996): Pyrenees (cells XN4 + YN2), western Cantabrian Mountain Range (QH1 + TN3), part of the Baetic System (Sierras of Mágina, Cazorla-Segura, and La Sagra, WG1 + WH2) and Majorca. They harbour a significant fraction of the south-western European flora, have probably served as refugia during the Quaternary glaciations, and constitute local hotspots of endemicity (Castro Parga et al., 1996; Araújo et al., 2007; Gómez & Lunt, 2007). Although the Sierra Nevada is considered among the richest floristic areas of the Mediterranean Basin (Gómez-Campo, 1985), it was not supported as a whole in our analysis by a high number of endemic species (Table 3). However, parts of Sierra Nevada are well supported as areas of endemism including single cells areas of endemism (VG2, VG4, WG2; Table 4). These areas of endemism in Sierra Nevada are so rich in endemic taxa that deserve to be considered, as the Sierra already was, an important area for conservation. Other single cells that contain a great number of endemics and therefore could be considered as relevant for conservation purposes are listed in Table 4 and mapped in Fig. 5.