PATTERNS OF ENDEMISM ALONG AN ELEVATION GRADIENT IN SIERRA NEVADA ( SPAIN ) AND LEFKA ORI ( CRETE , GREECE )

– Aim: High mountains in the Mediterranean region of Europe are particularly rich in endemic vascular plants. We aimed to compare the altitudinal patterns of vascular plant species richness and the proportion of endemic species in two Mediterranean region: Lefka Ori on the island of Crete (Greece) and Sierra Nevada on the Iberian peninsula. Location: Sierra Nevada, Granada (Spain); Lefka Ori, Crete (Greece). Methods: Data from standardised permanent plots settings on summit sites (comprising eight plot sectors, covering the upeermost 10 altitudinal metres) of different elevations were used (GLORIA Multi-Summit approach; www.gloria.ac.at). Species numbers, rates of endemic species, and soils temperature were compared by means of ANCOVA and linear regression. Results: The two regions, though climatically similar, showed strikingly different patterns: In Sierra Nevada, the proportion of endemic vascular plants (species restricted to Sierra Nevada) showed a stepwise increase from the lowest to Pirineos.Revista de Ecología de Montaña Vol. 168, 7-24 Jaca, Enero-Diciembre, 2013 ISSN: 0373-2568 eISSN: 1988-4281 doi: 10.3989/Pirineos.2013.168001 Recibido: 11, 18, 2011. Aceptado: 04, 05, 2012. 8 R. FERNÁNDEZ, D. GHOSN, M. GOTTFRIED, G. KAZAKIS, J. MOLERO, H. PAULI & A. MERZOUKI the highest summit. In contrast, the proportion of endemic species restricted to Crete was not significantly different between the four summits in Lefka Ori. In both regions the observed trends were largely consistent with the altitudinal distribution of the endemic species obtained from standard floras. Main conclusions: The geographic positions of the two regions, i.e. island versus mainland and the higher elevation of Sierra Nevada are suggested to be the primary causes of the observed differences. The high degree of endemism in the cold environments of Mediterranean mountains’ upper bioclimatic zones indicates a pronounced vulnerability to the impacts of climate change. A continued and intensified species monitoring in the mountains around the Mediterranean basin, therefore, should be considered as a priority research task.


Introduction
The degree of endemism of a region's flora is considered as a measure of the uniqueness of the flora which attracts and inspires botanists and amateurs (Van Der Werff and Consiglio, 2004).Mountain areas are often hot spots of plant diversity (Barthlott et al., 1996;Myers et al., 2000) and can host a large number of endemic species.Even though vascular plant species richness generally decreases with altitude, Europe's alpine life zone, covering around 3% of Europe, show a disproportionately high number of species making up approximately 20% of the continents native flora (Väre et al., 2003).This high plant diversity can be attributed to a compression of thermal zones, to high habitat diversity related to rugged topography above the forest line, and to dispersal barriers between mountain ranges supporting divergent speciation.Europe shows a general south-north gradient of mountain endemism, decreasing from the Mediterranean to the temperate and further to the boreal mountains (Favarger, 1972).Mountains of the Mediterranean region remained more isolated and were only partly glaciated during the Pleistocene, (compare Blanca et al., 1998;Pauli et al., 2003;Hughes & Woodward 2008), whereas mountains further north were more extensively or almost fully ice-covered and their margins were more connected to surrounding areas and offered migration pathways.The importance of endemic plants for the mountain vegetation in southern Europe is well documented, e.g. for Crete (Strid, 1993;Dafis et al., 1996;Phitos et al., 1996;Médail and Quézel, 1999;Bergmeier, 2002) or for the Sierra Nevada in southern Spain (Quézel, 1953;Rivas Goday and Mayor López, 1966;Molero Mesa et al,. 1996;Médail and Quézel, 1999;Mota et al., 2002).
In the present study we focus on altitudinal gradients of vascular plant species richness and the degree of endemism in two Mediterranean mountain ranges with a different geographic situation: Lefka Ori on the island of Crete and Sierra Nevada on the Iberian Peninsula.By using data from standardized permanent plot settings established on summit sites of different elevations (GLORIA Multi Summit approach; Pauli et al., 2004), we question if the two climatically similar mountain areas show similar patterns of altitudinal species distribution with respect to the proportion of endemic species.

Study area
The two selected Mediterranean mountain regions (Fig. 1) are part of an international network of permanent plots in alpine ecosystems (GLORIA -PATTERNS OF ENDEMISM ALONG AN ELEVATION GRADIENT IN SIERRA NEVADA (SPAIN)... Global Observation Research Initiative in Alpine Environments; www.gloria.ac.at).Plots were setup to assess the impacts of climate change on plant species composition along a gradient from the potential climatic treeline to the uppermost bioclimatic zone of each region.The two regions were selected for this study because of (1) their exceptionally high number of endemic vascular plant species, and (2) for their similar Mediterranean-type climate with predominant winter rainfall and a pronounced dry season in summer.
The Sierra Nevada of southern Spain extends over about 80 km in an E-W direction and reaches its highest point at 3481 m a.s.l.At altitudes above 2500 m mean annual precipitation amounts to around 800 mm (Delgado et al., 2001) and falls almost exclusively as snow (Molero Mesa et al., 1996).The four studied summits were in the western, higher part of the range on siliceous bedrock.The summits ranged from 2778 m in the upper oro-Mediterranean zone to 3327 m in the upper cryoro-Mediterranean zone; for a definition of zones compare Molero Mesa and Pérez-Raya (1987)., 2013, Vol. 168, 7-24, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2013.168001)   the number of species found on each summits; horizontal bars: the number out of all species found on the four summits that occur in each 50-m altitudinal band from sea level to the highest peaks (from literature data of vertical species ranges).d,c, triangles: the proportion (percentage) of local endemic species on the four summits; horizontal bars: the percentage share of locally endemic species of all summits species in the 50-m altitudinal bands (from literature data of vertical species ranges); e, f, as before, but for regional endemic species.

Figura 2. Distribución altitudinal del número de especies de plantas vasculares en las áreas cimeras de las dos regiones de estudio; izquierda, Sierra Nevada (España); derecha, Lefka Ori (Creta). a, b, triángulos: número de especies encontradas en cada cima; barras horizontales: número de entre todas las especies encontradas en las cuatro cimas que se da en cada banda altitudinal de 50-m desde el nivel del mar hasta las cimas más altas (procedente de datos bibliográficos sobre los rangos verticales de las especies). c, d, triángulos: proporción (porcentaje) de especies endémicas locales en las cuatro cimas; barras horizontales: porcentaje de especies endémicas locales de todas las cimas en cada banda altitudinal de 50-m (procedente de datos bibliográficos sobre los rangos verticales de las especies); e, f, como la anterior pero para las especies endémicas regionales.
The Lefka Ori region is located in the western part of Crete covering 385 km² above 1000 m and reaching 2453 m at Pachnes summit.The massif is the wettest place on the island with 1900-2000 mm mean annual precipitation (Rackham and Moody, 1996).The summits were in the eastern part on carbonatic bedrock (marble and dolomite) and ranged from 1664 m in the oro-Mediterranean zone to 2339 m in the alti-Mediterranean zones.

Data recording
The sampling design used on the selected summit sites followed the GLORIA Field Manual (http://www.gloria.ac.at;Pauli et al. 2004).On each summit, the area from its highest point to the 10-m contour line below the summit was surveyed.This area was subdivided into eight summit area sections (SASs).Four upper SASs extended downward to the 5-m contour line and the four lower SASs from the 5-m to the 10-m contour line.The SASs were orientated towards the cardinal geographical directions, with their boundaries set at NE, SE, SW, and NW.Within each SAS, all vascular plant species were recorded.A miniature temperature data logger (StowAway Tidbit; Onset Corporation, Bourne, Massachusetts) was buried at 10 cm below the soil surface in each upper SAS, approximately 4.5 m below the highest summit point.The temperature series measured at an interval of 1 hour between 2001 and 2002 were used in this study.

Data analysis
Vascular plants species numbers for each summit were derived from merging the eight SASs, and for each cardinal direction from merging the upper and the lower SASs of each aspect.Species numbers per summit were visually compared with the overall altitudinal distribution (obtained from standard floras) of the species found on the summits (Fig. 3).The relationship of percentage local endemics (le) and regional endemics (lese) on summits of different altitude was analysed by Analysis of Covariance (ANCOVA), using data of the four aspects on each summit (Fig. 4 a,b,c,d).Linear regression was used to analyse the relation between yearly mean temperature and percentage of local and regional endemics (Fig. 5 a,b,c,d).
A species was considered here as locally endemic when its overall distribution was restricted to Sierra Nevada or to the Island of Crete respectively; and regionally endemic when its overall distribution was restricted to the Baetic range of southern Spain or to the Aegean region, respectively (Annexe 1).R. FERNÁNDEZ, D. GHOSN, M. GOTTFRIED, G. KAZAKIS, J. MOLERO, H. PAULI & A. MERZOUKI The information about the distribution and altitudinal ranges of the species were extracted from several bibliographic references.In the case of Sierra Nevada, the literature sources were Flora Iberica (Castroviejo et al., 1986(Castroviejo et al., -1997) ) and Molero Mesa and Pérez-Raya (1987) in cases not yet treated in Flora Iberica.For Lefka Ori, sources were Jahn and Schönfelder (1995) and Strid and Tan (1991).Altitudinal species ranges in metres above sea level, indicated in the above standard floras and check lists, were used.The nomenclature of taxon names followed Flora Europaea (Tutin et al., 1964(Tutin et al., -1980)).

Results
Total vascular plant species richness on the study summits was decreasing with increasing elevation in both regions, but with the exception of the second-lowest peak in Sierra Nevada, showing slightly higher species numbers than the lowest one.The altitudinal distribution (obtained from literature sources) of the species found on the summits in Sierra Nevada ranged from almost sea level to the maximum altitude for around 3500 m a.s.l., but shows a pronounced peak between 2500 and 3000 m a.s.l.(Fig. 2a).A similar pattern, however, with a less pronounced peak between 1500 and 2000 m a.s.l.resulted from the Lefka Ori data (Fig. 2b).R. FERNÁNDEZ, D. GHOSN, M. GOTTFRIED, G. KAZAKIS, J. MOLERO, H. PAULI & A. MERZOUKI 14 (Pirineos, 2013, Vol. 168, 7-24, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989The proportion of locally endemic species along the full altitudinal gradient (vertical ranges obtained from literature sources) shows a distinct increase with elevation in Sierra Nevada from no endemic species at 1500 m to 100% endemics at 3500 m (Fig. 2c), whereas only a weak altitudinal trend was found for Lefka Ori, ranging from 20% endemics between 0 and 500m to and 35-40% endemics above around 2200 m (Fig. 2d).The patterns for regional endemics were similar to the local endemics, apart from the second highest peak in Lefka Ori (Fig. 2e, f).
The proportion of local endemic as well as of regional endemic vascular plant species in Sierra Nevada showed a significant increase from the lowest to the highest site (Fig. 3a, c; Table 1, p <0.001).This was not the case in Lefka PATTERNS OF ENDEMISM ALONG AN ELEVATION GRADIENT IN SIERRA NEVADA (SPAIN)... Table 1.Summary of the results from ANCOVA and linear regression.TRC: target region.LEO: Lefka Ori.SNE: Sierra Nevada.Le: local endemics.Lese: regional endemics (these generally include also the local endemics).Perc_le: percentage of local endemics.Perc_lese: percentage of regional endemics (these generally include also the local endemics).Sualt: altitude of summit.
Ori.Here the percentage of local endemics did not significantly change from the lowest to the highest site (Fig. 3b; Table 1, p = 0.45), and slightly decrease with elevation in the case of regional endemics (Fig. 3d; Table 1, p = 0.015).No significant differences in the percentage of local and regional endemics in relation to slope exposures were found (Fig. 3; Table 1).
In Sierra Nevada, the percentage of both local and regional endemic species was related to yearly mean soil temperature, showing an increase of the proportion of endemics with decreasing temperature (Fig. 4a, c; Table 1, p <0.001).In Lefka Ori , local endemics did not show this relation, and the proportion of regional endemics was negatively related to soil temperature, corresponding to their proportional decreases with altitude (Fig. 4b, d; Table 1).

Discussion
Although both Sierra Nevada and Lefka Ori are located in the southern European Mediterranean zone, we observed distinct differences regarding the altitudinal patterns of endemic vascular plant species: the proportion of endemic species (local and regional) was increasing with elevation in Sierra Nevada, whereas it remained rather constant along the altitudinal gradient in Lefka Ori.
The difference in the geographic position, island versus mainland, and the high elevation of Sierra Nevada are suggested to be the primary reasons for the strikingly different patterns.In Sierra Nevada, being by far the highest mountain range in the southern Iberian peninsula, historic (Pleistocene) as well as present orographic isolation appear to have supported endemism, whereas in Lefka Ori, being 1000 m lower than Sierra Nevada, dispersal barriers caused by the geographical distance to mainland Greece seem to be the predominant factor.
Of the factors not related to geographical distance, low temperature is considered to be the main cause contributing to orographic isolation in alpine environments (Chapin and Körner, 1994;Price and Barry, 1997;Körner, 2003;Pauli et al., 2003).Most of the endemic species recorded in Sierra Nevada are cryophilic plants restricted to low temperature environments (Molero Mesa et al., 1996).Blanca et al., (2002) report a share of 30-40% in the upper Sierra Nevada, but it can even reach 80% at some sites.Endemics were considered to be of fundamental importance for the zonal vegetation of the upper elevation levels, where they accounted for up to about 90% of vegetation cover on high summits (Pauli et al., 2003).In Lefka Ori, the shorter vertical distance and, hence, the less pronounced thermal gradient, should lead to a reduced potential of temperature-related ecological niches and altitudinally distinct distribution patterns.Further, as endemism here is principally favoured by the island position, orographic thermal isolation is of subordinate relevance for the proportion of endemics within a particular elevation zone.Low temperature per se, therefore, cannot be taken as a principal indicator for endemism, but in combination with orographic isolation and the Pleistocene history (i.e.not full glaciated mountains would favour endemism).Whereas the proportion of the endemics of the entire island of Crete did not much vary among the summits, a previous GLORIA study, comparing very narrowly distributed endemics (restricted to the Lefka Ori), however showed an altitudinal increase in endemics percentages, with 3% on the lower, 6% on the second highest and 9% on the highest summit (Kazakis et al., 2007).Values for the Lefka Ori given by Strid (1996), of 26.6% Cretan endemics and 10% single mountain endemics (i.e.restricted to Lefka Ori) above 1500 m a.s.l. are well in the magnitude found on the summit sites.
Besides altitude, varying substrate conditions in the Sierra Nevada massif -siliceous in the core part including the summits analysed in this study, calcareous in parts of the margins-may further enhance habitat separation (compare Kruckeberg and Rabinowitz, 1985;Médail and Verlaque, 1997).
Finally, as Bergmeier suggests (1995Bergmeier suggests ( , 2002)), the floristic composition of the vegetation in the Lefka Ori is influenced by constant land use impact over several thousand years.This may be a further reason explaining the less pronounced altitudinal differences of endemic percentages among the summit sites.
Traditional practices of pastoral farming, but, appear to be relevant in both regions (compare Mc Neill, J. R. 1992).Alterations of natural vegetation through human activities, however, were much more pronounced at lower elevations (compare Nogués-Bravo et al., 2008), but little is known about the actual impact of land-use practices on the species distribution in Mediterranean high mountain areas.
The generally large proportion of endemics in our two Mediterranean study regions, and their accumulation at the highest altitudes in the case of Sierra Nevada, appears to be a common feature of high mountain vegetation around the Mediterranean basin.Similarly, high rates of endemism are known form the Moroccan Atlas mountains (Faverger, 1972), and aboveaverage proportions of endemics in the highest vegetation zones were reported form the Central Apennines (Stanisci et al., 2005) as well as from the highly fragmented and patchily distributed subnival environments of the mountains of Iran (Noroozi et al., 2011).
This situation -a primarily endemic vegetation being concentrated in the uppermost bioclimatic zones-gives rise to the concern that the high mountain flora of the wider Mediterranean region is highly vulnerable to impacts of climate warming.Warming-induced biodiversity losses are to be expected, particularly, if temperature increases are paired with decreasing precipitation as was projected for the south of Europe during the upcoming decades (Christensen et al., 2007).A Europe-wide study showed that an increase of more thermophilous species in alpine vegetation across the continent is already discernible (Gottfried et al., 2012); our sites in Sierra Nevada and Lefka Ori were part of this study.An accelerating "thermophilisation" of alpine vegetation will threaten cryophilic species through colonisation of high-elevation habitats by usually more widespread species of lower altitudes.A recent paper provided evidence that species numbers were declining on the southern European summits within less than a decade (Pauli et al., 2012, in press).This deserves fostering comparative monitoring activities in Mediterranean mountain regions.

Figure 2 .
Figure2.Altitudinal distribution of vascular plant species numbers found in the summit areas of the two study regions; left, Sierra Nevada (Spain); right, Lefka Ori (Crete).a, b, triangles: the number of species found on each summits; horizontal bars: the number out of all species found on the four summits that occur in each 50-m altitudinal band from sea level to the highest peaks (from literature data of vertical species ranges).d,c, triangles: the proportion (percentage) of local endemic species on the four summits; horizontal bars: the percentage share of locally endemic species of all summits species in the 50-m altitudinal bands (from literature data of vertical species ranges); e, f, as before, but for regional endemic species.Figura 2. Distribución altitudinal del número de especies de plantas vasculares en las áreas cimeras de las dos regiones de estudio; izquierda, Sierra Nevada (España); derecha, Lefka Ori (Creta).a, b, triángulos: número de especies encontradas en cada cima; barras horizontales: número de entre todas las especies encontradas en las cuatro cimas que se da en cada banda altitudinal de 50-m desde el nivel del mar hasta las cimas más altas (procedente de datos bibliográficos sobre los rangos verticales de las especies).c, d, triángulos: proporción (porcentaje) de especies endémicas locales en las cuatro cimas; barras horizontales: porcentaje de especies endémicas locales de todas las cimas en cada banda altitudinal de 50-m (procedente de datos bibliográficos sobre los rangos verticales de las especies); e, f, como la anterior pero para las especies endémicas regionales.