Climatic Patterns in the Mediterranean region...ecologia mediterranea – Vol. 40 (1) – 2014 49 Abstract Climate and vegetation patterns in the Mediter - ranean region are known

49ecologia mediterranea – Vol. 40 (1) – 2014

Abstract

Climate and vegetation patterns in the Mediter-ranean region are known to be highly corre-lated and a summary of the major studies in thistopic is provided. Many of the studies rely onsimple indices based on simple climatic variablessuch as monthly averages of temperature andprecipitation. The availability to the public ofdata from FAO and World Climate with 30 yearaverages of monthly temperatures and precipi-tation for 2,086 weather stations around theMediterranean provided a good basis to modeltheir geographical distribution. Again, simplevariables were used as explanatory variables, aslatitude, longitude, elevation and distance tocoast. Models of the linear or logistic type wereused to explain monthly and annual averagetemperatures and precipitation indicating that,in many situations, the simple models providedcan already be used to understand the generalpatterns of climate and vegetation in theMediterranean Region. A small exercise illus-trates the possibilities of these models to simu-late the general effects of major climaticchanges in the changes of the geographical lim-its of Mediterranean species and vegetation.

Résumé

Dans la région méditerranéenne, le climat et ladistribution de la végétation sont connus pourêtre hautement corrélés et une synthèse desprincipales études sur le sujet est présentée ici.Pour la plupart, les études s'appuient sur desindices simples fondés sur des variables clima-tiques simples comme les moyennes mensuelles

de la température et des précipitations. La dis-ponibilité pour le grand public de données dela FAO et de World Climate comprenant desmoyennes trentennales de températures et deprécipitations pour 2 086 stations météorolo-giques autour de la Méditerranée a fourni unebonne base pour modéliser leur distributiongéographique. Des variables simples ont été uti-lisées comme variables explicatives, comme lalatitude, la longitude, l'altitude et la distance àla mer. Des modèles de type linéaire ou logis-tique ont été utilisés pour expliquer lesmoyennes des températures et des précipita-tions mensuelles et annuelles, indiquant que,dans de nombreux cas, les modèles simples pro-posés peuvent d'ores et déjà être utilisés pourcomprendre la caractérisation spatiale du climatet la distribution de la végétation en régionméditerranéenne. Un petit exercice illustre lespotentialités de ces modèles pour simuler leseffets des principaux changements climatiquessur les variations des limites géographiques desespèces et de la végétation méditerranéenne.

Introduction

Climate and vegetation

Mediterranean vegetation has been studied fora long time and its geographical distributionhas been represented in several papers, as inthe very influential work by Tomaselli (1977in de Montgolfier 2002) shown in Figure 1.

Climatic Patterns in the Mediterranean regionModèles climatiques dans la région méditerranéenne

Francisco C. REGO, Marta S. ROCHA

Centro de Ecologia Aplicada Prof. Baeta Neves (CEABN/INBIO)Instituto Superior de Agronomia, Universidade de Lisboa

Tapada da Ajuda, 1349-017, LisboaE-mail: [email protected]

Keywords: Mediterranean region, climate,climate interpolation.

Mots clés : région méditerranéenne, climat,interpolation climatique.

05-art-5-vol40(1)-corTC_Mise en page 1 26/08/14 09:39 Page49

Most early bioclimatic studies were charac-terized by the search of synthetic expressionsof climate to produce maps of homogeneousclimatic regions and to assess climate-vegeta-tion relationships (e.g. Lang 1915; de Mar-tonne 1926a, b, 1942) and several of thosestudies were performed in the Mediterraneanregion (Emberger 1930; Giacobbe 1949;Daget 1977). In most cases, climatic variables(in particular precipitation, evaporation andtemperature) were combined to allow climaticpatterns to be correlated with observed vege-tation patterns and to provide some quantita-tive expressions of climatic conditions suchas aridity and winter cold (Mazzoleni et al.1992).

Mediterranean vegetation was always consid-ered to be associated with a specific type ofclimate – the Mediterranean climate. In fact,climate has an obvious direct influence onvegetation, and the influence of other naturalfactors on vegetation is related with climateas soil development or fire regimes. Even theeffect of topography is not independent of cli-mate because of the influences of elevationand aspect on temperature and precipitation.

This close association between climate andvegetation results in that the geographical lim-its of the Mediterranean ecosystems are com-monly defined by climatic parameters (e.g.Emberger 1930; Gaussen & de Phillips 1958;Quezel 1977; de Montgolfier 2002).

The different climatic indices in the Mediterranean

A popular approach for the explanation of thedistribution of species in the Mediterraneanwas that of Emberger (1930), who proposeda system with two axis, one related with thecold limit (represented by the average of theminimum temperatures of the coldest month– m in ºC), and one coefficient (Q) related toaverage annual precipitation (P in mm), m andto the average of the maximum temperaturesof the warmest month (M in ºC):

Q = 2000 P/[(M + m)(M – m)]In a comparative study, Daget (1977) con-cluded that Emberger's (1930) system was thebest for defining Mediterranean climates, andthe approach was disseminated in severalMediterranean countries (e.g. Quézel 1977 orMhirit 1999).However, in spite of the wide use of theEmberger system in the Mediterranean, otheralternative systems were proposed based onsimple and more readily available data asmonthly averages of temperature and precip-itation.One popular approach is the displaying of cli-matic patterns as a combined double-scalerepresentation of temperature and precipita-tion values throughout the year (Bagnouls &Gaussen 1957; Gaussen 1956; Walter 1955).The use of these climatic diagrams was verysuccessively extended to the complete worldin the Klimadiagramm-Weltatlas (Walter &Lieth 1960).

ecologia mediterranea – Vol. 40 (1) – 201450

FRANCISCO C. REGO, MARTA S. ROCHA

Figure 1 – Distribution of the Mediterranean matorral (maquis) (Tomaselli 1977 in de Montgolfier 2002).


In this system, a dry month was defined bythe simple relation P/T < 2, where precipita-tion (P in mm) was less than two times theaverage temperature (T in ºC). This approachallowed for the calculation of the number ofdry months, which was the basis of theapproach taken by Gaussen and de Phillips(1958) in the preparation of the bioclimaticmap of the Mediterranean (UNESCO-FAO,1963).

Many other authors, using the same method,agree in that the Mediterranean climate ischaracterised by the presence, in the warmseason, of a dry period of at least one month(de Montgolfier 2002).

In spite of the obvious criticisms for the over-simplification of these empirical approachesbased on Walter temperature rainfall dia-grams, these are still very much used as didRivas-Martinez (1981) in his climatic classi-fications linked to phytosociological cate-gories. Rivas-Martinez further extendedGaussen’s concepts, suggesting the use ofratios between the sum of precipitations (Pi inmm) for a period of n months and the sum ofthe corresponding positive monthly tempera-tures (Ti in ºC):

In = Σn Pi/Σn TiThe period can range from one month, wherethe index is the same as that of Gaussen, aperiod of n summer months (Iosn), or the totalyear (Io). As before, the value of 2 for theratio between precipitations and temperaturesin summer months is the limit of a dry period.The values of Io allow for the classification ofthe climate as humid (above 3.6), dry (2.0-3.6), semi-arid (1.0-2.0), or arid (below 1.0).Originally used in Spain, this system has beenmore recently applied in other regions of theworld such as South America (Rivas-Martinez& Navarro 1994), or North America (Rivas-Martinez et al. 1999).

Regardless of the indices used, monthly aver-age temperatures and precipitation are thebasis for many analyses of climate and its cor-relation with vegetation. It is important nowto understand the factors associated with thoseclimatic variables.

The factors associated with temperature and precipitation

Temperature and precipitation are known tovary with spatially predictable patterns. Theyvary also in function of the elevation. If pre-

cipitation has a relatively complex relation-ship with elevation, temperature decreases lin-early according to elevation, following a meanlapse rate of – 6 to – 10 °C km-1 (MacArthur1984; Friend & Woodward 1990; Waring &Running 1998).

It is also well known that proximity to oceansor other large water bodies of water that areheat reservoirs has a great moderating effecton temperature; the interiors of continents, farfrom oceans, are said to have a “continental”climate, having lower temperatures in winterand higher temperatures in summer.

The issue of the timescales in modelling tem-perature and precipitation is important and ithas been addressed by various authors.Timescales used vary from daily (Johansson& Chen 2003), to monthly (Goodale et al.1998), or annual (Briggs & Lemin Jr 1992).

For annual precipitation, it has been long indi-cated that global rainfall patterns in relationto latitude can only be understood on the basisof major atmospheric circulation processes.Global circulation patterns have beendescribed by scientists since Halley in 1683or Hadley in 1735 following observationsmade in the discovery voyages throughout theoceans. Full explanations for such patterns arecomplex but good approximations are pre-sented in various books (e.g. MacArthur1984).

The information of the average annual pre-cipitation is however not enough to describefeatures of extreme relevance for the Mediter-ranean Region. In particular, it does notenable us to determine the degree of waterscarcity in the summer months. It is thereforenecessary to understand the processes associ-ated with the seasonal fluctuations of precip-itation.

In Europe and North Africa the atmosphericcirculation is established in the presence ofpermanent cyclonic centers, the IcelandicLow and the Azores High, located around lat-itude 30ºN in January and moving northtowards the British Isles in July. The Mediter-ranean type climate corresponds exactly tothis latitudinal belt north of the deserts, wherewinters are rainy (above 30ºN) and summersare dry (below 45ºN).

The effect of location (distance to the coastand longitude) on precipitation can be easilyunderstood by inspection of any Europeanmap of rainfall distribution (e.g. EEA & Bour-deau 1995).


Climatic Patterns in the Mediterranean region


The increase in precipitation with elevation isevident worldwide, and commonly approxi-mate a linear form (Osborn 1984). However,because of the various interactions of eleva-tion with wind direction, topography, latitudeand distance to the coast, it is very difficult toprovide a simple way to represent the inde-pendent effect of elevation, and authors asDaly et al. (1994) proposed a statistical-topo-graphic model for use over mountainous ter-rain where precipitation-elevation regressionswere estimated in independent slopes. Theseauthors further indicate that the coefficientsof the different regressions can be quite vari-able, ranging from increases of precipitationof 250 mm per kilometer of elevation in lowprecipitation regions to more than2,000 mm/km in high precipitation areas. As aconsequence, the authors proposed that theregression slopes could be expressed as a pro-portion of the average precipitation of thearea, indicating increases in precipitation inthe range of 0.6 to 1.3 km-1.

The above discussion refers to general pat-terns, as local topography influences local cli-mate patterns, therefore a portion of the vari-ability of temperature and precipitation maybe explained using topoclimatic models(Dobrowski et al. 2009). These models allowthe use of both regression modeling and inter-polation methodologies for climatic datasetcreation. In particular, multiple regression sta-tistics have proved to be useful in examiningclimate data and can be used to develop rela-tionships between topographic variables andtemperature and precipitation at locations ofmeasured climate data. These relationshipscan then be applied to areas possessing appro-priate topographic data but lacking measuredclimate information (Brown & Comrie 2002).

This study examines the relationship betweenterrain and location variables with averagemonthly temperature and precipitation byestimating multi-regression models. Theapproach aims to use simple models for a bet-ter understanding of the exogenous factorsthat can lead to an explanation of climatic pat-terns.

Going a step further, we discuss how the mod-els developed may be useful to define theboundaries of the Mediterranean region tak-ing into account the limits of temperature sug-gested by Pavari (1954) and the relationshipsbetween temperature and precipitation pro-posed by Gaussen and de Phillips (1958) in

the preparation of the bioclimatic map of theMediterranean (UNESCO-FAO, 1963).

The analysis is based on data from the Foodand Agriculture Organization of the UnitedNations (FAO) and World Climate.

Methods

Study area

While the focus of the research is the Mediter-ranean region, the network of meteorologicalstations used in the analysis was defined asthe area from 25º to 70°N and 35ºW to 50°E(Figure 2) to cover neighboring regions, sothat possible edge effects arising from inter-polation were minimized.

The study region is characterized byextremely varied physiology, ranging fromisolated and extended mountain ranges tolow-elevation deserts.

Climate data

Monthly mean temperature (°C) and total pre-cipitation (mm) data were obtained from FAOand World Climate Data, as averages for peri-ods varying between 30 and 90 years in thelast century, for a total of 2,086 weather sta-tions around Europe and North Africa (Fig-ure 2). FAO data accounted for one third ofthe stations considered and included almostall the stations in Europe. Stations located inNorth Africa and Euro-Asia were taken fromWorld Climate. In instances, where individualreporting stations contained partial or no data(e.g. a missing monthly value), they wereomitted from the analysis.

Explanatory Variables

Using the geographic information systemArcGIS, an initial set of terrain and locationvariables were developed in order to be usedas regression predictor variables. The 1 ×1 km DEM developed by the United StatesGeological Survey, the gtopo30 DEM, wasused to provide topographic data for the studyregion. However due to the great extension ofthe study area, a resample of 10 x 10 kmproved to be the manageable solution for vari-ables development. Latitude and longitudewere also calculated at 10 × 10 km resolutionfor each grid cell using the DEM grid as a ref-




erence. Using an ArcGIS extension, the neardistance to coast line was also calculated.

Summary statistics of latitude, longitude, ele-vation and distance to coast are provided inTable 1.

Results

Temperature Model

The strong relationship expected betweenaverage annual temperature and latitude isvery dominant in our dataset (Figure 3).

The effect of latitude (Lat in ºN) was thusmodeled as a constant cooling rate. The “con-tinentality” was expressed by two variables,distance to the coast (Dist in km), and longi-tude (Lon in ºE), and also assumed as havinga linear effect. Finally, as discussed above, theeffect of elevation (Elev in km) was assumedto be linear.

The average monthly temperatures (andannual average) were then estimated withSPSS Statistics for Windows, Version 22.0(IBM Corp. 2013) by a weighted multiple lin-

ear regression using the area of Thiessen poly-gons as weight, because the stations are notuniformly distributed across the study area.

The independent variables used in the regres-sion models were latitude, longitude, eleva-tion, and distance to coastline, as a stepwiseselection method, using p = 0.05, indicatedthat all independent variables should beincluded in the model. In fact, all variableswere statistically highly significant (p lowerthan 0.001).

The analysis was performed for every monthand for the annual average. The linear equa-tions fitted to the temperature data from the2086 weather stations was of the form:

Temp (ºC) = a + b Lat (ºN) + c Lon (ºE) + d Dist (km) + e Elev (km)



Figure 2 – The study area encompasses the Mediterranean region, with data also collected from portions of the surroundingcountries. Locations of FAO and World Climate weather stations are shown.

Mean Minimum Maximum Std. Deviation

Latitude (ºN) 45.7 25.5 69.7 13.5Longitude (ºE) 11.5 –31.1 50.0 28.1Elevation (m) 228.3 0.0 3109.0 493.6Distance (km) 132.5 0.0 964.0 206.4

Table 1 – Summary statistics of explanatory variables used in model development.


Table 2 shows the regression parameters esti-mated for each month, and for the annualaverage, as well as the corresponding coeffi-cient of determination.The analysis of the parameters estimated indi-cates that the intercept (a) behaves asexpected, following the seasonal variation ofthe temperatures, with the highest values inAugust and September.The negative effect of latitude in temperature(b) is clear throughout the year, but less pro-nounced in summer.The two variables associated with “continen-tality”, longitude (c) and distance to the coast(d) show the same seasonal pattern, with pos-itive values (higher temperatures) in summerand negative values (lower temperatures) inwinter, as expected.

The effect of elevation (e) in decreasing tem-perature is also obvious throughout the yearbut, as with latitude, the effect is also less pro-nounced in summer. The values estimated, adecrease of temperature between 3.5 ºC and6.5 ºC per kilometer of elevation are only alittle lower than the “moist adiabatic lapserate” suggested by MacArthur (1984) to bearound 6 ºC km-1.

Overall, the fit of the models can be consid-ered good (R square always above 0.89).

Globally, the exercise showed that simplemodels can really be used as a generalapproximation to the distribution of tempera-tures throughout the year.

The models were used, as explained below, tobuild the final maps.

Precipitation Model

The data from 2,086 weather stations aroundEurope and North Africa allow for a gooddemonstration of the influence of latitude onannual precipitation (Figure 4).

A preliminary exploratory analysis confirmeda non-linear relationship between annual pre-cipitation total and latitude (Figure 4). Dif-ferent non-linear alternatives were exploredand selection was made based on the knownrelationships between precipitation andexplanatory variables discussed above. Thecomponent of latitude was modeled as anequation of the logistic type:

exp {a/[1 + exp (b *(Lat – c))]}



Figure 3 – Decrease in average annual temperature with latitude based on data from the 2,086 weather stations analyzed. A reasonablegood fit of the equation (R square 0.77) indicates a linear trendwith a general decrease of 0.54ºC per degree of latitude.

Month a (ºC) b (ºC/ºN) c (ºC/ºE) d (ºC/km) e (ºC/km) R square

January 35.7 – 0.62 – 0.16 – 0.0082 – 6.28 0.924February 36.5 – 0.63 – 0.14 – 0.0058 – 6.23 0.912March 37.0 – 0.61 – 0.09 – 0.0020 – 5.79 0.924April 37.4 – 0.56 – 0.02 0.0037 – 5.81 0.937May 38.0 – 0.51 0.03 0.0066 – 5.19 0.931June 39.7 – 0.48 0.06 0.0069 – 4.75 0.915July 41.3 – 0.47 0.07 0.0059 – 3.48 0.892August 43.5 – 0.52 0.05 0.0045 – 3.97 0.902September 43.9 – 0.57 0.01 0.0016 – 5.10 0.932October 42.4 – 0.60 – 0.03 – 0.0022 – 5.97 0.936November 38.9 – 0.61 – 0.08 – 0.0050 – 6.17 0.946December 36.2 – 0.60 – 0.12 – 0.0078 – 6.53 0.936Annual 39.2 – 0.56 – 0.04 – 0.0002 – 5.44 0.942

Table 2 – Parameters of the equations relating temperature with latitude, distance to coast, longitude, and elevation for the various months, based on the 2,086 weather stations analyzed. Coefficients of all variables are highly significant (p < 0.001).


where a, b and c are fitted parameters. Thevalue of c has units of latitude and is relatedwith the latitude of the Azores High.

The components related to location and to ele-vation assume a multiplicative effect, with theparameters d, e, and h indicating proportionalincreases, or decreases:

(1 + d * Lon + e * Dist) * (1 + h * Elev)

The variables Lat, Lon, Dist, Elev are thesame as in the Temperature Model.

A full non-linear model with all the compo-nents was then used. As for temperature, andbecause weather stations were not uniformlydistributed across the study area, we used thearea of Thiessen polygons as a weight vari-able. The contribution of all variables washighly significant (p lower than 0.001). Modelfitting was done using the Non-Linear ModelOption in SPSS and the Levenberg-Marquardtmethod.

The full model is of the form:

Prec (mm) = exp {a/[1 + exp (b *(Lat – c))]}*(1 + d * Lon + e * Dist) * (1 + h * Elev)

In Table 3, we show fitted parameters as wellthe coefficient of determination (R2). Thesemodels were used, as explained below, tobuild the final maps.

It is concluded from this analysis that precip-itation is more predictable during the summermonths June, July and August (R square> 0.77) than in other periods of the year. Pre-cipitation patterns are very much determinedby the value of the parameter c, reflecting thechanges in the latitudinal position of thecyclonic centers. Changes in summer precip-

itation patterns based on changes of the lati-tudinal position of the Azores High can nowbe predicted.The influence of “continentality”, representedby the parameters associated to longitude (d)and to distance to the coast (e) show that thenegative effect is much more pronounced inwinter.The values for the parameter associated to theeffect of elevation (h) are generally lower thanthose presented in the literature, probablybecause of the large area considered in thisstudy, except for the summer months, wherethe values are similar to the lower limit of therange of values suggested by Daly et al.(1994).



Figure 4 – Average annual rainfall in 2,086 weather stations in Europe and North Africa as related to latitude between 20º and 70ºN. It is clearly apparent the latitude of the deserts below 30ºN.

Month a b c (ºN) d (ºE-1) e (km-1) h (km-1) R square

January 4.59 – 0.35 25.3 – 0.0128 – 0.000307 0.205 0.635February 4.36 – 0.30 25.2 – 0.0124 – 0.000319 0.337 0.630March 4.37 – 0.28 26.3 – 0.0120 – 0.000271 0.290 0.661April 4.10 – 0.31 28.9 – 0.0103 0.000004 0.527 0.702May 4.06 – 0.40 32.8 – 0.0100 0.000226 0.786 0.762June 4.10 – 0.34 34.9 – 0.0077 0.000543 0.783 0.791July 4.22 – 0.22 34.9 – 0.0047 0.000384 0.780 0.779August 4.44 – 0.21 33.2 – 0.0081 0.000117 0.343 0.770September 4.49 – 0.40 32.2 – 0.0089 – 0.000148 – 0.186 0.606October 4.70 – 0.26 26.8 – 0.0085 – 0.000496 – 0.049 0.561November 4.56 – 0.26 23.2 – 0.0088 – 0.000597 0.452 0.562December 4.58 – 0.30 24.2 – 0.0100 – 0.000487 0.324 0.591Annual 6.88 – 0.21 23.5 – 0.0101 – 0.000104 0.235 0.700

Table 3 – Parameters of the equations relating precipitation with latitude, distance to coast, longitude, and elevation for the various months, based on the 2,086 weather stations analyzed. Coefficients of all variables are highly significant (p < 0.001).


Discussion on the possibleuses of the models

On understanding the limits of the Mediterranean vegetation

The geographical distribution of sclerophyl-lous species of some typical Mediterraneangenus as Olea, Laurus, or Ceratonia, has beenstudied by various authors. One of the mostwell-known studies of the distribution of Oleain relation to the number of dry months waspresented by Desfontaines (1975 in de Mont-golfier 2002) and shown in Figure 5.

Applying our models to the DEM grid(10 x10 km), it was possible to predict aver-age monthly and annual temperature and pre-cipitation for the whole area, and to use spe-cific thresholds to estimate the geographicallimits of the Mediterranean climate. In orderto estimate such geographical limits it wasconsidered, based on the literature alreadymentioned, that the threshold for the “coldlimit” could be set as 4 ºC for the averagetemperature of the coldest month. For the“aridity limit”, the approach by Rivas Mar-tinez (1981) was followed, and the thresholdof 2 for the ratio between the total annual pre-cipitation (in mm) and the sum of all positivemonthly temperatures (in ºC) was used. The

result for the geographical distribution of theMediterranean climate based on our modelsand the above thresholds can be seen in Fig-ure 6. These limits are in general agreementwith the distribution of the Mediterranean“maquis” by Tomaselli (1977 in de Mont-golfier 2002) represented in Figure 1, andwith other studies about the Mediterraneanclimate or the distribution of Mediterraneangenus as Olea represented in Figure 5.

On predicting the consequences of climatic changes

The models can also be used as a first approx-imation to estimate the effects of climatechange in the Mediterranean region. In fact,during the 20th century, with an evident accel-eration since 1970, climate change hasalready led in the Mediterranean to an aver-age increase of temperatures of about 2 °C inthe south-western Europe (Iberian Peninsula,southern France) and North African territoriesand rainfall decreased up to 20% in someregions of southern and eastern Mediterranean(FAO 2013).

It is also expected that ongoing climatechanges continue and increase in the Mediter-ranean region in the next decades, mainlyaffecting air and sea temperatures and pre-



Figure 5 – Geographical distribution of the number of dry months in the Mediterranean region as compared with the limits of Olea(Desfontaines 1975 in de Montgolfier 2002). 1 = no dry months; 2 = one to three dry months; 3 = three to five dry months; 4 = more than five dry months.


cipitation patterns as described in many textsas in the Strategic Framework for Mediter-ranean Forest (FAO 2013). By 2100, theMediterranean climate is set to continue tochange with temperatures rising further by anaverage of 2°C to 4°C, while rainfalls drop by4% to 30% (IPCC 2007; FAO 2013). In accor-dance with the predictions, a scenario of ageneral increase of 2ºC and a decrease of 20%

in precipitation was used resulting in a differ-ent geographical distribution of the Mediter-ranean climate, as shown in Figure 7.

It is apparent that aridity increases in the southand east, shifting the dry limit of the Mediter-ranean climate to the north. On the otherhand, new areas in the north will change to aMediterranean type of climate, from theFrench Landes to large areas in Turkey.



Figure 6 – Current geographical distribution of the Mediterranean climate resulting from the models developed using the temperature and aridity limits mentioned in the text.

Figure 7 – Scenario of climate change with increased temperature and decreased precipitation as described in the text.


Other possible uses of the models

These models can also be used, for example,to assist in predicting future distribution ofspecies using, for example, niche-based mod-els of species climatic envelops as proposedby Thuiller et al. (2005). These distributionsare commonly associated to thresholds inaverage temperature or precipitation in spe-cific periods (months or seasons). Using thistype of models it is possible to evaluate thespatial distribution of a species as a functionof geographic location, distance to coast andelevation.

The results confirm, at the scale of theMediterranean region, that simple equationscan explain a relatively important proportionof the variations of temperature and precipi-tation in the Mediterranean region. The sim-plicity of these models relies on the fact thatthey use spatial variables that are widelyavailable or computed in GIS (latitude, longi-tude, elevation, and distance to the coast), andthat they confirm the general effects of thosevariables in temperature and precipitation.This is particularly important when globalmodels tend to be so complex that it becomesdifficult to decode the relative importance ofthe various factors involved in the vegetationdistribution.

References

Bagnouls F. & Gaussen H., 1957. Les climats biolo-giques et leur classification. Annales de Géographie.Société de géographie. Vol. 66 : 193-220.

Briggs R. D. & Lemin R. C. Jr, 1992. Delineation of cli-matic regions in Maine. Canadian Journal of ForestResearch 22: 801-811.

Brown D. P. & Comrie A. C., 2002. Spatial modeling ofwinter temperature and precipitation in Arizona andNew Mexico, USA. Climate Research 22: 115-128.

Daget P., 1977. Le bioclimat méditerranéen: caractèresgénéraux, modes de caractérisation. Plant Ecology34 : 1-20.

Daly C., Neilson R. P. & Phillips D. L., 1994. A statis-tical-topographic model for mapping climatologicalprecipitation over mountainous terrain. Journal ofapplied meteorology 33: 140-158.

De Martonne E., 1926a. L’indice d’aridité. Bulletin del’Association des géographes français 9 : 3-5.

De Martonne E., 1926b. Une nouvelle fonction clima-tologique. L’indice d'aridité. La Métérologie 2, 449-458.

De Martonne E. 1942. Nouvelle carte mondiale de l’in-dice d'aridité. Annales de Géographie, Société degéographie. Vol. 51 : 241-250.

De Montgolfier J., 2002. Les espaces boisés méditerra-néens : situation et perspectives. Economica, Paris,Ed. Sophia Antipolis, Paris,.

Dobrowski S. Z., Abatzoglou J. T., Greenberg J. A. &Schladow S. G., 2009. How much influence doeslandscape-scale physiography have on air tempera-ture in a mountain environment? Agricultural andForest Meteorology 149: 1751-1758.

EEA D. S. & Bourdeau P., 1995. Europe’s Environment:The Dobris Assessment. Copenhagen.

Emberger L., 1930. La végétation de la région méditer-ranéenne: essai d’une classification des groupe-ments végétaux. Revue de Botanique, Librairiegénérale de l’enseignement, 503 : 642-662 ; 504 :705-721.

FAO, 2013. Strategic Framework on MediterraneanForests. Consulted in http://www.fao.org/fores-try/36306-0496de1ce9c0218e588fd41ef7353afcd.pdf.

Friend A. D. &Woodward F. I., 1990. Evolutionary andecophysiological responses of mountain plants tothe growing season environment. Advances in Eco-logical Research 20: 59-124.

Gaussen H., 1956. Le XVIIIe congrès international degéographie, Rio de Janeiro, août 1956. Ann Geogr353 : 1-19.

Gaussen H. & de Phillips A., 1958. La limite eumédi-terranéenne et les contrées de transition. Sous com-mission de coordination de la FAO des quationsforestières méditerranéennes. FAO, Rome.

Giacobbe A., 1949. Le basi concrete per une classifica-zione ecologica della vegetazione italiana. Arch.Bot. XXV-vol. IX, Valbonesi Forli : 65-177.

Goodale C. L., Aber J. D. & Ollinger S. V., 1998. Map-ping monthly precipitation, temperature, and solarradiation for Ireland with polynomial regression anda digital elevation model. Climate Research 10: 35-49.

IBM Corp. Released, 2013. IBM SPSS Statistics forWindows, Version 22.0. Armonk, NY: IBM Corp.

IPCC, 2007. Climate Change 2007: Synthesis Report.Contribution of Working Groups I, II, and III to theFourth Assessment Report of the IntergovernmentalPanel on Climate Change (Core Writing Team,Pachauri R.K. & Reisinger A. (eds.)). IPCC,Geneva, Switzerland. 104 p.

Johansson B. & Chen D., 2003. The influence of windand topography on precipitation distribution in Swe-den: statistical analysis and modelling. Internatio-nal Journal of Climatology 23: 1523-1535.

Lang R., 1915. Versuch einer exacten Klassification derBoden Mitt. Klimatischer und geologischer Hin-sicht. Int., Bodenk 5 : 312-346.

MacArthur R. H., 1984. Geographical ecology: patternsin the distribution of species. Princeton UniversityPress.

Mazzoleni S., Porto A. L. & Blasi C., 1992. Multiva-riate analysis of climatic patterns of the Mediterra-nean basin. Plant Ecology 98: 1-12.

Mhirit O. 1999. Climats et bioclimats de la forêt. In: LeGrand Livre de la forêt marocaine. Ed. Mardaga,Sprimont, Belgium : 54-60.

Osborn H. B., 1984. Estimating precipitation in moun-tainous regions. Journal of Hydraulic Engineering110: 1859-1863.

Pavari A. 1954. Fundamentos Ecológicos e Técnicos daSilvicultura nos Países Mediterrânicos. Trad. A.Monteiro Alves. ISA, Lisboa.




Quézel P., 1977. Les forêts du bassin méditerranéen.Paris, UNESCO (Notes techniques du MAB nº 2).

Rivas-Martinez S., 1981. Les étages bioclimatiques dela végétation de la péninsule Ibérique. Anal. Jard.Bot. Madrid 37(2) : 251-268.

Rivas-Martinez S. & Navarro G., 1994. Mapa Biogeo-gráfico de Suramerica. In: Classificación Bioclimá-tica de la Tierra. Folia Botanica Madrilensis. Uni-versidad Complutense de Madrid 12 : 23.

Rivas-Martinez S., Sanchez-Mata D. & Costa M., 1999.North America Boreal and Western TemperateForest Vegetation. Itinera Geobotanica 12: 5-316.

Thuiller W., Lavorel S., Araújo M., Sykes M. & ColinPrentice I., 2005. Climate change threats to plantdiversity in Europe. Proc. Natl. Acad. Sci. U.S.A.102(23): 8245-8250.

UNESCO-FAO, 1963. Carte bioclimatique de la régionméditerranéenne. Notice explicative. (Recherchessur la zone aride, XXI). Unesco, Paris et FAO,Rome.

Walter H., 1955. Die Klimadiagramme als Mittel zurBeurteilung der Klimaverhältnisse für ökologische,vegetationskundliche und landwirtschaftliche Zwecke.Berichte der deutschen botanischen Gesellschaft 68 :331-344.

Walter H. & Lieth H., 1960. Klimadiagramm-Weltatlas.Gustav Fischer Jena.

Waring R. H. & Running S. W., 1998. Forest ecosys-tems: analysis at multiple scales, 2nd ed. AcademicPress, San Diego.




Documents

Climatic Patterns in the Mediterranean region...ecologia mediterranea – Vol. 40 (1) – 2014 49 Abstract Climate and vegetation patterns in the Mediter - ranean region are known