Materials and methods

Study area. The physiographic province of SMS is located southwest of Mexico, between 15° 40’- 19° 40’ N, and 94° 45’- 104° 40’ W. It has an extension of 195,700 km². It comprises the southeast of Jalisco, over 80 % of Colima, the southern region of the State of Mexico, Michoacán, Puebla and almost the complete extension of Guerrero, with the Istmo de Tehuantepec as the geographic limit in Oaxaca. The altitudinal gradient ranges from 0 to 3,710 m asl (Figure 1), (^{INEGI 2001}).

Figure 1 Location map of the study area the Sierra Madre del Sur, Mexico According with ^{INEGI (2001)}. 

Species richness and altitudinal zoning. Specimens of Quercus at National Herbarium (MEXU), the herbaria from Facultad de Ciencias (FCME), Universidad de Guadalajara (IBUG) and Instituto de Ecología A.C. (IEB) were revised and curated to create a database of 3,267 non-duplicated records, which was used to determine the species richness in the SMS and obtain information about the altitudinal distribution of the species diversity. Based on altitudinal data, a series of boxplots were calculated to identify the altitudinal variation patterns of each species and each section. We also evaluated the altitudinal distribution of species richness both by section and jointly. Analyses were conducted in ^{R Core Team (2013)}.

Using the Q₁ - Q₃ interval of the boxplot, a data matrix was created based on the presence/absence of each species in intervals of 100 m. The rows of the matrix represent altitudinal zones and the columns represent species. Zero equals absence and one equals presence of the species in the different altitudinal intervals. Based on this matrix, a similarity matrix was obtained using the Sorensen-Dice index applying the “ade4” package (^{Dray & Dufour 2007}). Afterwards, a cluster analysis was carried out using the Ward method. The resulting dendrogram allowed to recognize the altitudinal zones of the SMS based on the distribution of the oak species diversity. Using these altitudinal zones, we generated a heatmap that reflects the distribution of each species in the different altitudinal zones based on the percentage of records of each species for a given altitudinal zone. To eliminate the potential effects of outliers, we considered that a species belongs to a certain altitudinal zone only if their records in a given altitudinal zone were greater than 25 %, this percentage corresponds to the value obtained from the boxplot analysis. Finally, the altitudinal zoning was mapped using QGis 3.8.3 (^{QGIS Development Team 2019}).

Climate differentiation of the altitudinal zones. A total of 19 bioclimatic variables were downloaded for the 3,267 records from the WorldClim database v1.4 (^{Fick & Hijmans, 2017}) (www.worldclim.org). From these records, a correlation analysis was conducted to eliminate those highly correlated variables. A total of nine bioclimatic variables (Bio2, Bio3, Bio4, Bio5, Bio6, Bio14, Bio15, Bio16 and Bio18) were used to evaluate which climatic variables have greater discriminating power among different altitudinal zones detected through cluster analysis. However, since not all variables discriminate in the same way, a stepwise variable selection was conducted. As a general criterion to select a variable, F value associated with each variable was used, in such a way that if the F value was small, the variable would discriminate little, while large values of F suggest greater discriminating power. Initially, we selected the variable with the highest F value, subsequently the F value was evaluated for the unselected variables and a new variable with the highest F value was selected. This procedure was repeated until a variable with the highest F value was statistically non-significant. The discriminant analysis was conducted with the software ^{SAS (2012)} using a linear discriminant method.

Results

Species richness and altitudinal zoning. A total of 61 species of the genus Quercus were registered in the SMS, representing 37.9 % of the total species of the genus reported for Mexico (^{Valencia-A. 2004}). From these, 36 belong to the section Lobatae, 24 to the section Quercus and one to the section Virentes (^{Denk et al. 2017}). A total of 45 species are endemic to Mexico and 13 are endemic to the SMS (see Appendix 1). The altitudinal distribution of total species richness had its highest value (20-28 species) between 1,500 - 2,000 m with an important second peak of species richness (20 species) at 2,200 m, which coincided with the third peak of species richness detected exclusively for the Quercus section (Figure 2). From this altitude, the total species richness decreases constantly until reaching less than five species above to 2,800 m (Figure 2). When species richness was analyzed by section, a similar pattern was observed between section Lobatae and section Quercus. The section Lobatae had the highest species richness values between 2,300 - 2,400 m, while section Quercus showed three important peaks of species richness at 1,600, 1,800 and 2,200 m (Figure 2).

Figure 2 Altitudinal pattern of species richness of Quercus per altitudinal interval each 100 m in the Sierra Madre del Sur. Blue line = Total richness, Red line = Total richness of Lobatae, Green line = Total richness of Quercus. 

We found 18 oak species of the section Lobatae with altitudinal ranges > 1,000 m, Q. acutifolia, Q. aristata, Q. calophylla, Q. castanea, Q. conzattii, Q. crassifolia, Q. elliptica, Q. confertifolia, Q. grahamii, Q. laurina, Q. mexiae, Q. nixoniana, Q. planipocula, Q. salicifolia, Q. scytophylla, Q. trinitatis, Q. urbanii and Q. uxoris. In altitudinal ranges < 1,000 m we found Q. acherdophylla, Q. benthamii, Q. crassipes, Q. crispifolia, Q. cualensis, Q. depressa, Q. eduardi, Q. hintonii, Q. iltisii, Q. jonesii, Q. mulleri, Q. pinnativenulosa, Q. rubramenta, Q. sartorii and Q. skinneri. The species of the section Quercus with altitudinal range > 1,000 m were Q. corrugata, Q. glabrescens, Q. glaucescens, Q. glaucoides, Q. insignis, Q. laeta, Q. liebmanii, Q. magnoliifolia, Q. martinezii, Q. microphylla, Q. obtusata, Q. peduncularis, Q. resinosa, Q. rugosa, Q. segoviensis and Q. sororia. The species of the section Quercus with altitudinal ranges < 1,000 m were Q. centenaria, Q. deserticola, Q. frutex, Q. greggii, Q. macdougallii, Q. polymorpha, Q. sebifera and Q. subspathulata. While Q. oleoides, the only species of the section Virentes in the SMS, also had an altitudinal range < 1,000 m.

The cluster analysis segregated the SMS in three altitudinal zones (Figure 3A-B). The first altitudinal zone (Altitudinal A zone) corresponds to a level ranging from 500 to the altitudinal band of 1,100 m (orange color). Geographically, this zone comprises the Pacific Coastal Plain and the Balsas river Basin. In this altitudinal zone, we found 14 oak species (Figure 3C), two of them as exclusive to this level, Q. oleoides (the only species of the section Virentes of the region) and Q. tuitensis, and three endemic species to SMS (Q. iltisii, Q. salicifolia and Q. tuitensis). The altitudinal B level (green color) (Figure 3 A-B) corresponds to the interval between 1,200 m and the altitudinal band of 2,200 m, the largest species richness of oaks was registered in this level with 47 species, eight of them exclusive for this level (Q. benthamii, Q. eduardi, Q. hintonii, Q. jonesii, Q. mulleri, Q. pinnativenulosa, Q. sebifera, Q. skinneri) (Figure 3C) and nine endemic species to SMS (Q. centenaria, Q. cualensis, Q. liebmanii, Q. martinezii, Q. mulleri, Q. nixoniana, Q. rubramenta, Q. salicifolia, Q. uxoris ). Likewise, the altitudinal A and B zones showed seven species in common (Q. acutifolia, Q. elliptica, Q. confertifolia, Q. glabrescens, Q. iltisii, Q. salicifolia and Q. sororia). The altitudinal C level ranges from 2,300 to > 3,000 m (blue color). Its species richness resulted in 19 species, with four of them exclusive to the level (Q. acherdophylla, Q. crassipes, Q. depressa and Q. greggii) and four endemic species to SMS (Q. macdougallii, Q. martinezii, Q. nixoniana and Q. rubramenta).

Figure 3 A) Dendrogram produced by the classification of oak distribution based on the Sorensen-Dice index. Altitudinal zone A corresponds to zones ranging from 500 to 1,100 m asl (orange color), altitudinal zone B corresponds to zones ranging from 1,200 to 2,200 m asl (green color), altitudinal zone C corresponds to zones ranging from 2,300 to > 3,000 m asl (blue color), on the X axis = similarity index; B) Altitudinal zonation found in this study for the Sierra Madre del Sur based on the oak distribution; C) A heatmap of the altitudinal distribution for the 61 oak species across the three altitudinal zones. Dark colors represent low percentages of records for each species in each altitudinal zone while light colors represent high percentages of records for each species in each altitudinal zone. 

Based on the altitudinal zonation, the oak species richness in the SMS exhibited a maximum of 47 species in the altitudinal B zone showing a unimodal pattern throughout the altitudinal gradient. Therefore, we propose that the SMS has a core altitudinal zone, on an elevation between 1,200 and 2,200 m, which contains the highest amount of species richness and endemic oaks.

Climate differentiation of the altitudinal zones. The results of the discriminant analysis showed that mean diurnal range (Bio2; P = 0.000), max temperature of warmest month (Bio5; P = 0.000), precipitation of wettest quarter (Bio16; P = 0.0036) and precipitation of warmest quarter (Bio18; P = 0.000) are the most important variables to discriminate among the three altitudinal zones (Figure 4). The overall model was highly significant (Wilks = 0.68; P = 0.0001) and showed a moderate percentage of correct classification of the populations (Altitudinal zone A = 72 %; Altitudinal zone B = 66 %; Altitudinal zone C = 44 %), evidencing the existence of different temperature and precipitation regimes as important factors that contribute to the differentiation of species composition throughout the altitudinal belts.

Figure 4 Sample ordination produced by the discriminant analysis. The ellipses represent the 95 % confidence. Each point represents a species records for the different altitudinal zones. 

Discussion

The altitudinal zoning found in this study, reveals the presence of three different altitudinal zones and two important species turnover points. Each of these altitudinal zones showed a particular species richness, as was found by ^{Toledo-Garibaldi & Williams-Linera (2014)} who found three groups of sites along an elevation gradient in Veracruz; and also by ^{Olvera-Vargas et al. (2010)} in a Quercus dominated forest located towards the north of the SMS, who also found a floristic gradient with three floristic zones and two important species turnover points. Both studies were conducted including only trees of different species and genus.

The altitudinal A zone had the least oak diversity, with only 14 species presents. It showed the lowest elevation within the study area, and resulted closely related to tropical and dry climates, which are distinctive of tropical rainforest, dry oak forest, and tropical deciduous forest. This concurs with the presence of certain species such as Q. crispifolia, Q. glaucescens, Q. sapotifolia, Q. oleoides and Q. tuitensis which are characteristic of these altitudes and type of environment (^{Rzedowski 1978}, ^{González-Villarreal 2003}, ^{Valencia-A. 2004}). The discriminant analysis showed that the altitudinal A zone differs from the others altitudinal zones due mainly to a different precipitation regime during the warmest quarter (Bio 18; Figure 4). This suggests that the differences in the precipitation regimes of the warmest quarter among the different altitudinal zones could be one of the determining factors that limit the distribution of characteristic species in the altitudinal zone A such as Q. crispifolia, Q. glaucescens, Q. oleoides, Q. sapotifolia and Q. tuitensis.

On the other hand, the altitudinal B zone had the greatest species richness of oaks (47 species), the largest number of exclusive, and endemic species which is why it can be considered as the core region for the genus Quercus inside the SMS. In this region, the mean diurnal range (Bio2) and max temperature of the warmest month (Bio5) were the most important variables. Therefore, the temperature at certain times of year could be considered as an important environmental filter in the distribution of species diversity exclusive to the altitudinal B zone such as Q. benthamii, Q. eduardi, Q. hintonii, Q. jonesii, Q. mulleri, Q. pinnativenulosa, Q. sebifera. In the same way, these variables could be acting like environmental barriers that prevent the establishment of species with tropical and xeric affinities in this altitudinal zone.

This pattern of species richness known as mid-elevation peaks is the most common in the mountains (^{McCain 2009}, ^{McCain & Grytnes 2010}), and is generally attributed to the mid-domain effect, which proposes that this pattern is caused by the overlapping of the altitude ranges of the species as it approaches the central part of the mountain system (^{Sanders 2002}, ^{Colwell et al. 2004}). For Quercus species, a complementary hypothesis is one that is related to climate which state that in the altitudinal B zone there is a higher diversity of temperate and humid climates in relation with altitudinal A and C zones (^{García-CONABIO 1998}). Altogether, these hypotheses, in addition to the great availability of niches occasioned by heterogeneous topography of the zone, would explain the great richness of the oak species in this altitudinal zone (^{Hipp et al. 2018}).

Finally, the altitudinal zone C showed a significant decrease in the species richness of oaks, registering only 19 species (Figure 2). The low similarity coefficient between the altitudinal B and C levels (Figure 3A), reflects a low composition of oak species diversity shared in both levels, being the second important changing point in the species turnover diversity of Quercus in the SMS. This important point in the species turnover diversity of Quercus is apparently related to differences in the precipitation regime and the mean temperature of the warmest quarter, which prevents a large number of species distributed in the altitudinal B zone from being present in the altitudinal C zone.

Our study found that the largest species richness of species of Quercus is concentrated between 1,200 and 2,200 m, which adjusts itself to the pattern of species richness at mid-altitudes (mid-elevation peaks), and it also corresponds to one of the four distribution patterns of species richness proposed by ^{McCain (2009)} and retaken by ^{McCain & Grytnes (2010)}. A similar pattern with high species richness at mid-altitudes, though with a wider range, was cited by ^{Sabás-Rosales et al. (2015)} for oaks in the State of San Luis Potosí, where the altitude ranges from 1,201 and 2,700 m, and holds the largest species richness of species of the genus Quercus. Such a pattern was also found in the western region of Mexico for tree species (^{Olvera-Vargas et al. 2010}) and in a study of six genera of the Mexican cloud forest (^{Alcántara et al. 2002}). This pattern corresponds to the one described by ^{Guo et al. (2013)} for trees, and to the study of ^{Arenas-Navarro et al. (2020)} who found a similar pattern in Serranias Meridionales of Jalisco for Quercus species.

Our results showed that precipitation regime during the warmest quarter, defines the altitudinal A zone. In the altitudinal B zone, the mean diurnal range, and max temperature of warmest month were the most important variables, and in altitudinal C zone were the precipitation regime and the mean temperature of the warmest. Then, different climatic factors are important in driving elevational variation of plant species richness and taxa distribution along elevational gradients. This is a result obtained in other studies such as ^{Olvera-Vargas et al. (2010)}, ^{Salas-Morales et al. (2015)}, ^{Salas-Morales & Williams-Linera (2019)} who stated that the climate is fundamentally important in taxa distribution along elevational gradients; and ^{Arenas-Navarro et al. (2020)} pointed out that temperature and precipitation are important filters of species distribution along an elevational gradient. Quercus has species adapted to different environmental conditions, but this study supports that the main climatic factors driving the elevational distribution of richness oaks at SMS, are those related with temperature and the precipitation regime. However, studies of other environmental factors related to soil, relief, atmospheric moisture, herbivory, dispersion vectors and competition must be conducted in the future to gather more information in order to understand more accurately the environmental limits of oaks species.

In addition to the ecological factors mentioned above, particularly those climatic, which in part explain the unimodal pattern of oaks in the SMS, there are historical and evolutionary factors that allow proposing an explanation for the emergence of this pattern. These can be grouped into three: I) The main affinity of Quercus for temperate climates; II) the middle lands as receptors of multiple arrival events originated by the glacial cycles during the Pleistocene, and the permanence in situ of taxa that they found in the middle land refuges; and III) the ability of species of Quercus to colonize new habitat.

I) The genus Quercus has a northern temperate affinity (^{Hipp et al. 2018}) and is widely distributed in the northern hemisphere in a wide range of habitats and climates. Despite this ecological diversity even in tropical lands like section Virentes (^{Cavender-Bares et al. 2015}), the species show several traits that recall their temperate origin (^{Cavender-Bares 2019}). After their possible origin in North America in the Eocene, American oaks diversified southward, and they arrived in Mexico in the Miocene, where their speciation rate increased in Mexican mountains (^{Hipp et al. 2018}), and reached the highest world diversity, currently exceeding 160 species (^{Valencia-A 2004}).

II) Climate changes during Pleistocene had great impact in all existing biota, provoking its latitudinal and altitudinal migration, reduction or fragmentation of species, but also its survival in refugia, expansion, extinction and diversification of taxa (^{Hewitt 2004}, ^{Turner 2004}, ^{Ramírez-Barahona & Eguiarte 2013}). During the coldest periods of the Pleistocene, many tropical species became extinct, creating available habitats for species with temperate affinities that could migrate towards lower latitudes and altitudes, and that could persist due to their preadaptation to certain environmental changes, and their ability for rapid migration (^{Kremer & Hipp 2019}), reaching altitudes near to sea level, which is the case of some oak species (^{Hewitt 2004}, ^{Turner 2004}, ^{Soltis et al. 2006}, ^{Cavender-Bares 2016}) such as Q. oleides and Q. sapotifolia. During the interglacial periods, temperature raised to the current temperature or even 3 degrees above (^{Caballero et al. 2010}), provoking new migrations to greater altitudes and latitudes (^{Bush et al. 2004}, ^{Caballero et al. 2010}, ^{Ramírez-Barahona & Eguiarte 2013}).

^{Mastreta-Yanes et al. (2015)} indicated that Mexican mountains presented minor climatic changes during Pleistocene compared to those global, which provoked that temperate and semi warm-humid climates (which are preferred by oaks) were temporarily displaced. This enables organisms to migrate, primarily along an altitudinal gradient rather than a latitudinal one (^{McGlone 1996}, ^{Jetz et al. 2004}, ^{Turner 2004}) and survive the oscillations in the same region (^{Hewitt 2004}). Palynological studies mention the presence of montane taxa (Quercus, Alnus, Juglans, Liquidambar, Magnolia and Podocarpus) growing at lower altitudes than the present ones during the coldest periods of the Pleistocene (^{Bush & Colinvaux 1990}, ^{Colinvaux et al. 1996}, ^{Hodell et al. 2008}, ^{Bush et al. 2009}).

The mid altitudes acted as corridors or paths and even as transition zones between the highest and lowest lands and received multiple taxa arrival events. Some biota went across the mid regions as many times as climatic changes took place, that is, twice the cycles there existed. Another biota found in this middle zone a niche or shelters where they could settle, these migrations or stays increased the specific diversity in this level.

III) Quercus species are characteristic for presenting high genetic diversity produced by frequent hybridization among species (^{Petit et al. 2003}, ^{Hipp et al. 2020}), which can produce not only preadaptations, specific diversification, and high rates of evolutionary divergence (^{Kremer & Hipp 2019}), favoring species migration into new and different habitats (^{Cavender-Bares 2016}, ^{Hipp et al. 2020}), so they went up and down during the climatic changes. If slope differences and orientation, micro-orography, topography, soil, solar exposition, wind, precipitation, evaporation and the diversity of climates in this zone are considered, a vast collection of very heterogeneous habitats in the mid regions were produced (^{Turner 2004}, ^{Martinelli 2007}, ^{Espinosa et al. 2016}, ^{Sam et al. 2019}). According to ^{Tews et al. (2004}), there is a positive correlation between habitat heterogeneity and species diversity. These habitats become available for organisms and their heterogeneity is greater than in higher and lower altitudes, which is an advantage for species with great genetic variation and temperate affinity (such as many of the American species of Quercus), and also with the ability to establish themselves in new habitats.

Finally, we want to point out that the number of altitudinal zones, species turnover points, and the richness pattern found in this study depend on the method being used (^{Bach & Gradstein 2011}), the site of study (^{Mastreta-Yanes et al. 2018}) and the analyzed taxa (^{Arenas-Navarro et al. 2020}). However, a similar altitudinal pattern and altitudinal zones could be possibly found in the case of temperate taxa and those associated with oaks in the SMS. On the other hand, according to the limited available resources for the management and conservation of biodiversity, it is important to prioritize certain regions and to consider the importance of oaks in montane ecosystems of the northern hemisphere. Therefore, it is estimated that it would be a good starting point for further studies in management and conservation of species.