Spatial modeling of the ecological niche of Pinus greggii Engelm (Pinaceae): A species conservation proposal in Mexico

Studies in Mexico have shown that the genus Pine has always been under evolutionary changes, however currently they have accelerated as a result of human activities. Pinus greggii is a species restricted by particular environmental conditions of the Sierra Madre Occidental, which is of socio-economic importance in terms of wood production and provides environmental services to the ecosystem. Species distribution models are a relevant geospatial tool in decision making, and notable applications exist such as identifying areas of distribution and zones susceptible to climate change. The objectives of this study were: 1) model and quantify the current distribution, and possible future distribution under four scenarios of climate change; 2) identify the most relevant environmental variables that drive changes in distribution; and 3) to propose adequate zones for the species’ conservation in Mexico.

wherein the species capacity occupies a geographic space according to a set of variables in conjunction with continuous or categorical character of the region's climatology, edaphology, and topography (Guisan and Zimmermann 2000).
MaxEnt is used to predict the current and future spatial distribution (  . MaxEnt is one of the most-used spatial distribution algorithms and is considered to be the best method of estimating at spatial level in relation to the suitability of species (Ortega-Huerta and Peterson 2008; Kumar and Stohlgren 2009). Global circulation models are used to simulate future climate behavior (Delgado and Suárez, 2009; Fernández-Eguiarte et al. 2015), and are therefore a good alternative to estimating and evaluating climate changes and how they will affect future species distribution (Peterson et  Although it is known that a real threat exists due to the negative effect of climate change on the future of the majority of species in the Pinaceae family, the prognosis for species (such as P. greggii) that inhabit the arid and semiarid climates, could be more encouraging, which would imply the current and future environmental conditions will be similar. Thus, most areas of current distribution of the species could persist or be preserved through time (niche conservatism theory) (Soberón and Miller 2009;Peterson 2011).
In this study, we analyzed geographic and environmental records (climatic, topographic, and edaphic) of P. greggii to help us to delimit, estimate, and identify the most relevant environmental variables in current and future distributions, as well as to propose conservation areas within the species' natural distribution in Mexico. In response to these objectives, the following speci c goals were formulated: 1) delimit and estimate the area of distribution of the P. greggii; 2) identify the most relevant variables in the current and future (2041-2060) distribution; and 3) propose conservation areas for P. greggii to thrive within its current natural area of distribution in Mexico.

Study area
The present study includes the physiographic provinces of the Sierra Madre Oriental and a portion of the Neovolcanic Transversal Belt, speci cally in the subprovince Plains and Sierras of Queretaro and Hidalgo, and Lagoons and Volcanoes of Anahuac (Fig. 1), and between the longitudinal coordinates of 97°0'0''-105°0'0'' and latitudinal coordinates of 18°0'00''-30°0'00'' (Instituto Nacional de Estadística y Geografía [INEGI] 2018). The geographic region of the provinces and subprovinces were considered to be the spatial area of modeling M, which is the geographic and environmental space where the presence of a species has been registered and is delimited according to the biological knowledge of it and its dispersal capacity (Soberón and Peterson 2005) for P. greggii. The highest elevation point corresponds to the Pico de Orizaba (5,610 masl), while the lowest is at sea level on the coast of Veracruz (Instituto Nacional de Estadística y Geografía [INEGI] 1998). The precipitation ranges from 154 to 3,866 mm, with an average of 685.09 mm and a mean annual temperatures range from − 2 °C to 28 °C (Fick and Hijman 2017).

Geographic Records
In this study four different sources of records about P. greggii were used, three of which are from a database and are included in o cial pages for their free download: 262 records were retrieved from the Global Biodiversity Information Facility (GBIF) which is an international organization whose purpose is to accumulate and provide scienti c data (GBIF 2018) http://www.gbif.org/occurrence/download/0053032-160910150852091; 40 records of were retrieved from the database of the National Herbarium of the Universidad Nacional Autonoma de Mexico (MEXU 2019) http://www.ib.unam.mx/botanica/herbario ; 63 records were retrieved from the Global Network of Biodiversity Information by CONABIO that contain different national and international collections (REMIB 2019); and 73 records were generated through dendrochronological expeditions made by personnel of the National Dendrochronological Laboratory of the INIFAP CENID-RASPA in 2018-2019.
In total, 438 records were obtained, and then cleaned up via the Niche ToolBox platform of the National Commission for the Knowledge and use of Biodiversity (Osorio-Olvera et al. 2019) to eliminate double records and spatial separations of less than 1 km of lineal distance between records. This step helped to avoid the autocorrelation spatial effect and a subestimation of the distribution models (Peterson and Nakazawa 2008; Monterrubio-Rico et al.

2016).
Current And Future Climatic Variables Current climatic information was obtained from the 19 bioclimatic layers (Table 1)  Intercomparison Project Phase 5 2013) of the IPCC. The bioclimatic layers of these models were downloaded with two radiative forcings of 4.5 (constant trajectories of CO 2 ) and 8.5 (high trajectories of CO 2 ) for 2041-2060 (Table 1), with a spatial resolution of 30 seconds (~ 1 km 2 ).
In Table 1, the 19 current and future bioclimatic layers downloaded from WorlClim ver. 2.0 are presented. and soil organic carbon content were extracted and adapted to an ASCII standard format with a spatial resolution of 30 seconds (~ 1 km 2 ).

Variable Selection
For variable selection, a minimum convex polygon was generated according to the presence records of P. greggii

Current Distribution Modeling
The MaxEnt ver. 3.4.1 algorithm was used to model current distribution ). This algorithm was chosen because it is one of the most-used methods in the eld study of potential distribution and generates accurate geographic predictions from only presence records (Elith et al. 2006; Ortega-Huerta and Peterson 2008). Seventy-ve percent of records were used to train the model and 25% to validate it. The BIO1, BIO7, BIO11, BIO15, BIO17, BIO19, DEM, SLO, and pH variables are in ASCII format ( Table 2).
The modeling criterion comprise internal replication by cross-validation, 1,000 iterations, logistic output, 100 replicates, and a convergence threshold of 0  Model calibration was evaluated through the standardized coe cient of the Akaike information criterion (AICc), which provides model information, such as

Relevant environmental variables
The contribution of environmental variables in the current and future distribution were evaluated with the Jacknife test, which allowed us to identify and quantify the percent of contribution or relevancy of each variable in the growth and development of P. greggii during the current and future periods in the study area ).

Modeling of the current distribution
The results of the 100 replicates of the AUC were from 0.879 to 0.886 for training and from 0.797 to 0.93 for validation, which indicated the development of the models was good. For P. greggii in Mexico, the best model resulted in a partial ROC value of 1.90 (Table 3), an AUC of 0.881, and 0.930 for training and validation.
The results estimated the potential current distribution of P. greggii in an area of 617,706.04 ha (Fig. 2) within the study area.
The majority of the current estimated area of P. greggii is located in the states of Nuevo

Relevant Variables In The Current And Future Distribution
The relevant variables in the current distribution were BIO1, BIO11, SLO, BIO19 and BIO7, which contributed 81.2% of the model's variability (Fig. 3). The relevant variables for the 2041-2060 CNRM-CM5 RCP 4.5 model were BIO1, SLO, BIO19, BIO15 and BIO11, and the relevant variables for the RCP 8.5 model were BIO1, SLO, BIO15, BIO19 and BIO11, with contributions of 83.4% and 87.9% respectively (Fig. 4).
Current and future area of P. greggii in Mexico  The results of the CNRM-CM5 (RCP 4.5) foresee a raise of 0.7 °C in mean annual temperature, which will reduce the species' ecological niche by 21 The results of the four climate change scenarios predict a range from slight reduction (HadGEM2-ES RCP 8.5; 7.8%) to more extensive reduction (CNRM-CM5 RCP 4.5; 21.8%) in the ecological niche of P. greggii for 2041-2060. According to these models, the rise of the mean annual temperature is the principal variable responsible for the reduction in suitable ecological niche for P. greggii within their natural future distribution.

Conservation Of The Ecological Niche
The

Relevant Environmental Variables
The most important variable for this study was the BIO1 (mean annual temperature), coinciding with diverse studies related to the Pinus habitat, which have shown that this variable plays an important role for at least ten different species One aspect to consider is that an increase in temperature generates an accelerated development of the species, but with reduction in their growth (Gennaretti et al. 2014). This behavior is attributed to elevated temperatures promoting an increase in evapotranspiration and, consequently, a metabolic alteration that impacts the assimilation of photosynthates ( In the study by Aceves-Rangel et al. (2018), the variable BIO11 (mean temperature of coldest quarter) was found to be important for P. arizonica, which is ecologically associated with P. greggii and found in similar climate conditions (López-Peralta and Sánchez-Cabrera 1996).
García-Aranda et al. (2018) found the variable BIO11 plays an important role in the distribution of P. nelsonii in northwestern Mexico, a species associated with P. greggii in the Sierra Madre Oriental and the Neovolcanic Transversal Belt. The mean value of the BIO11 variable was 4.3 °C in the present study and is very similar to the mean value found for P. nelsonii of 4.6 °C -an adequate temperature for the growth and development of P. greggii .The SLO for Garcia-Aranda et al. (2018) found that this topographic variable presented great relevance in three Pinus species with distribution restricted to northeastern Mexico (P. cembroides, P. culminicola, and P. nelsonii), contributing to the model with 21.1% signi cance; close to that found in this study of P. greggii with 18.3% of contribution to the current distribution model.
The SLO is a signi cant variable for the establishment of tree species located on steep soils. Muñoz et al. (2012) mention that the P. greggii is present on slopes with up to a 5% grade; the mean slope value found in this study is 8%.
The BIO19 variable (precipitation of coldest quarter) was considered an important variable for at least seven Abies species in Mexico (Martínez-Méndez et al. 2016). According to the study of Aceves-Rangel et al. (2018) BIO19 was relevant for P. lumholtzii at a national level with 8.2% as the lowest value, in comparison to the 12.8% found for P. greggii. In this analysis, the mean BIO19 was 427 mm, however P. greggii is adapted to zones with low precipitation ranging from 293 to 747 mm (Ramírez-Herrera et al. 2005).
The annual oscillation of the temperature (BIO7) for the current distribution of P. greggii in the present study had a mean value of 25.4 °C, which, according to the study zone, is a suitable temperature in the ideal annual range for the species development. This variable is relevant in other studies as well: However, this type of study has not been widely applied to Pinus species that grow in the arid and semiarid regions of the country.
According to the increases of temperature forecast by the CNRM-CM5 and HadGEM2-ES models with two radiative forcings (RCP 4.5 and 8.5) for 2041-2060, the ecological niche will decrease between 7.8% and 21.8% within the endemic zone of P. greggii, but with a tendency to modify their distribution with consideration for the climatic change scenarios between RCP 4.5 and RCP 8.5 as are mentioned by Gavilán (2008).
In However, these types of studies are very scarce for the Mexican conifers and it is just beginning to be studied in the country.  2019a) for A. religiosa, which is due to a reduction in the amount of moisture and an increase in the temperature in the ecological niche of P. greggii.
According to the analysis of the interaction of the relevant variables in the current and future distribution, it was possible to delimit two conservation niche zones suitable for the generation of conservation and restoration activities of P. greggii inside their natural distribution in Mexico. Aguirre and Duivenvoorden (2010), in their study about the potential modeling of 56 Pinus species in Mexico mentioned as a conservation proposal that one of most important zones to establish new protection areas for this genus is the Sierra Madre Sur where several currently modeled species exist, as well as few areas that are under regulatory protection. However, this conservation proposal was focused on Pinus species in a more temperate climate, which differed from the arid to semiarid of P. greggii. Manzanilla et al. (2019b) proposed two zones of conservation and seed production for P. pseudostrobus and P. montezumae. Although they have different environmental requirements than P. greggii, the proposed zones of conservation and seed production are similar, with two delimited zones in both our study and Manzanilla et al. (2019b).
Most of the studies on this subject have been made on Pinaceae of temperate-cold climates, therefore, the ndings of our study are relevant because it allows us to de ne and propose two important areas for the conservation and restoration of P. greggii in Mexico.
Finally, according the objectives of this study, it was possible to delimit and estimate the current natural distribution of P. greggii in Mexico and the most relevant environmental variables were also analyzed. From this data we were able to determine the presence of this important ecological and economic species in Mexico, and from the niche conservation analysis between the current and future distributions, it was possible to propose conservation areas, which do not compromise the development and growth of the species because they would be under similar current-future environmental conditions.

Conclusions
The analysis in this study allowed us to delimit and estimate the natural geographic distribution of P. greggii and the occupied area for the species in the Sierra Madre Oriental and Neovolcanic Transversal Belt in Mexico. The delimitation and estimation of the species distribution under four scenarios of climate change from 2041 to 2060 was also developed; an area of study where few studies with species of arid and semiarid climates exist.
The most important variable that determined the suitable of the current and future ecological niche of P. greggii in Mexico was the mean annual temperature, which is a variable that is expected to increase in the future due to climate change. In this study, a decrease in area of P. greggii is expected according to the four projected scenarios (two with constant concentrations and two others with increased in greenhouse gas emissions). This projected decrease in area is in relation to the current area, however according to the present study, comparatively between modeled scenarios (RCP 4.5 to RCP 8.5), there will be a slight increase, which will favor the species in the future.
The area most affected by climate change would be the states of Hidalgo and Puebla, according to the four projected scenarios for P. greggii during the period of 2041-2060. It should be noted that the change in environmental conditions under the effects of climate change probably will limit the growth and development of tree species, but would not imply a local or regional extinction.
The niche conservation analysis of P. greggii allowed us to identify areas under similar environmental conditions that may become ideal for the subsistence of P. greggii in Mexico in the future. These areas are divided into north (Nuevo Leon) and center (Hidalgo) and could be used for conservation, restoration, and forest propagation purposes in situ.
Declarations Figure 2 Potential current distribution model of P. greggii in the study area. Relevant environmental variables of the current distribution models.

Figure 4
Relevant environmental variables of the CNRM-CM5 RCP 4.5 and 8.5 models.

Figure 5
Relevant environmental variables of the HadGEM2-ES RCP 4.5 and 8.5 models.