Mexico is home to the highest species diversity of pines: 46 species out of 113 reported around the world. Within the great diversity of pines in Mexico, Pinus culminicola Andresen et Beaman, P. jaliscana Perez de la Rosa, P. maximartinenzii Rzed., P. nelsonii Shaw, P. pinceana Gordon, and P. rzedowskii Madrigal et M. Caball. are six catalogued as threatened or endangered due to their restricted distribution and low population density. Therefore, they are of special interest for forest conservation purposes. In this paper, we aim to provide up-to-date information on the spatial distribution of these six pine species according to different historical registers coming from different herbaria distributed around the country by using spatial modeling. Therefore, we recovered historical observations of the natural distribution of each species and modelled suitable areas of distribution according to environmental requirements. Finally, we evaluated the distributions by contrasting changes of vegetation in the period 1991–2016. The results highlight areas of distribution for each pine species in the northeast, west, and central parts of Mexico. The results of this study are intended to be the basis of in situ and ex situ conservation strategies for the endangered Mexican pines.

Various spatial modelling methods and tools have been used in ecology and biogeography. The application of these options serves a dual function: first, they offer information about the potential distribution of species to understand the richness and diversity of unassessed areas. Second, spatial modelling methods employ these predictions to select relevant sites to determine natural conservation areas. In this study, we compared three methods for modelling the spatial distribution of Egg-cone Pine (Pinus oocarpa Schiede), an important non-timber pine in Mexico. The final goal is to estimate suitable areas for the conservation and reproduction of superior individuals (plus trees) of P. oocarpa as a conservation strategy outside the known distribution since this species possesses a high ecological and economic value. The model used were a generalised linear model (GLM) as a parametric regression method, random forest (RF) as a machine-learning method, and the MaxEnt model, a standard procedure, implemented using the Kuenm R package. The results suggest that the models used performed well since the AUROC was between 0.95 and 0.98 in all cases. MaxEnt and random forest approaches provided more conservative predictions for the distribution of suitable areas of plus trees of P. oocarpa than the generalised linear model, but the random forest algorithm achieved the best performance. The results of the study allowed the determination of ex situ conservation areas for P. oocarpa plus trees outside of their known distribution.

The design of a sampling unit, whether a simple plot or a subplot within a clustered structure, including shape and size, has received little attention in inferential forestry research. The use of auxiliary variables from remote sensing impacts the precision of estimators from both model-assisted and model-based inference perspectives. In both cases, model parameters are estimated from a sample of field plots and information from pixels corresponding to these units. In studies assisted by remote sensing, the shape of the plot used to fit regression models (typically circular) often differs from the shape of the population elements for prediction, where the area of interest is divided into equal tessellated parts. This raises interest in understanding the effect of the sampling unit shape on the mean of variables in forest stands of interest. Therefore, the objective of this study was to evaluate the effect of circular and square subplots, concentrically overlapped and arranged in an inverted Y cluster structure, over tree density, basal area, and aboveground biomass in a managed temperate forest in central Mexico. We used a Multivariate Generalised Linear Mixed Model, which considers the Gamma distribution of the variables and accounts for spatial correlation between Secondary Sampling Units nested within the Primary Sampling Unit. The main findings of this study indicate that the type of secondary sampling unit of the same area and centroid, whether circular or square, does not significantly affect the mean tree density (trees), basal area (m²), and aerial biomass.

Estimating aboveground biomass (AGB) is crucial for understanding the carbon cycle in terrestrial ecosystems, particularly within the context of climate change. Therefore, it is essential to research and compare different methods of AGB estimation to achieve acceptable accuracy. This study modelled AGB in temperate forests of central Mexico using active and passive remote sensing data combined with machine learning techniques (Random Forest and XGBoost) and compared the estimations against a traditional method, such as linear regression. The main goal was to evaluate the performance of machine learning techniques against linear regression in AGB estimation and then validate against an independent forest inventory database. The models obtained acceptable performance in all cases, but the machine learning algorithm Random Forest outperformed (R²_cv = 0.54; RMSE_cv = 19.17) the regression method (R²_cv = 0.41; RMSE_cv = 25.76). The variables that made significant contributions, in both Random Forest and XGBoost modelling, were NDVI, kNDVI (Landsat OLI sensor), and the HV polarisation from ALOS-Palsar. For validation, the Machine learning ensemble had a higher Spearman correlation (r = 0.68) than the linear regression (r = 0.50). These findings highlight the potential of integrating machine learning techniques with remote sensing data to improve the reliability of AGB estimation in temperate forests.

The scarcity of meteorological stations and the strong need for climatic information in alpine forests require the use of large-scale climatic algorithms but the lack of in situ information produces high uncertainty on their suitability. In this study, we used linear mixed models to study the topographic effect (elevation and aspect) and time variations (from hourly to monthly) on temperature (T) and relative humidity (RH) with a 5-year instrumental database. Furthermore, we compared climatic information from a geographical algorithm and our in-situ data. Our data covered two mountains (Tláloc-TLA and Jocotitlán-JOC, State of México), four elevation belts (from 3500 m to 3900 m a.s.l.), and two aspects (Northwest and Southwest). We found differences for average temperature (TLA = 7.56 °C ± 0.03 °C and JOC = 6.98 °C ± 0.02 °C), and relative humidity between mountains (TLA = 69.3% ± 0.12% and JOC = 72.5% ± 0.13%,). The most significant variables explaining T were the elevation (Δ= -0.36 °C by 100 m) and aspect, while the aspect was relevant for RH. May was the warmest month (9.50 °C ± 0.10 °C for average temperature) while September the wettest for both mountains (85.1% ± 0.30% and 87.4% ± 0.25 % RH, respectively). Despite the higher correlations between climatic sources (up to r = 0.83), the geographical algorithm overestimates T and underestimates RH. We propose that when climatic information from geographical algorithms is used in alpine forests, calibrations are needed whenever possible with in situ information. La escasez de estaciones meteorológicas y la necesidad de información climática en bosques de alta montaña implica el uso de algoritmos climáticos a gran escala, pero la falta de información in situ produce una alta incertidumbre en su idoneidad. En este estudio, se usaron modelos lineares mixtos para estudiar el efecto de la topografía (elevación y exposición) y tiempo (de nivel hora a mensual) en temperatura (T) y humedad relativa (RH) en una base de datos instrumental de cinco años. Además, se comparó información climática de un algoritmo geográfico y una base de datos in situ. Los datos incluyeron dos montañas (Tláloc-TLA and Jocotitlán-JOC, Estado de México), cuatro niveles altitudinales (de 3500 m a 3900 m s.n.m.) y dos exposiciones (Noroeste y Suroeste). Se encontraron diferencias para temperatura promedio (TLA = 7.56 °C ± 0.03 °C and JOC = 6.98 °C ± 0.02 °C), y humedad relativa entre montañas (TLA = 69.3% ± 0.12% and JOC = 72.5% ± 0.13%). Las variables más importantes que explicaron T fueron la elevación (Δ= -0.36 °C cada 100 m) y exposición, mientras que la exposición para RH. Mayo fue el mes más caluroso (9.50 °C ± 0.10 °C para temperatura promedio) mientras que septiembre, el más húmedo para ambas montañas (85.1% ± 0.30% y 87.4% ± 0.25% RH, respectivamente). A pesar de las altas correlaciones entre fuentes de información climática (hasta r = 0.83), el algoritmo geográfico sobreestima T y subestima RH. Se propone que cuando la información climática proveniente de algoritmos geográficos sea usada en bosques de alta montaña, es necesaria una calibración con información in situ cuando sea posible.