texto en 
Inglés (pdf)
Artículo en XML
Referencias del artículo
Traducción automática
Enviar artículo por email
Citado por SciELO 
Similares en
SciELO 
Permalink
Highlights:
Species distribution modeling (SDM) is used to estimate the impacts of climate change on arboreal species.
Pinus pinceana has been the most studied species (2.8 %) from 1990 to 2022.
The most used model/algorithm to analyze SDM was MaxEnt (72.8 %)
The variables used in SDM were 65; precipitation was the most used variable (4.8 %).
Introduction
The distribution of species, communities and populations depends on environmental, ecological and even social factors (Leal-Nares et al., 2012). Some tree species are distributed in particular environments; for example, Rhizophora mangle L. (red mangrove) inhabits exclusively aquatic environments (Barrantes-Leiva & Cerdas-Salas, 2015); other taxa preferentially inhabit temperate mountainous climates, as is the case of the genera Pinus (Barrantes-Leiva & Cerdas-Salas, 2015; Comisión Nacional para el Conocimiento y Uso de la Biodiversidad [CONABIO], 2022) and Quercus (CONABIO, 2022; Maciel-Mata et al., 2015). On the other hand, some are predominant in arid climates, as the case of most of the species of the Prosopis genus (Maciel-Mata et al., 2015). Thus, the distribution area of each species is restricted due to its environmental tolerance, a product of evolutionary processes that have shaped the organisms and, consequently, have determined their presence in certain areas (Chaves-Barrantes & Gutiérrez-Soto, 2017).
In recent decades, an increasing interest in analyzing of the relationship between the distribution of species, habitat and climate has grown (Borthakur et al., 2018; Li et al., 2016). Thanks to the recent development of geographic information systems (GIS) and applied statistical techniques, models that predict species distributions have been developed (Ramirez-Magil et al., 2020; Singh et al., 2020). Species distribution models (SDM) use information on environmental variables related to the presence of one or a group of species. With mathematical algorithms, the information is systematized for the development of correlative to complex models that analyze the ecological and environmental conditions affecting the distribution of species. Currently, these models are expanding with new methods and strategies for their treatment and interpretation. (Becerra-López et al., 2016; Rong et al., 2019; Zhao et al., 2020, 2020a, 2021).
SDM applications are relevant in research related to conservation biology, biogeography, epidemiology, potential impact of climate change, projections of geographic occurrence of invasive species, and identification of geographic regions that require exploration (López-Sandoval et al., 2015; Saupe et al., 2012). For example, in the state of Baja California, Palma-Ordaz and Delgadillo-Rodríguez (2014) studied the potential distribution of eight invasive alien species and identified sites that require further exploration, for conservation and species management purposes.
From an ecological point of view, species contribute to the generation of environmental goods and services. Some of the most important are climate regulation, water provision (quality and quantity) and oxygen generation, erosion control, as well as soil regeneration, conservation and recovery (Başkent, 2021), which provide economic, social and environmental benefits (Başkent, 2021), which provide economic, social and environmental benefits (System of Environmental Economic Accounting [SEEA], 2022).
There are currently no recent scientific systematic review papers that analyze global trends in tree-based SDM studies, so the objective of this study was to compile and analyze scientific information about the countries that have developed SDM, the tree species studied, types of analysis and algorithms/models used, as well as the type of variables used in the research during the period 1990-2022. Having a document that synthesizes the information, produced over 30 years, contributes to the understanding of the line of research in different fields such as biology, ecology, biological conservation and biogeography.
Materials and Methods
Compilation of information
A detailed systematic review of the global scientific literature on the use of arboreal species SDM published in the period 1990-2022 was completed. Seventy-six scientific databases and publishers were reviewed, corresponding to Dialnet, Crossref, Scilit, PubMed, ISI Web of Science, Springer Link, ScienceDirect, Journal Article, SciELO, INECOL, Polibotánica, Redalyc, Ecology and Evolution, Ecología Austral, Plant Ecology, New Forests, Library Online, Botanical Sciences and Semantic Scholar. A total of 233 scientific journals were consulted; in addition, 10 books and seven theses were reviewed. In the search for information, keywords (in English and Spanish) were used such as: ‘potential distribution’, ‘geographic distribution’, ‘geographic patterns’, ‘modeling’, ‘tree species’, ‘habitat suitability’, ‘climate change’ and ‘ecological niche’.
Organization and analysis of information
There were six categories of analysis of the information found: country of origin of the study, species studied, objective of the study (species, family or genus distribution), type of analysis (correlative models [simple correlation between distribution record and bioclimatic layers] and complex models [use of multiple physiological, climate and edaphological variables, habitat solar radiation, species genetics, disturbance of the study area and habitat suitability variables of the species studied]), type of model/algorithm and climate variables. The information was recorded and categorized in a Microsoft Excel database (Gómez-Tolosa et al., 2021; Ocampo‐González et al., 2020).
Results and Discussion
The overall trend of studies on arboreal species SDM in the last decades increased mostly in the period from 2010 to 2020 (70.4 %, n = 176; N = 250), followed by the periods 2020-2022 (15.2 %, n = 38) and 2000-2010 (13.6 %, n = 34); in contrast, for the years 1990 to 2000, studies related to this subject were very scarce (0.8 %, n = 2).
These types of studies have increased in the last 30 years and have covered not only spatial patterns of tree species occurrence, but also research topics of endemic species and plant suitability, such as Ávila-Sánchez et al. (2018) and Quipuscoa Silvestre et al. (2019). According to the results, the increasing impact of environmental factors (climate change) and anthropogenic damage (land use change, pollution, deforestation and extensive livestock farming) on species distribution has motivated an increase in research worldwide to understand species responses. Studies show significant differences in the effects and factors that determine the distribution of species within families and even within the same genus, influencing the diversity structure (relative abundance of species and ecosystems and degree of connectivity; Villaseñor, 2018).
Origin of the study
From 1990 to 2022, 250 arboreal species SDM studies were conducted in 48 countries. The countries with the highest number of studies were China (21.7 %, n = 54), Mexico (20.5 %, n = 51) and Spain (11.6 %, n = 29), while the countries with the lowest percentage were Senegal (0.40 %, n = 1), Syria (0.40 %, n = 1) and Nepal (0.40 %, n = 1; Figure 1). According to Martínez-Meyer et al. (2014) and Marchese (2015), 11 countries are considered the most diverse in vascular plants, among them, specifically China and Mexico, which, due to their location in the tropical zone and their topography with a great complexity of landscapes and microclimates, favor suitable environments for several tree species (Mokany et al., 2020). This may explain why more SDM studies are conducted in these two countries. Although Spain is not included in the list of nations with the greatest diversity of tree species, in the results it is one of the five countries that have carried out the most studies on this subject. Therefore, authors such as García-Valdés and Morales-Castilla (2016) suggest that in recent years special attention has been presented to SDM studies in that country, where they use the Spanish national forest inventory in order to highlight the geographical distribution capacity and the performance of tree species in the future.
Species studied
In the 250 studies reviewed, 163 tree species corresponding to 36 families and 69 genera were recorded (Table 1). The most studied species belong to the families Pinaceae (60 species and 86 genera) and Fagaceae (40 species and 53 genera). Among the frequently studied species were Pinus pinceana Gordon (2.8 %, n = 7), Pinus durangensis Martinez (2.0 %, n = 5) and Pinus sylvestris L. (2. 0 %, n = 5); while the least studied were Ostrya rehderiana Chun (0.4 %, n = 1), Oroxylum indicum (L.) Kurz (0.4 %, n = 1) and Polylepis racemosa Ruiz & Pav. (0.4 %, n = 1).
The Pinaceae family is the most represented probably because the research describes pines as elements of great ecological, economic and even social importance. Often, pines are the dominant component of the vegetation, strongly influencing functional processes of the forest ecosystem, such as biogeochemical cycles, hydrological cycles, and fire regimes; in addition, they are habitat and food source for a large number of invertebrates and vertebrates (Aceves-Rangel et al., 2018; González-Cubas et al., 2020). Pines provide important environmental services (water, oxygen, recreation, and carbon sequestration) and influence the regional climate (Cruz-Cárdenas et al., 2016; Martínez-Méndez et al., 2016). On the other hand, the Pinaceae family has great economic value, as pines are a source of timber, firewood, pulp, resins, edible seeds and other products of anthropogenic interest. All this results in the growing interest in understanding the distribution patterns of this family, such is the case of P. pinceana (species with the highest percentage of studies), a tree endemic to Mexico, recognized as one of the few forest resources of semi-arid zones due to its wide adaptability and resistance to adverse conditions, which gives it the status of a species with great potential for ecological restoration (Rosas-Chavoya et al., 2016).
Table 1 Tree families, genera and species included in 250 distribution modeling studies published in the period 1990-2022.
| Species studied | Number of publications | Genera studied | Families studied |
|---|---|---|---|
| Abies alba Mill. | 2 | 3 | 3 |
| Abies guatemalensis Rehder | 1 | 1 | 1 |
| Abies pinsapo Boiss. | 2 | 2 | 2 |
| Abies religiosa (Kunth) Schltdl. & Cham. | 2 | 3 | 3 |
| Abies vejarii Martínez | 1 | 3 | 3 |
| Acacia pycnantha Benth | 1 | 1 | 1 |
| Acer davidii Franch. | 2 | 1 | 1 |
| Acer campestre L. | 1 | 1 | 1 |
| Acer monspessulanum L. | 1 | 1 | 2 |
| Acer pseudoplatanus L. | 1 | 1 | 1 |
| Acer platanoides L. | 1 | 1 | 1 |
| Alnus glutinosa (L.) Gaertn. | 1 | 1 | 1 |
| Alnus cremastogyne Burkill | 1 | 1 | 1 |
| Aglaia bourdillonii Gamble | 1 | 1 | 2 |
| Araucaria angustifolia (Bertol.) Kuntze | 1 | 1 | 1 |
| Argania spinosa Roem. & Schult. | 1 | 1 | 1 |
| Arbutus xalapensis Kunth | 1 | 1 | 1 |
| Astronium graveolens Jacq. | 4 | 2 | 2 |
| Astronium graveolens Jacq. | 4 | 2 | 2 |
| Bursera simaruba (L.) Sarg. | 1 | 1 | 1 |
| Canacomyrica montícola Guillaumin | 1 | 1 | 1 |
| Cariniana decandra Ducke | 3 | 2 | 3 |
| Catalpa bungei C. A. Mey. | 1 | 1 | 1 |
| Carpinus tientaiensis W. C. Cheng | 1 | 1 | 1 |
| Castanopsis sieboldii (Makino) Hatus. | 1 | 1 | 1 |
| Cedrelinga cateniformis (Ducke) Ducke | 2 | 1 | 3 |
| Cedrus atlantica (Endl.) Manetti ex Carrière | 1 | 1 | 1 |
| Cedrela odorata L. | 2 | 2 | 2 |
| Cedrela montana Moritz ex Turcz. | 2 | 2 | 2 |
| Cedrela kuelapensis T. D. Penn. & Daza | 1 | 1 | 1 |
| Cedrus libani A. Rich. | 1 | 1 | 1 |
| Cedrela dugesii S. Watson | 1 | 1 | 1 |
| Cinnamomum camphora (L.) J. Presl | 1 | 1 | 1 |
| Cola lorougnonis Aké Assi | 1 | 1 | 1 |
| Cunninghamia lanceolata (Lamb.) Hook. | 3 | 2 | 3 |
| Dalbergia cultrata Benth. | 1 | 1 | 1 |
| Davidia involucrata Baill. | 1 | 1 | 1 |
| Dipteryx oleifera Benth. | 1 | 1 | 1 |
| Dialium guianense (Aubl.) Sandwith | 1 | 1 | 1 |
| Dipterocarpus turbinatus C. F. Gaertn. | 2 | 2 | 2 |
| Eusideroxylon zwageri Teijsm. & Binn. | 1 | 1 | 1 |
| Esenbeckia cornuta Engl. | 1 | 1 | 1 |
| Eremanthus erythropappus Gardner | 1 | 1 | 3 |
| Elaeagnus angustifolia L. | 1 | 1 | 1 |
| Fagus sylvaticaL. | 1 | 2 | 2 |
| Fagus grandifolia Ehrh. | 1 | 1 | 1 |
| Glyptostrobus pensilis (Staunton ex D. Don) K. Koch | 1 | 1 | 1 |
| Handroanthus impetiginosus (Mart. ex DC.) Mattos | 1 | 1 | 1 |
| Hydnocarpus kurzii (King) Warb. | 1 | 1 | 1 |
| Haloxylon ammodendron (C. A. Mey.) Bunge ex Fenzl | 1 | 1 | 1 |
| Hymenaea courbaril L. | 1 | 1 | 1 |
| Hippophae salicifolia D. Don | 1 | 1 | 1 |
| Hura crepitans L. | 1 | 1 | 1 |
| Iberian abies L. | 1 | 1 | 1 |
| Ilex aquifolium L. | 2 | 1 | 2 |
| Ilex pallida Standl. | 1 | 1 | 1 |
| Juniperus rigida Siebold & Zucc. | 1 | 1 | 1 |
| Juglans regia L. | 1 | 1 | 1 |
| Liriodendron chinense (Hemsl.) Sarg. | 1 | 1 | 1 |
| Liquidambar styraciflua L. | 1 | 1 | 1 |
| Malus pumila Mill. | 1 | 1 | 1 |
| Mangifera sylvatica Roxb. | 1 | 1 | 1 |
| Melaleuca quinquenervia (Cav.) S. T. Blake | 1 | 1 | 1 |
| Melaleuca cajuputi Powell | 2 | 2 | 2 |
| Milicia excelsa (Welw.) C. C. Berg | 1 | 1 | 1 |
| Minquartia guianensis Aubl. | 1 | 1 | 1 |
| Moringa peregrina (Forssk.) Fiori | 1 | 1 | 1 |
| Morus alba L. | 1 | 1 | 1 |
| Myristica dactyloides Gaertn. | 2 | 1 | 2 |
| Myrrhidendron donnellsmithi J. M. Coult. & Rose | 1 | 1 | 1 |
| Otoba parvifolia (Markgr.) A. H. Gentry | 2 | 2 | 2 |
| Ostrya rehderiana Chun | 1 | 1 | 1 |
| Oroxylum indicum (L.) Kurz | 1 | 1 | 1 |
| Olea ferrugínea Wall. ex Aitch. | 1 | 1 | 1 |
| Oreomunnea mexicana (Standl.) J.-F.Leroy | 1 | 1 | 1 |
| Pelliciera rhizophorae Planch. & Triana | 1 | 1 | 1 |
| Pinus arizonica Engelm. | 2 | 2 | 5 |
| Pinus cembroides Zucc. | 3 | 2 | 3 |
| Picea crassifolia Kom. | 3 | 2 | 3 |
| Picea chihuahuana Martínez | 2 | 1 | 2 |
| Pinus armandii Franch. | 2 | 1 | 2 |
| Pinus caribaea Morelet | 1 | 1 | 1 |
| Pinus culminicola Andresen & Beaman | 3 | 1 | 1 |
| Pinus chihuahuana Engelm. | 3 | 2 | 4 |
| Pinus durangensis Martínez | 5 | 4 | 8 |
| Pinus hartwegii Lindl. | 4 | 2 | 4 |
| Pinus herrerae Martínez | 2 | 2 | 2 |
| Pinus jaliscana Pérez de la Rosa | 1 | 1 | 1 |
| Pinus koraiensis Siebold & Zucc. | 2 | 2 | 2 |
| Pinus montezumae Lamb. | 1 | 2 | 3 |
| Pinus martinezii E. Larsen | 1 | 2 | 1 |
| Pinus nelsonii Shaw | 3 | 1 | 1 |
| Pinus nigra J. F. Arnold | 3 | 3 | 3 |
| Pinus pineaL. | 3 | 1 | 2 |
| Pinus pinaster Aiton | 1 | 2 | 3 |
| Pinus pinceana Gordon | 7 | 5 | 4 |
| Pinus pseudostrobus Lindl. | 3 | 2 | 2 |
| Pinus sylvestris L. | 5 | 3 | 6 |
| Pinus uncinata Ramond ex DC. | 5 | 2 | 2 |
| Pinus teocote Schied. ex Schltdl. & Cham. | 2 | 4 | 6 |
| Pinus rzedowskii Madrigal & M. Caball. | 2 | 4 | 7 |
| Pinus virginiana Mill. | 2 | 2 | 2 |
| Pinus tropicalis Morelet | 2 | 2 | 2 |
| Piper aduncum L. | 1 | 1 | 1 |
| Polylepis racemosa Ruiz & Pav. | 1 | 1 | 1 |
| Populus tremula L. | 1 | 1 | 1 |
| Prunus armeniaca L. | 1 | 1 | 1 |
| Prosopis cineraria (L.) Druce | 1 | 1 | 1 |
| Prosopis laevigata (Willd.) M. C. Johnst. | 1 | 1 | 1 |
| Prunus africana (Hook. f.) Kalkman | 2 | 1 | 2 |
| Pseudolarix amabilis (J. Nelson) Rehder | 1 | 1 | 1 |
| Pseudotsuga sp. L. | 1 | 1 | 2 |
| Pterocarpus santalinus L. f. | 1 | 1 | 1 |
| Quadrella incana (Kunth) Iltis & Cornejo | 1 | 1 | 2 |
| Quercus costaricensis Liebm. | 2 | 1 | 2 |
| Quercus coccolobifolia Trel. | 1 | 1 | 1 |
| Quercus canbyi Trel. | 1 | 1 | 1 |
| Quercus ithaburensis Decne. | 1 | 1 | 1 |
| Quercus cedrorum Kotschy | 2 | 2 | 2 |
| Quercus cerris L. | 4 | 2 | 4 |
| Quercus cornelius-mulleri Nixon & K. P. Steele | 1 | 1 | 1 |
| Quercus ilex L. | 5 | 2 | 5 |
| Quercus leucotrichophora L. | 1 | 1 | 1 |
| Quercus coccifera L. | 3 | 3 | 3 |
| Quercus laeta Liebm. | 1 | 1 | 2 |
| Quercus humilis Mill. | 1 | 1 | 2 |
| Quercus semecarpifolia Sm. | 1 | 1 | 1 |
| Quercus aegilops Scop. | 2 | 2 | 2 |
| Quercus libani G. Olivier | 1 | 1 | 1 |
| Quercus falcata Michx. | 2 | 2 | 2 |
| Quercus petraea (Matt.) Liebl. | 2 | 2 | 2 |
| Quercus suber L. | 4 | 5 | 6 |
| Rhizophora racemosa G. Mey. | 1 | 1 | 1 |
| Rhizophora mangle L. | 1 | 1 | 1 |
| Rhododendron arboreum Sm. | 1 | 1 | 1 |
| Sabina przewalskii (Kom.) W. C. Cheng & L. K. Fu | 1 | 1 | 1 |
| Sapindus L. | 1 | 1 | 1 |
| Santalum albumen (R. Br.) A. DC. | 1 | 1 | 1 |
| Sassafras tzumu (Hemsl.) Hemsl. | 1 | 1 | 1 |
| Semiliquidambar cathayensis Hung T. Chang | 1 | 1 | 1 |
| Sebastiana longicuspis Standl. | 1 | 1 | 1 |
| Schinus molle L. | 1 | 1 | 1 |
| Shorea guiso Blume | 1 | 1 | 1 |
| Sinadoxa corydalifolia C. Y. Wu, Z. L. Wu & R. F. Huang | 1 | 1 | 1 |
| Sorbus aria (L.) Crantz | 1 | 1 | 1 |
| Styrax sumatrana Pohl | 1 | 1 | 1 |
| Swietenia macrophylla King | 1 | 1 | 1 |
| Tamarindus indica L. | 2 | 1 | 2 |
| Taiwania cryptomerioides Hayata | 1 | 1 | 1 |
| Taxus baccata L. | 2 | 2 | 2 |
| Taxodium mucronatum Ten. | 1 | 1 | 1 |
| Taxus globosa Schltdl. | 1 | 1 | 1 |
| Tecoma rosaefolia Urb. | 1 | 1 | 1 |
| Tectona grandis L. f. | 2 | 1 | 2 |
| Tilia platyphyllos Scop. | 2 | 1 | 2 |
| Tilia cordata Mill. | 1 | 1 | 1 |
| Triadica sebifera (L.) Small | 1 | 1 | 1 |
| Triplochiton scleroxylon K. Schum. | 1 | 1 | 1 |
| Vaccinium consanguineum Klotzsch | 1 | 1 | 1 |
| Ximenia americana L. | 1 | 1 | 1 |
| Zelkova serrata (Thunb.) Makino | 1 | 1 | 1 |
| Zelkova schneideriana Hand. Mazz. | 1 | 1 | 1 |
Objective of the study
The main objective of most of the 250 studies was to study the distribution of tree species (76 %, n = 190) and to analyze SDM of tree families (20.4 %, n = 51), while only 3.6 % (n = 9) aimed to study SDM of tree genera. It is likely that the reason that the objects of study are species and not families is because the former possess very particular anatomical, morphological and physiological characteristics that make them unique. For example, a species can be located in a specific habitat, whose abiotic and biotic conditions interact dynamically to produce the complex entity showing the geographic distribution, since these conditions are favorable only to the species, while families gather groups of species that require the varied presence of biotic and abiotic conditions to explain the entire distribution of the group to be studied. Thus, the study of a single species provides the necessary knowledge to model its spatial and temporal distribution, as it has specific data on its physiology, evolutionary biology, ecology and conservation (Guisan et al., 2013; Vroh et al., 2016).
Type of analysis (correlative or complex)
Of the 250 publications, and according to Table 2, 78.4 % (n = 194) of the SDM studies used correlative analyses of species distribution with environmental variables such as precipitation (50.4 %, n = 126), temperature (49.6 %, n = 124) and mean annual temperature (46 %, n = 115).
Table 2 Frequency and percentage of use of variables included in correlative models of 250 publications of tree species distribution in the period 1990-2022.
| Variables | n | % |
|---|---|---|
| Precipitation | 126 | 50.4 |
| Temperature | 124 | 49.6 |
| Average annual temperature | 115 | 46.0 |
| Precipitation in the wettest quarter | 113 | 45.2 |
| Precipitation in the driest period | 107 | 42.8 |
| Average temperature | 105 | 42.0 |
| Quarterly average temperatures | 103 | 41.2 |
| Precipitation in the wettest period | 101 | 40.4 |
| Precipitation in the driest quarter | 98 | 39.2 |
| Minimum temperature of the coldest month | 97 | 38.8 |
| Seasonality of temperature | 92 | 36.8 |
| Precipitation of the warmest quarter | 89 | 35.6 |
| Precipitation of the coldest quarter | 89 | 35.6 |
| Minimum temperature of the warmest month | 86 | 34.4 |
| Minimum temperature | 86 | 34.4 |
| Maximum temperature | 80 | 32.0 |
| Average temperature of the winter quarter | 79 | 31.6 |
| Monthly precipitation | 78 | 31.2 |
| Average temperature of the rainiest quarter | 75 | 30.0 |
| Average temperature of the coldest month | 74 | 29.6 |
| Annual temperature range | 73 | 29.2 |
| Annual precipitation | 72 | 28.8 |
| Altitude | 70 | 28.0 |
| Precipitation in the dry season of the year | 67 | 26.8 |
| Precipitation in the rainy season of the year | 65 | 26.0 |
| Maximum temperature of the warmest month | 65 | 26.0 |
| Daily temperature range | 60 | 24.0 |
| Average temperature of the driest quarter | 59 | 23.6 |
| Seasonality of precipitation | 58 | 23.2 |
| Average temperature of warmest quarter | 58 | 23.2 |
| Average temperature of the coldest quarter | 54 | 21.6 |
| Elevation | 36 | 14.4 |
| Average diurnal range | 33 | 13.2 |
| Land topography | 19 | 7.6 |
| Average diurnal temperature range | 19 | 7.6 |
| Physiography | 17 | 6.8 |
| Slope | 16 | 6.4 |
| Isothermality | 14 | 5.6 |
| Latitude | 11 | 4.4 |
| Moisture | 11 | 4.4 |
| Composite topography and aspect | 7 | 2.8 |
| Lithology | 6 | 2.4 |
| Thermal oscillation | 5 | 2.0 |
| Global TiltedIrradiation | 5 | 2.0 |
| Annual temperature oscillation | 4 | 1.6 |
| Diffuse horizontal irradiation | 4 | 1.6 |
| Road density | 3 | 1.2 |
| Exposure | 3 | 1.2 |
| Safe frost | 3 | 1.2 |
| Actual annual evapotranspiration | 3 | 1.2 |
| Emberger's annual ombrothermal indices | 3 | 1.2 |
| Aridity index | 3 | 1.2 |
| Direct normal irradiance | 3 | 1.2 |
| topographic wetness index | 3 | 1.2 |
| Geomorphology | 2 | 0.8 |
| Global horizontal irradiation | 2 | 0.8 |
| Rock consolidation | 1 | 0.4 |
| Rock acidity | 1 | 0.4 |
On the other hand, only 21.6 % (n = 54) of the studies included complex analyses, using the effect of different variables on the distribution of tree species. According to Table 3, complex studies used soil variables (45.6 %, n = 114), habitat suitability (20 %, n = 50) and vegetation structure (15.6 %, n = 39).
Table 3 Frequency and percentage of use of variables included in complex models of 250 publications of tree species distribution in the period 1990-2022.
| Variables | n | % |
|---|---|---|
| Edaphology | 114 | 45.6 |
| Habitat suitability | 50 | 20.0 |
| Vegetation structure | 39 | 15.6 |
| Genetics | 12 | 4.8 |
| Solar radiation | 9 | 3.6 |
| Disturbance | 6 | 2.4 |
| Forest land cover | 5 | 2.0 |
Complex analyses stand out because few studies have solid data on the biotic and abiotic factors that a given species needs for an ideal physiological process, and therefore play a very important role in the distribution of tree species; however, most studies include correlative analyses between few bioclimatic variables and the species record (Soilhi et al., 2022; Ye et al., 2022). Most studies only count with data on organism presence, a few also count with absence data and, finally, there are occasions in which abundance data are available. Usually, the data come from non-targeted sampling and observations that consist of making the most of the circumstances to gain the greatest possible benefit (Becerra López et al., 2016). There are rare cases where there are samplings expressly designed to estimate the distribution of an organism (Maciel-Mata et al., 2015). That is why 78.4 % of the studies are simple correlation studies between the distribution record and environmental variables, among these we find: (1) climate variables, usually generated from the interpolation of data from climatological stations, using elevation as a covariate (Soilhi et al., 2022; Ye et al., 2022); (2) lithology and geology information, representing the dependence of vegetation on substrate type; (3) elevation and derived variables, such as slope topography, curvature or roughness, and microclimates; (4) remotely sensed variables, such as vegetation indices, surface temperature, or land cover classifications, which have been little employed although they have significant potential (Chen et al., 2022). However, 21.6 % of the publications have enough information to perform complex analyses (physiological, climatic, edaphological variables, solar radiation, species genetics, area disturbance and habitat suitability variables of the species studied) (Axer et al., 2021; Manzanilla-Quijada et al., 2020; Ramírez-Magil et al., 2020), and even have sufficient data on species abundance to allow their inclusion in the model (Bañuelos-Revilla et al., 2019).
Type of variables
In the 250 studies, the use of 65 variables was recorded, which are listed in Tables 2 and 3. The most frequently used variables were precipitation (50.4 %, n = 126), temperature (49. 6 %, n = 124) and mean annual temperature (46 %, n = 115); while the least used were geomorphology of the study area (0.8 %, n = 2), rock consolidation (0.4 %, n = 1) and rock acidity (0.4 %, n = 1). Other studies have shown that the potential geographic distribution of species can be inferred mainly from precipitation, annual temperature and relative humidity (Soilhi et al., 2022; Zhao et al., 2020a). Bañuelos-Revilla et al. (2019) and Díaz et al. (2012) report that these climate factors have direct and indirect effect on the distribution of forest species (Bañuelos-Revilla et al., 2019) at broad spatial scales (kilometers). For example, Díaz et al. (2012) mention that, in Big Bend National Park, south Texas, USA, the distribution and abundance of forest species are directly related to ambient temperature and soil moisture and indirectly to exposure to solar radiation. In the northeast of Mexico, González-Cubas et al. (2020) found that the annual temperature range and precipitation in the driest month are the most important environmental variables in determining the potential distribution of Abies vejarii Martínez (Pinaceae). Finally, Gutiérrez and Trejo (2014) affirm that changes in precipitation and temperature reduce the distribution range of five temperate forest tree species in Mexico, which would imply a considerable decrease in their populations and even some local extinctions are expected to happen.
Lira-Noriega et al. (2013) analyzed how temperature and precipitation factors explain the distribution area of the Phoradendron californicum Nutt. parasitic plant in the Sonoran and Mojave deserts and recommend integrating of environmental and biological factors in the geographic ranges, to understand the distribution patterns and processes. According to the authors, the study of environmental (temperature and precipitation), topographic and anthropogenic variables is essential in the distribution of species, because they help to understand the implications in the spatial and temporal development of their populations at a local scale, playing a very important role in the reproductive physiology, from flower formation to seed germination.
Type of model/algorithm
A total of 13 techniques for performing arboreal species SDM models/algorithms were recorded in the 250 studies, which are listed in Table 5. The most widely used were the MaxEnt algorithm (72.8 %, n = 182), the Random Forests model (8 %, n = 16) and Genetic Algorithm for Rule-Set Prediction or GARP (4.8 %, n = 12); in contrast, the least used were the English UKMOHADGEM1 model (0.4 %, n = 1), the ILWIS V. 3.3 multicriteria spatial assessment (0.4 %, n = 1) and cluster analysis (0.4 %, n = 1).
The MaxEnt algorithm (Phillips et al., 2006) is the most widely used for multipurpose ecological niche modeling in biogeography, conservation biology, and ecology, probably because this algorithm simulates potential geographic distributions of species using machine learning and the principle of maximum entropy, achieving good results even when data on the distribution of a species are scarce (Radosavljevic & Anderson 2014; Ramírez-Magil et al., 2020; Singh et al., 2020). The results of this model can provide information on how species are restricted by environmental conditions, the impact of environmental changes on their distribution, and the determination of potential areas for species reintroduction (Rong et al., 2019; Zhao et al., 2020, 2020a, 2021). In addition, this algorithm has been widely used for solving problems involving endangered species in a strict sense, invasive species, suitable habitat for rare species, and forest destruction (Deb et al., 2020; Zhu et al., 2020).
Table 5 Frequency and percentage of model/algorithm use in in geographic distribution modeling techniques for tree species in 250 publications in the period 1990-2022.
| Type of model | n | % |
|---|---|---|
| MaxEnt 3.3.3k | 182 | 72.8 |
| Random Forests | 20 | 8.0 |
| Genetic Algorithm for Rule-Set Prediction (GARP) | 12 | 4.8 |
| Ordinary Least Squares (OLS) | 8 | 3.2 |
| General Linear Models (GLM) | 7 | 2.8 |
| Logistic Regression Model | 6 | 2.4 |
| German Model (MPIECHAM5) | 3 | 1.2 |
| General Additive Models (GAM) | 3 | 1.2 |
| Generalized Boosting Models (GBM) | 3 | 1.2 |
| Universal Kringing Technique | 2 | 0.8 |
| English Model (UKMOHADGEM1) | 2 | 0.8 |
| Multicrtiterial Spatial Evaluation ILWIS V.3.3 | 1 | 0.4 |
| Cluster Analysis | 1 | 0.4 |
Conclusions
This review provides a perspective on global trends in species distribution modeling studies for the period 1999-2022, which are usually carried out in countries with a high diversity of vascular species such as China and Mexico, Spain being a particular case where this rule is not followed. The most studied species belong to the Pinaceae and Fagaceae. MaxEnt is the most widely used algorithm in the models, due to its easy application and speed to obtain results, in which precipitation and environmental temperature are the most used variables. Despite its limitations, the studies demonstrate the usefulness of the tree species distribution modeling, therefore, it should be considered a necessary tool for forecasting the potential impacts of climate change on species distribution. It should be emphasized that the performance of the modeling will depend on the technique used, as well as its interpretation, which should be made according to the species and its habitat.
