Modeling the current and future habitat suitability of Neltuma pallida in the dry forest of northern Peru under climate change scenarios to 2100

Abstract The development of anthropic activities and climate change effects impact worldwide species' ecosystems and habitats. Habitats' adequate prediction can be an important tool to assess current and future trends. In addition, it allows strategies development for their conservation. The Neltuma pallida of the forest region in northern Peru, although very significant, has experienced a decline in recent years. The objective of this research is to evaluate the current and future distribution and conservation status of N. pallida in the Peruvian dry forest under climate change (Location: Republic of Peru). A total of 132 forest presence records and 10 variables (bioclimatic, topographic, and soil) were processed and selected to obtain the current and future distribution for 2100, using Google Earth Engine (GEE), RStudio, and MaxEnt. The area under the curve values fell within the range of 0.93–0.95, demonstrating a strong predictive capability for both present and future potential habitats. The findings indicated that the likely range of habitats for N. pallida was shaped by factors such as the average temperature of wettest quarter, maximum temperature of warmest month, elevation, rainfall, and precipitation of driest month. The main suitable areas were in the central regions of the geographical departments of Tumbes, Piura, and Lambayeque, as well as in the northern part of La Libertad. It is critical to determine the habitat suitability of plant species for conservation managers since this information stimulates the development of policies that favor sustainable use programs. In addition, these results can contribute significantly to identify new areas for designing strategies for populations conserving and recovering with an ecological restoration approach.

Climate change can have a significant impact by altering the composition of terrestrial ecosystem communities and the performance of species (Bertrand et al., 2011;Forrest, 2016).It often affects the geographic distribution area of endangered species and reduces the size of their native habitats, ultimately resulting in a decline in population or even extinction (Anderegg et al., 2019;Khanal et al., 2022).However, this will depend on the habitat of each species (Bertin, 2008;Meir et al., 2015).The El Niño-Southern Oscillation (ENSO) alters global precipitation patterns, increasing temperatures in arid areas with less frequent rainfall under normal conditions, and temperatures may be more moderate.El Niño-Southern Oscillation indices, Pacific indices, Walker circulation, and the Humboldt Current are factors that control air temperature and climate on the northern coast of Peru (Rollenbeck et al., 2015).
ENSO on the northern coast of Peru has benefited desert vegetation by improving primary productivity levels (Vining et al., 2022) and reducing rural community poverty by 5% in this ecosystem (Pécastaing & Chávez, 2020).
Species distribution modeling (SDM) is an important tool for research because they provide spatial information on the current and future environmental suitability of species (Elith & Leathwick, 2009;Mammola et al., 2021;Santini et al., 2021;Wang et al., 2020).Species distribution modeling defines the species-environment relationship to estimate the geographical distribution of species under different climatic scenarios (Gobeyn et al., 2019).A variety of SDM has been developed (Climex, Genetic Algorithm for Rule-Set Production [GARP], BIOCLIM, and MaxEnt); however, comparative studies suggest that the MaxEnt model is the most suitable (Elith et al., 2011;Liu et al., 2021;Phillips & Dudík, 2008).In dry forest ecosystems, the potential distributions of current and future habitats of species such as Cavanillesia platanifolia, Cordia, Erythrina velutina, Handroanthus chrysanthus, Terminalia valverdeae, Euterpe edulis Mart, Prosopis juliflora, and Prosopis pallida (now N. juliflora and N. pallida) have been studied.Some of these habitats decrease under climate change, while others increase (Aguirre et al., 2017;Leal et al., 2022;Oliveira et al., 2018).
Neltuma pallida forms extensive forests in northern Peru (Zorogastúa et al., 2011).It provides multiple ecosystem benefits as an environment preserver, soil protector and fertilizer, and as a food source for goat farming (Cruzado-Jacinto et al., 2019).In addition, it is used in construction and firewood for populations settled in dry forest ecosystems (Cuentas & Salazar, 2017).However, this species has been reduced by deforestation, urbanization, and agricultural expansión (Depenthal & Meitzner, 2017).
Studies of N. pallida have focused mainly on evaluating its physiological characteristics and its ecological and economic valuation (Aguirre & Kvist, 2005;Cuentas Romero, 2015;Espinosa et al., 2012).
Others are related to the insect's identification associated with the species and its biodiversity (Cruzado-Jacinto et al., 2019;SERFOR et al., 2022), other researchers evaluated foliar functional traits and adaptability to extreme climatic events (Salazar et al., 2021;Salazar, Navarro-Cerrillo, Ancajima, et al., 2018;Salazar, Navarro-Cerrillo, Cruz, & Villar, 2018).There have been reports of studies with limited data regarding the present and future ecological suitability distribution of N. pallida.Hence, studying the behavior of its habitat with of La Libertad.It is critical to determine the habitat suitability of plant species for conservation managers since this information stimulates the development of policies that favor sustainable use programs.In addition, these results can contribute significantly to identify new areas for designing strategies for populations conserving and recovering with an ecological restoration approach.Information on the distribution of N. pallida, spatial arrangement, and climatically suitable habitats under current and future conditions can be assessed by using tools such as QGIS, RStudio, and MaxEnt (Bushi et al., 2022;Kalboussi & Achour, 2018;Shi et al., 2023).Hence, in this study, we aimed to model the present and future potential habitat distribution of N. pallida within the dry forests of Peru, elucidating the relationship between the species and its habitat through response curves derived from key environmental variables.The primary research objectives included:  Addressing these inquiries can not only provide theoretical support for strategic planning related to the introduction and cultivation of N. pallida in Peru but also establish a scientific foundation for its restoration and conservation efforts.

| Study area
The study area consisted of the Equatorial Dry Forest, located in the geographical departments of La Libertad, Lambayeque, Piura, and Tumbes in northern Peru (Figure 1) with an altitudinal range that varies from 0 to 1500 masl (Salazar et al., 2021).The mean annual precipitation ranges from 100 to 500 mm with the months of highest rainfall from January to March, in turn, the mean annual temperature varies from 24 to 27°C and is directly correlated with the intensity of rainfall and the thermal inertia of the Pacific Ocean (Rollenbeck et al., 2015).The study area presents a flat to semi-flat topography with soils originating from eolian or alluvial deposition (Salazar, Navarro-Cerrillo, Cruz, & Villar, 2018).

| Source of data for the presence of N. pallida
The 132 data points on the presence of N. pallida, duly georeferenced, were retrieved (longitude, latitude, and altitude) in CSV format, from the fieldwork that was carried out from June 15 to June 22, 2002 and downloaded from the GBIF only for the study area (https:// www.gbif.org/ , accessed September 24, 2022).Those lacking exact geographic coordinates were corrected through visual interpretation in Google Earth Pro (Zhang et al., 2021).Likewise, spatial analysis was applied in QGIS v. 3.16 to decrease the impact of spatial autocorrelation, ensuring that each pixel contains a single point of presence (Liu et al., 2021;Yang et al., 2022).

F I G U R E 2
Flowchart outlining the methodology to evaluate the spatial modeling of the present and future distribution of Neltuma pallida.

| Environmental variables
Environmental factors such as soil, climate, and topography play a key role in the development and distribution of flora (Yang et al., 2022).To address these considerations, a total of 36 variables were chosen (Table 1).Twenty-three of these variables constituted bioclimatic data for the current period spanning 1970-2000 (Fick & Hijmans, 2017), while terrain-related factors, specifically altitude, slope, flow direction, terrain roughness index, and topographical position index were obtained from digital elevation model (Hennig et al., 2007), downloaded from WorldClim 2.1 data set (https:// www.world clim.org/ ) and procesing in Google Earth Engine (GEE) (Gorelick et al., 2017).The remaining eigth variables, relating to soil properties, were obtained at 30 arcsecond (~1 km) resolution from the SoilGrids 0.5.3 database (http:// soilg rids.org) using GEE.
The bioclimatic, topographic, and edaphic variables were processed at 250 m resolution.To determine the importance of each variable, the Jackknife method was applied (Meza et al., 2022).

| Future scenarios of climate change
We use future bioclimatic data from the sixth version of the MIROC 6 (Tatebe et al., 2019).

| Model execution
The biogeographic distribution model for N. pallida was constructed using MaxEnt.This model assessed the likelihood of potential distribution for each species based on the presence data (locations) (OSINFOR, 2013(OSINFOR, , 2016)).To validate this model, 75% of randomly selected presence data points were used for training, while the remaining 25% were set aside for validation, as outlined by Liu et al. (2021).

| Suitable habitat classification
To improve the performance of the model, five replications of cross-validation were done (Liu et al., 2021).Subsequently, according to the results of habitat levels for each scenario, the maps were reclassified into four habitats suitable categories: highly (0.5 ≤ p ≤ 1.0), moderately (0.3 ≤ p < .5),little (0.1 ≤ p < .3),and unsuitable (p < .1).

| RE SULTS
Table 2 shows the results of the statistical method of the ROC curve that allowed to compare the average sensitivity with the specificity of N. pallida in current and future conditions.For the current and a standard deviation is 0.014 (Figure 3), while, for future scenarios, the AUC ranged between 0.93 and 0.95 (Table 2).

| Precision model and current distribution
The Jackknife test results indicated that the primary factors influencing the potential habitat distribution of N. pallida were the mean temperature during the wettest quarter (bio08), the maximum temperature in the warmest month (bio05), altitude (dem), precipitation, and precipitation during the driest month (bio14), as illustrated in Figure 4.
The modeling reported that the relative contributions of the two variables, the bio14 and the bio08, were the most influenced on the distribution of N. pallida, which explained 46.8% and 27% of the species habitat distribution, respectively.On the other hand, the variables of least contribution were the bio02, the bio18, and the flow direction with 0.7%, 0.7%, and 0.5%, respectively (Figure 5).

| Potential distribution of N. pallida under future climate scenarios
The unsuitable areas in the Equatorial Dry Forest of Peru for N. pallida are likely to be reduced in the coming years to become areas of "low," "medium," and "high" potential habitats.Table 3 shows that the "low" potential habitat in the future scenarios will increase by 12.78, 11.34, 10.16, and 11.65% by 2030 (SSP5-8.5),2050 (SSP5-8.5),2070 (SSP5-8.5),and 2090 (SSP2-4.5),respectively.
The potential current distribution of N. pallida overlapped with the future distribution under climate change scenarios (Figure 7), an habitat increase was observed by 2100 at all potentiality levels.The results showed that habitat characteristics are relatively consistent across the four SSPs, in the different forecast years.The high future potential is in the central part of the geographical departaments of Tumbes, Piura, and Lambayeque, distributed from north to south.
The potential distribution of the N. pallida current habitat overlapped with climate change scenarios to obtain characteristics of habitat loss, gain, or permanence spatially (Figure 8).The habitat areas of N. pallida showed a trend of area gain by 2100.The areas of surface gain were located mainly in the southwest, center-west, and north of the study area.On the other hand, the loss zones were located mainly in the center and center-south of the study area.In addition, the most stable zones under climate change scenarios are located in the central part of the dry forest.

| DISCUSS ION
The performance results of MaxEnt indicated high precision and reliability, as the AUC values ranged between 0.93 and 0.95.If the value exceeds this threshold, it represents exceptionally high predictive accuracy of the model (Swets, 1988).The current zones with "high," "moderate," and "low" potential are distributed in the central areas   current and potential distribution of N. juliflora was mapped, reporting an increase in surface area (Wakie et al., 2014).The use of satellite images such as Landsat 8 and Sentinel-2 has also contributed to mapping N. juliflora and demonstraated its long-term surface increase under climate change scenarios (Ahmed et al., 2021;Rembold et al., 2015).However, other species of dry forests could be affected in future scenarios, such as Anadenanthera colubrina, Aspidosperma pyrifolium, and Myracrodruon urundeuva (Rodrigues et al., 2015), as well as cacti (Cavalcante & Sampaio, 2022), Calycophyllum multiflorum (Alabar et al., 2022), Albizia multiflora, Ceiba trichistandra, and Cochlospermum vitifolium (Manchego et al., 2017).
The increase in the habitat distribution of N. pallida in the study area could be related to its adaptability to extremely dry and wet conditions.This climatic event creates two alternative states that promote survival and growth strategies, respectively (Holmgren et al., 2001;Salazar, Navarro-Cerrillo, Cruz, & Villar, 2018).Unlike other species of dry forests, these changes have allowed the species to develop physiological and morphological adaptations to survive in years of drought, grow after floods, and in saline soils.This adaptability appears to be strongly associated with the occurrence of El Niño on the northern coast of Peru (Palacios et al., 2012;Salazar et al., 2021).However, N. pallida is considered an invasive species in some areas such as Australia, Africa, Ethiopia, Brazil, and Pacific Islands (Ahmed et al., 2021;Gallaher & Merlin, 2010;Rembold et al., 2015;Sintayehu et al., 2020).The decline in N. pallida conditions may be related to pest and disease attacks, forest fires, and indiscriminate logging.The conservation of these species is vital because they provide ecosystem services that contribute to people's economy.Therefore, it is essential to implement measures that help mitigate the species' population decline and ensure its sustainable use (INIA, 2020).
This research determined the current and future habitat distribution areas of N. pallida.Based on this, potential species conservation areas were identified, and conservation plans were established (Li et al., 2020).Additionally, future studies could utilize other climatic models, such as the fusion of methodologies like Analytic Hierarchy Process (AHP), GEE, and GARP (Cotrina, Bandopadhyay, et al., 2021;Padalia et al., 2014;Rojas-Briceño et al., 2022;Zhang et al., 2021).
In this research, we integrated tools like QGIS, Rstudio, GEE, and MaxEnt to identify potential distribution zones of N. pallida under current and future conditions, which yielded favorable results.
Although MaxEnt is one of the most used models, it has certain disadvantages in considering climatic impact, topographical, and environmental factors (nonbiological factors), in areas heavily impacted by human activities (Wang et al., 2020).Therefore, this model only effect of climate change under different scenarios is necessary.
(1) assessing the current and future potential distribution of N. pallida in Peru under different climate scenarios, (2) investigating the influence of environmental factors on the spatial distribution of N. pallida habitat, and (3) understanding how climate change will impact the future distribution and spatial patterns of N. pallida.

F
The study area for Neltuma pallida, along with the boundaries of the different geographical departments and the georeferenced records.

Figure 2
Figure 2 describes the methodological flowchart to analyze the current and future spatial distribution of the N. pallida, based on the collection, cutting, and standardization of bioclimatic, topographic, and edaphic variables according to the study area.Likewise, the information on georeferenced presence data of N. pallida was collected.Finally, the MaxEnt algorithm was applied to model the current and future potential distribution for 2100 using the Model for Interdisciplinary Research on Climate (MIROC 6) and the different Shared Socioeconomic Pathways (SSP).

TA B L E 2
Assessment of the species distribution model's performance (measured by AUC) in both the current environmental conditions and various climate change scenarios.F I G U R E 3 Reliability test of the distribution model based on area under the curve (AUC) and mean sensitivity versus specificity for Neltuma pallida.

F
Relative contributions of the variables to the MaxEnt model to assess the potential habitat distribution of Neltuma pallida.F I G U R E 5 Analysis of variable contributions to the MaxEnt model to assess the potential habitat distribution of Neltuma pallida. of the five repetitions reported an AUC of 0.94, of the departments of Tumbes, Piura, and Lambayeque, as well as in the northern part of La Libertad, central to the study area.This study reports an increase in the surface areas from current to future conditions of N. pallida.The results are similar to those reported by Oliveira et al. (2018), who analyzed the dynamics of climatic niches of Prosopis juliflora (now N. juliflora) and Prosopis pallida (now N. pallida) in semiarid areas of Brazil.A similar trend is observed in Ethiopia, where the TA B L E 3 Current distribution range of Neltuma pallida and the percentage of change in various scenarios in northern Peru.
Distribution of potential habitat loss, gain, or permanence of Neltuma pallida habitat under climate change scenarios.
Within the arid forests of Peru, there is a variety of species, highlighting the need to adopt effective strategies aimed at their monitoring and preservation, as emphasized byCotrina, Bandopadhyay,   et al. (2021).N. pallida is one of the main species in this ecosystem; however, it is threatened by anthropogenic activities such as deforestation, land use change, and logging(Nieuwstadt & Sheil, 2005;Qarallah et al., 2021).Efforts are currently underway to conserve the biodiversity of the ecosystem and N. pallida individuals through the creation of protected areas (725.69 km 2 , Figure9) (SERNAP, 2023).

F
Distribution of the conserved area of Neltuma pallida under different climate change scenarios, according to habitat levels (a) 2030s, (b) 2050s, (c) 2070s, and (d) 2090s.