Prediction scenarios of past, present, and future environmental suitability for the Mediterranean species Arbutus unedo L.
Current and past scenariosThe Arbutus unedo distribution presence database and eight environmental variables were used to predict the species potential distribution under past, current, and future climatic conditions. The current fitted model predicted the A. unedo suitable habitat with good performance, when evaluated using AUC (AUC = 0.92)31.The results suggest that the species distribution was mainly determined by the following attributes: (1) mean temperature of the driest quarter (BIO9), (2) annual mean temperature (BIO1), (3) slope and (4) temperature annual range (BIO7), summing up to 78.6% of the model’s total contributions (Table 1). In a recent study for the same species in Portugal, the authors considered that the variables precipitation seasonality (BIO15) and slope had a net significant influence on the species’ habitat suitability16. The variable slope was expected to influence the A. unedo distribution, since this tree is ecologically adapted to rocky slopes, which are edaphically unsuitable for woodland, where the vegetation is limited to shrub communities. The Arbutus unedo is particularly intolerant to shade, and this high light requirement is particularly needed for fruit production, restricting this species to settlement with open habitats21. This aspect was confirmed in a study about a woodland disturbance, where the A. unedo widespread initially, and, afterward, the species declined as canopy woodland re-developed35.The Arbutus unedo was adapted to a wide range of climatic conditions considering the response curves analysis (Fig. 1). According to Santiso24, this species developed a conservative strategy in the use of nutrients and water, when they are scarce. The Arbutus unedo plasticity and evolvability explained the current species persistence throughout its distribution range, which will be decisive in future response to climate change. The species distribution prediction for the present is in agreement with the species distribution according to Caudullo et al.38 (Fig. 6). Indeed, the A. unedo is widespread in Portugal, Galicia, and southwest of Spain, occupying the coastal belt from Tunisia to Morocco along the North of Africa, and from the Iberian Peninsula to Turkey across southern Europe, yet occurring in the northern Iberian Peninsula, western France, and south-western Ireland. The results showed that MaxEnt fitted model suitability prediction was high in those regions, which was further supported by the high observed AUC. Furthermore, it is important to stress the current potential distribution meaningful accuracy, and, therefore, the ability to extrapolate past and future prediction scenarios for A. unedo distribution in the Mediterranean Basin. Not surprisingly, the highest predicted suitability for the species occurs mostly in its native range, in regions where the mean January temperature is above 4ºC, a limit required for the species’ survival23. Unfortunately, this variable is only available in the WorldClim database for the present, but not for the future or past periods. Nevertheless, the species’ current spatial distribution was overlapped to the areas with mean January temperature higher than 4ºC, and a high overlap is reckoned, thus confirming this factor ecological importance17,23. The analysis of the species’ predicted suitability show that high-suitability areas [0.8–1.0] are more frequent in current conditions, compared to other scenarios, and thus confirming the suitability of climate current conditions for the species’ development (Fig. 2d).Figure 6Arbutus unedo L. distribution area and occurrences. Black dots represent the species’ occurrences (Supplementary Table S1), and the green area represents the species native distribution range for A. unedo downloaded from the data sets in38. Equal-area projection EPSG:3035. The map was made using the sf 1.0–3 and the terra 1.4–14 (https://rspatial.org/terra/) packages in R 4.1.032,33.Full size imageThe predicted habitat suitability for the LGM and MH scenarios differed, as expected, from the obtained for the current climatic conditions. The climatic conditions in the MH, were more suitable than during the LGM, due to climate warming, thus allowing the species expansion. According to the LGM projections, the species occurred in the Mediterranean islands (Sicily, Sardinia, and Balearic), North of Italy, south and eastern coast of Spain, Catalonia, North Africa, and spots in Portugal, Greece, and Turkey (Fig. 2a). These were suitable putative cryptic refugia areas that remained during the LGM. Keppel et al.39 defined refugia as locations to which species retreated during periods of adverse climate, and could potentially expand from, when environmental conditions turn out to be more suitable for the species. These locations/habitats were responsible for the species’ survival under changing environmental conditions for millennia. Identifying and characterizing climate refugia provides an important context for understanding the modern species distribution development, traits, and local adaptation40. The Iberian, Italian and Balkan peninsulas, which remained relatively ice-free during the ice ages were identified as the main glacial tree refugia areas in Europe41,42.The Betic mountains, in the Iberian Peninsula (IP), were sought to be a species’ refuge from the last ice age, nearby the sea (Valencia region, Spain), where fossil pollen pieces of evidence were found in the Canal de Navarrés peat deposit4, with 34 ka, and in the Siles lake (ca. 19 ka) located in the Segura mountains of southern Spain43 (Fig. 5, numbers 1 and 4, respectively). These mountains were ice-free during the Late Pleistocene, and the persistence of a mild climate could explain that other thermophilic species have prevailed in this particular region during the Full Glacial, proved by high genetic diversity levels44, and with the pollen spectra suggesting the region as a glacial refugium for temperate and Mediterranean trees9,13,43.The Iberian Peninsula border could have been a refuge for the species, due to the importance of the sea as a temperature regulator. Indeed, pollen evidence was found in the Basque mountains (northern IP, Fig. 5, number 3: 21 ka)45. Despite the information given by the MaxEnt modelling, the A. unedo reconstruction for the LGM was observed in other areas, since paleo-evidences were found in unexpected northern sites (Donatella Magri, personal communication). Certainly, fossil charcoal pieces of evidence were found in the central region of Portugal (Serra do Sicó), nearby the sea, supporting the presence of thermophilous taxa, including A. unedo, during the Full Glacial (24 ka, Fig. 5, number 2), and the presence of other Mediterranean taxa, such as olive tree and Pistacia lentiscus L., suggested that this may have been refuge zones for thermophilous plants46. Corroboration of the species expansion later during the warm and humid period of the Holocene (see Fig. 3b; green area) was confirmed by pollen remains (Fig. 5), particularly inland and northwardly, in the IP, and, also, in a molecular population genetics study made in Portugal, with northward haplotype migration47.Santiso et al.17 in a study about the A. unedo phylogeography, concluded that this species had two clades, separated during the late Pleistocene, before the LGM, suggesting that it may have coincided with the hardest glaciations recorded in the Quaternary. One clade occupied the Atlantic Iberia and, possibly, North Africa, while the other occurred in the western Mediterranean Basin in Spain. Besides, the A. unedo possibly persisted in the late Quaternary in the western Mediterranean, based on chloroplast DNA observations, and the results from the current study supported this interpretation, and by the fossil evidence (Supplementary Table 3). The same strong genetic differentiation between the western and eastern Mediterranean Basin was also found in olive lineages, and this pattern is congruent in other Mediterranean shrubs and tree species9,13,15. In another species (maritime pine), Bucci et al.44 confirmed the existence of a genetic divide between eastern and western lineages, also previously described by Burban and Petit48 based on mitochondrial DNA. Additionally, maximum haplotypic diversity for this species was found in south-eastern and central Spain, which, therefore, may be considered a biodiversity hotspot and a strong signal for refuge, as long-term populations tend to harbour more diversity than recently expanding ones41. In their study with A. unedo in Portugal using cpSSR, Ribeiro et al.47 found signals of two putative refugia in southern and central littoral in the country, also supported by macrofossil and pollen remains. Furthermore, according to Médail and Diadema41, several regions in the Western Mediterranean (large Mediterranean islands, North Africa and Catalonia), could have played a role in the case of A. unedo, and sought as refugia locations.The Arbutus unedo suitability maps obtained for past conditions support the claim that MH climatic conditions were more suitable for the species expansion than the LGM ones (Fig. 2a,b, and d). Consequently, spatial distribution changes analysis between past and current scenarios (Fig. 3) showed that A. unedo potential distribution expanded extensively from the LGM-current and the MH-current periods, indicating that more suitable areas were available for the species at present. Pollen records during the Early-Middle Holocene supported the Mediterranean presence of this species, in particular, nearby the sea (Fig. 5), which is under the MaxEnt predicted eastward dispersion (Fig. 3a, b). In another species bird-dispersed, the Myrtus communis L. (myrtle), a spread from west to east was verified, following genetic differentiation during the Pleistocene15, since the western region of the Mediterranean Basin had a milder climate, compared to the eastern one, during the LGM.The Arbutus unedo had, probably, a considerable ability to disperse, migrating over thousands of kilometres and even crossing sea stretches, allowing the species expansion during the MH, when the climatic conditions became more favourable17. In the future, A. unedo migration possibility will depend on climatic change pace and, also, on seed dispersal, particularly long-distance dispersal events49, since long-distance migration fitness will be useful in future change scenarios, allowing species to progress to newly suitable areas, as happened in the past. Nevertheless, from past evidence in Ireland, Britain, and across Europe, trees migrated faster than would be expected, during the warming period at the beginning of the Holocene, and that biotic and/or abiotic vectors must have been involved in dispersal50. This is the case with A. unedo, the seeds are dispersed by different type of birds that eat the soft fruit and, some of them are migratory species, as the European robin (Erithacus rubecula L.)51.Future scenariosThe impact of climate change on the A. unedo potential distribution was assessed under two representative concentration pathways (RCP 4.5 and RCP8.5) for the years 2050 and 2070. The results showed that climate change, under both moderate and high emission scenarios, will affect the species distribution range. A decrease in the predicted suitable areas was, generally, observed, since the climate becomes less favourable for the species in the future.According to the results, under the RCP 4.5 scenario, the potential distribution area will increase up to 2050, and decrease afterward according to the prediction for 2070 (Fig. 2d, e, f). Considering the RCP 8.5 scenario, the suitable area will exhibit a net decrease from the present up to 2050 and continue this trend until 2070. It is also expected that he medium–high suitable areas will gradually decrease from the present to the future, especially for scenario RCP 8.5. These results are in agreement with those obtained by other authors, confirming that Mediterranean species (e.g. Quercus sp.) distribution areas will be negatively affected by future climate change52,53.Several studies concluded that global warming will influence species distributions by causing expansions, contractions, or shifts in the species ranges2. As expected, the results from the current study showed that suitable areas will contract under future climate scenarios when compared to the current conditions, though suitable areas will emerge. These effects’ impact will depend on the climate change scenario severity. Despite contraction and expansion effects, the presence of the species will gradually decrease from the 2050s to the 2070s. Moreover, a species’ shift toward the North will be verified, because of suitable areas emergence, observed mainly in the RCP 8.5 scenario. Those areas will mostly emerge in the North of France, South of the United Kingdom, and Ireland, implying species’ latitudinal migration. The species’ northward displacement is consistent with climate change studies’ results obtained by several authors, including A. unedo16,20,54,55. Nevertheless, according to Gerassis et al.20, under climate change, the expected habitat disruption and fragmentation could lead to very adverse conditions for A. unedo survival in the future, which could undoubtedly conduct to a possible species’ presence decline in most of the current distribution area. Moreover, distribution models that predict climate‐induced range shifts do not account for spatial dispersal variation56, but adaptive dispersal evolution always reduced neutral genetic diversity across the species’ range. This means that the species’ genetic pool might be erased, depending on the climate change velocity, amongst other conditions, like landscape fragmentation and competition with other species/crops16,57. Additionally, the species ability to migrate mainly through seeds, dispersed by migratory frugivorous birds, bird abundance, and the velocity of climate change, are key issues for future species survival23,58.These results suggested that future changes in environmental conditions may lead to suitable habitat loss in areas where the species had persisted and with a possible range shift towards the North. These findings also revealed that with continuous future climate warmth, the current potential distribution A. unedo areas will become unsuitable or contract, leading to significant changes in the species’ current distribution pattern and putative presence loss. The possibility of species’ migration will ultimately depend on its capability to keep pace with the changing conditions and the velocity to adapt to environmental changes, such as those presented by habitat and climate modifications56.The Mediterranean Basin is one of the most vulnerable climate change hotspots in the world, thus understanding how future climate changes will disturb Mediterranean plant species distribution will be key for tree management planning and conservation design. Nevertheless, further investigation is needed for species well adapted in this region to assess the impacts of climate change in their current and future potential distributions. Those studies including past climate impact on species distribution should be complemented with phylogeographic methods and paleoclimate reconstructions to locate refuges. Other species distribution models, besides MaxEnt, could be tested, although the lack of absence records data limits considerably the modelling approach. Additionally, the human influence magnitude on the predicted ecological niche should be further studied in detail, including urbanization, industrialization and putative tourism pressure, particularly in coastal areas. More
