Aim: Species distribution models have been used frequently to assess the effects of climate change on mountain biodiversity. However, the value and accuracy of these assessments have been hampered by ...the use of low-resolution data for species distributions and climatic conditions. Herein we assess potential changes in the distribution and community composition of tree species in two mountainous regions of Spain under specific scenarios of climate change using data with a high spatial resolution. We also describe potential changes in species distributions and tree communities along the entire elevational gradient. Location: Two mountain ranges in southern Europe: the Central Mountain Range (central west of the Iberian Peninsula), and the Iberian Mountain Range (central east). Methods: We modelled current and future distributions of 15 tree species (Eurosiberian, sub-Mediterranean and Mediterranean species) as functions of climate, lithology and availability of soil water using generalized linear models (logistic regression) and machine learning models (gradient boosting). Using multivariate ordination of a matrix of presence/absence of tree species obtained under two Intergovernmental Panel on Climate Change (IPCC) scenarios (A2 and B2) for two different periods in the future (2041-70 and 2071-2100), we assessed the predicted changes in the composition of tree communities. Results: The models predicted an upward migration of communities of Mediterranean trees to higher elevations and an associated decline in communities of temperate or cold-adapted trees during the 21st century. It was predicted that 80-99% of the area that shows a climate suitable for coldwet-optimum Eurosiberian coniferous and broad-leaved species will be lost. The largest overall changes were predicted for Mediterranean species found currently at low elevations, such as Pinus halepensis, Pinus pinaster, Quercus ilex ssp. ballota and Juniperus oxycedrus, with sharp increases in their range of 350%. Main conclusions: It is likely that areas with climatic conditions suitable for cold-adapted species will decrease significantly under climate warming. Large changes in species ranges and forest communities might occur, not only at high elevations within Mediterranean mountains but also along the entire elevational gradient throughout this region, particularly at low and mid-elevations. Mediterranean mountains might lose their key role as refugia for cold-adapted species and thus an important part of their genetic heritage.
Question: Will the predicted climate changes affect species distribution in the Iberian Peninsula? Location: Iberian Peninsula (Spain and Portugal). Methods: We modelled current and future tree ...distributions as a function of climate, using a computational framework that made use of one machine learning technique, the random forest (RF) algorithm. This algorithm provided good predictions of the current distribution of each species, as shown by the area under the corresponding receiver operating characteristics (ROC) curves. Species turnover, richness and the change in distributions over time to 2080 under four Intergovernmental panel on climate change (IPCC) scenarios were calculated using the species map outputs. Results and Conclusions: The results show a notable reduction in the potential distribution of the studied species under all the IPCC scenarios, particularly so for mountain conifer species such as Pinus sylvestris, P. uncinata and Abies alba. Temperate species, especially Fagus sylvatica and Quercus petraea, were also predicted to suffer a reduction in their range; also submediterranean species, especially Q. pyrenaica, were predicted to undergo notable decline. In contrast, typically Mediterranean species appeared to be generally more capable of migration, and are therefore likely to be less affected. Nomenclature: Amaral Franco (1990), Govaerts & Avishai (2002) for Quercus pubescens.
Mixed forests play an important role in promoting forest functions and services, and showed better resilience to perturbations than monospecific forests. In the Mediterranean Basin they constitute an ...important share of the total forest area. However, although mixed forests have been described, classified and mapped in most of the Mediterranean regions around the world, in the Mediterranean Basin they remain neglected, with direct negative consequences for forest management and conservation strategies. Our objective is to present a reliable, uniform classification along with a map of the Spanish Mediterranean Mixed Forests (MMF) as a tool for their management and conservation in Spain. The digital Spanish Forest Map was analyzed to identify MMF. The most frequent tree species combinations were identified and their representativeness in terms of the total forest area was analyzed. In addition, to ensure environmental homogeneity in the proposed classification, the arrangement of each tree species combination within the Spanish Mediterranean ecoregions was evaluated using the Pearson Chi-square test. Based on our results, Spanish MMF currently cover 27.07% of the Mediterranean natural forest area. They were divided into 9 main ecological groups and 23 subtypes. The classification of Spanish MMF and the distribution map represent a first step towards recognizing the importance of mixed forests in the vegetation of the Mediterranean Basin. Together they may provide a valuable basis to improve future forest management, monitoring and conservation strategies both at national and European level.
This paper reports a bioclimatic envelope model study of the potential distribution of 19 tree species in the Iberian Peninsula during the Last Glacial Maximum (LGM; 21 000 yr BP) and the ...Mid-Holocene (6000 yr BP). Current patterns of tree species richness and distributions are believed to have been strongly influenced by the climate during these periods. The modelling employed novel machine learning techniques, and its accuracy was evaluated using a threshold-independent method. Two atmospheric general circulation models, UGAMP and ECHAM3 (generated by the Palaeoclimate Modelling Intercomparison Project, PMIP), were used to provide climate scenarios under which the distributions of the 19 tree species were modelled. The results obtained for these scenarios were assessed by agreement measure analysis; they differed significantly for the LGM, but were more similar for the Mid-Holocene. The results for the LGM support the inferred importance of pines in the Iberian Peninsula at this time, and the presence of evergreen Quercus in the south. Important differences in the altitude at which the modelled species grew were also predicted. During the LGM, some normally higher mountain species potentially became reestablished in the foothills of the Pyrenees. The warm Mid-Holocene climate is clearly reflected in the predicted expansion of broad-leaved forests during this period, including the colonization of the northern part of the Iberian Peninsula by evergreen Quercus species.
Distribution area and surface are both parameters of paramount importance for habitat management, monitoring and conservation. Here we present the distribution of eight zonal forest types of mainland ...Spain that are consistent with the Habitat Types (HT) listed in Annex I of the European Union Habitats Directive 92/43 EC. Their dominant species and HT codes are Fagus sylvatica (9120, 9130 and 9150), Quercus robur and Q. pyrenaica (9230), Q. suber (9330), Pinus uncinata (9430), P. nigra ssp. salzmannii (9530) and P. pinea (subset of 9540). These distributions are based on tesserae from the 1:50,000 Spanish Forest Map and are the result of sorting assisted by supplementary databases. The distributions are presented as vector coverages, and provide three information levels of increasing detail: geographic distribution, basic forest type and structural forest patch. Two R scripts are also included with the dataset. They implement a segmented regression approach to investigate forest fragmentation on these or other patch-like data.
European beech is a temperate species with peripheral populations in the Mediterranean mountains. These Mediterranean beech forests have often been interpreted as climatic relicts. However, the use ...of this term does not always seem appropriate because their recent post-glacial expansion in the Iberian Peninsula makes it difficult to disentangle the effects of climate from those associated with land use. Moreover, the observed recovery of Mediterranean beech forests after land-use abandonment is not what might be expected of a climatic relict. Our objective is to clarify this issue by studying one of these Mediterranean “relict” populations located at the south-western margin of this species (Iberian Central Range), discriminating between subpopulations with dissimilar land-use legacies. Differences between the climatic range of this population and the rest of the European beech forests were explored. Historic and recent shifts in the species distribution area were studied by reviewing paleorecords from the literature and contrasting historical and recent cartography. Field data on population structure and demography were also examined. The climatic analyses do not point to climatic marginality of the peripheral population studied. Furthermore, the results revealed no evidence of reduction in area or confinement over recent decades. Contrary to what is expected for climatic relict species under current climate, beech is expanding in the studied mountain following the abandonment of traditional land uses, particularly since the middle of the twentieth century CE. Our results do not support the concept of a climatic relict but rather, they point to the importance of anthropogenic legacies to explain the distribution area, population dynamics and current conservation status of Mediterranean beech forests.
Aim: To improve our knowledge of the process of selection of important plant areas (IPAs), a recent requirement of the Global Strategy for Plant Conservation.The study was conducted at a hotspot of ...plant conservation in the European continent, using a comprehensive database of plant species distribution in the area. Location: Spain. Methods: We used range distribution data for 3218 vascular plants found in Spain, in the form of 10 km UTM squares, totalling 169,124 species occurrences across 5508 UTM cells. We identified IPAs by scoring threat status, endemism, rarity, phylogeny and species richness. We then performed two different analyses, with and without incorporating the species richness score of every square. Finally, a null model was used to obtain a general pattern of species occurrences, we computed an index of occurrence richness (SI), and then we selected a number of specific territories of different sizes to reveal differences in sampling effort within the study area. Results: We identified IPAs in Spain according to the proposed scoring method. We detected a positive relationship among richness and total score calculated with the rest of the criteria. However, endemism and threat status produced certain specific effects for species-poor squares. Regarding sample bias, we detected over-and under-recorded areas. This bias seems to be due to the accumulation of field prospecting in species-rich areas in detriment to poor areas. Main conclusions: We envisage two different approaches to address IPA selection in hotspots. First, we advocate a complementary scoring-mapping method for areas where a relatively large amount of range distribution data and plant knowledge is available. Secondly, as richness per se encompasses a great amount of biogeographical information, we suggest using species richness or any other environmental surrogate to delineate preliminary IPAs in poorly known but species-rich territories.
We model the past and future distribution of Pinus sylvestris in the Iberian Peninsula using the random forest algorithm, a machine learning technique that implements an automatic combination of tree ...predictors. In order to model the past, we chose two of the most climatically significant events recognized affecting the species distribution: the last glacial maximum (LGM, 21 000 cal. BP), and the mid Holocene (6000 cal. BP). In order to model the distributions, we based the future scenarios of climate change upon the four storylines projected by the Intergovernmental Panel on Climate Change (A1, A2, B1 and B2). The results obtained for the past show a contraction in the distribution area during the LGM and a subsequent expansion, coinciding with an improvement in climate in the mid Holocene. The presence of isolated populations of Pinus sylvestris at low altitudes could support the existence of refuges for Northern European taxa. The results obtained for the future, even in the least aggressive scenario, show a rapid decrease in this species' distribution and a vertical migration as a response to climate change. Pinus sylvestris populations would be restricted in the future to the higher altitudes of the mountains in the north of the Iberian Peninsula (mainly the Pyrenees), and the species would disappear from the central and southern mountain ranges. These results identify a risk to high-mountain forest species in the Iberian Peninsula from the effects of global warming.
We present a modelling framework for predicting forest areas. The framework is obtained by integrating a machine learning software suite within the GRASS Geographical Information System (GIS) and by ...providing additional methods for predictive habitat modelling. Three machine learning techniques (Tree-Based Classification, Neural Networks and Random Forest) are available in parallel for modelling from climatic and topographic variables. Model evaluation and parameter selection are measured by sensitivity-specificity ROC analysis, while the final presence and absence maps are obtained through maximisation of the kappa statistic. The modelling framework is applied at a resolution of 1
km with Iberian subpopulations of
Pinus sylvestris L. forests. For this data set, the most accurate algorithm is Breiman's random forest, an ensemble method which provides automatic combination of tree-classifiers trained on bootstrapped subsamples and randomised variable sets. All models show a potential area of
P. sylvestris for the Iberian Peninsula which is larger than the present one, a result corroborated by regional pollen analyses.