Forest Composition in Mediterranean Mountains is Projected to Shift Along the Entire Elevational Gradient Under Climate Change

Mountainous areas in southern Europe act as wet-cool microclimate refugia in a warm-dry region for genetically unique and endemic species. However, recent predictions suggest that species rich regions within the Mediterranean mountains are at great risk to climate change because changes in atmospheric dynamics will cause these areas to become warmer and drier. Therefore, evaluating the effects of future climate change on the diversity, composition, and distribution of Mediterranean mountainous species is essential to the persistence of their genetic heritage. By using high spatial resolution data, generalized linear models, and machine learning models, Ruiz-Labourdette et al. (2011) modeled the current and future distributions of 15 tree species in two connected mountain ranges in the Iberian Peninsula across the elevational gradient. The distributions were modeled as functions of climate, lithology, and soil water availability. Additionally, Ruiz-Labourdette et al. analyzed potential changes in the composition of tree communities. The authors’ models predicted an upward migration of Mediterranean tree communities to higher elevations, an expansion of drought and high temperature tolerant tree species’ ranges, and a decline in temperate, cold-adapted tree communities with moderate water requirements. Furthermore, the mountain lowlands showed the largest projected changes in species distribution. Ultimately, these findings indicate that climate change may result in a loss of tree species diversity within these mountain ranges.—Megan Smith

Ruiz-Labourdette, D., Dogues-Bravo, D., Ollero, H.S., Schmitz, M.F., Pineda, F.D., 2011. Forest Composition in Mediterranean Mountains is Projected to Shift Along the Entire Elevational Gradient Under Climate Change. Journal of Biogeography 39:1. DOI: 10.1111/j.1365-2699.2011.02592.x

The varied topography within the Mediterranean Mountains created a variety of microclimates that acted as refugia for species that thrived in warm environments during periods of adverse climatic conditions. Additionally, the mountains protected other species that grew in cold and wet environments from post-glacial warming period. As a result, mountain species’ isolation in refugia resulted in one of the highest rates of endemism in Europe.

The study area encompassed a mountainous region of 71,700 km2 in southern Europe and included two large mountain ranges: the Central Mountain Range, located in the central west part of the Iberian Peninsula, and the Iberian Mountain Range, located in the central east part of the peninsula. The Iberian Mountains are primarily composed of limestones, calcarenites, marls, evaporates, dolomites, and sandstones. The Central Mountains consist of granite, metamorphic materials, a homogenous siliceous substratum, and limestone. A map showing the location of the Central Iberian Mountain Ranges, as well as the elevations of the highest summits in these mountain ranges, was displayed.

Forest composition varies between the two mountain ranges. The wetter, northern areas contain forests of temperate and broadleaf species. Mountainous regions with a more continental climate consist of pine forests, while the lower mountainous regions contain sub-Mediterranean deciduous forests. Additionally, wet-warm Mediterranean forests are found to the west of the study area, while warmer and drier piedmont environments and the shaded low-elevation mountainous areas contain tree species resilient to summer droughts.

Spatial distribution models (SDMs) are used to project climate changes in areas that have a suitable climate for the species in question. However, most SDMs utilize low-resolution biological and climatic datasets that cannot identify small patches with suitable conditions. These small patches might reduce the negative effects of climate change by providing microrefugia for communities. Therefore, Ruiz-Labourdette et al. utilized high-resolution data in their study to examine the topographic complexities and microrefugia present in the Iberian mountain ranges. Ultimately, this method achieved a more accurate representation of species distribution and community composition.

The authors modeled the current and future distributions of 15 tree species in Iberian Peninsula mountain ranges as functions of climate, lithology, and soil-water availability using generalized linear models and data mining models. Ruiz-Labourdette et al. also mapped the variation in the composition of current and potential forest communities in the study region using a multivariate ordination of a matrix of presence/absence of tree species. Two IPCC Special Report on Emission Scenarios (SRES) (A2 and B2) between the years of 2041–70 and 2071–2100 were used to model the study area’s future forest composition.

Forest species data were collected from a forest map of Spain at a scale of 1:50,000. The map was rasterized to a 500 x 500 m grid to create a data matrix of the current presence/absence of tree species in 286,688 cells. Tree species that occurred in more than 500 cells were incorporated in the study. Lithological data were collected from a geological map of the Iberian Peninsula and were classified into six lithological groups. Additionally, the authors incorporated soil hydromorphology into their study by calculating the topographic ratio (TR), which assesses the effects of topography on water drainage.

Ten climatic variables were used to model the potential distribution of tree species. These variables were obtained from monthly measurements recorded at 752 rainfall stations and 197 temperature stations that belong to the observation network of the Spanish State Meteorology Agency. Current climate conditions were recorded for the period 1961–90. For the periods 2041–70 and 2071–2100, the IPCC SRES scenarios A2 and B2 were used. The scenarios differ in the amount of carbon emitted from energy and industrial sources by 2100, with A2 being the most severe. The global models were downscaled to a regional scale to make the information on current and future climatic conditions available at the level of individual weather stations. These data were used to develop climatic maps. To construct climatic maps in a 500 x 500 m grid format, current and future climate data were inserted over the study area using multivariate stepwise regressions that incorporated quantitative geographic data (elevation, distance from coast, latitude, longitude, and slope) and qualitative data (drainage basins and aspect). Hydromorphology of the soils in the future scenarios was calculated by applying the expected increase or decrease in yearly rainfall volume to the current hydromorphology map.

GLM and data mining (gradient boosting) species distribution models were used to create alternative species spatial projections. The models were validated by splitting the data into two groups. Two-thirds of the data were used for calibration and one-third was used to evaluate and validate the models. Two probability maps were constructed for each species, one using GLM and the other using the gradient boosting technique. Binary presence/absence maps were created from probability maps using a threshold to maximize the kappa statistic, which defines the similarity between the binary map and the available biological evidence. Ruiz-Labourdette et al. chose between the GLM and gradient boosting models based on the kappa value obtained with the validation subset. They selected the model that best explained the current species distribution. Species were only modeled if they achieved a kappa value > 0.4 in one of two models. The model that was selected and validated for each species was applied to the future climate scenarios. A table displaying the kappa values obtained for each tree species in the GLM and gradient-boosting models was constructed.

Using the results of the models, the authors created a presence/absence matrix for all the tree species in the five scenarios examined (current; B2 for 2041–70; A2 for 2041–70; B2 for 2071–2100; and A2 for 2071–2100). This matrix was analyzed by multiple correspondence analyses (MCA) to identify trends in the variation of the current and future compositions of species. A table displaying the contributions of the tree species to the first axis of the MCA analysis was constructed.

Ruiz-Labourdette et al. incorporated 12 of the 15 species in their study that achieved acceptable kappa values. Gradient-boosting models were selected for the majority of the species (8 species).

The authors observed the largest changes in distribution for Mediterranean species such as xeric conifers and sclerophyllous evergreen species that are tolerant of high temperatures and summertime drought. These species ranges were expected to increase by 350% under the most severe climate change scenario (A2). Interestingly, their future range only occurred in regions in which the elevation did not differ from the elevation of their original range. This suggests that these species may spread from their present range and colonize flat piedmont and low-elevation mountainous areas. In contrast, Ruiz-Labourdette et al. found that Eurosiberian coniferous and broad-leaved species (species that prefer cold and wet environments) would experience a significant decrease in range. This decrease ranged from 80–99%, depending on the climate scenario. The Eurosiberian species are also expected to undergo the largest elevational displacement: upward between 200 and 550 m. As a result, the tree line would rise, and these tree species would colonize treeless areas that are now occupied by high-mountain grasslands. Unfortunately, it is likely that areas with climatic conditions suitable for their survival would disappear from the central and southern massifs.

The models also predicted that the ranges of sub-Mediterranean species (semi-deciduous oak trees, sub-Mediterranean gymnosperms, and sub-Mediterranean phreatophytes) would remain constant or decrease slightly under different climate scenarios (between 5% and 70%). These species optimal growing conditions may also occur at slightly higher elevations, compensating for the warming and water deficits. A figure displaying the current and projected distributions of three tree species in the Central and Iberian Mountain Ranges under the A2 climate scenario was constructed. Additionally, a graph displaying the potential change in the distribution of all the tree species in each three groups was constructed.

The authors’ models revealed that shifts in tree communities would occur as a result of an increase in the area where the climate is Mediterranean. The proportion of Mediterranean forests, especially the perennial sclerophyllous species, will increase, as will the range of xerophyllous vegetation. This species currently occupies marginal areas in warm, dry, sheltered piedmont enclaves. However, in the future, they may become the dominant vegetation within these regions. Semi-deciduous and deciduous species now found in flat piedmont and low elevation areas that require a moderate amount of water are predicted to decline. Furthermore, coniferous forests that grow in cold regions, as well as the Eurosiberian broadleaf forests that grow in wet and cold conditions, will experience a latitudinal displacement towards the north and a reduction in range. They will disappear entirely from the westernmost mountains. Instead, oak forests will become the dominant forests within these regions. A figure displaying the changes in tree communities projected for the Central and Iberian Mountain Ranges in the Iberian Peninsula was constructed.

Overall, Ruiz-Labourdette et al.’s results indicate that climate warming and a decrease in the availability of water will alter the abundance and diversity of mountain tree biota. Tree species resilient to high temperatures and drought that occupy lower elevations could increase in range and elevation, while species that persist in cold and wet environments may decline as a result of water stress. However, changes in the spatial extent of species ranges and in community composition will be greatest at lower, rather than at higher, elevations. Therefore, changes in tree species’ distributions and forest communities will occur across the entire range of elevation among these mountainous regions.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s