Determining the potential distribution of highly invasive plants in the Carpathian Mountains of Ukraine: a species distribution modeling approach under different climate-land- use scenarios and possible implications for natural-resource management
MetadataShow full item record
Background: The Ukrainian Carpathian Mountains are one of the most species-rich ecoregions in Europe. High levels of biodiversity can be explained by the heterogeneous geomorphology of the mountain range, its geographic location bridging the Western and Eastern Carpathians, a moderate climate and ample precipitation, and traditional, extensive land use. The introduction and spread of non-native invasive plant species however threatens biodiversity and negatively impacts the structure and functioning of natural and semi-natural ecosystems. Highly Invasive Plants in the Ukrainian Carpathians: Eleven highly aggressive invasive plant species are the subjects of this study: Acer negundo L., Reynoutria japonica Houtt., Reynoutria x bohemica Chrtek.& Chrtková, Echinocystis lobata (Michx.) Torr. & Grey, Impatiens glandulifera Royle, Heracleum sosnowskyi Manden, Robinia pseudoacacia L., Helianthus tuberosus L., Solidago canadensis L., Solidago gigantea Aiton, and Ambrosia artemisiifolia L. These species share several interconnected characteristics common to successful invaders: (a) inherent competiveness, (b) occupation of an empty niche, and (c) strengthening of competitive ability due to release from enemies and (d) due to mutualism in the invaded range. Meanwhile, the invasibility of the study area is determined by a favorable climate, an extensive network of linear habitats (rivers and roads) and high levels of disturbances along these habitats as well as in and around human settlements. Simultaneously, invasive plant species are expected to profit from the interactions of future land-use intensification and increases in atmospheric CO2 concentrations. Species Distribution Modeling (SDM): This study aims to model/predict suitable habitats for the introduction and establishment of invaders and to determine the contribution of different climatic and anthropogenic factors to the particular patterns of predictions for current environmental conditions and potential future spread under different scenarios. SDM determines habitat suitability by relating the response variable, georeferenced occurrences of species in the Ukrainian Carpathian Mts., to underlying environmental predictor variables through various logistic functions and then extrapolating the fitted models to the entire study area. The choice of predictor variables and ecological interpretation of the outputs are based on the species niche concept and the theory of equilibrium of ecological states. Different algorithms are used on each species because results tend to reflect not only the species’ ecology but also differences in modeling approaches between algorithms. Methods: The study area on which the output of the models is defined is the entire Ukrainian Carpathian Mts.. The response variables are in presence-only format. Six predictor variables are chosen based on expert knowledge of the ecology of the species and on preliminary statistics on the goodness-of-fit of the models. They are: minimum temperature of coldest month (mintcold), maximum temperature of warmest month (maxtwarm), sum of active temperatures > 10°C (sat), proximity to water (s_dist_water), proximity to settlements and roads (s_dist_sett_r), and slope (slope). Two software applications, Maxent and BIOMOD, are used to fit the models and extrapolate to the entire study area. The accuracy of the model predictions is measured with the threshold-independent ROC curve. Predictions are projected onto four future scenarios: (i) climate change/low economic development by 2050; (ii) climate change/high economic development by 2050; (iii) climate change/low economic development by 2100; and (iv) climate change/high economic development by 2100. The significance of predictions and projections is tested with the Wilcoxon ranked sum test and paired Wilcoxon signed-rank test. Results: All AUC values are statistically significant, and the spatial distribution of predictions is similar in Maxent and BIOMOD. Under current climatic and land-use patterns, all algorithms predict suitable habitats for establishment to be aggregated in the southwest, east, and northeast along large rivers and roads at elevations up to approximately 600 m above sea level. Taking correlation between climatic variables into account, the spread of species is primarily limited by warm temperatures or proximity to humans, and the aggregation of favorable values of both groups at low elevations in the southwest, east, and southeast explains the spatial patterns of habitat suitability predictions. For all future change scenarios, suitable habitat ranges are predicted to increase significantly. The net gain of novel suitable habitat is significantly higher under scenarios ii and iv than under i and iii, suggesting that the higher the proposed rate of human development, the more suitable habitat is projected to be gained by all species. The differences between algorithms are significant for some species when projecting, but a general trend is discernible: species that become established in a variety of soils and habitats tend to gain more suitable habitat under future scenarios than more specialist species establishing exclusively along rivers. All species are expected to migrate to higher elevations along linear habitats and to expand laterally from habitats predicted as suitable for current conditions. General Implications for Natural-Resource Management: The eleven invasive plant species have already established viable populations in the regions predicted to contain suitable habitats. Invasion of protected areas is likely if current trends continue and monitoring of areas of high conservation values can be efficiently accomplished using the results of this study as guides to prioritize monitoring efforts. Furthermore, inquiries of why a species is predicted to occur at a certain location and which high- risk locations (in terms of invasibility) are in need of particularly sensitive natural-resource management that reduces anthropogenic pressures can be answered by analyzing the relative importance of predictor variables to habitat suitability. Lastly, the maps produced in this study can be utilized to educate the general public and demand forestry practices and forms of tourism development that minimize the chances of the invaders to spread farther into the mountains.