Аннотация:Over the last two decades Species Distribution Modelling a.k.a. SDM became a major tool for computa-
tion ecology. This technique takes advantage of the ecological niche concept that is basic to ecology as well as
biology and implemented with supervised machine learning (ML). Many researchers from various fields use
SDM in their work investigating shifts in species ranges under climate change, nature conservation manage-
ment, and many other fundamental as well as applied tasks. This paper aims at estimating the potential range of
two xerophilous land snails: Xeropicta derbentina (Krynicki, 1836) and Brephulopsis cylindrica (Menke,1828),
(Gastropoda: Stylommatophora). Both species currently expand their geographical ranges thus demonstrating
invasion potential. Being native to the steppe zone of the Black sea region (more particularly, B. cylindrica ini-
tially inhabited the Crimean peninsula), the gastropods are spread now across Ukraine (as west as Transcarpa-
thia in the case of X. derbentina) and in Belarus where one occurrence of B. cylindrica has been documented.
Both are sighted in southwestern Russia in the Belgorod region. Well fitted for arid environments, they are like-
ly to spread further in Europe.
In the presented work, we used information on ca. 300 actual localities, i.e. places where the species were
reported. The data came both from GBIF.org database, published materials, and unpublished field observations
by V. Adamova.To lessen sampling bias, caused by. uneven coverage of the study region by observations, the
initial dataset was reduced with spatial thinning procedure
Environmental predictors, i.e. variables describing ecological conditions, included: a) 19 bioclimatic co-
variates taken from worldclim.org; b) elevation; c) surface type, and d) enhanced vegetation index (EVI) de-
rived from remote sensing (MODIS). Only most impactful and least correlated predictors as evidenced by ex-
ploratory data analysis were further used for ML based on 4 different algorithms. Subsampling and repeated
model training were performed in each case. As a result, we obtained 20 models for X. derbentina and 20 for
B. cylindrica as well as their prediction i.e. estimates of distribution ranges. To summarize their predictions and
lessen the bias of a particular model, we also used ensemble modelling that produced a consensus prediction.
The predictions in our work suggest that the main area of potential range of xerophilous snails
X. derbentina and B. cylindrica covers the steppe and forest-steppe zones of the East European Plain. The high
suitability for X. derbentina falls in the Azov region and comprises the territory of the Donetsk ridge, the Dnie-
per lowland, and the Dnieper Upland. Highly suitable for X. derbentina invasion region spans to the south and
east of the Central Russian Upland.
The SDM results support the common notion that Xeropicta derbentina and Brephulopsis cylindrica are
likely to continue their resettlement thus posing a threat to native steppe ecosystems outside of their current
range. The obtained models provide means for invasion management and ecology of object species.