Keywords
How to Cite
Abstract
The mountain systems of Central Asia are important centres of floristic diversity; however, spatially representative and taxonomically verified data on the distribution of many wild taxa remain fragmentary. This study aimed to quantitatively assess the taxonomic struc-ture, spatial distribution, degree of clustering, documented local species richness, and quality of a long-term array of georeferenced observations of the genus Rosa L. in the Tarbagatai mountain system of eastern Kazakhstan. The initial sample included 127 ob-servations from 2019–2026; after taxonomic filtering, the analysis included 118 Rosa rec-ords corresponding to 109 unique coordinates. A total of 61 records (51.7%) were identi-fied to the species level, while 57 records (48.3%) remained at the Rosa sp. level. The da-taset includes R. spinosissima (33 observations), R. laxa (13), R. platyacantha (10), R. alberti (4), and R. canina (1). Ninety-six records were positionally accurate to within 100 m; after duplicate removal, they corresponded to 88 unique locations. Nearest neighbour analysis revealed strong spatial aggregation of high-precision locations: the observed mean nearest neighbour distance was 0.75 km, compared to a mean expected value of 3.02 km across 999 runs of complete spatial randomness (empirical p = 0.001). With grid thin-ning, the 96 high-precision observations corresponded to 32 occupied 1x1 km cells and 26 occupied 2x2 km cells, indicating significant route and observer bias. Species-identified, high-precision records occupied 14 5x5 km cells; the maximum documented richness, equal to four species, was observed in two cells in the central part of the studied spatial cluster. Only 15 records (12.7%) had Research Grade status, while 103 (87.3%) were classi-fied as Needs ID. These results characterise the structure of the documented array of finds, rather than the demographic structure of populations, and provide a reproducible spatial basis for further expert verification, standardised monitoring, and subsequent modelling of species distributions with mandatory control for spatial sampling bias.