Procladius (Holotanypus) sagittalis

Procladius (Holotanypus) sagittalis (Kieffer, 1909) Material examined. • 11 pupal exuviae, 1 male, Lake Vartopul 1 (M 3), 1 July 2024 • 25 pupal exuviae, 2 pupae, 1 pharate adult – male, Lake Vartopul 2 (M 4), 1 July 2024. Distribution. Palaearctic and Oriental. Distributed from Europe and North Africa through Iran to Japan and the Russian Far East. One record is known from China (Ashe and O’Connor 2009; de Jong 2016). Habitat. Generally, larvae of the subgenus Holotanypus are dwellers of stagnant and slow flowing waters regardless of size or volume. Langton (1991) noted that P. sagittalis typically occurs in shallow water under 2 m deep, which aligns with the findings from small-volume habitats (e. g., Velasco et al. 1993; Hirabayashi et al. 2004). However, the species has also been recorded from artificial ponds and reservoirs, as well as from backwaters, and large rivers (Bitušík 1993; Evrard 1994; Móra et al. 2010; Quintana et al. 2018). It should be noted that ecological information on the species could be more accurate if the identification of the preimaginal stages were more reliably resolved. Remarks. Identification of the pupal exuviae, and even adult males of Procladius (Holotanypus) is extremely challenging (Vallenduuk and Moller Pillot 2007). The extended key for exuvia (Langton et al. 2013) is not reliably applicable to Procladius material collected from the Maramures lakes due to the variability of the tergite armament. Notably, the distinctive “ fish scale ” armament typical of P. choreus can also appear in some specimens of Procladius Pe 3. The parameters of the thoracic horns appear to be more reliable characteristics for identification. Thus, we propose a model that classifies input data with 97 % accuracy, achieving 100 % for P. sagittalis and 96 % for P. choreus. Based on the decision tree trained on our dataset, we constructed an identification key for distinguishing the aforementioned Procladius species (Table 2). We are aware of the tentative nature of the key and acknowledge that a larger dataset would improve the tuning and evaluation of the proposed system. Therefore, the proposed key should be used with great caution.