Glyptotendipes Kieffer and Demeijerea Kruseman from Lake Winnipeg, Manitoba, Canada, with the description of four new species (Diptera: Chironomidae) Author Saether, Ole A. text Zootaxa 2011 2760 39 52 journal article 10.5281/zenodo.203816 ee54eab7-bca2-4d95-9496-ca2110bae795 1175-5326 203816 Glyptotendipes (Glyptotendipes) lobiferus (Say) ( Fig. 3 D–F) Material examined. CANADA : Manitoba, Lake Winnipeg, Victoria Beach, 12 males , 9.vii. 1969 ; Pine Dock, 3 males , 10.vii. 1969 ; Gull Harbour, 5 males , 2 females , 16.vii. 1969 ; Gimli Government Wharf, 1 female , 24.vii. 1969 ; 3 km off George Island, 16 males , 27.vii. 1969 ; Grand Rapids Government Wharf, 5 males , 28.vii. 1969 ; McBeth Harbour, 5 males , 30.vii. 1969 . When examining specimens of what was assumed to be this species, it was discovered that there apparently were three sibling species occurring together in Lake Winnipeg. At first it was thought that two of them might be merely male intersexes caused by mermithid infections. However, the "normal" males conforming to previously described G. lobiferus usually had a higher number of sensilla chaetica on the legs, not a lower number as would have been the case had the "unnormal" males represented intersexes ( Saether & Galloway 1980 ). Furthermore, nematodes were never found inside the imagines. One of the three sibling species Glyptotendipes (Glyptotendipes ) sensilis sp. n. , was relatively easily separable. The other two (including G. lobiferus ), however, were overlapping in all characters. Therefore a statistical computer analysis was initiated by the author and performed by Dr. R.H. Green, Department of Zoology, University of Manitoba. Eleven characters were used on 41 specimens . First a straight hierarchical classification using Ward's clustering method was done. Two different allocation procedures were used, one starting from the three group solution of Ward's clustering method, the other a random allocation into three groups. A principal components analysis showed that all specimens with 100% efficiency could be allocated into the two groups using an index of weighted values of the 11 characters. One of the 41 specimens had a 5% probability of having been incorrectly classified, while one other had a probability of 1%. The remaining all had a 100% probability of having been correctly classified. The optimal discriminant function was: - 0.02432184 total length + 1.61533684 total length/profemur + 0.52223190 AR + 0.01362620 dorsocentrals + 0.05081318 prealars - 0.06179725 parascutellars + 0.00603972 scutellars + 0.01182867 setae on squama - 0.00197175 sensilla chaetica on p2 + 0.04968852 sensilla chaetica on p3 + 0.17558169 HR - 8.17067772. If this function comes out positive the specimen is Glyptotendipes (G.) lobiferus , if it comes out negative the specimen belongs to G. (G.) paulisensus sp. n. The ranges and means of the eleven characters in G. lobiferus were (n = 24): Total length 3.9–5.0, 4.27 mm ; Total length/length of profemur 2.68–2.95, 2.79; AR 3.85–4.60, 4.24; dorsocentrals 34–52, 40; prealars 7–15, 12.0; supraalars 1–4, 2.1; setae on scutellum 35–61, 48; setae on squama 28–45, 36; sensilla chaetica on p 2 11–44 , 27.6; sensilla chaetica on ta1 of p3 0–3, 0.42; HR 0.95–1.20, 1.09. The hypopygium is illustrated in Fig. 3 D. There also appeared to be 2 types of females, one with 10–12 setae on gonocoxite IX and a narrow ventrolateral lobe, another with 4–5 setae on gonocoxite IX and a broader ventrolateral lobe. It is assumed that the first type belongs to G. (G.) lobiferus and the second to G. (G.) paulisensus because of the somewhat higher chaetotaxy of the former. The females of G. (G.) l o b i f e r u s have (n = 2): Total length 4.3 mm , total length/length of profemur 2.70– 2.91, AR 0.43–0.47, 55–81 dorsocentrals, 9–11 prealars, 1–2 supraalars, 20 scutellum with 44 setae, squama with 46–65 setae, 58–61 sensilla chaetica on ta1 of p2, 37 sensilla chaetica on ta1 of p3, 10–12 setae on each gonocoxite IX, 88 setae on T IX, 15–20 setae on each side of segment X, 250–252 µm long notum, 16–20 µm wide ventrolateral lobe, 54–60 µm long apodeme lobe, and 200–203 µm long semimal capsules. The female genitalia, are illustrated in Fig. 3 E, F (from Saether 1977 fig. 80 E, F). There still is the possibility that the two species only represent a dimorphic population of G. lobiferus . However, the distribution in the light traps, where G. lobiferus is present alone in several samples while G. paulisensus is clearly dominant in others, make this a highly unlikely proposition. Distribution and ecology. G. l o b i f e r u s is known from all over North America ( Townes 1945 : 143, Oliver et al. 1990 : 47). In view of the presence of at least 3 sibling species in a species agglomerate keying to G. lobiferus , however, the records need re-examining.