Fish Community Structure, Diversity, and Their Environmental Correlates in Two Arid-Region Lakes of Northwest China: Bosten Lake and Taitema Lake

Bosten Lake and Taitema Lake are important inland lakes in arid Northwest China, but recent fish-assemblage variation in these systems remains poorly resolved. We surveyed fish communities at 10 sites in Bosten Lake and four sites in Taitema Lake from 2023 to 2025 using standardized conventional capture methods. Across the three years, 20 species from 7 orders and 12 families were recorded in Bosten Lake and 12 species from 6 orders and 9 families in Taitema Lake; Cyprinidae was the most species-rich family in both lakes. The cumulative observed species pool and effort-standardized richness were higher in Bosten Lake, but year-specific site-level richness contrasts were not significant after Holm adjustment. Community composition differed between lakes after adjustment for year using exact whole-site permutations (Bray–Curtis PERMANOVA: pseudo-F1,38 = 37.92, partial R2 = 0.499, p = 0.0010), although multivariate dispersion also differed (PERMDISP: F = 5.46, p = 0.030). Site-restricted PERMANOVA detected year effects in Bosten Lake (pseudo-F2,18 = 29.37, partial R2 = 0.765, p < 0.001) and Taitema Lake (pseudo-F2,6 = 156.36, partial R2 = 0.981, exact p = 0.0046). Bosten Lake had higher Margalef richness (Taitema minus Bosten: −0.818, 95% CI −1.451 to −0.185, p = 0.016) and Pielou evenness (−0.056, 95% CI −0.105 to −0.008, p = 0.027), whereas Shannon–Wiener diversity did not differ detectably between lakes (−0.169, 95% CI −0.673 to 0.335, p = 0.478). After controlling for site and year, no diversity–environment correlation remained significant after false-discovery-rate correction, and the partial RDA test of within-site temporal environmental–community covariation was nonsignificant (pseudo-F2,24 = 1.285, partial R2 = 0.097, p = 0.1754). Thus, the study documents clear spatial separation between lakes and interannual variation within lakes, but does not identify the retained environmental variables as statistically robust drivers of those patterns.