High-throughput DNA sequencing (HTS) and photosynthetic marker-pigment analysis can provide cost-and time -efficient alternatives for morphology-based microscopic identification and counting of lacustrine phytoplankton. However, application of particularly HTS in tropical lakes is still uncommon. We analyzed the community composition of cyanobacterial and eukaryotic phytoplankton in relation to trophic status and other environ-mental factors in 15 tropical crater lakes in western Uganda, using both high-performance liquid chromatog-raphy (HPLC) of photosynthetic marker pigments and HTS of 16 S and 18 S small subunit rRNA genes. We then compared the results with phytoplankton composition data based on microscopic identification and cell counting to evaluate whether the two former methods can be alternative or complementary approaches to analyze these lakes' phytoplankton communities. HTS recorded 105 operational taxonomic units (OTUs) of cyanobacteria, and 197 OTUs of autotrophic eukaryotes mainly from dinoflagellates, cryptophytes and green algae. Ochrophyta (chrysophytes and diatoms) were present in low abundances only, and the few remaining sequences belonged to Streptophyta and Haptophyta. Marker pigment distribution among samples reflected the widespread and often dominant presence of cyanobacteria (primarily zeaxanthin, myxoxanthophyll and echinenone), followed by green algae (chlorophyll b, lutein, antheraxanthin, neoxanthin) and diatoms (fucoxanthin, diato-diadinoxanthin), and with cryptophyta (alloxanthin) important in some of the lakes. Direct and indirect ordi-nations of the OTUs and marker pigments revealed that, similar to patterns observed in cell counts, phyto-plankton community composition differed between deep oligotrophic lakes and shallower more productive lakes, even though the distribution of cyanobacterial OTUs was more related to lake water conductivity and season than to trophic status. Procrustes analyses of the composition of green algae and Ochrophyta, the two dominant eukaryotic groups, based on HTS data showed patterns significantly similar to composition data based on cell counts, but no significant correlations were found between the complete eukaryotic or cyanobacterial HTS data and cell count data. While the latter discrepancies can be attributed to method-related biases in both HTS and microscopy, in contrast to the microscopic approach HTS failed to identify trophic indicator species and did not assign all lakes to the correct trophic category. Composition data based on marker pigments clustered the 15 lakes according to their trophic level only when pigment abundances were expressed as absolute concentrations, not when expressed as fractional abundances. We attribute this lower discriminating power partly to the wide distribution of cyanobacteria in these tropical crater lakes, besides the pigments' low taxonomic resolution. Nevertheless, as the compositional patterns in relation to environmental variation observed in marker pigments and OTUs of the dominant eukaryotic phytoplankton groups are similar to those in cell count data, both methods hold potential for ecosystem-level monitoring of tropical crater lakes. However, the high-resolution HTS approach is handicapped by non-uniform amplification of gene sequences, while the low-resolution marker -pigment approach is handicapped by the widespread dominance of cyanobacteria in these systems irrespective of their trophic status.
