This study shows that temperature was one of the main parameters driving picocyanobacterial abundance. The correlation between temperature and increase in picocyanobacterial abundance is well known in both marine 8,9 and freshwater systems 61,62. Picocyanobacterial abundance started increasing at > 10°C, in line with other temperate ecosystems (e.g. 63) and previous records in the Baltic Sea 9. Peak abundances at the coastal K-station and offshore LMO during 2019 and 2020 were in the same range, with the exception of the summer of 2018, when K-station peak abundances reached 4.7 × 105 cells mL− 1. These numbers are comparable to other observations in the Baltic Sea Proper during summer (105 cells mL− 1) 55,64, and confirm previous observations that picocyanobacterial abundances are as high in coastal and offshore locations 53. It is also important to note that the picocyanobacterial abundances at < 10°C reported in this study were notably higher than previous reports in the Gulf of Finland and in other temperate ecosystems 8,9. The ecological niche for SYN has recently been defined as > 5°C 19. However, in this study abundances of > 104 cells mL− 1 were recorded during winter time at both K-station (0-5.7°C) and LMO (2.7–5.8°C), suggesting that the strains present in the Baltic Sea are well adapted to low temperatures, in line with previous observations by Paulsen et al., 65.
According to the PLS models independent variables only explained 38 and 40% of the total variation of PE-rich and PC-rich respectively. This indicates other important controllers such as light quality 4, grazing by ciliates and flagellates 66 or viral lysis 67 may be important drivers of picocyanobacterial abundances should be included in future models.
SYN was divided into PE-rich and PC-rich depending of the pigment content. PE-rich and PC-rich coexisted at similar abundances during the summer, confirming previous observations based on cpcBA and cpeBA libraries 30. PE-rich is better adapted to blue light which can penetrate deeper in the mixing layer, while PC-rich is adapted to red light, which is dissipated in the surface 4. As a result, PE-rich was equally prevalent in both the K-station and the LMO. On the other hand, PC-rich abundance variation was strongly linked to the stratification index (N2) and was more prevalent in the coastal K-station compared to the offshore LMO, in line with observations in other estuaries and freshwater lakes 31,68. These results suggest a horizontal gradient on the Baltic Sea, where PC-rich is more prevalent in coastal shallow areas while in offshore areas abundances are lower and tightly joined to stratification seasonality. In future, an increase in the stratification periods as a result of global warming could reinforce PC-rich dominance on the picocyanobacterial community. PC-rich has been observed to have a negative effect on co-occurring filter-feeders, the ability to avoid predation and viral lysis, which can affect the energy flow to upper trophic layers 69–71. Thus, understanding the physiology and ecology of PC-rich, a generally understudied group of SYN, is of high importance for the understanding of current and future climate scenarios.
Nutrient availability, particularly nitrogen species, was correlated to picocyanobacterial dynamics according to the PLS models. For example, both PE-rich and PC-rich showed moderate negative correlation with NO3 abundance. Picocyanobacteria (both PE-rich and PC-rich) showed a negative correlation with NO3 concentration. Several studies have documented the preference of picocyanobacteria for NH4 over NO3 at high temperature conditions (> 15°C) 53,72. Thus, newly fixed nitrogen in the form of NH4 from N2-fixers may be a main driver supporting picocyanobacterial growth during summer 33–35, in line with the positive correlation between N2-fixers and PE-rich. However, PE-rich peak abundances at the K-station and LMO were in the same range, although N2-fixers were only observed at the K-station during 2018. This suggests that PE-rich can benefit from newly fixed nitrogen, but it is not a requirement to achieve peak abundances and thus other nitrogen pools should also be considered. In fact, the main nitrogen source for picocyanobacteria during the summer could be originated from benthic regeneration, which in coastal and shallow water areas (< 50 m depth) can represent up to 97% of the nutrient requirements 35,73. In addition, peak abundances at the LMO are sustained during the first half of the autumn at < 10°C, which indicates that picocyanobacteria can uptake NO3 at low concentrations efficiently.
The community composition was studied using 16S rRNA gene sequences amplified using specific primers that target almost exclusively picocyanobacteria 60. The results corroborate that the V5-V7 region of the 16S rRNA gene showcases higher variability in picocyanobacteria than the V3-V4 region, revealing an unprecendented high strain diversity in the Baltic Sea with a particularly high number of ASVs at the coastal K-station. Most of the previously defined clades and clusters were described 60, but some clusters were altered. For example, clades A and B clustered together (clade A/B) contrasting with phylogeny based on other regions of the 16S rRNA gene 12,74. All ASVs in clade A/B displayed similar seasonal variation in relative abundance, suggesting a similar ecophysiology whithin the clade. The increase in contribution of S5.2 took place in June-July, when temperature was > 18°C and NO3 concentration was low. This result suggests that S5.2 affiliated picocyanobacterial strains are adapted to high temperatures and may use NH4 as a primary nitrogen source 35,53,72. On the other hand, clade A/B dominated during the colder months, indicating that picocyanobacteria strains in this clade are well adapted to low temperatures and high NO3 concentration.
This study indicates a coastal to offshore differentiation in picocyanobacterial community composition. The coastal K-station presented higher ASV diversity than the offshore LMO. Moreover, the clades KS2, KS5 and KS6 were only present in the coastal K-station, which suggest that some clades are only present in the coastal region. These results contradicts previous observations in the Baltic proper where no differences in community composition were observed in coastal offshore gradients 6. One explanation is that the higher resolution achieved with the primers in this study have revealed differences that could not be detected with less specific primers. Community composition seasonal dynamics in the coastal and offshore stations also showed major differences. At the K-station picocyanobacterial peak abundances took place when S5.2 was dominating the community while at the LMO peak abundances took place under clade A/B dominance. The different dynamics at the coastal compared to the offshore station could be driven by low PO4 levels as S5.2 was positively linked to PO4 concentration. At the LMO, PO4 was low during summer (0-0.32 µM) while it remained high at the K-station (0-1.5 µM), explaining the lower contribution of S5.2 in the LMO. The potential PO4 limitation could also explain why at the LMO S5.2 has lower contributions during summer to the community composition compared to the K-station. However, to fully understand picocyanobacterial dynamics, other parameters such as light hours 75, NH4 recycling rates 35 or nutrient competition with specific phytoplankton groups 72 should also be considered.
The most abundant ASV in the dataset, ASV_00001, was identical to the V5-V7 region of a metagenome-assembled genome (MAG) reconstructed from the Baltic Proper (BACL30) and has been identified as dominant in the Baltic Sea 16,58. Phylogenetic classification based on amino-acid classification included BACL30 in the S5.2 clade 16; however this classification may have been biased by the lack of genome sequenced estuarine and freshwater SYN strains. In this study, ASV_00001 had 99% identity with the strain MW73D5, a freshwater strain included in clade A/B 12. Most of the ASVs of the picocyanobacterial community in the Baltic Sea were more similar to freshwater strains rather than estuarine and marine strains, which suggest a freswater origin as opposed to a global brackish microbiome 58. ASV_00001 showed high contributions during the summer, particularly at the LMO (up to 78%), in line with observations in other offshore locations 15,16. However, the highest contributions took place in the cold months, indicating that BACL30 is well adapted to low temperature conditions.
This study provides a detailed description of picocyanobacterial seasonal abundance, biomass contribution and community composition during three years at a coastal and an offshore station in the Baltic Sea Proper, showing that SYN are highly adaptable and diverse. The results bridges the gap between phylogenetic classification and ecology. In a climate change scenario, longer and warmer summers could result in earlier picocyanobacterial blooms in the coast as a consequence of achieving optimal temperatures for S5.2 ecotypes earlier in the spring/summer season. This effect could be further magnified by earlier and more extensive blooms of N2-fixing cyanobacteria resulting in higher NH4 availability, that are projected as a consequence of global warming 76,77. However, at offhore locations in the Baltic Proper, the picocyanobacterial summer bloom could be delayed since optimal temperature for clade A/B would take place later in the year. The results in this study highlight that besides temperature, water column stratification and nutrient availability also play an important role in picocyanobacterial dynamics and community composition.