Recently formed Antarctic lakes host less diverse benthic bacterial and diatom communities than their older counterparts

Abstract Glacier recession is creating new water bodies in proglacial forelands worldwide, including Antarctica. Yet, it is unknown how microbial communities of recently formed “young” waterbodies (originating decades to a few centuries ago) compare with established “old” counterparts (millennia ago). Here, we compared benthic microbial communities of different lake types on James Ross Island, Antarctic Peninsula, using 16S rDNA metabarcoding and light microscopy to explore bacterial and diatom communities, respectively. We found that the older lakes host significantly more diverse bacterial and diatom communities compared to the young ones. To identify potential mechanisms for these differences, linear models and dbRDA analyses suggested combinations of water temperature, pH, and conductivity to be the most important factors for diversity and community structuring, while differences in geomorphological and hydrological stability, though more difficult to quantify, are likely also influential. These results, along with an indicator species analysis, suggest that physical and chemical constraints associated with individual lakes histories are likely more influential to the assembly of the benthic microbial communities than lake age alone. Collectively, these results improve our understanding of microbial community drivers in Antarctic freshwaters, and help predict how the microbial landscape may shift with future habitat creation within a changing environment.

. Alpha rarefaction curves for the 29 bacterial communities based on the Observed ASVs Table S1. Hydrochemistry of the lakes Table S2. Comparison of old and young lake groups in selected parameters Table S3. Different diversity metrics for bacterial and diatom communities of 'old' and 'young' lakes Note S1. On the ages of the JRI lakes In general, the geomorphological evolution (e.g., Davies et al., 2013;Jennings et al., 2021) and the obtained deglaciation chronologies (e.g., Johnson et al., 2011;Glasser et al., 2014;Nývlt et al., 2014 including unpublished data of Daniel Nývlt, which are currently being put together for two publications), as well as the ages of the origin of some of the lakes (e.g., Bjorck et al., 1996;Hjort et al., 1997;Píšková et al., 2019;Čejka et al., 2020 and further unpublished data from Monolith Lake of Daniel Nývlt, which are currently being put together for a publication) show a rather simple geomorphological and deglaciation history of the currently deglaciated parts of James Ross Island (JRI).
The deglaciation of the lowest lying parts of northern James Ross Island started during the Termination I (Pleistocene-Holocene transition), and coastal zones became glacier-free by 12.9 ka ago (Nývlt et al., 2014). This is the area of Lachman lakes (LA1 and LA2), and the dating of basal lacustrine sediments suggests their origin to be approximately 11.9 ka ago (Hjort et al., 1997). These are by far the oldest lakes on the island, even though they are rather shallow and may dry out during some summer seasons. The rapid early deglaciation during the Holocene lead to the splitting of local glacier cover on James Ross Island from the Antarctic Peninsula Ice Sheet and local glaciers behaved independently since the early Holocene (Glasser et al., 2014;Nývlt et al., 2014Nývlt et al., , 2020. The climatic conditions of most of the Holocene were very similar for the current climate (average for 1950-2000 CE) as calculated by the temperature anomalies from the James Ross Island Ice Cap (Mulvaney et al., 2012). This led to a slow glacier recession of local glaciers between 8 and 2 ka ago (Glasser et al., 2014;unpublished data of Daniel Nývlt) to an extent smaller than at present (Nývlt et al., 2020). The only prominent cooling leading to the Neoglacial phase of local glaciers advances began in this area 2.5-2.0 ka ago (Sterken et al., 2012;Mulvaney et al., 2012;Čejka et al., 2020). Local glaciers advanced from their accumulation areas and deposited prominent frontal and lateral moraines during the Neoglacial phase. The advance culminated approx. 1.0-0.8 ka ago with a second less prominent advance 0.4-0.3 ka ago as evidenced from the Lookalike Glacier (unpublished data of Daniel Nývlt). Since 0.3-0.2 ka ago, all local glaciers retreated with a prominent speedup during the last decades as evidenced by the studies of Carrivick et al. (2012), Engel et al. (2012), and Kaplan et al. (2020).
Basing on the geomorphological evolution outlined above, the lakes which are located outside of the Neoglacial moraines (i.e. our "old" lakes) must be older than 2.0 ka, and some of them are likely even older as evidenced by the dating of basal sediments in the Lachman Lakes. On the contrary, lakes associated with the deglaciation after the Neoglacial culmination (i.e. our "young" lakes) must be younger than approx. 300 years, some of them evolving only during the last decades. The youngest lakes on the Peninsula are kettle lakes, the origin of which could be seen directly in the field evolving in the Neoglacial moraines of local glaciers in connection with the deepening of lakes due to the thermal effect of freshwater on the underlying ice, which still forms the largest proportion of the moraines. A glacier lake outburst of one such kettle lake was recently documented by a Japanese team (Sone et al., 2007).

Note S2. On the water temperature and sampling times
In our analyses, water temperature consistently appeared among the variables potentially contributing to the observed variation between microbial communities of the old and young lakes. An average of the measured water temperatures for the old lakes (excluding the anomalous VO4 -see Discussion for more details) was 7.7 ± 3.5 °C while the young lakes averaged at 1.7 ± 1.5 °C (or 1.2 ± 0.6 °C when BLU young lake outlier with 5.5°C is excluded). Although water temperature (as well as pH in response to a rate of photosynthesis) typically changes with daytime (and weather conditions -see below), our sampling times were not drastically different between the lake groups (with medians ± standard deviations of 15:45 ± 70 min and 16:30 ± 165 min of the local time for old and young lakes, respectively).
While we lack information on the weather/air temperature/sunlight conditions on the sampling days (between 8 th and 17 th February 2017), lakes from both old and young groups were typically sampled on the same day (e.g., 5 old lakes and 2 young on the 12 th of February, 5 old and 2 young on the 13 th of February, 2 old and 4 young on the 15 th of February; Table 1). Moreover, the rogue lake VO4 was sampled on the 9 th of February along with 5 other old lakes, all six between 14:20 and 16:40). Therefore, in the case of this study, potentially different weather conditions on different sampling days and times seem highly unlikely to cause the observed differences between old and young lake groups. Figure S1. Alpha rarefaction curves for the 29 bacterial communities based on the Observed ASVs. Shannon Diversity revealed the same (not shown).