High-latitude polar oceans are covered in an extensive layer of sea ice, which grows and shrinks throughout seasonal cycles. Covering 3–6% of Earth’s surface area (Comiso, 2003), the high albedo of sea ice and snow cover reflects large fractions of incoming solar radiation, playing a crucial role in regulating Earth’s global climate system (Hall, 2004). However, the anthropogenic increase in atmospheric greenhouse gas concentrations is increasing Earth’s surface temperature and driving variability in sea ice dynamics (e.g., Thomas and Dieckmann, 2008). In the Arctic, the once dominant thicker and more persistent multi-year sea ice (MYI) is declining and transitioning into thinner and more ephemeral first-year sea ice (FYI; Polyakov et al., 2012; Meier et al., 2014). Recent climate models have estimated that within the Arctic, an ice-free summer is likely to occur before the end of this century (Overland and Wang, 2013; Notz and Stroeve, 2018). In the Antarctic, an overall gradual increase in sea ice extent was observed from 1979 to 2014, with the greatest increases occurring in the Ross Sea. However, since 2014, strong declines in overall Antarctic sea ice extent have greatly exceeded rates in the Arctic, with a record low recorded in 2017 (Maksym, 2019; Parkinson, 2019). Current climate models predict that substantial continental scale losses of sea ice will occur in the Antarctic before the end of this century (Bracegirdle et al., 2008; Collins et al., 2013; Smith et al., 2014).

Sea ice provides an important habitat for diverse under-ice microalgal communities that are typically dominated by diatoms (Arrigo, 2014). Primary productivity is closely associated with variations in sea ice conditions and its control on sunlight availability, consequently influencing the amount of under-ice algal biomass and detrital material falling to the seafloor (Dayton et al., 1986; Lizotte, 2001; Boetius et al., 2013). Once settled on the seafloor, the algal-derived organic matter serves as a major food resource to benthic primary consumers (Norkko et al., 2007; Wing et al., 2012, 2018). Predictions for the Antarctic suggest that declines in sea ice and snow cover will lead to increased primary productivity (Lohrer et al., 2013; Smith et al., 2014; Leung et al., 2015; Kaufman et al., 2017). There have been observations of rapid and long-lasting reorganizations of benthic macrofauna (Thrush and Cummings, 2011; Dayton et al., 2013; Clark et al., 2017) associated with changes in sea ice conditions and organic matter deposition. However, much less is known about the effects on benthic microbial communities, the key drivers of organic matter remineralization processes and biogeochemical cycling (Falkowski et al., 2008). Previous polar studies have shown that the quality and quantity of organic matter are major drivers of microbial diversity and community composition, with certain taxonomic groups showing strong affiliations with environmental parameters such as sediment algal pigment concentration and total organic carbon (TOC; Bienhold et al., 2012; Ruff et al., 2014; Learman et al., 2016; Cho et al., 2020). However, few studies in Antarctica have characterized the benthic microbial communities, and none have compared the effects of FYI and MYI, which may provide a unique perspective on marine ecosystem functioning in these difficult to study and rapidly changing polar environments.

The Ross Sea represents the southernmost seasonally ice-free ocean region in the world, and sea ice dynamics in this region have so far been among the slowest on Earth to react to climate change – unlike regions further north, such as the Antarctic Peninsula (Kim et al., 2018). The McMurdo Sound, in the southwestern Ross Sea (77°S latitude), is covered by land-fast sea ice (ice fastened to the coastline) for most of the year. Sea ice thickness and persistence conditions within McMurdo Sound differ across the east and west sides which, along with circulation patterns, drive regions of low and high primary productivity, generating measurable differences in the supply of labile organic matter to the seafloor and benthic macrofauna assemblages (Dayton and Oliver, 1977; Barry, 1988; Thrush et al., 2006). In this study, we used the strong east–west food supply gradients at Cape Evans and New Harbour as tractable means of determining whether the structure and diversity of the benthic microbial communities are reflective of FYI and MYI conditions, respectively. Due to the proximity of New Harbour to the terminus of the Taylor Valley, one of the McMurdo Dry Valleys, we also considered the potential influence of local terrestrial inputs on the microbial communities at New Harbour. Terrestrial surface material is known to be transported by katabatic winds down the valley to the surface of the near-shore MYI (Murray et al., 2013). Additionally, we investigated the hypothesis proposed by Mikucki et al. (2015) that a sub-permafrost brine aquifer underlying the Taylor Valley discharges at New Harbour by investigating the presence of extreme halophiles.

McMurdo Sound in the southwestern Ross Sea exhibits contrasting sea ice conditions that drive regions of low vs. high primary productivity (Dayton and Oliver, 1977; Lohrer et al., 2013), providing a unique setting to investigate and compare the impact of sea ice thickness and persistence on shallow-water coastal benthic microbial communities and their potential role as sentinels for climate change. Land-fast FYI breaks out annually in the east at Cape Evans, and the area is typically ice-free from mid-December to late March, whereas thicker MYI in the west at New Harbour rarely breaks out. Sea ice dynamics at these two sites are well-known to influence food supply, benthic faunal densities, community composition and food web structure (Dayton and Oliver, 1977; Kim et al., 2019), and microbially mediated rates and processes (Lohrer et al., 2013). Here, we show that there is a strong association between the legacy of sea ice dynamics and benthic microbial community structure and composition. With Antarctic and Arctic sea ice predicted to decline this century (Bracegirdle et al., 2008; Collins et al., 2013; Overland and Wang, 2013; Smith et al., 2014; Notz and Stroeve, 2018), this highlights the capacity of benthic microbial communities to be sentinels for climate change within polar marine ecosystems.

We found that the origin and quantity of organic matter on the seafloor at Cape Evans and New Harbour reflected the legacy of local sea ice dynamics. The measured stable isotopes of carbon within the sediments at both sites (δ¹³C –20.3‰ to −22.7‰) were suggestive that the origin of the organic matter was likely to be the same at both sites [i.e., diatoms (−20.3‰; Meyers, 1994)], and were comparable to sea ice associated δ¹³C microbial communities previously measured at Cape Evans [−20.2‰ (Norkko et al., 2007); −19.8‰ (Wing et al., 2018)]. Sediment δ¹³C at Cape Evans and New Harbour was marginally lower than that previously reported (−18.7‰ and −19.7‰, respectively) and higher than δ¹³C found in pelagic phytoplankton (−23.8‰ and −24.0‰, respectively; Norkko et al., 2007). When combined, these results indicate that settling sea ice algae provided the primary source of organic matter to the benthos at both sites at the time of sampling, not advected phytoplankton from the water column. This finding supports previous observations that organic matter originating from sea ice is the primary food source for benthic macrofauna under sea ice at Cape Evans (Dayton et al., 1986; Norkko et al., 2007; Wing et al., 2018). We cannot rule out that some primary production does come from benthic diatoms. However, Wing et al. (2018) and Norkko et al. (2007) have also previously reported that sea ice-associated algae is the primary source of organic matter to the benthos within McMurdo Sound.

The impact of FYI vs. MYI on the sediment environment was evident in the concentrations of chlorophyll a, pheophytin, TOC, and TN, which were higher at the FYI site (Cape Evans) than the MYI site (New Harbour). These differences reflect the legacy of ice thickness and persistence at these two sites as sea ice conditions (including ice opacity and snow cover) affect sea ice algal productivity, the flux of ice algal-derived organic matter to the seafloor, and in situ microphytobenthos production (Dayton et al., 1986; Norkko et al., 2007; Thrush and Cummings, 2011). Consistent with higher chlorophyll a concentration and TOC content, heterotrophic bacteria associated with phytoplankton blooms and which participate in the degradation of algal polysaccharides (Teeling et al., 2012; Buchan et al., 2014; Cardman et al., 2014; Krüger et al., 2019) were enriched at Cape Evans compared to New Harbour. For example, Cape Evans sediments had higher abundances of Flavobacteriaceae, which are well-known specialists in the initial break down of algal-derived organic matter (Teeling et al., 2012; Buchan et al., 2014; Krüger et al., 2019). Studies in the Antarctic Polar Front (Ruff et al., 2014), the Antarctic Peninsula (Learman et al., 2016; Li et al., 2020), and the Mertz Glacier Polynya, East Antarctica (Bowman and McCuaig, 2003) support the role of Flavobacteriaceae, including the genus Ulvibacter, in the degradation of diatom-derived organic matter. The sloughing of ice algae from the under-ice layer and melting of the FYI itself at Cape Evans could also play a role in structuring the benthic microbial community, with the relative sequence abundance of taxa associated with sinking algal aggregates [e.g., Ulvibacter and Winogradskyella (Rapp et al., 2018)] higher than in New Harbour. Taxa within the family Rubritaleaceae (genera Haloferula and Luteolibacter) were also more common at Cape Evans. An Arctic fjord (Smeerenburgfjord, Svalbard) study suggested that both Haloferula and Luteolibacter were associated with the breakdown of laminarin, the major storage glucan in diatoms (Cardman et al., 2014). Laminarin is one of the most well-studied algal polysaccharides and contributes substantially to carbon export in the ocean (Becker et al., 2020). Together, these taxa may also be playing a central role within the community at Cape Evans by breaking down complex algal polysaccharides.

In support of increased organic matter deposition at Cape Evans, strictly anaerobic sulfate-reducing bacteria [e.g., Desulfocapsaceae (CE 4.0%±1.5; NH 0.6%±0.5)] were more abundant at this site compared to New Harbour and the PICRUSt2 pathway prediction tool identified that this community had a greater capacity for sulfate reduction I (assimilatory). Higher organic matter deposition rates and increased degradation of organic matter can result in greater compression of redox zones which affects microbial community distribution (Rasigraf et al., 2020). It is therefore likely that the anoxic sediment layer is shallower at Cape Evans, and that anaerobic organic matter degradation processes are more important within this community than at New Harbour. Altogether, our results suggest that at Cape Evans, the legacy of annual FYI breakout and intense pulse fluxes of organic matter have selected for heterotrophic taxa that specialize in degrading algal detrital material within the sediments. In regions where MYI is predicted to breakout more frequently, heterotrophic benthic bacteria are likely to play an increasingly important role in transferring energy through the food web.

At New Harbour, windblown terrestrial sediment is known to accumulate on the near-shore MYI at the terminus of the ice-free Taylor Valley (Murray et al., 2013), thus it was important that we assess the possible influence of terrestrial inputs on the benthic microbial community at New Harbour. Our analysis showed that the composition of the microbial community collected from sediment within the shoreline moat, where the MYI meets the Taylor Valley coastline, was composed of taxa that reflected both terrestrial (Taylor Valley) and marine (New Harbour) input. For example, similarities with the Taylor Valley included the presence of Actinobacteria, Blastocatellia, and Thermoleophilia, whereas the presence of Gammaproteobacteria and Planctomycetes reflect more marine influences as would be expected in an intertidal marine environment impacted by katabatic Taylor Valley winds. These observations show that aeolian inputs have had minimal impact on the composition of the microbial community at our New Harbour marine site, consistent with the findings by Chewings et al. (2014) who calculated low sedimentation rates of aeolian material in New Harbour. Nevertheless, accumulations of terrestrial sediments within and on the surface of the MYI in Explorers Cove, while very evident, is also very patchy (authors’ pers. obs.) and more widespread seafloor sampling is recommended to confirm the consistency of this observation.

In comparison with Cape Evans, the presence of MYI and low organic inputs at New Harbour selected for a community enriched in chemolithoautotrophic taxa. New Harbour had comparatively higher abundances of taxa capable of ammonia oxidation [e.g., Nitrosopumilaceae (Könneke et al., 2014)] and nitrite oxidation [e.g., Nitrospira (Daims et al., 2015)], and those with known metabolic potential for sulfur and hydrogen oxidation [e.g., Woeseiaceae (Mußmann et al., 2017; Hoffmann et al., 2020)]. Our PICRUSt2 analysis also predicted higher representations of CO₂ fixation pathways at New Harbour including the rTCA cycle I and CBB cycle, and included predictions for a higher representation of the nitrate reduction VI (assimilatory) and nitrifier denitrification pathways. The enrichment of ammonia-oxidizing Nitrosopumilaceae in the low organic matter sediments at New Harbour was consistent with reports that this archaeal group thrives in nutrient limited environments (Könneke et al., 2014) including deep-sea sediments of the Ross Sea (Learman et al., 2016). Deep-sea sediment studies in the Guaymas Basin, Gulf of California (Baker et al., 2013) and North Pond, Mid-Atlantic Ridge (Zhao et al., 2019) also suggested the role of nitrite-oxidizing bacteria Nitrospira as important contributors to nitrogen cycling, carrying out the second step of nitrification coupled to CO₂ fixation via the rTCA cycle (Lücker et al., 2010; Bayer et al., 2020). The higher relative importance of chemoautotrophy at New Harbour was further supported by the enrichment of Woeseiaceae (previously JTB225 Marine Benthic Group) which are prevalent in deep-sea sediments (Bienhold et al., 2016; Dyksma et al., 2016) and are capable of CO₂ fixation via the CBB cycle (Mußmann et al., 2017; Hoffmann et al., 2020). Consistent with lower sediment chlorophyll a and TOC at New Harbour, it is likely that these chemolithoautotrophic taxa have adapted to life under a legacy of limited organic resource availability by specializing in using alternative inorganic energy sources. Thus, the selection of these taxa at New Harbour is reflective of the legacy of the prevailing MYI conditions. Under current climate-change scenarios, the role that chemolithoautotrophic taxa play in supporting primary productivity within the community may become less important if organic matter fluxes increase.

In 2015, Mikucki et al. (2015) proposed that a sub-permafrost brine system may exist in the Taylor Valley, discharging at New Harbour. In searching for known extreme halophiles within the dataset, we identified two halophiles [Thiohalorhabdales (NH 1.5%±0.6; CE 0.5%±0.2) and Geopsychrobacteraceae (NH 0.4%±0.1; CE 0.0%±0.0)] that were significantly more abundant at New Harbour than at Cape Evans, lending taxonomic support for the presence of sub-surface brine. Given that we showed the communities at New Harbour were significantly more diverse than at Cape Evans, these differences between relative sequence abundances of Thiohalorhabdales and Geopsychrobacteraceae across sites are noteworthy. PICRUSt2 also predicted increased glycine betaine degradation I pathways at New Harbour compared to Cape Evans, a compound widely used by the majority of bacteria for osmoprotection (Sleator and Hill, 2002). Extreme halophiles grow in high-osmolarity environments, and their presence at New Harbour supports the proposal of a submarine discharge of sub-permafrost brines from the Taylor Valley. This finding is also consistent with earlier work of the Dry Valley Drilling Project in the 1970s, during which boreholes drilled 183m below sea level near the shoreline of New Harbour filled with liquid brine (Cartwright and Harris, 1981). Further investigations of the subsurface hydrology of the McMurdo Dry Valleys and the associated microbial communities are needed to better understand the degree and implications of such hypothesized connectivity between the two systems.

Our analyses show that the legacy of local sea ice dynamics has a strong influence on the diversity of the benthic microbial communities at Cape Evans and New Harbour. Our observation of increased diversity at New Harbour compared to Cape Evans is potentially reflective of a more stable environment driven by a history of MYI cover and low productivity vs. a more dynamic environment driven by ephemeral and productive FYI. Both the NRI and NTI also revealed greater phylogenetic overdispersion (coexisting taxa are more distantly related than expected by chance) within the communities at New Harbour than at Cape Evans. The low temperatures, prolonged darkness, and consistently low nutrient inputs at New Harbour are analogous to conditions of the deep-sea (Dayton and Oliver, 1977). In deep-sea sediments, infaunal communities exhibit remarkably high diversity despite only intermittent fluxes of detrital material falling from the water column (Hessler and Sanders, 1967). More recently, microbial communities were also shown to be extremely diverse in deep-sea sediments, such as the eastern South Atlantic Ocean (Schauer et al., 2010), the Pacific Ocean in the Arctic (Li et al., 2009), and Fram Strait in the Arctic (Jacob et al., 2013). It has been theorized that the co-existence of many species in deep-sea sediments arises from past competitive interactions within a stable environment, thus promoting niche differentiation and enabling many taxa with narrower metabolic niches to co-exist (Sanders, 1968). Our observations support those by Dayton and Oliver (1977), and we propose that the persistent MYI conditions and low carbon inputs at New Harbour, which provide stable conditions that parallel the deep-sea, selects for more diverse and specialized microbial taxa. These findings support our overarching hypothesis that the legacy of sea ice conditions structure the benthic microbial communities at Cape Evans and New Harbour.

Future climate change scenarios predict substantial sea ice decline this century in both the Antarctic and Arctic (Bracegirdle et al., 2008; Collins et al., 2013; Overland and Wang, 2013; Smith et al., 2014; Notz and Stroeve, 2018; IPCC, 2019). Based on the taxonomic assessment made in this study, increased fluxes of organic matter to the seafloor may cause significant community shifts with heterotrophic taxa (e.g., Flavobacteriaceae, unclassified Gammaproteobacteria, and Rubritaleaceae) becoming more dominant and playing an increasingly important role in transferring energy through the food web. An enrichment of Flavobacteriaceae at New Harbour may be a sign that change is already underway, as sea ice breakout events at this site have been increasing in frequency recently (authors’ pers. obs.). Changes to microbial community structure and functioning are likely to affect nutrient cycling between the sediments and water column, with unknown effects on primary production and marine ecosystem functioning. We highlight the role of the connectivity between sea ice and shallow water benthic microbial communities and demonstrate the importance of microorganisms as sentinels of climate change (Cavicchioli et al., 2019) in polar marine ecosystems globally.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found at: https://www.ncbi.nlm.nih.gov/genbank/, PRJNA721518.

The project was conceived and designed by SC, AL, VC, and AM. Support for the project was obtained by SC, AL, and VC. Samples were collected by AL, and all subsequent laboratory work and analyses were conducted by AC and AM. The manuscript was written by AC, AM, and SC with contribution and editing by SS, AL, and VC. All authors contributed to the article and approved the submitted version.

This research was conducted as part of the Resilience in Antarctic Biota and Ecosystems Project, funded by the New Zealand Antarctic Research Institute (RFP2016-2) to SC, VC, and AL. Support for field logistics was provided by Antarctica New Zealand. Support for AC came directly through a University of Waikato Postgraduate Scholarship. AM was supported by Smart Ideas Award (UOWX1602) from the New Zealand Ministry of Business, Innovation and Employment and the Rutherford Foundation Royal Society Te Aparangi Postdoctoral Fellowship (20-UOW006).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.