Sea ice is one of the largest biomes on earth. The area covered by Arctic (15.6 × 106 km2) and Antarctic (18.8 × 106 km2) sea ice is roughly 4 and 5% of the global ocean surface (361.9 × 106 km2) at their respective maximum extents (Meier, 2017; Stammerjohn and Maksym, 2017). Sea ice is a very diverse and potentially very productive habitat, with primary production estimated to amount to 2–24% of total production in sea-ice covered marine areas (Arrigo, 2017). Sea ice is especially productive in spring and summer when, locally, carbon biomass can be ten times higher in the bottom ice than in the seawater, with values greater than 3 mg chlorophyll a L–1 (Chl a L–1) in bottom layers (e.g., Corneau et al., 2013). On some occasions, ice algae may contribute up to 50–60% of total primary production (Gosselin et al., 1997; McMinn et al., 2010; Fernandez-Mendez et al., 2015). Sympagic (ice-associated) microalgae (see Horner et al., 1992, for terminology) are relevant for global biogeochemical cycles (Vancoppenolle et al., 2013a), especially through their uptake of carbon dioxide (CO2) and role as food source for specialized sympagic and pelagic zooplankton (Søreide et al., 2010; Bluhm et al., 2017; Thomas, 2017). Their contribution to a food chain that supports seabirds (Ramirez et al., 2017) and seals and whales (see Thomas, 2017, and references therein) is especially important due to ice-algal growth prior to any significant phytoplankton blooms. Algal carbon that is not consumed in the ice, water column, or by the benthos, is remineralized or permanently buried at the seafloor. Large export fluxes of carbon biomass from the sea ice into the deep ocean, up to 6.5 g Cm–2 year–1, have been recorded in the Arctic (Boetius et al., 2013). Ice algae released from the sea ice may also form the seeding population for sub-ice algal and ice-edge blooms (Arrigo, 2014).
The contribution of sympagic communities to polar marine biogeochemical cycles is still poorly described, despite its general importance (Vancoppenolle and Tedesco, 2017). This shortcoming is partly due to the fact that when studied in detail the production and composition of microalgae in sea ice range widely and are thus hard to quantify. Algal cell concentrations in sea ice vary by up to six orders of magnitude and algal production rates show similar variation (Arrigo, 2017). Biomass accumulation and production depend on the vertical position of sea-ice algae in the ice cover, and are controlled by various environmental parameters like light, nutrients, temperature and salinity that change with the season (e.g., Cota et al., 1991). Here we summarize available data on sea-ice microalgal biodiversity and production to derive parameters that may serve to improve the functional diversity aspect of sea-ice biogeochemical models. Although this approach may sometimes result in only rough averages, model improvement requires further parameterization. First the various habitats for microalgae in the sea ice are described briefly and modeling aspects discussed, after which a synthesis of over 55 studies on algal community structure and primary production is presented and the strength of the derived parameters is evaluated.
1.1. Sea ice as a habitat for microalgae
Sea ice is a complex habitat that is highly heterogeneous over space and time. The structure of sea ice has been described extensively in a number of reviews (Horner et al., 1992; Ackley and Sullivan, 1994; Arrigo, 2014; Petrich and Eicken, 2017). Several distinct layers in terms of both biochemical and physical properties can be defined. Each layer forms a specific habitat for a variety of microalgal communities with different physiological characteristics and production capacities.
At the ice surface, two different types of communities may evolve. So-called infiltration layers or gap layers can develop following the flooding of surface sea ice with seawater (Haas et al., 2001; Kattner et al., 2004). Flooding occurs when the ice is suppressed below sea level due to snow accumulation. Alternatively, ice floes may be pushed downwards by the pressure of overriding ice floes. The infiltration layer provides a good habitat for microalgae, as the inflowing seawater provides fresh nutrients, and light conditions near the surface are generally not limiting. Production in these layers may be high, and reach more than 2 g C m–2 day–1 (Lizotte and Sullivan, 1991). Whereas infiltration layers are quite common in the Antarctic, melt ponds are more characteristic for the Arctic. Their role in biogeochemical cycles is not well defined. Melt ponds are estimated to contribute less than 5% of the total annual production in the Arctic (Lee et al., 2012). Yet, occasionally they can host large aggregates of diatoms that may form an important carbon source for pelagic and benthic systems (Fernandez-Mendez et al., 2014).
Within the sea ice, dense interior communities may develop, with Chl a concentrations higher than 300 mg m–2 (Archer et al., 1996). Interior ice is structurally different between landfast and pack ice. Landfast ice is formed predominantly from columnar growth that creates a denser structure as it grows under more quiet conditions (Ackley and Sullivan, 1994). Pack ice has a more heterogeneous structure, due to the deformation of sea ice by mechanical stress of wave action. Algal biomass in the interior of Antarctic pack ice may contribute ca. 25% of the depth-integrated ice-algal standing stock within the ice column (Meiners et al., 2012). Turbulent conditions during ice formation result in frazil ice. The loose structure and high brine volume of frazil ice forms a suitable, well-protected habitat for sympagic communities. More biomass accumulates within frazil ice compared to columnar ice (Horner et al., 1992; Ackley and Sullivan, 1994).
Ice-algal production in the interior layers is controlled by a number of stressors that include extreme conditions of light, nutrients, temperature, salinity, and pH (Arrigo, 2014, and references therein). During the freezing process, brine pockets are formed that make a sometimes hostile habitat for sympagic algae. With the extraction of fresh water from the seawater during the freezing process, salinities in the brine pockets in upper ice may increase to more than 200 and temperatures can drop below –20°C (Thomas and Dieckmann, 2002; Petrich and Eicken, 2017). If microalgae can sustain these conditions and biological activity continues, the pH will slowly increase and can reach extreme values higher than 10 (Thomas and Dieckmann, 2002). Some algal species can survive these conditions but production in the sea-ice interior is generally low and concentrated at the seawater interface (Kottmeier and Sullivan, 1987; Lizotte, 2001).
Dense bottom communities may develop in the bottom layers of sea ice. Values higher than 50 g C m–2 have been recorded (Hsiao, 1980; Arrigo and Sullivan, 1992; Arrigo, 2017). Some of the most favorable habitats are found beneath landfast ice in Antarctica, where advection of ocean currents underneath ice sheets depresses the freezing point, resulting in the production of supercooled water and the formation of platelet ice (Smetacek et al., 1992; Arrigo, 2014). The sheltered, yet permeable conditions allow free exchange of nutrients. Platelet ice may support communities of more than 6 g Chl a m–2, which is an order of magnitude higher than concentrations in columnar ice (Arrigo et al., 1995).
Biological production in the bottom communities can be high. The bottom layer is strongly influenced by seawater conditions, and temperature and salinity are moderate. Bottom communities are well adapted to the ambient light climate. Early in the season, irradiance levels below sea ice are low, due to the low angle of incoming solar radiation and as snow cover may attenuate irradiance (Palmisano et al., 1987). Microalgae can acclimate to low irradiance levels by expansion of the light-harvesting complexes and adjustment of the pigment composition (Van Leeuwe et al., 2005; Van Leeuwe and Stefels, 2007; Alou-Font et al., 2013). Changes in the structure of the photosynthetic units may be accompanied by a decrease in their numbers (Barlow et al., 1988; Thomas et al., 1992). In addition, sympagic algae not only acclimate well to changes in light quantity, but also show chromatic acclimation to changes in the light spectrum with strong attenuation in the red, as light penetrates through the ice (Robinson et al., 1995).
The structure of sea ice is dynamic and shows strong vertical gradients in its physical and chemical properties. It is shaped by processes of ice melt and ice formation that altogether govern the biological processes in sea ice. In addition, nutrient supply (e.g., Gradinger, 2009; Fripiat et al., 2017) and snow cover (e.g., Gosselin et al., 1986; Mundy et al., 2007) exert external control on microalgal growth. The algal communities that inhabit sea ice are subsequently subject to seasonal changes in composition, as will be discussed in this paper, as well as biomass accumulation, as has been reviewed in Meiners et al. (2012) for Antarctic sea ice and in Leu et al. (2015) for the Arctic.
1.2. Organizing data for modeling purposes
Sea-ice biogeochemical models are composed of i) state variables, measurable quantities that vary in time and space, such as bulk microalgal biomass, and ii) biophysical parameters, constant values such as algal maximum growth rate and the light utilization coefficient (Vancoppenolle and Tedesco, 2017). However, biophysical parameters can vary in space and time and between different species or taxonomic groups (e.g., Cota and Sullivan, 1990; Campbell et al., 2016). This variability limits the development and general applicability of current sea-ice biogeochemical models across different systems. Simple biogeochemical models require lower levels of detail and thus fewer parameters than more complex biogeochemical models. Most of the existing sea-ice biogeochemical models feature only one group of algae resembling mostly diatoms (see Vancoppenolle and Tedesco, 2017, for a complete review of models). However, the sea-ice ecosystem is diverse and may not be represented realistically by a single group of algae. There have been few attempts to introduce biological diversity into sea-ice biogeochemical models, all by Tedesco et al. (2010, 2012, 2010). Their models include only two functional groups, distinguished by specific growth characteristics. The complexity of sea-ice models is further constrained by parameterization of biogeochemical processes. The major processes that can be defined in sea-ice algae models are nutrient uptake, primary production, respiration, lysis, exudation and predation (Tedesco and Vichi, 2014). Each process can be parameterized with a different level of complexity, from a simple linear equation to a complex set of equations that require numerical methods to solve.
There is an urgent need to improve biogeochemical sea-ice models, as our ability to predict ecological responses upon climate change is still limited: a comparison of models predicting change in Arctic primary production during the 21st century did not even agree on the sign of change (Vancoppenolle et al., 2013b). We need to find a proper balance between the level of detail required and the level of simplicity that is eventually adopted in these models. Finding this balance requires constraining the specific scientific question we want to answer with the computational resources that are available. If we want to consider the bulk properties of sea-ice Chl a on a scale that includes both poles, such as with a global biogeochemical model, the use of one generic group of sea-ice algae representing mostly fast-growing and high nutrient-demanding diatoms, might be a valid or useful approximation. If instead, we want to look at the regional carbon fluxes from sea ice, their fate and the sea ice–pelagic–benthic coupling, then more diversity in the composition of the biological community is desirable.
One major shortcoming to the possibility of increasing diversity in models is the limited data available for model calibration and validation. To fill this gap, in this paper we review and combine data on biodiversity and photosynthetic activity of sea-ice microalgae into a systematic framework. This paper does not contain an exhaustive list of the more than 1000 species of single-celled eukaryotes that have been reported in sea ice (Poulin et al., 2011); for a more detailed description of the composition of sea-ice algal communities, we refer the reader to the available extensive reviews (e.g., Garrison, 1991; Poulin et al., 2011). Likewise, reviews containing long lists of studies have been published on primary production and photosynthesis-irradiance relationships in polar regions (Cota and Smith, 1991; Legendre et al., 1992; Lizotte, 2001; Arrigo et al., 2010, 2014). In this paper, variables and parameters are summarized to expand the potential for increased model complexity. The aim is to update the available studies, to provide a more comprehensive overview of the composition and photosynthetic capacity of microalgae in sea ice, and to derive generic parameters that may facilitate the inclusion of further complexity in sea-ice biogeochemical models (Tedesco and Vichi, 2014; Steiner et al., 2016).
2. Data collection and analysis
In this work, we reviewed and combined published and unpublished data collected over 40 years in both the Arctic and Antarctic sea-ice regions (Table 1). Data from the sub-Arctic (e.g., Sea of Okhotsk and Baltic Sea) have not been included, as they comprise their own unique system.
|Type of data||Location||Month||Year||Ice typeb||Reference|
|Community composition||East Antarctica||11||1993||L||Archer et al., 1996|
|McMurdo Sound||10||1989||L||Arrigo et al., 1995|
|McMurdo Sound||11||2011||L||Carnat et al., 2014|
|Arctic Archipelago||5||2014||L||Campbell et al., 2017|
|Canadian Arctic||3||2008||P||Comeau et al., 2013|
|East Antarctica||10||2003||P||Dumont et al., 2009|
|Terre Adelie, Antarctica||5||1998||L||Fiala et al., 2006|
|Weddell Sea||11||1983||P||Garrison and Buck, 1989|
|Weddell Sea||2||1992||P||Gleitz and Thomas, 1993|
|Arctic Ocean||7||1994||P||Gosselin et al., 1997|
|Greenland Sea||5||1994||P||Gradinger et al., 1999|
|Canadian Archipelago||5||1972||L||Hsiao, 1980|
|Canadian Arctic||4||1998||L||Michel et al., 2002|
|Kobbefjord, Greenland Sea||11||2005||L||Mikkelsen et al., 2008|
|Canadian Arctic||6||2008||L||Mundy et al., 2011|
|Beaufort Sea||1||2008||L||Niemi et al., 2011|
|Arctic Ocean||7||2001||P||Ratkova and Wassmann, 2005|
|McMurdo Sound||1||2003||L||Remy et al., 2008|
|Terre Adelie, Antarctica||5||1995||L||Riaux-Gobin et al., 2003|
|Beaufort Sea||2||2004||L||Rozanska et al., 2009|
|Ross Sea||11||2003||P||Ryan et al., 2006|
|West Antarctic Peninsula||11||2014||P||Selz and Arrigo, unpubl|
|Arctic Ocean||6||2010||P||Selz and Arrigo, unpubl|
|Resolute Passage||4||1992||L||Sime-Ngando et al., 1997|
|McMurdo Sound||11||1995||L||Stoecker et al., 1998|
|Arctic Ocean||7||2003||P||Tamelander et al., 2009|
|East Antarctica||11||1996||L||Thomson et al., 2006|
|Marguerite Bay||12||2014||L||Van Leeuwe, unpubl|
|Weddell Sea||11||2004||P||Tison et al., 2010|
|Arctic Ocean||12||2003||P||Werner et al., 2007|
|Photosynthetic parametersa||McMurdo Sound||11||1988||L||Arrigo and Sullivan, 1992|
|Hudson Bay||4||1985||L||Barlow et al., 1988|
|Canadian Arctic||5||1986||L||Bergmann et al., 1991|
|Northwest Passage||4||1985||L||Cota and Horne, 1989|
|McMurdo Sound||11||1985||L||Cota and Sullivan, 1990|
|Arctic Ocean||8||2012||P||Fernandez-Mendez et al., 2015|
|Weddell Sea||10||1988||P||Gleitz and Kirst, 1991c|
|Hudson Bay||5||1983||L||Gosselin et al., 1986|
|Barents Sea||5||1988||P||Johnsen and Hegseth, 1991|
|Terra Nova Bay||11||1999||P||Lazzaro et al., 2007|
|Weddell Sea & Peninsula||3||1987||L||Lizotte and Sullivan, 1991|
|Terra Nova Bay||11||1997||P||Mangoni et al., 2009|
|Greenland Sea||5||1997||P||Mock and Gradinger, 1999d|
|McMurdo Sound||12||1983||L||Palmisano et al., 1985|
|McMurdo Sound||11||1987||L||Palmisano et al., 1987|
|McMurdo Sound||11||1988||L||Robinson et al., 1995|
|McMurdo Sound||8||1989||L||Robinson et al., 1998|
|West Antarctic Peninsula||11||2014||P||Selz and Arrigo, unpubl|
|Arctic Ocean||6||2010||P||Selz and Arrigo, unpubl|
|Resolute Bay||5||1988||L||Smith and Herman, 1991|
|Resolute Bay||5||1985||L||Smith et al., 1988|
|McMurdo Sound||11||1995||L||Stoecker et al., 2000|
|Resolute Passage||4||1992||L||Suzuki et al., 1997|
|East Antarctica||9||2015||P||Ugalde et al., 2016|
2.1. Microalgal community structure
The dataset on algal species composition consists of data that were collected from 32 regions in 280 ice cores, divided into 626 sea-ice sections (Figure 1; Table 1). We included only data that were based on a quantitative assessment of microalgae through microscopy, flow cytometry, FlowCam or Imaging Flow Cytobot.
A limited number of functional algal groups was defined according to their perceived importance related to an explicit role in biogeochemical cycles. The number of groups was constrained by their quantitative role in biogeochemical cycles and by the limited availability of data. The four following functional groups of algae were distinguished:
- pennate diatoms: often dominating algal biomass and therefore playing a major role in carbon fluxes;
- centric diatoms: typically less abundant, but often rich in carbon and relevant for carbon fluxes;
- autotrophic flagellates, including autotrophic dinoflagellates: occasionally highly abundant (e.g., in surface blooms) and important for their specific role in biogeochemical cycles (e.g., as producers of dimethylsulphide); and
- heterotrophic protists: specifically functioning as demineralizers that are important during periods of low light levels. This group includes heterotrophic (dino-) flagellates and ciliates. This group was not always defined in the literature, and therefore may be underestimated in this study.
In the analyses presented here, we have included only quantitative data based on cell counts, and thus some algal species may be underestimated. The progress made in molecular techniques has introduced new information on species abundance in sea ice (e.g., Bachy et al., 2011; Piwosz et al., 2013; Torstensson et al., 2015; Hardge et al., 2017). These new approaches are a major strength in studying smaller eukaryotic groups. As studies using these approaches are limited in number and generate a different type of data, we have not incorporated them into our analyses.
We did not discriminate between collection and processing techniques (e.g., method of fixation, type of microscope, cell retrieval; for discussion of melting techniques, see Rintala et al., 2014, and Miller et al., 2015). The data retrieved from the literature are expressed as cell numbers, or units of carbon or Chl a. In the Arctic, 95% of the data for landfast ice and 69% for pack ice are based on cell counts. In the Antarctic, these percentages are 72% for landfast ice and 30% for pack ice. The functional groups in this study are presented as percentages of the provided unit. Normalization to carbon biomass was not possible, as the majority of studies on community composition do not report carbon content and cell size. We note that presenting the functional groups as percentages makes generalizations, and that the various groups likely contribute differently to the total carbon inventory of sea ice.
Sub-ice colonial centric diatoms like Melosira arctica and Berkeleya adeliensis form strands and comprise a specific group of sea-ice algae. M. arctica is omnipresent in the Arctic Ocean (e.g, Poulin et al., 2014), while B. adelienesis is mainly associated with landfast ice in Antarctica (Riaux-Gobin et al., 2003; Belt et al., 2016). Both species may be of local importance in late spring when long strands of > 8m can be formed. Melosira strands can contribute over 85% of carbon export to the sea floor (Boetius et al., 2013). Because of the heterogeneous distribution and the limited amount of data available, strand-forming sea-ice diatoms were not included in our study.
2.2. Photosynthetic parameters
The photosynthetic parameter dataset consists of data collected from 23 regions in 90 ice cores, divided into 137 sea-ice sections (Figure 1; Table 1). Parameters were derived from photosynthesis-irradiance relationships as defined by Platt et al. (1980):
- the maximum photosynthetic capacity, Pmax (µg C µg Chl a–1 h–1);
- the slope of photosynthesis versus light, α (µg C µg Chl a–1 h–1 (µmol photons m–2 s–1)–1); and
- the index for photoadaptation, Ik (µmol photons–1 m–2 s–1),
where α is conventionally used as an indication for light affinity and Ik as an index for light saturation. Data were taken only from studies that established photosynthesis-irradiance relationships by means of 14C-incorporation in an effort to minimize variability associated with methodology. We did not discriminate among experimental conditions (e.g., extraction of samples from the ice, time of incubation). By combining all data, a certain measure of variability was introduced to the dataset. The choice of incubation technique (e.g., in situ versus in vivo) can have a strong effect on photosynthetic performance. Due to alterations in light and nutrient availability and variations in temperature, differences in photosynthetic parameters up to an order of magnitude may occur (Smith and Herman, 1991). Adequate tracer diffusion is another requirement for accurate analyses (Mock and Gradinger, 1999) and was not guaranteed in all studies. Furthermore, differences in the timing of incubations can potentially affect the results, as many microalgae have a diurnal rhythm. Diurnal cycles may affect microalgal photophysiology; however, various studies have shown that the diurnal signal in polar regions is of secondary importance relative to other driving parameters, given the reduced daily dynamics in polar algae that experience extended periods of daylight during spring and summer (e.g., Palmisano et al., 1987; Johnsen and Hegseth, 1991).
2.3. Data analysis
Three layers were distinguished for their different physical and biochemical properties:
- the infiltration/gap layer at the surface;
- the interior layer; and
- the bottom layer, varying in thickness from 0.01 to 0.10 m, depending on the method of sample collection.
These layers are partly isolated from each other, but also influence each other over the seasons. Each layer is affected in different ways by the various environmental parameters that shape sea-ice communities (Tedesco et al., 2012; Duarte et al., 2015). Consequently, each layer has its own characteristics, with a specific algal composition and specific photosynthetic capacities (Grossi and Sullivan, 1985; Manes and Gradinger, 2009).
Melt ponds form a separate habitat at the sea-ice surface. As very few quantitative data are available to allow for accurate parameterization, this habitat has not been distinguished from infiltration communities in our analyses. Given the potential importance of melt ponds, however, we have discussed the role of melt ponds specifically in the discussion. Pressure ridges, which may have the potential for large biomass accumulations (Horner et al., 1992), were also not incorporated. As these features are quite unique in their character, we chose not to merge the limited available data with the other communities.
Land fast ice was distinguished from pack ice because of differences in structure and hence habitat (as briefly discussed in Section 1.1). For similar reasons, the Arctic was separated from the Antarctic. A distinction can also be made between first year ice (FYI) and multi-year ice (MYI). MYI can reach a thickness of several meters, and tends to be more ridged. Several layers of microalgae may be incorporated (e.g., Lange et al., 2015; Werner et al, 2007). We did not distinguish FYI from multiyear ice MYI, however, because the limited dataset did not allow for proper statistical analyses.
Data were categorized per ice layer for statistical testing. To establish significant effects of hemisphere, ice type and season (analyzed as month-of-year for each hemisphere separately) on the microalgal community structure and photosynthetic parameters, and to establish interactions, data were analyzed by linear modeling in R (RStudio, 0.99.902). Within the datasets (Figures 2, 3, 4, 5, 6), significant differences were established by one-way ANOVA on ranks (Kruskal-Wallis), followed by Dunn’s Multiple Comparison test, assuming non-Gaussian distribution. Seasonality effects on community structure and photosynthesis were tested by Spearman rank correlation. A significance level of p < 0.05 was applied. Data are presented as mean ± standard error (SE), which best indicates the accuracy of a parameter as an estimate of the population mean.
3. Results and Discussion
Although sea ice is a very complex biome with many different micro-habitats subject to extreme variations over the seasons, the general patterns of algal distribution were found to be remarkably consistent, with clear patterns for the different layers (Section 3.1.). The succession of functional groups over time had a strong impact on community composition, especially as recorded in the bottom layer, and is discussed separately (Section 3.1.2.). Patterns for photosynthetic parameters were more difficult to structure than community composition (Section 3.2.), most likely because these parameters are subject to short-term environmental perturbations and therefore exhibit more variance (Section 3.2.1.). Overall, the data analysis presented in this paper confirm general concepts of distribution of functional groups and photosynthetic activity that previously had not been quantified.
3.1. General patterns in microalgal community structure
Different sympagic communities were observed to characterize each sea-ice layer. The most significant effects on the distribution of algal groups over the ice column were by the ice type (landfast or pack ice) and hemisphere (Table 2). Diatoms were the most susceptible to interactions of the various factors (Table 2).
|Algal group||Ice layer||Hemisphere (H)||Ice type (T)||H*T||Month (M)||M*T|
|Flagellates||Surface||n.s.a||p < 0.005||p < 0.05||n.s.||p < 0.0005||n.s.||p < 0.0005|
|Interior||p < 0.0005||n.s.||n.s.||n.s.||n.s.||p < 0.0005||n.s.|
|Bottom||p < 0.05||n.s.||n.s.||n.s.||n.s.||n.s.||n.s.|
|Pennate diatoms||Surface||n.s.||n.s.||p < 0.0005||n.s.||p < 0.0005||p < 0.05||p < 0.0005|
|Interior||p < 0.0005||p < 0.0005||p < 0.05||p < 0.005||p < 0.005||p < 0.0005||p < 0.0005|
|Bottom||p < 0.05||n.s.||n.s.||n.s.||n.s.||n.s.||p < 0.0005|
|Centric diatoms||Surface||n.s.||p < 0.05||p < 0.0005||n.s.||p < 0.0005||p < 0.05||p < 0.0005|
|Interior||p < 0.0005||p < 0.0005||p < 0.05||n.s.||n.s.||p < 0.0005||n.s.|
|Bottom||p < 0.005||p < 0.0005||n.s.||n.s.||n.s.||p < 0.0005||n.s.|
|Heterotrophs||Surface||n.s.||p < 0.005||n.s.||n.s.||n.s.||n.s.||n.s.|
|Interior||n.s.||p < 0.005||n.s.||n.s.||p < 0.05||p < 0.0005||p < 0.05|
|Bottom||p < 0.05||n.s.||n.s.||n.s.||n.s.||p < 0.05||n.s.|
The surface layer is dominated by flagellate species in both the landfast (about 60% of the whole community) and pack ice (about 45% of the whole community) datasets (Figure 2a, b). Light conditions can be extreme at the ice surface, and especially in spring UV-levels can be quite high (UVA > 5 Wm–2 and UVB > 0.35 Wm–2; Mundy et al., 2011). The surface dataset contains data collected largely from infiltration layers; only three of the included sites were referred to as having true surface melt ponds. These infiltration layers, which thus determine the signature of the surface layer, contain communities that are generally mixed (e.g., Garrison and Buck, 1989; Horner et al., 1992), though sometimes dominated by a single species, like the flagellate Phaeocystis antarctica (Garrison et al., 2005; Dumont et al., 2009). The interaction between hemisphere and ice type had a significant effect on the surface community structure (Table 2), which we attribute to snow-loading on pack ice in Antarctica, mostly characterized by infiltration communities, versus melt pond communities more frequently observed in the Arctic.
The limited dedicated (Arctic) studies on melt ponds show that the species distribution in ponds is more homogeneous than in infiltration layers, especially early in the melt-pond season, with observations of the freshwater alga, Chlamydomonas nivalis (Melnikov et al., 2002; Lee et al., 2011), and of Pyramimonas sp. (e.g., Mundy et al., 2011). The absence of regular intrusions of fresh seawater supply might well promote the dominance of a single species (Stoecker et al., 1992). The relatively strong contribution of flagellates in the ponds is noteworthy, as it contrasts with the pelagic system, where the paradigm is that diatoms can better tolerate high levels of irradiance than flagellates (Richardson et al., 1983). Melt ponds change in character later in the season, when brine channels gradually open up during melt (Mundy et al., 2011). Mixed communities may develop then, benefitting from a renewed nutrient supply from below. Strong melt may open the connection between surface melt ponds and seawater below the ice, creating open saline ponds. Communities in these ponds are similar to those in seawater (Lee et al., 2012).
The interior layer consists of more mixed communities, with a slight prevalence of flagellates and pennate diatoms in both landfast and pack ice (Figure 2c, d), with various taxa including Nitzschia sp., Navicula sp., Pyramimonas sp. and Gymnodinium sp. (e.g., Mikkelsen et al., 2008). These communities can originate from algae trapped in the sea ice during ice formation that continue growing. The interior layer shows the most significant difference between the Arctic and Antarctic and between ice type (Table 2). In the ice interior, the distribution of pennate diatoms is quite different for pack ice versus landfast ice (Table 3). In the Arctic, pennate diatoms dominate with a 63% relative abundance in interior communities in landfast ice; in pack ice they make up only 7% (Table 3). This pattern is reversed for Antarctic interior layers, with 10% pennate diatoms in landfast ice versus 36% in pack ice (Table 3). The interaction of ice type and time of year had a strong impact on the community structure in the ice interior (Month * Ice type; Table 2; Section 3.1.2.), which may partly explain this hemispheric difference.
|Type of ice||Ice layer||Hemisphere||Flagellates||Pennate diatoms||Centric diatoms||Heterotrophs|
|Landfast ice||Surface||Arctic||40 (6, 0, 100)a||39 (6, 0, 91)||15 (2, 0, 57)||3 (2, 0, 40)|
|Antarctic||76 (5, 0, 100)||3 (1, 0, 47)||1 (1, 0, 41)||3 (1, 0, 47)|
|Combined||63 (4)b||16 (3)||6 (1)||3 (1)|
|Interior||Arctic||22 (3, 0, 100)||63 (4, 0, 100)||14 (2, 0, 90)||1 (1, 0, 5)|
|Antarctic||65 (4, 1, 100)||10 (2, 0, 73)||1 (1, 0, 47)||5 (1, 0, 60)|
|Combined||48 (3)||31 (3)||6 (1)||3 (1)|
|Bottom||Arctic||15 (2, 0, 100)||65 (3, 0, 100)||18 (2, 0, 72)||2 (1, 0, 40)|
|Antarctic||31 (5, 0, 98)||28 (5, 0, 100)||4 (2, 0, 52)||10 (3, 0, 91)|
|Combined||21 (2)||51 (3)||12 (1)||5 (1)|
|Pack ice||Surface||Arctic||66 (12, 0, 100)||1 (1, 0, 1)||1 (1, 0, 1)||15 (6, 0, 40)|
|Antarctic||43 (5, 0, 97)||19 (4, 0, 98)||14 (3, 0, 84)||11 (3, 0, 88)|
|Combined||46 (4)||16 (3)||12 (3)||11 (3)|
|Interior||Arctic||20 (3, 0, 100)||7 (3, 0, 91)||1 (1, 0, 11)||15 (3, 0, 100)|
|Antarctic||54 (4, 0, 100)||36 (4, 0, 86)||4 (2, 0, 47)||1 (1, 0, 20)|
|Combined||34 (3)||20 (3)||2 (1)||9 (2)|
|Bottom||Arctic||21 (3, 0, 68)||32 (6, 0, 93)||5 (2, 0, 47)||16 (4, 0, 74)|
|Antarctic||16 (7, 0, 99)||83 (7, 4, 100)||1 (1, 0, 14)||6 (4, 0, 73)|
|Combined||16 (3)||55 (5)||3 (1)||8 (2)|
Pennate diatoms are observed to dominate the bottom layers (Figure 2e, f). The most ubiquitous species are Fragilariopsis cylindrus (formerly Nitzschia cylindrus) and N. frigida (Horner and Schrader, 1982; Rozanska et al., 2009; Leu et al., 2015). Pennate diatoms are fast-growing and probably the most efficient at nutrient utilization under low light levels (Hegseth, 1992; Gradinger et al., 1999; Lizotte, 2001). The Arctic and Antarctic bottom communities again are quite different in landfast ice, with significant differences in the relative abundance of flagellates in the Antarctic (31%) compared to the Arctic (15%) (Table 3).
Heterotrophic protists (or protozoa) do not appear as a dominant group in our datasets (Figure 2; Table 3). The relative abundance of heterotrophic protists is greatest in early spring (Figure 3e), when light availability is still low and autotrophs are at low abundances (see also Rozanska et al., 2009). Numbers of protozoa are likely often underestimated in field studies. They may be overlooked in microscopic analyses because they are small and do not show up brightly as they often lack chlorophyll. In fact, phylogenetic analysis has revealed the dominance of a heterotrophic dinoflagellate (SL163 A10 clade) in summer interior sea ice in the Amundsen and Ross seas (Torstensson et al., 2015). Heterotrophic protists as clustered in this review represent a mixed group of species. Many of the protozoa (flagellates as well as ciliates) do not necessarily have a fully heterotrophic lifestyle. Some species are autotrophic or mixotrophic (Torstensson et al., 2015; Caron et al., 2017, and references therein). Phylogenic analysis of protists in sea ice has revealed a wide variety of mixotrophic taxa at the end of the Arctic winter (Bachy et al., 2011; Paterson and Laybourn-Parry, 2017). These protists feed mainly on organic detritus or bacteria, rather than other microalgae (Michel et al., 2002). As such, they contribute to the remineralization of nutrients, specifically nitrate and phosphate (Arrigo et al., 1995). The remineralization of silicate is much slower (Cota et al., 1991). As nutrient concentrations in the semi-enclosed ice interior drop with algal consumption, remineralization activity by heterotrophic flagellates plays an important role locally (e.g., Tamelander et al., 2009). In pack ice, heterotrophic protists are more abundant in the Arctic than the Antarctic (e.g., Sime-Ngando et al., 1997; Table 3). This comparatively high abundance is possibly related to the more extended period of darkness at high Arctic latitudes, and may also be linked to greater concentrations of dissolved organic matter that support active microbial foodwebs (Meiners and Michel, 2017).
Defining the importance of sea ice in seeding the pelagic community is difficult, and very few dedicated studies are available. Species identified in sea ice have often been found to be the same as the ones in the water column (e.g., Horner and Schrader 1982; Mundy et al., 2011), yet the opposite has also been observed (e.g., Riaux-Gobin et al., 2003; Barber et al., 2015). Even if species in the pelagic system are similar to the sympagic community, the seeding capacity of sympagic microalgae is hard to demonstrate. A microscopic study in the Beaufort Sea by Horner and Schrader (1982) showed that many of the cells released from ice were unhealthy. A brief pulse in pelagic production was recorded, but the bloom did not last long. The fate of sympagic algae might depend on the timing and size of the pulse released (Tedesco et al., 2012; Selz et al., 2017). In two consecutive seasons in the high Arctic, an early and rapid release of sympagic algae initiated a pelagic bloom, yet in another season when the release of sympagic microalgae occurred only towards summer, no such effect was recorded (Galindo et al., 2014). This difference in fate may be related to the physiological state of the cells when they are released from the sea ice. Export of sympagic algae into deeper waters has been related to aggregation of cells and the production of extracellular polysaccharide substances (EPS; Riebesell et al., 1991). The importance of sea-ice algae in seeding the water column requires more dedicated studies that focus on early spring.
3.1.1. (Dis)similarities in the biodiversity of microalgae in various ice types
Sympagic microalgae show a high degree of biodiversity. Pennate diatoms are the most diverse group: 446 pennate versus 122 centric species have been recorded (Arrigo, 2014). Maximum biodiversity is observed in pack-ice interior layers (Stoecker et al., 1992; von Quillfeldt et al., 2003). The relatively porous structure of pack ice is influenced by infiltration of seawater, occasionally bringing in new species (Syvertsen and Kristiansen, 1993). This layer is also the layer with the most variable and extreme conditions, which hints at an evolutionary adaptation to the various niches in sea ice. In the Canadian Arctic, only 27 diatom species were recorded in FYI versus 55 diatom species in MYI (Melnikov et al., 2002), the latter providing a potentially more stressful habitat due to lower levels of nutrients. Likewise, a phylogenetic study in the central Arctic Ocean revealed the highest biodiversity in MYI (Hardge et al., 2017). Successful survival of microalgae in the heterogeneous sea-ice habitat is apparently based on evolutionary adaptation and high plasticity (Sackett et al., 2013).
Although sympagic communities show a high biodiversity, there is also overlap between species found in landfast and pack ice. Some species, like Polarella glacialis (Montresor et al., 2003; Thomson et al., 2006) and Fragilariopsis cyclindrus (Roberts et al., 2007, and references therein; Poulin et al., 2011), are distributed in both polar regions, which supports the hypothesis that many ice-algal species living in sea ice are in fact pelagic phytoplankton species (see Table 2 in Garrison, 1991, for an extensive species list). A similar correspondence in species composition at the genus level has been observed between pack ice in the Arctic and the Antarctic, with the genera Fragilariopsis, Nitzschia and Navicula illustrating the cosmopolitan characteristics of many microalgal genera (McMinn and Hegseth, 2003). Similarities were also found when comparing FYI with MYI (e.g., Melnikov et al., 2002).
Comparing landfast and pack-ice surface communities, the composition of heterotrophic protists also appears remarkably similar in two separate and remote sites in Antarctica (Archer et al., 1996). Whereas commonly observed choanoflagellates were considered less abundant, the heterotrophic nanoflagellate Cryothecomonas sp. appears to be a ubiquitous species in the Antarctic (see Archer et al., 1996, and references therein).
3.1.2. Seasonal variation
The data presented in Figure 2 and Table 3 are average values, without discrimination over time. However, significant correlations were established between microalgal functional groups and time of year, especially in Antarctica (Table 4; Figure 3a–f). The interior layer appears as the most susceptible to seasonal variation, dictated by changes in the physicochemical environment related to ice formation. The phenology of sea-ice communities has been described in only a few papers (Rozanska et al., 2009; Leu et al., 2015; Campbell et al., 2017a). In our datasets, the patterns are hard to retrieve (Figure 3). The data are too scattered in time and space, and seasonal patterns can be related to specific phenomena in ice formation. A schematic depiction of seasonal fluxes is presented in Figure 4.
|Type of data||Arctic landfast ice||Arctic pack ice||Antarctic landfast ice||Antarctic pack ice|
|Algal groups||Flagellates||p < 0.05||n.s.a||p < 0.05||p < 0.05|
|Pennate diatoms||n.s.||n.s.||p < 0.05||p < 0.05|
|Centric diatoms||p < 0.05||p < 0.05||p < 0.05||n.s.|
|Heterotrophs||n.s.||n.s.||n.s.||p < 0.05|
|Photosynthetic parameters||Pmax||n.s.||p < 0.05||n.s.||n.s.|
Previous studies have suggested that in early spring, the sea-ice community consists mainly of pennate diatoms (e.g., Leu et al., 2015). This pattern is indeed shown for the Arctic data compiled here, with high pennate diatom abundance in May (Figure 3a, b). The dominance of pennate diatoms in the Antarctic is more pronounced in pack ice than in landfast ice (Figure 3c, d). In the Arctic (e.g., Mikkelsen et al., 2008) and Antarctic (e.g., Stoecker et al., 1998), flagellates can also be present, although seasonal patterns of flagellates are less clear (Figure 3a–d). High abundance of centric diatoms in spring is most obvious in Arctic landfast ice, and less so in the Antarctic and in pack ice (Figure 3a–d). Notably, early spring may witness a succession of diatom species; in the Arctic, centric diatoms can become more abundant than pennate diatoms with the progressive increase of light availability (Melnikov et al., 2002; Campbell et al., 2017a, and references therein). This phenomenon may be related to high resistance to UV-radiation (Karentz et al., 1991) and high nutrient affinity (Campbell et al., 2017a) in centric species.
The composition of the spring community is largely determined by species that survive the winter (Tedesco et al., 2010), often in the ice interior. The winter algal community originates in autumn, when algae are trapped during ice formation. They may continue to grow, encapsulated in the grazer-protected ice environment (Garrison et al., 1983), despite the fact that over winter temperature drops and salinity increases. Newly formed sea ice may contain an algal community with high biodiversity (Gleitz and Thomas, 1993). Larger cells, especially EPS-producing species, are likely the most easily scavenged by the rising ice crystals that lead to ice formation and therefore tend to dominate in young sea-ice communities (Syvertsen, 1991; Figure 4). Larger pennate diatoms are scavenged preferentially (Garrison et al., 2005; Rozanska et al., 2008), forming occasional autumn sea-ice blooms (Fiala et al., 2006). Besides pennate diatoms, centric diatoms (Hsiao, 1980; Gleitz et al., 1998; Garrison et al., 2005) and flagellates (Gradinger and Ikävalko, 1998; Rozanska et al., 2008) have also been observed in newly formed sea ice.
When the ice column thickens, due to ice growing at the bottom or to ice rafting, interior layers are formed (Grossi and Sullivan, 1985; Ackley and Sullivan, 1994). The zonation of microalgae within the ice column is partly the result of a passive process, as algae are simply trapped in position even as the ice continues to grow. Pennate diatoms, however, are also known to be motile and can migrate within the ice column along favorable physicochemical gradients (Grossi and Sullivan, 1985; Aumack et al., 2014). The confined conditions within the ice may involve morphological adaptation (Ratkova and Wassmann, 2006) and stimulate selection for smaller species (Gleitz and Thomas, 1993; Krembs et al., 2000). Naviculoid diatoms like Navicula glaciei, observed to be the most motile (Grossi and Sullivan, 1985), and are also the most likely species to occur in interior layers (Figure 4).
Over winter, acclimation and natural selection occurs. The few winter data that were available to include in our dataset (Arctic: November, December; Antarctic: May–August) show high abundances of flagellates as well as mixed diatom communities (Figure 3a–d). Various adaptations are required to endure the extreme temperature, light and salinity conditions characteristic of the winter season. Energy storage, mixotrophy and heterotrophy may be the most important traits when photosynthetic activity is reduced (Zhang et al., 1998). Secondarily, the production of EPS, particularly by pennate diatoms, can aid their incorporation into the ice and subsequent persistence by altering the physical structure of sea ice (Rozanska et al., 2008; Krembs et al., 2011). Ice-binding proteins and cryo-osmolytes such as dimethylsulphonioproprionate (DMSP) produced by a number of psychrophilic microalgae may also offer protection against low temperatures and high salinity (Stefels, 2000; Krell et al., 2008). Finally, resting spore and cyst formation are often mentioned as important traits that enable microalgae to survive the winter. Microalgae may survive the dark period by formation of resting cells or spores that can be used for short-term dormancy in the case of diatoms or of cysts that often require a dormancy period before the germination in the case of dinoflagellates. Such spore or cyst formation, however, remains a poorly characterized trait of sea-ice algae. Various microalgal species have been suggested to form resting stages, but not all of them are prominent ice-algal species (see Appendix in Quillfeldt et al., 2003). Further investigation into resting stages may reveal not only more about survival strategies, but also the link between the sea-ice and pelagic biomes (Rintala et al., 2007).
The community composition in winter can be quite diverse, with up to 140 taxa reported (Werner et al., 2007; Niemi et al., 2011). Pennate diatoms are the most adapted to conditions in winter, specifically the more specialized species, like Nitzschia frigida (Syvertsen, 1991; Gradinger et al., 1999; Leu et al., 2015). An increasing number of observations, however, has made clear that besides pennate diatoms, flagellates also subsist through the winter (Ikävalko and Gradinger, 1997; Stoecker et al., 1998). Fewer observations are available for centric diatoms. They may survive only in the bottom layers (Kottmeier and Sullivan, 1987), where relatively high abundances have been observed in early spring in Arctic landfast ice (Figure 3a). In the ice interior, however, centric diatoms observed in high abundance in young pack ice (Figure 3d) apparently disappear from the ice column over time (Gleitz et al., 1998; Ratkova and Wassmann, 2006; Olsen et al., 2017). Heterotrophic protists are present throughout the year in pack ice (Figure 3f). In landfast ice, highest abundances have been recorded in early spring in the Antarctic. The relative abundance of protozoa declines when brine channels open up and fresh nutrients are provided, initiating the ice-algal spring bloom (Archer et al., 1996).
When ice-melt starts in spring, the sub-ice community of diatoms is the first to be sloughed off the bottom of the sea ice (Figure 4). Generally, flagellate species (autotrophic and heterotrophic) persist longer in sea-ice bottom communities (e.g., Tamelander et al., 2009; Torstensson et al., 2015). Nutrient limitation towards the end of sympagic blooms can also explain a dominance of flagellates later in the year (Figure 3d). Whereas nitrate gets depleted before silicate in the sea-ice interior (Fripiat et al., 2017), in bottom layers silicate often becomes the first limiting nutrient (Riaux-Gobin et al., 2003). In the final phases of the sea-ice cover, flagellates may thus contribute significantly to carbon fluxes (Tamelander et al., 2009).
3.2. Primary production
The photosynthetic parameters compiled in this study are highly variable (see Section 3.2.1.), and fewer statistically significant differences were detected (Figure 5) than for community composition. Hemisphere had slightly more significant impact on photosynthesis than did ice type (Table 5). Only bottom layers in the Arctic were susceptible to the interaction of ice type and time-of-year (Table 5).
|Parameter||Ice layer||Hemisphere (H)||Ice Type (T)||Month (M)||M*T|
|Bottom||n.s.||p < 0.005||n.s.||p < 0.005||n.s.|
|α||Surface||p < 0.05||n.s.||n.s.||n.s.||n.s.|
|Bottom||n.s.||p < 0.0005||n.s.||p < 0.0005||n.s.|
|Ik||Surface||p < 0.05||p < 0.05||n.s.||n.s.||n.s.|
Surface communities showed high photosynthetic capacity, Pmax (Figure 5a, b), with highest average values recorded in Arctic pack ice (Table 6). At the same time, maximum values for light saturation, Ik, were measured at the surface, with average values of almost 100 µmol photons m–2 s–1 recorded in Arctic pack ice (Figure 5e, f; Table 6). These high values for Ik confirm that the communities were adapted to high light. Only one study provided quantitative data on photosynthesis-irradiance relationships in melt ponds. The Pmax of 2.8 µg C µg Chl a–1 h–1, an α of 0.05 µg C µg Chl a–1 h–1 (µmol photons m–2 s–1)–1 and Ik of 139 µmol photons m–2 s–1 are worthwhile mentioning (Fernandez-Mendez et al., 2015), as the Ik is the highest in our dataset. Melt-pond communities appear well adapted to high light. Production, however, may be inhibited by depletion of nitrate and phosphate (Stoecker et al., 2000). An increase in nutrient limitation can be reflected in a seasonal decline in Pmax and α (Stoecker et al., 2000; Fernandez-Mendez et al., 2014). Later in the season, when melt ponds can become connected to the ice interior and even to underlying seawater, a fresh supply of nutrients may be provided (e.g., Mundy et al., 2011). Production in infiltration layers is seldom inhibited by nutrients, as they are regularly supplied by flooding seawater (Fripiat et al., 2017). Melting and freezing processes in sea ice, however, can affect nutrient supply (Fripiat et al., 2017). In MYI, surface drainage through brine channels and subsequent refreezing results in nutrient depletion in surface layers. Alternatively, nutrients that are captured in organic material will be retained upon refreezing (Fripiat et al., 2017). Altogether, melting and freezing processes make it difficult to link nutrient availability to any specific ice condition (Vancoppenolle et al., 2013a; Meiners and Michel, 2017).
|Type of ice||Ice layer||Hemisphere||Pmax||α||Ik|
|Landfast ice||Surface||Arctic||0.83 (0.15, 0.2, 1.5)a||0.048 (0.015, 0.018, 0.134)||32.4 (8.6, 11, 75)|
|Combined||0.76 (0.15)c||0.044 (0.014)||29.0 (8.2)|
|Antarctic||0.10 (0.02, 0.06, 0.14)||0.018 (0.010, 0.004, 0.038)||8.9 (3.2, 2.8, 13.5)|
|Combined||0.10 (0.02)||0.018 (0.010)||8.9 (3.2)|
|Bottom||Arctic||1.25 (0.35, 0.19, 5.20)||0.189 (0.033, 0.014, 0.45)||7.4 (1.7, 1.4, 21.9)|
|Antarctic||0.16 (0.02, 0.08, 0.27)||0.021 (0.003, 0.007, 0.037)||7.9 (1.1, 4.2, 13.4)|
|Combined||0.89 (0.25)||0.133 (0.027)||7.6 (1.1)|
|Pack Ice||Surface||Arctic||2.00 (0.80, 1.20, 2.80)||0.040 (0.010, 0.030, 0.050)||98.5 (40.5, 58, 139)|
|Antarctic||1.12 (0.18, 0.04, 3.14)||0.021 (0.003, 0.005, 0.074)||57.4 (7.1, 3, 134)|
|Combined||1.28 (0.17)||0.022 (0.003)||57.4 (6.8)|
|Interior||Arctic||0. 07 (0.03, 0.01, 0.16)||0.037 (0.032, 0.005, 0.100)|
|Antarctic||0.37 (0.08, 0.03, 1.20)||0.007 (0.002, 0.001, 0.030)||71.1 (13.2, 14, 274)|
|Combined||0.31 (0.07)||0.012 (0.004)||71.1 (13.2)|
|Bottom||Arctic||0.15 (0.02, 0.02, 0.39)||0.011 (0.003, 0.002, 0.100)||27.8 (5.0, 6.7, 111.5)|
|Antarctic||0.32 (0.17, 0.02, 2.17)||0.031 (0.022, 0.001, 0.268)||25.5 (6.3, 7, 91)|
|Combined||0.20 (0.05)||0.016 (0.007)||27.0 (3.9)|
Surface blooms are often terminated by the deleterious effects of decreasing salinity and increasing irradiance in the course of spring (e.g., Campbell et al., 2014). Various adaptive mechanisms towards exposure to high irradiance and UV have been recorded, including adjusting pigment composition or the production of mycosporine-like amino acids (Mundy et al., 2011; Kauko et al., 2016). However, it is the low salinity shock in combination with high irradiance levels that especially suppresses algal growth (Cota and Smith, 1991; Ralph et al., 2007; Ryan et al., 2011). Conditions appear more severe for landfast than for pack-ice communities (Figure 5e, f). For instance, a relatively low Ik of < 30 µmol photons m–2 s–1 was determined for landfast ice at McMurdo Sound, which probably related to overexposure to high irradiance and ensuing strong downregulation of photosynthesis (see also Palmisano et al., 1987). The difference between the two ice types may be related to differences in snow cover, but not enough data are available to confirm this relationship.
Very few data exist on production by interior communities. For landfast ice, only one dataset was available, with Pmax values of < 0.1 µg C µg Chl a–1 h–1 recorded at McMurdo Sound (Palmisano et al., 1987). This maximum photosynthetic capacity is less than 10% of the average production rates in surface and bottom sea ice (Figure 5a, b). Interior communities are subject to extreme conditions of salinity and temperature, which suppress photosynthetic activity (Palmisano et al., 1987; Gleitz and Kirst, 1991; Arrigo and Sullivan, 1992). These communities are also the most subject to temporal changes in sea-ice structure and biogeochemistry. In FYI, production may also be controlled by nutrient limitation, but this limitation is less likely to occur in the ice interior where nutrients are recycled rapidly and therefore in ample supply (Gleitz et al., 1995; Fripiat et al., 2017; Roukaerts et al., 2016). Algal biomass in the ice interior can build to concentrations of 193 µg C L–1, accounting for 50% of total biomass (Archer et al., 1996). Production to such levels, however, ultimately leads to nutrient depletion, as recorded in the Canadian Arctic where silicate and phosphate values in both FYI and MYI did not exceed 0.1 µmol l–1 (Melnikov et al., 2002). Production rates in MYI have been recorded as lower than in FYI, but other factors contribute to this difference, particularly light, salinity and space limitations (Mock and Gradinger, 1999). Lower production in older ice will result in a shift of the system towards heterotrophy, as observed in comparative bottom ice communities of McMurdo Sound (Remy et al., 2008). Though production rates may be low, integrated over the ice column, production may be a factor of significance, especially in Antarctic pack ice (Table 6) but also in the Arctic (Mock and Gradinger, 1999), and therefore should not be ignored in biogeochemical models.
The highest value for Pmax of 5.2 µg C µg Chl a–1 h–1 was recorded in a bottom community in Arctic landfast ice (Gosselin et al., 1986). The high Pmax values in the bottom layer of our complied datasets (Figure 5a) indicate optimal photoacclimation. Shade adaptation by the bottom communities is also reflected by very low Ik values, with lowest values just over 1 µmol photons m–2 s–1 recorded in early spring in Arctic landfast bottom communities (Suzuki et al., 1997). Overall a 60–70% reduction in Ik was recorded for the bottom layers versus the surface (Figure 5e, f). Shade adaptation will also shape α, the index of light affinity; however, data were highly variable and significant differences between the various layers could not always be detected (Figure 5c, d). A minimum α of 0.0012 µg C µg Chl a–1 h–1 (µmol photons m–2 s–1)–1 was recorded in bottom pack-ice communities in East Antarctica (Ugalde et al., 2016). Low values for α are the result of high Chl a/carbon ratios that are typical under shade adaptation (see Johnsen and Hegseth, 1991, and references therein). Light affinity α is ten times lower in pack-ice bottom communities compared to landfast communities (Figure 5a–d). This difference can be attributed to relatively low light levels under pack ice, as a result of various factors such as snow cover and ice thickness (Lazzara et al., 2007; Ugalde et al., 2016; Arndt et al., 2017) as well as self-shading (Johnsen and Hegseth, 1991).
The average value for α of 0.13 µg C µg Chl a–1 h–1 (µmol photons m–2 s–1)–1, as established for landfast bottom communities (Figure 5c), is near the proposed theoretical maximum (see Cota and Smith, 1991, and references therein). This average indicates relatively high growth efficiency under the ambient light conditions. It is an order of magnitude higher than values for α generally observed in natural communities of polar phytoplankton (see also Cota and Smith, 1991, for comparison), which confirms the high capacity of photoacclimation in sympagic microalgae.
The relatively high Pmax of communities in the bottom layer of landfast ice is most likely linked to differences in nutrient availability. Landfast ice forms in coastal and shelf areas, where current regimes are often more dynamic (Cota et al., 1991) and nutrient concentrations in the water are often higher than in open ocean areas (see also Gradinger, 2009, and references therein). A tight relationship between increased production and enhanced nutrient availability as provided by tidal currents was recorded in the Arctic Archipelago (Cota and Horne, 1989). In the Antarctic, platelet-ice systems appear especially productive, partly because their porous structure facilitates high rates of nutrient exchange (Grossi et al., 1987; Arrigo et al., 1995). In contrast, pack ice in the Arctic often forms over open ocean areas that are generally low in nutrients (e.g., Lee et al., 2010; Fernandez-Mendez et al., 2015). In addition, stagnation of currents underneath pack ice can result in stratification that inhibits the nutrient exchange between deeper layers and surface waters (Cota et al., 1991; Barber et al., 2015). Especially in the Arctic, a strong halocline is persistent (Cottier et al., 2017), which results more often in nutrient limitation within the sea ice.
Comparing landfast bottom communities in the Arctic and Antarctic confirms that conditions other than ice type control production capacity. Photosynthesis rates in the Arctic are on average ten times higher than in the Antarctic (Table 6). These differences can be linked directly to the local light climate, as studied closely in the Antarctic by Lizotte and Sullivan (1991). They compared pack-ice with landfast ice communities, observing higher values for Pmax in the pack ice, and related this difference to higher irradiance levels under the pack ice. Additional research on Antarctic pack ice led to the same conclusion: light availability more so than structure of the sea ice made the difference between low and high production rates (Ugalde et al., 2016).
3.2.1. Acclimation to variations in light and temperature
Whereas significant patterns can be established for the microalgal community composition, identifying similar common regularities for photosynthetic parameters is difficult (Table 6). The only significant effect recorded for time of year, for example, was on Pmax in Arctic pack ice (Table 4). Photosynthetic parameters are highly variable, and variations of an order of magnitude within one dataset are not exceptional (e.g., Lizotte and Sullivan, 1991). Values for Pmax can vary by >200%, variations that are larger than the average estimates presented in Figure 5. In sea ice, small-scale differences in structure are associated with large variations in environmental factors like temperature, salinity and light, accounting for the heterogeneity of the habitat. These small-scale variations in space and time are the features that control algal physiology and biomass (e.g. Lizotte and Sullivan, 1991), more than general characteristics that define sea ice as either landfast or pack ice.
The most dominant factor that governs photosynthesis is light. The light climate in sea ice is controlled to a large extent by ice thickness and snow (Mundy et al., 2005, 2007; Arndt et al., 2017; Perovich, 2017). A thick snow cover prevents light penetration into the sea ice, which can result in light limitation of sympagic microalgal communities (Arrigo, 2014, 2017). Seasonal changes in photosynthetic activity due to changes in irradiance are in fact observed (e.g., Gosselin et al., 1986; Hawes et al., 2012; Campbell et al., 2017b), but they are difficult to grasp in meta-analyses. Environmental conditions that determine the onset of the spring bloom are too site-specific.
Photoacclimation and associated changes in photosynthetic parameters may occur on a time scale of days, i.e., the generation time of a community, to weeks (Grossi et al., 1987; Cota and Horne, 1989; Cota and Smith, 1991; Juhl and Krembs, 2010). Fluorescence analyses have clearly shown how local photosynthetic activity can be triggered in spring by increasing temperature (Hawes et al., 2012) and improvement of the light climate (Manes and Gradinger, 2009; Campbell et al., 2017b). Similarly, a 3–5-fold increase in Pmax and α was recorded in response to gradual snow thinning over the course of two months (Gosselin et al., 1986; Barlow et al., 1988). Gradients can be observed not only in time but also in space. In the ice column, vertical gradients in photosynthetic activity were recorded by fluorescence analyses at Point Barrow, Alaska (Manes and Gradinger, 2009). The highest activity was measured in bottom communities. Vertical patterns became stronger with increasing photosynthetic activity over the course of time.
Snow clearing that improves light penetration into sea ice does not always stimulate microalgal growth. Differences in light history and site-specific variations in the light climate introduce further scatter in a meta-analyses as presented here. For instance, sudden exposure to increased light intensities can trigger migratory movement away from the light (Aumack et al., 2014; Lund-Hansen et al., 2014). Rapid removal of snow can also have more deleterious effects. In a number of experiments, snow clearance resulted in a decline in photosynthetic activity and a decrease in biomass as the change in light climate was apparently too abrupt for an adequate response (McMinn et al., 1999; Juhl and Krembs, 2010). A sudden change in photosynthetic parameters in bottom layers can also be due to mechanical damage and losses associated with a rapid deterioration of the sea ice habitat induced by snow melt and increased temperatures (Smith et al., 1988; Mundy et al., 2005; Campbell et al., 2014, 2015). Recent studies on Arctic landfast ice have shown a seasonally changing influence of snow cover on ice-algal bottom communities (Mundy et al., 2005; Campbell et al., 2014, 2015; Leu et al., 2015) and demonstrated how snow dynamics can alter the timing, duration and magnitude of ice-algal spring blooms. Similar concepts have recently been postulated for Antarctic pack ice (Meiners et al., 2017), but a comparable snow-driven ice-algal phenology is currently lacking for Antarctic fast ice.
In sea ice, both light and temperature influence algal growth. The strong ability of photoacclimation in sympagic algae does not necessarily imply that growth is efficient. Sympagic microalgae perform rather poorly in terms of biomass-specific production. Pmax in sympagic algae rarely exceeds 1 µg C µg Chl a–1 h–1. Phytoplankton may produce more than 5 µg C µg Chl a–1 h–1 (Cota and Smith, 1991). Sympagic microalgae appear specifically adapted for growth at low temperatures (Kottmeier and Sullivan, 1988). Sympagic algae from polar sea ice have growth optima between –0.5°C and 4°C (Rochet et al., 1985; Palmisano et al., 1987; Cota and Smith, 1991), which implies that sympagic microalgae often grow under suboptimal (colder) conditions, as confirmed by fluorescence and oxygen-based production studies. In studies at Cape Evans (Antarctica), low production efficiency was observed, which according to the authors indicated that the landfast sea-ice algae were not functioning at maximum capacity (Robinson et al., 1998; McMinn et al., 2003).
Despite these apparent shortcomings, sympagic algae in the sea-ice interior survive under conditions that temperate species could not endure. The polar winter allows only for very low photosynthetic rates (Lizotte, 2001; Berge et al., 2015). Growth can be maintained because respiration rates are equally low due to the low temperatures. With increasing temperatures, respiration rates increase faster than photosynthetic rates (Tilzer and Dubinsky, 1987; Thomas et al., 1992; Regaudie-de-Gioux and Duarte, 2012). When metabolic costs increase with the rise in temperature in spring, a positive growth balance can still be maintained because carbon assimilation increases even more as a consequence of the simultaneous extension of daytime (Tilzer and Dubinsky, 1987). It is because of this relative high temperature sensitivity that sympagic microalgae can survive the polar winter.
4. A future perspective on climate change and sea-ice algae
In the past three decades, the summer minimum sea-ice extent in the Arctic has decreased by 45% (Arrigo, 2014; Thomas, 2017, and references therein). In 1991, the sea ice cover in the Arctic would range from 8.5 × 106 km2 in September to a maximum of 15 × 106 km2 in March (Cota et al., 1991). In 2014, the ice cover in the Arctic varied between 5 × 106 and 15 × 106 km2 (Arrigo, 2014). Associated with this decrease in ice cover, the sea ice is thinning, and a marked shift has already occurred with FYI replacing MYI (Meier, 2017; Stammerjohn and Maksym, 2017). The Arctic is also likely to witness a reduction of landfast sea ice (Tamelander et al., 2009). Changes in the Antarctic are more variable. The Western Antarctic Peninsula has shown a 41% decline relative to the mean sea-ice extent over the period 1979–2007 (Ducklow et al., 2012). On the other hand, sea-ice extent in East Antarctica has not changed over the last several decades. Overall the Antarctic sea-ice extent has slightly increased over the satellite observing period, but this trend is masking strong regional changes in extent and duration of the sea-ice cover (Stammerjohn and Maksym, 2017).
Changes in the extent and structure of the sea ice will have consequences for sympagic communities. So far, the data have been too sparse geographically to record broadly significant changes, but a number of papers have summarized the potential effects of climate change (e.g., Post et al., 2013; Barber et al., 2015). In the Arctic, thinning of the sea ice is a factor that can already be recognized as affecting sympagic algae. Bottom communities develop earlier in the season because light penetration increases with decreasing ice thickness (Lazzara et al., 2007; Barber et al., 2015). In addition, snow cover is thinner on FYI compared to MYI as there is less time for snow accumulation, which results in more irradiance penetrating the sea ice. Larger surface areas are covered by melt ponds on FYI compared to MYI, which also makes FYI more transparent (Nicolaus et al., 2012). Because of enhanced light availability bottom sea-ice communities may reach higher biomass, though more condensed in time, as nutrients will be consumed more rapidly (Leu et al., 2015). An increase in the frequency of storms is predicted to result in the formation of more leads (open water between ice floes), which upon refreezing support high algal growth rates, though overall refrozen leads are less productive than thicker ice floes (Duarte et al., 2017).
The ecological consequences of the changes in ice structure and algal biomass are enigmatic. Current distribution patterns show that flagellates characterize surface communities, mixed communities inhabit interior ice, and pennate diatoms dominate bottom communities. Changes in sea-ice conditions, as established and predicted for specific regions of the Antarctic sea-ice zone, may alter the vertical distribution of ice-algal communities in pack-ice floes (Meiners et al., 2012). Productivity of sympagic communities will be affected by the foreseen changes in community structure (Campbell et al., 2017a). In addition, climate change can be expected to impact primary production directly, as the physiology and photosynthesis of microalgae are very responsive to environmental perturbations. Production may in fact increase with changes in the light climate. This increase may not necessarily be beneficial for biogeochemical fluxes, as the expected shift forward in time of the algal bloom can result in a mismatch in trophic relations. Pelagic herbivores feeding on sea-ice algae may not be able to synchronize with the algal bloom, with negative consequences for their reproduction (Leu et al., 2011; Søreide et al., 2010). The coupling between primary and secondary producers becomes more important as sympagic blooms may become shorter. Arctic sea-ice thinning and consequent ablation of bottom communities is likely to happen earlier in the season (Barber et al., 2015). As a result, fluxes of carbon into the deep ocean upon ice melt may increase (Tamelander et al., 2009), with positive consequences for benthic communities that feed on them (Boetius et al., 2013). The consequences of sea ice seeding the pelagic community are hard to predict, as not enough is known about the current relevance of this process let alone future changes.
Over the last two decades, the occurrence and extent of melt ponds in the Arctic has increased (Lee et al., 2011). At the same time, the nature of these ponds has changed. With the thinning of sea ice, more ponds become connected with the ice interior and the seawater below the ice, turning closed pond-systems into open ponds (Lee et al., 2011). Consequently, melt ponds lose their unique character and become more like seawater communities. Predicting how much the contribution of melt ponds to carbon fluxes will change is difficult. Melt ponds currently contribute less than 5% to total annual production in the Arctic. However, locally they can contribute to more than 30% of annual production, thus an increasingly important role of melt ponds in biogeochemical cycles can be anticipated (Fernandez-Mendez et al., 2015). Closed ponds will likely be prone to nutrient limitation (Lee et al., 2012). With a limited nutrient supply, the composition of algal communities will shift towards species of smaller size, which generally contribute less to carbon export than larger species. The role of smaller species in biogeochemical cycles may be more significant as potential producers of DMSP and halocarbons (e.g., Stefels et al., 2007). The importance of sea ice as a link with the atmosphere may thus increase. Model studies show that especially the Arctic foodweb is sensitive to species loss that may result from climate change (Carscallen and Romanuk, 2012), implying negative consequences of sea-ice habitat loss for connected biomes.
Our analysis has made clear that state-of-the-art sea-ice biogeochemical models require more complexity than currently adopted. At least a 3-layer model featuring bottom, internal and surface communities would be needed to reproduce both Arctic and Antarctic habitats, and both landfast and pack-ice conditions (Figures 2 and 3). Besides diatoms, our analysis shows that at least one more group of algae, namely autotrophic flagellates, deserve consideration both spatially, in both hemispheres, and temporally, for playing an important role in community succession. Regionally, both centric diatoms and heterotrophic/mixotrophic populations may need to be considered. The analysed photosynthetic parameters (Figures 5 and 6) further show the large range of photoacclimation that characterizes sea-ice algae, which stresses the importance of considering mechanisms of photoacclimation in sea-ice algal models. The inclusion of such mechanisms is still rarely done, but with the analysis presented here, sea-ice algal models can now be greatly and soundly improved.