Understanding the structure and functioning of polar pelagic ecosystems to predict the impacts of change

The determinants of the structure, functioning and resilience of pelagic ecosystems across most of the polar regions are not well known. Improved understanding is essential for assessing the value of biodiversity and predicting the effects of change (including in biodiversity) on these ecosystems and the services they maintain. Here we focus on the trophic interactions that underpin ecosystem structure, developing comparative analyses of how polar pelagic food webs vary in relation to the environment. We highlight that there is not a singular, generic Arctic or Antarctic pelagic food web, and, although there are characteristic pathways of energy flow dominated by a small number of species, alternative routes are important for maintaining energy transfer and resilience. These more complex routes cannot, however, provide the same rate of energy flow to highest trophic-level species. Food-web structure may be similar in different regions, but the individual species that dominate mid-trophic levels vary across polar regions. The characteristics (traits) of these species are also different and these differences influence a range of food-web processes. Low functional redundancy at key trophic levels makes these ecosystems particularly sensitive to change. To develop models for projecting responses of polar ecosystems to future environmental change, we propose a conceptual framework that links the life histories of pelagic species and the structure of polar food webs.


Supplementary Table S1
Environmental characteristics of the Arctic Ocean and Southern Ocean that are important influences on metazoan pelagic biodiversity and ecosystem structure and functioning [1,2].

Feature
Arctic Ocean Southern Ocean Topography [1] Almost completely surrounded by land, occupying the high-latitude polar region and is connected north-south to both the Pacific and Atlantic Oceans.
The continental shelf areas of North America and Europe are extensive and generally shallow (<100 m), while the central Arctic Basin is 2000 to 4000m deep.
Continuous water mass surrounding the Antarctic continent links all of the world's major oceans.
Deep continental shelf areas (> 500m). Submarine mountainous arcs extend to the surface and island archipelagos occur with locally extensive shelf regions. Generally greater than 2000 m deep.
Major currents and structure [1] The northward flowing Norwegian Atlantic Current between Greenland and Norway brings Atlantic waters into the Barents Sea region and shelf areas to the east. Farther west, there is the southward flowing East Greenland Current, which brings polar waters south to areas around southern Greenland and further west towards the Canadian coast.
On the Pacific side, exchanges with the Arctic are important but more restricted by the shallow (~50 m) Bering Strait, which separates the Bering Sea to the south from the Chukchi Sea to the north. Where conditions for growth are less favourable, production is generally low and smaller autotrophs (diatoms and dinoflagellates) dominate the plankton communities.** ** Intense blooms dominated by diatom species. The prymnesiophyte, Phaeocystis, can also be important.

**
Light and iron are the main limiting factors [6].
Production generally constrained by light and macronutrient availability.
Where there is little snow cover during spring, light penetration through the ice can generate large early season blooms below the ice [5].
In the Southern Ocean, iron is crucial for bloom development, generating extensive and sustained blooms in areas high in macronutrients in continental shelf waters south of the Southern Boundary of the ACC [7] and around various Southern Ocean archipelagos [8].
Substrate for breeding seabirds and marine mammals Require access to appropriate substrates for nesting, or haul-out while ice-obligate species haul-out on sea-ice or breed on iceshelves. The availability of these conditions and their vicinity to appropriate food supply limits the potential distribution of these species [9].** **

Variability
Marked interannual variability in ocean and ice conditions associated with hemisphere scale atmospheric variability (e.g. North Atlantic Osscillation (NAO) and El Niño Southern Oscillation (ENSO). Regional footprint varies throughout the Arctic [10].
Marked interannual variability in ocean and ice conditions associated with hemisphere scale atmospheric variability (e.g. Southern Annular Mode (SAM) and ENSO). Regional footprint varies throughout the Southern Ocean [10].

Change
Overall reduction in winter sea ice extent.
Increased influence of North Atlantic waters in Barents Sea region [10].
Small overall increase in circumpolar winter seaice extent.
Large regional differences. Increases in Ross Sea region and decreases around West Antarctic Peninsula [10]. ** Statements apply to Arctic and Antarctic Supplementary Figure S1 Consumption flows derived from Ecopath mass balance models constructed for: (a) west Antarctic Peninsula [11], (b) South Georgia in the northern Scotia Sea [12] and (c) the Barents Sea [13]. The food webs were derived using the balanced consumption matrix from each model and flows aggregated into the groups shown. Flows are expressed as percentages of the total primary production supporting the pelagic food-web. Benthic flows are shown separately and expressed as percentage of the total pelagic production. The coloured lines indicate the source group for flows.
The three studies analysed the food webs at very different levels of taxonomic or functional resolution. As far as possible we mapped the available information onto the groups shown, but this was not always simple and required some interpretation based on the information presented in the individual studies. The exact flow rates do change based on the aggregation decisions, but the general pattern, relative magnitudes and routes of energy flow are not sensitive. For some flows disaggregation into the required structure was not possible so aggregated flows are illustrated. For example, the Barents Sea model does not distinguish meso-and macrozooplankton so the associated flows are combined in the food web figure ( Figure S1c). Numerical values of combined fluxes are shown within an outlined box -the outline colour indicates the combined source. Focusing on the trophic flow between boxes and connections to the highest trophic levels, we do not show most return flows and internal box consumptions (e.g. fish eating fish). The original studies show the more detailed interactions in each system.
For comparative purposes the direct phytoplankton flux into the benthos in the Barents Sea model was excluded from the calculation of the total pelagic primary production required to support the food web. The direct flux was accounted for in the calculation of overall benthic fluxes. Detrital and benthic flows were aggregated and are shown as total inputs and outputs. Flows into the detrital and benthic pools were distinguished in the west Antarctic Peninsula and South Georgia models. For the Barents Sea the benthic system components and associated flows were analysed at much finer resolution, but detrital flows were not fully distinguished. For comparative purposes aggregated flows in a combined detrital and benthic pool are shown, while pelagic food web associated flows involving benthic organisms are shown as a separate set of aggregated flows (purple arrows and outline). The South Georgia model included two offshelf source boxes representing import/influx of i. krill and ii. other zooplankton or fish. For the calculation of relative consumption from different sources shown in the text, we totalled all inputs and calculated percentages of the consumption from different sources (including benthic and internal box consumptions (e.g. fish eating fish). These flows are not shown on the figures.