The history of seabird colonies and the North Water ecosystem: Contributions from palaeoecological and archaeological evidence

The North Water polynya (NOW) marine ecosystem is host to the largest seabird populations in Greenland. The community is diverse with 14 regular breeders and a few more species occurring as non-breeding summer visitors. The seabirds are almost exclusively present in the spring and summer, with the exception of some black guillemots (Cepphus grylle), which can winter in the NOW. Here, we focus on the three most abundant seabird species: the little auk (Alle alle), the thick-billed murre (Uria lomvia), and the common eider (Somateria mollissima) (Fig. 1). These species have the largest biomass and the greatest importance of the locally harvested seabird species. The breeding population of the little auk in the NOW region is immense, estimated at 33 million pairs (Boertmann and Mosbech 1998; Egevang et al. 2003) and corresponding to more than 80% of the global breeding population. The thick-billed murre colonies along the Greenland coast of the NOW consist of approximately 225 000 breeding pairs (Merkel et al. 2014) representing two thirds of the breeding population in Greenland. Currently, the NOW is the only area in Greenland where the thick-billed murre population is not in decline. The common eider breeding population in the NOW was estimated at 25–30 000 pairs in 2009, and has had a fivefold increase between 1997 and 2009 (Burnham et al. 2012). This increase is related to the stricter harvest regulations which came into force in 2001, especially the restricting of spring harvest near the colonies, which sparked a general population increase in all West Greenland populations following a decline in the 20th century related to overharvesting (Merkel 2010). Thus, despite the large uncertainties in the estimates, the NOW is clearly of great international importance to the species.

Information on the past seabird populations in the High Arctic is generally scarce, with a few investigations providing some information on past populations. A study from the east coast of Greenland (Wagner and Melles 2001) used lake sediments to track the history of a nearby little auk colony over much of the Holocene. They found that large numbers of birds arrived ca. 7500 cal. year bp, but suggest that they have not been continuously present since. The colony appears to have been present there from 7500 to 1900 cal. bp, from 1000 to 500 cal. bp, and then again over the last 100 years to the present. The authors suggest that the absence of birds was related to periods of colder climate in East Greenland. Expanding to the wider Arctic, there is evidence of the presence of Arctic tern (Sterna paradisaea), glaucous gull (Larus hyperboreus), and barnacle goose (Branta leucopsis) in Svalbard as far back as 9400 cal. bp (Yuan et al. 2009). Palaeoecological studies in the Canadian Arctic found no evidence of variation in the northern fulmar (Fulmarus glacialis) colonies over the short time period (maximum 200 years) covered by the sediment cores (Michelutti et al. 2009; Keatley et al. 2011). Work on peat deposits in Hudson Bay showed that the thick-billed murre colonies were at least 1500 and 3800 years old (Gaston and Donaldson 1995). In the NOW area, there is very little information on the past populations of seabirds. A study using the sedimentary archive of peat deposits on the Carey Islands demonstrated the presence of a seabird colony, probably the Atlantic puffin (Fratercula arctica), during the period from c. 7100 to 5100 cal. bp (Bennike et al. 2007). Apart from this study, there is no information available on the arrival times and change in abundance of seabirds in the NOW within a long-term perspective.

Seabirds feed in the marine ecosystem, but use the terrestrial environment for breeding. As a result, some species transport large quantities of nutrients from sea to land, which transforms the landscape around the colonies, leaving unequivocal signatures of their presence (González-Bergonzoni et al. 2017; Mosbech et al. 2018). These signatures in the landscape, both terrestrial and fresh water, open up the possibility of investigating past dynamics of seabird populations. The reliance of our three species upon the marine ecosystem varies and largely depends on their different feeding strategies and habitats.

All three species require open water in which to feed upon arrival in spring. Thus, all three species benefit from the early season open water of the polynya. The little auk is a zooplanktivore and is highly dependent on the abundant copepods within their diving range (approx. 50 m) and foraging range from the colonies (approx. 100 km). They feed these large, lipid-rich copepods to their chicks (Frandsen et al. 2014) and also rely on the copepods for their own foraging, supplemented with larger zooplankton (Karnovsky et al. 2008). The high density of large Calanus copepods is a key factor driving the abundance of the little auk colonies in the NOW area. The diet of the adult thick-billed murre during summer is dominated by the pelagic amphipod (Parathemisto libellula) and arctic cod (Boreogadus saida), supplemented with a variety of other invertebrates and fish (Gaston and Hipfner 2000; Karnovsky et al. 2008). It performs pursuit dives down to 150 m of depth within a foraging range of approx. 110 km around the breeding colonies (Mosbech et al., unpublished GPS-tracking data). Even though adult murres feed on a variety of food items, they are strongly dependent on abundance of forage fish for breeding success. The murre is a “single-prey loader”, capable of bringing home only one food item at a time to feed its chick, and this renders food items smaller than fish energetically unsuitable for raising the chick. (Elliott et al. 2009). The main diet of nestlings in Canadian Arctic is arctic cod, but also capelin (Mallotus villosus) and sculpin (Triglops, Gymnocanthus, Myoxocephalus spp.) (Elliott et al. 2009). In contrast to the little auk and the thick-billed murre, the common eider is exclusively a benthic feeder, targeting mussels, crustaceans, and polychaetes at water depths often below 10 m (Merkel et al. 2007). In the spring, the eider needs open water around the small islands and skerries on which they breed to ensure that foxes are excluded. After laying the eggs, only the female attends the nest, and neither the female nor the chicks feed during the brooding period. This leaves a much smaller nutrient imprint in the colonies, but as males and non-breeding eiders also spend time in the colonies, some nutrients are deposited on land.

Analysis of the NOW food web, using stable isotopes, has further demonstrated that the three seabirds investigated here are linked to the marine ecosystem in different ways (Hobson et al. 2002). Thus, each species may display a different degree of dependence on the particular ecological conditions of the polynya, which is in part reflected by the distribution around Greenland and the wider Arctic. Polynyas are characterised by sustained periods (spring through summer) of open water and relatively high levels of primary production, from which all seabirds potentially benefit. However, whilst the common eider and the thick-billed murre are found in many areas around Greenland, large little auk colonies are found solely in relation to productive polynyas (NOW and Scoresbysund).

Here, we set out to elucidate the history of the seabird populations in the NOW area. Given the dependence of the little auk on the polynya, it may be possible to use information on the past population dynamics to provide insights into the history of the polynya itself. To do this, we used palaeoecological methods to investigate the history of seabird colonies on the east side of the NOW. The key questions were: (1) How long have the colonies been present? (2) Is there any evidence of variation over time? And (3) how do the arrival times and any variation in abundance over time relate to climatic variation and the history of human settlement in the region?

Identification of the point of arrival of seabirds in a particular catchment is relatively straightforward as the transport of marine-derived nutrients (MDN) transforms the landscape (González-Bergonzoni et al. 2017; Mosbech et al. 2018). The estimated time of arrival can be determined by dating basal samples from peat cores, or the point of marked increases of δ¹⁵N in lake sediments. Values of δ¹⁵N in organic matter in lake sediments not affected by seabirds from the High Arctic are generally not higher than 3‰ and seldom rise higher 4–5‰ (e.g. Janbu et al. 2011; Perren et al. 2012). A rise in δ¹⁵N of 2–3‰ δ¹⁵N has been used in other studies to track millennial scale change in sockeye salmon population in Alaska (Finney et al. 2002) and increases to levels similar to those reported here, up to and > 20‰, have been used to track human influence on fresh waters, via transport of MDN, on the Canadian side of the NOW (Michelutti et al. 2013). Furthermore, there is an almost total absence of peat accumulation in the NOW region outside bird colonies (Mosbech et al. 2018), and in addition, the δ¹⁵N values of the peat cores, though variable between sites, are much higher than values reported for non-bird driven peat accumulation (Skrzypek et al. 2008). Thus, the combination of data presented here provides an unequivocal marker of bird arrival. The evidence indicates that the earliest arrival of the little auk in the NOW region was around 4400 cal. years bp, whereas the thick-billed murre and the common eider have been present for at least 1500 years longer. The three species discussed here breed in completely different habitats and landscape settings, so there is no possibility of a change in bird community at a particular site.

A striking feature of the arrival times of the little auk at the sites across the NOW, compared with the thick-billed murre and the eider duck, is the correspondence with periods of cooler climate, as reflected by the oxygen isotope data from the Aggasiz ice core (Fig. 4) (Vinther et al. 2009). Though, it should be noted that not all cool periods correspond to an arrival event, for example the period of 1900 cal. years bp, but then, only four of more than a hundred little auk colonies in the region were dated (Boertmann and Mosbech 1998). In contrast, the earliest record of the arrival of the thick-billed murre and the common eider correspond with a period of warmer climate around 5500 cal. years bp (Fig. 4). The common eider is a shallow water, benthic feeder, which needs ice-free (and thus fox free) conditions around its breeding series. The thick-billed murre needs fish to feed its chick—it is only capable of bringing one food item home at a time, and thus needs large food items for foraging to be feasible. In particular, the common eider, but also to some degree the thick-billed murre, have an extensive breeding distribution around Greenland, and seem capable of inhabiting many different habitats. In contrast, breeding colonies of little auks in Greenland are almost exclusively found in proximity to High Arctic polynyas, first and foremost the NOW but also in connection with the polynya at the mouth of Scoresby sund. To complete its breeding cycle, the little auk requires a sustained population of large copepods (C. hyperboreus and C. glacialis) in the upper 50 m of the water column during chick rearing in July/August, and presumably also in May/June during the early stages of the breeding season. This is provided only by a high Arctic marine ecosystem with open water and sustained primary production throughout spring and summer, combined with limited competition from fish predation on the copepods. In Greenland today, these conditions are exclusively found around polynyas, and thus, it is exceedingly unlikely that little auks would be present in large enough numbers to transform the landscape in the absence of a polynya. This means that our oldest date of a little auk colony 4400 cal. years bp may be seen as a minimum age of the NOW polynya ecosystem. It may well be possible that the NOW polynya first formed at this point in time, which corresponds to the end of the mid-Holocene Thermal Maxima and the onset of the cooling associated with the Neoglacial period (Vinther et al. 2009; Briner et al. 2016).

The idea that the NOW polynya first developed at a time of climate cooling is in good accord with the existing knowledge on the formation of the polynya. The formation of the NOW polynya primarily rests on the establishment of an arch or bridge of fast ice across Nares Strait in the southern part of Kane Basin, which blocks the southerly flow of drift ice from the Polar Sea (Melling et al. 2010). South of the ice arc in Smith Sound/Northern Baffin Bay, new ice forms on the sea surface but is continuously blown southwards by the prevailing northerly winds and currents. This creates the open water area, and at the same time, this ‘ice factory’ promotes brine formation and downwelling/upwelling, which seeds the system with nutrients, perhaps, supplemented by the West Greenland Current. The formation of the ice bridge may be a classic tipping point (Jeppesen et al. 2018), where a small change in temperature results in a large alteration of the surrounding ecosystems. In this case, perhaps, somewhat counter intuitively, the open water area of the NOW polynya is reliant on colder temperatures. In addition, colder air temperatures would result in a larger amount of new ice production in the ‘ice factory’, more brine formation, and greater circulation of the system. The fact that the little auk population arrived in and, perhaps, expanded during colder periods may indicate a link between the success of the little auk and climate. Thus, we speculate that a colder climate overall results in a stronger and more consistent ice bridge, more open water, more nutrients, increased primary production, and larger Calanus populations—all beneficial to the little auk. Lower temperatures may further imply reduced competition from fish predation on the large Calanus spp., as fish species like Capelin (Malotus vilosus) may be metabolically limited and retract their distribution southwards during colder periods (Rose 2005).

Caution must be exercised in interpreting the sediment record from the lake at Qeqertaq and also the peat cores, though their lower temporal resolution is already acknowledged. Whilst the dating of the arrival of the little auk is relatively straightforward, inferring changes in abundance can be more difficult. This is because the nitrogen in the system stored in the lake itself but also in the extensive peat developed in the catchment can be continually recycled, and thus, the ‘memory’ of past bird populations may be preserved in the δ¹⁵N signal, reducing its utility as an indicator of bird abundance. This may be particularly true for the lake at Qeqertaq where there is no other source of water input to flush the system in the event that birds were absent. The catchments soils also provide a large buffer/smoother for the lake as nitrogen enriched in δ¹⁵N will continue to leach from the soils into the lake for decades to centuries even in the absence of a bird colony. However, the combination of a number of indicators of bird abundance, in this case δ¹⁵N, δ¹³C (which in the lake reflect inputs of marine derived carbon), and Cd and P appear to provide a relatively robust semi-quantitative means of tracking relative change in bird abundance.

Cadmium and P data have been used in other studies (Wagner and Melles 2001) to infer change in little auk populations as both are abundant in little auk droppings. Both these elements are more abundant in sea water and are concentrated by marine zooplankton, an important food source of the little auk. Here, we also used the ratio of these elements to Ti. Ti content of seabird guano is low, whereas Ti is an indicator of input from catchment erosion which, dependent on geology, may supply both Cd and P. Thus, the ratios of Cd and P to Ti are more likely to reflect the input of the little auk, with any catchment input removed. Cd and P both suffer, to some degree, the same problem as δ¹⁵N as they can be stored in catchment soils and can also be mobile in the sediment cores. However, both seem to provide some evidence for variations in little auk numbers. It is, however, difficult to discuss absolute numbers with any certainty, but the records suggest that there has been a relatively large variation in little auk numbers since their arrival circa 4400 cal. years bp. The records, the peat core from Annikitisoq and the lake core from Qeqertaq, do not entirely agree, but combining the records, taking into account δ¹⁵N, δ¹³C, Cd/Ti, and P/Ti the data from Annikitisoq suggest that there was a decline in bird numbers around 4000 to around 3600–3700 cal. years bp, whereupon numbers rose again. This latter date is the point at which the colony was established at Qeqertaq (Fig. 2). The very high accumulation rates of the peat at Annikitisoq (GL3) suggest that there were abundant nutrients until at least 3500 cal. years bp, at which point nitrogen content of the peat cores was extremely low. It is difficult to interpret the records for the next 1000 or so years, but the general indication is a period of relative stability with some increases in bird influence around 3200 and then a decline around 3000 cal. years bp. At around 2500 cal. years bp, a number of the records indicate change, at Annikitisoq accumulation rate falls, P and P/Ti fall to zero, although Cd is still present. At Qeqertaq, P, Cd, and their ratios with Ti fall also fall around 2500 cal. years bp, but the isotope records and % N do not indicate change. This may be due to a mismatch in temporal resolution of the observations. P, Cd, and Ti are XRF data measured at higher temporal resolution (50 mm), whereas the isotope data are measured every 2 cm, so large short-term variation may be missed by the isotope data. The data do, however, agree that there was a peak at 2200 cal. years bp followed by a fall in indicators of bird input at Qeqertaq. The peak at 2200 cal. years bp corresponds with the arrival time of the colony at Kuukkat. Post-2200 cal. years bp, there follows a period of instability with sharp increases and falls in these key elements with a notable low point in P, Cd, and their ratios with Ti post 2000 cal. years bp, which remained low until 1500 cal. years bp.

The changes in bird numbers suggested by the data must be treated with some caution, as there is a relatively large degree of uncertainty associated with inferred changes. However, if we try and summarise the change in little auk numbers and by inference the polynya size and productivity the data suggest a more productive polynya at 4400, 3600, 2700, 2200, and around 1500 cal. years bp and a less productive or smaller polynya at 3800, 2500, and 2100–1550 cal. years bp.

This study is the first to investigate the long-term patterns in the presence, absence, and abundance of seabird colonies in the NOW across multiple locations and to provide direct evidence of the timing of the onset of colony formation of the three of the key sea bird species the region. The data, particularly when synthesised together, provide indirect evidence on the state, or ‘strength’ of the polynya through time. Some remarkable correlations between cold periods, bird arrival, and number, inferred polynya condition, and major human demographic events are evident. We should caution that these are inferred polynya conditions, and that there are no simple one-to-one relations between the polynya and demographic developments. For example, periods where the polynya is inferred to be large and productive appear coincide with the absence of humans. The present study certainly encourages further investigations along the same lines and in collaboration with other disciplines exploring polynya formation.