Global Change Biology (2015), doi: 10.1111/gcb.12882 Too much of a good thing: sea ice extent may have forced emperor penguins into refugia during the last glacial maximum JANE L. YOUNGER1,2*, GEMMA V. CLUCAS3,4*, GERALD KOOYMAN5 , † BARBARA WIENECKE6 , ALEX D. ROGERS4 ,PHILIPN.TRATHAN7 ,TOMHART4 and † KAREN J. MILLER8,9 1Institute for Marine and Antarctic Studies, University of Tasmania, Private Bag 129, Hobart 7001 Tas., Australia, 2Australian School of Advanced Medicine, Macquarie University, 2 Technology Place, Sydney 2109 NSW, Australia, 3Ocean and Earth Sciences, University of Southampton Waterfront Campus, Southampton SO14 3ZH, UK, 4Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, UK, 5Scripps Institution of Oceanography, University of California San Diego, San Diego CA, USA, 6Australian Antarctic Division, 203 Channel Highway, Kingston 7050 Tas., Australia, 7British Antarctic Survey, High Cross, Madingley Road, Cambridge CB3 0ET, UK, 8Australian Institute of Marine Science, The UWA Oceans Institute, 35 Stirling Highway, Crawley, WA 6009, Australia, 9School of Biological Sciences, University of Tasmania, Private Bag 5, Hobart 7001 Tas., Australia Abstract The relationship between population structure and demographic history is critical to understanding microevolution and for predicting the resilience of species to environmental change. Using mitochondrial DNA from extant colonies and radiocarbon-dated subfossils, we present the first microevolutionary analysis of emperor penguins (Aptenodytes forsteri) and show their population trends throughout the last glacial maximum (LGM, 19.5–16 kya) and during the subsequent period of warming and sea ice retreat. We found evidence for three mitochondrial clades within emperor penguins, suggesting that they were isolated within three glacial refugia during the LGM. One of these clades has remained largely isolated within the Ross Sea, while the two other clades have intermixed around the coast of Antarc- tica from Adelie Land to the Weddell Sea. The differentiation of the Ross Sea population has been preserved despite rapid population growth and opportunities for migration. Low effective population sizes during the LGM, followed by a rapid expansion around the beginning of the Holocene, suggest that an optimum set of sea ice conditions exist for emperor penguins, corresponding to available foraging area. Keywords: Antarctica, Aptenodytes forsteri, climate change ecology, molecular ecology, paleoecology, phylogeography, polynya, Ross Sea Received 24 July 2014; revised version received 13 November 2014 and accepted 30 December 2014 shifts, fluctuations in abundance, species extinctions, Introduction and also in the formation of genetically distinct popula- Genetic data from both modern and subfossil samples, tions (Hewitt, 1996). As climate change and habitat deg- paleoecological niche modeling, and fossil evidence radation potentially take us into the 6th mass extinction have become vital tools for reconstructing demographic (Barnosky et al., 2011), it is critical that we understand histories (e.g., woolly mammoths (Mammuthus primige- how species have coped with change in the past to be nius) (Nogues-Bravo et al., 2008) and lions (Panthera leo) able to assess their likely responses and resilience to (Barnett et al., 2014)). Indeed, such studies have shown future climate change (Hoelzel, 2010). that species’ patterns of genetic diversity and distribu- Emperor penguins (Aptenodytes forsteri) are an iconic tion have varied dramatically under different climatic Antarctic species whose population genetic structure regimes (Carstens & Richards, 2007). Climatic shifts has not been studied to date. We know little about dis- have been one of the major drivers of species’ range persal among colonies or how historical climate change may have affected their range and abundance. Thus, Correspondence: Jane Younger, tel. +61 439 869 364, we have limited capacity to predict how these birds e-mail: [email protected]; Gemma Clucas, tel. +44 1865 may fare in the future. Projections for continent-wide 281991, e-mail: [email protected] declines of emperor penguins have been made based *These authors have contributed equally to the manuscript and are on the demographic responses of the Pointe Geologie joint first and corresponding authors. † These authors have contributed equally to the manuscript and are colony to changes in sea ice conditions (Barbraud & joint last authors. Weimerskirch, 2001; Jenouvrier et al., 2009, 2012, 2014; © 2015 John Wiley & Sons Ltd 1 2 J. L. YOUNGER et al. Ainley et al., 2010). However, decadal monitoring data emperor penguins may also have to contend with are only available for this single site out of 46 known altered prey availability and face new threats from pre- emperor penguin colonies; as such, the climate change dators as changing conditions differentially affect spe- responses of emperor penguins across their entire dis- cies at other trophic levels (Trathan et al., 2011). tribution and over millennial timescales are currently During the last glacial maximum (LGM, 19.5–16 kya), unknown (Ainley et al., 2010). the winter sea ice extent was approximately double the Emperor penguins are highly reliant on sea ice present-day values, and seasonal variation in sea ice throughout most of their breeding cycle, and mating and extent is thought to have been greater (Gersonde et al., incubation takes place on land-fast sea ice in most of the 2005). It is unclear how this would have affected known colonies (Fretwell et al., 2012, 2014). During the emperor penguins. Thatje et al. (2008) suggested that breeding season, emperor penguins feed on prey that is they may have migrated with the sea ice to lower lati- also sea ice dependent (Gales et al., 1990). Significant tudes, staying within energetic migration thresholds of areas of open water exist year-round within the Antarc- the ice edge, and could have maintained breeding pop- tic sea ice zone in the form of leads and polynyas (Zwal- ulations around Antarctica by foraging in the marginal ly et al., 1985 and references therein). These areas are ice zone at the sea ice edge. Alternatively, they could often important in providing emperor penguins access have remained associated with polynyas. Sediment to their underwater foraging habitat when the fast ice cores suggest the existence of LGM polynyas in several extends far from their colonies (Dewasmes et al., 1980). locations, including the northwestern Ross Sea, the Polynya formation is driven by either upwelling of cir- southeastern Weddell Sea off Dronning Maud Land, cumpolar deep water or by the outflow of katabatic and the northwestern Weddell Sea (Mackensen et al., winds that push sea ice away from the coastline (Martin, 1994; Brambati et al., 2002; Thatje et al., 2008; Smith 2001). Polynyas are associated with enhanced primary et al., 2010). In either case, reductions in overall primary production, as the reduction in sea ice volume facilitates productivity within what is today’s seasonal sea ice an earlier spring melting of sea ice and a coincident ear- zone (Domack et al., 1998; Kohfeld et al., 2005) would lier start in photosynthetic primary productivity (Martin, likely have been detrimental to emperor penguin popu- 2001). Some polynyas are permanent features of the sea lations (Ainley et al., 2010). ice zone and create areas of hyperproductivity, such as Little is known about the level of natal philopatry or the Ross Sea polynya (Smith & Gordon, 1997), while migration among emperor penguin colonies. Under- most are smaller and ephemeral features depending on standing philopatry is particularly important in light of wind stresses and currents. population models that suggest that emperor penguins Changes in the extent and duration of sea ice around may be declining as a result of local climatic shifts (Je- Antarctica show highly regionalized trends with some nouvrier et al., 2009, 2014). High emigration rates are areas increasing or remaining stable while others are conceivable among emperor penguin colonies; satellite decreasing (Zwally et al., 2002; Vaughan et al., 2013); tracking has shown that they travel thousands of kilo- this has an effect on the population dynamics of meters on their juvenile journeys, often passing other emperor penguins as both positive and negative sea ice colonies (e.g., Kooyman et al., 1996; Wienecke et al., anomalies can result in negative population growth 2010; Thiebot et al., 2013). Generally, philopatry is high rates at the local scale (Massom et al., 2009; Ainley et al., among penguins (Saraux et al., 2011; Dehnhard et al., 2010; Barbraud et al., 2011; Jenouvrier et al., 2014). 2014), but population structure is absent in many spe- Despite uncertainties over the rate and extent of ice loss cies (e.g., Chinstrap penguins (Pygoscelis antarctica) that will occur around Antarctica, all climate models (Clucas et al., 2014)) as even low levels of migration can project a reduction in the extent and duration of Ant- be sufficient to homogenize populations (Hartl & Clark, arctic sea ice by the end of the century (Collins et al., 1997). 2013). As sea ice declines, we might expect emperor We analysed the population structure among eight penguins to be disadvantaged by a lack of breeding extant emperor penguin colonies (Fig. 1) using mito- habitat (Jenouvrier et al., 2014), unless they have the chondrial DNA sequences and inferred population tra- capacity to alter their preferred choice of breeding site