MARINE ECOLOGY PROGRESS SERIES Vol. 262: 19–25, 2003 Published November 7 Mar Ecol Prog Ser Late-Holocene initiation of ice-free ecosystems in the southern Ross Sea, Antarctica Steven D. Emslie1,*, Paul Arthur Berkman2, David G. Ainley3, Larry Coats4, Michael Polito1 1Department of Biological Sciences, University of North Carolina, 601 S. College Road, Wilmington, North Carolina 28403, USA 2Byrd Polar Research Center, Ohio State University, 1090 Carmack Road, Columbus, Ohio 43210, USA 3H. T. Harvey & Associates, 3150 Almaden Expressway, Suite 145, San Jose, California 95118, USA 4Quaternary Sciences Program, Department of Geology, Northern Arizona University, Flagstaff, Arizona 86011, USA ABSTRACT: New data on marine sediments, seawater paleotemperatures, and the occupation his- tory of Adélie penguins indicate that modern ice-free conditions in the southern Ross Sea developed only within the last 1000 yr. Here we show that penguins permanently abandoned the southern Vic- toria Land Coast 2000 yr ago when extensive sea-ice cover blocked access to ice-free terrain for breeding. The first colonization of Ross Island in East McMurdo Sound, where over 300 000 penguins breed today, did not commence until after 1170 yr BP when coastal areas became exposed from under the Ross Ice Shelf. Our results demonstrate that investigations of abandoned penguin colonies can provide increased resolution to Holocene paleoclimatic records and paleoceanographic conditions in Antarctica. KEY WORDS: Antarctica · Adélie penguins · Late Holocene · Paleoclimate · Ross Sea · Sea ice Resale or republication not permitted without written consent of the publisher INTRODUCTION Current penguin distribution patterns in the south- ern Ross Sea result from meterologic and oceano- The Adélie penguin Pygoscelis adeliae is well graphic conditions that produce persistent open water known as an ice-obligate indicator species that is in East McMurdo Sound, in contrast to the multi-year sensitive to environmental change (Trivelpiece et al. fast ice that persists in West McMurdo Sound (Dayton 1990, Smith et al. 1999, Wilson et al. 2001, Ainley & Oliver 1977, Barry & Dayton 1988). In East McMurdo 2002). Over 744 000 breeding pairs, or 30% of the Sound, wind conditions (Stonehouse 1967, Ainley 2002) world population, occur on the Victoria Land Coast and warm, southward-moving currents and ocean bordering the Ross Sea (Ainley 2002), one of the south- swells from the Ross Sea influence the annual break- ernmost marine ecosystems on Earth. Adélie penguins up of sea ice, which in turn leads to eutrophic marine require open water near their breeding colonies, with communities adjacent to penguin colonies on Ross and no more than 2 to 3 km of persistent fast ice that must Beaufort Islands. In West McMurdo Sound, super- be crossed on foot to reach ice-free terrain; any addi- cooled currents moving northward from under the tional fast ice, especially lacking tide cracks associated McMurdo Ice Shelf promote multi-year fast ice, which with grounded icebergs, and colonies are very small or leads to oligotrophic marine communities and a lack of absent (Watanuki et al. 1997, Ainley 2002). All extant penguin colonies along the adjacent mainland (Dayton colonies on Ross Island (see Fig. 1), the largest occu- & Oliver 1977, Barry & Dayton 1988). pied area in the southern Ross Sea (defined as the area Preserved organic remains from ornithogenic soils at south of the Drygalski Ice Tongue, 75° 24’ S), are near abandoned penguin colonies, recognized by accumu- open water, and penguins appear to respond quickly lations of juvenile penguin bones, eggshell fragments, when new areas become accessible with a shift in sea- nest pebbles, and guano on the Victoria Land Coast, ice conditions (Wilson et al. 2001, Ainley 2002). provide evidence for past occupations by Adélie pen- *Email: [email protected] © Inter-Research 2003 · www.int-res.com 20 Mar Ecol Prog Ser 262: 19–25, 2003 McMurdo Sound east west Cape Barne Cape Crozier * DVDP Cores 8-10 Ross Island Marble Point Cape Royds M oder Spike Cape Cape Bird n m Dunlop Island ulti-yea Cape Roberts r fast ice Cape Ross Beaufort Island Depot Island 1170 - 790 yr B.P. Cape Day Franklin Island Cape Hickey Prior Island 2000 - 1000 yr B.P. Coast Land ctoria Vi Drygalski Ice Tongue Fig. 1. Map of Antarctica showing the location of the Victoria Land Coast and Ross Island (in box) with detail of the southern Ross Sea and sampling locations (named capes and islands) discussed in the text. The location of the Dry Valley Drilling Project (DVDP) Cores 8 to 10 in a prograding delta at the mouth of Taylor Valley, west McMurdo Sound, is also provided. Bold lines indi- cate ice tongues and the edge of the Ross Ice Shelf; bold dashed lines represent proposed extent of permanent or multi-year fast ice at 3 time periods including modern, 1170 to 790 yr BP, and 2000 to 1000 yr BP. This last period is when permanent abandon- ment of the southern Victoria Land Coast occurred, and predates the first occupation of Ross Island by breeding penguins guins (Baroni & Orombelli 1994, Lambert et al. 2002). Table 1, Appendix 1). At each site we excavated 1 × 1m These data also indicate that populations north of the test pits in 5 cm levels. Excavations ceased when Drygalski Ice Tongue have remained relatively stable bedrock or the bottom of the ornithogenic soils (recog- since the mid-Holocene (Berkman et al. 1998). This nized by a change in sediment texture and color) was stability is due in part to the presence of the Terra reached. The pit was then backfilled. All sediments Nova Bay, Passage, and Ross Sea polynyas, which were screened through 3 fine-mesh sieves (mesh sizes, developed after retreat of the West Antarctic Ice Sheet from top to bottom, of 0.64, 0.32, and 0.025 cm2) to (WAIS) 7600 yr ago, based on dates from Cape Bird, recover most organic remains preserved in these sites. Ross Island, and Franklin Island (Conway et al. 1999, Thirty-six samples of penguin bone, eggshell and but see also Andrews et al. 2002). The grounding line of feather fragments recovered from these excavations WAIS had retreated to the south of Cape Ross and Ross were submitted for radiocarbon analysis. In addition, Island, and southward along the continental coast to 83 previously published radiocarbon dates on Adélie Hatherton Glacier (80° S), by 6800 yr ago (Licht et al. penguin remains from this region were recalibrated 1996, Conway et al. 1999, Hall & Denton 1999). In con- for comparison to our data (Table 1, Appendix 1; see trast to the northern penguin colonies, those south of below). the Drygalski Ice Tongue apparently have been alter- Due to upwelling of old carbon in Antarctica, all nately occupied and abandoned in conjunction with radiocarbon dates must be adjusted for the marine- shifts in ocean circulation and persistence of fast ice carbon reservoir effect. There so far exists no simple or that influenced penguin access to ice-free terrain. reliable method for correcting these dates, especially given the diversity of materials that often are dated (e.g. marine plants, vertebrates, and calcareous fos- MATERIALS AND METHODS sils). This problem has received much attention (e.g. Björck et al. 1991, Gordon & Harkness 1992, Berkman To further investigate the occupation history of Adélie & Foreman 1996, Andrews et al. 1999, Domack et al. penguins in the southern Ross Sea, we sampled 13 1999a), and will not be easily resolved. Here, methods abandoned and 3 active colonies at Marble Point and to adjust a conventional radiocarbon-derived age of named capes and islands along the Victoria Land penguin remains include either (1) subtracting a pre- Coast, and at 4 capes on Ross Island, in 2001 (Fig. 1, determined number of years from all dates, or (2) cor- Emslie et al.: Ice-free ecosystems in the Ross Sea 21 recting and calibrating each date using a ∆R (reservoir) method of Berkman & Forman (1996). Calcitic Antarc- value that incorporates changing differences in reser- tic scallop Adamussium colbecki fossils were further voir effects between the atmosphere and ocean selected for stable oxygen-isotope analyses (Klein et through time (Stuiver et al. 1986, 1998). Method 1 pro- al. 1996) because they represent a comparable paleo- vides a corrected date in radiocarbon years, while ecological and geochemical framework to interpret Method 2 produces a date calibrated to ‘calendar years ambient oceanographic conditions through time. Esti- before present’ (yr BP). Both methods are based on mated seawater paleotemperatures, based on the δ18O radiocarbon dates obtained on modern (before nuclear values (±0.1‰), were derived according to the equa- testing, pre-1950) specimens of marine organisms that tions of Craig (1965). yield ages from 100s to >1000 yr due to ingestion of Unless otherwise stated, all radiocarbon dates listed old carbon during life. herein from ice cores, debris bands, marine sediments, Sediments and calcareous marine fossils that were and calcareous fossils from the published literature (as deposited along the coast in West McMurdo Sound referenced) are corrected to radiocarbon yr BP (for were interpreted from the Dry Valley Drilling Project details on how these dates were corrected, see refer- (DVDP) Cores 8 to 10, which were extracted from the enced literature). We assume these corrected dates same hole at the base of Taylor Valley (Fig. 1). Radio- roughly correspond to our calibrated ages for the pen- carbon dates on calcareous marine fossils recovered guin remains presented below. We do not consider our from DVDP Cores 8 to 10 were corrected by subtract- correlations between corrected dates versus calibrated ing 1300 yr from the conventional date, following the dates to be a problem, as discrepancies between them are minimal for material dating to the mid- to late Holocene (Stuiver et al.