An Introduction to the Biocomplexity of Sanak Island, Western Gulf of Alaska1

Herbert D. G. Maschner,2,12 Matthew W. Betts,3 Joseph Cornell,4 Jennifer A. Dunne,5 Bruce Finney,4 Nancy Huntly,4 James W. Jordan,6 Aaron A. King,7 Nicole Misarti,2 Katherine L. Reedy-Maschner,8 Roland Russell,9 Amber Tews,10 Spencer A. Wood,11 and Buck Benson2

Abstract: The Sanak Biocomplexity Project is a transdisciplinary research effort focused on a small island archipelago 50 km south of the Alaska Peninsula in the western Gulf of Alaska. This team of archaeologists, terrestrial ecologists, social anthropologists, intertidal ecologists, geologists, oceanographers, paleoecolo- gists, and modelers is seeking to understanding the role of the ancient, historic, and modern Aleut in the structure and functioning of local and regional ecosys- tems. Using techniques ranging from systematic surveys to stable isotope chem- istry, long-term shifts in social dynamics and ecosystem structure are present in the context of changing climatic regimes and human impacts. This paper presents a summary of a range of our preliminary findings.

The greater North 1 This research was funded with generous support Pacific region hosts from NSF BE/CNH 0508101, NSF ARC 0326584, one of the world’s most important fisheries. NSF ARC 9996415, and NSF BCS 0119743. Manuscript Recent dramatic decreases in the numbers of accepted 3 February 2009. many species in this region have perplexed bi- 2 Department of Anthropology and the Center for Archaeology, Materials, and Applied Spectroscopy (CA- ologists, caused sweeping and underinformed MAS), Idaho State University, 921 South 8th Avenue, management decisions, and are provoking Stop 8005, Pocatello, Idaho 83209-8005. cultural disintegration of many Alaska Native 3 Archaeology and History Division, Canadian Mu- villages. But twentieth-century data show seum of Civilization, 100 Laurier Street, Box 3100, Station B. Gatineau, Quebec, J8X 4H2 Canada. clearly that there have been other periods of 4 Department of Biological Sciences, Idaho State low productivity for species, often followed University, 921 South 8th Avenue, Stop 8007, Pocatello, by periods of abundance (Trites et al. 2007; Idaho 83209-8007. H.D.G.M., A.W.T., K.L.R.-M., M.W.B., 5 Santa Fe Institute, 1399 Hyde Park Road, Santa Fe, A.T., and M.L., unpubl. data), as do histori- New Mexico 87501. 6 Antioch University New England, 40 Avon Street, cal, paleoecological, and archaeological data Keene, New Hampshire 03431-3552. from the last 5,000 years (e.g., Moser 1902, 7 Ecology and Evolutionary Biology, University of Finney 1998, Maschner 1998, Finney et al. Michigan, Ann Arbor, Michigan. 2000, 2002, Savinetsky et al. 2004, Maschner 8 Department of Anthropology, Idaho State Univer- sity, 921 South 8th Avenue, Stop 8005, Pocatello, Idaho et al. 2008). The global problem is thus: un- 83209-8005. der varying population cycles of humans, 9 Columbia University, Center for International salmon, groundfish, sea mammals, birds, and Earth Science Information Network (CIESIN), 61 Route many other species in the North Pacific re- 9W, Palisades, New York 10964. gion, and with varying harvests by the higher 10 1013 South 800 East, Salt Lake City, Utah 84105. 11 Department of Zoology, University of British Co- trophic levels, how do we sustain populations, lumbia, 2370 University Boulevard, Vancouver, British ecosystems, and the peoples and cultures who Columbia V6T 1Z4, Canada. depend on them for their social, political, 12 Corresponding author (e-mail: [email protected]). economic, and cultural survival? Most modern investigations of this ques- Pacific Science (2009), vol. 63, no. 4:673–709 tion have had limited temporal data. For ex- : 2009 by University of Hawai‘i Press ample, studies of possible long-term cycles All rights reserved in populations of Steller sea lions in the

673 . 674 PACIFIC SCIENCE October 2009

North Pacific were handicapped by limited is not only possible but necessary to our un- and short-term data, although cycles may be derstanding of the region. The implications much longer (Ocean Studies Board and Polar of this approach are profound and require Research Board 2003). Further, attempts to integration of anthropology, archaeology, model the effects of human fishing on the geology, ecology, mathematics, climatology, population dynamics of cod, pollock, or sea and history; the perspective of many spatial lions, for example, typically begin with an as- and temporal scales; and the seamless merg- sumed prefishing or prehunting state (e.g., ing of theoretical approaches from many Castellini 1993). Yet indigenous peoples have fields. Because indigenous peoples have been harvested large quantities of these species for harvesting resources in the North Pacific for thousands of years, requiring that we assume thousands of years, we must include them in that there is no ecologically meaningful time models and reconstructions of this ecosystem, period without humans, at least not since in which they may function as ecosystem deglaciation 16,000 years ago. Thus, archaeo- engineers, as more ordinary components of logical data provide the essential temporal food webs and landscapes, or as passive re- perspective for such an analysis. sponders to a world that is primarily driven From both anthropological and ecological by largely external forces such as climate and viewpoints, and building on the work of geomorphic evolution, or some combination Crumley (1994 and papers therein), the roles of all three. We expect this human role to of humans must be considered as one of vary temporally and spatially, with cultural many linkages within ecosystems that include context, and our work provides a framework other biotic components, as well as abiotic and a test case for understanding the history constraints and drivers. A sweeping article in of people as components of ecosystems. Science ( Jackson et al. 2001) suggested large- This project is being conducted in the scale roles of humans in ecosystems, arguing Sanak Island archipelago, 50 km south of the that changes in marine ecosystems are com- tip of the Alaska Peninsula (Figure 1), and plex, systemic, and have a long historical con- draws on past and ongoing work on the text (e.g., see papers in Rick and Erlandson nearby western Alaska Peninsula and Aleutian 2008). However, most of the suggested rela- regions. Sanak Island is about 120 km2 in tionships between trophic structure, species area, with 92 km of shoreline, and is the larg- extinctions, marine productivity, and human est island of its archipelago. The islands have exploitation remain untested. Our research a 6,000-year archaeological record. They was developed to directly test the sorts of re- were populated by a number of villages at lationships suggested by Jackson et al. (2001), Russian contact, were the center of sea-otter as well as provide fundamental data on the harvesting in the nineteenth century and role of humans in ecosystem dynamics in the cod harvesting in the early twentieth century, North Pacific over millennia. This project and supported active cattle ranches until the focuses on whole-ecosystem complexities of 1960s. The islands are surrounded by a mas- the region, which are founded on predator- sive reef system that supported some of the prey and other food-web interactions, cul- largest populations of groundfish and sea tural harvesting strategies, long-term changes mammals in the region. Millions of tons of in the North Pacific ecosystem and climate, cod were harvested here in the early twenti- and direct human impacts on coastal environ- eth century, and, today, commercial fisher- ments. men from False Pass and King Cove fish We have constructed a transdisciplinary here for cod in spring or halibut in summer. approach to studying humans as part of the These islands supported the last major popu- northern ecosystem, incorporating the mod- lations of sea otters to be commercially har- ern and prehistoric, the terrestrial and ma- vested. At least three river drainages have rine, the local and regional, and used both runs of sockeye, humpback, and chum empirical and theoretical exploration. We salmon. In the last 100 years, the landscape suggest that this multidimensional approach has been altered by large herds of now feral Figure 1. Map of western Gulf of Alaska. Sand Point, False Pass, Nelson Lagoon, and King Cove are modern Aleut communities. Sanak and Pauloff Harbor are abandoned. . 676 PACIFIC SCIENCE October 2009 cattle that are an important part of local ecological theory addresses the dynamics subsistence in the region. Finally, local peo- and interactions of species within ecosystems ples lived on these islands for many years (e.g., Chesson and Huntly 1994, 1997, Jones before their abandonment in the 1960s. Fam- and Lawton 1995, Polis and Winemiller ilies in Sand Point, King Cove, and False 1996, King and Schaffer 2001, Dunne et al. Pass, Alaska, have provided a wealth of local 2002, Turchin 2003, Worm and Myers 2003, knowledge about the landscape. Given the ar- Williams and Martinez 2004a), and we are chaeological, historic, and ethnographic data, applying such approaches to the study of the the local knowledge of the region, and the Aleut (sometimes referred to as Unangan) relative isolation of this archipelago from the and their ecosystems of the western Gulf of mainland peninsula, the Sanak Island region Alaska. is a perfect location to investigate the com- Evidence suggests that humans have been plex dynamics of human-landscape relations key components of the regional ecosystem and the role that humans have played in the for millennia. Standing crop, diversity, spe- structure of the North Pacific ecosystem cies composition, and the abundances of edi- over the last several thousand years. ble and medicinal plants all differ between Humans function within ecosystems in areas inhabited by Aleuts and those that many ways; they are one of many consumers, have not been (Bank 1953, Shacklette 1969, whose roles within food webs may range from Amundsen 1972, Johnson 2003). The distri- strong top predators, to omnivorous and gen- butions of plants typically influence the dis- eralized consumers, to builders and husbands tributions of animals (Morrison et al. 1998), of domesticated or industrialized resource including mammals and birds that, in turn, supply systems. They can be ecological engi- influence human populations and cultures. neers ( Jones and Lawton 1995, Jones et al. Additional ecosystem-level interactions are 1997, Power et al. 1998), changing the physi- caused by cultural systems of exploitation, cal structure of ecosystems, and can link ter- such as fisheries and the keeping of domestic restrial and marine ecosystems, as do other livestock. Moreover, humans have repeatedly organisms (e.g., seabirds or salmon) and removed and exterminated species, which abiotic processes (Polis and Hurd 1996, Polis may have altered food webs and ecosystems et al. 1997). For instance, they may alter or profoundly. For instance, in the 1780s, Rus- mediate movements of resources from ma- sian documents reported three species of rine to terrestrial regions and may modify foxes on Sanak Island, but in the 1920s, geological or hydrological linkages deliber- no foxes were reported. In recent decades, ately or as unanticipated side effects of their arctic foxes were reintroduced only to be ex- activities. Humans draw from the Ecosystem terminated in the last few years. In 1823, the Services of nature (e.g., supply of freshwater Russian Orthodox Church removed the Aleut and food, medicinals, esthetics), and these people from the islands to protect one of the have played strong historical roles in patterns last commercially viable sea otter populations of human settlement, health, and culture in the North Pacific. We suspect that this re- (Dale et al. 2000). Nevertheless, humans moval again reorganized ecosystems, which have only rarely been evaluated explicitly as again were altered when Scandinavian fisher- integral components of ecosystems (e.g., Ellis men and their Aleut wives recolonized Sanak and Swift 1988, Ellis and Galvin 1994) with in the 1870s. properties that can be compared with those Our research is different than the more of other biotic and abotic components within traditional (and highly successful) approaches an evolving system, the trajectory of which previously attempted in the region. Yesner’s may be disproportionately influenced by peo- (1980, 1981, 1987, 1988, 1998) studies, for ple and their behaviors. The interdependence example, were some of the first to identify of people and their ecosystems makes explicit the role of the Aleutian ecosystem in the study and evaluation of humans as integral structure of ancient Aleut society. He was parts of ecosystems critical to study. Much also one of the first to attempt to identify . Biocomplexity of Sanak Island Maschner et al. 677 human impacts on that ecosystem. Simenstad portunity to view both of the above research et al. (1978) made an early foray into the themes in finer detail and with comparisons complex dynamics between human harvesting that include modern domestic agriculture and the restructuring of the marine ecosys- (e.g., cattle) and technology (e.g., canneries, tem in the Aleutians. Building on these early regulated commercial fisheries). Here, we are studies, one focus of our work has been the investigating contrasts between traditional linkages involving humans within the Sanak and modern domestic-technological exploita- ecosystem. We are investigating major biotic tion ecosystems. Given this context, we pro- components, both terrestrial and marine, of vide some of our preliminary results here. the food webs and ecosystems that include Although still in the early phases of analyses, the Aleut, and include primary abiotic drivers some distinct patterns are beginning to such as climate (e.g., Pacific Decadal Oscilla- emerge from these investigations. tion, Aleutian Low) and geomorphic factors. The first section of our preliminary find- We are testing hypotheses as to the link- ings covers the ethnohistory and history of ages of humans within ecosystems, beginning the last 200 years. This is followed by a geo- with a series of descriptors of the effects of logical and paleoenvironmental review. The humans on terrestrial and marine populations archaeological data are then used to assess and the landscape, and following through to the temporal distribution of sites and the dis- analyze potential ramifying effects of these tribution of sites with shell middens to recon- to other ecosystem components and pro- struct the population history of the island, to cesses. Hypotheses are being addressed from investigate cod and otariid histories, to com- contemporary to ancient times, using sites on pare the modern intertidal ecosystem with Sanak and the western Alaska Peninsula that the prehistoric exploitation of shellfish, to date back to 6,000 years ago to address pat- track regional paleoenvironmental data using tern and process in deep time. modern and prehistoric samples for isotopic A second focus is the dynamic importance analysis. We conclude with our first explora- of humans to other components of the food tion into the reconstruction and analysis of web and landscape over the last several mil- food webs. lennia, as well as the dynamics of the human Finally, restating our research question in population itself. Here, the emphasis is on more specific and testable form: What have population and community dynamics, as pos- been the roles of prehistoric, historic, and sible to reconstruct from long-term indicators modern peoples in the structure and func- and contemporary, more precise, measures, tioning of the North Pacific ecosystem, and and from models. This focus more directly is it possible for that role to continue and to tests Jackson et al.’s (2001) assertion that hu- viably sustain the communities that live in mans may have cascading effects on marine this ecosystem today? This paper provides a (or other) food webs that are realized over first assessment of a number of specific find- decades, centuries, and millennia and over ings that contribute directly to this funda- large spatial scales. This work additionally mental question of the project, especially in examines population and social dynamics of regard to how humans shaped the North humans in correlation with resource and Pacific region, and, in turn, how the complex- landscape dynamics over millennia. Thus, we ities of the North Pacific conditioned the be- are directly addressing the conditions under haviors, adaptations, and histories of the which humans have sustainable patterns of people who lived there. resource supply and exploitation. We are in- vestigating pattern and process at multiple points in time and in space, as well as the preliminary results linkages and transitions through time and An Economic and Social History across space. The modern history of Sanak Island offers An analysis of the ethnohistoric data, as well a final important research focus, a unique op- as many hours of interviews with former . 678 PACIFIC SCIENCE October 2009 residents of the island, demonstrates the dy- O’Leary 2002), and later, peaceful contacts namic and volatile economic history of the that included baptisms and trade (Black Sanak archipelago. This historic volatility is 1999:66). In 1778, Captain James Cook an- partially a product of changing political chored his ship Discovery off Sanak Island economies and partly a product of the real- and took so many large halibut that he named ities of human impacts on these islands the group the ‘‘Halibut Islands’’ (Beaglehole creating changing economic and social op- 1967). But the Russians had a foothold on portunities (Reedy-Maschner 2001, 2004, Alaska, and after 1808 Sanak was reorganized 2007). into an artel, the last of six primary artels, or The inhabitants of this region today, at workers cooperatives under Imperial Russia, the time of historic contact, and as far as we for the purposes of mass harvest and export can surmise for at least the last 6,000 years, of sea-otter skins. Artels were led by Russian are the Aleut. In the eighteenth century, the baidarshchiks (captains), with Russian and Aleut lived in large, sedentary villages. They Aleut employees supplying the company with were organized into lineage-based households skins. that included from six to 10 families in a sin- Sea-otter hunting was conducted using gle dwelling, perhaps larger in the eastern Aleut labor, methods, and gear, in parties of Aleutian Islands (Lantis 1984, McCartney up to 15 baidarkas (kayaks) in the calmer 1984, Liapunova 1996, McCartney and summer months (Veniaminov 1984:330). Rec- Veltre 1999). The Aleut were ranked, with ords from that era are sparse, but Khlebni- hereditary nobility, a middle class made up kov’s report in 1818 ranks Sanak as a top of kinsmen to the nobility, people without producer compared with the other artels property or kinsmen, and slaves, who were in (1994:170). Aleut hunters were also held in a separate class altogether (Townsend 1980, debt to the Russian America Company. In 1983, Veniaminov 1984). Because most trans- 1823, after a decline in otters on the islands, portation and communication were by kayak, hunters of the Sannakh Artel were trans- the Aleut participated in trade and warfare ferred to Bel’kovskaia Bay on the mainland over extremely long distances (Maschner and and placed under the Unga baidarshchik, Reedy-Maschner 1998). Ethnographic and even though Sanak continued to be hunted archaeological investigations have demon- from a distance. strated that the Aleut lived corporately, shar- Sanak was not resettled until after the ing supplies, work areas, and other aspects U.S. purchase of Alaska in 1867. Unga and of household maintenance (Hoffman 2002). Belkofski Aleuts gradually returned to Sanak Economically, they were oriented specifically alongside Euro-American hunters who were toward the sea and utilized most edible re- hunting the few remaining sea otters. Former sources. Only on the Alaska Peninsula and villages were resettled, and new villages were adjacent nearby islands are terrestrial re- established. The Alaska Commercial Com- sources found and only in a few time periods pany built a trading post on the north end did the Aleut regularly exploit caribou, bear, of the island at Sanak village. Scandinavian and smaller terrestrial mammals found lo- immigrants took a particular interest in the cally. region, first for sea otters (until they were The Aleut of the Sanak Islands were a hunted out in the 1890s), and then more in- ranked polity of perhaps a few hundred living tensively for the codfish industry. These men in several villages across the islands before married locally and have left an indelible leg- Russian contact. They were in an antagonistic acy within the eastern Aleut community. trade relationship with the Alaska Peninsula The cod fisheries began in the Aleutians in Aleut and frequently at war with Kodiak the 1870s, with Scandinavian dory fishermen Islanders. The first historical reference to the arriving in the Shumagin Islands, Sanak Is- islands is in 1762, describing their sea otter lands, and up to Port Moller in the Bering wealth, which initially provoked Russian Sea. Schooners from San Francisco to Van- interest and hostilities (Black 1999:61–64, couver sailed to the region stacked with do- . Biocomplexity of Sanak Island Maschner et al. 679 ries and fishermen who fished for cod with Fox farming and cattle industries were also handlines (Shields 2001). This developed attempted on the Sanak Islands with mixed into a shore-based fishery, and codfish sta- success but allowed some residents to stay tions for salting and storing in barrels were until 1980. Today, the island supports wild established at Company Harbor (townsite of cattle and horses from those economic ex- Sanak), Pavlof Harbor (now Pauloff ) and periments; the Aleut harvest cattle annually several other locations on Sanak Island, as for subsistence. Sanak Island is now owned well as several on the Alaska Peninsula and and managed by the Sanak Corporation and in the Shumagin Islands (Figure 1). By 1915, Pauloff Harbor Tribe in Sand Point, and there were 17 shore stations operating in the many Aleuts in the region continue to visit region taking over 1 million fish each year the island and fish its waters. (Shields 2001:20). Scandinavian fishermen The historic Aleut were adept at taking and Aleuts shared communities to some ex- advantage of changing social, political, and tent and worked alongside one another. environmental dynamics. Regardless of the After 1915, fish began to disappear around global political and economic dominance of the Shumagins and Sanak, and by 1930 shore the time, the Aleut translated their long his- stations began to close (Shields 2001). But by tory of harvesting the North Pacific and 1899, salmon canneries had been built by the southern Bering Sea into adaptations that Pacific American Fisheries Company on the reflected the larger forces around them but Alaska Peninsula, offering a viable economic maintained their identify as coastal foragers. alternative, but with local environmental con- As we show in the following sections, this sequences (Figure 2). Villages were estab- flexibility was present in the prehistory of lished around these canneries, and gradually the Sanak region as well. residents of Sanak and several other former sea otter and codfish based communities Environmental Background moved to these new villages, which remain the viable fishing villages of King Cove, geological context. The Sanak Island Sand Point, False Pass, and Nelson Lagoon archipelago is located at the edge of the today. continental shelf above the Aleutian Trench

Figure 2. Long-term effects of the sockeye salmon harvest on Sanak Island, 1911 to 1927. Today there are fewer than 3,500 fish returning to the combined streams of Sanak Island (data from Rich and Ball 1930). . 680 PACIFIC SCIENCE October 2009 subduction zone and is separated from the the Alaska Peninsula, its proximity is critical Alaska Peninsula by 50 km of open water. in the paleoenvironmental history and dy- The island group is geographically part of namics of the Sanak archipelago, primarily the greater eastern Aleutian Island archipel- due to the influence of the peninsula on ago, but its location south of the western glaciation of the regional continental shelf Alaska Peninsula effectively separates it from and the proximity and eruption frequency of the Bering Sea, a physiographic characteristic its closely spaced Holocene volcanic centers, that distinguishes it from the rest of the Aleu- and seismic province. New paleoenvironmen- tians. The Sanak archipelago is geologically tal evidence from Sanak Island (sea level and related to the Shumagin Island group to the vegetation and climate proxies) is remarkable northeast (Moore 1974) and is composed because it provides spatial context and tempo- principally of rocks of the Cretaceous-aged ral depth to the record of postglacial environ- Shumagin Formation—continental shelf mental change in the region. slope and turbidite deposits consisting of Before this project, our view of the glacial medium- to fine-grained sandstone inter- history of the western Gulf of Alaska was bedded with mudstone. Fault-bound Jurassic- dominated by a handful of mapping or mod- to Cretaceous-aged chert, volcanic and ma- eling studies (Detterman 1986, Mann and Pe- rine sedimentary rocks occur between Sanak teet 1994, Dochat 1997, Wilson and Weber and Long Island (Figure 3). Sanak Island 1999), none of which included field observa- proper is topographically dominated by Sanak tions in the Sanak Islands. The Sanak archi- Peak, a Tertiary intrusive granitic massif at pelago was overrun by ice from a source area the northwestern end of the island. Although to the north during the last glacial maximum, its bedrock geology is distinct from that of but field evidence indicates that ice thickness

Figure 3. Sanak Island paleoenvironmental sites: lake and peat core sites (circles), glacial striations or grooves cut on granodiorite (arrows), and ice override (drumlinoid) features (dashed lines). Stippled areas indicate location of beach ridge complexes. . Biocomplexity of Sanak Island Maschner et al. 681 did not exceed 70 m above modern sea level cene terraces and beach ridges on Sanak are (asl) ( Jordan et al. 2007), an important down- driven by nonisostatic processes (e.g., chang- ward revision of ice thickness modeled by ing storm regimes, seismic events, tsunamis, Mann and Peteet (1994). Glacial striations eustatic sea level change). A well-developed and ice flow indicators oriented north-south shore platform cut in sedimentary bedrock across Sanak, the spatial pattern and limited deposits around the islands suggests vertical thickness of till on the island, and till lithol- stability of the crust at time scales on the or- ogy distinct from that observed in Cold and der of one to two millennia (Trenhaile 2002). Morzhovoi bays on the peninsula all indicate Sand and gravel storm beach ridge sets have that the Sanak Group was outboard of a re- been deposited in bay heads on Sanak during gional ice sheet axis ( J.W.J., unpubl. data), the past 1,000 years, particularly in Sandy Bay the thickness and position of which was re- (Figure 3), which was rapidly infilled with an constructed by Mann and Peteet (1994) and extensive sandy beach ridge plain during that suggested by Dochat (1997) to have been time. Historic records and field observations centered near Sanak. Basal ages from lakes (Lander 1996, Fryer et al. 2004, Jordan et al. cored on Sanak during our project indicate 2007) document the effect of tsunamis on that ice had disintegrated from the archipel- Sanak; run-up heights exceeded 10 m during ago by about 16,000 cal years B.P., several the 1788 and 1946 events, and although those millennia earlier than on the peninsula based events were locally catastrophic, the impact on the current chronology there. (All dates in alongshore was constrained by the direction this paper are calibrated years B.P., unless the from which the tsunamis originated. date concerns near-modern events.) paleoenvironment. Pollen stratigra- New insights from field evidence of glaci- phies were compiled from lake cores (Figure ation on Sanak are critical for understanding 3) and represent the vegetational history of the record of postglacial sea-level change in Sanak Island as it evolved in response to the archipelago and across the adjacent shelf. climatic and environmental change since A thin last glacial maximum ice margin at deglaciation. Artemisia, adapted to cool/dry Sanak constrains our estimates of glacial- conditions, was abundant during initial degla- isostatic depression and uplift to values lower ciation. Artemisia was also abundant during than those of the peninsula and is consistent the subsequent Younger Dryas. Cyperaceae, with the elevation of marine terraces docu- generally adapted to wetter climates, in- mented on the island (4 to 5 m asl) relative creased as Artemisia decreased. Ericaceae to the isostatically uplifted shorelines ob- peaked between @7,000 and 4,000 cal B.P., served in areas around Pavlof, Cold, and before the onset of neoglaciation. Sedges Morzhovoi bays (25 m þ) (Figure 1) (Wilson and grasses then dominated from 4,000 to et al. 1998, Jordan 2000, Jordan et al. 2007). 1,000 B.P., and Ericaceae decreased. Sedges Shoreline and ice thickness data from the declined after 1,000 B.P., and Ericaceae in- Alaska Peninsula (Mann and Peteet 1994, creased. The Sanak pollen data are in general Jordan 2000) included in the regional post- agreement with sites on the peninsula (Table glacial rebound model of James (2001) sup- 1 and associated references) but extend the port his findings of relatively low mantle regional paleovegetation record by several viscosity and limited glacial-isostatic rebound millennia. Changes in pollen stratigraphy are values relative to those of previous global- regionally conspicuous during the Holocene, scale models (cf. Peltier 1994). An important and suggest a shift at @4500 BP from gener- implication of James’ work for the western ally warmer (and drier) to cooler conditions Gulf of Alaska and the Sanak archipelago is in the eastern Aleutian Islands, the Shumagin that current-day vertical motions due to ice Islands, and on the central Alaska Peninsula mass changes in the late Pleistocene are ex- (Heusser et al. 1985). Localized dunes were pected to be of the magnitude of @1mmor active at that time based on the lack of stable less per year ( James 2001), suggesting that surfaces in regional eolian deposits. After sea-level fluctuations recorded by late Holo- 3,200 B.P. vegetation changes indicate the . 682 PACIFIC SCIENCE October 2009

TABLE 1 Climatic Regime Shifts in the Western Gulf of Alaska and Sanak Archipelago

Period Age Range Approximate Climatic Conditions

Early Holocene 9,000–6,200 B.P. Cool and dry 6,200–4,500 B.P. Warm and wet Neoglacial 4,500–3,200 B.P. Cool and wet 3,200–2,100 B.P. Cool and wet (increased storminess) Pre-Medieval climatic anomaly 2,100–1,700 B.P. Warmer 1,700–1,100 B.P. Cold Medieval climatic anomaly 1,100–700 B.P. Warmer and dry Little Ice Age 700–100 B.P. Cold and wet Recent era 100 B.P.–Present day Very warm

Note: Shading highlights periods recording important changes in air and sea-surface temperature, storminess, and enhanced ocean productivity reconstructed from various climatic proxies for the North Pacific and Gulf of Alaska (Huesser et al. 1985, Mason and Jordan 1993, Mann and Hamilton 1995, Calkin et al. 2001, Hu et al. 2001, Mann 2001, Finney et al. 2002, Jordan and Krumhardt 2003, D’Arrigo et al. 2005, Loso et al. 2006, Misarti 2007). There is a cool period at approximately 5,400 B.P. and a warm period be- tween 3,000 and 2,600 B.P. that have been documented in multiple Arctic records. Neither of these is currently visible in our regional proxy data, but both are possibly present in our settlement and midden data. onset of cooler and wetter conditions that identify regional changes in land use and site initiated neoglacial advances on the peninsula distributions; reconstruct the population his- and higher peaks of the Aleutians, and the re- tory of the island based on measures of site gional stabilization of dunes. Modern coastal size, density, and numbers of house depres- tundra developed after 1,000 B.P., broadly sions; sample middens for a fine-grained fau- reflecting the influence of the region’s mari- nal record to be analyzed in the context of time climate. Multiple tephras preserved in modern species distributions and in light of lake cores imply that vegetation on Sanak the paleoclimatic record; to use faunal data, responded periodically to this disturbance and the isotopic analysis of those specimens, mechanism (Heusser 1990), and organic car- to track changes in the structure and func- bon analysis of a peat core from Unimak tioning of the marine ecosystem; and to use Cove (Figures 3, 4) indicates cyclical inter- the faunal data to reconstruct prehistoric ruption of marsh productivity during the food webs and to assess changes in the mod- past 5,500 years, which may relate to periods ern marine ecosystem. The first two of these of storminess and the delivery of inorganic goals are reviewed in this section, and the re- shoreline sediment onto the marsh. maining goals are discussed in the following sections. This is considered a very prelimi- nary assessment of the archaeological data in Archaeological Research light of these project goals. The archaeology of Sanak Island was com- Critical to all of these analyses is that set- pletely unknown before our first preliminary tlement archaeology is feasible in the region. survey in 2002. Since that time, over 120 ar- House depressions are clearly visible on the chaeological sites spanning nearly 6,000 years surface, and most sites are single component, have been surveyed, mapped, and most have or if multicomponent, the components are been tested (Figure 5). The goals of our ar- spatially or stratigraphically distinct. This al- chaeological research have been to create a lows population reconstructions based on site local prehistory of the Sanak Islands that size or surface depressions to be possible. It could be correlated with the better-known also contributes to our ability to distinguish prehistory of the western Alaska Peninsula spatial and temporal trends in the settlement to investigate macroregional changes in so- system. cial, economic, and political interactions; The archaeology of the western Alaska . Biocomplexity of Sanak Island Maschner et al. 683

Figure 4. Total organic carbon (TOC) measurements for the Unimak Cove peat core. Low organic carbon is a proxy measure of increased storminess; higher TOC is a measure of calmer oceanic conditions.

Peninsula has been the focus of intensive one that shared some sort of social, political, research for nearly 15 years (Maschner 1998, and probably ethnic identity. Everything that 1999a,b, 2000, 2004a,b, 2008, Maschner and we have discovered so far for Sanak Island Reedy-Maschner 1998, 2005, Jordan 2000, falls squarely within our understanding of 2001, Jordan and Maschner 2000, Maschner the prehistory of the western Gulf of Alaska and Jordan 2001, 2008, Hoffman 2002, region. Over 150 Accelerator Mass Spec- Maschner and Bentley 2003, Maschner and trometry (AMS) dates on charcoal from Hoffman 2003, Maschner et al. 2008, 2009). house floors, shell middens, and other fea- When these data are combined with previous tures demonstrate a continuous 6,000-year studies from the region (Weyer 1930, Work- occupation of the island, matching our re- man 1966, McCartney 1969, 1974, 1984, gional prehistory for the mainland to the 1988, Okada 1980, Yesner 1985, Johnson north (Figure 6). 1988, 1992, Johnson and Winslow 1991, Hol- The earliest sites are small groups of house land 1992, Johnson and Wilmerding 2001), a depressions with square stone slab box- rather complete regional overview is possible. hearths along one wall. These houses may Extending from Unimak Pass in the west have been covered with skins as opposed to to the central Alaska Peninsula in the east is more permanent structures. Artifacts include a region that shares a number of important large laurel-leaf lance blades, cruciform- historical trajectories in technology, house based bone and ivory harpoons, and a small form, art, and other factors that permit its number of microblades (Maschner 2008). classification as an archaeological region— There are few sites with faunal preservation . 684 PACIFIC SCIENCE October 2009

Figure 5. Map of the Sanak Islands showing archaeological sites in the project area.

between 6,000 and 4,000 years ago because disruptions in the use of shellfish, and a plot there appears to have been little early of all dates on preserved middens demon- emphasis on shellfish. The earliest preserved strates this clearly (Figure 7). Thus, although midden deposits are dated to approximately there appears to be a continuous occupation 4,500 B.P. of the island based on Figure 6, the distribu- By 3,800 years ago, shell middens are tion of shell-based middens is discontinuous found across the island, much like the main- at best. land. Although there does not appear to be a At approximately 2,400 B.P., on both the greater number of sites, the economic focus mainland and the islands, there is a period of has clearly shifted to a greater diversity of village formation where much larger groups species. Shell middens again disappear after and larger households arise for a 400-year 3,300 years ago, only to become common af- span. Although villages are smaller on the ter 2,600 years ago. In fact, there are regular islands than on the mainland, the trend is . Biocomplexity of Sanak Island Maschner et al. 685

large shell middens are rare. The years be- tween 1,000 B.P. and 600 B.P. are tumultu- ous, with the rapid rise and fall of villages, a brief abandonment of the region, and the development of the historic house form (Maschner 1999a,b). About 400 years ago, the development of large, corporate multi- roomed houses is continuous across the entire region as is a massive population increase (Hoffman 2002, Maschner and Hoffman 2003). Currently, the analysis of the archaeo- logical data is only in its infancy. A brief summary of the regional climate history based on our data and a review of the literature shows a strong association between climate shifts and the distribution of shell middens in time (e.g., Table 1), especially in regard to storminess described earlier (Figure 4). It will become obvious in the next section that the demographic trends on the island partly mimic the distribution of shell middens Figure 6. A temporal density plot of 2s ranges of all ra- as well and the preliminary climatic history diocarbon dates in the Sanak archipelago. All radiocarbon presented here. One major disruption in the dates in the paper are presented as calibrated years before cultural sequence on the western Alaska Pen- present. insula mainland is a disappearance of most villages and their populations during the Me- dieval climatic anomaly approximately 750 B.P. (Maschner et al. 2009). All indications are that Sanak Island went through a similar population decline. A full evaluation and analysis of these trends is in preparation. Regionally, the artifacts and house styles, with a trend toward increasing technological complexity, increasing house size, and in- creasing village size, show the same patterns as those on the Alaska Peninsula (Maschner 1999a,b, Maschner and Bentley 2003, Masch- ner and Hoffman 2003), although the trends are not necessarily linear and our first im- pression is that the trends appear to be con- ditioned by marine productivity. Specific artifact types at key chronological transitions over the last 6,000 years point to regular con- Figure 7. Histogram of mean calibrated dates of indi- tact between the Sanak archipelago and the vidual shell midden components in the Sanak archipel- western Alaska Peninsula, a connection that ago. was observed among historic peoples by early explorers in the region as well. But these data also lead us to recognize, especially in har- much the same. On both Sanak Island and poon and end blade types, that there were re- the Alaska Peninsula, the period between gional interactions spanning over 1,000 km of 1,900 and 1,000 B.P. is poorly known, and the North Pacific Ocean from Kodiak Island although there are a large number of sites, in the east to Umnak Island in the west. . 686 PACIFIC SCIENCE October 2009

A Population Prehistory of Sanak Island tainty introduced by carbon dating. Modern techniques of carbon dating can produce Critical to our long-term understanding of very precise date estimates, but these esti- the role of humans in the structure and func- mates always come with an estimated error tioning of the Sanak ecosystem, and, in turn, range. Usually, only the midpoint carbon the role of that ecosystem in human adaptive date is used to place an archaeological sample capabilities is modeling the history of human into one time period or another. Given the populations on the island. One common fact that the estimated error associated with method for estimating changes in human any given carbon date may well be plus or populations from archaeological data is to minus 100 years or so, it is possible to con- convert analyses of temporal frequency distri- struct multiple estimates of population den- butions of house or settlement size into esti- sity, based on the estimated error of all the mates of human population (Hassan 1978, carbon dates. One method for dealing with Chamberlain 2006). This method relies upon this problem is to use the estimated error as- ethnographic parallels that allow past popula- sociated with each carbon date to define a tion densities to be estimated from modern or possible window of occupation for each site historical information concerning the rela- based upon probabilistic resampling. tionship between house size and population To estimate changes in human occupation density (Hassan 1981, Kolb 1985) or settle- on Sanak Island we developed a sophisticated ment size and population density (Wiessner model called ‘‘wiggle.sites’’ in R (version 1974). An important assumption is that 2.4.1, R Core Development Team, 2006) higher frequencies of dated sites represent to calculate a ‘‘human occupation index’’ higher population densities (Brantingham et (HOI) using temporal frequency data. The al. 2004, Holdaway et al. 2005). Temporal wiggle.sites model combines the idea that it frequency distributions, however, may be should be possible to extract persistent short- affected by taphonomic bias where the loss term demographic trends from temporal of data increases with time, which Surovell frequency distributions despite potential taph- and Brantingham (2007) suggested may be onomic bias with the idea of a dimensionless largely responsible for the positive curvilinear index that can describe trends in human pop- frequency distributions often observed in ulation over time without having to estimate archaeological, paleontological, and geologi- precisely the magnitude of change. cal data. One potential improvement to pop- The model uses probabilistic resampling ulation models based on temporal frequency to assign an ‘‘occupation window’’ from distributions could be to focus on short-term within the estimated carbon date range for variations, which presumably occur at differ- each village. The size of the window deter- ent rates than either the time-dependent ef- mines the amount of overlap in time between fects of taphonomic bias or the long-term sites, with a longer window resulting in more trends in other factors that affect human pop- overlap and a greater smoothing of the index. ulations, such as climate change. Surovell and In this case we used a window length of 100 Brantingham (2007) ended their paper with years. Using a smaller window typically re- an intriguing aside concerning the possibility sults in less smoothing but does not substan- of extracting population trends from biased tially change the shape of the resulting index. data by converting temporal frequency data After assigning an occupation window to into ‘‘dimensionless’’ ratios or proportions, each site the model calculates the total occu- such as the ratio of juvenile to adult humans, pied village area active in each 10-year period. which should preserve information on demo- The model uses a conversion factor of one graphic change despite the potential effects of person per 100 m2 of site area (from regional taphonomic bias. ethnographic data) to estimate population Another, largely ignored, problem with density from the occupied area. The model population estimates based on temporal fre- repeats the entire process 1,000 times and re- quency distribution models is the uncer- ports the results as 25th, 50th, and 75th quan- . Biocomplexity of Sanak Island Maschner et al. 687

Figure 8. The Human Occupation Index (HOI). The HOI is a model of possible ranges of human population on Sanak Island based on site area and numbers of houses. tiles. Because the goal of the model is to show the lower Alaska Peninsula, ca. 1,000 B.P. and how human occupation changes over time, 700 B.P. (see later in this section). This is be- the shape of the resulting index is more im- cause there are several small sites on Sanak portant than the absolute numbers derived that all date to the period surrounding ca. from the index. 1,000 B.P., and the probabilistic model is The HOI for Sanak Island shows three unable to track the sudden and precipitous relatively well-defined periods of human pop- decline in sites directly after that time. ulation growth and decline (Figure 8). The Like many temporal frequency distribu- first period is between approximately 3,000 tions, the HOI does exhibit what appears to and 1,800 B.P. with a peak around 2,000 be an exponential increase over time, which B.P. The second period peaks at about 1,000 may be due in part to taphonomic bias. How- B.P., and the third period peaks at about ever, although taphonomic bias might be 400 B.P. The 25th quantile shows that, given partly responsible for the differences in peak the possible range of carbon dates for all the population between the three periods noted sites, the population of Sanak could have earlier, bias cannot account for the general been zero at some points during the timeline, shape of the HOI. Instead, the results show such as between approximately 1,800 and that regardless of possible taphonomic bias, 1,200 B.P. and even 1,100 to 700 B.P., but it the Aleut population on Sanak Island proba- also shows that using even the most conserva- bly experienced large fluctuations over time. tive estimate of population still reveals the Given the HOI we can now ask questions three distinct periods of population growth that link changes in human population on and decline. Sanak with changes in ecology, climate, and One potential issue with the HOI is that it culture. As discussed earlier, the distributions does not appear to accurately track the pau- of sites, and the distributions of sites with city of human occupation on both Sanak and middens, is at least partially correlated with . 688 PACIFIC SCIENCE October 2009 the occupation index. But this is not tauto- et al. 2005). As a result, allometric analysis logical. There are many sites, some quite can provide a crucial means to compare the large, that have no middens leading to large preindustrial fisheries record with modern population estimates between the midden fisheries management data. peaks described in the previous section. The As outlined in Maschner et al. (2008) (see implication is that although there is some re- also Betts et al. in press a), we reconstructed lationship between population size and cooler a deep time series of Pacific cod (Gadus climates, and a relationship between cooler macrocephalus) length frequencies over our se- climates and midden use, there is a large quence and compared this with modern fish- amount of variation that is not measured by eries lengths taken from the Gulf of Alaska in a simple climate model. 2005 (Thompson et al. 2006). Here we pro- vide an overview of this analysis and provide an updated and in-depth discussion of rela- Faunal Studies tionships between changes in cod size and The archaeofaunal remains described in the climatic shifts. following analyses are derived from seven dis- Measurements on the cod remains were crete archaeological contexts (many more are derived from the length of the ascending pro- currently being analyzed) that form a time se- cess of premaxillas and the anterior-posterior ries spanning the period ca. 4,500 to 500 B.P. length of the centra of abdominal (trunk) All were recovered from single stratigraphic vertebrae. We estimated the size of each units dated by multiple radiocarbon determi- measured element using regression equations nations on wood charcoal. These deposits developed by Orchard (2001, 2003). Sample were screened through 8 mm (1/4 inch) sizes for the Sanak assemblages ranged be- mesh, a sieve gauge we consider appropriate tween a low of 10 to a high of 507 elements. to the ‘‘specimen sizes’’ for the taxa under in- The modern average length of Pacific cod vestigation (e.g., Cannon 1999). was obtained from data obtained from long- To assess the changes in relative taxo- line surveys done by the Alaska Fisheries nomic abundance through time, we used the Science Center in the Gulf of Alaska in 2005 climatic model for the western Gulf of Alaska (n ¼ 3,308) (Thompson et al. 2006). Al- presented in Table 1. From a marine ecosys- though data from baited jigs (the fishing tech- tem perspective, these perturbations may be nique used by the prehistoric Aleut) were not referred to as macrolevel regime shifts (Ben- available, modern longline-obtained cod are son and Trites 2002, Polovina 2005, Finney likely to be most comparable with fish taken et al. in press) and are used here to assess ma- by prehistoric jigging practices. The mean jor shifts in the distributions of key resources length recorded for the 2005 longline survey, and their harvests. which occurs as frequencies in size bins, was cod population histories. The mid- calculated using procedures in Gedamke and dens on Sanak Island contain important in- Hoenig (2006). formation on the population histories of fish Figure 9 displays a time series of average taxa. For example, fish bones are allometri- Pacific cod lengths spanning the preindustrial cally linked to animal size and weight, and (prehistoric) to recent period. To us, the therefore measurements from preserved fish graph indicates that: (1) preindustrial mean bones can be utilized to reconstruct length length varies in a cyclical fashion of increas- frequency distributions for prehistoric fish ing and then decreasing size, in apparent populations (e.g., Wheeler and Jones 1989). multicentennial cycles; and (2) the modern The ability to do such analysis is critical be- mean length of Pacific cod appears to fall cause fish grow throughout their life span, within the range of precontact (prehistoric) and thus most life history traits in fish stocks variability. In fact, t-tests reveal that the mod- are correlated with size (Reiss 1989). Size is ern mean is not significantly different from therefore an important metric in modern that of five of the precontact assemblages fisheries management research (e.g., Shin (P <:01). . Biocomplexity of Sanak Island Maschner et al. 689

Figure 9. Gulf of Alaska reconstructed mean cod lengths with the modeled climatic reconstruction demonstrating that there is a relationship between climatic regimes and cod length, especially after 2,000 B.P. (adapted from Betts et al. in press a and Maschner et al. 2008). Light gray bars are cooler regimes; dark gray bars are warmer regimes.

This latter point is not an unexpected cor- respondence between size and the Medieval relation; we know from modern management climatic anomaly (1,000–700 B.P.) and Little records that Gulf of Alaska cod populations Ice Age (700–100 B.P.), respectively the two are not overexploited (e.g., Thompson et al. climatic periods immediately preceding the 2006, 2007) and therefore should not exhibit modern era. Following these trends, it is a significant harvesting-mediated size reduc- tempting to speculate that the slight increase tion (for a discussion of the relationship be- in cod length at the end of the sequence may tween average size and exploitation pressure in fact reflect a stock response to global see Shin et al. [2005]). We have previously warming, similar to the increase in mean size suggested (Maschner et al. 2008) that modern that occurred during the Medieval climatic fisheries practices are largely maintaining the anomaly. The implications of this correlation preindustrial population structure of Pacific are unclear, but we propose that global warm- cod, and, therefore, using sized-based indica- ing may be a complicating factor in the recent tors as a metric, modern cod harvest practices population histories of Pacific cod. appear to be ‘‘sustainable.’’ otariids, climate, and human inter- What is perhaps equally intriguing is the actions. We assess the relationship be- apparent oscillation in cod size over the last tween Steller sea lions (Eumetopias jubatus) 4,500 years. When we compare this profile and humans by tracking the relative frequen- with our climatic model (light gray bars are cies of otariid bones in middens and compar- cold regimes, dark gray bars are warmer re- ing shifts in the resulting profiles. We gimes), we see a correlation (Figure 9). In measure shifts in the frequency of Steller general, the trends appear to suggest an in- sea lions over the sequence by means of sev- crease in mean length with warm regimes eral abundance indices, a procedure that we and a decrease in mean length with cold re- note is necessary to control for potential gimes. We draw specific attention to the cor- perturbations in the frequency of the com- . 690 PACIFIC SCIENCE October 2009

Figure 10. Otariid abundance indices (AI, Index Value on chart) as compared with climatic model. Cooler regimes (light gray bars) appear to positively influence otariid abundance (from Betts et al. in press b, Lech et al. in press; H.D.G.M., A. W. Trites, K.L.R.-M., M.W.B., A.T., and M. Livingston, unpubl. data). Dark gray bars are warmer regimes. parison taxa and thus eliminate the issue of encounter rate caused by availability of prey ‘‘closed arrays’’ (e.g., Betts and Friesen 2006; (e.g., Charnov et al. 1976, Stephens and see Lyman [2008] for discussion about closed Krebs 1986). arrays). In Figure 10, we assess changes in two An abundance index (AI) represents a otariid abundance indices against changes normed ratio of a highly ranked taxon (A), in in climate. Here the dark boxes represent terms of body size, to a lower-ranked taxon warmer and less-productive climatic regimes, (B), measured as AI ¼ A/A þ B. Decreasing and the lighter boxes indicate cooler and index values imply a decline in the abundance stormier periods (from Table 1). A relatively of a taxon, and higher values indicate an in- clear correlation occurs between the climatic crease in abundance. Because the measures model and the post-Neoglacial sequence after incorporate caloric relationships based on ca. 2,100 B.P. Specifically, we note an in- body size (i.e., larger-bodied animals have a crease in the otariid abundance index during higher caloric return), shifts in the index can a cold period from ca. 1,700 to 1,100 B.P., a reflect changes in return/encounter rate and decline in otariid abundance during the Me- thus foraging efficiency (e.g., Hildebrandt dieval climatic anomaly ca. 1,100–700 B.P., and Jones 1992, Broughton 1994, 1997, and an increase in otariid abundance during 1999, Janetski 1997, Butler 2000, Cannon the Little Ice Age ca. 700 to 100 B.P. 2000, Grayson 2001; for discussion see Ly- We interpret this correspondence as ota- man [2003], Betts and Friesen [2006]). As a riid populations responding to climate and result, when judiciously calculated, AIs can marine productivity shifts. In fact, the graph be used to track the occurrence of resource suggests that during times of warmer and depressions, defined as a reduction in return/ less-productive climatic regimes, otariids . Biocomplexity of Sanak Island Maschner et al. 691 experienced a form of resource depression, positive correspondence between the number likely a classic ecological depression (e.g., of humans and the number of otariids. To us, Charnov et al. 1976). In general the two AIs this suggests two things. First, it appears that correspond well, although the otariid–sea ot- humans were not overexploiting sea lions ter AI appears to correspond more favorably in the past, at least on Sanak Island. If they with the climatic model (Table 1), especially were, declines in otariids should be associated after the post-Neoglacial period ca. 2,100 with an increase in the number of predators B.P. Specifically, the otariid–sea otter AI in- (humans), the hallmark of a classic exploita- dicates an increase in the otariid abundance tion depression (Charnov et al. 1976). The during a cold period from ca. 1,700 to 1,100 exact opposite is evident here. The second B.P., a decline in otariid abundance during implication of this relationship is that humans the Medieval climatic anomaly ca. 1,100–700 and sea lion populations appear to respond to B.P., and an increase in otariid abundance climate changes in exactly the same fashion, during the Little Ice Age ca. 700 to 100 B.P. at least post-Neoglacial. This strongly implies The discordance between the two AIs, ca. that changes in climate and oceanic produc- 2,300 B.P. and 1,500 B.P., is explored in a de- tivity were an important force in the popu- tailed manner in Betts et al. (in press b); how- lation histories of large-bodied, top-tiered ever it is believed that the discordance is the predators, which includes both humans and result of disproportionate decreases and in- sea lions. creases in phocid seals during those periods relative to other taxa. Given the close rela- Stable Isotopes of C and N and Changing Pacific tionship of phocid seals to ice platform, it Regimes may be noteworthy that the decrease in pho- cid seals ca. 2,300 B.P. (resulting in a slight Stable isotopes have been used in ecological increase in the otariid-phocid AI) was associ- and archaeological studies to trace diet and ated with stormier environments, and the po- geographic location for the past few decades. tential increase in phocid seals (resulting in a In this study, we integrated both paleoeco- decrease in the otariid-phocid AI) was associ- logical studies and archaeological studies to ated with very cold environments (see Table determine changes through time in marine 1). It may be that the otariid–sea otter index ecosystems utilized by the Aleut for the last more closely tracks the natural abundance of 4,500 years. Stable isotope signatures are pre- otariids, especially as this relates to climatic served in many different types of tissue, in- shifts (see Betts et al. in press b for discus- cluding bone collagen, and bone is often well sion). preserved in shell middens in coastal Alaska. Using the human occupation index de- Bone is also particularly well suited to studies scribed in Figure 8, the archaeological record that span hundreds or thousands of years indicates that human populations were rela- because it has a slower turnover rate than tively stable during the Neoglacial but began organs, blood, and muscle tissue and there- to rise steadily, and at a relatively predictable fore reflects a longer period in an organism’s rate for hunter-gatherers, during the pre- life (Ambrose and Norr 1993, Lambert and Medieval climatic anomaly ca. 2,100–1,100 Grupe 1993). This allowed us to compare an B.P. During the Medieval climatic anomaly, organism’s overall trophic levels without con- populations declined sharply. This was fol- cerns about seasonal or short-term foraging lowed by a notable increase in population area changes. d13C and d15N are used in tro- during the Little Ice Age. phic level and dietary reconstructions, and Comparing our schematic population in theory, if changes are discerned in either curve (derived from Figure 8) against otariid isotope, they can be linked to changes in for- abundance indicates interesting correlations aging habits (Post 2002). Samples of Pacific between otariid and human populations (Fig- cod, sockeye salmon, harbor seal, northern ure 11). The trend during the Neoglacial is fur seal, sea otter, and Steller sea lion were again murky, but post-Neoglacial there is a collected from archaeological middens span- . 692 PACIFIC SCIENCE October 2009

Figure 11. Otariid abundance indices (AI, Index Value on chart) compared with the relative human occupation index (HOI) from Figure 8 (without scale). Although there are many more sites included in the HOI than in the faunal data, patterns do emerge. The increase in the otariid AI during the Little Ice Age, the decrease during the Medieval climatic anomaly, and the higher AI value during the general population increase between 1,700 and 1,100 B.P. (another cool period) track closely, indicating some relationship between otariid abundance, human populations, and climate after 2,000 B.P. (see also Betts et al. in press b). Before 2,000 B.P., there is no obvious relationship. ning the last 4,500 years on Sanak Island. arti 2007; Misarti et al. in press). d15N is uti- Bone collagen was extracted from 250 indi- lized as a marker for trophic position, and viduals from all six species and analyzed for d13C is more often linked to the type of plant d13C and d15N. at the base of a food web, geographic location In general, mean d13C and d15N of all six of an organism, and productivity levels (Fry species from 4,500 to 200 cal years B.P. main- and Sherr 1984, Schoeninger and DeNiro tain expected trophic positions when com- 1984, Michener and Schell 1994, Doucett pared with modern prey species (Figure 12). et al. 1996, Hobson et al. 1996, Hirons et al. Three species, salmon, cod, and sea otter, 2001, Schell 2001, Post 2002, McCroy et al. were compared within six archaeological 2004). These data suggest that otters, with time periods. There are significant changes changes in both carbon and nitrogen isotopes, in isotope concentrations within that time may have been feeding at slightly different frame for salmon and sea otter but not for trophic levels through time. For example, sea Pacific cod. Sea otters had a significant urchins, a favored prey item, are secondary difference in d13C (single-factor ANOVA, producers in a kelp forest ecosystem while F ¼ 3:644, df ¼ 120, P ¼ :07) and in d15N pelagic fish, also otter prey species, are not (single-factor ANOVA, F ¼ 3:767, df ¼ 120, only in higher trophic positions but also have P ¼ :003), and changes in d13C of salmon phytoplankton as the base of their food web were significant (single-factor ANOVA, (Estes and Duggins 1995, Watt et al. 2000, F ¼ 3:012, df ¼ 32, P ¼ :028), but no Steneck et al. 2002, Estes et al. 2003, Bodkin changes were found in d15N of salmon (Mis- et al. 2004, Reisewitz et al. 2006). The . Biocomplexity of Sanak Island Maschner et al. 693

Figure 12. Trophic position of archaeological mammals and fish with some selected modern prey species. All samples corrected for fractionation from muscle tissue to bone collagen. Prey data compiled from Schoeninger and DeNiro (1984), Duggins et al. (1989), Boutton (1991), Hobson and Welch (1992), Hobson et al. (1997), Satterfield (2000), Hirons (2001), Kline and Willette (2002). Archaeological data from Misarti (2007).

changes in sea otter carbon and nitrogen iso- has been recorded (Albers and Anderson topes may be explained by a change in per- 1985, Yang 2004, Ciannelli and Bailey 2005). centage of diet represented by each of these If all prey items, even though changing prey items. Only d13C changed in salmon throughout time, held the same trophic posi- over time, suggesting that there was a change tions and were in the same foraging areas, in productivity throughout the Gulf of Alaska then there would be no discernable change or a change in geographic area inhabited by in either d13Cord15N. the fish but not a change in trophic position of salmon. The lack of change in either d13C The Intertidal Landscape or d15N of cod in the archaeological record is interesting. Studies over the past 50 years The intertidal habitats of the Sanak archipel- have noted a change in diet for Pacific ago provide a wealth of human resources that cod during regime shifts in the northeastern were a key source of food for residents dur- Pacific, but no change in foraging location ing times of low oceanic productivity or . 694 PACIFIC SCIENCE October 2009 decreased offshore access such as storms. Pri- biological communities around Sanak Island. marily these intertidal regions are rocky For comparison between ancient and modern benches and cobble beaches, with occasional data we use whole-organism biomass, calcu- sand or pebble shores. Food resources of the lated from samples of modern abundance intertidal areas are relatively plentiful, quickly and ancient shell mass (using experimentally renewed, and predictable; all ethnographic derived regressions to back-calculate animal and archaeological sources agree that there is biomass from shell mass), as our response. thus little doubt that the intertidal has been There are clear differences in the biomass critical to the history of this region. The pre- composition of modern intertidal community served middens, described earlier, provide in- assemblages and ancient shellfish middens on formation on the structure of the prehistoric the island (Figure 13). This pattern is driven intertidal. primarily by differences in abundances of chi- Results from experimental collections on tons, snails, and mussels. These three taxa Sanak indicate that an hour of collection of account for 66% of the dissimilarity between contemporary intertidal resources (i.e., shell- modern and ancient assemblages in a similar- fish) provides up to 23 kg of meat mass (an ity percentage (SIMPER) analysis (Clarke average of 4 kg and a median of 316 g across 1993), which quantifies species’ contributions all shellfish taxa collected). Given such con- to sample dissimilarity (further results not temporary resource abundance it would be presented). Snails and mussels are more feasible to meet a reasonable daily protein abundant in the ancient midden records, but requirement for humans (e.g., 40 g/day [Er- chitons are more characteristic of the modern landson 1988]), especially in concert with intertidal assemblages. What is driving these a readily accessible source of carbohydrate differences? We ought to be able to explain (e.g., seeds or bulbs from terrestrial plants; the differences between modern and ancient or even marine algae, which contain roughly assemblages by shifts in either (1) intertidal 3 kcal per gram of dry mass [Paine 1971]). habitat types (e.g., from rocky to sandy); (2) Some of these resources are quickly gathered preferences of harvesters; or (3) productivity, in large volume (e.g., littorine snails, marine climate, or weather that alter ecosystems or algae); others (e.g., sea urchins) are slower to the accessibility of certain habitats. Here we harvest but may provide a substantial food explore the first two drivers of change and value in a reasonable amount of time (i.e., draw attention to earlier sections of this 1 hr of harvesting time could provide well paper for discussion of the climatic changes in excess of one individual’s requirements). on the island. Thus, similar to elsewhere in Alaska (e.g., Er- First, there is no evidence of shifts in landson 1988), contemporary intertidal com- the proportions of rocky versus sedimented munities could provide substantial resources shoreline habitat. The midden remains are for human populations on Sanak Island. So, dominated by rocky shore species, with no do we see evidence that Sanak Islanders relied evidence of sandy shore taxa dominating the upon the intertidal for such resources, and is assemblage, throughout the course of avail- there evidence that these intertidal communi- able data. A substantial shift in shoreline ties have changed through time—whether an habitat types would be expected to be evident artifact of human behavior or otherwise? in the relatively continuous midden record Using archaeological and contemporary available here; thus we conclude that it is un- biological surveys, we now explore whether likely that there have been large-scale shifts there have been shifts in the composition of the fundamental habitat types around the of intertidal communities around the island island over the last 5,000 years (see similar over the last 5,000 years. Our contemporary analyses with opposite results in Graham et surveys involve exhaustive vertical transects al. [2003] or Masters and Gallegos [1997]). (sampled using 0.25 m2 quadrats in which all To assess the harvesting preferences for macroscopic organisms were recorded) pro- particular taxa, we quantified return on col- viding a robust sample of the intertidal lection effort using empirical data from Sanak Figure 13. Comparison of ancient midden (a) and modern intertidal (b) abundances of major taxa around Sanak Island. The length of each wedge represents the relative abundance of the taxon in the sample unit (a vertical transect survey of the modern assemblage and one midden sample of the ancient assemblage). . 696 PACIFIC SCIENCE October 2009

Island. Total meat weight collected over time zooarchaeology of Sanak Island middens, (taxon-specific timed collections by knowl- ethnographic data, and interviews with mod- edgeable individuals repeated at different ern Aleuts in the region, a detailed picture of locations at appropriately low tide levels) is the Sanak Island intertidal food web, and the normalized by modern abundance (from sur- position of humans in it, begins to emerge veys mentioned earlier) for each group of taxa (Figure 14). Initial results reveal a species- to estimate a standardized return on collect- rich and complex food web with 725 feeding ing. The results indicate that three taxa are links among 164 taxa represented, making expected to be preferred: snails, mussels, and this the most diverse intertidal food web ever chitons. This helps explain the abundance of compiled and one of the few food webs to snails in the middens relative to the modern explicitly include humans as a node (see also shores; snails provide more meat for a given Link 2002). time spent collecting. Further, littorinid snails The Sanak Island intertidal food web is a are often found in massive aggregations and cumulative community food web—it repre- occur on wave-sheltered high intertidal sents the feeding relationships among co- shorelines that are accessible much of the occurring taxa that live or feed in the inter- year. It is interesting that although chitons tidal, as integrated over space (multiple Sanak are a rewarding and thus likely a preferred intertidal sites) and time (the last 5,000 years). shellfish, they are less abundant throughout For initial analyses of the network structure the midden record. If the midden record re- of trophic interactions among species that flects the natural abundance of chitons, this live or feed in the Sanak intertidal system, result indicates that rocky intertidal commu- we assume that the composition of the nities on the island may have varied over the Sanak Island intertidal communities has not course of settlement history. changed dramatically over the last few thou- The intertidal record for Sanak Island in- sand years in terms of what species are dicates that: (1) humans relied heavily upon present. For example, all intertidal species the intertidal for food resource needs (based collected in the middens are also present in on the rich and extensive midden record on the current intertidal community. Although the island); (2) the intertidal communities abundances of those and the other species could nutritionally support substantial hu- clearly change through time, abundance data man populations on the island (based on are not necessary to analyze basic properties reconstructions of prehistoric biological com- of food-web structure. This initial version of munities); and (3) although the modern the web is incomplete because it still lacks communities appear different from ancient some of the terrestrial taxa, particularly birds, midden assemblages, there is little evidence that feed in the intertidal, and inasmuch as no for clear temporal trends in the harvesting of food web can ever be entirely complete be- intertidal communities over the course of the cause there are always species and links that 5,000-year midden record on Sanak Island remain unidentified. However, it does give a (data forthcoming). This implies that the useful first look at the structure of the food structure and dynamics of the intertidal eco- web and how the Aleut fit into it. system has likely had a strong influence on On average, each species in this initial human history on Sanak Island, yet human version of the Sanak intertidal food web has history does not appear to have had a power- 4.4 trophic links to other species (i.e., links ful influence on the intertidal system. per species, or L=S ¼ 4:4) through links to resource (prey) species plus those from con- sumer (predator) species. The connectance Role of Humans in the Intertidal Food Web ðCÞ of the food web, which describes the pro- Through a combination of field observations portion of possible links among all species of the current intertidal communities of that actually occur, usually calculated as links Sanak Island, additional trophic (feeding) in- per species2 ðL=S2Þ, is 0.027, or 2.7%. This formation culled from the literature, detailed falls at the low end of the range observed for . Biocomplexity of Sanak Island Maschner et al. 697

Figure 14. Sanak Island intertidal food web. Spheres represent 164 taxa and links represent 725 feeding interactions. Basal taxa (autotrophs such as algae as well as detrital categories) are shown at the bottom, with the highest trophic level taxa shown at top. Aleut are indicated by the labeled black sphere and are located in the background due to their high number of links (40) to resource taxa. Image produced with FoodWeb 3D, www.Foodwebs.org. other food webs of @2% to 30% (Dunne (including humans). There is excellent reso- et al. 2004). The mean trophic level of the lution of basal and invertebrate taxa, unlike species in the food web is 2.2. Trophic level most published marine and estuary food web represents how many steps energy must take data sets (Dunne et al. 2004). In terms of tro- to get from an energy source to a species. By phic roles, 14% of the taxa are herbivores or convention, primary producers are assigned detritivores, 6% are cannibals, and 56% are a trophic level of 1, and obligate herbivores omnivores (i.e., feed on taxa at more than have a trophic level of 2. Trophic level can one trophic level). The harbor seal (Phoca be calculated in a variety of ways; here we vitulina) has the highest trophic level (3.7), use the short weighted trophic level algo- and the most common resource is detritus, rithm (Williams and Martinez 2004b). On which is consumed by 60 taxa. average, the shortest chain of feeding links Until they left the island in the late 1970s, connecting each pair of species is 2.6 links humans played important roles in the Sanak long. This ‘‘path length’’ gives an indication intertidal food web. They fed directly on 40 of how quickly trophic effects can spread taxa, making them the most general con- through a food web, with 2.6 indicating a sumer of intertidal taxa, with the next most longer mean path, and thus longer potential general species, a sea star (Pycnopodia helian- propagation time, than seen in most other thoides), consuming 35 taxa. They were strong webs (Williams et al. 2002). Twenty-three omnivores, feeding on everything from algae percent of the taxa in the web are basal (i.e., to a variety of invertebrates to the highest autotrophs or detrital categories), including a trophic level taxon, the harbor seal, making diverse set of algae taxa, 65% are inverte- their own trophic level intermediate at 2.9. brates, 9% are fishes, and 3% are mammals In addition to the 40 taxa humans fed on . 698 PACIFIC SCIENCE October 2009 directly, 92 additional taxa occurred within and Maschner 1999, Fitzhugh 2003), and par- human-topped food chains, with only 31 taxa allel with new data from the South Pacific (only one basal species) having no direct or (Nunn 2007), is directly a product of rapid indirect trophic linkages to humans. Thus, and rather radical regime shifts affecting humans played unique roles in the Sanak primary productivity. This appears to be es- Island intertidal food web through their ex- pecially true of the higher trophic levels, af- treme generality and omnivory, and there fecting human and sea mammal populations was a large potential for humans to affect the directly, and is even perceptible in the shifts intertidal ecosystem directly and indirectly in the size of Pacific cod (e.g., Maschner through hunting and gathering, as well as et al. 2008). through mechanisms such as food storage, In historic times this does not appear to technology, planning, and prey switching. have had the same impact. The ethnohistoric Future work will include analyses of the data imply that Sanak Island populations were topological roles of humans in the other ma- not so constrained or influenced by marine jor habitat types of Sanak Island (i.e., terres- productivity but rather by the mitigating trial, freshwater, marine) and comparisons of effects of global economic and political re- the structure of those habitat-specific webs gimes. But marine productivity does appear with each other, with the larger metaweb en- to influence which economic forces have pre- compassing all four habitats, and with other cedence at any one time. For example, the food webs (Dunne et al. 2008). Early-stage collapse of the nineteenth-century sea otter research is making use of the initial intertidal stocks, as well as the decline in the Atlantic topological analyses to create scenarios for cod fishery (both human caused), certainly modeling the dynamics of complex ecologi- led to the success of the early twentieth- cal trophic networks (Brose et al. 2006) to century cod fishery in the North Pacific. But explore how humans’ roles and capabilities the collapse of the North Pacific cod fishery tended to promote or inhibit species and in the late 1930s, leading to cattle and other community persistence and other measures economic diversification on Sanak, and cod’s of ecosystem stability in this type of pre- resurgence in the mid-1970s as a dominant industrial coupled human-natural system. fishery appears to have been driven by oce- anic regime shifts (Maschner et al. 2008; discussion and conclusions Reedy-Maschner et al. unpubl. field notes). It should be recognizable that there is The late Holocene paleoecology of the west- some correlation between the human occupa- ern Gulf of Alaska, and the Sanak Island tion index (Figure 8), the distribution of sites region in particular, has shown considerable with preserved shell middens (Figure 7), and variation in general climate regimes, particu- the total organic carbon data (Figure 4). For- larly in regard to the Pacific Decadal Oscilla- aging models would predict that higher pop- tion (PDO). Trites et al. (2007) have shown, ulations would result in an expansion of the as have Finney et al. (in press), Misarti subsistence base, requiring an increase in (2007), and others, that these changing cli- the use of lower-ranked, small-package foods, matic regimes alter regional marine produc- such as shellfish. The correlation between the tivity in a number of complex ways and that HOI and the distribution of middens implies basic marine productivity is key to under- this to be true. But the periods with the great- standing the population histories of most est numbers of middens occur during cooler higher mammals (including humans) and oceanic regimes, the periods with the highest other species (Maschner et al. 2009). marine productivity. The total organic car- We suspect that the human occupation in- bon (TOC) data provide a key to this discrep- dex, which is identical to population trends in ancy. Decreasing TOC in nearshore cores the greater Alaska Peninsula region (Masch- is likely a product of increasing storminess. ner et al. 2009), highly similar to those in the Storms are the only condition that keeps tra- entire North Pacific (Maschner 1997, Ames ditional hunters and fishermen from going . Biocomplexity of Sanak Island Maschner et al. 699 offshore. So although marine productivity is is another line of evidence that leads us to up during cooler periods, increasing stormi- the same conclusion. But more important, the ness results in fewer days out hunting and size structure of modern populations does fishing, requiring the use of lower-ranked not appear to have been affected by modern species during storm periods. Conversely, commercial fishing, a strong case for the sus- this implies that during warmer, less-stormy tainability for that species and fishery (e.g., periods, the decreased marine productivity Maschner et al. 2008), a situation unlike the provided sufficient resources, except in the history of the North Atlantic (Betts et al. in worst of climatic regime shifts. press a). Elsewhere, we have argued (Maschner In nearshore areas, it appears that humans et al. 2009; H.D.G.M., A.W.T., K.L.R.-M., harvested large quantities of shellfish dur- M.W.B., A.T., and M.L., unpubl. data) that ing particular time periods, especially 3,600, one of the key factors in Aleut interactions 2,200, and 400 years ago, and that these shell- with sea mammals was their constant need fish were important contributions to the diet. for Steller sea lion skins, the only viable cov- Our first attempts to place the local and pre- ering for ocean-going kayaks. We wonder historic peoples in the structure of a dynamic if the close relationship over the last 1,500 food web show that humans have complex years between the human population index interactions at multiple levels. Harvesting and the otariid AI is not simply a product of over 40 taxa of intertidal species, the ancient the fact that Aleut males had to harvest up to Aleut are the most connected species in the six Steller sea lions every year to cover their intertidal food web. This means that re- kayaks, meaning that the actual sea lion har- moving the Aleut from the food web, as oc- vest is not dictated by diet but by the number curred both in the 1820s and then again in of Aleut males owning kayaks. the 1970s, probably had measurable organiz- Likewise, it is possible that the changes ing effects on the structure of the intertidal in the archaeological record of otter nitrogen system. and carbon stable isotopes are driven by Aleut It is also clear from the species harvested harvesting practices (Misarti 2007). Sea otters (M.W.B. and H.D.G.M., unpubl. data) that are a more-localized, nearshore species, and humans did have local impacts on the struc- hunting may have changed population struc- ture of the intertidal zone, harvesting lower- ture over time. If population size affects the ranked species as the intertidal zone was availability of otter prey then the Aleut may increasingly exploited. On the other hand, have inadvertently affected what food items there were periods when, despite large human were available to otters depending on how populations, shellfish were rarely harvested or heavily otters were hunted at any one time. not harvested at all. But it is important to Salmon on the other hand, although heavily note that although the modern intertidal harvested along the Alaska Peninsula, seem zone is different from the intertidal regimes to have been more affected by marine harvested by the prehistoric Aleut between productivity and climate regimes. Perhaps 5,000 and 400 years ago, the modern regime salmon were less impacted prehistorically by does not appear to be a product of human ac- numbers harvested by the Aleut and more by tivities but rather may be more closely caused changes to the marine ecosystems out in the by changing oceanic conditions. deep ocean. These observations are at best preliminary. The archaeological data concerning Pacific We are just now to the stage in the investiga- cod allow us to draw two critical conclusions. tions that more complex modeling efforts will The first is that minor variations in the popu- be possible and we may with confidence be- lation structure and size structure of Pacific gin addressing in detail the hypotheses put cod appear to be driven by overall marine forward at the beginning of this paper. But productivity, as measured in changing cli- one observation is clear—modern 25- to 50- matic regimes. The lack of change over year fisheries and climate data sets for the 4,500 years in the stable isotope data for cod North Pacific region represent little of the . 700 PACIFIC SCIENCE October 2009 dramatic shifts in climate, productivity, and Ames, K., and H. D. G. Maschner. 1999. human feedbacks that have occurred over the Peoples of the Northwest coast: Their last 5,000 years. Although there is some evi- prehistory and archaeology. Thames and dence forthcoming that humans have had Hudson, London. long-term impacts on the structure of the in- Amundsen, C. C. 1972. Amchitka bioenvir- tertidal zone, or even on offshore fisheries, onmental program: Plant ecology of what is clear is that much of the behavioral Amchitka—final report. Battelle Memorial ecologies of many species such as sea lions, Institute, Columbus, Ohio. or of the interactions among species around Bank, T. P. 1953. Ecology of prehistoric village sites, are products of both passive and Aleutian village sites. Ecology 34:246–264. active human engineering of those behaviors. Beaglehole, J. C. 1967. The journals of Cap- As such, at least for the last few thousands tain James Cook on his voyages of discov- of years, there is no ‘‘natural’’ North Pacific ery: The voyage of the Resolution and ecosystem without humans as one of the key Discovery, 1776–1780. 2 vols. Cambridge species in that ecosystem. University Press, Cambridge. Benson, A. J., and A. W. Trites. 2002. Eco- acknowledgments logical effects of regime shifts in the Be- ring Sea and eastern North Pacific. Fish We gratefully acknowledge the support and Fish. 3:95–113. access provided by the Sanak Corporation Betts, M., and T. M. Friesen. 2006. Declining and the Pauloff Harbor Tribe. We would es- foraging returns from an inexhaustible pecially like to thank the many undergraduate resource?: Abundance indices and beluga and graduate students who helped with field- whaling in the western Canadian Arctic. J. work and laboratory analyses used in this pa- Anthropol. Archaeol. 25:59–81. per; among these the contributions of Dieta Betts, M., H. D. G. Maschner, and D. Clark. Lund, Garrett Knudsen, Krystal Williams, In press a. Zooarchaeology of the ‘Fish Krista Williams, and Julie Kramer are ac- that Stops’: Using archaeofaunas to con- knowledged. We also thank the Idaho State struct long-term time series of Atlantic University Office of Research and the Re- and Pacific cod populations. In M. Moss search Vice-President for financial and infra- and A. Cannon, eds. The archaeology structure support. All opinions, findings, and of North Pacific fisheries. University of conclusions or recommendations expressed Alaska Press, Anchorage. in this material are those of the authors and Betts, M., H. D. G. Maschner, and V. Lech. do not necessarily reflect the views of the Na- In press b. A 4500 year time-series of ota- tional Science Foundation. riid abundance on Sanak Island, western Gulf of Alaska. In T. Braje and T. Rick, Literature Cited eds. Sea dogs of the North Pacific: The ar- chaeology and historical ecology of seals, Albers, W. D., and P. J. Anderson. 1985. Diet sea lions, sea cows, and sea otters. Univer- of Pacific cod, Gadus macrocephalus, and sity of California Press, Berkeley. predation on the northern pink shrimp, Black, L. 1999. Sanak. Pages 59–77 in L. Pandalus borealis, in Pavlof Bay, Alaska. Black, S. McGowan, J. Jacka, N. Taksami, Fish. Bull. 83:601–610. and M. Wright, eds. The history and eth- Ambrose, S. H., and L. Norr. 1993. Experi- nohistory of the Aleutians East Borough. mental evidence for the relationship of The Limestone Press, Kingston, Ontario, the carbon isotope ratios of whole diet Canada. and dietary protein to those of bone colla- Bodkin, J. L., G. G. Esslinger, and D. H. gen and carbonate. Pages 1–38 in J. B. Monson. 2004. Foraging depths of sea Lambert and G. Grupe, eds. Prehistoric otters and implications to coastal marine human bone archaeology at the molecular communities. Mar. Mamm. Sci. 20:305– level. Springer-Verlag, New York. 321. . Biocomplexity of Sanak Island Maschner et al. 701

Boutton, T. W. 1991. Stable carbon isotope seabird declines. Alaska Sea Grant Rep. ratios of natural materials: II. Atmo- 93-01. spheric, terrestrial, marine, and freshwater Chamberlain, A. 2006. Demography in ar- environments. Pages 173–185 in D. C. chaeology. Cambridge Manuals in Archae- Coleman and B. Fry, eds. Carbon isotope ology. Cambridge University Press, techniques. Academic Press, San Diego. Cambridge. Brantingham, P. J., K. W. Kerry, A. I. Charnov, E., G. Orians, and K. Hyatt. 1976. Krivoshapkin, and Y. V. Kuzmin. 2004. Ecological implications of resource de- Time-space dynamics in the early-Upper pression. Am. Nat. 110:247–259. Paleolithic of northern Asia. Pages 255– Chesson, P., and N. Huntly. 1994. Temporal 283 in D. B. Madsen, ed. Entering Amer- hierarchies of variation and the mainte- ica: Northeast Asia and Beringia before nance of diversity. Plant Species Biol. the Last Glacial Maximum. University of 8:195–206. Utah Press, Salt Lake City. ———. 1997. The roles of harsh and fluctu- Brose, U., R. J. Williams, and N. D. Marti- ating conditions in the dynamics of eco- nez. 2006. Allometric scaling enhances sta- logical communities. Am. Nat. 150:519– bility in complex food webs. Ecol. Lett. 553. 9:1228–1236. Ciannelli, L., and K. M. Bailey. 2005. Land- Broughton, J. 1994. Declines in mammalian scape dynamics and resulting species in- foraging efficiency during the late Holo- teractions: The cod-capelin system in the cene, San Francisco Bay, California. J. southeastern Bering Sea. Mar. Ecol. Prog. Anthropol. Archaeol. 13:371–401. Ser. 291:227–236. ———. 1997. Widening diet breadth, de- Clarke, K. R. 1993. Non-parametric multi- clining foraging efficiency, and prehistoric variate analyses of changes in community harvest pressure: Ichthyofaunal evidence structure. Aust. J. Ecol. 18:117–143. from the Emeryville Shellmound, Califor- Crumley, C. L., ed. 1994. Historical ecology: nia. Antiquity 71:845–862. Cultural knowledge and changing land- ———. 1999. Resource depression and in- scapes. School of American Research tensification during the late Holocene, Advanced Seminar Series. School of San Francisco Bay: Evidence from the American Research Press, Santa Fe, New Emeryville Shellmound vertebrate fauna. Mexico. Univ. Calif. Anthropol. Rec. 32. Dale, V. H., S. Brown, R. Haeuber, N. T. Butler, V. 2000. Resource depression on the Hobbs, N. Huntly, R. Naiman, W. Rieb- Northwest coast of North America. Antiq- same, M. Turner, and T. J. Valone. 2000. uity 74:649–661. Ecological principles and guidelines for Calkin, P. E., G. Wiles, and D. J. Barclay. managing the use of land. Ecol. Appl. 2001. Holocene coastal glaciation of 10:639–670. Alaska. Quat. Sci. Rev. 20:449–461. D’Arrigo, R., E. Mashig, D. Frank, R. Wil- Cannon, M. 1999. A mathematical model of son, and G. Jacoby. 2005. Temperature the effects of screen size on zooarchaeo- variability over the past millennium in- logical relative abundance measures. J. Ar- ferred from northwestern Alaska tree chaeol. Sci. 26:205–214. rings. Clim. Dyn. 24:227–236. ———. 2000. Large mammal relative abun- Detterman, R. L. 1986. Glaciation of the dance in Pithouse and Pueblo Period Alaska Peninsula. Pages 151–170 in T. D. Archaeofaunas from southwestern New Hamilton, K. M. Reed, and R. M. Thor- Mexico: Resource depression among the son, eds. Glaciation in Alaska: The geo- Mimbres-Mogollon? J. Anthropol. Ar- logic record. Alaska Geological Society, chaeol. 19:317–347. Anchorage. Castellini, M. 1993. Report on marine mam- Dochat, T. 1997. The glacial history of the mal working group. Pages 4–13 in Is it lower Alaska Peninsula. Ph.D. diss., Uni- food?: Addressing marine mammal and versity of Wisconsin, Madison. . 702 PACIFIC SCIENCE October 2009

Doucett, R. R., G. Power, D. R. Barton, R. J. ity in marine fish populations: A long- Drimmie, and R. A. Cunjak. 1996. Stable term perspective on the impacts of climate isotope analysis of nutrient pathways lead- change on marine fisheries. J. Mar. Syst. ing to Atlantic salmon. Can. J. Fish. Aquat. GLOBEC Special Issue. Sci. 53:2058–2066. Finney, B. P., I. Gregory-Eaves, M. S. V. Duggins, D. O., C. A. Simenstad, and J. A. Douglas, and J. P. Smol. 2002. Fisheries Estes. 1989. Magnification of secondary productivity in the northeastern Pacific production by kelp detritus in coastal Ocean over the past 2,200 years. Nature marine ecosystems. Science (Washington, (Lond.) 416:729–733. D.C.) 245:170–173. Finney, B. P., I. Gregory-Eaves, J. Sweetman, Dunne, J. A., R. J. Williams, and N. D. M. S. V. Douglas, and J. Smol. 2000. Im- Martinez. 2002. Network structure and pacts of climatic change and fishing on biodiversity loss in food webs: Robustness Pacific salmon abundance over the past increases with connectance. Ecol. Lett. 300 years. Science (Washington, D.C.) 5:558–567. 290:795–799. ———. 2004. Network structure and robust- Fitzhugh, J. B. 2003. The evolution of com- ness of marine food webs. Mar. Ecol. plex hunter-gatherers: Archaeological evi- Prog. Ser. 273:291–302. dence from the North Pacific. Kluwer- Dunne, J. A., R. J. Williams, N. D. Martinez, Plenum, New York. R. A. Wood, and D. E. Erwin. 2008. Fry, B., and E. B. Sherr. 1984. d13C Mea- Compilation and network analyses of surements as indicators of carbon flow in Cambrian food webs. PLoS. Biol. 5:e102. marine and freshwater ecosystems. Con- doi:10.1371/journal.pbio.0060102. trib. Mar. Sci. 27:13–47. Ellis, J. E., and K. Galvin. 1994. Climate Fryer, G. J., P. Watts, and L. F. Pratson. patterns and land use practices in the 2004. Source of the great tsunami of 1 dry zones of Africa. BioScience 44:340– April 1946: A landslide in the upper Aleu- 349. tian forearc. Mar. Geol. 203:201–218. Ellis, J. E., and D. M. Swift. 1988. Stability Gedamke, T., and J. M. Hoenig. 2006. Esti- of African pastoral ecosystems: Alternate mating mortality from mean length data paradigms and implications for develop- in nonequilibrium situations, with applica- ment. J. Range Manage. 41:450–459. tion to the assessment of goosefish. Trans. Erlandson, J. M. 1988. The role of shellfish Am. Fish. Soc. 135:476–487. in prehistoric economies: A protein per- Graham, M. H., P. K. Dayton, and J. M. spective. Am. Antiq. 53:102–109. Erlandson. 2003. Ice ages and ecological Estes, J. A., and D. O. Duggins. 1995. Sea transitions on temperate coasts. Trends otters and kelp forests in Alaska: General- Ecol. Evol. 18:33–40. ity and variation in a community ecological Grayson, D. 2001. Explaining the develop- paradigm. Ecol. Monogr. 65:75–100. ment of dietary dominance by a single un- Estes, J. A., M. L. Riedman, M. M. Staedler, gulate taxon at Grotte XVI, Dordogne, M. T. Tinker, and B. E. Lyon. 2003. Indi- France. J. Archaeol. Sci. 28:115–125. vidual variation in prey selection by sea Hassan, F. A. 1978. Demographic archaeol- otters: Patterns, causes and implications. ogy. Pages 49–103 in M. B. Shiffer, ed. J. Anim. Ecol. 72:144–155. Advances in archaeological method and Finney, B. P. 1998. Long-term variability theory. Vol. I. Academic Press, New York. in Alaskan sockeye salmon abundance de- ———. 1981. Demographic archaeology. termined by analysis of sediment cores. Academic Press, New York. North Pac. Anadromous Fish. Comm. Heusser, C. J. 1990. Late Quaternary vegeta- Bull. 1:388–395. tion of the Aleutian Islands, southwestern Finney, B. P., J. Alheit, K. C. Emeis, D. B. Alaska. Can. J. Bot. 68:1320–1326. Field, D. Guitierrez, and U. Struck. In Heusser, C. J., L. E. Heusser, and D. M. press. Paleoecological studies on variabil- Peteet. 1985. Late-Quaternary climate . Biocomplexity of Sanak Island Maschner et al. 703

change on the American North Pacific last two millennia. Proc. Natl. Acad. Sci. coast. Nature (Lond.) 315:485–487. U.S.A. 98:10552–10556. Hildebrandt, W., and T. Jones. 1992. Evolu- Jackson, J. B. C., M. X. Kirby, W. H. Berger, tion of marine mammal hunting: A view K. A. Bjorndal, L. W. Botsford, B. J. Bour- from the California and Oregon coasts. J. que, R. H. Bradbury, R. Cooke, J. Erland- Anthropol. Archaeol. 11:360–401. son, J. A. Estes, T. P. Hughes, S. Kidwell, Hirons, A. C. 2001. Trophic dynamics of C. B. Lange, H. S. Lenihan, J. M. Pan- pinniped populations in Alaska using stable dolfi, C. H. Peterson, R. S. Steneck, M. J. carbon and nitrogen isotope ratios. Ph.D. Tegner, and R. R. Warner. 2001. Histori- diss., University of Alaska, Fairbanks. cal overfishing and the recent collapse of Hirons, A. C., D. M. Schell, and B. P. Finney. coastal ecosystems. Science (Washington, 2001. Temporal records of d13C and d15N D.C.) 293:629–638. in North Pacific pinnipeds: Inferences James, T. S. 2001. Postglacial rebound analy- regarding environmental change and diet. sis for better determination of earthquake- Oecologia (Berl.) 129:591–601. related deformation and rheology at sub- Hobson, K. A., D. M. Schell, D. Renouf, and duction zones. USGS/NEHRP Award E. Noseworthy. 1996. Stable carbon and 01HQGR0005. Program Element II: Re- nitrogen isotopic fractionation between search on Earthquake Occurrence and Ef- diet and tissues of captive seals: Implica- fects. http://earthquake.usgs.gov/research/ tions for dietary reconstructions involving external/reports/01HQGR0005.pdf marine mammals. Can. J. Fish. Aquat. Sci. Janetski, J. 1997. Fremont hunting and re- 53:528–533. source intensification in the eastern Great Hobson, K. A., J. L. Sease, R. L. Merrick, and Basin. J. Archaeol. Sci. 24:1075–1088. J. F. Piatt. 1997. Investigating trophic Johnson, D. 2003. Species composition, di- relationships of pinnipeds in Alaska and versity, and productivity of plant commu- Washington using stable isotope ratios of nities on Prehistoric Aleut village sites on nitrogen and carbon. Mar. Mamm. Sci. the western Alaska Peninsula. M.S. thesis, 13:114–132. Idaho State University, Pocatello. Hobson, K. A., and H. E. Welch. 1992. De- Johnson, L. L. 1988. Archaeological surveys termination of trophic relationships within of the outer Shumagin Islands, Alaska, a high arctic marine food web using d13C 1984, 1986. Arct. Anthropol. 25:139–170. and d15N analysis. Mar. Ecol. Prog. Ser. ———. 1992. Prehistoric settlement patterns 84:9–18. and population in the Shumagin Islands. Hoffman, B. W. 2002. The organization of Anthropol. Pap. Univ. Alsk. 24:73–88. complexity: A study of late prehistoric Johnson, L. L., and E. G. Wilmerding. 2001. village organization in the eastern Aleutian Bringing down the house: Modeling con- region, Alaska. Ph.D. diss., University of struction and deconstruction of Aleut Wisconsin, Madison. semisubterranean houses. Pages 127–149 Holdaway, S., P. Fanning, and J. Shiner. in D. E. Dumond, ed. Archaeology in the 2005. Absence of evidence or evidence of Aleut zone of Alaska: Some recent re- absence? Understanding the chronology search. Univ. Oreg. Anthropol. Pap. 58. of indigenous occupation of western New Johnson, L. L., and M. A. Winslow. 1991. South Wales, Australia. Archaeol. Oceania Paleoshorelines and prehistoric settle- 40:33–49. ments on Simeonof and Chernabura Is- Holland, K. M. 1992. Rethinking Aleutian lands, outer Shumagin Islands, Alaska. In prehistory as viewed from a lithic data- L. L. Johnson, ed. Paleoshorelines and base. Ph.D. diss., Arizona State University, prehistory: An investigation of method. Tempe. CRC Press, Boca Raton, Florida. Hu, F. S., E. Ito, T. A. Brown, B. B. Curry, Jones, C. G., and J. H. Lawton. 1995. Link- and D. R. Engstrom. 2001. Pronounced ing species and ecosystems. Chapman and climatic variations in Alaska during the Hall, New York. . 704 PACIFIC SCIENCE October 2009

Jones, C. G., J. H. Lawton, and M. Shachak. physical Research, Documentation No. 1997. Positive and negative effects of or- 31. National Geophysical Data Center, ganisms as physical ecosystem engineers. Boulder, Colorado. Ecology 78:1946–1957. Lantis, M. 1984. Aleut. Pages 161–184 in D. Jordan, J. W. 2000. Coastal paleogeography Damas, ed. Handbook of North American of the western Alaska Peninsula. Ph.D. Indians. Vol. 5. Arctic. Smithsonian Insti- diss., University of Wisconsin, Madison. tution Press, Washington, D.C. ———. 2001. Late Quaternary sea level Lech, V., M. Betts, and H. Maschner. In change in southern Beringia: Post-glacial press. An analysis of seal, sea lion, and sea emergence of the western Alaska Penin- otter consumption patterns on Sanak Is- sula. Quat. Sci. Rev. 20:509–523. land, Alaska: A 4500 year record on Aleut Jordan, J. W., B. P. Finney, and H. D. G. consumer behavior. In T. Braje and T. Maschner. 2007. Geomorphic history of Rick, eds. Sea dogs of the North Pacific: the Sanak Islands: Implications for ecosys- The archaeology and historical ecology of tem change and human settlement. Society seals, sea lions, sea cows, and sea otters. for American Archaeology, 72nd Annual University of California Press, Berkeley. Meeting, Austin. Liapunova, R. G. 1996. Essays on the ethnog- Jordan, J. W., and A. Krumhardt. 2003. Post- raphy of the Aleuts (at the end of the eigh- glacial climate and vegetation history of teenth and the first half of the nineteenth the western Alaska Peninsula. Alsk. J. century). J. Shelest, transl. University of Anthropol. 1:16–33. Alaska Press, Fairbanks. Jordan, J. W., and H. D. G. Maschner. 2000. Link, J. 2002. Does food web theory work for Coastal paleogeography and human occu- marine ecosystems? Mar. Ecol. Prog. Ser. pation of the lower Alaska Peninsula. Geo- 230:1–9. archaeology 15:385–414. Loso, M. G., R. S. Anderson, S. P. Anderson, Khlebnikov, K. T. 1994. Notes on Russian and P. J. Reimer. 2006. A 1500-year rec- America, Parts II–V: Kad’iak, Unalashka, ord of temperature and glacial response Atkha, the Pribylovs. Compiled by R. G. inferred from Varved Iceberg Lake, Liapunova and S. G. Fedorova. M. Ram- Southcentral Alaska. Quat. Res. 66:12–24. say, transl. R. A. Pierce, ed. Limestone Lyman, R. L. 2003. The influence of time Press, Kingston, Ontario, Canada. [Origi- averaging and space averaging on the ap- nally published in 1979 in Russian.] plication of foraging theory in zooarch- King, A. A., and W. M. Schaffer. 2001. The aeology. J. Archaeol. Sci. 30:595–610. geometry of a population cycle: A mecha- ———. 2008. Quantitative paleozoology. nistic model of snowshoe hare demogra- Cambridge University Press, New York. phy. Ecology 82:814–830. Mann, D. H. 2001. Climate during the past Kline, T. C., and T. M. Willette. 2002. millennium. Weather 56:91–102. Pacific salmon (Oncorhynchus spp.) early Mann, D. H., and T. D. Hamilton. 1995. marine feeding patterns based on N-15/ Late Pleistocene and Holocene paleo- N-14 and C-13/C-12 in Prince William environments of the North Pacific coast. Sound, Alaska. Can. J. Fish. Aquat. Sci. Quat. Sci. Rev. 14:449–471. 59:1626–1638. Mann, D. H., and D. M. Peteet. 1994. Extent Kolb, C. C. 1985. Demographic estimates and timing of the last glacial maximum in in archaeology: Contribution from ethno- southwestern Alaska. Quat. Res. 42:136– archaeology on Mesoamerican peasants. 148. Curr. Anthropol. 26:581–599. Maschner, H. D. G. 1997. Settlement and Lambert, J. B., and G. Grupe. 1993. Prehis- subsistence in the later prehistory of Te- toric human bone: Archaeology at the mo- benkof Bay, Kuiu Island. Arct. Anthropol. lecular level. Springer-Verlag, New York. 34:4–99. Lander, J. F. 1996. Tsunamis affecting ———. 1998. Salmon run volatility, subsis- Alaska: 1737–1996. NGDC Key to Geo- tence, and the development of North Pa- . Biocomplexity of Sanak Island Maschner et al. 705

cific societies. Pages 11–28 in Hokkaido Maschner, H. D. G., B. Finney, J. Jordan, A. Museum of Northern Peoples, ed. Pro- Tews, N. Misarti, and G. Knudsen. 2009. ceedings of the 12th International Abashiri Did the North Pacific ecosystem collapse Symposium: Salmon fishery in the north in AD 1000? In H. D. G. Maschner, and its change through time. Association O. W. Mason, and R. McGhee, eds. The for the Promotion of Northern Cultures, northern world AD 900–1400. University Abashiri, Hokkaido, Japan. of Utah Press, Salt Lake City. Pp. 33–57. ———. 1999a. Prologue to the prehistory of Maschner, H. D. G., and B. W. Hoffman. the lower Alaska Peninsula. Arct. Anthro- 2003. The development of large corporate pol. 36:84–102. households along the North Pacific Rim. ———. 1999b. Sedentism, settlement and Alsk. J. Anthropol. 1:41–63. village organization on the lower Alaska Maschner, H. D. G., and J. W. Jordan. 2001. Peninsula: A preliminary assessment. The Russell Creek manifestation of the Pages 56–76 in B. Billman and G. Fein- Arctic Small Tool Tradition on the west- man, eds. Settlement pattern studies in ern Alaska Peninsula. Pages 151–172 in the Americas: Fifty years since Viru. D. E. Dumond, ed. Archaeology in the Smithsonian Institution Press, Washing- Aleut zone of Alaska: Some recent re- ton, D.C. search. Univ. Oreg. Anthropol. Pap. 58. ———. 2000. Catastrophic change and re- ———. 2008. Catastrophic events and punc- gional interaction: The southern Bering tuated culture change: The southern Be- Sea in a dynamic world system. Pages ring Sea and North Pacific in a dynamic 252–265 in M. Appelt, J. Berglund, and global system. In D. Papagianni, R. H. H. C. Gulløv, eds. Identities and cultural Layton, and H. D. G. Maschner, eds. contacts in the Arctic. Proceedings from a Time and change: Archaeological and an- Conference at the Danish National Mu- thropological perspectives on the long seum, Copenhagen, 30 November–2 De- term. Oxbow, Oxford, United Kingdom. cember 1999. Danish National Museum Pp. 95–113. and Danish Polar Center, Copenhagen. Maschner, H. D. G., and K. L. Reedy- ———. 2004a. Redating the Hot Springs Maschner. 1998. Raid, retreat, defend (re- Village site. Alsk. J. Anthropol. 2:100–116. peat): The archaeology and ethnohistory ———. 2004b. Traditions past and present: of warfare on the North Pacific. J. Anthro- Allen McCartney and the Izembek Phase pol. Archaeol. 17:19–51. of the western Alaska Peninsula. Arct. ———. 2005. Aleuts and the sea. Archaeol- Anthropol. 41:98–111. ogy. March/April, 63–70. ———. 2008. Fishtails, ancestors, and old is- Mason, O. K., and J. W. Jordan. 1993. landers: Chirikof Island, the Alaska Penin- Heightened North Pacific storminess dur- sula, and the dynamics of western Alaska ing synchronous late Holocene erosion of prehistory. Alsk. J. Anthropol. 6:171–184. Northwest Alaska beach ridges. Quat. Res. Maschner, H. D. G., and R. A. Bentley. 2003. 40:55–69. The power law of rank and household on Masters, P. M., and D. R. Gallegos. 1997. the North Pacific. Pages 47–60 in R. A. Environmental change and coastal adapta- Bentley and H. D. G. Maschner, eds. tions in San Diego County during the Complex systems and archaeology: Empir- Middle Holocene. Pages 11–22 in J. M. ical and theoretical applications. Univer- Erlandson and M. A. Glassow, eds. Ar- sity of Utah Press, Salt Lake City. chaeology of the California coast during Maschner, H. D. G., M. W. Betts, K. L. the Middle Holocene. Institute of Ar- Reedy-Maschner, and A. W. Trites. 2008. chaeology, University of California, Los A 4500 year time series of Pacific cod Angeles. (Gadus macrocephalus): Archaeology, re- McCartney, A. P. 1969. Prehistoric Aleut in- gime shifts, and sustainable fisheries. Fish. fluences at Port Moller, Alaska. Anthropol. Bull. 106:386–394. Pap. Univ. Alsk. 14:1–16. . 706 PACIFIC SCIENCE October 2009

———. 1974. Prehistoric cultural integration tions in 1901. Pages 173–398 in Bull. U.S. along the Alaska Peninsula. Anthropol. Fish Comm. Vol. XXI for 1901. Pap. Univ. Alsk. 16:59–84. Nunn, P. D. 2007. Climate, environment, ———. 1984. Prehistory of the Aleutian and society in the Pacific during the last region. Pages 119–135 in D. Damas, ed. millennium. Vol. 6. Developments in Handbook of North American Indians. earth and environmental sciences. Elsevier Vol. 5. Arctic. Smithsonian Institution Science, Amsterdam. Press, Washington, D.C. Ocean Studies Board and Polar Research ———. 1988. Maritime adaptation in south- Board. 2003. The decline of the Steller ern Alaska. Pages 19–51 in H. Okada, ed. sea lion in Alaskan waters: Untangling Proceedings of the International Sympo- food webs and fishing nets. National Aca- sium on Maritime Adaptations in the demies Press, Washington, D.C. North Pacific. City of Abashiri, Hokkaido, Okada, H. 1980. Prehistory of the Alaska Japan. Peninsula as seen from the Hot Springs McCartney, A. P., and D. W. Veltre. 1999. Site, Port Moller. Pages 103–112 in Y. Aleutian Island prehistory: Living in insu- Kotani and W. B. Workman, eds. Alaska lar extremes. World Archaeol. 30:503– Native culture and history. Senri Ethno- 515. logical Studies No. 4. National Museum McCroy, C. P., R. Gradinger, A. M. of Ethnology, Osaka. Springer, B. Iverson, S. Iverson, and S. O’Leary, M. 2002. Estimating casualties for Budge. 2004. Stable isotopes reveal food the Aleut resistance, 1763–1775. Paper web shifts due to arctic sea ice decline. presented at the Annual Alaska Anthropo- Paper presented at AGU Joint Assembly, logical Association meeting, April 2002, Montreal, Canada. Anchorage, Alaska. Michener, R. H., and D. M. Schell. 1994. Orchard, T. 2001. The role of selected fish Stable isotope ratios as tracers in marine species in Aleut palaeodiet. M.S. thesis, aquatic food webs. Pages 136–156 in K. University of Victoria, Victoria, British Lajtha and R. H. Michener, eds. Stable Columbia. isotopes in ecology and environmental ———. 2003. An application of the linear science. Blackwell Scientific Publication, regression technique for determining London. length and weight of six fish taxa: The Misarti, N. 2007. Six thousand years of role of selected fish species in Aleut pa- change in the Northeast Pacific: An inter- laeodiet. Archaeopress, Oxford, United disciplinary view of maritime ecosystems. Kingdom. Ph.D. diss., University of Alaska, Fair- Paine, R. T. 1971. The measurement and ap- banks. plication of the calorie to ecological prob- Misarti, N., B. Finney, H. Maschner, and M. lems. Annu. Rev. Ecol. Syst. 2:145–164. Wooller. In press. Changes in a Northeast Peltier, W. R. 1994. Ice-age paleotopog- Pacific marine ecosystem over 4500 years: raphy. Science (Washington, D.C.) 265: Evidence from stable isotope analysis of 195–201. bone collagen from archaeological mid- Polis, G. A., W. B. Anderson, and R. D. Holt. dens. The Holocene. 1997. Toward an integration of landscape Moore, J. C. 1974. Geologic and structural and food web ecology: The dynamics of map of the Sanak Islands, southwestern spatially subsidized food webs. Annu. Rev. Alaska. U.S. Geol. Surv. Misc. Invest. Ser. Ecol. Syst. 28:289–316. Map I-817. Polis, G. A., and S. D. Hurd. 1996. Linking Morrison, M. L., B. G. Marcot, and R. W. marine and terrestrial food webs: Allochth- Mannan. 1998. Wildlife-habitat relation- onous input from the ocean supports high ships, concepts and applications. 2nd ed. secondary productivity on small islands University of Wisconsin Press, Madison. and coastal land communities. Am. Nat. Moser, J. F. 1902. Alaska salmon investiga- 147:396–423. . Biocomplexity of Sanak Island Maschner et al. 707

Polis, G. A., and K. O. Winemiller. 1996. mal and bird populations of the Bering Food webs: Integration of patterns and Sea region over the last several millennia. dynamics. Chapman and Hall, New York. Palaeogeogr. Palaeoclimatol. Palaeoecol. Polovina, J. J. 2005. Climate variation, regime 209:335–352. shifts, and implications for sustainable fish- Schell, D. M. 2001. Carbon isotope ratio eries. Bull. Mar. Sci. 76:233–244. variations in Bering Sea biota: The role Post, D. M. 2002. Using stable isotopes to es- of anthropogenic carbon dioxide. Limnol. timate trophic position: Models, methods, Oceanogr. 46:999–1000. and assumptions. Ecology 83:703–718. Schoeninger, M. J., and M. DeNiro. 1984. Power, M. E., D. Tilman, J. A. Estes, B. A. Nitrogen and carbon isotopic composition Menge, W. J. Bond, L. S. Mills, G. Daily, of bone collagen from marine and terres- J. C. Castilla, J. Lubchenco, and R. T. trial animals. Geochim. Cosmochim. Acta Paine. 1998. Challenges in the quest for 48:625–639. keystones. BioScience 46:609–620. Shacklette, H. T., ed. 1969. Vegetation of Reedy-Maschner, K. L. 2001. Aleut identity Amchitka Island, Aleutian Islands, Alaska. and indigenous commercial economies: U.S. Geol. Surv. Prof. Pap. 648. Local responses under global pressures in Shields, Captain E. 2001. Salt of the sea: The the eastern Aleutians. Alsk. J. Anthropol. Pacific Coast cod fishery and the last days 1:62–82. of sail. Pacific Heritage Press, Lopez Is- ———. 2004. Aleut identity and indigenous land, Washington. commercial fisheries. Ph.D. diss., Cam- Shin, Y.-J., M.-J. Rochet, S. Jennings, J. G. bridge University, Cambridge, United Field, and H. Gislason. 2005. Using size- Kingdom. based indicators to evaluate the ecosys- ———. 2007. The best-laid plans: Limited tem effects of fishing. ICES J. Mar. Sci. entry permits and limited entry systems 62:384–396. in eastern Aleut culture. Hum. Organ. Simenstad, C. A., J. A. Estes, and K. W. Ken- 66:210–225. yon. 1978. Aleuts, sea otters, and alternate Reisewitz, S. E., J. A. Estes, and C. A. Simen- stable-state communities. Science (Wash- stad. 2006. Indirect food web interactions: ington, D.C.) 200:403–411. Sea otters and kelp forest fishes in the Steneck, R. S., M. H. Graham, B. J. Bourque, Aleutian archipelago. Oecologia (Berl.) D. Corbett, J. M. Erlandson, J. A. Estes, 146:623–631. and M. J. Tegner. 2002. Kelp forest eco- Reiss, M. J. 1989. The allometry of growth systems: Biodiversity, stability, resilience and reproduction. Cambridge University and future. Environ. Conserv. 29:436–459. Press, Cambridge, United Kingdom. Stephens, D., and J. Krebs. 1986. Forag- Rich, W. H., and E. M. Ball. 1930. Statistical ing theory. Princeton University Press, review of the Alaska salmon fisheries. Part Princeton, New Jersey. I: Bristol Bay and the Alaska Peninsula. Surovell, T. A., and P. J. Brantingham. 2007. Bull. U.S. Bur. Fish. Vol. XLIV, 1928. A note on the use of temporal frequency Rick, T., and J. Erlandson. 2008. Human distributions in studies of prehistoric de- impacts on ancient marine ecosystems: A mography. J. Archaeol. Sci. 34:1868–1877. global perspective. University of Califor- Thompson, G. G., M. W. Dorn, and D. G. nia Press, Berkeley. Nichol. 2006. Assessment of the Pacific Satterfield, F. R. 2000. A comparison of sock- cod stock in the Gulf of Alaska. Pages eye salmon (Oncorhyncis nerka) in two cli- 147–220 in Stock assessment and fishery mate regimes in the North Pacific Ocean evaluation report for the groundfish re- using stable carbon and nitrogen isotope sources of the Gulf of Alaska. North ratios. Institute of Marine Science, Fair- Pacific Fisheries Management Council, banks, Alaska. Anchorage, Alaska. Savinetsky, A. B., N. K. Kiseleva, and B. F. Thompson, G. G., J. N. Ianelli, M. W. Dorn, Khassanov. 2004. Dynamics of sea mam- and M. Wilkins. 2007. Assessment of the . 708 PACIFIC SCIENCE October 2009

Pacific cod stock in the Gulf of Alaska. Port Moller, Alaska. Am. Mus. Nat. Hist. Pages 169–194 in Stock assessment and Anthropol. Pap. 31:239–279. fishery evaluation report for the ground- Wheeler, A., and A. Jones. 1989. Fishes. fish resources of the Gulf of Alaska. North Cambridge University Press, Cambridge, Pacific Fisheries Management Council, United Kingdom. Anchorage, Alaska. Wiessner, P. 1974. A functional estimator of Townsend, J. B. 1980. Ranked societies of population from floor area. Am. Antiq. the Alaskan Pacific Rim. Senri Ethnologi- 39:343–349. cal Studies 4:123–156. Williams, R. J., E. L. Berlow, J. A. Dunne, ———. 1983. Pre-contact political organiza- A. L. Baraba´si, and N. D. Martinez. 2002. tion and slavery in Aleut societies. Pages Two degrees of separation in complex 120–132 in E. Tooker, ed. The develop- food webs. Proc. Natl. Acad. Sci. U.S.A. ment of political organization in Native 99:12913–12916. North America. Proc. Am. Ethnol. Soc. Williams, R. J., and N. D. Martinez. 2004a. 1979. Stabilization of chaotic and non- Trenhaile, A. S. 2002. Rock coasts, with par- permanent food-web dynamics. Eur. Phys. ticular emphasis on shore platforms. Geo- J. B 38:297–303. morphology 48:7–22. ———. 2004b. Trophic levels in complex Trites, A. W., A. J. Miller, H. D. G. Masch- food webs: Theory and data. Am. Nat. ner, M. A. Alexander, S. J. Bograd, J. A. 163:458–468. Calder, A. Capotondi, K. O. Coyle, E. D. Wilson, F. H., and F. R. Weber. 1999. Qua- Lorenzo, B. P. Finney, E. J. Gregr, C. E. ternary geology of the Cold Bay and False Grosch, S. R. Hare, G. L. Hunt, J. Pass quadrangles, Alaska Peninsula. Geo- Jahncke, N. B. Kachel, H.-J. Kim, C. logic Studies in Alaska by the U.S. Geo- Ladd, N. J. Mantua, C. Marzban, W. Mas- logical Survey. U.S. Geol. Surv. Prof. lowski, R. Mendelssohn, D. J. Neilson, Pap. 1633:51–71. S. R. Okkonen, J. E. Overland, K. L. Wilson, F. H., F. R. Weber, T. M. Dochat, Reedy-Maschner, T. C. Royer, F. B. T. P. Miller, and R. L. Detterman. 1998. Schwing, J. X. L. Wang, and A. J. Win- Revised geologic map of the Cold Bay ship. 2007. Bottom-up forcing and the and False Pass quadrangles, Alaska Penin- decline of Steller sea lions (Eumetopias sula. Open File Report 97-866. U.S. Geo- jubatus) in Alaska: Assessing the ocean cli- logical Survey, Washington, D.C. mate hypothesis. Fish. Oceanogr. 16:46– Workman, W. B. 1966. Prehistory at Port 67. Moller, Alaska Peninsula, in the light Turchin, P. 2003. Historical dynamics: Why of fieldwork in 1960. Arct. Anthropol. states rise and fall. Princeton Studies in 3:132–153. Complexity. Princeton University Press, Worm, B., and R. A. Myers. 2003. Meta- Princeton, New Jersey. analysis of cod-shrimp interactions reveals Veniaminov, I. 1984. Notes on the islands of top-down control in oceanic food webs. the Unalaska District. [1840]. L. T. Black Ecology 84:162–173. and R. H. Goeghegan, transls. R. A. Yang, M. S. 2004. Diet changes of Pacific cod Pierce, ed. Alaska history, No. 27. The (Gadus macrocephalus) in Pavlof Bay associ- Limestone Press, Kingston, Ontario, Can- ated with climate changes in the Gulf of ada. Alaska between 1980 and 1995. Fish. Bull. Watt, J., D. B. Siniff, and J. A. Estes. 2000. 102:400–405. Inter-decadal patterns of population and Yesner, D. R. 1980. Maritime hunter- dietary change in sea otters at Amchitka gatherers: Ecology and prehistory. Curr. Island, Alaska. Oecologia (Berl.) 124:289– Anthropol. 21:727–750. 298. ———. 1981. Archaeological applications of Weyer, E. M., Jr. 1930. Archaeological mate- optimal foraging theory: Harvest strategies rial from the village site at Hot Springs, of Aleut hunter-gatherers. Pages 148–170 . Biocomplexity of Sanak Island Maschner et al. 709

in B. Winterhalder and E. A. Smith, eds. 285–310 in M. Harris and E. Ross, eds. Hunter-gatherer foraging strategies. Uni- Food and evolution. Temple University versity of Chicago Press, Chicago. Press, Philadelphia, Pennsylvania. ———. 1985. Cultural boundaries and eco- ———. 1988. Effects of prehistoric human logical frontiers in coastal regions: An ex- exploitation on Aleutian sea mammal pop- ample from the Alaska Peninsula. Pages ulations. Arct. Anthropol. 25:28–43. 51–91 in S. W. Green and S. M. Perlman, ———. 1998. Origins and development of eds. The archeology of frontiers and maritime adaptations in the Northwest Pa- boundaries. Academic Press, New York. cific region of North America: A zooarch- ———. 1987. Life in the Garden of Eden: aeological perspective. Arct. Anthropol. Causes and consequences of the adoption 35:204–222. of marine diets by human societies. Pages