A sub-centennial, Little Ice Age climate reconstruction using beetle subfossil data from Nunalleq, southwestern Alaska

This research was funded through an Arts and Humanities Research Council grant (AH/K006029/1) awarded to Drs. Rick Knecht, Charlotta Hillerdal and Kate Britton, the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 703322, and two NERC Radiocarbon Facility grants (NF/2015/1/6 and NF/2015/2/3) awarded to Drs. Rick Knecht and Paul Ledger. Our work at Nunalleq also benefitted from the support of the local community who made us all feel at home in Quinhagak and provided consistently warm hospitality. Logistical support from Warren Jones and Qanirtuuq Incorporated was also invaluable. VF would like to thank Anthony Davies for his invaluable help with the identification of Stenus species and for verifying her identification of other staphylinid taxa. Patrice Bouchard, Yves Bousquet, Henri Goulet and Ales Smetana are also thanked for help provided with beetle identifications, as well as Professor Ian Foster, for kindly providing Pb210 dating support. Finally, we would like to thank Peter Jordan for the opportunity to publish in this special issue, as well as Philippe Ponel and an anonymous reviewer, whose comments helped us improve the original manuscript.

To date, there are no palaeo-temperature reconstructions from the vast expanse of 23 the Y-K delta. Therefore, in order to facilitate an exploration of how past climatic change 24 affected late prehistoric Yup'ik society at the Nunalleq site, and to provide a contextual 25 baseline to modern climate change, local palaeoclimatic data is essential. To this end, this 26 paper presents the results of a high-resolution palaeoclimate reconstruction from a sediment 27 profile closely associated with this late prehistoric Yup'ik archaeological site. By applying the 28 Mutual Climate Range method (Atkinson et al. 1986) this paper aims to: (i) reconstruct past 29 summer and winter temperature variation in the vicinity of Nunalleq; (ii) compare these data 30 to modern long term meteorological observations from Bethel Airport, and; (iii) explore how 31 this record compares to other regional data.

M A N U S C R I P T
A C C E P T E D The study site is located on the coast of the Bering Sea in southwestern Alaska, 6 approximately 5 km south of the village of Quinhagak, in the vast Yukon-Kuskokwim delta 7 (Fig. 1). The relief of this deltaic wetland is relatively flat, and elevations in the vicinity of the 8 study location range from 2-3 m above mean sea level. The region is characterised by 9 numerous small lakes/ponds and meandering rivers and sloughs (Jorgenson, 2000) and is 10 underlain by discontinuous permafrost (Jorgenson et al., 2008). Local geology is 11 uncomplicated and comprises recent alluvial deposits of sands, silts and clays of Quaternary 12 age capped by a thin layer of peat (Wilson et al., 2015). Vegetation is dominated by tundra 13 communities (Babcock and Ely, 1994) characterised by a mixture of graminoid-rich meadows 14 and dwarf shrub tussock tundra of Betula nana, Rubus chamaemorus and Empetrum nigrum. Climatically, the region is located in the sub-arctic and the west coast climate division 19 of Bienek et al. (2012). The nearest long-term observational climatic data comes from Bethel 20 airport, located 120 km northeast of Nunalleq (NOAA, 2017). Continuous observation of 21 summer (July) temperatures at this station date from 1923, while winter (January) 22 temperature observations date from 1929. The early part of this instrumental record, until 23 the 1980s, is characterised by a summer climate approximately 1ᵒC cooler than the modern 24 summer mean of 13.4ᵒC (Table 1). Although the period from the mid-1980s is characterised 25 by sustained summer warming (Fig. 2), it is also notable for a sharp increase in mean summer 26 temperature variability. Mean winter temperatures are more variable than those from the 27 summer and record a marked increase in the winter mean from -14.7ᵒC (1929-1958) to -13.7 28  in the mid-20 th century, which falls again towards the modern day (Table 1).
29 Table 1 here The density of Coleoptera subfossils in any substrate will vary according to the nature 25 of that substrate. Palaeoentomologists typically collect samples varying between 2 and 5 l (or 26 1 and 5 kg) volumes (e.g. Bain, 2001;Kenward, 2009;Panagiotakopulu and Buckland, 2013;27 Vickers et al., 2011) but there has not been an investigation of the influence of sample size on 28 the representativity of beetle subfossil assemblages. Generally, less organic substrates 29 produce fewer beetle fossils than more organic ones (Elias, 2010) and due to the 30 accumulation of organic matter and nutrient enrichment in and close to human settlements, 31 anthropogenic sediments also tend to harbour higher densities of beetle (and other M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT 7 arthropods) than naturally accumulated deposits located further away from human 1 occupation sites. Following preliminary analysis of samples from archaeological floors at 2 Nunalleq (Forbes et al., 2015), our approach has been to privilege increased temporal 3 resolution over higher counts for single samples. Despite the smaller volume of samples 4 analysed in the present study, our MNIs are consistently equal to, or higher than, those 5 obtained from other studies of Coleoptera subfossils extracted from peat bogs (e.g. Buckland 6 et al., 2009;Khorasani et al., 2015;Panagiotakopulu and Buckland, 2013;Vickers et al., 2011). In total, twelve contiguous 2-cm thick samples, from a depth of 39 to 15 cm, and a 11 further sample from 13-11 cm, were processed and analysed for Coleoptera. In the first 12 instance samples were placed into large buckets and soaked in a cold solution of weak (<2%) 13 NaOH for a period of up to one week to promote disaggregation of the sediment matrix.
14 Samples were then processed though paraffin floatation (Coope and Osborne, 1967;15 Kenward et al., 1980) to separate the Coleoptera from the bulk of the plant macrofossil 16 content. The flotants were hand-sorted under a binocular microscope to allow the collection 17 of beetle sclerites. Insect fossils were stored in vials of ethanol prior to identification.

M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT
Based on this classification, three different percentage diagrams were generated to 1 allow an appreciation of changes in local palaeoenviromental conditions through time ( Fig.   2   6). The first (Fig. 6a) was produced using all five ecological groups, the second (Fig. 6b) 3 excluding the 'Eurytopic' group in order to better display patterns obscured by the presence 4 of wide-tolerance taxa, and the third (Fig. 6c) displaying the proportion of taxa associated 5 with microhabitats available in organic matter. Range, or MCR). Only predator and scavenger beetle taxa are utilised as these groups are not 14 bound to particular plant species or communities and respond rapidly to climatic change 15 (Elias, 2010). We employed continent-wide climate envelopes based on North American 16 species records and associated climatic data (Elias, 1996;Elias et al., 1996). In order to 17 incorporate taxa that could only be identified to the level of genus, subgenus or group, we 18 also constructed a series of 'grouped envelopes'. Here, the principle was to account for the 19 maximum climatic range of species occurring within the same taxonomic group, allowing us 20 to include taxon such as for example Pterostichus brevicornis group (sensu Ball 1966),

21
Eucnecosum spp. and Pycnoglypta spp., which are particularly difficult to identify to species. This was achieved by combining all established climate envelopes for species of a given    including Olophrum spp. and Boreaphilus henningianus C.R. Sahlberg, are associated with 5 moss and emergent subaquatic vegetation in such settings (Campbell, 1978(Campbell, , 1983b. Others, 6 such as Acidota crenata (Fabricius) and A. quadrata (Zetterstedt), are also common in organic 7 materials such as leaf litter, carrion and dung (Campbell, 1982). All members of the 'Aquatic' with dytiscids forming about 5% of these assemblages (Fig. 6). From the time of the site's 9 occupation to the end of the sequence, assemblages are dominated by the 'Mesic' group and 10 'Hygro-riparian' groups, which suggest moist to humid conditions, such as occurs in the area 11 at the present-day, which is characterised by wet tundra. The proportion of beetles 12 associated with decomposing organic matter microhabitats is high throughout the profile, 13 but it is remarkably higher (reaching more than 80 % of the assemblage) in the interval 14 between 20 and 32 cm, which correspond to the period of occupation of the site. We 15 interpret this as evidence for nutrient-enrichment of the bog in the vicinity of the occupation 16 site, likely derived from trampling, disturbance and the transport, processing and deposition 17 of organic materials (e.g. hunted animals, sod, grass, dog and human faeces) on the ground.

18
The detailed palaeoecological implications of this dataset will be the object of a forthcoming 19 paper, which will compare faunas from the peat profile with those extracted from occupation 20 floors layers from the sod dwelling's interior at Nunalleq.  For the most part, the reconstructed mean summer temperatures throughout the 1 sequence were marked by wide MCRs. A majority of samples produced T MAX estimates with 2 ranges of between 3.0˚C and 6.5˚C and results that encompassed both the early 20 th century 3 and modern summer means (Fig. 7). Nevertheless, there are three samples from which the 4 Coleoptera assemblages permitted reconstructions of T MAX with MCRs of between 0.4˚C to 5 2.7˚C.

6
The temperature reconstruction can be sub-divided into three main parts: the pre- There is little change for the period immediately following site abandonment. much lower than observational data. This is likely a result of beetles adapted to living in the 4 boreo-arctic zone being able to enter a phase of dormancy and/or seek shelter from the cold 5 during the winter (Elias, 2010;Elias et al., 1999a). In addition, it has been shown that modern 6 beetle faunas from coastal regions in Alaska represent species that are able to withstand 7 extreme cold weather events that occur for a few days or weeks of the winter, rather than 8 species that would normally be associated with the relatively mild winter temperatures 9 associated with oceanic climate (Elias et al., 1999a). The reconstruction of mean winter 10 temperatures must therefore be viewed as a crude estimate of winter temperature variation.

11
Our results suggests that from the mid-15 th to the early 16 th century, prior to the 12 human occupation at Nunalleq, January mean temperatures varied from c. -14.7˚C to -34.2˚C.

13
Data for the latter half of the 16 th century ( Fig. 8; The earliest and longest instrumental temperature series for the Y-K Delta is from   and modern 2 (1983-2016) means (Fig. 9).
3 Figure 9 here, 190mm 4 The clearest trend from the dataset is that the summer climate of the Y-K delta from 5 the mid-15 th to late 19 th century was, in all likelihood, cooler than that of the 20 th century.

6
There are no instances where the minimum values of the reconstructed T MAX exceed either 7 the early 20 th century or modern means. Indeed, in all instances (with exception of the late 8 19 th /early 20 th century), the majority of the range of T MAX is lower than both 20 th century 9 reference periods (Fig. 9a, c). There are two points when mean summer temperatures in the 10 Y-K delta were definitively cooler than in the early 20 th century. The first of these occurs at The nearest available palaeo-temperature reconstruction is from Farewell Lake, 6 approximately 540 km northeast of Nunalleq (Fig. 1). Hu et al. (2001) present evidence for 7 significant temperature change from c. AD 1550 that culminated at AD 1700 with 8 temperatures 1.7˚C lower than present. Whilst the beginning of LIA cold at Farewell Lake is 9 comparable from observations from Nunalleq, the deepest cold of our record, 1.3˚C below 10 the modern mean, is significantly warmer and occurs over a century later at c. AD 1815 11 ( Figure 9, Table 2). This is perhaps unsurprising given the coastal aspect of Nunalleq relative  Table captions  Table captions  Table captions  Table captions 1 2               Figure 1. Figure 1. Figure 1. Sheenjek; [11] Iceberg Lake.        Figure 5. Figure 5. Figure 5. Figure 5. Grouping of identified taxa according to their habitat/ecology. In red font are those taxa that are eurytopic or belong to mesic, hygrophilous and riparian environments, but that are known to be associated with microhabitats available in decaying organic matter (e.g. rotting wood, leaf litter, floor debris, dung, carrion).  The dashed and dotted lines respectively illustrate the early 20 th century (AD 1923(AD -1952 and modern (1983-2016) July means.   Summer  Winter  Summer  Winter (1989 -2016) (ᵒ ᵒ ᵒ