Millennia-Long Co-Existence of Two Major European Whitefish ( Coregonus spp . ) Lineages in Switzerland Inferred from Ancient Mitochondrial DNA

Archaeological fish remains are an important source for reconstructing past aquatic ecosystems and ancient fishing strategies using aDNA techniques. Here, we focus on archaeological samples of European whitefish (Coregonus spp.) from Switzerland covering different time periods. Coregonus bones and scales are commonly found in archaeological assemblages, but these elements lack species specific features and thus inhibit morphological species identification. Even today, fish taxonomy is confusing and numerous species and ecotypes are recognized, and even more probably existed in the past. By targeting short fragments of the mitochondrial d-loop in 48 morphologically identified Coregonus scales and vertebrae from 10 archaeological sites in Switzerland, endogenous d-loop sequences were found in 24 samples from one Neolithic, two Roman, and four Medieval sites. Two major mtDNA clades, C and N, known from contemporary European whitefish populations were detected, suggesting co-occurrence for at least 5000 years. In the future, NGS technologies may be used to explore Coregonus or other fish species and ecotype diversity in the past to elucidate the human impact on lacustrine/limnic environments.

In Swiss sub-Alpine lakes, up to six whitefish species coexist [9,10].They are genetically distinct, vary in body size, growth rate, gill-raker numbers, food preferences, and spawning behaviour, and occupy littoral, pelagic, and benthic habitats [7,[10][11][12][13].In Europe, two diverse mitochondrial (mtDNA) d-loop clades were detected in whitefish (C.lavaretus complex), which coexist across Swiss lakes and are shared among species: one so-called 'northern lineage' (N clade) with a predominant distribution range from north-west Russia to Denmark and one 'central European lineage' (C clade) with a higher frequency from Denmark to the European Alps [12,14].The co-occurrence of both lineages suggests that whitefish populations originating from at least two different refugia colonized the Swiss lakes after the last glacial maximum (LGM), giving rise to new whitefish species [12].
However, human impact, such as eutrophication, overfishing, regulations, barrages, or sewage plants, led to a drastic decrease in diversity and the loss of species, e.g., lacustrine and river-spawning coregonids including those supposedly living exclusively in rivers [10,11,15,16].Today, the European whitefish is one of the major fish groups in Switzerland declared endangered, and thus all species are protected by the 1982 Bern Convention on the Conservation of European Wildlife and Natural Habitats.Biological research into the processes that maintain and generate biodiversity may help to understand and prevent such extinctions [10].Yet, archaeology, archaeozoology, and ancient DNA (aDNA) analyses of whitefish remains offer a direct view on the life history of a species during past cultural periods in relation to the human impact and thus permit a long-term perspective on biological conservation efforts [17].
In Switzerland, the archaeological remains of whitefish are frequently found in different time periods from the Neolithic onwards [18][19][20][21][22][23][24][25].They are well preserved both in waterlogged and dry conditions.In the Neolithic, whitefish, along with other fish remains, document the regular exploitation and on-site consumption of fish depending on seasonal availability.These finds, along with dendrochronological dated lake shore settlements, provide rich and detailed chronological insights into prehistoric fishing techniques, e.g., harpoons for pike fishing or gillnets for cyprinids in shallow water, while the fishing of whitefish in open water required watercraft and more refined catching techniques [18,19].
During the Roman era, whitefish was on the menu in military camps, as well as in wealthy households across Switzerland [20,21].As lakes are the usual habitat of whitefish [22], the finding of their remains at Roman sites close to the Rhine River and its tributaries [21] suggest that migratory populations spawning in the Rhine existed in the past, some of which may have survived until recently [16].
The Middle Ages was a period of increasing fish consumption [23,24], including whitefish.An example for the transportation of whitefish over 100 km from Lake Lucerne to the city of Basel is known from a written source dated to the 12th century AD [25].However, the consumption of fresh fish from sources at long distances may have been limited to wealthy people, e.g., members of the clergy and upper classes and the demands of most people were likely satisfied by local fisheries [23,24].
Archaeological sites for later periods are rare in Switzerland.Therefore, no data are included from periods between the ca.16th century AD and modern times.Today, whitefish are very important economically for professional fishers, who supply local restaurants around the lakes that flourish during times of eutrophication.However, the recent re-oligotrophication of the lakes may reduce whitefish catches, causing economic loss to the fishers [15].
With the routine wet-sieving of archaeological sediments in Switzerland, using small mesh sizes, small fish remains such as cranial bones, vertebrae, and scales, provide evidence for a much more important role of fish as a food source throughout history than previously believed [18,21,23].However, the species identification of Coregonus based on the morphological criteria of skeletal elements is not possible as it lacks diagnostic features, and in the case of cranial bones, scales, or vertebrae most commonly preserved in the archaeological record, identification is only possible on the genus level, preventing the assessment of species diversity in the past.Therefore, the application of DNA-based methods to distinguish modern Coregonus species may help to overcome this limitation provided that DNA has survived in the archaeological remains.
Ancient DNA studies have been successfully applied to a variety of fish from marine and freshwater environments of cold and temperate regions, but rarely from tropical regions [26][27][28][29][30][31], using maternally inherited mitochondrial DNA markers.These papers support the potential of aDNA to address past diversity, historic trade routes, and economies, to reconstruct expansions and colonization routes of past wild fish populations, and to detect climatic effects on fish species.Nuclear markers have been targeted in a few publications to study historic genetic fish diversity [29,30,32].The application of SNP typing, NGS, and DNA capture-enrichment methods to ancient fish, however, awaits the development of genomic tools for population studies that are, unfortunately, still not readily available for most modern fish groups [33].Nuclear markers (SNPs) and NGS are currently being developed and adopted for extant Coregonus species [1,34], and they will be available for species determinations in ancient Coregonus remains in the near future.
Here, we show that small Coregonus remains are a source of aDNA by the PCR amplification of mtDNA d-loop fragments in 48 individual elements from two waterlogged and eight dryland archaeological sites in Switzerland.The presence of two main maternal lineages, C and N, was found from the Neolithic onwards.This study provides a basis for further genetic research using archaeological Coregonus and other fish remains to reconstruct fishing practices, subsistence, trade, economy, or fish speciation and diversity in the context of the past and present anthropogenic impact.

Sampling
Fish remains are recovered from 4, 1, and 0.35 mm mesh size sieves after the wet-sieving of sediment samples together with archaeobotanical and other small faunal remains.An archaoezoological analysis of Coregonus samples was performed at Integrative Prehistory and Archaeological Science (IPAS), University of Basel, Switzerland.Coregonus vertebrae and scales were obtained and morphologically determined following standard procedures at the IPAS [23,24,35].The specimens were stored at IPAS for up to 20 years in a dark, dry place at room temperature.From this collection, 40 morphologically well preserved Coregonus remains (34 vertebrae, six scales) were selected for aDNA analysis.Additionally, eight Coregonus elements (four vertebrae, four scales) from the Neolithic lake-shore settlement Arbon Bleiche 3 were "freshly" re-sampled from wet sediments that had been cold-stored for about 20 years.These waterlogged preserved remains were dried at room temperature for one week, after which DNA was extracted (Tables 1 and S1).

Archaeological Sites
Fish remains were recovered from two Neolithic waterlogged sites, and three Roman, one Early Medieval, and four Medieval dryland sites in Switzerland.Neolithic sites were dated by dendrochronology, and all others were dated by a typo-chronological analysis of the artefacts (coins and pottery) in the same layers (Figure 1, Tables 1 and S1).

Stansstad-Kehrsiten
The lakeshore settlement Stansstad-Kehrsiten is located at Lake Lucerne, in the Canton of Nidwalden and dendro-dates point to occupation between ca.3500-3400 BC.The settlement was established on a lake beach and is a key site of the Neolithic transitional time period between the Cortaillod and Pfyn cultures [19].Underwater excavations took place between 2003 and 2011.The animal remains at the site are 75% fish, 20% mammals, and 5% amphibians after wet-sieving.The final assessment of the site is ongoing.

Arbon-Bleiche 3
Arbon Bleiche 3 is a single-phase settlement of the Horgen culture on Lake Constance, in the Canton of Thurgau and has been dated dendrochronologically to between 3384 and 3370 BC [36].Excavations started in 1993, and the final evaluation of the site took place from 1997 to 2002.After wet-sieving, a high proportion of fish remains were recovered: 79% fish, 20% amphibians, and <1% birds and mammals, which underlines the importance of fish as a protein source in these Neolithic cultures.
Both sites belong to the UNESCO World heritage site "Prehistoric Pile dwelling around the Alps" (http://sites.palafittes.org/home).Thanks to well-preserved timber, layers of these sites have been dated dendrochronologically, and uncarbonised organic material is mostly well preserved under waterlogged conditions, providing an excellent opportunity for archaeobiological studies of early agricultural societies [48][49][50].

Roman Era
The sites Römerblick and Breite at Windisch, in the Canton of Aargau, belong to a large Roman legionary camp excavated in several, still ongoing, campaigns lasting more than 100 years.Both are dated typologically to the 1st century AD, and samples are preserved in dry conditions.Excellently preserved samples were obtained from the floor of a high-ranking centurion's kitchen

Arbon-Bleiche 3
Arbon Bleiche 3 is a single-phase settlement of the Horgen culture on Lake Constance, in the Canton of Thurgau and has been dated dendrochronologically to between 3384 and 3370 BC [36].Excavations started in 1993, and the final evaluation of the site took place from 1997 to 2002.After wet-sieving, a high proportion of fish remains were recovered: 79% fish, 20% amphibians, and <1% birds and mammals, which underlines the importance of fish as a protein source in these Neolithic cultures.
Both sites belong to the UNESCO World heritage site "Prehistoric Pile dwelling around the Alps" (http://sites.palafittes.org/home).Thanks to well-preserved timber, layers of these sites have been dated dendrochronologically, and uncarbonised organic material is mostly well preserved under waterlogged conditions, providing an excellent opportunity for archaeobiological studies of early agricultural societies [48][49][50].

Roman Era
The sites Römerblick and Breite at Windisch, in the Canton of Aargau, belong to a large Roman legionary camp excavated in several, still ongoing, campaigns lasting more than 100 years.Both are dated typologically to the 1st century AD, and samples are preserved in dry conditions.Excellently preserved samples were obtained from the floor of a high-ranking centurion's kitchen (Windisch Römerblick) and from three pits (Windisch Breite), and the latter were filled with 45% fish from all faunal remains [21,37].
The site Neftenbach, in the Canton of Zürich, is a Roman villa rustica dating to the 3th/4th century AD [38,39].Samples are from a cesspit of construction 25.

Medieval Period
Samples are from a floor insulation of the annex constructions G and F of an early Christian church (7th century AD; Sogn Murezi) at Tomils, in the Canton of Grisons [40][41][42].The constructions were physically connected with each other and with the church.The samples had been moved from their primary and unknown archaeological context to their final destination within the annex constructions.Faunal samples were retrieved during a student field course of IPAS in 2005.Excavations were carried out between 1994 and 2011.
Zürich Fraumünsterstrasse, in the Canton of Zürich (1010-1160 AD) was a living quarter close to the lake of Zürich and the abbey of Fraumünster.The samples were taken from the first occupation layers with an unclear context and from the filling of a fire pit [43].
The fish remains from Basel Bäumleingasse 14, in the Canton of Basel-Stadt (13th century AD) [44] were from the upper layer of a cesspit outside the Medieval building.
Further samples were taken from cesspit 2 from the excavation at Basel Museum der Kulturen, Im Schürhof, in the Canton of Basel-Stadt (15th/16th century AD) [45,47].An assessment of the site is ongoing.
Fish remains were retrieved from waste ditches (German Ehgraben) built between houses for the disposal of faeces and other waste at Weesen Rosengärten, in the Canton of St. Gallen (14th century AD) [45,46].A final assessment of the site is ongoing.

Methods
All DNA extractions and pre-PCR steps were performed in dedicated aDNA facilities and strictly followed the accepted standards as established at IPAS and in aDNA research (see e.g., [51,52]).

DNA Extraction
Single Coregonus vertebrae and scales (samples Cor1 to Cor48) were ground in 360 µL buffer ATL (provided from DNeasy ® Blood & Tissue Kit, Qiagen, Basel, Switzerland) using a mortar and pestle, and for every eight samples, two blank extracts were included in the preparation.All extraction steps followed the "User-Developed Protocol: Purification of total DNA from compact animal bone" for ≤100 mg bone using the DNeasy ® Blood & Tissue Kit.Thereafter, extracts were further purified with Ultra-0.5 mL 30 kDa centrifugal filters (Amicon/Millipore, Zug, Switzerland) with buffer AE (provided from DNeasy ® Blood & Tissue Kit) to a final volume of 100-150 µL per sample.Samples were stored at −20 • C.

Primer Design, PCR, and Sequencing
European whitefish populations including those from Switzerland belong to two divergent mitochondrial d-loop lineages: the Northern and the Central European clade [12,14].Compared to the reference sequence NC002646 of the Central (C) European lineage, the Northern (N) European lineage typically shows one nucleotide insertion, 15,847_15,848insT or 15,846_15,847insT (SNP1), and three substitutions, 15,887C>T (SNP2), 16,498A>G (SNP3), and 16,726G>A (SNP4).The whole region spans about 992 bp and as aDNA is highly fragmented, five primer pairs were designed to target SNP sites by generating short non-contiguous d-loop fragments and allowing for high specificity to Coregonus (Table 2).The fragment lengths generated were less than 100 bp for most primer pairs, except for the combination CORb1F and COR1R, which produced a larger fragment of 135 bp.However, this primer pair was less successful and was therefore not further used in this study.PCR amplifications were performed in a Mastercycler pro S (Eppendorf, Allschwil, Switzerland) in 25 µL volumes with 5-8 µL DNA sample, 2 µM of each primer, and 400 µM dNTP Mix (Promega, Dübendorf, Switzerland), plus 1.5 U AmpliTaq Gold, 1×GeneAmp PCR Gold buffer (150 mM Tris-HCl, 500 mM KCL, pH 8.0) and 2 mM MgCl 2 (all from Applied Biosystems, Hombrechtikon, Switzerland).PCR started with a 12-min initial denaturing step, followed by 70 cycles of denaturing at 94 • C for 1 min, annealing at 54-55 • C for 1 min, extension at 72 • C for 1 min, and a final extension at 72 • C for 5 min.Amplifications were performed in sets of six to eight samples per primer pair including at least one blank extract and one non-template control.None of the controls produced a product after PCR amplification.
PCR products were run on 3% agarose gel and bands of an expected size were cut and purified using a MinElute Gel Extraction Kit (Qiagen, Basel, Switzerland).Products were directly sequenced in both directions by Microsynth (Balgach, Switzerland) using Sanger technology with modified PCR primers (the same as used in amplification reactions) to which a non-specific 40-bp nucleotide tail (5 -AACTGACTAAACTAGGTGCCACGTCGTGAAAGTCTGACAA-3 ) had been added to the 5 -end [53].Three independent repeat amplifications from the same extracts were performed per sample and targeted to ensure the reproducibility of the sequencing and genotyping results.All ancient Coregonus sequences are available at GenBank accession numbers MF441228-MF441251.

Results and Discussion
PCR reactions were performed on 48 archaeological Coregonus remains using primers that target four small non-contiguous segments of Coregonus d-loop DNA.Amplifications were successful and sequences were reliably retrieved for all d-loop segments in 50% of elements (five scales, 19 vertebrae) from seven out of 10 sites, covering the whole time range tested, except for the Early Medieval period (Figure 1, Table S1).This success rate is in the order of magnitude observed for PCR-based approaches when targeting the mtDNA of archaeological fish [32,55] or other animal remains from central European climate zones [56].Even from unusually small quantities of starting material (50% of specimen weighing <10 mg and the smallest weighing <2 mg), vertebrae and scales provided a source of aDNA (Table S1).Sequences generated from all segments were repeatedly identical within one individual and SNPs were consistently typed.All fragments were identical to published Coregonus d-loop sequences.Therefore, a morphological and d-loop-based determination of Coregonus spp.gave congruent results.
For 19 vertebrae and five scales from three sites, amplifications completely failed for all segments/primer combinations.This applied in the first instance to the Neolithic lake-dwelling settlements of Stansstad-Kehrsiten and Arbon Bleiche 3 when stored samples were processed (Table S1).This agrees with earlier observations that PCR-based aDNA amplification from animal remains from waterlogged Neolithic contexts in Switzerland failed [56], with few exceptions [52].The performance of advanced DNA techniques (NGS) when using the same type of material is as yet unknown.Interestingly, when we used specimens "freshly" re-sampled from cold stored sediments from Arbon Bleiche 3, d-loop sequences were obtained for seven out of eight samples tested (Table S1).We observed similar behaviour at other Neolithic lake shore settlements (e.g., Zürich Opéra, data not shown).Therefore, material for DNA analyses from waterlogged Neolithic specimens in the Alpine foreland is preferably taken directly from the excavations or whole sediment blocks (not individuals) should be cold stored [57,58].Quite unexpectedly, all six samples of the Early Medieval dryland site at Tomils (Figure 1, Table S1) also failed to yield any amplifiable aDNA, suggesting the existence of taphonomic particularities associated with this site (e.g., specimens were not found in situ and were used to construct a floor insulation structure).
Sequences of ancient and modern concatenated d-loop segments from the whitefish radiation across Europe, including Switzerland, were compared to each other: Median-joining network analysis (Figure 2) showed that all ancient sequences fit within the haplotype diversity of the two major maternal lineages known from modern European whitefish (C.lavaretus complex), i.e., the C and N lineage [12].Two ancient haplotypes (20 samples and one sample, respectively) affiliated to the C lineage and one haplotype (three samples) clustered with the N lineage (Figure 2).This is consistent with contemporary lineage distribution in the sub-Alpine region and, more specifically, within Switzerland (228 samples C lineage, 42 samples N lineage, see Table S2).In the future, SNP genotyping and NGS may be used to discern single Coregonus individuals, as well as species, properly and allow for more accurate quantitative estimates of diversity and population sizes, for e.g., from sediments [59].
For 19 vertebrae and five scales from three sites, amplifications completely failed for all segments/primer combinations.This applied in the first instance to the Neolithic lake-dwelling settlements of Stansstad-Kehrsiten and Arbon Bleiche 3 when stored samples were processed (Table S1).This agrees with earlier observations that PCR-based aDNA amplification from animal remains from waterlogged Neolithic contexts in Switzerland failed [56], with few exceptions [52].The performance of advanced DNA techniques (NGS) when using the same type of material is as yet unknown.Interestingly, when we used specimens "freshly" re-sampled from cold stored sediments from Arbon Bleiche 3, d-loop sequences were obtained for seven out of eight samples tested (Table S1).We observed similar behaviour at other Neolithic lake shore settlements (e.g., Zürich Opéra, data not shown).Therefore, material for DNA analyses from waterlogged Neolithic specimens in the Alpine foreland is preferably taken directly from the excavations or whole sediment blocks (not individuals) should be cold stored [57,58].Quite unexpectedly, all six samples of the Early Medieval dryland site at Tomils (Figure 1, Table S1) also failed to yield any amplifiable aDNA, suggesting the existence of taphonomic particularities associated with this site (e.g., specimens were not found in situ and were used to construct a floor insulation structure).
Sequences of ancient and modern concatenated d-loop segments from the whitefish radiation across Europe, including Switzerland, were compared to each other: Median-joining network analysis (Figure 2) showed that all ancient sequences fit within the haplotype diversity of the two major maternal lineages known from modern European whitefish (C.lavaretus complex), i.e., the C and N lineage [12].Two ancient haplotypes (20 samples and one sample, respectively) affiliated to the C lineage and one haplotype (three samples) clustered with the N lineage (Figure 2).This is consistent with contemporary lineage distribution in the sub-Alpine region and, more specifically, within Switzerland (228 samples C lineage, 42 samples N lineage, see Table S2).In the future, SNP genotyping and NGS may be used to discern single Coregonus individuals, as well as species, properly and allow for more accurate quantitative estimates of diversity and population sizes, for e.g., from sediments [59].The presence of both divergent mtDNA lineages C and N at the Neolithic site of Arbon Bleiche 3 (Lake Constance) and at Fraumünsterstrasse (Lake Zürich) from the 11th century AD (Figure 1) suggests co-existence for more than 5000 years in the sub-Alpine region.The Neolithic Coregonus from Lake Constance are probably among the ancestors of the recent adaptive radiation and both

Figure 2 .
Figure 2. Median-joining network of concatenated Coregonus d-loop sequences (140-141 bp) displaying lineage distribution of archaeological samples from Switzerland (n = 24) (solid black) compared to published modern sequences from the European pre-Alpine/Alpine region including France (n =12), Switzerland (n = 270) and South-Germany (n = 57) (backward diagonal lines), and Northern Europe (n = 34) (cross lines).Size of nodes are proportional to haplotype frequencies.Numbers denote number of mutations between nodes.

Figure 2 .
Figure 2. Median-joining network of concatenated Coregonus d-loop sequences (140-141 bp) displaying lineage distribution of archaeological samples from Switzerland (n = 24) (solid black) compared to published modern sequences from the European pre-Alpine/Alpine region including France (n =12), Switzerland (n = 270) and South-Germany (n = 57) (backward diagonal lines), and Northern Europe (n = 34) (cross lines).Size of nodes are proportional to haplotype frequencies.Numbers denote number of mutations between nodes.

Table 1 .
Specifications of archaeological sites from which whitefish remains were taken for this study.

Table 2 .
Primer pairs used in this study.