Molecular archaeoparasitology identifies cultural changes in the Medieval Hanseatic trading centre of Lübeck

Throughout history, humans have been afflicted by parasitic worms, and eggs are readily detected in archaeological deposits. This study integrated parasitological and ancient DNA methods with a large sample set dating between Neolithic and Early Modern periods to explore the utility of molecular archaeoparasitology as a new approach to study the past. Molecular analyses provided unequivocal species-level parasite identification and revealed location-specific epidemiological signatures. Faecal–oral transmitted nematodes (Ascaris lumbricoides and Trichuris trichiura) were ubiquitous across time and space. By contrast, high numbers of food-associated cestodes (Diphyllobothrium latum and Taenia saginata) were restricted to medieval Lübeck. The presence of these cestodes and changes in their prevalence at approximately 1300 CE indicate substantial alterations in diet or parasite availability. Trichuris trichiura ITS-1 sequences grouped into two clades; one ubiquitous and one restricted to medieval Lübeck and Bristol. The high sequence diversity of T.t.ITS-1 detected in Lübeck is consistent with its importance as a Hanseatic trading centre. Collectively, these results introduce molecular archaeoparasitology as an artefact-independent source of historical evidence.

PGF, 0000-0002-3153-3411; OGP, 0000-0002-8797-2667; ALS, 0000-0002-7657-6191 Throughout history, humans have been afflicted by parasitic worms, and eggs are readily detected in archaeological deposits. This study integrated parasitological and ancient DNA methods with a large sample set dating between Neolithic and Early Modern periods to explore the utility of molecular archaeoparasitology as a new approach to study the past. Molecular analyses provided unequivocal species-level parasite identification and revealed location-specific epidemiological signatures. Faecal -oral transmitted nematodes (Ascaris lumbricoides and Trichuris trichiura) were ubiquitous across time and space. By contrast, high numbers of foodassociated cestodes (Diphyllobothrium latum and Taenia saginata) were restricted to medieval Lü beck. The presence of these cestodes and changes in their prevalence at approximately 1300 CE indicate substantial alterations in diet or parasite availability. Trichuris trichiura ITS-1 sequences grouped into two clades; one ubiquitous and one restricted to medieval Lü beck and Bristol. The high sequence diversity of T.t.ITS-1 detected in Lü beck is consistent with its importance as a Hanseatic trading centre. Collectively, these results introduce molecular archaeoparasitology as an artefact-independent source of historical evidence.

Introduction
Enteric parasites have afflicted humans throughout history and remain common in parts of the developing world. Indeed, prior to improvements in hygiene and medicine in the eighteenth century, enteric parasites were prevalent throughout the world, which makes them an attractive target for investigation of historical events. Previous molecular studies of historical infections have largely focused on high-impact diseases such as plague, leprosy, smallpox, malaria, and tuberculosis [1][2][3][4][5]. These studies typically focus on one, or very few, samples from a site and are targeted largely towards the identification of the pathogen under investigation [3,6 -12]. Highly pathogenic infectious diseases often cause acute disease, interfere with the daily lives of infected individuals (in extreme cases killing them), and can be difficult to detect in archaeological contexts if identification relies on analysis of mummified or skeletal remains.
By contrast, enteric worms (helminths) do not have devastating clinical effects and the eggs are readily detectable in a wide variety of archaeological contexts associated with human faecal material [13 -16]. Helminth is a collective term encompassing nematodes (roundworms), trematodes (flatworms), and cestodes (tapeworms), and there are many examples from each group that infect a range of hosts, including humans. Helminths exhibit wide variation in their life cycles: for example, Trichuris trichiura and Ascaris spp. (both nematodes) are transmitted via the faecal -oral route whereas the cestodes (tapeworms) Taenia spp. and Diphyllobothrium latum enter humans by consumption of undercooked red meat or freshwater fish, respectively. Hence, the type and number of eggs found in a deposit can be used to interpret levels of hygiene and/or dietary habits. In our experience, greater than 95% of pre-eighteenth century latrine and other communal deposit samples contain helminth eggs. The high prevalence of enteric helminths, their low pathogenicity, and the robustness of eggs encapsulating contained ancient DNA (aDNA), are features that underpin their broad applicability to historical and archaeological questions. A small number of studies of parasites in archaeological contexts have reported the use of aDNA as a diagnostic tool. These include the detection of protists in preserved tissue samples such as Trypanosoma cruzi from Peru, Northern Chile and Brazil or Plasmodium falciparum in Egyptian mummies and a fifth century CE Roman infant [17 -20]. In terms of intestinal helminth infections, PCR detection and sequence data have been reported for Trichuris trichiura, Ascaris spp., Taenia spp., Diphyllobothrium latum, Fasciola hepatica, Clonorchis sinensis, Schistosoma spp., and Enterobius vermicularis [15,16,[21][22][23][24]. However, in all of these cases, analyses were focused on diagnosis and often involved only a single archaeological site. Here, we employ molecular methodologies for species-level diagnosis and identification of epidemiological patterns by integrating data from multiple sites.
We analysed the occurrence of helminth parasite eggs within 152 samples that dated between Neolithic (ca 3600 BCE) and the Post Medieval (seventeenth century) which provided insight into historical living conditions and cultural practices. Parasite aDNA was amplified and sequenced to provide species-level diagnosis and, with Trichuris trichiura (the most frequently detected parasite), epidemiological patterns including genetic diversity provided a unique, artefact-independent source of historical information. The novel approaches developed in this work demonstrate the wide applicability and potential for molecular archaeoparasitology to broadly impact the study of historical events ranging from health to socio-economic and dietary practices.

(a) Site and sample description
The current study examined 152 samples from six sites in the UK, Germany, the Czech Republic, and Switzerland dating between Neolithic and Post Medieval periods (electronic supplementary material, figure S1) which included a large structured sample set from Medieval Lü beck. This city was one of the most influential trading cities in Medieval Europe and the founding centre of the Hanseatic League. The samples from Lü beck were obtained from the recent UNESCO excavation in the Grü ndungsviertel (founding quarter) between Fischstrasse and Alfstrasse, consisting of 31 stratigraphed latrine samples from eight houses dating between twelfth and seventeenth century CE. Further samples obtained for this study include a set from a medieval communal waste ditch in Bristol (UK, 26 samples), a Neolithic stilt settlement in Zurich (CH, 15 samples [25]), a Viking-age latrine in York (UK, one sample), and samples from the intestinal area of single burials in Břeclav-Pohansko (CZ, 61 early medieval samples) and Ellwangen-Jagst (DE, 28 medieval to early modern samples). The excavating archaeologists for all sites reported good to excellent morphological preservation of organic material in the archaeological deposits.

(b) Microscopic detection of helminth parasites
Helminth eggs are robust and display morphological characters that allow genus-level identification of parasites in archaeological deposits. The eggs of parasitic nematodes, Ascaris spp. and Trichuris spp., were detected in samples from all sites, and 94.5% of latrines or communal deposits (73 samples) were positive for eggs of at least one nematode. Cestode eggs were detected in samples from Neolithic Zurich, Viking York (Taenia spp.), and Medieval Lü beck (Taenia spp. and Diphyllobothrium latum). Representative micrographs of helminth eggs from Lü beck are depicted in  figure 1f, respectively) and in 25 of 26 samples from the medieval port area of Bristol (Finzels Reach; ranges of 78 -8 559/g and 76 -1 162/g for Trichuris and Ascaris, respectively). The samples from Lü beck also contained substantial numbers of eggs from two cestode parasites, 14 of 31 positive for D. latum (range 49 -1 414/g, figure 1g) and with 19 of 31 samples positive for Taenia spp. (range 133-8 310/g, figure 1h). Although Taenia spp. eggs were detected in samples from Neolithic Zurich and Viking York, these were too rare to allow a reliable estimation of density.

(c) Molecular identification of parasites
To confirm/identify microscopic diagnosis of parasite species, aDNA was extracted and a PCR-amplified fragment was sequenced for targets within Trichuris (ITS-1 and b-tubulin, figure 2a for ITS-1), Ascaris (CytB and COX1, figure 2b for COX1), Taenia (CytB, figure 2c), and Diphyllobothrium (COX1, figure 2d). Putative parasite sequences were identified using a pipeline involving BLAST (against the NCBI GenBank database) with their identity confirmed by constructing maximum-likelihood phylogenies (figure 2). The Transition/Transversion (Ts : Tv) ratios for aDNA targets were greater than those from modern sequences (electronic supplementary material, tables S1 and S2). However, with Trichuris, although the TsA:TsB ratio for Lü beck and York was higher than that for modern samples it was lower for rspb.royalsocietypublishing.org Proc. R. Soc. B 285: 20180991 other sites indicating non-aDNA damage-associated variation. For Trichuris, the sequence data obtained from aDNA grouped with the human parasite T. trichiura and were distinct from other Trichuris spp., including those typically found in pigs, dogs, and rodents (T. suis, T. vulpis, and T. muris, respectively; representative data depicted in figure 2a). Sequence analysis of PCR products targeting Taenia (CytB, figure 2c) and Diphyllobothrium (COX1, figure 2d) from Lü beck samples identified these as T. saginata and D. latum, respectively.

(d) Temporal distribution of helminth eggs
As the two nematode parasites Trichuris and Ascaris were commonly found, we evaluated the correlation between the numbers of these two parasites in Lü beck and Bristol using separate Generalized Linear Mixed Models (GLMMs). A strong positive correlation between the numbers of the two parasites were found in Lü beck (figure 1j, estimate + s.e.: 0.31 + 0.07, x 2 1 ¼ 16:40, p , 0.001) and Bristol (electronic supplementary material, figure S2, Lü beck was the only location where cestode eggs were detected consistently throughout the sampling period. Despite both cestode species being present at high numbers in Lü beck their occurrence was largely non-overlapping, with only five samples containing eggs from both parasites. The prevalence of Diphyllobothrium and Taenia eggs showed a clear time dependency (figure 1g,h; GLMM with Poisson error distribution model, Diphyllobothrium estimate + s.e.: 23.15 + 1.26, x 2 1 ¼ 9:24, p ¼ 0.0023 and Taenia estimate + s.e.: 3.34 + 1.17, x 2 1 ¼ 10:12, p ¼ 0.0015) with D. latum more prevalent in earlier samples and Taenia in later samples. To assess whether these shifts are best explained by a gradual change over time or by a sudden shift in parasite prevalence, we used a model using the date as a predictor of parasite status. Bootstrapping strongly supports a structural break in the data (greater than 99% of bootstrap replicates for D. latum, and 96% for Taenia). For D. latum, the break most likely occurs between 1300 and 1325 CE (92% of replicates, with other possible breakpoints each suggested by less than 1.5% of replicates). For Taenia, the breakpoint was slightly less abrupt, with a most likely timing of between 1300 and 1325 CE (60% of replicates) or between 1350 and 1400 CE (35% of replicates). A binomial test gives a p-value of less than 0.001 for the 1300/1325 CE breakpoint for D. latum, and p-values of less than 0.001 for either of the possible breakpoints for Taenia sp. These changes in parasite occurrence and abundance across the ca 1 300 CE boundary were evident only for the cestode parasites. We analysed variation in parasite occurrence for Lü beck using GLMMs with binomial error distribution. For each parasite, we entered separate GLMMs using various factors to assess the correlations. All models were checked for over-or under-dispersion. In Poissondistributed data, we added an observation-level random factor in the model whenever over-dispersion was detected [26,27]. No under-or over-dispersion was detected for binomially distributed data. Because Bristol had only four Ascaris

(e) The source of cestode infections in Lübeck
The sample set from medieval Lü beck provided a unique opportunity to investigate aspects of diet, as the prevalence of D. latum and Taenia spp. indicates the widespread consumption of contaminated raw or undercooked freshwater fish (D. latum) and red meat ( pork for T. solium or beef for T. saginata). The changing prevalence of these two parasites may have resulted from changes in consumption of contaminated freshwater fish or red meat. A wide range of freshwater fish species could be the source of D. latum, and the change in prevalence at around 1300 CE might be due to changes in the consumption of particular freshwater fish species. Previous studies have successfully amplified genetic material of plants and animals from archaeological sediments and used these to identify resident species [28,29]. We employed PCR amplification of a short fragment of the eukaryotic mitochondrial 16S gene and sequencing (MiSeq) to identify putative food species [30]. Sequences matched a range of targets, including domesticated and wild animals as well as  humans. The most common sequences were human, followed by a range of potential food species including cattle (Bos taurus), pigs (Sus scrofa), sheep and goat (Caprinae), fowl (Galloanserans) as well as various freshwater and marine fish species. Although extreme care must be exercised when interpreting the vertebrate aDNA food signature, the patterns obtained in samples from Lü beck were distinct to those detected in any other site, including Bristol (figure 3a).
As an additional control, we also included DNA extracted from six sub-surface (approx. 20 cm deep) soil samples from locations within Oxford which did not yield any food or human-related sequences. Although the aDNA food signature can only be used as a semi-quantitative measure, the composition and estimated proportions of food-related sequences identified within this study of Lü beck were comparable with osteological analyses from previous excavations [31][32][33][34][35]. Cod (Gadus morhua) and herring (Clupea haerengis) were the most prevalent fish sequences detected in Lü beck, but were not detected in other sites. This was not unexpected because cod and herring were major bulk trading goods of the Hanseatic League (air-dried or salted). The Lü beck samples also contained aDNA that identified a wide range of freshwater fish as potential sources of D. latum infection, including cyprinids, coregonids, perch, pike, and eel. The most straightforward explanation would be a relative decline in freshwater fish consumption. However, neither the occurrence nor prevalence of all freshwater fish or any individual fish species changed dramatically across the ca 1 300 CE boundary after which D. latum largely disappeared from the Lü beck samples. Furthermore, there was no correlation between D. latum prevalence and the proportion or type of freshwater fish aDNA (figure 3b; electronic supplementary material, table S3). In Lü beck, the most common freshwater fish family by sequence frequency were the Cyprinidae that could be subdivided into two subfamilies (Cyprininae and Leuciscinae). When considering the number of samples where each freshwater fish was detected, the most frequently detected were Coregonus spp. (f ) Genetic analysis of Trichuris trichiura aDNA ITS-1 sequences have previously been used for species identification of Trichuris [36]. As T. trichiura was present in a wide range of sites, we used ITS-1 (744 unique aDNA sequences, identical replicates were clustered) and b-tubulin (34 unique aDNA sequences) sequence alignments to analyse and compare T. trichiura populations in the different sampling sites. First, median-joining haplotype networks were inferred for the ITS-1 and b-tubulin (figure 4a) alignments, using NETWORK 4.6.6 (fluxus-engineering.com, [37]) with 1 ¼ 0. The haplotype distributions for T. trichiura ITS-1 (TtITS-1) comprised two large groups of sequences, one that contained linked sequences from all sites (comprising of 46 547 aDNA sequences; this group would include all modern reference sequences, but these were omitted from the analysis) and a second smaller group (comprising of 14 563 aDNA sequences) that was dominated by haplotypes present in the Lü beck site, which also contained a small number of haplotypes from Medieval Bristol (figure 4a). The Lü beck-dominated group 2 can be segregated from group 1 by virtue of six conserved nucleotide polymorphisms ( figure 4b). Although care should be exercised when identifying a novel haplotype based upon aDNA data, the high frequency and nature of the haplotype sequence (six conserved polymorphisms with only one C-T transition and no G-A transitions) indicate that it is highly unlikely that this haplotype group was derived from aDNA-associated damage. Despite the much smaller number of sequences available for T. trichiura b-tubulin, these also display two groups with one dominated by sequences only found in Lü beck (figure 4a). Hence, Lü beck appears to be associated with a different pattern of genetic variability than the five other locations and a subset of the sequences from Lü beck forms a distinct cluster within the haplotype network. For TtITS-1, sub-networks were constructed to assess spatial or temporal patterns. Only location-associated patterns were observed and no patterns were evident when sub-networks were generated for shorter time periods confirming that the two groups were present in Lü beck throughout the sampling period (electronic supplementary material, figure S3a,b). Some samples were dominated (greater than 99%) with either group 1 or 2 TtITS-1 sequences while others included mixed populations. Intriguingly, while samples with relatively few eggs (less than 500) contained mixed populations (5/6 samples) those that contained larger numbers of eggs were dominated (greater than 99%) by a single group (5/9 with group 2 and 4/9 with group 1 sequences). Lü beck contained the greatest nucleotide diversity [38] of TtITS-1 sequences; almost two times higher than in Bristol, Zurich, and Břeclav-Pohansko, and around four times higher than in York and Ellwangen. Furthermore, Lü beck was the only location with a relative nucleotide diversity [39] greater than one and this was at least three times higher than detected for the other locations (electronic supplementary material, table S4). Regarding genetic differentiation among locations, estimates of pairwise F ST and associated significance tests revealed that Lü beck as the only location which is significantly differentiated from the other locations (electronic supplementary material, table S4). Only Lü beck and Bristol had a significant spatial structure, as measured by the Association Index statistic ( p-value , 0.001). The maximum monophyletic clade size for Lü beck (41.590) was also much higher than that measured for Bristol (5.446, electronic supplementary material, table S4). It is important to note that these diversity measures are both corrected for unequal sample size, which makes possible their comparison among populations/sampling locations for which we have different sample sizes.

Discussion
The high occurrence of infection, low pathogenicity, and robust nature of the eggs from intestinal helminths support rspb.royalsocietypublishing.org Proc. R. Soc. B 285: 20180991 their use in archaeological contexts, including aDNA studies. Here, we integrated parasitological and aDNA methodologies on parasite material derived from a variety of sites ranging from ca 5600 to 350 years old. All sites contained eggs from two nematode parasites Trichuris spp. and Ascaris spp., and the samples from Lü beck contained high numbers of eggs from two cestodes D. latum and Taenia spp. Sequences of amplified aDNA confirmed parasite identity and clearly established that they were of human origin. Parasite aDNA sequence was obtained from 75.9% of samples where eggs were identified microscopically, independent of soil condition (e.g. age or context), which indicates the broad applicability of the approach to a wide range of sample types. Parasite eggs are environmentally robust [40] and because intact eggs are likely to protect aDNA, it is not surprising that it is possible to amplify parasite aDNA by PCR from most samples.
The Medieval Hanseatic city of Lü beck had a unique parasitological character in terms of the numbers and temporal distribution of cestodes and the population genetics of the nematode T. trichiura. First, considering the cestodes, there was a clear temporal pattern for both D. latum and Taenia spp. eggs, with the former being present in most pre-1300 CE samples and in only two of the post-1300 CE samples. By contrast, Taenia spp. eggs increased in occurrence and prevalence in later samples. The overall numbers of D. latum and Taenia spp. eggs in Lü beck were much larger than reported in any archaeological study (e.g. [41]) which indicates that the sampled Medieval population of Lü beck engaged in food preparation practices that supported the acquisition of such infections (i.e. consumption of uncooked or undercooked fish and red meat).
The source of human Taenia infection was most likely from undercooked beef because the sequences that were obtained matched the beef tapeworm T. saginata rather than the pork tapeworm T. solium. The source of D. latum infection in the people of Medieval Lü beck will have been freshwater fish. Analyses of eukaryotic mitochondrial 16S sequences revealed a range of fish species that would be suitable second intermediate hosts for D. latum, either by consumption of infected first intermediate host (copepod) or other fish that had previously acquired the infection. The importance of local freshwater fisheries to Lü beck is supported by imperial charters dating from 1188 and 1226 awarding the city formal control of fishing rights over the local rivers and lakes [42,43].
It is important to consider the factors that may have led to the dramatic changes in the occurrence and prevalence of both cestode parasites. For D. latum there was a clear boundary ca 1300 CE, with most samples prior to this date containing D. latum eggs. The change in occurrence may have resulted from a change in dietary preferences, food preparation practices, or availability of contaminated freshwater fish. Some explanations are confounded by the detection of D. latum in two samples dated c1500 CE, although only one of these contained substantial numbers of parasites. The simplest explanation would be that less freshwater fish was consumed after 1300 CE. However, there was no detectable change in total freshwater fish or any single species of fish (by morphology or aDNA signature with the resolution provided here) that would explain the dramatic reduction of D. latum. With the expansion of the city around 1300 CE, the river west of Lü beck, the Wakenitz, became increasingly polluted by waste from the increased production of meat and leather [44] which could have affected the availability/attractiveness of fish from the Wakenitz, or numbers of copepods. The supply chain of freshwater fish could also have been affected by other factors including the conversion of the Benedictine monastery to a Cistercian nunnery and the relocation of the Benedictine monks about 50 km away to Cismar (1245-1247 CE) [43]. As the monastery had riverine fishing rights and stock ponds this change could have interrupted the availability of contaminated fish within the city. It is also possible that the availability of locally caught fish was affected by a shortage of fishermen following the Black Death (1346 -1350) or by changes in trading patterns as Lü beck matured as an economic centre [33,45,46]. Although we cannot exclude any single factor affecting the incidence of D. latum the maintenance of freshwater fish aDNA signature across the 1300 CE boundary supports a hypothesis of a reduction in consumption of (a)   infected fish which is most easily explained by pollution reducing the numbers of infected copepods. The continued presence of D. latum in two households indicates that at least some contaminated fish remained available post-1300 CE, although at a much-reduced level. The temporal change in occurrence and prevalence of T. saginata was less pronounced but highly significant, suggesting an increase in the consumption of raw or undercooked beef from the mid-thirteenth century CE on. This change may reflect the increasing wealth of Lü beck's citizens or an increased supply of beef to the markets. In other medieval cities, the Black Death resulted in an increased focus on cattle due to shortages in manpower for arable farming which may have influenced the source of beef in Lü beck. The introduction of Rinderpest virus into Europe (early 1300s) which affected a large proportion of cattle [47] may also have affected the availability of infected beef. Whatever factors affected the prevalence of food-transmitted cestodes in Lü beck the magnitude of infection is without precedence and indicates the importance of local social/dietary practices affecting the prevalence of food-transmitted diseases.
The widespread incidence of faecal -orally transmitted nematode parasites (Trichuris and Ascaris), affords a novel opportunity to use molecular epidemiology to interrogate historical events. We used molecular approaches to identify parasite species and generated a considerable dataset of T. trichiura ITS-1 sequences to identify epidemiological signatures. The most diverse groups of TtITS-1 sequences were found in Lü beck and Bristol, both port sites where parasite diversity might be promoted by trade-based connectivity. Of these two locations, TtITS-1 sequences from Lü beck were substantially more diverse reflecting its high level of connectivity with other sites. Indeed, Lü beck was one of the most important ports in Medieval Europe controlling a considerable proportion of the trade in the Baltic Sea and was also the founding site and leading city of the Hanseatic League. We propose that the intra-specific parasite diversity seen in Lü beck was a consequence of extensive trade with, and unintentional import of parasites from, other parts of Europe. Intriguingly, Medieval Bristol contained the second most diverse population of TtITS-1 sequences, supporting the hypothesis that the connectivity associated with ports increased parasite diversity. Hence, parasite diversity may be useful as a marker for the level of interaction that any particular site has with other regions. A full examination of this hypothesis requires exploration of a wide range of sites but is supported by the fact that Bristol was a less prominent port (and contained less TtITS-1 diversity) than Lü beck and both of these contained more diversity than other non-port sites.  Phylogenetic analyses revealed two groups of TtITS-1 sequences. While group 1 sequences were ubiquitous, group 2 sequences were prominent within Lü beck, rare in Bristol, and absent from all other study sites and modern sequences. Despite a smaller number of sequences, a similar two clade pattern was observed for T. trichiura b-tubulin sequences with Lü beck represented in both groups. Hence, Lü beck contained a unique, genetically distinct parasite lineage that was circulating in the medieval period but is no longer present in modern T. trichiura populations. Importantly, the only non-Lü beck site where group 2 T. trichiura parasites were detected was Bristol, albeit at very low numbers. This distribution supports the premise that group 2 parasites were more likely to have been transferred to Bristol from Lü beck or that these ports were interacting very differently with an independent source location for group 2 parasites. Interestingly, within Lü beck group 1 and group 2 T. trichiura sequences were not evenly distributed although there were no temporal or spatial patterns. Further sequence data, including a much larger set of locations and DNA targets would be needed to confirm the nature and distribution of the, apparently extinct, group 2 lineage.
A combination of parasitological and molecular approaches was used to identify changes in eating habits and trading patterns centred on Lü beck, a dominant medieval port and founding city of the Hanseatic League. We propose that the integrated application of molecular parasitology and epidemiology to the study of archaeological sites represents a novel addition to the repertoire of approaches that can be used to interrogate historical events.

Methods
(a) Sample handling and preparation workflow aDNA handling practices have been outlined in a range of publications [48 -50]. None of the parasites targeted in this study is endemic in the UK or any country where the material was received from. None of the laboratories has ever handled or stored modern samples containing these parasites, hence modern contamination is extremely unlikely. The processing pipeline involved sample preparation and PCR set up in a 'clean' laboratory within dedicated UV hoods using dedicated equipment with all handling of amplified material undertaken in a separate laboratory. Researchers also followed a strict unidirectional protocol, never entering the 'clean area' or handling samples after entering any downstream processing laboratory.

(b) aDNA extraction
Soil subsamples (5 g) were re-hydrated in 20 ml of PCR-grade water (Qiagen, Hilden, Germany) as described above. The samples were then sieved through a series of disposable, single-use nylon mesh sieves with decreasing aperture size (1 030 mm, 500 mm, 100 mm; Plastok Associates Ltd, Birkenhead, UK). The eggs present in the filtrate were pelleted by centrifugation (400g, 10 min) and the pellet homogenized in a BeadBeater (BSP BioSpec, Bartlesville, USA) using 1 mm glass beads (Hecht, Sondheim, Germany). aDNA was extracted using a Qiagen Blood&Tissue or Mericon Food kit (Qiagen). Blank extraction controls were routinely included as contamination checks.

(c) Microscopic diagnosis
Aliquots of the initial subsample were analysed microscopically using a Nikon Eclipse E400 with 10Â and 40Â lenses (Nikon UK, Kingston-Upon-Thames, UK). Photographs were recorded on a QImaging MP5.0 RTV camera (QImaging, Surrey, Canada). Parasite egg counts were extrapolated from replicated counts to the dry weight of the sample.

(d) PCR amplification
The PCR was performed in two stages; with the first stage boosting the starting concentration using a very robust polymerase (AmpliTaq Gold 360, ThermoFisher, Loughborough, UK) and a second stage to produce sufficient amounts of target DNA using a high-fidelity polymerase and to attach the barcodes for MiSeq sequencing (Phusion Hot Start Flex, New England Biolabs, Hitchin, UK). PCR primers were designed based on published sequences or reported previously (electronic supplementary material, Methods). Multiple water controls were included with each set of PCRs, producing either no product or primer concatemers.

(e) Sequencing
Second stage PCR products were cleaned up prior to sequencing (MinElute, Qiagen). Initial Sanger sequence data were generated using the BigDye Terminator (ThermoFisher).
Parallel sequencing libraries were made from pooled PCR products with barcoded primers. For the library prep NEB-Next Ultra DNA Library Prep Kit for Illumina with NEBNext Multiplex Oligos for Illumina (NEB) were used. The libraries were sequenced on a MiSeq (Illumina, Chesterford, UK) using the MiSeq reagent kit v. 3 (600 cycle). Sequences were separated according to library barcode using Basespace (Illumina) and then processed using custom python scripts, which identified amplicons with the correct forward and reverse primers, and sorted according to sample barcode. Sample barcodes consisted of 7 bp attached to one primer (based on Bystrykh, [51]) with an edit (Hamming) distance of 2, enabling single error correction in the barcode. Sequencing was at the Wellcome Trust Centre for Human Genetics (MiSeq) and the Department of Zoology (MiSeq/Sanger), University of Oxford.
A total of 419 936 MiSeq reads were obtained from parasite PCR amplicons resulting in 129 494 total parasite-specific reads (electronic supplementary material, table S5). The panvertebrate 16S amplicons were sequenced to an overall depth of 13 460 879 with 1 768 632 vertebrate specific reads and 146 813 food-species-derived reads (electronic supplementary material, table S6).

(f ) Temporal distribution of helminth eggs
The correlations between the numbers of the two nematode parasites Trichuris and Ascaris in Lü beck and Bristol were evaluated using GLMMs. To assess whether these shifts are best explained by a gradual change over time or by a sudden shift in parasite prevalence, we used Akaike's information criterion [52].
(g) Molecular phylogenetic analysis Phylogenetic history was inferred by using the Maximum-Likelihood method under on the Tamura -Nei substitution model [53] using MEGA7 [54].

(h) Haplotype networks
Median-joining networks were inferred for the ITS1 and the additional b-tubulin gene fragments using NETWORK 4.6.6 (fluxus-engineering.com) [37] with 1 ¼ 0. For the haplotype rspb.royalsocietypublishing.org Proc. R. Soc. B 285: 20180991 network inferences, sequences including sequencing ambiguities were removed from the alignment and indels (insertiondeletion) were coded as substitutions. Haplotype networks inferred for ITS1 and b-tubulin were separately coloured by sampling origin and by estimated dating. Furthermore, for ITS1, sub-networks were inferred for each time period with more than five identified sequences and coloured by sampling origin.
(i) Genetic diversity and population structure Nucleotide diversity and the relative nucleotide diversity were estimated for ITS1 within each location using SPADS 1.0 [55]. We also used SPADS to estimate the pairwise F ST statistics for ITS1 [56] among the different locations.
Data accessibility. All ancient DNA sequences have been deposited on the NCBI GenBank, accessible under the accession numbers MH599138 -MH600059.