Archaeological analyses

Excavations at the Sorcé site (18˚05’ 56” Latitude North and 65˚29’ 34” Longitude West) in the Island of Vieques, Puerto Rico from 1977 to 1984 recovered faunal osseous remains from deposits Z and YTA-2, while the coprolites were recovered from deposits Z, ZT, YTA-1 and YTA-2 (Fig 1). No permits were required, the excavations took place in a private property (Sorcé, Vieques) and the owner’s consent was provided for the excavations by the Archaeological Research Center at the University of Puerto Rico, which complied with all relevant regulations. Coprolites ages (Table 1) were determined by radiocarbon dating of material associated with the samples [2]. Faunal osseous remains were identified by Narganes-Storde [3] via comparative analysis with a synoptic collection from the Zooarchaeology Laboratory of the Florida State Museum and the Center for Archaeological Research of the University of Puerto Rico (S2 Table). The coprolites and faunal osseous remains belong to an archaeological collection and they are permanently deposited in the Center for Archaeological Research of the University of Puerto Rico.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Description of the coprolites used in the molecular analysis.

https://doi.org/10.1371/journal.pone.0227810.t001

Microscopy

Portions of the coprolites (n = 20) were processed for microscopy using traditional flotation methods which allowed for the systematic observation of helminth eggs (S1 Table) [12]. Briefly, one gram of each sample was rehydrated in 14ml of 0.5% trisodium phosphate for 72 hours [13], shaken vigorously and filtered (1500μm mesh) to eliminate debris. Subsequently, 1ml of 10% acetic formalin solution was added per 10g of filtrate [14]. The samples were allowed to settle for 72 hours [15], and ten microscope slides were prepared using 50μl of sediment mixed with a drop of glycerin. Each slide was covered with a 20x20 cover slip and scanned microscopically in a serpentine fashion [16].

Preparation, DNA extraction and sequencing

Extraction and sequencing of ancient DNA (aDNA) was performed as described by Rivera-Perez et al. [17]. Briefly, nine coprolites, namely Huecoid (n = 5) and Saladoid (n = 4) were selected for shotgun metagenomic sequencing. Coprolites were processed in a reserved area of the laboratory for ancient DNA to avoid contamination. The coprolites were processed separately in a class II biosafety cabinet that was routinely disinfected with 70% ethanol and exposed to UV light for at least 30 minutes before and after use. All instruments were autoclaved and baked overnight at >100°C to denature any extraneous DNA. To reduce the presence of soil microbiota, the exterior layer was removed and the core of each coprolite was used. The cores were grounded separately using a sterile mortar and pestle. Ancient DNA was isolated using a PowerSoil DNA Extraction Kit (Mo Bio Laboratories, Carlsbad, CA) according to manufacturer’s instructions. All samples were hydrated overnight in sterile C1 buffer at 4˚C prior to extraction. Using standard glycogen precipitation, 10ul of aDNA were pooled according to ethnic group (MixS1 and MixH1) to compensate for low concentrations of aDNA. The concentration of aDNA was assessed using a Qubit® dsDNA HS Assay Kit (Life Technologies) and sequenced at a commercial facility (MR DNA Research lab in Shallowater, TX). REPLI-g Midi kit (Qiagen) was used for non-targeted whole genome amplification (WGA) followed by Nextera library preparation kit and sequencing with Illumina MiSeq system.

Putative parasite sequences and phylogenetic assignment

Fastq files produced by Illumina MiSeq were assessed through MG-RAST [18] for quality control and read length exclusion based on default parameters. Amino acid predictions of the metagenomic datasets were conducted using BLASTX [19] against a non-redundant protein NCBI database (National Center for Biotechnology Information). To address the damage present in aDNA [20], the cut-off value for functional identification was set at an E-value of <-15. Sequences with specific functions and high alignment scores to parasite reference reads were verified using MEGA 7 [21] for accurate taxonomic identification of BLASTX homologous results. BLAST hits were downloaded from NCBI and concatenated into fasta files with the putative parasite read extracted from the metagenomic dataset. The fasta file were imported to MEGA 7 for multiple sequence alignment using MUSCLE [22]. Substitution model was selected using Find Best DNA/Protein Models. The suggested model was then used to create a pairwise distance matrix and construct maximum likelihood phylogenetic tree using 1000 bootstrap iterations (see Supplementary Material).

Modeling parasite-host interaction

HelminthR [23] and rglobi (global biotic interactions) [24] curated databases for host-parasite interactions were used to verify zooarchaeological data as potential hosts for the identified parasites. Parasite-host interactions were reconstructed by generating a directional dataset of the identified parasites (detected by microscopy or molecular analysis) and general descriptions of potential host detected from the zooarchaeological data. Parasite-host interactions were modeled using the network graphical R package ‘igraph’ [25].

Supplementary materials

Parasite sequences are available in S1 Data. HelminthR and rglobi search results are available in S2 & S3 Data. The generated dataset and Rscript used in this study are included in S4 & S5 Data. BLAST homology of putative reads of the consumed animal are included in S6 & S7 Data.

Fecal-oral transmission

Giardia intestinalis is a zoonotic protozoan parasite with a wide host range transmitted by the fecal-oral route [32]. Cryptosporidium spp. and Giardia spp. life cycles are direct and develop to completion within one host [32–33]. Ortega and Bonavia [30] detected Giardia spp. and Cryptosporidium spp. in pre-Columbian Peruvian coprolites, although co-infection was not observed in the samples. Giardia spp. and Cryptosporidium spp. were identified in our metagenomic datasets. However, Cryptosporidium was not identified to the species level and therefore was not included in the network analysis as each of its species has a varied host range [32]. In our study G. intestinalis was assigned the highest Eigen value of the parasite nodes in the predicted network model, indicating that G. intestinalis could have been the most easily transmitted zoonotic parasite in the settlement.

Consuming raw or undercooked infected animal hosts

Eimeria spp. and Perkinsus marinus are regarded as epizootic diseases that are easily transmitted to hosts that live in close proximity, particularly livestock [34–35]. Neither are zoonotic infections, for instance, birds are the definitive hosts of E. necatrix whereas bivalves (Crassostrea and Mya) are the definitive host for P. marinus, and there is no evidence of Eimeria spp. and P. marinus being pathogenic to humans [36–37]. Bird bones were extracted from the Huecoid (n = 1,185) and Saladoid (n = 1,727) archaeological deposits. Most of these osseous remains were identified as Columbidae in Huecoid (n = 748) and Saladoid (n = 306) deposits. Other bird bones were detected in both deposits, including Rallidae, Pelicanidae, Ardeidae, Anatidae, Phoenicopteridae and Psittacidae among others (S2 Table). Few Psittacidae (Huecoid n = 1 and Saladoid n = 13) remains were found, though they were occasionally consumed, the cultures most likely kept Parrots and Parakeets as pets because of their colorful plumage [4].

Eimeria oocysts have been identified in prehistoric ruminant coprolites from Brazil [38]. Clearly Eimeria spp. was present in pre-Columbian America, unlike Perkinsus marinus which to the best of our knowledge has not been detected in pre-Columbian America samples. Perkinsus marinus is highly seasonal, transmission is influence by warmer climates [35]. A variety of bivalve shells were isolated from the deposits, including Crassostrea. Huecoid and Saladoid cultures scavenged beach shorelines for bivalves and other forms of small marine life [3]. Eimeria spp. and P. marinus putative reads detected in the metagenomic dataset may be the result of consumption of raw or undercooked infected hosts and thus their presence in the feces may be transient Finding Eimeria spp. and P. marinus in human coprolites provided additional evidence that birds and bivalves were part of their diet.

Diphyllobothrium spp. infects fish-eating mammals (definitive host). Diphyllobothrium eggs were previously detected in human and canid pre-Columbian coprolites from Peru and Chile [39–40]. Recently Diphyllobothrium spp. has been recognized as Dibothriocephalus spp., thus Diphyllobothrium spp. is a synonymized name. The parasite has two successive intermediate hosts; the first is a copepod and the second a freshwater or marine fish [41–42]. Diphyllobothriidea infection is associated with the ingestion of its raw or undercooked secondary intermediate host [41–42]. In the case of Diphyllobothriidea, cooking the fish kills the plerocercoid larvae in the muscle tissue [41–42], thus the detection of Diphyllobothrium spp. eggs could only have occurred by ingesting raw or under cooked infected fish. The Sorcé settlement was 70 meters from the sea, and a variety of marine fish osseous remains were extracted from the Huecoid (n = 1,941) and Saladoid (n = 43,660) archaeological deposits, suggesting both cultures supplemented their diets by fishing [4].

Transmission by ingestion of intermediate hosts

Canids are the definitive hosts for Dipylidium caninum, subsequently canids are reservoirs for the double-pored dog tapeworm. A total of 107 and 168 Canis familiaris bones were extracted from the Huecoid and Saladoid deposits (S2 Table), four whole canid remains were found laid out in a burial disposition suggesting endearment towards an animal companion, while scattered canids osseous remains without anatomical context were also dispersed throughout the deposits suggesting canids as occasional food source [4]. Thus far, there is no evidence of D. caninum human infections associated to the ingestion of infected canids, rather human infections are currently associated with accidental ingestion of its flea vector Ctenocephalides spp. [43]. Theoretically, prehistoric cultures may have controlled lice infections by ingesting them while grooming [44–46]. Similar infections have been described in pre-Columbian America in 1,400 year-old coprolites found in Mexico where Dipylidium spp. and Hymenolepis spp. eggs were detected [10].

Hymenolepis spp. sequences were not detected in the metagenomic dataset, however, microscopy suggested the presence of hymenolepidids. Hymenolepididae associated with human infections can also infect rodents, for instance, H. nana frequently infects humans whereas H. diminuta infections are uncommon in human hosts [47]. In the case of H. microstoma, the mouse bile duct tapeworm is questioned as whether to be regarded as a potential zoonotic parasite to humans [48]. Some rodent hymenolepidids are of health interest to humans, since they can cause infections in immunosuppressed individuals [49].

Throughout history, some rodent species have lived as commensals in human settlements [50]. Spanish historians reported that indigenous cultures raised rodents in small corrals [51] to ensure a constant supply of the animal for dietary purposes [4]. Examples include Isolobodon portoricensis commonly known as the Puerto Rican Hutia (Huecoid n = 1 and Saladoid n = 323) and the Spiny Rat known as Heteropsomys insulans (Huecoid n = 47 and Saladoid n = 19), whose osseous remains were identified in the archaeological deposit (S2 Table). Numerous unidentified rodents bones were also accounted for in Huecoid (n = 151) and Saladoid (n = 50) deposits. It is likely that the rodents that lived in close proximity or within the settlement influenced the transmission of zoonotic infections such as hymenolepidids.

Animal osseous remains found in Sorcé suggest the cultures possible diets, likewise detecting animal and plant aDNA in the gastrointestinal tract (GI) of these individuals is suggestive of their consumption. DNA-based methods have been applied to extant feces to assess the diets of herbivores, carnivorous and omnivorous animals [52–55]. As the ingested tissue passes through the GI tract, DNA from the prey species is substantially degraded [52]. Putative homologous sequences of canids, rodents and fish were observed in the metagenomic datasets (S5 Data). Several putative sequences resulted in a significantly low quality alignment to an animal reference reads which was possibly due to DNA damage inflicted by taphonomical processes [20] and the digestive process of the ingested tissue [52]. A similar study performed metagenomic sequencing of calcified dental plaque of medieval human skeletons, BLASTN alignment to a chloroplast and mitochondrial database revealed putative plant and animal reads [56]. Confirming the putative reads, microscopy examination detected conserved dietary microfossils fragments (such as plant fibers, starch granules, and animal connective tissue) in the dental calculus and zooarchaeological analysis of the medieval site confirmed the presence of animal osseous remains (Suidae, Caprinae, cattle and equids) as the potential protein food source [56]. Detecting putative sequences of animal and plant in the GI tract of ancient cultures via shotgun sequencing is suggestive and must be validated using diverse methods.

Schistosomatidae

Schistosoma spp. has two free-swimming stages that penetrates the intermediate and definitive host skin (Fig 2A and Table 3) [57]. Phylogenetic inference clustered the sequences to Schistosomatidae group. The sequences may correspond to S. mansoni since the eggs are excreted in the host feces. However, it is possible the sequences may also be associated to another member of the digenean taxon, for example Trichobilarzia spp. since it was a close second to Schistosoma spp. in a pairwise distance matrix (S11 Table). Trichobilarzia spp. definitive hosts are waterfowl, infections have been reported worldwide as migratory waterfowl (such as Anatidae) facilitate the spread of avian schistosomiasis [58–60]. Consequently Anatidae osseous remains were identified in the deposits. Trichobilarzia spp. does not mature in humans (accidental host) and instead causes an allergic skin reaction [60–62]. If the cultures ingested the infected tissue of waterfowls, theoretically the parasite would pass through the human GI tract without causing infection. If the sequence is related to Trichobilarzia spp., this would consequently alter the network model making canids the principal source of zoonotic infections in the settlement (S4 Table). Trematode eggs were detected by microscopy in a Saladoid coprolite, but could not be identified to the genus level.

Schistosoma spp. were not discarded as a potential pathogen. This parasite requires slow-flowing or a still water source for its life cycle to progress and a stream ran near the settlement (Fig 1). Therefore, inhabitants could have potentially acquired the Schistosomatidae and other fresh water related parasites while engaged in activities in the mentioned stream. Low prevalence in wild small mammals (such as rodents) were reported as reservoirs of zoonotic schistosomiasis in West Africa, namely S. mansoni, S. bovis and S. haematobium [63]. If supported, then it could be argued that rodents inhabiting the Sorcé settlement could have been reservoirs for both zoonotic schistosomiasis and Hymenolepid tapeworms. Thus far, ancient Schistosoma spp. have mainly been detected in mummies from Egypt and China [14]. To the best of our knowledge Schistosoma spp. infections have not been described in pre-Columbian America. Although false taxonomical assignment is a possibility regarding aDNA (see Author’s Statement), we believe that this is not the case seeing as phylogenetic inference strongly suggested the Schistosomatidae group. Thus, if further analyses support these data, this would be the first report of Schistosomatidae in pre-Columbian ethnic groups in America.