SEASONAL SUCCESSION OF CESTODE METACOMMUNITIES IN MUSEUM COLLECTIONS OF TWO NORTH AMERICAN SHREWS (SOREX SPP.)
ABSTRACT
Seasonal patterns in parasite diversity and prevalence are determined by a suite of factors ranging from environmental conditions (e.g., temperature, precipitation) to host behavior and traits. The forces shaping these patterns are increasingly unstable due to anthropogenic climate change, emphasizing the importance of leveraging natural history collections to understand how historical environmental conditions contributed to the spatiotemporal distributions of parasites and parasitic diseases. We implemented frameworks from metacommunity analyses to test for seasonal trends in parasite community structure attributable to host and abiotic environment in 2 North American shrew species, Sorex cinereus and Sorex monticola. Shrews were collected using snap traps and pitfall traps periodically from 2009 to 2018 at 2 sites in Cowles, New Mexico. Whole gastrointestinal tracts from shrews were screened and tapeworm taxa were identified to genus using high-throughput amplicon sequencing of the 28S rRNA gene. Elements of metacommunity structure (EMS) revealed coherent, dynamic cestode metacommunities indicative of idiosyncratic responses of parasite taxa to seasonal trends in climate variables. We used boosted regression trees to identify latent variables predictive of cestode metacommunity structure and identified “day of the year” as having the greatest relative influence on parasite community structure, followed by precipitation and host body size. This work demonstrates for the first time the utility of the EMS framework for detecting fine-scale seasonal dynamics in parasitic helminth communities.
Detangling the relationships between host species, host environment, and the abundance and diversity of parasitic organisms is critical for disease ecology in a changing world. These factors interact across varying temporal scales with patterns of disease outbreaks spanning from those that cycle annually such as the classic example of measles (Soper, 1929) and seasonal flu in humans (Petrova and Russell, 2018) to diseases with multi-annual cycles like phocine distemper virus that move through populations of hosts over the course of many years (Härkönen et al., 2006). This wide breadth of temporal scales at which disease dynamics can manifest requires investigators to consider not only timescales that span across years, but also infection dynamics occurring between seasons in the same year, especially for host organisms with short lifespans.
In parasitic helminth communities, previous investigations incorporating seasonal sampling have relied on coarse, mainly categorical classifications of seasonality such as “rainy and dry” or “spring, summer, fall, and winter,” to determine if parasite species richness, prevalence, and intensity respond to seasonal cues from environmental and/or host traits (Altizer et al., 2006; Albery et al., 2018). Studies of parasite communities infecting bats (Nickel and Hansen, 1967; Blankespoor and Ulmer, 1970), as well as rodents (Simões et al., 2010), have demonstrated that the richness of some helminth communities is greater in dry seasons than in rainy seasons. However, a study of gastrointestinal (GI) helminths infecting sheep and rams (Sissay et al., 2008), as well as a more recent study of bat helminth communities (Salinas-Ramos et al., 2017), found no effect of rainy vs. dry conditions on helminth community composition. The conflicting consensus on the degree to which parasite communities experience seasonal succession prompts further investigation into the seasonality of parasite communities in host populations (Poulin, 2020).
Historically predictable abiotic features that vary on a seasonal basis, such as temperature and rainfall, are increasingly unstable and prone to new extremes due to anthropogenic climate change (Parmesan, 2006; Urban, 2021). Seasonal climate trends now deviate from historical norms, signaling that it will be increasingly difficult to generate baseline observations with which to compare both contemporary and future distributions of host species and their parasites (Harmon et al., 2019). A solution to this problem has been proposed through the utilization of natural history collections to study historical disease patterns (Bell et al., 2018), with 1 study successfully reconstructing detailed temporal patterns in parasite abundance in a single fish species over a period of 86 yr (Howard et al., 2019).
The millions of host organisms archived by museum collections globally over the last 2 centuries are a largely untapped resource for establishing baseline observations of the historical diversity and distribution of parasites, since many endoparasites were inadvertently preserved during the preservation of their hosts in formalin or ethanol. Preserved parasites in natural history collections serve as excellent snapshots, with measures of parasite diversity within hosts providing unique insight into changing biotic and abiotic conditions. Richer parasite communities often reflect diversity of prey communities, food web dynamics, host community diversity, and host population health (Poulin, 2010; Mihaljavec et al., 2018). Quantifying trends in parasite species richness can also inform estimates of population scale resistance and tolerance to infection, as multiple coinfections are often associated with increased energetic costs due to allocation of resources to immune response (Hicks et al., 2018). However, richness and abundance are often insufficient for characterizing meaningful but heterogeneous responses to ecological conditions at the community level since parasite community composition may change over time while maintaining constant species richness. These challenges highlight the additional need for statistical tools capable of teasing apart the often heterogeneous spatial and temporal facets of parasite community dynamics, not just richness and intensity.
In addition, museum-based studies may require statistical approaches that better accommodate presence/absence data. Identifying parasites from preserved host specimens can be challenging, since some endoparasites degrade prior to preservation or preserve poorly. However, presence/absence data from these taxa can often still be generated through techniques such as TaqMan probe based real-time PCR, high-throughput amplicon sequencing, or metagenomic approaches from gastrointestinal or other tissue DNA extractions (Greiman et al., 2018, 2020, 2022; Papaiakovou et al., 2022). Although presence/absence data represent a single axis of parasite infection prevalence in a population and parasite species richness within an individual host, these data can nevertheless be leveraged to study community wide trends in host occupancy.
Metacommunity ecology, namely, the EMS (elements of metacommunity structure) framework, provides an effective avenue for using presence/absence data to pursue mechanistic understanding of patterns of species distributions, capable of identifying species occurrence patterns that characterize “metacommunities” within larger communities of organisms (Leibold and Mikkelson 2002; Presley et al., 2010). This framework has recently emerged as a useful tool in infectious disease ecology as a means to characterize the structure of, and mechanisms shaping, parasite metacommunities (Dallas and Presley, 2014; Cardoso et al., 2018), with 1 study successfully identifying broad-scale spatial predictors of metacommunity structure in the disease vector Culecoides (Cleveland et al., 2023). The successful utilization of the EMS framework in the study of the diversity and distribution of disease in response to climate variables like temperature and precipitation demonstrates its utility for identifying environmental and host factors that dynamically shape parasite communities.
In this study, we sampled frozen GI tracts of 2 syntopic shrews, Sorex cinereus and Sorex monticola, obtained from the Museum of Southwestern Biology to investigate if tapeworm communities in these hosts undergo seasonal cycles in community composition. Sorex monticola is native to North America, and its range extends from the north of Mexico to northern Alaska. Sorex monticola is primarily insectivorous but occasionally eats plant material such as seeds, as well as lichens and fungi (Smith and Belk, 1996). Sorex cinereus is also a native of North America and is more widely distributed than S. monticola, ranging across diverse boreal habitats from Alaska and northern Canada south to the mountains of Appalachia (Hope et al., 2016). Shrews are short lived, and these 2 species breed roughly during the same time of year. Most litters are born in the spring with the potential to birth multiple litters across their lifespan of approximately a year. These species also have similar diets (Whitaker, 2004), and their high surface-to-volume ratios and high metabolic rates result in an extreme demand for prey items, meaning that shrews must be constantly foraging and consuming diverse diets of arthropods, fungal, and plant material (Smith and Belk, 1996; Whitaker et al., 2004). The diversity and volume of arthropod prey items consumed by shrews facilitates contact with diverse communities of intestinal parasites; North American shrews are known to host more than 97 helminth species, including 9 trematodes, 34 cestodes, 50 nematodes, and 4 acanthocephalans (Kinsella and Tkach, 2009).
We utilized the EMS framework paired with high-throughput amplicon sequencing of museum specimens to (a) examine intestinal cestode occurrence in preserved S. cinereus and S. monticola specimens, (b) determine if cestode communities undergo fine-scale seasonal changes in community composition, and (c) determine the degree to which environmental and host factors influence changes in parasite community composition over time.
MATERIALS AND METHODS
Shrew specimen collection
Shrews were collected every 3 wk in September 2009, from May to October in 2016 and 2017, and April and May of 2018 via snap traps and pitfall traps from two closely associated sties in the Sangre De Cristo Mountains near Cowles, New Mexico (NMGF permit ID: Cook#3300); Jacks Creek (35°49′50.88″N, 105°39′32.76″W, elev. 2,404–2,618 m) and Winsor Creek (35°48′57.96″N, 105°40′43.32″W, elev. 2,469–2,652 m). Selected sites were separated by a distance of 2.52 km to ensure that specimens were from the same population but far enough apart to avoid oversampling from one location. Proximity of the selection of collection sites to one another and similar habitat composition eliminated any significant confounding variables that might have been attributed to environmental and population differences. Trapping was largely done overnight with the collection of specimens occurring in the early morning. Traps that were set during the day were checked every 3 hr to ensure minimal degradation of tissue. Collected shrews were identified as either S. monticola or S. cinereus using morphological characteristics, including dentition, and verified using cytochrome b mtDNA barcode sequences PCR amplified and sequenced using forward primer L14734 and reverse primer H15985 (Raspopova and Shchipanov, 2011). Cytochrome b sequences were blasted in NCBI with 100% sequence match to either S. monticola or S. cinereus. Living specimens were euthanized using chloroform inhalation following University of New Mexico IACUC protocol (85-R-0002). Collected shrews were dissected in the field (Galbreath et al., 2019), and the entire GI tract of each specimen was removed and immediately frozen in liquid nitrogen. Holistic voucher specimens were accessioned into the MSB Division of Mammals at the University of New Mexico (Suppl. Table S1).
DNA extraction of helminth communities
DNA was extracted from the whole GI tracts of shrews using a ZR Fecal DNA miniPrep kit (Zymo Research, Irvine, California) following the methods of Greiman et al. (2018). In short, GI tracts were removed from their storage at −80 C and placed on a sterile Petri dish under a dissecting microscope. The GI was cut in half and opened lengthwise to evenly distribute tissue between 2 vials during the DNA extraction steps. To ensure that as much of the gut contents were extracted as possible without overburdening the spin columns, the internal contents of each half of the GI were scraped into separate vials with sterile forceps. Extracted DNA from each half was combined following the elution steps (150 μl for each elution). All tools and containers were soaked in a 10% bleach solution and rinsed with water between each sample to prevent cross-contamination.
Cestode community library preparation
Prior to amplicon library preparation, GI tract DNA extractions were quantified using a QubitTM dsDNA Broad Range Assay Kit (ThermoFisher Scientific, Waltham, Massachusetts) on a Qubit fluorometer. Samples were then diluted and standardized to 50 ng/μl for library preparation. Amplicon sequence library preparation of the large ribosomal subunit 28S rRNA gene was completed following the protocol of Greiman et al. (2018). In short, the 28S rRNA gene for cestodes was amplified in triplicate via PCR using the 28S_F_Ces (5′-GAGTAAACAGTACGTGAAGC-3′) and 28S_R_Ces (5′-CCACCGGTCGTGGTGTTC-3′) primers. PCR primers were appended with Illumina Nextera (Illumina Inc., San Diego, California) P5 and P7 adapters with dual indices. Following amplification of the target genes, libraries were pooled and normalized. PCR amplification was confirmed via gel electrophoresis, and aliquots of the pooled libraries were cleaned using 1 μl ExoSAP-IT (ThermoFisher Scientific) per aliquot to remove impurities. Following the ExoSAP-IT protocol, the product was quantified via a Broad Range Quant-iT dsDNA assay kit (ThermoFisher Scientific) and normalized to a concentration of 50 ng/μl. The libraries were then pooled. A 25 μl aliquot from each normalized pool underwent gel purification to remove extra primers using a Qiagen MiniElute Gel Purification Kit (Qiagen, Valencia, California). A High Sensitivity Quant-it dsDNA assay kit (ThermoFisher Scientific) was used to quantify the cleaned samples which were then diluted to 4 nM. Libraries were then sequenced using an Illumina MiSeq sequencer through the University of Georgia Genomics and Bioinformatics Core. Raw amplicon sequence data are available through the NCBI SRA under the BioProject accession number PRJNA1177369 and assembled 28S sequences of each unique operational taxonomic unit (OTU) are provided in the supplemental data (.fasta) file.
Bioinformatics analysis
Sequence libraries were processed using Mothur software package (v. 1.37.3) as described in Kozich et al. (2013) for quality control, sequence filtering and alignment, clustering, and identification of operational taxonomic units (OTUs). For data from both cestode samples, ambiguous nucleotide base calls were removed, as well as forward and reverse sequences, and paired-end reads were combined by extracting sequence reads and quality scores. The command “make.contigs” was used to extract the sequence reads and create contigs from the samples. The command “screen.seqs” was used to eliminate ambiguous base calls found outside of the anticipated sequence length (Kozich et al., 2013). Cestode 28S rRNA sequences that were over 300 base pairs in length were removed. Duplicate sequences were merged and grouped using the command “unique.seqs,” and a table quantifying the number of appearances of each unique sequence in each group was created using the command “count.seqs” (Kozich et al., 2013). The command “pcr.seqs” was used to match sequences to reference alignments working in synergy with “align.seqs” and “screen.seqs” to make certain that the sequences are all roughly the same size and that the amplicons are all accurately positioned in accordance with each other (Kozich et al., 2013).
Cestode analysis in Geneious
Single-gene reference sequences from morphologically identified cestode species collected from shrews at the same sites were assembled and aligned into reference libraries using the Geneious Prime Build 2023.2 software (Biomatters Ltd., Auckland, New Zealand). Prior to analysis in Geneious Prime, some processing was performed in Mothur. Contigs were created from extracted sequence reads using the command “make.contigs.” The command “screen.seqs” was then used to eliminate ambiguous base calls found outside of the anticipated sequence length (Kozich et al., 2013). In Geneious Prime, a global BLASTN search against a database constructed from reference libraries of morphologically identified cestode species was conducted to assign haplotypes. The species reads identified in Geneious were grouped by genera, and reads with less than 98% identity were placed in groups titled “Unknown Hymenolepididae” or “Unknown Dilepididae.”
Environmental variables
We used the PRISM (PRISM Climate Group, 2014) database to access measures of environmental variables known to influence helminth distribution and abundance including average monthly precipitation (Shearer and Ezenwa, 2020) and average monthly maximum and minimum temperature (Cooper and Hollingsworth, 2018). Although climate variables 2 mo prior to parasite sampling have been shown to influence parasite dynamics in ungulates (Shearer and Ezenwa, 2020), that system presumably has a much slower turnover than the host-parasite system studied here. Therefore, all climate variables were averages for the month and year in which each individual specimen was collected.
Analysis of metacommunity structure
We constructed a host-parasite presence/absence matrix as outlined in Dallas and Presley (2014), where host corresponds to individual shrew observations and the presence of parasites are coded as either “1” for present or “0” for absent within an individual host. We ordered the site-host-parasite matrix via reciprocal averaging, assigning a “host score” to each host individual based on similarity of cestode infection patterns between hosts and parasites (Gauch, 1982) using the “metacom” package (Dallas and Presley, 2014) in R. Host score corresponds to parasite community infection similarity, where shrews with similar parasite infection patterns are assigned scores that are close together and shrews with dissimilar infection patterns have scores distant from each other. Following the ordination of the host-parasite matrix, we characterized idealized structures within the parasite communities using measures of parasite community coherence, the number of embedded absences (i.e., interruptions in the distributions of species or in the compositions of hosts) in a matrix; parasite community turnover, the number of times 1 species replaces another between 2 sites for each possible pair of parasites and for each possible pair of hosts; and parasite community boundary clumping, determining if parasite communities replace each other along a shared gradient or respond individually to environmental and host features resulting in an overall metacommunity structure (Leibold and Mikkelson, 2002).
Generalized boosted regression models
To investigate the drivers of parasite metacommunity structure (host score) in S. cinereus and S. monticola, we used boosted regression trees (GBM) to estimate the relative influence of factors possibly predictive of our response variable of host score: host mass (g) (Sherrard-Smith et al., 2015) and sex (Krasnov et al., 2012; Hostert et al., 2019), precipitation (Nickel and Hansen, 1967; Blankespoor and Ulmer, 1970; Simões et al., 2010; Shearer and Ezenwa, 2020), and maximum/minimum temperature (Cooper and Hollingsworth, 2018) using the R package “gbm” (Elith et al., 2008; Ridgeway, 2013). Because reciprocal averaging was used to obtain host score, values for host score are both negative and positive, requiring the rescaling of host score by adding the minimum value to each score, resulting in a new minimum score of zero. We then square root transformed scores into a roughly normal distribution. We used cross-validation to determine the optimal number of trees with 70% training data and 30% testing data. We accounted for interactions by setting interaction depth to 4 and fit the model using 3,000 trees with shrinkage = 0.01 and evaluated model performance using root mean squared error.
Generalized additive models
Since our boosted regression trees only reveal predictors of overall metacommunity structure (host rank), and helminth metacommunities comprise assemblages of individual parasite taxa, we used binomial generalized additive models (GAM) with a logit function to model host and environmental predictors of parasite prevalence for the 8 parasite genera with greater than 20 observations using the R package “mgcv” (Wood, 2011) with smooth terms applied to each predictor variable.
RESULTS
A total of 166 shrews were collected for this study between September 2009 and May 2018 (Suppl. Data, Table S1). In total, 73 shrews were collected from Jacks Creek (13 S. cinereus; 60 S. monticola), and 93 were collected from Winsor Creek (40 S. cinereus; 53 S. monticola). The communities of cestodes infecting S. cinereus and S. monticola at these localities largely belong to the genera Lineolepis, Mathevolepis, Monocercus, Staphylocystoides, Urocystis, Ditestolepis, and a potential new genus (Hymenolepididae gen. nov.) (S. E. Greiman, unpub. morphological data) within Hymenolepididae. In S. monticola and S. cinereus, cestode community composition and relative prevalence of individual parasite taxa were similar. We used a Welch’s t-test to compare prevalence of cestode taxa between host species. Of the 10 common cestode genera present, 4 showed biases in prevalence between S. cinereus and S. monticola. Notably, Mathevolepis, Staphylocystoides, and Hymenolepididae gen. nov. were more prevalent in S. cinereus than in S. monticola, while cestodes from Unknown Dilepididae were most prevalent in S. monticola. There were no cestode genera specific to either host species (Table I; Tables S2–S10), reinforcing the ecological similarity of these 2 host species and their paired analysis here.
Through the analysis of the elements of metacommunity structure (EMS) framework, we evaluated the cestode metacommunity for evidence of community coherence, species turnover, and boundary clumping. We found that the parasite metacommunity consisted of coherent cestode communities identifiable within the larger cestode species pool infecting shrews at these localities (z = 1.65, P < 0.0001) (Fig. 1). The cestode metacommunity contained a significant number of species replacements (z = 3.89, P < 0.001), suggesting parasite species turnover through time. We found that the ranges of ranked host individuals in which a given cestode genus was detected did not overlap strongly with the ranges of other cestode genera indicating the absence of boundary clumping among parasites (Morisita's index = 1.94, P = 0.12). When considered in concert, these results indicate that these cestode taxa display individualistic responses to environmental and host gradients, as opposed to a community-wide response to a shared suite of climate conditions (Gleason, 1926; Leibold and Mikkelson, 2002) (Fig. 1).
Citation: The Journal of Parasitology 111, 6; 10.1645/24-45
Based on relative contribution (RC) of each model covariate to cestode community structure, we found that day of the year (DOTY) (RC = 34.31), mean monthly precipitation (RC = 18.32), shrew mass (RC = 17.15), and monthly maximum temperature (RC = 16.45) were the most influential predictors of cestode community structure (Fig. 2), while collection year was the least predictive (RC = 2.57). Values for the response variable, host score, ranged from ∼0 to 2.5. Measures of model performance resulted in an adjusted R2 of 0.865, and a root-mean-squared error of 0.022 (Figs. 2, 3).
Citation: The Journal of Parasitology 111, 6; 10.1645/24-45
Citation: The Journal of Parasitology 111, 6; 10.1645/24-45
The results of our GAMs revealed that the prevalence of some cestode taxa including Urocystis, Monocercus, and Staphylocystoides changed in response to abiotic conditions such as precipitation and temperature. Prevalence of Mathevolepis was determined by combinations of environmental conditions and host mass (Fig. 4). Other cestode taxa, namely, Staphylocystis, Ditestolepis, and the group of Unknown Hymenolepididae, demonstrated seemingly random infection patterns unrelated to environmental factors, host traits, or day of the year (Table II; Suppl. Data, Figs. S1–S7; Tables S2–S10). Two cestode genera, Hymenolepididae gen. nov. and Staphylocystis, were site specific, where Hymenolepididae gen. nov. occurred only at Winsor Creek, and Staphylocystis was specific to Jacks Creek.
Citation: The Journal of Parasitology 111, 6; 10.1645/24-45
DISCUSSION
Elements of metacommunity structure (EMS) approaches revealed strong evidence of coherent, dynamic cestode metacommunities in Sorex spp. with fine-scale, seasonal turnover in parasite community composition progressing with day of the year (Fig. 5). The parasite taxa detected here displayed individualistic responses to both environmental and host traits. These results are consistent with previous studies reporting patterns of parasite community responses to environmental and host traits, where factors contributing to trends in infection prevalence as well as intensity in helminths are idiosyncratic and vary among parasite and host taxa (Albery et al., 2018; Cirino et al., 2022). At first glance, the metacommunity structure may appear Clementsian in nature due to the overall parasite metacommunity changing with DOTY. However, most covariates measured vary inherently with day of the year, particularly those predictive of individual parasite prevalence. A meta-analysis of seasonal dynamics of parasite infections across a broad range of both host and parasite taxa in aquatic communities found that seasonal infection dynamics are not easily generalizable, as they are more often controlled by contrasting parasite, host, and environmental traits (Poulin, 2020). We used boosted regression trees to detect latent variables predictive of cestode metacommunity structure, measured as host score generated by reciprocal averaging of a site × species matrix, and identified DOTY as having the greatest relative influence on parasite community structure, followed by average monthly precipitation and host body size (Figs. 2, 4). These results are consistent with other studies of seasonal helminth dynamics that implicate precipitation as a main driver of seasonal helminth infection patterns. A study of helminth parasites in free-ranging Grant’s gazelles found that rainfall in the 2 mo preceding sample collection was the best predictor of infection patterns for all 3 helminth taxa measured (Shearer and Ezenwa, 2020). Future investigations should account for El Niño and La Niña events, which are important ecological drivers of disease distribution and abundance in the southwestern United States.
Citation: The Journal of Parasitology 111, 6; 10.1645/24-45
Host factors such as reproductive status (Shearer and Ezenwa, 2020), host age (Albery et al., 2018), and immunocompetence (Fair and Whitaker, 2008) fluctuate with seasonality and might explain at least some of the temporal trends in cestode community structure in S. cinereus and S. monticola (Felis and Esch, 2004; Šimková et al., 2005). Host mass (g) had the third highest relative contribution to cestode metacommunity structure (RC = 17.5) behind DOTY and average temperature. Albery et al. (2018) found in their study of seasonal helminth infections in red deer that strongyle nematodes were more prevalent in calves than adults, with prevalence decreasing with age. We observed a significant relationship between host mass and parasitism in S. cinereus and S. monticola for only 1 of the cestode taxa detected, where Mathevolepis prevalence decreased dramatically with both host mass and DOTY. We included shrew mass as a proxy for host age, as the 2 are strongly correlated within Sorex spp. The relationships observed here may signal mechanisms associated with host age, since most litters in these host taxa are born in early spring and live for only approximately 1 yr. However, the observed negative relationship between host mass and Mathevolepis prevalence could also be due to heterogeneity in prey items utilized by large vs. small host individuals (Hostert et al., 2019) (Fig. 5).
There is some evidence that temporal trends in helminth abundance and diversity might be closely tied to changes in abundance and diversity of arthropod prey items that serve as their intermediate hosts (Lord et al., 2012). Moisture availability and average temperature are factors demonstrated to influence seasonal patterns in terrestrial arthropod diversity and abundance (Zhu et al., 2010; Liu et al., 2013). An analysis of a long-term dataset of soil arthropod communities collected over 18 yr in Zackenberg, Greenland, found cyclical patterns in soil arthropod community composition that responded primarily to seasonal variations in average temperature and winter duration, with dryer habitats displaying the most drastic changes in arthropod communities over time (Koltz et al., 2018). The seasonal turnover of arthropod communities could alter the distribution and abundance of cestodes infecting species of Sorex. Cestodes have indirect life cycles that utilize intermediate hosts, and in Hymenolepididae these hosts usually are an arthropod. Although cestode life cycles often vary slightly between species, they largely follow the same life history strategies. For Hymenolepididae infecting Sorex spp., intestinal worms shed eggs that are released in shrew feces and then ingested by the appropriate arthropod intermediate host, where they hatch in the intestines, penetrate and enter extraintestinal space, and develop as infective metacestode larvae. Following consumption of infected arthropods, metacestodes develop into adult tapeworms in the host intestine where they reproduce sexually. We identified DOTY, temperature, and precipitation as key predictors of parasite metacommunity structure in this system, and it is likely that the relationship between DOTY and cestode metacommunity structure is, at least in part, a manifestation of arthropod community turnover in response to seasonal variations in temperature and precipitation. Characterization of the local diversity and seasonal dynamics of arthropod prey communities utilized by these species could provide useful insight into the micro-niches occupied by S. cinereus and S. monticola as well as the relative contributions of host traits, abiotic conditions, and prey communities to seasonal trends in parasite metacommunity structure. In turn, temporal monitoring of these parasite communities provides an important window into community dynamics that links together helminths and arthropod and mammalian hosts (Vidal-Martinez et al., 2010).
We found that collection year contributed minimally to parasite metacommunity structure, with a relative contribution of ∼2.4% (Fig. 2). This could be an artifact of sampling bias, as monthly sampling effort was not consistent between all collection years. Specimens collected in 2009 were all from the month of September, and the specimens from 2018 included individuals collected only in April and May due to extensive forest fires that year that prevented field collections. However, when considered in concert with other predictors and their relative influence on metacommunity structure such as day of the year (32.9%), precipitation (18.6%), and maximum temperature (16.7%), the relative influence of collection year on parasite metacommunity structure was minimal. This is a promising finding for future, museum-based studies as these resources are often highly heterogeneous in sampling over space and time.
The cestode community composition observed here very closely matches previous reports for S. cinereus in North America (Kinsella and Tkach, 2009), including species of Cladotaenia, Ditestolepis, Lineolepis, Mathevolepis, Soricinia, Skrjabinacanthus, Staphylocystis, and Staphylocystoides. Kinsella and Tkach (2009) did not report a cestode community assemblage for S. monticola, although some of their cestode records may have been misclassified because S. monticola was formerly considered a subspecies of Sorex vagrans (Senger, 1955; Kinsella and Tkach, 2009). Also, at the time of Kinsella and Tkach (2009), none of the shrews surveyed for helminth fauna were collected in New Mexico. Of the cestode genera reported by Kinsella and Tkach (2009) for S. cinereus, all but Cladotaenia are found in both S. cinereus and S. monticola in New Mexico. Also, Monocercus and Urocystis are prevalent in both S. cinereus and S. monticola in this study but were unreported in S. cinereus by Kinsella and Tkach (2009). Although cestode taxa are considered somewhat host specific, both species of shrew surveyed here housed many of the same cestode genera. This is not surprising considering that sympatric or geographically proximate species of Sorex share much of their cestode fauna (Kinsella and Tkach, 2009; Binkienė et al., 2011; Haukisalmi, 2015). However, S. cinereus and S. monticola differed in overall cestode species richness, with average cestode richness for S. cinereus 4.32 (SD = 1.97) and S. monticola 3.47 (SD = 1.81), t(94.42) = 2.67, P = 0.009, d = 0.45. Although their diets have a large degree of overlap, other differences between these shrew species such as average body size might result in asymmetrical exposure to some helminth taxa as a function of prey community composition being limited by host size. With arthropod prey items serving as intermediate hosts for the cestodes detected here, we once again identify the need for investigations into the associations between cestode infection and arthropod prey items utilized by these species as an illuminating future direction.
Mathevolepis, Staphylocystoides, Urocystis, and the Hymenolepididae gen. nov were all more prevalent in S. cinereus than S. monticola, likely related to the observed greater cestode species richness in S. cinereus discussed above. Decreased host fitness resulting from parasite infection is a selective pressure for resistance strategies. However, those same adaptive strategies of the host increase selection on parasites (Buckling and Rainey, 2002). The answer concerning which factors determine host infection bias in this system likely lies in the evolution of local populations of S. cinereus and S. monticola in this region and requires consideration of how S. cinereus and S. monticola differ in exposure and susceptibility to the described cestode communities (Park et al., 2017). Furthermore, our study only diagnosed cestode taxa to the genus level, but it is possible that a particular genus was represented by multiple species in the samples collected. Taxonomic resolution of this study was limited primarily by the nature of short read amplicon sequencing <300 bp fragments of a gene. Identification of cestode taxa to species could provide clarity to nuanced mechanisms responsible for changes in cestode community composition over time. Previous studies have shown that, although many cestode genera are found across multiple congeneric shrew species, there are cases where cestode species infect some Sorex species but not others (Kinsella and Tkach, 2009; Binkienė et al., 2011; Haukisalmi, 2015). Predictors of parasite community composition and prevalence of parasite taxa are not universal (Poulin, 2020). The spatial extent of this study area and the diversity of host taxa considered are narrow, and these insights and preliminary hypotheses should be tested in other systems.
Using preserved specimens from the Museum of Southwestern Biology collected over several years, we provide evidence that there are clear, temporal trends in parasite community composition in S. cinereus and S. monticola in the mountains of northern New Mexico. Future work should endeavor to characterize the arthropod prey communities utilized by these host taxa (e.g., Gibson et al., 2022) and identify factors that contribute to exposure and susceptibility to cestode infections. As more genetic sequences from parasitic worms are submitted to data repositories such as GenBank, future researchers will be better able to utilize molecular techniques to identify cestode species from natural history collections, a task not easily accomplished through morphological examination of poorly preserved parasites. Improving parasite taxonomic resolution will in turn increase our ability to tease apart complex mechanisms shaping parasite community composition and driving temporal and spatial patterns in parasite diversity and abundance. This work further demonstrates the utility and critical role that natural history collections can play in studying changes in parasitism within host communities over time. The increasing deviation of abiotic conditions from historical norms emphasizes the importance of detangling contemporary effects of those forces on parasite communities and harnessing new insights to better forecast future spatiotemporal disease distributions in wildlife species.
Presence/absence matrix generated through reciprocal averaging of cestode communities in museum-archived Sorex cinereus (n = 53) and Sorex monticola (n =113) collected at 3-wk intervals in 2009 and 2016–2018 from 2 localities in the Sangre De Cristo Mountains near Cowles, New Mexico, with cestode genera identified through 28S rDNA metabarcoding. On the y-axis, hosts are ranked based on parasite community similarity (similar communities uppermost), and the x-axis shows cestode taxa. Black ticks represent cestode presence, white regions indicate cestode absence.
Bar plot of relative influence of boosted regression model covariates on the cestode communities in museum-archived Sorex cinereus (n = 53) and Sorex monticola (n =113) collected at 3-wk intervals in 2009 and 2016–2018 from 2 localities in the Sangre De Cristo Mountains near Cowles, New Mexico, with cestode genera identified through 28S rDNA metabarcoding.
A partial dependency plot for the top predictor of host rank from the boosted regression model. Host rank based on cestode community similarity within museum-archived Sorex cinereus (n = 53) and Sorex monticola (n = 113) collected at 3-wk intervals in 2009 and 2016–2018 from 2 localities in the Sangre De Cristo Mountains near Cowles, New Mexico, in relation to day of the year, is shown on the y-axis and is plotted as a line. Cestode genera were identified through 28S DNA metabarcoding.
Dual axes plot of presence of specific cestode genera and host rank as they relate to day of the year showing shrew host individuals ranked by cestode community similarity on the y-axis and parasite genera on the x-axis. Shrews, Sorex cinereus (n = 53), and Sorex monticola (n = 113), were collected at 3-wk intervals in 2009 and 2016–2018 from 2 localities in the Sangre De Cristo Mountains near Cowles, New Mexico. Black boxes indicate cestode presence. Color gradient corresponds to the day of the year that each host individual was collected. The top panel contains observations collected near the Winsor creek site, and the bottom panel contains observations from the Jacks creek site. Color version available online only.
Partial dependency plots for generalized additive model covariates (maximum temperature, day-of-the-year, and host mass) as they relate to the prevalence of the cestode genus Mathevolepis within museum-archived Sorex cinereus (n = 53) and Sorex monticola (n = 113) collected at 3-wk intervals in 2009 and 2016–2018 from 2 localities in the Sangre De Cristo Mountains near Cowles, New Mexico. “s(variable)” indicates that a smoothing function was applied to the respective term to account for nonlinear trends in continuous data.
Contributor Notes