ABSTRACT
The life cycle and ecology of the horsehair worm Chordodes morgani (Nematomorpha) in Nebraska remain unknown. To identify its definitive host, we installed a series of pitfall traps along 3 first-order streams at 4 sites: Elk Creek, Upper Elk Creek, Maple Creek, and West Oak Creek, all located northwest of Lincoln, Nebraska. In addition, we opportunistically hand-collected insects at these sites, including wood cockroaches (Parcoblatta virginica), and maintained them in the lab until they passed adult worms. Two of these field-collected wood cockroaches each yielded 1 adult worm, which was identified as C. morgani by microscopy, showing that P. virginica serves as a definitive host. Experimental infections of captive-reared Parcoblatta americana supported this result. The wood cockroach was found at all 3 creeks, but C. morgani was not found at West Oak Creek, suggesting that the definitive host does not limit the distribution of C. morgani. Physical properties of the streams were measured to examine how these properties influenced the distribution of the worm. Flow rate and pH differed between the 3 sites where C. morgani was found and the West Oak Creek site, suggesting an important role for these abiotic factors in the distribution of this horsehair worm species.
The ecology of nematomorphs, also known as gordian or horsehair worms, is poorly understood (Chandler, 1985; Poinar, 1991), in part because, unlike other parasites, adult horsehair worms exit their definitive host when they are sexually mature. They do this by manipulating their definitive arthropod hosts to enter water where the adults emerge and mate (Thomas et al., 2002, 2003). Females lay eggs in the water column, in vegetation, on sticks or rocks, or in the substrate. Larvae develop within eggs and encyst in an aquatic invertebrate host (Hanelt and Janovy, 1999, 2004; Schmidt-Rhaesa, 2001; Bolek et al., 2015). The life cycle continues when a terrestrial insect consumes the aquatic paratenic host (Hanelt and Janovy, 2003). This complex combination of free-living and parasitic stages makes the study of the horsehair worm difficult. Of the more than 360 described freshwater Nematomorpha species (Yadav et al., 2018), less than 30% of the definitive arthropod hosts are known (Harkins et al., 2016), and definitive host specificity has been tested only in Paragordius varius, Chordodes morgani, Gordius robustus (Hanelt and Janovy, 2003), and Chordodes kenyaensis (Bolek et al., 2013). In addition, few studies have examined the factors that correlate with free-living hairworm abundance in aquatic habitats, and such information is lacking for most species of gordiids, including most species of Chordodes (Poulin, 1996; Bolek and Coggins, 2002; Salas et al., 2011).
Seasonality impacts horsehair worm abundance in a variety of species in different locations, such as in Argentina among Chordodes brasiliensis, Noteochorododes cymatium, Noteochorododes talensis, and Pseudochordodes dugesi (Salas et al., 2011), in Wisconsin with Gordius difficilis (Bolek and Coggins, 2002), in France with Paragordius tricuspidatus and Spinochordodes tellinii (Schmidt-Rhaesa et al., 2005), and in New Zealand with Gordius dimorphus (Poulin, 1996). For example, Schmidt-Rhaesa et al. (2005) noted that the abundance of the 2 horsehair worms matches the abundance of their respective definitive hosts. However, Salas et al. (2011) noted that seasonal differences in abundance observed in subtropical Argentina, in contrast to the differences observed in temperate regions where water bodies freeze in the winter, suggest that horsehair worm abundance, in addition to the presence of their definitive hosts, could also be explained by the presence of their aquatic paratenic hosts.
With few exceptions, the genus Chordodes is predominantly restricted to tropical and neotropical regions of the world (Schmidt-Rhaesa, 2002; Achiorno et al., 2008; Chiu et al., 2011; Schmidt-Rhaesa and Lalramliana, 2011; Bolek et al., 2013; Schmidt-Rhaesa et al., 2015; Schmidt-Rhaesa and Perissinotto, 2016). Although little documentation exists on definitive host use for most species of Chordodes, when known, mantids, roaches, or orthopterans have been documented as definitive hosts (Montgomery, 1907; Studier et al., 1991; Poinar, 1999; Schmidt-Rhaesa and Ehrmann 2001; Schmidt-Rhaesa, 2002; Chiu et al., 2011; Bolek et al., 2013). However, and as mentioned above, it remains unclear as to the range of species of definitive hosts most species of Chordodes can infect.
Chordodes morgani has been found in eastern North America and as far west as Nebraska (Studier et al., 1991; Poiner and Chandler, 2004; Robison et al., 2012). It has been reported from orthopterans and cockroach definitive hosts across its range (Schmidt-Rhaesa et al., 2003). Specifically, Studier et al. (1991) reported cave crickets, Hadenoecus subterraneus, and camel crickets, Ceuthophilus stygius, as definitive hosts for C. morgani from Kentucky. In contrast, Montgomery (1907) and Schmidt-Rhaesa et al. (2003) reported C. morgani from an unidentified cockroach (location unknown) and from the smoky brown cockroach, Periplaneta fuliginosa, from South Carolina and nymphs of an unidentified species of wood cockroaches in the genus Parcoblatta from South Carolina and Missouri, respectively. Finally, McAllister and Hanelt (2012) found C. morgani in the oriental cockroach, Blatta orientalis, in Oklahoma. The paratenic host remains unknown.
Chordodes morgani adults were found in Elk Creek near Lincoln, Nebraska, between 1999 and 2000 (Hanelt et al., 2001), wrapped around dead twigs, laying eggs, but not in nearby creeks in the same county of Lancaster (B. Hanelt, pers. comm.). This fact allows us to ask what factors correlate with the presence of C. morgani. To answer this, we sampled Elk Creek and nearby creeks for adult horsehair worms and terrestrial insects in July 2017 and 2018. We measured the chemical and physical parameters of each creek as well as vegetation mass. We hypothesized that the presence of the definitive arthropod host will correlate with the presence of C. morgani.
MATERIALS AND METHODS
Study sites
We sampled 4 sites in Lancaster County, Nebraska, representing 3 different creeks. Elk Creek (40°53′11.1″N, 96°50′4.5″W) is located 5.8 km downstream from Upper Elk Creek (40°54′56.8″N, 96°53′29.0″W). Maple Creek (40°54′12.0″N, 96°55′47.5″W) is located 8.2 km northwest of Elk Creek, while West Oak Creek (40°55′49.36″N, 96°51′16.7″W) is located 5.1 km north of Elk Creek (Fig. 1). These sites were chosen because they are all first-order streams that flow into the Platte River via Salt Creek and are within 10 km of each other, and 2 of the 4 sites have historical data (Hanelt et al., 2001).
Map of the 4 sampling sites, representing 3 different first-order streams: Elk Creek, Upper Elk Creek, Maple Creek, and West Oak Creek. The wood roach (Parcoblatta virginica) was found at all 4 sites, but Chordodes morgani was not found at West Oak Creek.
Map of the 4 sampling sites, representing 3 different first-order streams: Elk Creek, Upper Elk Creek, Maple Creek, and West Oak Creek. The wood roach (Parcoblatta virginica) was found at all 4 sites, but Chordodes morgani was not found at West Oak Creek.
Collection efforts
To collect insects that are potentially infected with C. morgani, we placed 74 pitfall traps and 10 baited traps at the 4 sites in July of 2017. Pitfall traps consisted of polyethylene containers (15 × 15 × 5 cm) filled with a mixture of 3:1 water:propylene glycol solution following Ernst et al. (2016). Traps were placed in 7-cm-deep holes, made flush with the ground, and covered with cardboard squares (16 × 16 × 1 cm) secured with 4 steel nails (10 cm). Traps were serviced daily to remove specimens and debris and to refill the mixture as needed. Specimens collected from the traps were placed into plastic bags and identified to genus (Froeschner, 1954; Bland, 2003; Brust et al., 2008; Evans, 2014). They were then dissected and examined for the presence of horsehair worms.
Baited traps consisted of polyethylene containers (25 × 25 × 13 cm) placed in 15-cm-deep holes, made flush with the ground, and covered with cardboard squares (26 × 26 × 1 cm) secured with 4 steel nails (16 cm). The bait mixture consisted of fermented beer, garlic, onions, oats, and peanut butter (modified from Greenslade and Greenslade, 1971). Live specimens collected from the baited traps were placed in plastic vials for examination in the lab.
While servicing the traps, arthropods were also collected opportunistically by hand, maintained in the lab in separate vials (100 ml culture container, Carolina Biological Supply Co., Burlington, North Carolina), and provided with a moist cotton ball, pieces of autoclaved bark for cover, and cat food. We maintained all arthropods until either worms emerged from the arthropods or they died. If they died, we dissected the remains to check for the presence of horsehair worms. We made no effort to prompt the worms to exit the roach (i.e., we did not submerge the roaches), and worms that emerged from living arthropods were identified by microscopy following descriptions and keys (Schmidt-Rhaesa et al., 2016).
We also collected aquatic snails (Physa sp.) from each site using strainers and by hand. These were dissected and cysts were counted (Hanelt et al., 2001) but not identified to genus based on morphology (Szmygiel et al., 2014).
To collect adult C. morgani from the streams, we took advantage of their tendency to wrap around sticks to lay their eggs by constructing artificial stick traps. These consisted of 6 wooden skill sticks (Creatology, 11.4 × 0.9 × 0.2 cm) tied with fishing line, which served to collect both adult C. morgani and floating vegetation. We secured one end of the line to a steel garden stake, which was placed in the creek, and the other end was free to float in the water. Each day, we checked both the artificial stick traps as well as naturally occurring sticks and vegetation for horsehair worms. After 18 days, 10 artificial stick traps, with their accumulated vegetation, were randomly selected from each site, placed in a paper bag, and dried in an oven for 24 hr at 60 C. The dried organic matter was pulled or scraped off the stick traps and weighed to the nearest 0.01 g (Scout Pro SP202, Ohaus Corp., Pine Brook, New Jersey) to estimate the amount of vegetation as available habitat for adult horsehair worms. To account for the bounded and continuous nature of the variable vegetation dry weight, we fit a generalized linear model with a gamma error structure using the glmmTMB package in R (Brooks et al., 2017; R Core Team, 2020). In cases where the recorded values were 0 (n = 3), we added 0.0001 because the gamma error structure allows only for positive values. Confidence intervals were extracted with the ggeffects package (Lüdecke, 2018).
Site properties
To measure the physical properties of the stream we constructed a 10-m transect and measured the width of the creek at 0, 5, and 10 m. We then measured depth at 3 equidistant points (0, 5, and 10 m) and measured surface velocity by recording the time a fishing buoy takes to float 5 m, repeated 5 times. The average depth and width were used to calculate the cross-sectional area of the creek, which was multiplied by a velocity adjustment coefficient of 0.8 and the average surface velocity to determine the creek's flow according to the float method (Harrelson et al., 1994). A YSI-Xylem ProPlus probe (Yellow Springs, Ohio) measured pH, dissolved oxygen, nitrates, salinity, and temperature at each site once a week for 4 wk in July. Data were averaged (n = 4) and analyzed for differences among sites using a 1-factor ANOVA.
In the summer of 2018, we sampled for C. morgani adults from each of the 4 sites and measured the diameter of the sticks and vegetation upon which they are found. Measurements were taken in the field with calipers to the nearest 0.01 mm (UBANTE digital calipers ISO9001:2000, Shenzhen, China).
Experimental infection
In the lab, 34 captive-reared wood cockroaches (P. americana), purchased from a roach supplier who had maintained them until most were second-generation subadults, were individually housed in vials, provided with a moist cotton ball and a cat food pellet. Roaches were starved 24 hr before exposure to snails (Physa) collected from Elk Creek. Microscopy first confirmed that snails contained C. morgani cysts using identifying characteristics described in Szmygiel et al. (2014). The soft parts of the remaining snails were removed, homogenized, and evenly deposited among the roaches. Multiple exposures were necessary because we failed to find cysts in most snails. Repeat exposures occurred 4, 8, 17, 23, 30, 38, 45 and 52 DPE (days post-exposure).
RESULTS
Collection efforts
Over the 25 days of sampling in 2017, we collected 54 C. morgani adults: 26 from Elk Creek, 19 from Maple Creek, and 9 from Upper Elk Creek. We did not collect C. morgani adults from West Oak Creek. We collected 15 C. morgani adults from artificial stick traps, while the rest were collected from vegetation and naturally occurring twigs. Of the 40 worms identified by microscopy, 8 were too decayed to allow proper identification, 5 were Gordius cf. robustus, and 27 were C. morgani. Samples were deposited at the Museum of Southwestern Biology, Parasitology Division, University of New Mexico, Albuquerque, New Mexico, MSB: PARA:31791–31796, for parasites and MSB: HOST:24226–24231.
Over the 25 days of sampling in 2018, we collected 71 C. morgani adults: 17 from Elk Creek, 38 from Maple Creek, and 16 from Upper Elk Creek. We did not collect C. morgani adults from West Oak Creek. We collected 16 C. morgani adults from artificial stick traps, while the rest were collected from naturally occurring sticks. The average diameter of sticks with C. morgani adults (n = 28) was 6.9 mm.
Pitfall traps yielded 476 terrestrial arthropods from the 4 sites (Table I; 235 Coleopterans, 135 Orthopterans, and 106 assorted arthropods such as spiders, flies, millipedes, ants, and isopods). None were infected with C. morgani. Baited traps yielded 5 wood cockroaches (Parcoblatta virginica) collected from all 4 sites, and none were infected with C. morgani. Two hand-collected wood cockroaches from Elk Creek were identified as P. virginica (Froeschner, 1954) and maintained in the lab until, 3 days later, each yielded 1 adult worm, which was confirmed as C. morgani by microscopy. Lab-infected wood roaches (Parcoblatta americana) yielded 2 adult worms 133 DPE and 178 DPE. Dissected snails (Physa) from Elk Creek averaged 10 cysts (n = 15, range = 0–74), while snails from Maple Creek averaged 18 cysts (n = 18, range = 0–3). Snails from West Oak Creek (n =28) had no cysts, and we failed to collect any snails from Upper Elk Creek.
Site properties
Sites did not differ in dissolved oxygen, temperature, or salinity (Table II). The flow rate was significantly slower at West Oak Creek than at both Elk Creek and Maple Creek but did not significantly differ with Upper Elk Creek (ANOVA, P < 0.01). The pH at Upper Elk Creek was significantly higher than West Oak Creek (ANOVA, P < 0.05). Nitrate level at Elk Creek was higher than at Maple Creek, but Maple Creek had a lower nitrate level than Upper Elk Creek (ANOVA, P < 0.05). West Oak Creek significantly differed from Upper Elk Creek in vegetation dry weight (P = 0.00357, glmmTMB).
Summary statistics with the mean (and 95% confidence interval) of the physical and chemical parameters at each of the 4 sites. Significant differences between sites are indicated with symbols (*, †, ‡, §, ||). The total number of adult horsehair worms collected in 25 days in 2017 and 2018 is also reported.

DISCUSSION
Although C. morgani has been reported in Nebraska (Robison et al., 2012), the biotic and abiotic factors that correlate with the presence of this species remain mostly unknown. The emergence of 1 C. morgani adult from each of the 2 field-collected wood cockroaches (P. virginica) shows that the worms use these roaches as their definitive host and are the first confirmed species of a wood roach acting as a host for C. morgani. Parcoblatta americana occurs in the same genus as the field-collected roaches, and both are found in North America (Hebard, 1917). Exposure of these captive-reared wood cockroaches to aquatic snails from Elk Creek supports the suitability of the wood cockroach as a definitive host for C. morgani, which is consistent with the findings of Schmidt-Rhaesa et al. (2003), who report C. morgani in a Parcoblatta sp. nymph as well as in the more distantly related Periplaneta fuliginosa. It is also consistent with previous observations of 4 other Chordodes spp. using roaches as a definitive host (Poinar, 1999). Failure to find other arthropods infected with C. morgani does not leave out the possibility that other arthropods could also serve as a definitive host. Likewise, failure to find C. morgani adults or cysts at West Oak Creek after 2 summers of sampling does not necessarily mean this horsehair worm is absent from West Oak Creek. More extensive sampling, especially during other parts of the year, is required. Given that we collected wood cockroaches from all 4 sites but did not collect adult C. morgani at West Oak Creek suggests that the presence of the definitive host does not necessarily correlate with the presence of C. morgani.
Since C. morgani lays its eggs on sticks, some vegetation must be present in its habitat. Stream vegetation occurred at all 4 sites, and its accumulation on the artificial stick traps varied as measured by its dry weight from 0 to 49.7 g among the 4 sites. Despite the statistically significant difference between Upper Elk Creek and West Oak Creek, Elk Creek averaged the largest amount of stream vegetation with West Oak Creek having the second-largest amount (Table II), suggesting a poor correlation between vegetation and C. morgani abundance. Other factors such as the paratenic host or abiotic factors may limit the horsehair worm's distribution.
The pH at Upper Elk Creek was significantly higher than the site without C. morgani (West Oak Creek). This result is consistent with Salas et al. (2011), who found that the abundance of Chordodes brasiliensis in Argentina was greatest in autumn, which also had a high pH. The impact of pH on the free-swimming horsehair worm adult, its egg, or larval stages has not been studied. Adult horsehair worms spend relatively little time in the water, especially compared to its other life cycle stages. However, pH can impact possible paratenic hosts. For example, lab experiments showed that caddisflies are more tolerant of low pH than mayflies (Bell, 1971; Gaufin, 1973). The paratenic host of C. morgani may prefer a high pH environment.
Flow rate at Elk Creek and Maple Creek, where C. morgani was abundant, was significantly faster than at West Oak Creek, suggesting that flow rate may correlate with C. morgani abundance. This result is consistent with Salas et al. (2011), who found that the abundance of C. brasiliensis in Argentina was greatest in autumn, which also had the fastest flow rate. Flow rate correlates with dissolved oxygen, and lab experiments show that low oxygen impacts the survival and emergence of mayflies, midges, and caddisflies (Nebeker, 1972), which often serve as paratenic hosts for horsehair worms (Schmidt-Rhaesa, 2012). However, we failed to find differences in dissolved oxygen among the 4 sites. Flow rate, along with other interconnected factors such as food quantity and river depth, also impacts the distribution of filter-feeding insects (Richardson and Mackay, 1991), which could serve as paratenic hosts. Identification of the paratenic host would assist future studies on the relationship between flow rate and the abundance of C. morgani.
Various stages of at least 1 species (Chordodes nobilii) are susceptible to herbicides and other toxins (Achiorno et al., 2008, 2009, 2010). Nitrates, like herbicides, are associated with agriculture, and excessive nitrate levels can lead to algal blooms, eutrophication, degradation of aquatic systems (Pauwels and Talbo, 2004; Xu et al., 2016), and negatively impact human health (Ward et al., 2005). The impact of nitrates on the adult, egg, or larval stages of horsehair worms has not been studied. However, nitrates' negative effects on aquatic insects, which could serve as paratenic hosts, have been demonstrated only in levels that exceed 4.5 mg/L (Camargo and Ward, 1995), which is higher than that observed in this study (Table II). Further, the higher nitrate level at Elk Creek does not seem to correlate with horsehair worm presence since we also found them at the site with the lowest nitrate level (Maple Creek; Table II), but not at the site with the second-lowest nitrate level (West Oak Creek).
Water temperature, pH, flow rate, and other abiotic factors change with the seasons, especially in temperate regions. Knowing how these abiotic factors impact the abundance of horsehair worms and their definitive and paratenic hosts will help clarify the relationship between horsehair worms and water quality. For example, P. varius requires high-quality water to maintain water balance (Yoder et al., 2014). The complex life cycle of Nematomorpha may make it more susceptible to environmental degradation as has been observed in larval trematodes (Huspeni and Lafferty, 2004; Huspeni et al., 2005; Shea et al., 2012). Ehlke (1979) found that Trichopterans, Ephemeropterans, and Nematomorphs were eliminated from a stream that was downstream from a landfill, but this observation was based on the collection of a genus that was not identified. Thus, we do not know if this horsehair worm species relied upon the presence of Trichopterans, Ephemeropterans, or both to complete its life cycle.
On the other hand, Hanelt et al. (2001) found no correlation between horsehair worm prevalence and land use and human population density, suggesting that Nematomorpha is resistant to anthropocentric disturbances. In a sampling of snails in Oklahoma, cysts and larvae of P. varius were found at 50% of the sites, while cysts and larvae of C. morgani were found at 11% of sites (Harkins et al., 2016), suggesting large species differences in prevalence. To identify areas suitable for snail infections with gordiids, these data were then used with the environmental variables of elevation, stream flow direction, streamflow accumulation, land use data, and distance from stream to construct an ecological niche model. Although the model successfully predicted the presence of encysted snails, Harkins et al. (2016) did not include measures of water quality.
Collecting more data on the abiotic and biotic factors associated with horsehair worm abundance will assist in elucidating its complex life cycle, allow researchers to better predict its presence in an environment, and help researchers rear horsehair worms in the lab.
ACKNOWLEDGMENTS
We thank Yoonjae Im, Amy Tanoue, Janet Huynh, Ralph Mendez, Chris Smith, and Izzy Utschig for helping to check traps and take measurements in 2017. We also thank McKay and Kane Carstens, Simeon Gboun, Sydney Dang, Quinn Sano, Elijah St. John, and Patrick Kuwong for helping to check traps and take measurements in 2018. This study was partially funded by a Center for Undergraduate Research and Scholarship Faculty grant. We thank Jon Larson and Kyle Koch of University of Nebraska–Lincoln Extension for confirming the insect identifications. Fieldwork was conducted under Nebraska Scientific and Educational Permits 1030 and 1103. The authors would also like to thank the 2 anonymous reviewers for feedback that improved this paper.