Parasitol Res (2015) 114:895–901 DOI 10.1007/s00436-014-4254-5
ORIGINAL PAPER
Factors influencing spatial variation and abundance of a mermithid parasite in sand hoppers Trent K. Rasmussen & Haseeb S. Randhawa
Received: 7 August 2014 / Accepted: 3 December 2014 / Published online: 14 December 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract The impact of parasites on host population dynamics depends on local abundance of the parasites, which may vary considerably across spatial scales. In sand hopper populations, mermithid parasites have major impacts on host dynamics, which may vary among spatially separated populations due to the sand hopper’s wide, patchy distribution. The present study compared the abundance and biomass of a mermithid parasite (Thaumamermis zealandica Poinar et al., 2002) in sand hoppers (Bellorchestia quoyana (MilneEdwards)) both within and among disconnected beaches. In addition, several variables were measured and tested as potentially important predictors of the parasite abundance and biomass. It was found that geographic isolation may only be responsible for minor differences in parasite populations compared with other factors. Host size was identified as the most important predictor of mermithid parasite abundance, but epibiont abundance, kelp patch mass and host density were poor predictors of abundance. These factors were also poor predictors of parasite biomass in hosts. This study further supports the notion that studies aiming to elucidate population dynamics or patterns should sample thoroughly across both spatial and temporal scales. Keywords Parasite abundance . Parasite biomass . Spatial distribution . Bellorchestia quoyana . Body size . Epibionts
Introduction Parasites can regulate the population dynamics of their hosts in various ways, whether by causing mortality to host T. K. Rasmussen (*) : H. S. Randhawa Ecology Degree Programme, Department of Botany, University of Otago, PO Box 56, Dunedin 9054, New Zealand e-mail: [email protected]
individuals, reducing their fecundity or altering their behaviour (e.g. Anderson 1979; Ebert et al. 2000; Lafferty 2004). Such impacts of parasites on host populations depend largely on local abundance of parasites, which can vary considerably across spatial scales (Bates et al. 2010). Differences in parasite abundance among spatially separated host populations may reflect differences in susceptibility to parasites, a consequence of past co-evolutionary history with the parasite (Webster et al. 2004; Lohse et al. 2006). On the other hand, spatial differences in parasite abundance may also represent differences in infection risks faced by hosts, resulting from local abiotic and biotic factors (Thieltges et al. 2009). Whatever the case may be in a system, understanding causes of spatial variation for parasites is clearly critical to understanding host population dynamics. The talitrid amphipod Bellorchestia quoyana (MilneEdwards) (previously known as Talorchestia quoyana) provides a good system for looking at parasite spatial variation. B. quoyana is the most common sand hopper species living on New Zealand’s beaches (Morton and Miller 1973). It plays an essential role in consuming beach-stranded kelp (Inglis 1989; Dufour et al. 2012) and in nutrient cycling through the marine environment (Brown 2001). For these sand hoppers, spatial variation in parasite abundance acts on both small and large scales. On a small scale, they are distributed in patchy aggregations beneath available kelp, which allows for high-scale aggregations of their associated symbionts (Poulin and Rate 2001). On a large scale, sand hopper populations on different beaches can be separated by many kilometres, which could potentially allow marked differences in parasite abundance. The sand hopper B. quoyana is a known host of three symbiont species: a digamasellid mite (Dendrolaelaps sp.), a rhabditid nematode (unidentified) and a mermithid nematode Thaumamermis zealandica Poinar et al., 2002 (Poulin and Rate 2001). The former two species are epibionts that are not parasitic on the sand hopper host, whereas the mermithid
896
T. zealandica is an internal parasite (Poulin and Rate 2001). The infective stage of the parasite penetrates the body wall of the sand hopper and develops in its body cavity until maturity and then emerges into a moist sand environment as a freeliving adult, killing the host in the transition (Poinar et al. 2002; Currey and Poulin 2006). Not only does this parasite directly impact host populations, but also the mermithid may have further consequences on host dynamics via manipulation of the sand hopper’s burrowing behaviour. However, evidence for this behavioural effect has been mixed thus far (Poulin and Latham 2002; Currey and Poulin 2006, 2007). For instance, an experiment suggested that parasites may alter the burrowing depth of sand hoppers (Poulin and Latham 2002), but this result was not confirmed in field studies (Currey and Poulin 2006). Previous research has revealed that the mermithid parasite and epibionts of B. quoyana are each distributed heterogeneously within and among high-density host patches occurring across a beach (Poulin and Rate 2001). However, to our knowledge, no studies prior to the present one have examined whether the sand hopper mermithid parasites are distributed heterogeneously among separated beach populations. Moreover, little is known about what factors influence the variation of the mermithid within or among beaches. The present study aimed to address these knowledge gaps. In this study, we investigated the distribution of the mermithid T. zealandica among naturally infected sand hopper populations, both within and among separated beaches. We compared mermithid abundance and biomass between beaches and with several other variables of interest to determine which factors were important determinants of its variation. We measured host features (host length and sex), host density, epibiont abundance (mites and rhabditid nematodes) and the mass of the kelp patch covering each sample site. We hypothesised that mermithid abundance (the number of mermithids per sampled host) would be more different among populations on separated beaches than among those on a single beach. This would make sense if the populations on separate beaches were in isolation for enough time to cause differences in their genetic or population structure. In line with this, in our analyses, we predicted that beach identity would account for a large proportion of variation in abundance and biomass of the parasite.
Materials and methods Study sites and sampling procedure Sand hopper (B. quoyana) individuals were collected from six beaches during summer (December 2013–January 2014) around Dunedin, New Zealand. The sampled sites were the following: Brighton Beach (45.949° S, 170.333° E),
Parasitol Res (2015) 114:895–901
Tomahawk Beach (45.906° S, 170.546° E), Smails Beach (45.905° S, 170.564° E), Aramoana Spit (45.780° S, 170.709° E), Allans Beach (45.876° S, 170.698° E) and Long Beach (45.754° S, 170.649° E). Sand hoppers were collected from beneath three patches of kelp (three separate sites) per beach. To assess spatial variation within each beach, these three sites were selected randomly from across a stretch of a few hundred metres along the strand line or just above the high water mark during low tide. All selected kelp patches were in a similar stage of decay but varied in size (so they were collected and later weighed). There were six sample collection trips in total, one per beach, and all trips were done in sunny weather (strong winds and rain which may have affected their behaviour were avoided). Directly below each kelp patch site, an open metal cylinder (20 cm long with a diameter of 7.5 cm, a volume of approximately 884 cm3) was pushed into the sand to collect a volume of sand containing hoppers. The collected volume of sand was rinsed through a sieve (mesh size of 1.405 mm) with water to remove sand, leaving all the contained sand hoppers alive. Following this, the sand hoppers were transferred into closed buckets (one per kelp patch), transported to the laboratory and processed immediately. Only one cylinder sample was collected from each site on Brighton Beach, but for all the other beaches, two samples were collected from each site to increase the size of the sand hopper samples. This difference was accounted for in density estimates by standardising the density to the number of sand hopper individuals per cubic meter. Lab procedure A total of 946 sand hoppers was examined for symbionts. We found no mermithid-infected juveniles in our samples, and the number of infected juveniles was low in a previous study (Currey and Poulin 2006), suggesting that the importance of juveniles as hosts for mermithid parasites is negligible. For this reason, we restricted our analyses of mermithid abundance and biomass patterns to include only adult sand hoppers (n=583). Each sand hopper was measured (length from the front of the head to the tip of the telson in mm) and sexed on the basis of the male’s specialised gnathopods. All individuals shorter than 10 mm were classified as juveniles to avoid falsenegative classification of males. Following measurements, individuals were immediately decapitated and placed in a few drops of seawater. After allowing a few minutes, all symbionts (mermithids, rhabditid nematodes and mites) were counted under a dissection microscope. Mermithids were preserved in 70 % ethanol, and their wet weights were recorded (in mg to two decimal places). The total wet mermithid biomass per host was used as a response in addition to abundance in the analyses because biomass may show patterns not reflected in abundance; e.g. biomass may better reflect energy demands of the host (George-Nascimento et al. 2002).
Parasitol Res (2015) 114:895–901
Moreover, parasite biomass may be a more accurate measure of infection intensity than the number of parasites since it gives increased weighting to larger parasite individuals that are likely to have greater impacts on the host (by using more space and host resources). Analysis of results Mermithid abundance and biomass were compared between male and female sand hoppers using standard t tests (assuming unequal variance). Two main statistical models were used to analyse the results in this study, both utilised in the R environment (R Development Core Team 2008). The first one was a generalised linear mixed model used to analyse parasite abundance (the response variable) against eight predictors, including three with random effects (beach identity, kelp site (nested within beach) and host sex) and five with fixed effects (host length, host density at site (individuals/m3), mite abundance, rhabditid nematode abundance and kelp patch mass (kg)). This model was assigned a negative binomial distribution and accounted for zero inflation using the “glmmADMB” package (Bolker et al. 2012). The corrected Akaike information criterion (AICc) and Akaike model-averaged weights were calculated using the “MuMIn” package (Barton 2013) to compare component models with all combinations of the above predictors. These techniques provide a useful measure of relative variable importance (Anderson 2008). The second model used was a linear model with residuals of the regression between the hosts’ combined mermithid mass and host length as the response variable (with Gaussian distribution). The residuals between biomass and length were used instead of just biomass to account for collinearity between the variables (residuals were not used in the first model since abundance and length were not collinear). Only observations from the infected individuals were used in this model, but the predictors (excluding length) were the same as those used in the first model. As with the first model, all possible component models were compared by AICc. Both models (mermithid abundance and biomass) were re-run, but instead of having mite and nematode abundance as predictors, we used residuals of regression between their respective abundance and host length as predictors. The re-run first model also had mermithid abundance (the response) replaced with residuals of the regression between abundance and host length. The aim of re-running the models in this way was to observe whether variable effects changed when accounting for collinearity between symbiont abundance and host length.
Results Across the beaches sampled for this study, the mermithid parasite T. zealandica was distributed with considerable
897
heterogeneity both among separated beaches and patches on individual beaches. Prevalence of the mermithid parasite was 6.16 % in the entire sample (n=583) but ranged from 0 % up to 14.58 % per beach. There were clear differences among beaches in both abundance and biomass of the parasite; no parasites were found in the samples from two of the six beaches, and of the four beaches where mermithids were detected in the samples, parasite abundance and biomass were considerably higher at certain beaches. Notably, parasite abundance was twice as high at Aramoana Beach compared with Allans Beach and Tomahawk Beach (Table 1). Also, average biomass of the parasite was substantially low at Tomahawk Beach compared with the others (Table 1). However, it is worth noting that parasite abundance and biomass were also highly variable across each individual beach, as shown by large standard deviations (Table 1). Like the mermithid parasite, the sand hopper epibionts displayed great heterogeneity. Mites were particularly abundant at Long Beach with nearly six mites on average per sand hopper, compared with the majority of other beaches sampled which, on average, had one or fewer mites per individual. Rhabditid nematodes were also remarkably more abundant at certain beaches; abundance ranged from around three nematodes per individual at Allans Beach up to around 21 per individual at Tomahawk Beach. However, like the mermithid parasite, both epibionts showed highly variable abundance among individuals of the same beach (Table 1). In terms of host sex, mermithid abundance and biomass displayed different patterns. Across the entire data set, the parasite was significantly more abundant in female sand hoppers (0.16) compared to males (0.05) (p=0.03, t=2.23, degrees of freedom (df)=529.8). This pattern appeared consistent for each individual beach, but the effect of sex was also subject to high variability (Fig. 1a). In contrast, average mermithid biomass was actually higher for males (0.93 mg) than females (0.62 mg), though not significant (p=0.37, t= 0.95, df=6.7), and the effect of sex was again subject to high variability (Fig. 1b). It should also be noted that due to low infection rate in the sand hoppers, mermithid biomass comparisons were subject to low sample size. The best AIC models for mermithid abundance were very different from those for the parasite biomass (Table 2). The top AIC model (AICc =367.73) for parasite abundance included only the host length. However, the second best AIC model, which also included rhabditid nematode abundance, and the third best AIC model, which also included kelp patch mass, were very close to the best model (ΔAICc
ORIGINAL PAPER
Factors influencing spatial variation and abundance of a mermithid parasite in sand hoppers Trent K. Rasmussen & Haseeb S. Randhawa
Received: 7 August 2014 / Accepted: 3 December 2014 / Published online: 14 December 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract The impact of parasites on host population dynamics depends on local abundance of the parasites, which may vary considerably across spatial scales. In sand hopper populations, mermithid parasites have major impacts on host dynamics, which may vary among spatially separated populations due to the sand hopper’s wide, patchy distribution. The present study compared the abundance and biomass of a mermithid parasite (Thaumamermis zealandica Poinar et al., 2002) in sand hoppers (Bellorchestia quoyana (MilneEdwards)) both within and among disconnected beaches. In addition, several variables were measured and tested as potentially important predictors of the parasite abundance and biomass. It was found that geographic isolation may only be responsible for minor differences in parasite populations compared with other factors. Host size was identified as the most important predictor of mermithid parasite abundance, but epibiont abundance, kelp patch mass and host density were poor predictors of abundance. These factors were also poor predictors of parasite biomass in hosts. This study further supports the notion that studies aiming to elucidate population dynamics or patterns should sample thoroughly across both spatial and temporal scales. Keywords Parasite abundance . Parasite biomass . Spatial distribution . Bellorchestia quoyana . Body size . Epibionts
Introduction Parasites can regulate the population dynamics of their hosts in various ways, whether by causing mortality to host T. K. Rasmussen (*) : H. S. Randhawa Ecology Degree Programme, Department of Botany, University of Otago, PO Box 56, Dunedin 9054, New Zealand e-mail: [email protected]
individuals, reducing their fecundity or altering their behaviour (e.g. Anderson 1979; Ebert et al. 2000; Lafferty 2004). Such impacts of parasites on host populations depend largely on local abundance of parasites, which can vary considerably across spatial scales (Bates et al. 2010). Differences in parasite abundance among spatially separated host populations may reflect differences in susceptibility to parasites, a consequence of past co-evolutionary history with the parasite (Webster et al. 2004; Lohse et al. 2006). On the other hand, spatial differences in parasite abundance may also represent differences in infection risks faced by hosts, resulting from local abiotic and biotic factors (Thieltges et al. 2009). Whatever the case may be in a system, understanding causes of spatial variation for parasites is clearly critical to understanding host population dynamics. The talitrid amphipod Bellorchestia quoyana (MilneEdwards) (previously known as Talorchestia quoyana) provides a good system for looking at parasite spatial variation. B. quoyana is the most common sand hopper species living on New Zealand’s beaches (Morton and Miller 1973). It plays an essential role in consuming beach-stranded kelp (Inglis 1989; Dufour et al. 2012) and in nutrient cycling through the marine environment (Brown 2001). For these sand hoppers, spatial variation in parasite abundance acts on both small and large scales. On a small scale, they are distributed in patchy aggregations beneath available kelp, which allows for high-scale aggregations of their associated symbionts (Poulin and Rate 2001). On a large scale, sand hopper populations on different beaches can be separated by many kilometres, which could potentially allow marked differences in parasite abundance. The sand hopper B. quoyana is a known host of three symbiont species: a digamasellid mite (Dendrolaelaps sp.), a rhabditid nematode (unidentified) and a mermithid nematode Thaumamermis zealandica Poinar et al., 2002 (Poulin and Rate 2001). The former two species are epibionts that are not parasitic on the sand hopper host, whereas the mermithid
896
T. zealandica is an internal parasite (Poulin and Rate 2001). The infective stage of the parasite penetrates the body wall of the sand hopper and develops in its body cavity until maturity and then emerges into a moist sand environment as a freeliving adult, killing the host in the transition (Poinar et al. 2002; Currey and Poulin 2006). Not only does this parasite directly impact host populations, but also the mermithid may have further consequences on host dynamics via manipulation of the sand hopper’s burrowing behaviour. However, evidence for this behavioural effect has been mixed thus far (Poulin and Latham 2002; Currey and Poulin 2006, 2007). For instance, an experiment suggested that parasites may alter the burrowing depth of sand hoppers (Poulin and Latham 2002), but this result was not confirmed in field studies (Currey and Poulin 2006). Previous research has revealed that the mermithid parasite and epibionts of B. quoyana are each distributed heterogeneously within and among high-density host patches occurring across a beach (Poulin and Rate 2001). However, to our knowledge, no studies prior to the present one have examined whether the sand hopper mermithid parasites are distributed heterogeneously among separated beach populations. Moreover, little is known about what factors influence the variation of the mermithid within or among beaches. The present study aimed to address these knowledge gaps. In this study, we investigated the distribution of the mermithid T. zealandica among naturally infected sand hopper populations, both within and among separated beaches. We compared mermithid abundance and biomass between beaches and with several other variables of interest to determine which factors were important determinants of its variation. We measured host features (host length and sex), host density, epibiont abundance (mites and rhabditid nematodes) and the mass of the kelp patch covering each sample site. We hypothesised that mermithid abundance (the number of mermithids per sampled host) would be more different among populations on separated beaches than among those on a single beach. This would make sense if the populations on separate beaches were in isolation for enough time to cause differences in their genetic or population structure. In line with this, in our analyses, we predicted that beach identity would account for a large proportion of variation in abundance and biomass of the parasite.
Materials and methods Study sites and sampling procedure Sand hopper (B. quoyana) individuals were collected from six beaches during summer (December 2013–January 2014) around Dunedin, New Zealand. The sampled sites were the following: Brighton Beach (45.949° S, 170.333° E),
Parasitol Res (2015) 114:895–901
Tomahawk Beach (45.906° S, 170.546° E), Smails Beach (45.905° S, 170.564° E), Aramoana Spit (45.780° S, 170.709° E), Allans Beach (45.876° S, 170.698° E) and Long Beach (45.754° S, 170.649° E). Sand hoppers were collected from beneath three patches of kelp (three separate sites) per beach. To assess spatial variation within each beach, these three sites were selected randomly from across a stretch of a few hundred metres along the strand line or just above the high water mark during low tide. All selected kelp patches were in a similar stage of decay but varied in size (so they were collected and later weighed). There were six sample collection trips in total, one per beach, and all trips were done in sunny weather (strong winds and rain which may have affected their behaviour were avoided). Directly below each kelp patch site, an open metal cylinder (20 cm long with a diameter of 7.5 cm, a volume of approximately 884 cm3) was pushed into the sand to collect a volume of sand containing hoppers. The collected volume of sand was rinsed through a sieve (mesh size of 1.405 mm) with water to remove sand, leaving all the contained sand hoppers alive. Following this, the sand hoppers were transferred into closed buckets (one per kelp patch), transported to the laboratory and processed immediately. Only one cylinder sample was collected from each site on Brighton Beach, but for all the other beaches, two samples were collected from each site to increase the size of the sand hopper samples. This difference was accounted for in density estimates by standardising the density to the number of sand hopper individuals per cubic meter. Lab procedure A total of 946 sand hoppers was examined for symbionts. We found no mermithid-infected juveniles in our samples, and the number of infected juveniles was low in a previous study (Currey and Poulin 2006), suggesting that the importance of juveniles as hosts for mermithid parasites is negligible. For this reason, we restricted our analyses of mermithid abundance and biomass patterns to include only adult sand hoppers (n=583). Each sand hopper was measured (length from the front of the head to the tip of the telson in mm) and sexed on the basis of the male’s specialised gnathopods. All individuals shorter than 10 mm were classified as juveniles to avoid falsenegative classification of males. Following measurements, individuals were immediately decapitated and placed in a few drops of seawater. After allowing a few minutes, all symbionts (mermithids, rhabditid nematodes and mites) were counted under a dissection microscope. Mermithids were preserved in 70 % ethanol, and their wet weights were recorded (in mg to two decimal places). The total wet mermithid biomass per host was used as a response in addition to abundance in the analyses because biomass may show patterns not reflected in abundance; e.g. biomass may better reflect energy demands of the host (George-Nascimento et al. 2002).
Parasitol Res (2015) 114:895–901
Moreover, parasite biomass may be a more accurate measure of infection intensity than the number of parasites since it gives increased weighting to larger parasite individuals that are likely to have greater impacts on the host (by using more space and host resources). Analysis of results Mermithid abundance and biomass were compared between male and female sand hoppers using standard t tests (assuming unequal variance). Two main statistical models were used to analyse the results in this study, both utilised in the R environment (R Development Core Team 2008). The first one was a generalised linear mixed model used to analyse parasite abundance (the response variable) against eight predictors, including three with random effects (beach identity, kelp site (nested within beach) and host sex) and five with fixed effects (host length, host density at site (individuals/m3), mite abundance, rhabditid nematode abundance and kelp patch mass (kg)). This model was assigned a negative binomial distribution and accounted for zero inflation using the “glmmADMB” package (Bolker et al. 2012). The corrected Akaike information criterion (AICc) and Akaike model-averaged weights were calculated using the “MuMIn” package (Barton 2013) to compare component models with all combinations of the above predictors. These techniques provide a useful measure of relative variable importance (Anderson 2008). The second model used was a linear model with residuals of the regression between the hosts’ combined mermithid mass and host length as the response variable (with Gaussian distribution). The residuals between biomass and length were used instead of just biomass to account for collinearity between the variables (residuals were not used in the first model since abundance and length were not collinear). Only observations from the infected individuals were used in this model, but the predictors (excluding length) were the same as those used in the first model. As with the first model, all possible component models were compared by AICc. Both models (mermithid abundance and biomass) were re-run, but instead of having mite and nematode abundance as predictors, we used residuals of regression between their respective abundance and host length as predictors. The re-run first model also had mermithid abundance (the response) replaced with residuals of the regression between abundance and host length. The aim of re-running the models in this way was to observe whether variable effects changed when accounting for collinearity between symbiont abundance and host length.
Results Across the beaches sampled for this study, the mermithid parasite T. zealandica was distributed with considerable
897
heterogeneity both among separated beaches and patches on individual beaches. Prevalence of the mermithid parasite was 6.16 % in the entire sample (n=583) but ranged from 0 % up to 14.58 % per beach. There were clear differences among beaches in both abundance and biomass of the parasite; no parasites were found in the samples from two of the six beaches, and of the four beaches where mermithids were detected in the samples, parasite abundance and biomass were considerably higher at certain beaches. Notably, parasite abundance was twice as high at Aramoana Beach compared with Allans Beach and Tomahawk Beach (Table 1). Also, average biomass of the parasite was substantially low at Tomahawk Beach compared with the others (Table 1). However, it is worth noting that parasite abundance and biomass were also highly variable across each individual beach, as shown by large standard deviations (Table 1). Like the mermithid parasite, the sand hopper epibionts displayed great heterogeneity. Mites were particularly abundant at Long Beach with nearly six mites on average per sand hopper, compared with the majority of other beaches sampled which, on average, had one or fewer mites per individual. Rhabditid nematodes were also remarkably more abundant at certain beaches; abundance ranged from around three nematodes per individual at Allans Beach up to around 21 per individual at Tomahawk Beach. However, like the mermithid parasite, both epibionts showed highly variable abundance among individuals of the same beach (Table 1). In terms of host sex, mermithid abundance and biomass displayed different patterns. Across the entire data set, the parasite was significantly more abundant in female sand hoppers (0.16) compared to males (0.05) (p=0.03, t=2.23, degrees of freedom (df)=529.8). This pattern appeared consistent for each individual beach, but the effect of sex was also subject to high variability (Fig. 1a). In contrast, average mermithid biomass was actually higher for males (0.93 mg) than females (0.62 mg), though not significant (p=0.37, t= 0.95, df=6.7), and the effect of sex was again subject to high variability (Fig. 1b). It should also be noted that due to low infection rate in the sand hoppers, mermithid biomass comparisons were subject to low sample size. The best AIC models for mermithid abundance were very different from those for the parasite biomass (Table 2). The top AIC model (AICc =367.73) for parasite abundance included only the host length. However, the second best AIC model, which also included rhabditid nematode abundance, and the third best AIC model, which also included kelp patch mass, were very close to the best model (ΔAICc
