Immunity changes through ontogeny and can mediate facilitative and inhibitory interactions among co-infecting parasite species. In amphibians, most immune memory is not carried through metamorphosis, leading to variation in the complexity of immune responses across life stages. To test if the ontogeny of host immunity might drive interactions among co-infecting parasites, we simultaneously exposed Cuban treefrogs (Osteopilus septentrionalis) to a fungus (Bactrachochytrium dendrobaditis) and a nematode (Aplectana hamatospicula) at tadpole, metamorphic, and post-metamorphic life stages. We measured metrics of host immunity, host health, and parasite abundance. We predicted facilitative interactions between co-infecting parasites as the different immune responses hosts mount to combat these infectious are energetically challenging to mount simultaneously. We found ontogenetic differences in IgY levels and cellular immunity but no evidence that metamorphic frogs were more immunosuppressed than tadpoles. There was also little evidence that these parasites facilitated one another and no evidence that A. hamatospicula infection altered host immunity or health. However, Bd, which is known to be immunosuppressive, decreased immunity in metamorphic frogs. This made metamorphic frogs both less resistant and less tolerant of Bd infection than the other life stages. These findings indicate that changes in immunity altered host responses to parasite exposures throughout ontogeny.

Animal Husbandry

Cuban treefrog (Osteopilus septentrionalis) tadpoles were collected from 140-L pools filled with rainwater at the University of South Florida botanical gardens (Tampa, FL, USA) in the summer of 2017. Individuals collected early in the season (May), midseason (July), and later in the summer (August) were all reared to post-metamorphic, metamorphic, and tadpole life stages respectively. See supplemental methods for additional details.

Experimental Design

To examine how the life stage of amphibians affect host-parasite dynamics, we exposed O. septenrionalis to one of four treatments at each of the three amphibian life-stages. Treatments included no parasite exposure, and exposures to either Bd alone, A. hamatospicula alone, or Bd and A. hamatospicula together. These four treatments were applied at the tadpole (Gosner development stage 28-35; Gosner 1960), metamorphic (stage 42-45 frogs), and post-metamorphic stages. In total, there were 5,8, and 9 control individuals for tadpole, metamorphic, and post-metamorphic frogs, respectively, and 18 individuals for each of 9 treatment groups (3 life stages and 3 treatment groups), for a total of 184 individuals.

Bd (SRS-JEL 212 strain) was cultured in a 1% tryptone solution and A. hamatopsicula larvae were harvested from adult worms and developed in Petri dishes (see supplement for more details). Exposures were applied to post- and mid-metamorphic frogs in Petri dishes (25 x 100 mm) with 1 mL and 10 mL of ASW water, respectively, for 24 h. Tadpoles were exposed in cups containing 30 mL of ASW water. Hosts were exposed to 1 mL of DI water containing either 30 J3 A. hamatopsicula larvae or 105 zoospores of Bd. All frogs received sham exposures for parasites to which they were not exposed.

Assessing amphibian responses to parasites

To assess Bd abundance, hosts were swabbed eight days after treatment exposures because it is late enough for parasite establishment, but early enough so few hosts have experienced parasite-induced mortality (McMahon et al. 2014). Frogs were either swabbed 5 times from hip to toe on rear legs (if frogs had rear legs) or around the mouth for tadpoles. DNA was extracted from swabs (Qiagen DNEasy Blood & Tissue Kit) and analyzed using quantitative polymerase chain reaction (qPCR) to quantify Bd abundance (Boyle et al. 2004). A. hamatospicula abundance were assessed by counting adult worms in the gastrointestinal tract of the host after either mortality or euthanasia (for individuals surviving the entire experiment).

Subsets of experimental animals were euthanized on day 11, 19 or 28 to test host immunity during early, but detectable, Bd infections and the peak immune response for A. hamatospicula and Bd, respectively. Blood was drawn using micro-hematocrit capillary tubes (Fisher Scientific, Pittsburgh, PA) and used to create blood smears. Blood smears were stained using Camco Quik Stain II (Cambridge Diagnostic Products, Inc, Fort Lauderdale, FL). Three separate researchers counted 2000 cells on each slide and identified white blood cell types and the results were averaged (Heatley and Johnson 2009). Enzyme-linked immunosorbent assays (ELISA) from remaining plasma were used to detect IgY antibody presence (adapted from Knutie et al. 2017b; see supplemental methods).

Individuals were measured (snout-vent length [SVL]) and weighed weekly for four weeks. Amphibians were also checked twice daily for mortality. If mortality occurred, frogs were weighed, measured, swabbed and dissected as described above. Twenty-eight days after treatment exposures, all the remaining frogs were euthanized. Growth rates were calculated as the final mass or SVL minus the initial mass or SVL divided by weeks alive (g or mm/w).

Statistical Analyses

All analyses were run with R version 3.6.1 (R Core Team 2019) and figures were created using the visreg function and package (Breheny and Burchett 2019) or the ggplot function and ggplot2 package (Wickham et al. 2020). For all models host life stage, exposure to Bd, and exposure to A. hamatospicula were included as interacting independent variables unless stated otherwise. Tukey’s post-hoc tests to compare across life stages were run using the multcomp package and using the glht function (Hothorn 2010).

To investigate how host immune response and parasite abundance are affected by host life stage and co-infection, we ran models with all types of white blood cells counted (i.e., lymphocytes, neutrophils, basophils, monocytes, thrombocytes, and eosinophils), total white blood cell count, neutrophil to lymphocyte ratio, IgY antibody level, and Bd or A. hamatospicula abundance as separate dependent variables. Neutrophil, monocyte, eosinophil, and thrombocyte counts and Bd abundance were analyzed with negative binomial error distributions. The proportion of J3 larvae that hosts were exposed to that successfully penetrated the hosts’ skin and the proportion of J3 larvae that penetrated the hosts’ skin that successfully established as an adult worm in the amphibian guts were run as separate analyses with binomial distributions error. Weight was also used as a predictor variable in these models to control for differences in host size.  All other immune responses were analyzed as generalized linear models with normal error distributions. Life stage was based on the Gosner stage of the amphibian when immune metrics or parasite abundance were quantified.

We also investigated how co-infections and host life stage alter host health. We conducted a survival analysis using the survival package and the coxph function (Therneau and Lumley 2019). Individuals that survived to the end of the experiment were right-censored. Total development and growth rate (g or mm) were divided by four weeks or time to metamorphosis for development (e.g., g/w) and used as dependent variables in separate models. To test how host life stage and identity of the co-infecting parasite affected host tolerance, all the above-described models with host health and immunity were rerun with the addition of parasite abundance (Bd and A. hamatospicula abundance were run separately) as a fourth, interacting independent variable. Thus, tolerance is being measured as the slope of survival and growth against parasite abundance.