Laboratory assays show that parasites often have lower heat tolerance than their hosts. But how physiological tolerances and behavioral responses of hosts and parasites combine to affect their ecological interactions in heterogeneous field environments is largely unknown. We addressed this challenge using the model insect system of the braconid wasp parasitoid, Cotesia congregata, and its caterpillar host, Manduca sexta. We used experimental manipulations of microclimate in the field to determine how elevated daytime temperatures altered the behavior, performance, and survival of the host and parasite. Our experimental manipulation increased daily maximum temperatures on host plants, but had negligible effects on overall mean temperature. These increased maximum temperatures resulted in subtle, biologically relevant, changes in physiology and behavior of the host and parasitoid. We found that parasitism by the wasp did not significantly alter caterpillar thermoregulatory behavior, while experimentally increased daily maximum temperatures resulted in both parasitized and unparasitized caterpillars being found more frequently in cooler microhabitats. Overall, we did not observe the complete parasitoid mortality seen at extreme temperatures in laboratory studies, but gained insight into sublethal effects of increased daily maximum temperatures on host and parasitoid behavior and physiology. Climate change will alter both the biotic and abiotic environments that organisms face, and we show here that empirical experiments in the field are important for understanding organismal response to these new environments. 

Animals

Manduca sexta were collected from several field locations prior to the start of the experiment (see below). Cotesia congregata were sourced from the UNC colony, which was established from a colony at Virginia Commonwealth University (Richmond, VA) in 2017, originally sourced from wasps collected from Nottoway Co., NC (37.0817 °N, 77.9755°W) (Moore et al., 2021). Parasitoid wasps were maintained using M. sexta from the UNC colony as hosts. Adult wasps, cocoons, and parasitized caterpillars were kept at room temperature (~24°C - 26°C) under a 14L/10D light cycle.

 Collection and rearing

Manduca sexta used in this experiment were collected as eggs from two field locations (Upper Coastal Agricultural Research Station, Rocky Mount, NC: 35.8937 °N, -77.6810 °W and Mason Farm Biological Reserve, Chapel Hill, NC: 35.8890 °N, -79.0156 °W) in July 2019. Eggs were placed in small petri dishes at low densities (< 20 eggs) on an artificial diet containing 1.79% tobacco powder and were kept at 25°C in an environmental chamber (Percival 36VL). Upon hatching, 1st instar M. sexta larvae were transferred to fresh-cut tobacco leaves in water picks (20-30 hatchlings per leaf, 1-2 leaves per bin) and placed in plastic bins.  Bins were closed with either bridal veil clipped to edges, or lids with holes covered with metal mesh. First and 2nd instar larvae were kept at 25°C±10°C and monitored daily for leaf quality (replaced as needed, ad libitum) and for newly molted 3rd instars.

On the day of molt into the 3rd instar (day 0), M. sexta caterpillars were assigned randomly to a parasitization treatment (NP = unparasitized; P = parasitized) and to an experimental field plant. Each caterpillar was weighed and parasitized (if in P treatment). Each experimental plant was assigned two caterpillars, and each member of the pair was marked with different color fingernail polish on the last segment of the abdomen to differentiate individuals (Diamond and Kingsolver, 2010). After marking, pairs of 3rd instars were placed in large petri dishes with sections of tobacco leaf, and returned to the 25°C±10°C rearing condition. Manduca sexta caterpillars were then placed on their assigned plant in the field plot on day 1 of the 3rd instar (1 day after molt).

Field plots

Our field site consisted of two plowed plots, each 12m X 10m with 6 rows of 10 tobacco plants (60 plants per plot). Rows were spaced 2m apart, and tobacco plants were planted 1m apart to prevent possible migration of caterpillars between plants. Weed barrier fabric was installed to prevent weed growth, and different colors of weed barrier were used to manipulate the temperature of the field plots. Half of each plot was lined with grey, non-reflective weed barrier, and the other half with black, non-reflective weed barrier. During sunny conditions the temperature of the black weed barrier was greater than that of the grey barrier, as confirmed through IR imaging. 

Tobacco plants were grown from seeds in a climate-controlled green house (26°C) for 7-8 weeks before being transported in pots to Mason Farm Biological Reserve (Chapel Hill, NC). Tobacco plants were covered with bridal veil netting to exclude insect herbivores and were placed under shade to prevent sun bleaching for a 2-3 day acclimation period. Plants were then placed in direct sunlight until planting in experimental plots. While in pots, plants were watered daily by hand. Experimental Plot 1 was planted from 6/29/19 to 7/3/19, and Plot 2 was planted from 7/12/19 to 7/17/19. Once plants were planted in the soil, they were watered for an hour twice daily via an automatic sprinkler system. Tobacco plants were covered with bridal veil to prevent feeding by other insect herbivores, and loss of experimental caterpillars to predation. Insect herbivores that were too small for exclusion by the mesh were removed daily. 

Operative temperature in the field was recorded in Plot 1 only, using 20 thermoconductive, 3D printed caterpillar models with thermocouples connected to a datalogger (Campbell 21X) inserted into the models lengthwise (Kingsolver et al., 2001). Models were the size of late 5th instar M. sexta (6.35 cm length x 1.25 cm diameter), were painted leaf green to approximate the solar absorptivity of real caterpillars (Kingsolver, 2000), and placed on the underside of tobacco leaves along the midvein. Models were placed at three height locations on the plants (high, n = 6; middle, n = 11; low, n = 3) to account for spatial variation in temperature. Half of the models were placed on plants in the grey weed barrier treatment, and half on plants in the black weed barrier treatment. Temperature was recorded from the models by the datalogger every 15s and averaged over 10-min intervals. Caterpillar model accuracy was validated during the summer of 2020 by comparing the recorded internal temperatures with infrared surface images of real 5th-instar caterpillars (FLIR T400) during daytime conditions. Surface caterpillar temperature and internal model temperature were closely correlated across a range of environmental conditions. 

Caterpillar observation in the field

Manduca sexta were transported to the experimental plots on day 1 of the 3rd instar in coolers to regulate temperature. Each pair of caterpillars (either NP-P, or P-P) were placed on their assigned plant, on the upper leaf surface of two leaves in the middle of the plant, and in the shade if possible. It was not possible to place them on the underside of leaves (the preferred resting location of M. sexta caterpillars), as they did not adhere to the leaf surface until several minutes after handling. The majority of caterpillars had migrated to the underside of the leaf by the time of their first census (24 hours after placement). The date and time of placement in the field were recorded for each individual.

Caterpillar location was recorded daily during the experiment to measure thermoregulatory behavior through microhabitat selection. Daily censuses were taken of experimental plots from 10 am to 4 pm, and the time of data collection was alternated between morning and afternoon daily. During the censuses, each experimental plant was examined for caterpillar presence, location on the plant, and developmental stage. Caterpillar location was recorded with 3 metrics during daily census: height on the plant (high, middle, low), leaf surface (upper, under, edge), and sunlight (sun, shade). The categorization of sun versus shade was determined by if 50% or more of the caterpillar was in the shade, and vice versa. The date and time of the daily census were recorded for each individual caterpillar. If an individual caterpillar was not found for two consecutive days, it was assumed dead, and the date of death was recorded as the 3rd day of absence. Unparasitized caterpillars were removed from the field and returned to the laboratory after censusing on the 4th day after the molt to 5th instar to prevent caterpillar loss due to wandering (prepupal stage). Parasitized caterpillars were returned to the lab on the 5th-7th day of the 5th instar, to prevent loss of parasitoid data. Date of removal from field was recorded for each individual.

Data collection in lab

Change in mass (Δmass) and development time (Δage) for surviving M. sexta caterpillars from molt into the 3rd instar (the beginning of the experiments) to wandering or wasp emergence were calculated as metrics of caterpillar performance. After caterpillars were returned to the lab, they were placed in large, individual petri dishes, provided tobacco leaves from the plant from which they were collected, and placed in climate control chambers at 25°C±10°C. Caterpillars were monitored daily for wandering (NP) or wasp emergence (P). Once caterpillars exhibited wandering behavior, they were weighed and culled. At wasp emergence, caterpillars and all wasp larvae were transferred to a clean petri dish and left undisturbed for 48 hours to allow for cocoon formation and hardening. Cocoons were then removed from the caterpillar and counted, as were the number of parasitoids that failed to spin cocoons. Parasitized hosts were weighed and frozen for future dissection. Parasitoid cocoons were placed into closed condiment cups and returned to the climate control chamber, and were monitored daily for adult wasp eclosion. The date of wasp eclosion was recorded, and eclosion cups were left undisturbed for 48 hours to allow for all surviving adults to eclose. Cups were then frozen for future counting, sexing, and weighing. Adult mass was measured separately for male and female wasps, as females are generally larger (pers. obs.). All male and female wasps for each host were weighed en masse to the nearest 0.01 mg. Mean individual adult mass was obtained by dividing the total weight by the number of male or female wasps for that host. Parasitized hosts were dissected and the number of wasp larvae remaining in the hemocoel was used to calculate the proportion of parasitoid survival to emergence and eclosion (Moore et al., 2021). Development time and survival to emergence and eclosion, as well as adult mass, were calculated as metrics of C. congregata performance during the experiments. 

Statistical analyses:

Temperature

The effect of weed barrier treatment on model caterpillar temperature was examined by conducting a linear mixed effects model, using the residual temperature of the caterpillar models as the response variable. The residual temperatures of model caterpillars indicate the deviation from the expected temperature for each model. The residual temperatures were calculated by taking the mean temperature across all caterpillar models for each 10-minute time period, and then subtracting the mean model temperature from the recorded model temperature at each time point. The caterpillar model residuals were analyzed using the lme function from the ‘nlme’ package in R (v 4.0.2), with weed barrier treatment, location on plant, and their interaction term as fixed effects, and a random intercept of plant ID. 

Caterpillar behavior

As experimental plots differed in both qualitative (plant quality and structure) and quantitative (mean daily temperature, caterpillar, and parasitoid performance) ways, the data collected from each plot was analyzed separately for all analyses. The location of M. sexta on the host plants at each daily census was analyzed in two ways: 1. generalized linear models with binomial distributions of separate location metrics using the glm function in the ‘stats’ package in R (v 4.0.2), and 2. linear mixed effects models of the daily probability of movement for each location metric using the lme function from the ‘nlme’ package. The height and leaf surface metrics were compressed from a multinomial response to a binomial response by combining factors with the lowest occurrence (height: middle and low, leaf surface: upper and edge). The binomial metrics of location (height, leaf surface and shade) were used as the response variables in separate models, with weed barrier treatment, parasitization treatment, caterpillar instar and their interaction terms included as fixed effects. Experimental plots were analyzed separately. Parasitism by C. congregata is known to decrease the locomotion of M. sexta caterpillars prior to parasitoid emergence (Adamo et al., 2016), but whether the parasitoid affects caterpillar movement throughout development is unknown. To test this, we calculated the proportion of location state changes across the experiment, or the daily probability of movement, for each caterpillar. The daily probability of caterpillar movement was calculated separately for each location metric by determining the total number of observations, calculating the number of state changes (ex: high to middle, upper leaf surface to lower), then dividing the number of state changes by the total number of observations. The daily probability of state change was used as the response variable in separate linear mixed effects models for each location metric, with weed barrier treatment, parasitization, experimental plot, and their interaction terms included as fixed effects. All models had a random intercept of plant ID. 

Caterpillar performance and survival

The survival of M. sexta caterpillars in our experimental field plots was analyzed with a generalized linear model with a binomial distribution, with the ‘glm’ function in the ‘stats’ package in R (v. 4.0.2). Death before removal from the field was used as the response variable, and weed barrier treatment and parasitization treatment as interactive fixed effects. Experimental plots were analyzed separately. Manduca sexta mass and development time were analyzed using linear mixed effects models with the ‘lme’ function from the ‘nlme’ package in R (v. 4.0.2). The change in mass (Δmass) and age (Δage) through the experiment were calculated by final mass (at wandering (NP) or wasp emergence (P)) minus mass at 3rd instar, and final age minus age at 3rd instar.  Δmass and Δage were used as numerical response variables in two separate linear mixed effects models, and each model included weed barrier treatment and parasitization treatment as interactive fixed effects, and plant ID as random intercepts Experimental plots were analyzed separately.

The subset of parasitized hosts was analyzed separately to determine the effects of parasitoid load (number of C. congregata larvae that survived to at least the 2nd instar) on host growth and developmental time. Parasitoid load differed significantly between experimental plots (see below), so the Δmass and Δage of parasitized hosts in each plot were analyzed separately. Δmass and Δage were analyzed in separate linear mixed effects models, with weed barrier treatment and parasitoid load as interactive fixed effects, and plant ID included as a random intercept. 

Parasitoid performance and survival

Parasitoid load was analyzed with a linear model, with weed barrier treatment and experimental plot as interactive fixed effects. Load differed significantly between experimental plots, so all analyses of wasp life history traits were analyzed separately for each plot (Fig S4). Cotesia congregata survival to emergence and adult eclosion was analyzed with generalized linear mixed effects models with binomial distributions, using the ‘glmer’ function in R (v. 4.0.2). The number emerged (or eclosed) versus the number unemerged (or died before eclosion) were used as the response variable, and weed barrier and parasitoid load were included as interactive fixed effects. A random intercept of host ID was included in all models. Adult mass of C. congregata was analyzed using linear models, with the mean individual mass of C. congregata adults (mass of all male or female wasps per host / number of male or female wasps per host) used as the response variable. Weed barrier treatment, sex, and parasitoid load were included as interactive fixed effects, and host ID was included as a random intercept. 

This data set consists of behavioral and life history data taken from live caterpillars on tobacco plants, as well as temperature data taken from thermocouples inserted into thermoconductive caterpillar models, attached to a data logger. Both data sets have been processed to remove/correct erroneous data, and undergone necessary manipulations for combining these data and for statistical analyses and plotting. The R code used to process and manipulate data is included, and is notated, as well as the code for all statistical analyses and figures. An excel file is provided, with readmes for all data sets. Data sets do contain NAs, some of which are functional (caterpillar was removed from the field, therefore, there is no more census data), and some of which indicate missing data. The methods used to deal with both of these are described in the R code.