Long‐term variation in environmental conditions influences host–parasite fitness

Abstract Long‐term data on host and parasite fitness are important for predicting how host–parasite interactions will be altered in an era of global change. Here, we use data collected from 1997 to 2013 to explore effects of changing environmental conditions on bird–blowfly interactions in northern New Mexico. The objectives of this study were to examine what climate variables influence blowfly prevalence and intensity and to determine whether blowflies and climate variables affect bird fledging success. We examined how temperature, precipitation, and drought affect two parasitic blowflies and their hosts, Western Bluebirds (Sialia mexicana) and Ash‐throated Flycatchers (Myiarchus cinerascens). We found that blowfly prevalence did not change over time. Blowfly intensity increased over time in bluebird nests, but not in flycatcher nests. More blowflies result in slightly higher fledging success in bluebirds, but not flycatchers. There was a significant interaction between blowflies and precipitation on bluebird fledging success. For flycatchers, there was a significant interaction between blowflies and temperature and between blowflies and drought severity on fledging success. Given that the southwest is projected to be hotter and have more frequent and prolonged droughts, we predict that flycatchers may be negatively impacted by blowflies if these trends continue. Future work should focus on investigating the role of both blowflies and climate on fledging success. Climate patterns may negatively impact host fitness through altered parasite pressure.

Additionally, warming arctic temperatures are increasing infection pressure of nematodes in muskoxen, most likely due to faster development rates (Kutz, Hoberg, Polley, & Jenkins, 2005). High temperatures coupled with other climate conditions may act against each other. Romo and Tylianakis (2013) found rates of aphid parasitism increased as temperature increased. During high temperatures and drought, parasitism decreased and aphid population growth increased (Romo & Tylianakis, 2013). Ectoparasites and parasites with free-living life stages may be disproportionately affected by climate.
These parasites are more exposed to environmental conditions than parasites that are always in a host (Brooks & Hoberg, 2007;Martinez & Merino, 2011).
Birds are often subject to parasitism by a variety of ectoparasites, including hematophagous larvae. The larvae of flies of the genus Protocalliphora and closely related Trypocalliphora (Diptera: Calliphoridae) are common parasites in bird nests and feed on the blood of nestlings (Eeva, Lehikoinen, & Nurmi, 1994;Heeb, Kolliker, & Richner, 2000;Hurtrez-Boussès, Perret, Renaud, & Blondel, 1997;Marshall, 1981;Rothschild & Clay, 1952;Sabrosky, Bennett, & Whitworth, 1989). These parasites are often included under the same common name, avian blowflies. Female blowflies tend to oviposit from April through July in occupied nests and larvae likely hatch within 72 hr. Larvae feed on the nestlings and pupate within approximately 9-14 days and adults emerge around 14-21 days (Bennett & Whitworth, 1991). The difference between these genera is that Trypocalliphora is subcutaneous for the entire larval stage, while larvae of Protocallihpora feed intermittently on nestling blood (Sabrosky et al., 1989). As Trypocalliphora near maturity, they emerge from the skin and drop into the nest to pupate. Both emerge as adults to lay eggs in new nests (Sabrosky et al., 1989).
Protocalliphora spp. overwinter as adults; pupae cannot tolerate low winter temperatures (Bennett & Whitworth, 1991). There is evidence that activity of adult flies is somewhat temperature dependent. Bennett and Whitworth (1991) noted that adult blowflies were inactive at 7-10°C and rarely became active until a threshold ambient temperature of approximately 15.5°C was met. This temperature threshold differs between species. Furthermore, there is evidence that larval densities in occupied nests are somewhat temperature dependent. After experimentally manipulating temperature inside nests of tree swallows, Dawson et al. (2005) found that larval densities followed a curvilinear trend with temperature.
Larval densities were low at both higher and lower temperatures but peaked at around 25°C. Pavel, Chutný, Petrusková, and Petrusek (2008) also found Trypocalliphora spp. only infested Meadow Pipit and Bluethroat nestlings during warmer summers with temperatures frequently exceeding 18°C.
While there is still some discussion as to the breadth of effects Protocalliphora spp. have on nestlings, it seems that nestlings tolerate blowfly parasitism (DeSimone et al., 2018;Roby, Brink, & Wittmann, 1992;Simon et al., 2004). The mechanisms for tolerating parasitism by blowflies are not widely understood. There is some evidence that resistance is also an important defense strategy against blowflies (DeSimone et al., 2018). When female tree swallows had higher antibody responses, nests had fewer blowflies, and the nestlings had slightly higher antibody levels (DeSimone et al., 2018). Whether environmental conditions such as temperature, drought conditions, or precipitation have an influence on nestlings' ability to tolerate or resist parasites has not been greatly explored. There is evidence that weather may play a role in a host's fitness costs of parasitism (De Lope, González, Pérez, & Møller, 1993;Merino & Potti, 1996;Møller et al., 2014) and that climate may affect a host's defenses against parasites (Møller et al., 2014). Further studies are needed to gain a better understanding of the variable and complex effects changing environmental conditions have on host-parasite interactions.
The effects of climatic changes on the complexity of host-parasite interactions are challenging to forecast due to the lack of data over a long time period. We use a 17-year dataset of avian blowflies and cavity-nesting birds in northern New Mexico, an area experiencing rapid environmental change. While the southwestern United States commonly experiences periods of drought, there is consensus that rising temperatures will result in more intense, frequent, and prolonged droughts (Garfin, Jardine, Merideth, Black, & LeRoy, 2013). Average daily temperatures for 2001-2010 were the highest in the southwest from 1901 to 2010, and average temperatures since 1950 have been warmer than any comparable length of time in at least 600 years (Garfin et al., 2013). Dramatic changes to southwestern ecosystems over the last few decades and climatic variation among years offers a study system to explore environmental conditions and host-parasite dynamics.
Most studies examine the effects of blowflies over one or two seasons, which makes accessing the effects of environmental conditions difficult. We examine survival to fledging over a much longer timescale by using data collected from 698 nests over a 17-year period. We investigate how climate change affects the interactions between Western Bluebirds (Sialia mexicana) and Protocalliphora occidentalis as well as Ash-throated Flycatchers (Myiarchus cinerascens) and Trypocalliphora braueri. Our main goal was to determine how the environment influences the interactions between blowflies and fledging success. The three specific objectives of this study were to (a) determine whether blowfly prevalence and intensity have changed over time from 1997 to 2013; (b) identify the environmental variables that influence changes in blowfly prevalence and intensity; and (c) identify the effects of both environmental variables and blowfly parasitism on fledging success of these species. Based on evidence that blowfly prevalence is somewhat temperature dependent, we hypothesize that blowfly prevalence and intensity have increased over time since there has been a general increase in temperature (Dawson et al., 2005;Pavel et al., 2008). Our second hypothesis is that fledging success of both species will be negatively affected by the combination of blowflies, higher temperatures, and drought (De Lope et al., 1993;Merino & Potti, 1996;Møller et al., 2014).

| Study area and field design
This study was conducted on Los Alamos National Laboratory (LANL) property in Los Alamos County in north central New Mexico.
The Los Alamos National Laboratory is a multidisciplinary research institution but was established in 1943 as part of the Manhattan project to design atomic weapons. LANL has an ongoing assessment of potential site-related contamination and ecological risks.
All observed concentrations of constituents in areas at LANL that wildlife have access to are below the lowest observable adverse effect levels (LANL, 2018). Furthermore, reports from 2018 indicate that chemical concentrations detected in eggs of Western Bluebirds and Ash-Throated Flycatchers were below levels associated with adverse effects . The results of these studies indicate that it is unlikely historical contamination at the study site would influence the results of the present study.
The 111 km 2 laboratory is situated on the Pajarito Plateau and consists of a series of relatively narrow mesas separated by deep, steepsided canyons that decrease in elevation from the Jemez Mountains down to the Rio Grande (2,400-1,650 m). Six major vegetation types are found in Los Alamos County along the west-to-east elevational gradient: subalpine grassland, spruce-fir forest, mixed conifer forest, ponderosa pine forest, piñon-juniper woodland, and juniper grasslands (Foxx, Pierce, Tierney, & Hansen, 1998). The nest boxes are located primarily in pinon-juniper woodlands and ponderosa pine forests.

| Nestling monitoring
Nest boxes were only monitored during the breeding season. Nest boxes were not visited during nonbreeding seasons. We visited nest boxes every two weeks beginning in May 1997 until August 2013. Nests with eggs were considered active and visited once a week. Nestling age and hatch date were based on the physical development of the nestlings (mass, tarsus length, and culmen) (Sanchez et al. in preparation). Each nestling was handled for less than five minutes in accordance with the Guidelines for the Use of Wild Birds in Research (Fair, Paul, & Jones, 2010). The Institutional Animal Care and Use Committees of both LANL and the University of Missouri-St. Louis approved all protocols. Nests in which a nestling was missing before the expected fledge date (before 15 days) were presumed to be predated. Dead nestlings or unhatched eggs in the nest were removed from the nest and collected for laboratory analysis.

| Parasite quantification
Nest boxes were cleaned of rodent nests and other material prior to spring nesting. We collected nest material from all nest boxes within one week of fledging and the nests were stored in plastic bags at room temperature. We removed adult flies the day they emerged from pupae, which was usually 2-5 days from collection. Nests were then stored in a freezer until sifted for pupae that did not emerge. Nests were searched thoroughly for pupae of P. occidentalis and T. braueri.
Emergent adult insects were stored in methyl alcohol until identification. Other arthropod parts, seed food items, or unusual objects were removed and stored. All fly pupae were identified as P. occidentalis (in bluebird nests only) or Trypocallihpora braueri (in flycatcher nests only) by T. Whitworth in the early years of the study. Prevalence was defined as the number of nests that contained larvae, and parasite intensity was determined to be the number of larvae in only infested nests.

| Climate data
All the environmental variables chosen were selected a priori to be biologically relevant factors for both blowflies and bird fledging success. These data were collected from the same centrally located weather station on the Pajarito Plateau (35.861°W 106.31°N, 2,263m.) (LANL, 2014). Temperature and precipitation data were collected from 1997 to 2013 for the summer breeding season (April-August), from the weather station. Adult female flies tend to oviposit in nests containing very young nestlings (Sabrosky et al., 1989), and the nestling period lasts for 18-25 days for bluebirds and 13-17 days for flycatchers. Therefore, we chose mean monthly temperature, total monthly precipitation, and monthly Palmer Drought Severity Index (PDSI) values that corresponded with the month of the hatch date of the nest. PDSI was obtained using a Drought Atlas weather station location at 35.86°W and 106.321°N in Los Alamos, New Mexico (NDMC, 2018). PDSI is a meteorological drought index F I G U R E 2 Adult Western Bluebird (Sialia mexicana) on the left and an adult Ash-throated Flycatcher (Myiarchus cinerascens) on the right based on calculations of precipitation, temperature, and local available water content of the soil.
Because we were interested in determining how environmental variation influences blowfly prevalence and intensity in nests, we also used mean winter temperature and mean precipitation of the winter season preceding the breeding season (November-February).
This was done to determine whether winter conditions influence blowfly prevalence and intensity during the breeding season since blowflies overwinter as adults (Sabrosky et al., 1989). We took the mean winter precipitation and mean winter temperature over the entire winter, from November through February because we wanted a snapshot of the environmental conditions during the preceding winter months.
All LMMs and GLMMs were done at the nest level, and the two species were analyzed separately. The number of blowflies in each nest was divided by the number of nestlings present in the nest to get a measure of parasites per nestling. These data were used when determining the best predictors of blowflies (i.e., a response variable) and when blowflies were predictor variables in models predicting fledging success. Blowfly intensity is the number of blowflies per nestling in only nests infested with blowflies. Only nests with complete data were used. The number of nests with complete data for bluebirds was 494, while there were 112 nests for flycatchers.

| Changes in blowflies over time
To determine whether blowflies have increased or decreased over time, we used linear models and generalized linear models with year as the only predictor variable. Two separate tests were done to determine whether blowfly prevalence or blowfly intensity changed over time. These different tests allow us to understand more of the biology behind potential changes in blowfly populations over time.
Generalized linear models with binomial distributions were used to determine whether blowfly prevalence changed over time. The second test was done on only infested nests, and the response variable was the number of blowflies per nestling. To make the model residuals more normal, blowflies per nestling was log transformed.
We also split up PDSI by drought category for each nest because we were specifically interested in how drought affects blowfly prevalence and intensity, even if PDSI was not a top predictor of blowflies. This is important because there are different severities of drought, and we wanted to do coarse comparisons among moist years, normal years, and three levels of drought. These categories were based on the following PDSI cutoffs: <−4 (extreme drought); −4 to −3 (severe drought), −3 to −2 (moderate drought), −2 to 2 (normal), 2-3 (moist); >3 (very moist). We plotted the prevalence in each category and mean blowfly intensity (±SE) in each category. To test for differences in prevalence, we used a generalized linear model with a binomial distribution. To test for differences in blowfly intensity among the drought categories, we ran a nonparametric Kruskal-Wallis test because an ANOVA produced residuals that were highly nonnormal. Transforming blowfly intensity did not improve normality of model residuals. Pairwise comparisons were analyzed using pairwise Mann-Whitney U tests with Bonferroni corrected p-values, instead of comparing p-values to a corrected significance level.

| Model selection for blowflies
We used linear mixed models and GLMM to find the best set of variables to predict blowflies in the nests of each species. First, GLMM with binomial distributions allowed us to determine the best set of variables that influence whether nests are infested with blowflies (prevalence). Fixed effects in the models included the breeding season and winter environmental variables as well as hatch date of the nest. Year was the random effect in the models since several nests were sampled each year. If multicollinearity existed between two predictor variables (Pearson correlation coefficient, r > 0.7), the predictor variable that we predicted to have the biggest effect on the response variable was chosen. A global model was created with all environmental variables and hatch date, including interactions between environmental variables. Predictor variables were all standardized prior to model selection using the standardize function in the arm package (Gelman and Su, 2018). This was done to facilitate model convergence and interpretation of the parameter estimates (Grueber, Nakagawa, Laws, & Jamieson, 2011 Model selection was also done separately for parasite intensity (i.e., only those nests infested with blowflies). The response variable was number of blowflies per nestling for these linear mixed models.
The residuals in initial models were highly nonnormal. Therefore, the blowflies were log transformed in models predicting blowfly intensity to make residuals more normal. The same model selection procedure using linear mixed models and the model averaging process was completed as described above.

| Model selection for fledging success
Model selection was also used to determine the effects of blow- Year was included as a random effect in the mixed models to control for variation among years since multiple nests were sampled each year. Models were ranked using AICc and models within delta AIC values of 2 were averaged to get more precise parameter estimates, as described above.
Model selection was also done separately for blowfly intensity (i.e., only those nests infested with blowflies). Again, GLMM with binomial distributions were used. Interactions between each environmental variable and the number of blowflies per nestling were included. Model selection and model average were used to find the best set of predictor variables that predicted fledging success as described above.  Table 1 and Table 2, respectively.

| RE SULTS
Also, listed are the number of nests infested with flies (prevalence) and summary information regarding fly intensity in nests each year.
Here, mean intensity is the mean number of blowflies in infested nests; the number of nestlings has not been taken into account. Data from ninety-two nests could not be used for the following modeling analyses due to missing data.

| Changes in blowflies over time
In bluebird nests, there was no significant change in blowfly preva-

| Model selection for blowflies
The best predictors of blowfly prevalence and intensity were determined using model selection based on ranking candidate models using AICc values. The results of the top models from the model selection for blowflies of both bluebirds and flycatchers are presented in

| Model selection for fledging success
Next, we determined the best set of environmental variables for bluebird and flycatcher fledging success. For bluebird fledging success using all nests (i.e., those infested with blowflies and those not infested), all three of the top three models included the interaction between blowflies per nestling and total monthly precipitation   (Table 5). Blowflies per nestling, total monthly precipitation, and the interaction between them were each significant after model averaging and each had a relative importance of 1.00 (Table 6). The same models were also the best models explaining bluebird fledging success on only nests that were infested with blowflies (Table 5).
These models suggest that the effect of the number of blowflies per nestling on fledging success varies according to environmental conditions, specifically the amount of precipitation during the month the nest hatched. According to the models, as the number of blowflies per nestling increases, the higher the fledging success of bluebirds. The significant negative interaction with total monthly precipitation means that higher monthly precipitation decreases the positive effect of blowflies fledging success (Figure 4). Mean monthly temperature and PDSI were also variables in the top three models.
TA B L E 3 The top models (delta AICc < 2) for predicting blowfly prevalence and intensity for bluebirds and flycatchers. Monthly environmental variables (PDSI, mean temperature and total precipitation) are from the breeding season. However, these variables were not significant after model averaging and were determined to have the least relative importance of all the variables (Table 6).
For flycatchers, the top models for fledging success did not include the interaction between blowflies per nestling and total monthly precipitation (Table 5). Using all nests (i.e., those infested and not infested), the top models contained other variables, such as mean monthly temperature and PDSI (Table 5)

| D ISCUSS I ON
The main goal of this study was to use a long-term dataset to determine how environmental conditions influence the interactions between blowflies and fledging success through three specific objectives. We determined whether blowfly prevalence and intensity changed over time from 1997 to 2013, identified environmental variables that influence blowfly prevalence and intensity, and identified effects of both environmental variables and blowfly parasitism on fledging success.
We did not find a detrimental effect of blowflies on bluebird and flycatcher fledging success similar to other studies (Hannam, TA B L E 4 Model-averaged parameter estimates from the top models (delta AICc < 2) predicting blowfly prevalence and blowfly intensity for bluebirds and flycatchers. The intercept only models were among the best models to explain blowfly intensity for both species TA B L E 5 The top models (delta AICc < 2) for predicting fledging success for bluebirds and flycatchers. Fledging success using all nests and those using only infested nests were modeled separately. Monthly environmental variables (PDSI, mean temperature and total precipitation) are from the breeding season. Model weights were calculated from the set of top models  ). This may be due to compensatory feeding behaviors or other strategies such as accelerating growth toward the end of the nestling period (Killpack, Tie, & Karasov, 2014). Delaying fledging can also compensate for the negative effects of parasitism, (Hannam, 2006;Simon et al., 2004) although this was not analyzed in this study. Flycatcher nests in our study had a fairly high rate of parasitism (61% of nests were parasitized) and had no effect, positive or negative, on fledging success. provide convincing evidence that this occurs. Our study examines the relationships among three key players (i.e., the environment, parasites, and hosts) in the direction of environment-parasites-hosts.
We do not examine these relationships in the direction of environment-hosts-parasites. Host defense in these systems is needed to get a better understanding of the entire process.
One would predict that compensatory strategies would be unsuccessful when other environmental stressors exist (Dufva & Allander, 1996;De Lope et al., 1993;Pavel et al., 2008), such as a lack of resources due to drought conditions. This may be the case, however, our results suggest there is considerable variability between species and environmental conditions. For bluebirds, the in-  (Pavel et al., 2008). Similarly, other studies have suggested adverse weather (heavy rainfall) can negatively affect a host's ability to defend against parasites (Howe, 1992 Blowfly prevalence and intensity were not affected by environmental variables as has been shown in other studies (Dawson et al., 2005;Merino & Potti, 1996;Pavel et al., 2008;Poulin, 2006 Dawson et al., 2005). Larger sample sizes and additional years of data may be required to identify these conditions in the context of environmental change.
Blowfly prevalence and intensity in flycatcher nests did not differ among the six drought categories, but blowfly intensity of bluebirds did differ among categories. The extreme drought category had significantly more blowflies than the moderate drought and very moist categories. During periods of drought, a host may be more susceptible to parasites (Møller, Erritzøe, & Saino, 2003;Møller, Martín-Vivaldi, Merino, & J. Soler, 2006;Plischke, Quillfeldt, Lubjuhn, Merino, & Masello, 2010). It is possible that the parasite load in bluebird nests increased under extreme drought conditions because host condition was weakened. Other host-parasite systems such as prairie dogs and fleas in New Mexico have shown similar results (Eads, Biggins, Long, Gage, & Antolin, 2016). However, we were unable to collect host condition data for this study. Future work should focus on examining the effects of extreme drought on host condition and susceptibility to parasitism.
Drought conditions and higher temperatures are predicted to continue in the southwest due to climate change (MacDonald, 2010). Drought affects birds at the community level, such as decreases in species abundance, richness, and composition (Albright et al., 2010;Fair, Hathcock, & Bartlow, 2018). Drought also affects individuals. A lack of resources, as a result of a disturbance such as drought, often results in suppressed immune system, low weight, and lower reproductive success (Alonso-Alvarez & Tella, 2001). For instance, nestling bluebirds and flycatchers on the Pajarito Plateau experience a decrease in cell-mediated immune responsiveness during unusually dry weather conditions (Fair & Whitaker, 2008). Given that the southwest is projected to be hotter and have more frequent and prolonged droughts (Cayan et al., 2013), we predict that flycatchers may be more negatively impacted by blowflies than bluebirds. Future work should focus on investigating the role of both blowflies and climate on fledging success between species.
Altered parasite pressure through new climate patterns and increased variation of environmental conditions may be an unforeseen consequence of climate change for many species. Relatively stable host-parasite systems may become unstable, resulting in population declines. Our results generally concur with previous studies that nestlings tolerate parasitism by blowflies. However, our results also indicate there may be an interaction between climate conditions and parasitism on fledging success that could ultimately lead to a decrease in fledging success. Our results also suggest that environmental variables and blowflies were better predictors of fledging success than just blowflies or environmental variables alone. These results provide opportunities for future studies to explore these relationships in greater depth. Using fledging success as the only indicator of tolerance to parasitism was a limitation in this study. Future studies would benefit from using nestling condition to examine in depth how the host condition may influence parasite prevalence and intensity.