Experimental reduction of haemosporidian infection affects maternal reproductive investment, parental behaviour and offspring condition

Abstract

When hosts have a long coevolutionary history with their parasites, fitness costs of chronic infection have often been assumed to be negligible. Yet, experimental manipulation of infections sometimes reveals effects of parasites on their hosts, particularly during reproduction. Whether these effects translate into fitness costs remains unclear. Here, we present the results of an experimental study conducted in a free-ranging population of red-winged blackbirds (Agelaius phoeniceus) naturally experiencing a high prevalence of haemosporidian infections, with more than 95% of breeding adults infected with parasites from one or more haemosporidian genus. To assess effects of infection during reproduction, we manipulated adult red-winged blackbird females’ parasite burden by administering an anti-haemosporidian medication before onset of egg-laying. Experimental reduction of infection resulted in significant benefits to mothers and their offspring. Medicated females laid heavier clutches, invested more in incubation and provisioning behaviour, and produced more fledglings than control females. Nestlings of medicated females had higher haematocrit, higher blood glucose, and lower reactive oxygen metabolites than nestlings of control females. Overall, our results provide evidence that, even in a species with high prevalence of infection, parasites can lead to decreased maternal investment and offspring quality, substantially reducing fitness.

1. Introduction

Parasites can affect their hosts in multiple ways [1–4], but how detrimental these effects are might depend on the history of interactions between parasite and host. In naive host populations, costs of infection can be high, with high virulence and mortality rates [5,6]. By contrast, costs of infection in host populations with a long history of interaction with a parasite have often been assumed to be negligible [7,8], because parasites generally rely on living and active hosts for transmission to other hosts and for completion of their life cycle [9]. Over evolutionary time, we also expect hosts to evolve strategies to better tolerate or resist the fitness effects of parasitism [10]. Yet, even in systems where parasites and hosts have co-evolved, we might still expect fitness costs as parasite and host compete for same resources [11].

Parasitic infections are expected to be particularly costly during energetically demanding life-history stages, such as reproduction because they can shift the balance of resource allocation within a host [12]. Some evidence suggests investment in reproduction results in decreased or altered immune function [13–15] and increased risks of infection [16–18]. Similarly, disease-mediated decreases in maternal body condition can increase allocation to immune function, reducing reproductive investment [19–21]. Manipulation of infection in captive animals largely confirms the notion that maternal parasitism results in costs to mothers and their offspring (e.g. [22,23]).

Findings in natural populations have not been as consistent in detecting costs of parasitic infections in parents and their offspring [16,24,25]. For example, red deer (Cervus elaphus) mothers with naturally higher helminth parasitism during lactation have reduced fecundity and survival [26], but European common lizard (Lacerta vivipara) mothers with naturally higher haematophagous mite loads (family Laelapidae) produce yearlings with increased growth rates and higher reproductive investment at first reproduction [27], and Hawai'i 'amakihi (Hemignathus virens) mothers naturally infected with avian malaria (Plasmodium relictum) do not differ from uninfected mothers in their reproductive success [28]. However, costs of chronic infections in natural populations are difficult to detect without experimental manipulation [29,30]. Further, causation can be hard to discern without experiments, as higher reproductive investment can cause increased susceptibility to infection or increased parasite burden [16,31,32], and infection can paradoxically cause increased reproductive investment (e.g. terminal reproductive investment, [33]). Yet, even in experimental studies where parasite burden has been manipulated in natural populations, results remain mixed, with some studies finding no costs of infections and others finding effects in mothers but not in offspring, and vice versa [34–38].

Vector-borne Haemosporidians have been widely used to study host-parasite interactions partly because of their ability to infect a wide range of vertebrate hosts [39–41]. Costs of haemosporidian infections appear to vary considerably among individuals, populations and species. In naive host populations, virulence of haemosporidian infections can be severe, resulting in high mortality [30,42], whereas in populations where these parasites have a longer coevolutionary history with their hosts, parasitic effects on host fitness remain unclear, even during energetically demanding life-history stages [43–45]. Because pathogenicity of chronic haemosporidian infection is often not detected in the wild [40], costs of chronic infections in experienced hosts have generally been assumed to be low [30,46]. Yet, some evidence points to costs of chronic infection even in hosts where Haemosporidians are endemic. For example, haemosporidian infections can relapse when hosts experience stressful conditions [40], including during reproduction [47], suggesting that allocation to immune defences against these parasites is costly and can compromise investment in other demands.

Although most of the literature exploring the effects of haemosporidian infections on parental fitness and offspring phenotype has been correlative, the few field studies where parasite load has been experimentally manipulated largely confirm the notion that these parasites can impose costs. For example, a medication experiment in blue tits (Cyanistes caeruleus) found that treated females had higher fledging success than control females, but nestlings’ body mass and tarsus length were unaffected [34]. Experimental manipulation of parasite loads in a second blue tit population also showed that medicated females had higher hatching success, provisioning rates, and fledging success than control females [38]. Similarly, Haemoproteus-infected female house martins (Delichon urbicum) given antimalarial medication before breeding produced larger clutches and had higher hatching and fledging success than control females, but their nestlings had similar body mass, haematocrit and T-cell-mediated immune responses [36]. The lack of consistent findings of costs of maternal infection for offspring is particularly interesting given that during the nesting cycle, chicks of these altricial species are highly dependent on their parents for resources. Most experimental studies have, however, used body size as an indicator of offspring quality and/or condition [34,38,48], which might not be sufficient for detection of costs of maternal infection, as nestlings may invest resources in growth at the expense of other functions [49], leading to a potential underestimation of the costs of maternal haemosporidian infections for individuals, and, consequently, for the population. A more integrative assessment of nestling condition could provide a more accurate assessment of costs of maternal infection.

Here, we present the results of an experimental study conducted in a free-ranging population of red-winged blackbirds (Agelaius phoeniceus) naturally experiencing high rates of chronic infection with haemosporidian parasites, with more than 95% of breeding birds parasitized with one or more haemosporidian genera [50]. Birds in this population have experienced relatively high prevalence of infection with haemosporidian parasites in the genus Haemoproteus, Leucocytozoon and Plasmodium since at least the late 1980s [47]. Red-winged blackbirds at our field site appear to tolerate infection with haemosporidian parasites relatively well, with infected individuals breeding successfully and surviving across years [51]. Yet, previous captive, experimental work with males in this population revealed costs of chronic haemosporidian infection [52]. To assess whether haemosporidian infection affects maternal fitness and offspring phenotype, we experimentally manipulated infection in adult female red-winged blackbirds before onset of egg-laying by administering an anti-haemosporidian medication. We measured effects of treatment on maternal reproductive investment and offspring phenotype, relative to controls. If experimental reduction of infection leads to higher maternal investment, we expected medicated females to have higher reproductive success (lay eggs earlier, produce larger or heavier clutches, produce more fledglings) and increased investment in maternal care behaviours (incubation and provisioning) than control females. Because male red-winged blackbirds also provision their nestlings [53], we measured paternal care behaviour to determine whether reduced parasite load in females affected their mates' behaviour. If maternal haemosporidian infections also result in costs to the offspring, we expected offspring from medicated females to have improved condition, reflected in higher haematocrit, glucose, scaled body mass index (SMI, an integrative measure of condition incorporating both body mass and body size [54]) and lower baseline corticosterone concentrations. We also predicted offspring from medicated females would show better oxidative balance, expressed in higher total antioxidant capacity and/or lower reactive oxygen metabolites. Further, we predicted nestlings from medicated females would have lower incidence of infection than nestlings from control females.

2. Material and methods

(a) . Study population

We studied a free-ranging population of red-winged blackbirds during two breeding seasons (between April and July 2017 and 2018) at the Queen's University Biological Station in eastern Ontario, Canada (44°34 N, 76°19 W, approx. 135 m.a.s.l.). Previous studies conducted with the same population since 2013 have shown that adults experience high prevalence of infections with parasites from one or more haemosporidian genera, including Haemoproteus, Leucocytozoon and Plasmodium [50], with Plasmodium being the most common Haemosporidian in our population [52]. These polymerase chain reaction (PCR)-based results parallel previous less sensitive microscopy-based findings, indicating that high infection prevalence has persisted in this population for several decades [47].

(b) . Experimental parasite burden manipulation

To assess the effects of infection on maternal reproductive investment and offspring phenotype, we experimentally manipulated parasite burdens in adult red-winged blackbird females before onset of egg laying by administering an anti-haemosporidian medication, alternating with a control solution with each new female caught. At capture, we orally administered 100 μl of 10% sugar water to individuals in our control group and the same volume of sugar water mixed with antimalarial medications (Primaquine and Chloroquine; for dosage and validation see the electronic supplementary material, methods) to birds assigned to the medication group. Both solutions were labelled as solution A or solution B by colleagues who were not part of the study, so experimenters were blind to treatment type until after data analysis. Previous experimental work conducted in the same population has shown that this medication is successful in reducing parasitaemia for at least two weeks [52], so, we expected this dose to effectively decrease parasitaemia in females, but not fully clear their haemosporidian infections. We aimed to administer multiple doses to each female following methods used in a previous captive experiment conducted in males in the same population [52] that were derived from published veterinary and parasitology literature [55–57]. In 2017, each female received one dose of medication or control solution at the time of capture and was then released immediately. We aimed to recapture these females later to administer secondary doses, but we were unable to consistently recapture females (only five females received second doses in 2017), and so we modified our protocol during the 2018 field season to ensure all birds were captured, handled and treated similarly. During 2018, each female was given one dose of the allocated solution at the time of capture and then held in an outdoor aviary for two days so we could administer two additional doses of the same solution (one additional dose per day in captivity). Immediately following the third dose (day two of captivity), females were released at the original capture site. We statistically account for this difference in methods between years but note that results do not qualitatively differ if we only analyse data from 2017 (the year with the larger sample size and more conservative medication treatment).

(c) . Data collection

We caught all birds with either mist nets or seed-baited walk-in (Potter) traps, both in the marsh near nesting sites and at foraging areas nearby. After capture, we collected a small blood sample (less than 400 μl) via puncture of the brachial vein and fitted each bird with a uniquely numbered Canadian Wildlife Service leg band (CWS banding permit 10771) and a unique combination of up to three colour bands. We recorded morphological measurements including tarsus length (mm), head-bill length (mm), bill length (mm), flattened wing length (mm) and body mass (g). We used the blood from one capillary tube to measure blood glucose using a portable glucometer (One Touch Ultra, LifeScan Inc.) and to prepare blood smears by rapidly spreading a droplet of approximately 5 µl of blood on a glass microscope slide (Fisherbrand Premium Frosted Microscope Slides, 25 × 75 × 1 mm, Fisher Scientific). We immersed each slide in a methanol fixative and left them to air dry, before storing them in a slide box. We stored all remaining blood samples in a cooler box and later used them to assess haematocrit and haemosporidian infection status via PCR (electronic supplementary material, methods).

After release, we monitored females regularly to observe their breeding behaviour and nesting success. Some females were not re-sighted and might have dispersed away from the study site or nested in inaccessible parts of the marsh (number of experimental females not resighted = 24 out of 40, number of control females not resighted = 26 out of 42). On day two of incubation, we visited each nest, weighed each egg in it, and placed temperature loggers to record maternal incubation behaviour (electronic supplementary material, methods). We returned to the nest on the predicted hatch date, 11–12 days after the start of incubation, to determine hatching success.

On day six after hatching, we returned to the nest and set up video cameras approximately 7–8 m away from a nest in deep vegetation to record parental provisioning behaviour (electronic supplementary material, methods). On day 10 after hatching, we returned to the nest and sampled all surviving offspring (males and females). We measured all standard morphometrics as with adult females (see above), other than wing length. We also collected blood samples (up to 200 µl) from all nestlings to quantify the same blood parameters we assessed in females (details above), as well as plasma corticosterone concentrations and oxidative status (electronic supplementary material, methods). We obtained these blood samples within 3 min of disturbing the nest to minimize the influence of capture on corticosterone levels, which can increase rapidly after this period [58]. We focused on these blood parameters and SMI to assess effects of maternal antimalarial treatment on offspring because several of these metrics have been linked to both individual condition (e.g. [59,60]) as well as haemosporidian infection (e.g. [46,52,61,62]). For example, nestlings with higher SMI have been shown to have higher survival [60], and individuals infected with Haemosporidians display reduced total antioxidant capacity and increased reactive oxygen metabolites when their immune response is activated, causing oxidative damage [61]. We continued to monitor nest progress by assessing parental activity at the nest (e.g. parents observed flying to and from the nest carrying food) from a distance over the next few days to minimize disturbance until we could confirm nest failure or the number of nestlings that successfully fledged. Here, we include only nestling measurements collected in 2017, as in 2018 we used nestlings for a separate experiment, which necessitated sampling of individuals at a younger age. This, together with the high nest failure typically experienced by this species prior to day 10, resulted in a smaller nestling sample size compared to the maternal sample size.

(d) . Statistical analysis

We performed all statistical analyses using R v. 4.1.2 [63]. We created all graphs using ggplot2 [64] and jtools [65]. We applied a Box Cox transformation to timing of laying (number of days between initial treatment and laying of the first egg), a square root transformation to paternal provision rates, and a natural log transformation to corticosterone concentrations to make them more normally distributed and improve model fit [66].

We used a generalized linear model (GLM) with a binomial distribution and a GLM with a negative binomial distribution to determine if medicated females differed from control females in their haemosporidian infection status and their total haemosporidian parasitaemia at the time of initial capture, prior to treatment. We entered maternal infection status or maternal parasitaemia (number of parasites) as the response variables in separate GLMs with treatment (control, medicated) and year (2017, 2018) as the categorical fixed effects to determine if parasitaemia differed among years or treatment group.

We assessed pairwise correlations (Hmisc; [67]) between our numerical estimates of reproductive success and parental behaviour to determine if any of these factors were highly correlated and excluded those variables with r > 0.7 from the analysis (electronic supplementary material, results).

We used a principal component analysis (PCA) (factoextra; [68]) of haematocrit (%) and maternal parasitaemia to estimate maternal infection burden at the time of treatment (electronic supplementary material, results). We used a PCA of haematocrit (%) and maternal parasitaemia because this integrative measure of parasite burden has been used in the past to estimate tolerance of infection, with individuals with higher hematocrit than expected given their parasitemia interpreted as being less negatively affected by infection (i.e. more tolerant; [50]). Past experimental work in this population has suggested that infection causes a reduction in haematocrit [52]. As such, we considered the resulting composite trait of PC1 as a measure of maternal infection burden, which we predicted might influence the magnitude of treatment effects of anti-haemosporidian medication. We multiplied PC1 by −1 for analyses and presentation of results, so that high values correspond to high maternal infection burden.

We used linear models (LMs) to test for the effects of treatment on timing of laying (number of days between treatment date and laying date), total mass of all eggs in a clutch (clutch mass, g), maternal provisioning rates, and paternal provisioning rates. We used a GLM with a Poisson distribution to test for the effects of treatment on total number of off-bouts (the number of times a female left her nest during an 8 h incubation period), overall number fledglings produced, and number of fledglings produced from nests, excluding those that failed owing to predation. Excluding depredated nests in this complementary analysis allows investigation of the mechanism of variation in fledging number between medicated and control females. We used a GLM with a binomial distribution to test for effects of treatment on percent of time a female spent incubating her eggs. We entered each trait as response variables in separate LMs or GLMs, with treatment and infection burden (−PC1) as fixed effects. Calendar day of laying of the first egg and sampling year were included as cofactors in these models. Number of nestlings present at the time of provisioning measures was included as a covariate in analyses of maternal and paternal provisioning rates.

We used a generalized linear mixed model (GLMM) with a binomial distribution to test for the effects of maternal treatment on offspring infection status (presence or absence of haemosporidian-infected cells) and linear mixed models (LMMs; lme4 [69]) to test for effects of maternal treatment on offspring condition. Traits assessed as measures of nestling condition included haematocrit (%), glucose (mmol l⁻¹), SMI (electronic supplementary material, methods), total antioxidant capacity (Carratelli units), reactive oxygen metabolites (mMol of H₂O₂ equivalents) and corticosterone (ng ml⁻¹). Each trait was entered as a response variable in a separate GLMM or LMM, with treatment and calendar day of laying of the first egg as fixed effects. We did not include year as a cofactor in these models, because this analysis was limited to data collected from nestlings in 2017. Nest identity was included as a random effect in all models.

To test for the effects of the fixed factors on the response variables in each of our models, we adopted a model selection approach (MuMin; [70]). Initially, we constructed a global model that included all fixed effects. Then, we compared the global model and the null model with different candidate models with or without various fixed effects, but always including treatment, using Akaike information criterion, corrected for small sample size (AICc; see the electronic supplementary material, tables S1–S15 for details on candidate models and model selection approach). We report results from the model from this candidate set that had the lowest AICc. Comparison of models with and without random effects using AICc revealed that inclusion of nest identity improved model fit only for nestling total antioxidant capacity, so we retained the random effect in these models only and simplified all other LMMs and GLMMs to LMs or GLMs, respectively.

We assessed model fit and assumptions of the global and top-ranking models by visually inspecting plots of normalized residuals versus fitted values, distributions of normalized residuals, and relationships between normalized residuals and fixed effects, assessing Cook's distance (predictmeans; [71]) to detect influential datapoints, and checking for over-dispersion, under-dispersion and zero-inflation (DHARMa; [72]). We report standardized effect sizes and confidence intervals for each effect [65].

3. Results

(a) . Effects of treatment on maternal reproductive investment

We measured effects of treatment with anti-haemosporidian medication in 32 females (16 control and 16 medicated). Infection status at the time of treatment did not differ between treatment groups (medicated: 81.2% infected, control: 87.5% infected; binomial GLM, exp(β) = 0.63 [0.09–4.41], z = −0.48, p = 0.64) or between years of treatment (2017: 85.7%, 2018: 81.8%; binomial GLM, exp(β) = 0.77 [0.11–5.54], z = −0.26, p = 0.78). The most common parasite genus was Plasmodium with 81.2% prevalence, followed by Haemoproteus with 28.1%, and Leucocytozoon with 12.5%. Maternal parasitaemia did not differ between medicated and control females at the time of treatment (mean ± s.e., medicated: 459 ± 178, control: 481 ± 131 parasites per 10 000 cells; negative binomial GLM, exp(β) = 1.20 [0.41–3.53], z = 0.33, p = 0.74), or between years of treatment (2017: 594 ± 152, 2018: 234 ± 100 parasites per 10 000 cells; negative binomial GLM, exp(β) = 0.37 [0.12–1.16], z = −1.70, p = 0.09).

Medicated females laid heavier eggs (LM, β = 0.80 [0.12–1.48], t = 2.42, p = 0.02, n = 13 control and 13 medicated; figure 1a), spent more time on their nests incubating their eggs (binomial GLM, exp(β) = 1.35 [1.16–1.57], z = 3.82, p = 0.0001, n = 15 control and 13 medicated; figure 1b), and provisioned their chicks at higher rates (LM, β = 3.12 [0.58–5.66], t = 2.55, p = 0.02, n = 11 control and 13 medicated; figure 1c). Interestingly, males attending nests of medicated females tended to provision their chicks at lower rates than males paired with control females (LM, β = −0.72 [−1.53–0.09], t = −1.84, p = 0.08, n = 11 control and 13 medicated; figure 1d). In addition, medicated females fledged more nestlings compared to control females (Poisson GLM, exp(β) = 2.00 [1.11–3.60], z = 2.32, p = 0.02, n = 16 control and 16 medicated; figure 1e), including when depredated nests were excluded from the analysis (Poisson GLM, exp(β) = 2.10 [1.17–3.78], z = 2.49, p = 0.01, n = 13 control and 12 medicated; figure 1f).

Figure 1.

Female red-winged blackbirds treated with anti-haemosporidian medication laid heavier clutches (a), spent more time incubating their eggs (b), and provisioned their offspring at higher rates (c) than control females. Males paired with medicated females tended to visit their nests at lower rates than males paired with control females (d). Females treated with anti-haemosporidian medication produced more fledglings than control females (e), even when depredated nests were excluded from the analysis (f). Range plots show the mean (black dot) and 95% upper and 5% lower percentiles (black line ranges) of the response for control and medicated females. Grey circles represent raw data and have been jittered on the x-axis to reduce overlap.

We found no evidence of an effect of treatment on the interval between treatment date and laying date (medicated: 17.0 ± 2.75, control: 17.3 ± 3.56 days; LM, β = 0.12 [−0.44–0.68], t = −0.44, p = 0.66, n = 16 control and 16 medicated) and number of off-bouts (i.e. the number of times a female left her nest during an 8 h incubation period; medicated: 21.2 ± 1.35, control: 20.9 ± 1.48 off bouts per 8 h period; Poisson GLM, exp(β) = 0.97 [0.82–1.14], z = −0.37, p = 0.71, n = 15 control and 13 medicated). However, females with higher infection burdens had shorter intervals between treatment date and laying date (LM, β = −0.39 [−0.67 – −0.11], t = −2.84, p = 0.008, n = 16 control and 16 medicated; electronic supplementary material, figure S1). We also found several relationships between reproductive investment and timing of breeding (electronic supplementary material, figure S2) and a difference in maternal incubation rates between years (2017: 60.40% ± 1.32; 2018: 64.80% ± 1.68 time a female spent sitting on her eggs; exp(β )= 1.19 [1.01–1.40], z = 2.11, p = 0.03, n = 15 control and 13 medicated).

(b) . Effects of maternal treatment on offspring phenotype

We measured effects of maternal treatment with anti-haemosporidian medication in 19 ten-day-old nestlings (seven control nestlings from four nests and 12 medicated nestlings from four nests). Nestlings from medicated females did not differ from nestlings from control females in infection status (medicated: 7 of 12 infected, control: 5 of 7 infected; binomial GLM, exp(β) = 0.56 [0.08–4.14], z = −0.57, p = 0.57).

Nestlings from medicated females had higher haematocrit (LM, β = 1.01 [0.12–1.90], t = 2.39, p = 0.03; figure 2a), higher blood glucose (LM, β = 1.11 [0.25–1.97], t = 2.71, p = 0.02; figure 2b), and lower reactive oxygen metabolites (LM, β = −0.11 [−0.14 – −0.09], t = −9.45, p < 0.0001; figure 2c) than nestlings from control females. Some of these parameters also varied with timing of breeding, independent of treatment effects (electronic supplementary material, figure S3).

Figure 2.

Nestling red-winged blackbirds from females treated with an anti-haemosporidian medication had higher haematocrit (a), higher glucose (b), and lower reactive oxygen metabolites (ROM; c) than nestlings from control females. Range plots show the mean (black dot) and 95% upper and 5% lower percentiles (black line ranges) of the response for control and medicated females. Grey circles represent raw data and have been jittered on the x-axis to reduce overlap.

Nestlings from medicated females did not differ from nestlings from control females in their SMI (mean ± s.e., medicated: 27.99 ± 1.08, control: 30.32 ± 1.26 SMI; LM, β = −0.63 [−1.61–0.35], t = −1.36, p = 0.19), total antioxidant capacity (mean ± s.e., medicated: 413.0 ± 1.64, control: 410.0 ± 2.80 Carratelli units; LMM, β = 0.73 [−0.47–1.93], t = 1.19, p = 0.29), or plasma corticosterone concentrations (mean ± s.e., medicated: 2.46 ± 0.32, control: 2.94 ± 0.71 ng ml⁻¹; LM, β = −0.24 [−1.22–0.74], t = −0.51, p = 0.62).

4. Discussion

In systems where hosts and parasites have coevolved, chronic infection has often been assumed to have minimal fitness costs because parasitaemia is typically low and parasite virulence can be reduced by selection to minimize harm to hosts [30,40,46]. By contrast, we provide experimental evidence that infection with haemosporidian parasites results in marked costs for mothers (figure 1) and their offspring (figure 2). Females treated with anti-haemosporidian medication invested more in maternal care behaviour and had higher reproductive success. Medicated females also produced offspring in better condition, reflected in their higher haematocrit and glucose, and lower reactive oxygen metabolites than nestlings of control females. Our results suggest that chronic infection with haemosporidian parasites during reproduction has important transgenerational effects, potentially leading to significant population-level effects.

Experimental reduction of haemosporidian infections increased maternal reproductive investment. In our study, medicated and control females laid clutches that had similar numbers of eggs (mean ± s.e., medicated: 3.74 ± 0.18, control: 3.79 ± 0.10 eggs clutch⁻¹), but medicated females' eggs weighed on average 0.28 g more than those of control females (mean ± s.e., medicated: 4.24 ± 0.06, control: 3.96 ± 0.07 g, 14% heavier clutches). Avian mothers provide resources to their growing offspring via the transfer of nutrients through the egg yolk [73]. By producing larger, more nutrient-rich eggs, avian mothers can, therefore, confer direct benefits to their offspring in terms of growth and survival [74]. Yet, producing larger, more nutrient-rich eggs is costly, particularly when individuals are burdened by infections. Reducing infection in medicated females allowed those mothers to invest more resources in producing larger, more nutrient rich eggs, thus placing their offspring at an advantage compared to offspring of control females.

Experimentally reducing parasite burden also affected behavioural investment in parental care. Medicated females spent on average 12% more time on the nest incubating their eggs and provisioned their chicks approximately 31% more than control females. As incubation is critical for maintaining optimal temperatures for the development of the growing embryo [75,76], avian mothers spend considerable time on their nests during incubation. Yet, incubation is also costly for mothers because they lose heat to the eggs and cannot forage while sitting on the eggs [77]. Similarly, parental provisioning of nestlings is essential, and very demanding in altricial species [78,79], because of the costs associated with intense flight and foraging activities [80]. During parental stages, an avian mother must, therefore, balance optimal investment in offspring with self-maintenance. Reduced female condition resulting from parasitic infections can significantly alter this balance [12,81], with infected females typically decreasing parental effort [25,38]. In our experiment, medicated females increased investment in parental care by spending more time at the nest during each incubation bout and bringing food to nestlings at a higher rate, relative to controls. These findings suggest that haemosporidian infections might increase metabolic rates and caloric intake needs in parasitized females [82], requiring them to spend more time away from their nests foraging. Parasitic infections could also have compromised control females’ condition so that they were unable to maintain the same levels of parental care provided by medicated females [83]. Our finding that males paired with control females tended to provision their chicks more than males paired with medicated females further strengthens the idea that haemosporidian infections limit maternal care behaviour [77,78]. The difference we see in paternal care in males becomes even more striking when we consider that red-winged blackbirds are polygynous and males often provision offspring only at some nests [84], suggesting that effects of infections with Haemosporidians can extend beyond an individual host.

While treatment with anti-haemosporidian medication affected several measures of maternal reproductive investment, we found no evidence of an effect on the interval between treatment date and laying date. We found that females with higher infection burden at the time of treatment laid their clutches sooner after initial capture compared to females with lower infection burden. Females can adjust their reproductive effort based on current conditions [85,86], with individuals in poor condition sometimes investing more in current than in future reproduction when the likelihood of surviving to the next breeding event is low (‘terminal investment hypothesis'; [87]). Highly parasitized red-winged blackbird mothers in our study could have perceived their chances of survival to the next breeding season to be low and adjusted their reproductive effort accordingly. Breeding early in the season could also have allowed heavily infected females to better align their own and their nestlings' needs with times when resources are most abundant [88], improving their chances of survival and offsetting some of the costs associated with maternal investment. Alternatively, increased parasite burden might simply reflect advancing reproductive development, given previous findings that parasitemia increases during breeding, possibly owing to reallocation of resources towards reproduction and away from immune function [16,31].

Experimental reduction of maternal haemosporidian infection also affected several measures linked to offspring condition, including haematocrit, blood glucose and oxidative status. Nestlings of medicated females had higher haematocrit compared to nestlings of control females. Haematocrit is linked to metabolic demands [89] and can be related to energy availability, with low hematocrit levels indicating anemia, which can result from insufficient diet and vitamin intake [90,91]. Nestlings of medicated females also had higher blood glucose compared to nestlings of control females. Because glucose responds to frequency and quality of feeding, lower levels of glucose in nestlings of control females suggest that parasitized females were unable to provide the same level or quality of provisioning to their chicks as females with reduced parasite burdens, reflecting our finding of lower provisioning rates in control females. Nestlings of medicated females also had lower reactive oxygen metabolites compared to nestlings of control females. In red-winged blackbirds, nestlings depend largely on their mothers for feeding and care [84]. A reduction in the amount of food and/or care provided by their mothers may have left nestlings of control females unable to balance the energetic demands of growth and other costly functions, such as self-maintenance [92,93], which might be supported by our nestling body condition results. By day 10 post-hatching, nestlings of control and medicated females attained similar body sizes (as measured by SMI), indicating that nestlings in both treatment groups invested heavily in growth at this stage. An experimental supplementation of dietary antioxidants in red-winged blackbird nestlings has shown that chicks allocate extra resources to growth rather than to reducing oxidative damage [94]. Maternal parasitism might have, therefore, magnified the trade-off between growth and self-maintenance in nestlings of control females.

We detected clear fitness costs of haemosporidian infection. Medicated females produced almost twice as many fledglings (94% more) as control females. This difference between medicated and control females was larger when we excluded depredated nests from the analysis, indicating that intrinsic causes of nest failure were an important driver of differences in fledging success between control and medicated females. We also found that medicated females had a lower partial brood failure compared to control females (medicated: 10% versus control: 60% of nests that fledged at least one offspring), further reflecting differences in nestling quality and survival between the two treatments. Lower offspring survival reduces the number of individuals available for recruitment in the population. In our red-winged blackbird population, the number of nestlings fledged from a cohort each year correlates strongly with the number of individuals returning the following year [95]. Additionally, because the early environment experienced by offspring can have important fitness consequences that persist across life-history stages, including affecting future reproduction and senescence [96,97], costs of maternal infections in nestlings may further amplify costs of infection at the population level.

Here, we have shown that the fitness effects of haemosporidian infections can be substantial, even in a species with high levels of exposure to the same genera of parasites. Chronic haemosporidian infections negatively affected both maternal reproductive investment and offspring condition in red-winged blackbirds. Avian parents can dramatically increase the chances of survival and reproduction of their offspring by investing resources at key stages during their early growth and development. Our experimental manipulation shows that parasites can substantially hinder this investment at early stages of the offspring development, leading to important transgenerational effects that could scale up to impact the entire population.

Ethics

All procedures were approved by the Queen's University Animal Care Committee (protocol no. 2016-1646).

Data accessibility

All data and scripts related to this article are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.34tmpg4pk [98].

The data are provided in the electronic supplementary material [99].

Authors' contributions

I.S.: conceptualization, data curation, formal analysis, investigation, methodology, validation, visualization, writing—original draft, writing—review and editing; S.O.: conceptualization, data curation, formal analysis, investigation, methodology, visualization, writing—original draft, writing—review and editing; I.T.M.: conceptualization, funding acquisition, resources, writing—review and editing; F.B.: conceptualization, formal analysis, funding acquisition, methodology, project administration, resources, supervision, validation, visualization, writing—original draft, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This work was supported by funding from the U.S. National Science Foundation (to F.B. and I.T.M., no. IOS-1145625), the Natural Sciences and Engineering Research Council of Canada (Discovery Grant no. 05883–2014 to F.B.), a Canadian Foundation for Innovation award (no. 32672 to F.B.) and the Queen's University's Summer Work Experience Program as well as a Research Leader's Fund Award (to F.B).