Maternally invested carotenoids compensate costly ectoparasitism in the hihi

Abstract

Dietary ingested carotenoid biomolecules have been linked to both improved health and immunity in nestling birds. Here, we test whether maternally invested egg carotenoids can offset the cost of parasitism in developing nestling hihi (Notiomystis cincta) from the bloodsucking mite (Ornithonyssus bursa). Our results reveal clear negative effects of parasitism on nestlings, and that maternally derived carotenoids compensate this cost, resulting in growth parameters and ultimate mass achieved being similar to nonparasitized young. Our results offer an unique example of a direct positive relationship between enhanced maternal investment of carotenoids and an ability to cope with a specific and costly parasite in young birds. As O. bursa infestations reduce population viability in hihi, our findings also highlight the importance of key nutritional resources for endangered bird populations to better cope with common parasite infestations.

The quality of young birds at fledging depends not only on the nutrition they receive from their parents, but also on the microenvironment of their nest. Parasites are frequently observed in nestling birds and can have a severe impact on their health, condition, and immediate survival (1), potentially even causing populations to decline (2, 3). It is not surprising therefore, that parasite-induced mortality is a major selection pressure on nestling growth and development time (4). Parent birds provide specific resources to developing embryos and young nestlings that can help them overcome such costly parasitism (5, 6), and any interaction between these beneficial resources and costly parasite load will combine to determine the quality of fledgling birds and hence their subsequent survival and reproductive ability (7–9).

Maternal effects, involving the direct investment of beneficial resources into the egg, have received increasing attention in the field of evolutionary biology and are thought to impact embryonic development, offspring quality and survival (5, 6). Through maternal deposition of resources into the egg, breeding females can control the variability in size of nestlings within a clutch by the point of hatching and influence the nestlings' competitive ability, subsequent growth trajectories and fledging time (4, 10). In addition, maternally invested resources can provide protection to the developing nestling from parasites via an optimal immune response (11–13). Carotenoids are one such biomolecule invested by breeding females into the egg yolk (14). This maternal investment of carotenoids into eggs has been positively associated with health, vigor, and survival of offspring (12, 15–19). Of the beneficial roles carotenoids play, their positive actions in the immune system are thought to be of primary importance (20). However, the only direct tests of the hypothesis that maternally derived carotenoids compensate the costs associated with specific nest parasites is the well-studied great tit (Parus major)–hen flea (Ceratopyllus gallinae) host–parasite system, where no such relationship was found (18, 19).

Here, we test whether nestling condition and survival is influenced by parasites in a wild population of a threatened New Zealand passerine, the hihi (Notiomystis cincta), and further, we test whether the maternal investment of carotenoids into the egg yolk compensates the expected cost of parasitism. We have previously shown that enhanced carotenoids have little effect on hatchability of eggs, nestling survival, or nestling immunity in parasite controlled environments (21). To introduce a natural stressor to the nest environment we manipulated exposure of nestlings to the bloodsucking mite (Ornithonyssus bursa), a common parasite of hihi (2). O. bursa are known to alter host behavior and detrimentally effect host fitness in some species (22, 23). Our experimental manipulations (i) quantified the effects of O. bursa on survival, growth, and condition of nestlings to the point of fledging, and (ii) tested whether maternal investment of carotenoids can compensate any negative effect O. bursa have on survival, growth, and condition.

Results

We monitored 81 clutches (53 first and 28 second clutches), 61 (75%) of which successfully produced fledglings. O. bursa removal was applied to 39 clutches (O. bursa removal clutches) (91 fledglings), leaving 42 clutches (70 fledglings) untreated (O. bursa untreated clutches). Females associated with 79 of the 81 clutches were provided with either carotenoid-enriched (36 clutches, 69 fledglings) or sugar-water-only supplements (43 clutches, 92 fledglings); the remaining 2 clutches were discovered too late for food supplement to be assigned. Food supplement was allocated by alternating distinct forest patches to either carotenoid-enriched or sugar-water only, followed by alternating clutches within each forest patch between O. bursa removal or untreated (see Materials and Methods). The allocation of carotenoid-enriched and sugar-water supplements were approximately balanced with respect to O. bursa treatment. Monitoring of O. bursa presence showed that 34 of the 42 untreated clutches (81%) developed O. bursa infestations, whereas we detected presence of O. bursa in 26 of the 39 removal clutches (67%) (which were immediately treated). The small number of nests where no O. bursa were seen were left in their treatment groups for all analyses. Biochemical analysis of the small numbers of infertile eggs confirmed that females feeding on carotenoid enriched diets deposited more carotenoids into the egg yolk (sugar-only supplement, n = 6 eggs from 6 nests, total carotenoids concentration = 19.2 ± 1.9 SE μg/mL; carotenoid-enriched supplement, n = 6 eggs from 5 nests, total carotenoids concentration = 105.5 ± 20.1 SE μg/mL) (also see ref. 24 for further evidence of enhanced egg yolk carotenoids with identical carotenoid supplementation).

Mean clutch size was significantly greater in the O. bursa removal clutches compared with O. bursa untreated clutches and in first clutches compared with second clutches, but not between carotenoid-enriched and sugar-water supplements (Table 1). First clutches had a 14% increase in clutch size over second clutches, and O. bursa removal clutches had an 8% increase in clutch size over untreated clutches (Table 1), with no significant interaction between these effects (P = 0.680). Hatching probability was not significantly affected by any of the 3 variables considered (Table 1).

Table 1.

Generalized linear models examining the effects of mite (O. bursa) treatment, carotenoid supplementation, and first (versus second) clutch laid on reproductive parameters of clutch size, number of eggs hatched (proportion of number of eggs laid), and number of chicks fledged (proportion of number of eggs hatched)

Reproductive parameters	Effect	Estimate (SE)		Fixed effects
		No	Yes	Type III F-test	DF	P-value
Clutch size, normal variable	Mite treatment	3.93 (0.120)	4.24 (0.108)	4.25	1,70	0.043
	Carotenoid supplementation	4.17 (0.110)	4.00 (0.118)	1.23	1,70	0.272
	Second clutch	4.36 (0.091)	3.81 (0.142)	10.65	1,70	<0.002
Hatching probability, proportion ∼ binomial variable	Mite treatment	0.776 (0.044)	0.750 (0.040)	0.20	1,70	0.654
	Carotenoid supplementation	0.757 (0.041)	0.769 (0.043)	0.04	1,70	0.848
	Second clutch	0.789 (0.031)	0.734 (0.056)	0.77	1,70	0.382
Fledging probability, proportion ∼ binomial variable	Mite treatment	0.431 (0.070)	0.629 (0.060)	4.80	1,70	0.032
	Carotenoid supplementation	0.489 (0.065)	0.573 (0.067)	0.90	1,70	0.346
	Second clutch	0.774 (0.038)	0.273 (0.068)	32.84	1,70	<0.001

Survival of nestlings to fledging was significantly greater in the O. bursa removal clutches compared with O. bursa untreated clutches and in first clutches compared with second clutches, but not between carotenoid-enriched and sugar-water supplements (Table 1). For a first clutch, the odds of fledging were 3.5× higher than the odds of fledging for second clutches, and 1.4× higher for O. bursa removal clutches than untreated clutches, with no significant interaction between these effects (P = 0.130).

Controlling for clutch identity, nestling sex and clutch size there was a significant effect of O. bursa treatment on nestling condition (Log₁₀ transformed body mass at day 24 controlling for body size by including tarsus length as a covariate), being 8.1% heavier, on average, in the O. bursa removal clutches (39.9 g) versus the untreated clutches (36.9 g) (Table 2). Neither plasma carotenoid concentration at 24 days nor total white blood cell counts were significantly different between O. bursa removal and untreated clutches (Table 2). Carotenoid supplementation had no significant effect in any of these models and no interaction terms were retained (Table 2).

Table 2.

Generalized linear mixed models examining the effects of nest mite (O. bursa) treatment and carotenoid supplementation on nestling condition, plasma carotenoid concentration, and total white blood cell count

Models	Effect	Estimate (SE)		Fixed effects
		No	Yes	Type III F-test	DF	P-value
Nestling condition, Log10 g normal variable*	Mite treatment	0.263 (0.104)	0.308 (0.106)	8.53	1,89	0.004
	Carotenoid supplementation	0.251 (0.106)	0.266 (0.107)	3.04	1,89	0.085
	Mite–carotenoid interaction			0.66	1,89	0.418
Plasma carotenoid concentration, Log10 ug·L−1 normal variable	Mite treatment	0.951 (0.042)	0.974 (0.034)	0.19	1,86	0.665
	Carotenoid supplementation	0.940 (0.035)	0.997 (0.040)	1.14	1,86	0.289
	Mite–carotenoid interaction			1.15	1,86	0.286
Total white blood cell count, Poisson distributed count variable	Mite treatment	1.425 (0.099)	1.440 (0.082)	0.01	1,74	0.910
	Carotenoid supplementation	1.429 (0.089)	1.441 (0.097)	0.01	1,74	0.932
	Mite–carotenoid interaction			2.29	1,74	0.135

Model estimates are least square means controlling for clutch identity as a random effect and nestling sex and clutch size as fixed effects.

*Body mass controlling for body size by including tarsus length as a covariate.

The best growth model was the one with compensatory effects of carotenoid supplementation and O. bursa treatment (Eq. 4, see Materials and Methods), and this model gave a good fit to the data (Fig. 1). The Akaike's Information Criterion (AIC_c) value of this compensatory model was 9.5 units lower than the model with additive effects alone (Eq. 2, see Materials and Methods) and 4.8 units lower than the model with additive effects plus interactions (Eq. 3, see Materials and Methods), giving AIC weights of 0.91, 0.08 and 0.01 for the 3 models. All of the fixed effects included in the best model were statistically significant (P < 0.05) based on approximate t values (Table 3), and the AIC_c increased when any effect was removed. Under this model, carotenoid supplementation and/or O. bursa treatment is estimated to increase the growth parameters A and K by 4–6% (Table 3), resulting in an 8–9% increase in asymptotic size (Fig. 1).

Fig. 1.

Mean growth of male (black symbols) and female (white symbols) hihi nestlings in 4 treatments: O. bursa treatment and carotenoids (■), O. bursa treatment but no carotenoids (♦◇), carotenoids but no O. bursa treatment (▴▵), and neither O. bursa treatment nor carotenoids (●○). Bars accompanying the symbols are standard errors. The curves show mean projected growth under the model shown in Eq. 4, with the parameters estimates shown in Table 3. Solid lines show projected growth of males (black) or females (gray) nestlings that receive mite treatment and/or carotenoids, and dotted lines show projected growth of nestlings receiving neither mite treatment nor carotenoids.

Table 3.

Parameter estimates for the best model (Eq. 4) fitting to growth data for hihi nestlings

Parameter	Estimate	SE	t	P	L95%CL	U95%CL
A	11.013	2.396	4.600	<0.0001	6.278	15.747
βAS	1.362	0.448	3.040	0.003	0.477	2.246
βAT	−0.602	0.292	−2.060	0.041	−1.179	−0.024
K	0.260	0.011	23.250	<0.0001	0.238	0.282
βKS	0.008	0.003	2.340	0.021	0.001	0.014
βKT	−0.011	0.002	−5.460	<0.0001	−0.015	−0.007
i	7.468	0.473	15.790	<0.0001	6.533	8.402
βiS	0.380	0.091	4.180	<0.0001	0.200	0.559
b	1.503	0.138	10.870	<0.0001	1.230	1.776
s2μi	1.187	0.784	1.510	0.132	−0.363	2.737
s2e	0.010	0.000	24.820	<0.0001	0.010	0.011

Discussion

Our results clearly show the negative impacts of O. bursa on developing hihi nestlings. Survival of nestlings was compromised, as was the asymptotic mass of those individuals that reached fledging age. Similar effects of O. bursa have been reported in nestling barn swallows (Hirundo rustica), where experimental infestations resulted in a lower number of fledglings and fledglings that were of lower mass (22). This result supports the conclusions of a recent study of hihi on Mokoia Island, where the estimated effects of O. bursa infestations were influential in models predicting population viability (2). Research on barn swallows also suggests poor quality adult birds carry O. bursa and subsequently infect their nestlings with them (22, 25). The prevalence of clutches naturally infected with O. bursa in our study is very high (74% of clutches), and similar numbers of clutches had O. bursa detected in both our O. bursa removal group and in our O. bursa untreated group, indicating little potential for adult quality based differences between our O. bursa treatment groups.

We also show that maternal investment of carotenoids can offset the costs associated with parasitism with O. bursa. Our best growth model highlights a compensatory effect of either direct treatment of O. bursa or maternal deposition of carotenoids (via the egg yolk) on increased growth parameters and ultimate nestling size. Interestingly, we did not find evidence that carotenoid supplementation resulted in improved condition of nestlings at the point of fledging despite clear evidence for increased mass. This result could arise because the effects of carotenoids apply similarly to skeletal growth and overall mass. It would be interesting to further examine how nestlings choose to allocate resources available to them when confronted with costly parasitism. Altered growth priorities have been revealed in studies examining the trade offs in competing developmental demands on parasitized nestlings. For example, experimental Hippoboscid fly (Diptera) infestations on nestling barn swallows caused more rapid feather development but compromised mass increase and tarsus length (allowing nestlings to leave the nest earlier but in poorer condition) (26). Similarly, O. bursa parasitized nestling barn swallows managed to attain similar body size to parasite free individuals but had reduced mass, again suggesting priority investment allowing early fledging (22). Our results suggest carotenoids allow compensation for otherwise altered growth prioritization under parasitism.

Life history theory predicts a tradeoff in nestling development time between the energetic demand of rapid growth (that can compromise a nestling's immune system) and age-specific mortality (younger with prolonged exposure to ectoparasitism in the nest) (27, 28). It appears that parents of some bird species are able to manage this tradeoff through altered maternal and parental investment, so that productivity is maximised. In the house finch (Carpodacus mexicanus), it has been shown that females minimized costs of ectoparasitism by the nest mite Pellonyssus reedi by altering the sex of nestlings across the laying order. Specifically, males were laid later in nests that were exposed to mites, resulting in reduced rearing times for sons (whose survival is compromised more than daughters). In addition, sons in mite exposed nests hatched heavier, had faster initial growth and completed growth earlier than sons in mite-free nests (4). The above compensatory strategies of maternal investment and growth trajectories lowered mite-induced mortality (4). Our results show that parasitism retards both the rate of growth and ultimate size and survival of nestling hihi. Although maternally invested carotenoids did not reduce mite induced mortality in hihi, they did allow surviving parasitized nestlings to grow at rates similar to nonparasitized nestlings and to attain similar ultimate fledging weights.

There may be alternative forms of maternal investment that can aid nestlings in their resistance, or ability to cope, with parasites. In great tits, for example, there is evidence that maternal investment of androgens into the egg yolk can promote nestling growth even when ectoparasite hen fleas (Ceratophyllus gallinae) are present (10), and, furthermore, that androgen deposition into egg yolk mediates this coevolutionary host–parasite system by affecting the subsequent dispersal strategies of young birds (29). In the kittiwake (Rissa tridactyla), there is a positive relationship between the antibody concentrations in maternal serum against the Lyme disease agent Borrelia burgdorferi sensu lato and the same antibodies in these females' eggs and chick serum, indicating an adaptive maternal effect (11).

Given the known health benefits of carotenoids, we predicted that an increased maternal investment of these biomolecules would assist nestlings in coping with the presence of ectoparasitic mites (either directly by improving the immune system or indirectly by facilitating rapid development). We found that carotenoids invested by female hihi to her offspring, via the egg yolk, do provide a benefit when these offspring are coping with O. bursa parasitism. We note that our experimental design does not allow us to assess the mechanism whereby maternally invested carotenoids impart their benefits; for example, altered nestling begging intensities (18) or enhanced immunity (20). Our experimental design does show, however, that the growth model with the strongest support is one where maternal investment of carotenoids acted in a beneficial and compensatory manner on nestling development when those nestlings were faced with O. bursa parasitism. These results differ from a recent study which tested similar predictions in the great tit–hen flea host–parasite system (19). Although no beneficial affects of maternally invested carotenoids were detected with the presence of a parasite stressor, they did find higher mass within carotenoid supplemented nestlings in experimentally enlarged broods (19). This reported mass gain is approximately the same as we report here with hihi nestlings [8–9% increase in mass; see figure 1 in Berthouly et al. (19)]. In an associated study, Helfenstein et al. (18) also found that hen flea presence along with carotenoid supplementation increased the begging intensity of nestlings in interaction with altered brood sizes, suggesting intensified sibling competition under parasitism.

Most studies showing health benefits to nestling birds from experimentally increased access to carotenoids typically assess general immune response measures (e.g., 12, 17, 30), and few have directly tested for an enhanced ability to cope with a specific (and costly) parasite (see ref. 19). Our study is important for host–parasite disease ecology in revealing a direct relationship between a common and costly parasite and maternal strategies involving carotenoid biomolecules that can offset the negative effects of this parasitism. Although it is immensely difficult to quantify natural carotenoid availability in the diet of any wild vertebrate species, we have previously shown that natural variation in a hihi nestlings' carotenoid based pigmentation is condition dependant (21) and that carotenoid supplementation enhances the coloration of these pigmented regions (21, 31). These results, together with the current study, suggest carotenoid availability (and their interactions with stressors) are limiting in this environment. These results have direct relevance to conservation biology, not only by showing how costly ectoparasitism can be, quantitatively confirming [Armstrong et al. (2)], but how parental dietary quality (linked to habitat quality) is essential in a species' ability to cope with otherwise common and widespread parasites.

Materials and Methods

Study Site and Experimental Design.

We conducted the experiment on Tiritiri Matangi Island during the breeding season of September 2005–March 2006. Details of the study site and history of this population are published elsewhere (21). All breeding occurs in nest boxes placed throughout the forest habitat on the island, and females have up to 2 clutches per season. Conservation management of this population involves 2 main protocols: (i) provisioning of supplementary food at 4 centralized feeding stations (20% by weight sugar water, provided year round), and (ii) control of O. bursa infestations in nests and on chicks. O. bursa management involves regular checking of nests (every 2 days) and, where O. bursa are detected, treatment of nestlings with the miticide Frontline (Merial Ancare New Zealand). Where infestations are heavy, the nest is also changed to an artificial nest cup in a clean box. Nest changes are readily accepted by the breeding adults. All nest boxes are thoroughly cleaned and disinfected after each use.

We allocated each breeding pair into 1 of 2 treatment groups in a fully crossed design to simultaneously assess the impacts of O. bursa and the compensatory affects of carotenoid supplementation on nestling survival and quality. Treatment 1 involved either removing O. bursa ectoparasites (O. bursa removal clutches) (see below) or leaving natural infestations to develop (O. bursa untreated clutches). Treatment 2 involved providing normal supplementary food (i.e., 20% sugar water placed 10 m from the nest) to all pairs during nest building through to when the ultimate egg was laid. Half of these supplementary feeders were then enriched with carotenoids (i.e., 20% sugar water with lutein and zeaxanthin added, to a final concentration of 100 μg/mL) (21). The amount of time supplementary food was available was similar between groups (sugar = 6.7 ± 0.9 days; sugar + carotenoids = 7.1 ± 0.6 days), and we have previously shown that use of supplementary food does not differ between treatment groups (21). The study island has 16 distinct forest fragments, so fragments were alternately assigned as sugar or carotenoid-enriched to avoid pairs feeding on the nonassigned food supplement. Within each forest patch, pairs were then further assigned to O. bursa treatment groups in an alternating fashion. Because female age has been shown to influence reproductive success in hihi (2, 32), we tested to make sure that the mean age (years) of females did not differ between either O. bursa removal and untreated groups (least square means ± std err; treatment = 3.0 ± 0.4 versus control = 3.3 ± 0.4, F_1,62 = 0.27, P = 0.605) or between carotenoid-supplemented and sugar-water-only groups (least square means ± std err; treatment = 2.9 ± 0.4 versus control = 3.3 ± 0.4, F_1,62 = 0.82, P = 0.369).

O. bursa Monitoring and Management.

Nests were checked daily from the day before hatching was expected through to nest failure or fledging (at ≈30 days). At each visit, the nest cup, the nest box, and each nestling, was examined for O. bursa. Hands and forearms were then cleaned thoroughly with an alcohol-based gel disinfectant and disposable cloth to prevent transmission of O. bursa between nests.

In the groups where O. bursa infestations were removed, we followed the standard protocol used in hihi management. If <100 mites were seen, then the miticide Frontline was gently rubbed on each nestling with a piece of cotton wool. If there were >100 mites, then the nest was changed in addition to the treatment of nestlings. Nest changes were undertaken on only 7 (4 in each of carotenoid-enriched and sugar-water-only supplemented groups) of the 39 O. bursa removal clutches. Our results show increased survival of nestlings in O. bursa removal clutches, and increased asymptotic mass of those surviving nestlings, supporting the conclusion that artificial nests are suitable management tools in this system.

Monitoring of Eggs, Nestlings, and Juveniles.

During daily nest checks, we also recorded clutch size, date and success of hatching, nestling weights and survival, and took blood samples. Nestlings were weighed (0.01 g) at 1, 3, 6, 8, 10, 12, 15, 18, 20, and 24 days old, were banded with a unique color combination at 21 days, and were blood sampled at 24 days. Blood (≈150 μL) was collected from each chick through brachial venipuncture with 1 drop used for a smear (for white blood cell counts), and the rest was immediately centrifuged to separate plasma and hematocrit. These samples were then frozen, the plasma being subsequently used to measure carotenoid levels and the hematocrit for sexing. Any unhatched eggs were also collected 3 days after their expected hatching date, and the yolk from infertile eggs frozen for carotenoid analysis. Nest checks after 24 days were used to confirm fledging.

Laboratory Analysis.

Total carotenoid concentration in plasma and egg yolk samples was determined by HPLC as described by Ewen et al. (24). White-blood-cell counts from smears were conducted by the commercial laboratory Alpha Scientific (Hamilton). Nestlings were sexed by using a PCR-based molecular sexing technique, with DNA extracted from blood or tissue using the ammonium acetate method (http://www.nerc-molgen.org/protocols) and sex identified using fluorescently labeled Z002a and Z037b (Operon) primers (33).

Statistical Analysis.

The effects of O. bursa removal, carotenoid supplementation and first versus second clutches on clutch size (normal variable), proportion of eggs hatched (binomial variable) and survival of nestlings to fledging (binomial variable), were analyzed using generalized linear models in SAS v9.02. The effects of O. bursa removal and carotenoid supplementation on nestling condition at fledging (Log₁₀ body mass in g with tarsus length included as covariate), plasma carotenoid concentration (Log₁₀ ug·L⁻¹) and total white-blood-cell count (poisson variable), were analyzed using generalized linear mixed models in SAS v9.02. For analyses involving measurements of individual chicks from a single clutch, clutch identity was included as a random effect and sex and clutch size were included as fixed effects. We included clutch size as a covariate in the model assessing nestling condition due to the larger clutch sizes found in the O. bursa treatment group (see results). These larger clutch sizes are considered a chance artifact of experimental assignment as O. bursa treatment was applied only at the nestling stage (after clutch size was determined).

We analyzed changes in nestlings weights (W) with age (t) using a variation of the logistic function

which is a standard model for avian postnatal growth (34). Fitting this model to data involves estimating 3 parameters: A, the asymptotic size; K, the rate constant; and i, the inflection point of the curve. Preliminary analysis showed that males and females differed in all 3 parameters, that the model gave a better fit to the data if an exponent was added to the equation (as is often the case with weight data), and that it was important to allow for individual random variation in A. We then added terms to the model to test for the effects of O. bursa removal and carotenoid supplementation on A and K, giving the model

where μ_j is the individual random effect, S is the nestling's sex (1 = male, 0 = female), M is the O. bursa treatment (1 = removed, 0 = untreated), C is the food treatment (1 = carotenoids, 0 = no carotenoids), and the β terms describe the effects of sex, O. bursa removal, and carotenoids on the growth parameters. We also considered the more complex model

which allows for interactions between the effects of O. bursa removal and carotenoids. Finally, because it was plausible that these effects would be compensatory rather than additive, we also considered the simpler model

where T = 1 if carotenoids were provided or O. bursa removed, and T = 0 otherwise.

The models were fitted using Proc NLMIXED in SAS v9.1 (SAS 2004), with the log₁₀ of the observed weight modeled as a function of log₁₀ of the expected weight. We specified normal distributions for both the model and random effect, meaning the residuals and individual variation were both taken to be log-normally distributed. We compared the models based on differences in AIC_c values (Akaike's Information Criterion corrected for bias), which are calculated from the models' approximate likelihoods and their relative numbers of parameters (SAS 2004). We assessed the significance of each parameter based on the approximate t statistics provided and by the change in AIC when the parameter was removed.