Niche construction through a Goldilocks principle maximizes fitness for a nest-sharing brood parasite

Abstract

Generalist brood parasites that share nests with host nestlings can optimize resource acquisition from host parents by balancing the benefits that host nest-mates provide, including attracting increased provisions to the nest, against the costs of competing with the same host young over foster parental resources. However, it is unclear how parasitic chicks cope when faced with more nest-mates than are optimal for their survival upon hatching. We suggest that, in the obligate brood parasitic brown-headed cowbird (Molothrus ater), chicks use a niche construction strategy and reduce larger, more competitive host broods to maximize the parasites' survival to fledging. We experimentally altered brood sizes to test for Goldilocks principle patterns (i.e. a ‘just right’ intermediate brood size) of cowbird survival in nests of prothonotary warbler (Protonotaria citrea) hosts. We found that intermediate brood sizes of two host nestlings maximized cowbird fledging success relative to 0 or 4 host nest-mates at hatching. Specifically, cowbird nestlings lowered host brood sizes towards this optimum when placed in broods with more host nestlings. The results suggest that cowbirds reduce, but do not eliminate, host broods as a niche construction mechanism to improve their own probability of survival.

1. Introduction

In the early social environment of many species with dependent young, parent–offspring and sibling–sibling interactions play determining roles in offspring survival [1]. In birds, diverse factors including brood size, parental provisioning and thermoregulation of nestlings all contribute to the development, growth and survival of offspring, and are typically regulated by the genetic linkages between parents and offspring [2]. However, in the case of obligate brood parasitism, female parasites lay their eggs in the nests of other species and co-opt care from these genetically unrelated hosts [3]. Among generalist brood parasites, there is great variation across host species in the early social environment that individual parasitic offspring may be exposed to, including differences in care and competition at both the species and individual nest level. For host-specialist parasites, by contrast, this variation is confined to a single host species [4].

To combat this host variation and unpredictability, some generalist brood parasitic chicks, such as parasitic cuckoos and honeyguides, take the extreme approach of eliminating all host offspring to avoid the burdens of sharing the nest. Critically, in these species, accidental or experimentally induced competition with host nestlings reduces parasitic provisioning and fitness [5,6], and instead these parasites use exaggerated visual and auditory cues to receive enough care to survive after eliminating host nestlings [7,8]. Many other nestling brood parasites, such as cowbirds, are likely constrained from taking this level of direct action against host nestlings, as they appear unable to stimulate host parents to provide care sufficiently on their own [9,10]. Instead, these generalist brood parasites have evolved a nest-sharing strategy, whereby the brood parasitic chick allows the survival of host young to gain assistance in procuring itself more parental care than the parasite is capable of eliciting alone [11]. A cost of a nest-sharing strategy, especially for generalist brood parasites, is that it increases variation in the developmental milieu that parasitic nestlings experience through variable care duration, host nestling numbers and the intensity of begging competition, as well as the responsiveness of the host parents to begging displays of parasitic versus host chicks [12,13].

Given the sheer variability of these socio-ecological contexts, offspring of nest-sharing brood parasites are exposed to a wider breadth of early life conditions than other avian species including other brood parasites. Thus, parasitic nestlings regularly hatch into nests that are not optimal for parasitic development and survival. Certain aspects of host suitability are contingent on host choice and are not manipulable by a nestling brood parasite, including diet, host body size, incubation time and other aspects of life history related to care that may impact survival. However, aspects of the nesting environment often thought to be regulated by parent-offspring conflict, such as provisioning rate, thermoregulation and overall survival of nest-mates may be susceptible to ‘niche construction’ by nestling brood parasites. Niche construction is a Darwinian concept [14] in which an organism alters its own or others' environment for its own fitness benefit [15].

Brown-headed cowbirds (Molothrus ater; hereafter: cowbirds) lay their eggs in well over 200 host species' nests and, thus, cowbird nestlings, which share nests with host young, are left to experience a wide swath of early environmental and social variation [4]. To determine whether cowbird chicks can manipulate host brood sizes to mitigate the costs of nest-sharing, we experimentally parasitized the nests of host prothonotary warblers (Protonotaria citrea) and manipulated their brood sizes to represent the range of scenarios into which a cowbird might hatch. In doing so, we investigated how competitively diverse rearing environments affect the survival of generalist brood parasitic chicks and, in turn, whether brood parasitic chicks can mitigate this variability by causing reduction of the host brood.

Attributing brood reduction to the cowbird nestling is challenging to test empirically because it requires divorcing the effects exerted by the nestling cowbird from environmental factors (both inside and outside the nest) that may lead to host nestling death. For example, adult cowbirds may influence the number of host nest-mates their offspring has by pecking or removing host eggs [16]. Ectoparasitism by blowfly (Protocalliphora spp.) nest parasites, which feed on the blood of nestlings, may disproportionately affect either hosts or cowbirds to the apparent benefit of the other species [17].

To test for a case of niche construction experimentally, we first needed to establish whether host brood size per se influences the fitness of cowbird nestlings. We experimentally parasitized and manipulated host brood sizes of prothonotary warblers, within the range of brood sizes that cowbirds are known to experience at hatching, to determine the effect on cowbird survival [18]. Manipulations were performed at hatching to generate parasitized nests containing a cowbird with 0, 2 and 4 host nestlings.

Prothonotary warbler hosts lend themselves to a test of the niche construction hypothesis in cowbird chicks because they readily nest in nest-boxes (electronic supplementary material, figure S1) and typically produce brood sizes of 4–6 nestlings at hatching [19], generating natural variation in host brood size. Likewise, the body size of prothonotary warblers is representative of more than 80% of host species, which are smaller than brown-headed cowbirds [4]. Critically, the nest-box dwelling behaviour of prothonotary warblers also allowed us to exclude retaliatory nest destruction by adult cowbirds and ectoparasitism as causes of nestling mortality to host and parasitic nestlings in our experiment [16,20]. Likewise, the nest-box also reduced the risk of total brood failure caused by ground and aerial predators, even though cowbird begging is known to increase predation risk [21]. Nests were then monitored throughout the nestling period. Treatments were not statistically biased in either the seasonal onset of breeding attempts or the sex of the focal cowbird nestlings per treatment (electronic supplementary material, table S1).

2. Methods

(a) . Study system

Nests of prothonotary warblers were experimentally manipulated during the 2019–2021 breeding seasons within the Cypress Creek National Wildlife Refuge and Cache River State Natural Area in the Cache River watershed of southern Illinois, USA. The prothonotary warbler is a neotropical migratory songbird and obligate secondary cavity nester that breeds in central and eastern North America. This species preferentially nests in natural or artificial cavities above standing water in habitats such as bottomland and swamp forests. Prior to the 2019 breeding season, 200 nest-boxes were set up across a 150 hectare (ha) site of connected wetland forest with an additional 100 nest-boxes being added to the system prior to the 2020 breeding season. Prothonotary warblers readily breed in nest-boxes, often preferring them over natural cavities [19].

Our goal with this experiment was to form rearing environments where only internal nest dynamics of nest-mate–nest-mate and parent–offspring interactions could determine survival differences for both cowbird and warbler nestlings. We focused on cowbird parasitism in nest-boxes because it was not possible to predator-proof natural cavities to the same degree. Rates of parasitism by brown-headed cowbirds across these 3 breeding seasons were as high as 60% annually in the nest-boxes. Adult cowbirds (35–45 g) are much larger than the adult warblers (14–16 g), and even at day 4 post-hatch, cowbird nestlings are ∼2.3× the size of warbler nest-mates (ND Antonson 2021, unpublished data). In nests with day 8–9 warblers, parasitized nests typically receive ∼1.75× the feeds provisioned to unparasitized nests, demonstrating an increased demand for provisioning [19]. Prothonotary warblers are acceptors of cowbird eggs and nestlings and apparently lack any adaptive defences against parasitism after the egg is laid [15,22].

(b) . Nest-box design and predator-proofing

We used nest-boxes for this study. They have been described previously and consist of cleaned wax-coated cardboard drink cartons [19]. Two aluminium conduit poles driven at a shallow angle into the ground formed the support for the nest-boxes. The nest-boxes were then strapped to the poles with filament tape below the point where the poles crossed (electronic supplementary material, figure S1a). Cowbird guards (electronic supplementary material, figure S1b), consisting of a flat piece of cardboard that could be inserted into the nest-box to change the diameter of the entrance from 44 mm to 32 mm, were added to the nest-boxes during incubation to prevent the mafia behaviour known in this cowbird population [16]. They were removed on day 9 of the nestling period to allow surviving cowbird chicks to fledge.

Anti-raptor guards, consisting of green garden wire formed into a domed shape (electronic supplementary material, figure S1a), were also attached on the day of cowbird hatching. They prevented attack by aerial predators which typically depredate the nests by removing the tops of the nest-boxes or by perching on top and reaching into the entrance to remove nestlings. The raptor guards have slotted holes that warblers and cowbirds can easily navigate in flight when returning to the nest. Finally, at the beginning of each breeding season, multi-purpose grease was applied to all nest-box poles below the level of the box to prevent ground predation by small mammals and snakes. The previous season's grease was removed using paint scrapers. Finally, we changed nesting material on the 4th day after hatching to prevent chick mortality due to blow fly larvae ectoparasitism [20].

Together, these measures meant that any brood reduction we measured could be attributed to the dynamics of host–brood parasite interactions themselves [17]. Any nest where it was not possible to change nesting material due to inclement weather or flooding was excluded from this study.

(c) . Assignment of treatments

To separate the effects of a cowbird from the effects of brood sizes on host chick mortality, we varied brood size treatments in both experimental parasitized and unparasitized nests. Nest-boxes were experimentally parasitized by adding a cowbird egg that was laid in a different prothonotary warbler nest when the focal nest contained 3 or 4 warbler eggs. Prothonotary warblers typically lay five eggs, and incubation times for cowbirds (10–11 days) and prothonotary warblers (11–12 days) are similar but the cowbird typically hatches one day before the warblers in naturally parasitized nests [19]. On the day of warblers' hatching, treatments were randomly generated by fostering newly hatched warbler nestlings out to the nearest nest-box with similarly aged (1–2 days) nestlings and never producing brood sizes of more than six warblers. Nests where warblers were added were subsequently excluded from this study. Unparasitized nests used in this study contained only natal host young and no foreign chicks. Experimental parasitism resulted in three treatments: (1) a cowbird with no host nest-mates – n = 27, (2) a cowbird with two warbler nest-mates – n = 15 or (3) a cowbird with four warbler nest-mates – n = 28. We also generated unparasitized nests of 1 (n = 7), 3 (n = 10) or 5 (n = 10) warbler hatchlings following the same protocol, thereby enabling a direct comparison between warblers and cowbirds experiencing the same developmental environment to disentangle the effects of brood size on brood reduction, and other traits, from those of the cowbird.

(d) . Nest and video monitoring

Nest-boxes were monitored daily to the extent it was logistically possible, or every other day when external factors such as weather and flooding prevented a daily nest check. Daily nest checks consisted of removing the lid from the top of the nest-box and simply recording the number of host nestlings still alive and whether the cowbird was alive. On day 3 post-hatching for the cowbird, small infrared video cameras were placed on top of the nest-boxes where the drink carton's cap could be unscrewed, and cameras were held in place with adhesive poster putty (Duck Brand). Day 3 post-hatch was chosen for measuring provisioning because we wanted to choose a time before much brood reduction could occur. The camera apparatus consisted of a 12 V rechargeable sealed lead acid battery (Mighty Max) sitting in a waterproof container (Attwood) attached to a digital video recorder (DVR) that recorded footage from the cameras to a 128 GB SD card. Video data were collected for 1 h and 15 min from the time the nest-box was departed by NDA. Filming took place only between 08.00 and 11.00 h to standardize potential variation in provisioning that may occur over the course of a day [23]. After filming, cameras were removed, and the cap was screwed back into place on top of the nest-box.

Brood size treatments were based on the natural range of brood sizes experienced by cowbirds in the nests of prothonotary warblers. It should additionally be noted that cowbirds do not typically remove a host egg prior to laying in our specific host–parasite system [19], so parasitized brood sizes with six nestlings (five warblers and one cowbird) are relatively common. Parasitism on top of a full brood, however, may cause additional physiological strain on the host parents [24] and could contribute additional variation that may obscure results when investigating the interactions between brood parasitism and total brood size. Thus, we opted to use host brood sizes up to 4 warbler nest-mates + 1 cowbird since five warbler nestlings is the modal brood size for prothonotary warblers [19].

Warbler nest-mates of cowbirds were never cross-fostered into treatments and were fostered out to the nearest age-matched nest-box, so experimental broods were always the same size, or smaller, than the original brood size and only comprised of natal offspring. Rates of conspecific parasitism and extra-pair young are low among prothonotary warblers in southern Illinois [25]. All nest manipulations were approved by the University of Illinois at Urbana-Champaign Institutional Animal Care and Use Committee (IACUC #19032).

(e) . Video analysis

The first 15 min of video footage were not analysed due to potential lingering effects of human interference at the nest-boxes, though adult warblers still readily provisioned within this timeframe. Provisioning on day 3 post-hatch was scored as the number of prey items brought to the nest by parents and fed to either the cowbird or the warbler nestlings during 1 h of observation. Because the camera was mounted from above the nest, and the parents leaned downwards to feed the chicks, it was not possible to assess the size and taxa of the prey items, relative to the parental beak length, as has been done in other similar studies of mixed host–parasite broods (e.g. [26], who recorded footage laterally at the level of the nest cup). These videos were analysed both by NDA as well as undergraduate students at the University of Illinois at Urbana-Champaign. Interobserver correlations were high for both cowbird provisioning (r = 0.941, p < 0.001) and warbler provisioning (r = 0.976, p < 0.001). Brooding time was also assessed during these day 3 provisioning videos. Brooding time was scored as the total amount of time the adult (always the female) warbler spent stationary and covering the nestlings in the nest-box over the course of the same hour as provisioning.

(f) . Statistical analysis

All analyses were conducted in R [27] using the annotated script included in the open access material that provides details of all packages and datasets used. In comparisons of parasitized and unparasitized nests, two-factor analyses were used to separate the effects of brood size from the effect of a nestling brood parasite. A small number of broods (n = 13) were excluded where total nest failure, defined as the death of the cowbird and all warblers, occurred, and the resulting sample sizes are indicated in figure 1. Several nests at each brood size were excluded from the final survival analysis because of additional experimental manipulations that occurred on day 4 post-hatching. These were chosen at random and are denoted in the publicly available dataset.

Figure 1.

Experimental survival of single cowbirds in the nests of prothonotary warblers at different hatching brood sizes (n = 27 for the 0 host nestling treatment, n = 15 for the 2 host nestling treatment and n = 28 for the 4 host nestling treatment). Circles represent individual data points with the quadratic regression line shown in purple and 95% CI in grey shading. (Online version in colour.)

To assess the relationship between cowbird survival during nestling period and experimental brood size at hatching, a polynomial quadratic regression with a binomial response variable (1 = cowbird survived to fledging, 0 = cowbird died prior to fledging) was used to fit the data. Cox proportional-hazard models were used to perform high-resolution survival analyses and to determine when risk was highest in each of the brood sizes. Treatment bias on brood size across ordinal date and nestling sex was assessed using multinomial logistic regression and a two-tailed z-test. Day 3 post-hatch provisioning, brood reduction and brooding time across brood sizes were assessed with ANOVAs and Tukey post hoc tests were used to evaluate multiple comparisons across brood sizes. Brood failure between parasitism treatments with 2 and 4 host nestlings was assessed with a logistic regression (1 = no nestling survived, 0 = at least one nestling survived).

3. Results

We found that brown-headed cowbirds in the nests of prothonotary warblers appear to conform to a Goldilocks principle [28], with the greatest cowbird survival predicted in this host's nests with an intermediate brood size of 2.2 host nest-mates. Reduced cowbird survival was detected in broods with 0 or 4 host nestlings relative to broods with two host nestlings (figure 1; quadratic binomial regression: R² = 0.293, z = −3.488, p < 0.001). When the timing of the cowbird's mortality risk was compared using Cox proportional-hazard models, we found that cowbird nestlings died significantly earlier in post-hatch development when raised alone (median age at death = 3 days; 95% CI = 2, 6 days) than when raised in nests with 2 host nestlings (figure 2a; Hazard Ratio (HR) = 2.694, z = 3.63, p < 0.001). Cowbirds raised with four host nestlings (median = 7 days; 95% CI = 5 days, 10 days—i.e. survived to fledging) were also more likely to die than cowbirds raised with two nestlings (figure 2a; HR = 1.713, z = 2.28, p = 0.023). Notably, sole cowbird nestlings also had lower survival than experimentally matched nests with a sole prothonotary warbler nestling in the nest (electronic supplementary material, figure S2; HR = −1.576, z = −2.55, p = 0.011).

Figure 2.

(a) Cox proportional-hazard models showing survival across the first 8 days of the post-hatching period for the cowbird nestling. Lines represent proportion of nestlings in each treatment alive on a given day with 95% CIs. (b) Hazard tables denoting number of cowbird nestlings at risk and (c) the number of cumulative death events for cowbirds across three parasitized treatments. Note: there was not enough variation present in the data for the two host nestling treatment to resolve the median and 95% CI for age of death because cowbird nestlings survived to fledging at such a high rate. (Online version in colour.)

We next assessed whether cowbirds employed a niche construction strategy of brood reduction to optimize host brood sizes to their benefit. To do so, we quantified nestling warbler mortality across both of our two parasitized brood treatments (2 or 4 warbler nestling broods + 1 cowbird) and our experimentally generated broods of unparasitized treatments, which contained the same total number of nestlings (3 or 5 warblers). This enabled us to disentangle the effects of the cowbird on host nestling mortality from the effects of brood size alone (e.g. [21]).

Consistent with a niche construction hypothesis for brood reduction, we found that the presence of a cowbird nestling significantly increased the likelihood of host brood reduction (F_1,50 = 7.022, p = 0.011). Brood size at hatching was also positively related to the chance of brood reduction (F_1,50 = 22.593, p < 0.001), and host young in larger broods were even more likely to die when the cowbird nestling was present as did the interaction between brood size and cowbird nestling presence (figure 3; brood size×parasitism interaction: F_1,50 = 5.827, p = 0.019). Specifically, cowbirds reared with four host nestlings reduced host brood sizes by a mean of 1.95 ± 0.23 warblers, whereas 0.14 ± 0.28 warbler nestlings died when cowbirds were raised with two host nestlings (figure 3; Tukey post hoc test t = −4.965, p < 0.001). Again, strikingly, this effect was isolated to broods including cowbirds, as experimentally generated unparasitized broods of both 3 and 5 warbler chicks experienced similarly nestling mortality, with 3-chick warbler broods being reduced by 0.10 ± 0.33 and 5-chick warbler broods being reduced by 0.44 ± 0.35 (post hoc t = −0.710, p = 0.893), akin to parasitized broods with two warbler nestlings (figure 3).

Figure 3.

Brood reduction occurred significantly more in parasitized nests with four host nestlings plus a cowbird (n = 21) than in unparasitized broods of the same size with five warbler nestlings (n = 10), and parasitized broods with two host nestlings and a cowbird (n = 14), while there was no statistical difference between parasitized and unparasitized broods with 3 (n = 10) or 5 (n = 10) nestlings. Mean ± SE shown with statistically significant post hoc differences denoted by different letters. (Online version in colour.)

Median age of death for the first warbler to die in the four host nestling parasitized treatment was 3 days after hatching (95% CI = 2 days, 10 days, i.e. survived to fledging), while the median age of death for the second warbler to die was 7 days after hatching (95% CI = 4 days, 10 days) (electronic supplementary material, figure S3). Focusing on the first warblers to die in each nest, parasitized nests of four warblers and a cowbird had a higher mortality risk than unparasitized broods of five warblers (HR = 1.491, z = 2.36, p = 0.019; electronic supplementary material, figure S3). The risk of a second warbler dying was not significantly greater in parasitized broods of five compared to unparasitized ones (HR = 1.419, z = 1.87, p = 0.062; electronic supplementary material, figure S4), but the direction of the pattern suggests a shared impact of brood size and brood parasitism.

When comparing provisions delivered to the cowbird, day 3 cowbirds reared with two host nest-mates received significantly more prey items (figure 4a; ANOVA: F_2,47 = 5.391, p = 0.008) than cowbirds with either zero host nest-mates (Tukey post hoc t-test = −3.254, p = 0.006) or four host nest-mates (t = 2.485, p = 0.043) as determined by Tukey post hoc comparisons. Single cowbird and single warbler nestlings elicited almost identical provisioning rates (post hoc t = −0.294, p = 0.999; electronic supplementary material, figure S5).

Figure 4.

(a) Mean provisioning to the cowbird in each brood treatment complemented the quadratic survival (Goldilocks) pattern demonstrated with cowbird nest survival. (b,c) Mean provisioning to the least fed and second least fed warbler nestlings in each treatment was greater in smaller brood sizes regardless of parasitism status. (b,c) are shown on the same scale. Mean ± SE shown with unique letters denoting significant differences in post hoc comparisons, where present. Individual points represent individual samples. (Online version in colour.)

We carried out a similar analysis for warbler chicks, focusing on provisioning to the least fed and second least fed warblers in each nest, because we reasoned that these chicks were the most likely to die. Mean provisioning for the least fed warbler, in nests with multiple chicks, was not significantly different between parasitized and unparasitized nests (F_1,44 = 0.027, p = 0.869; figure 4b). However, the rate at which this nestling was fed decreased with increasing brood size (F_1,50 = 5.78, p = 0.019). The second least fed warbler in each nest followed a similar pattern, with no effect of parasitism on mean provisioning (F_1,47 = 1.423, p = 0.238; figure 4c), but with a significantly negative effect of hatched brood size on the rate at which it was provisioned (F_1,50 = 5.037, p = 0.029).

Filming also allowed the assessment of brooding time by female warblers as a metric of early thermoregulation. Female warblers spent significantly more time brooding when they raised fewer chicks (ANOVA: F_2,72 = 31.877, p < 0.001). This was not dependent on the presence of a cowbird nestling (F_1,72 = 0.448, p = 0.505) though there was a significant interaction between cowbird presence and host brood size (F_2,72 = 10.395, p < 0.001; figure 5). Specifically, lone cowbird nests were brooding for significantly more time than broods of a cowbird and two host nest-mates (Tukey post hoc t = −4.295, p = 0.001), but nests with a cowbird and either 2 or 4 host nest-mates received similar amounts of brooding time (post hoc t = −2.302, p = 0.208). In unparasitized nests, single warbler nests received significantly more brooding time than nests with three warblers (post hoc t = 6.975, p < 0.001) or five warblers (post hoc t = 7.373, p < 0.001). There was no significant difference in brooding times between parasitized and unparasitized nests of equal brood sizes, except that broods with a single warbler received even more thermoregulation than broods with a single cowbird (post hoc t = −4.011, p = 0.002; figure 5).

Figure 5.

Mean amount of time (seconds/hour) that female warblers spent brooding chicks on 3 days post-hatch was significantly higher in nests with a single chick compared to most nest treatments with multiple chicks. Mean value ± SE are represented with unique letters representing significant post hoc differences and individual points represent samples. (Online version in colour.)

4. Discussion

The brood reduction strategy of parasitic cowbird nestlings detected here experimentally appears to represent an adaptive niche construction strategy since it alters the social environment experienced by nestling parasites and results in an outcome that maximizes the parasite's survival [15]. Here, we have shown that the survival probability for nestling cowbirds is highest in broods with two host nestlings and lower when alone or with four host nestlings. We have also demonstrated that cowbirds in broods with four host nestlings reduce host broods toward the optimal brood size of two host nestlings, while cowbirds with two host nestlings along with unparasitized warbler broods of 3 and 5 nestlings do not experience significant brood reduction. This suggests that brood reduction is not simply a function of brood size but rather caused by the presence of a brood parasite in large brood sizes.

An often-made claim about nest-sharing brood parasitic nestlings, such as cowbirds, is that they do not survive in the absence of host nestlings because they are unable to stimulate the parents to provision without the aid of host nestlings [9,10,23,29]. However, few studies have compared the stimulatory abilities of single brood parasite nests to those containing a single host nestling. Unlike the common cuckoo (Cuculus canorus), with its supernormal calling behaviour, a cowbird will probably have low survival prospects if it causes the death of all host young and no longer has social help to motivate parental provisions (electronic supplementary material, figures S2 and S5, see also [10,23]). However, our provisioning and survival data from sole nestling broods demonstrate that single cowbird nestlings were provisioned at the same rate as single warbler nestlings (figure 4), directly confirming that cowbirds do not represent a supernormal stimulus for foster parents of this host species. As such, survival is probably lowest in this treatment because a cowbird's growth cannot be sustained by an equivalent level of provisioning that a single warbler nestling attracts, due to the parasite's greater mass and faster growth rate [30]. Coupled with reduced host maternal investment in brooding time (see Results), it is likely that cowbirds raised alone died due to a combination of insufficient early provisioning and inadequate thermoregulation, while cowbirds raised with 4 host nestlings die due to insufficient provisioning.

The killing of all host nestlings is therefore not a viable strategy for cowbird survival and partial brood reduction, and instead represents an alternative pathway by which a nest-sharing brood parasite may maximize parasitic feeding and survival. Given the results of our survival analyses, future studies should consider how the timing of brood reduction by the cowbird in larger broods affects their survival, as it is likely that the extent of brood reduction by cowbirds needs to be finely timed and balanced to not risk losing host aid in stimulating parents to provide sufficient care.

How might cowbirds cause brood reduction that leads to an optimal niche? Reduced provisioning to the least fed and second least fed warblers does not appear to be a mechanism for the brood reduction we detected. On the other hand, our focal provisioning data provides only a detailed snapshot for day 3 after the cowbird's hatching, whereas brood reduction occurred evenly throughout the nestling period (electronic supplementary material, figures S3 and S4). Perhaps if we had measured daily provisioning, as well as growth, over the entirety of nestling development, we might have detected a protracted reduction in the warbler chicks' provisioning in large broods, cumulating in starvation before the end of the nestling period.

Clearly, the mechanism by which cowbirds cause host brood reduction requires further study. For example, nestling parasites may hijack parental provisioning rules in the nests of species where the largest nestling is preferentially fed. These provisioning rules vary between species and between brood sizes intraspecifically [31–33]. It is also possible that, in other host species, niche construction toward an optimal brood size for the cowbird is the result of strategic egg destruction (for example, in adult shiny cowbirds (Molothrus bonariensis) [34]) though any synergistic action between the efforts of adult and nestling cowbirds remains to be investigated. To further elucidate the mechanism of niche construction, it would be interesting to determine brood sizes that are optimal from the cowbird's perspective for its other host species.

We expect that our estimates for optimal cowbird survival are conservatively high due to the experimental exclusion of predation. Predation in prothonotary warbler nests almost always leads to the death of all nestlings, which would have caused the exclusion of a greater number of nests. Predation is likely to be higher in large brood sizes due to more intense begging behaviours [21], but this would not have changed the increased mortality of cowbirds with four host nestlings when compared to cowbirds reared with two host nestlings. Finally, it is possible that blow fly parasitism may also affect host–brood parasite dynamics but this remains to be determined in future work.

Our results have broad implications for understanding the evolution of generalist brood parasitic strategies among birds and other social parasites [35]. While many factors of host suitability are certainly contingent on host choice by adult brood parasites and are not manipulable by a nestling brood parasite, young cowbirds appear to reduce host broods when hatching into brood sizes too large for their optimal survival and do not appear to do so when in brood sizes optimal for their survival. This demonstrates that host manipulation by nestling parasites extends beyond direct acts of physical aggression and eviction to more subtle and indirect forms of social niche construction that benefit the selfish interests of the parasite [18].

Ethics

All research conducted as part of this study was approved by the University of Illinois at Urbana-Champaign Institutional Animal Care and Use Committee (IACUC #19032) as well as state and federal permitting.

Data accessibility

The data are available at these links: Experimental Data (doi:10.6084/m9.figshare.19230060) [36], R Code (doi:10.6084/m9.figshare.19319846) [37].

Electronic supplementary material is available online [38].

Authors' contributions

N.D.A.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, visualization, writing—original draft, writing—review and editing; W.M.S.: funding acquisition, investigation, methodology, project administration, resources, supervision, writing—review and editing; D.T.: investigation, writing—review and editing; R.M.K.: investigation, writing—review and editing; M.E.H.: conceptualization, data curation, funding acquisition, investigation, methodology, project administration, resources, supervision, 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 an Animal Behaviour Society student research grant, University of Illinois Clark research grants, University of Illinois Dissertation Travel Grant and a DOE GAANN fellowship to N.D.A., as well as a grant from the National Science Foundation (IOS #1953226) and the Harley Jones Van Cleave Professorship to M.E.H. Additional support was provided by the Humboldt Foundation, Germany [to M.E.H.].