Individual Optimization of Reproductive Investment and the Cost of Incubation in a Wild Songbird

Despite avid interest in life history trade-offs and the costs of reproduction, evidence that increased parental allocation reduces subsequent breeding productivity is mixed. This uncertainty may be attributable to environmental heterogeneity in space and time, necessitating experiments across a range of ecological contexts. Over three breeding seasons, we cross-fostered clutches between nests to manipulate incubation duration in a wild population of Carolina wrens, a species in which only females incubate, to test for a cost of incubation on current and future reproduction. Prolonged incubation affected maternal productivity in a manner dependent on the current environment and initial investment in eggs, suggesting that incubation is optimized according to other components of reproduction and individual quality. Effects of incubation duration on foster nestling condition varied between years, being costly in one, beneficial in another, and neutral in the third. The proportion of young fledged, females’ probability of breeding again within seasons, and subsequent clutch sizes all declined with increasing incubation effort—effects that became more pronounced as seasons progressed. Therefore, costs of incubation were almost entirely dependent on maternal quality and environmental variation, illustrating the importance of conducting experiments across a range of environmental settings for understanding the costs of reproduction and evolution of life histories.


Introduction
Understanding individual differences in the allocation of resources to competing life history demands continues to be a principal focus of theoreticians and empiricists (Williams 1966;Zera and Harshman 2001;Fowler and Williams 2017).In iteroparous species, fitness maximization is thought to require that parents temper investment in current young in order to reproduce again (Stearns 1992;Royle et al. 2012).Thus, it is generally accepted that increased parental investment constrains future reproduction and induces trade-offs when caring for altricial young, where a host of factors differ between individuals, breeding attempts, and seasons, with the potential to exacerbate or ameliorate trade-offs (Stearns 1992).
In birds, producing eggs and rearing altricial young are expected to entail greater energy expenditure than maintaining embryonic development, so the costs of producing eggs and provisioning offspring have historically been more obvious research foci than costs of tending or incubating eggs (Reid et al. 2002;Nord and Williams 2015;Williams 2015;Miller and Bowers 2021).Recent research, however, reveals nontrivial energetic demands associated with incubation (Ardia et al. 2009;Nord and Williams 2015), which may impose fitness costs within and between cycles by ultimately reducing a parent's ability to raise current or future young (Heaney and Monaghan 1996;Reid et al. 2000aReid et al. , 2000b;;Martin 2002;Remeš 2007).Nonetheless, recent studies investigating costs of incubation yield mixed results (Dobbs et al. 2006;Lothery et al. 2014;Williams 2018).For example, despite costs of incubating enlarged clutches of eggs (Heaney and Monaghan 1996;Cichoń 2000;Reid et al. 2000a;Ilmonen et al. 2002), when Sakaluk et al. (2018) manipulated incubation duration while holding clutch size constant, females with prolonged incubation periods neither suffered fitness costs nor were their offspring in below-average condition, a common predictor of postfledging survival and recruitment into breeding populations (Tinbergen and Boerlijst 1990;Both et al. 1999;Bowers et al. 2014).
Fitness costs of incubation may depend on specific life history strategies.For example, eggs in open nests may lose heat more rapidly than those in cavities that maintain comparably stable temperatures (Hilton et al. 2004;Martin et al. 2017).Cycles of cooling and rewarming eggs may affect parents, as rewarming a clutch of cold eggs can be more demanding than maintaining steady temperatures (Biebach 1986), or developing young by decreasing hatching success, nestling condition, or survival to fledging (Wiebe and Martin 1997;Reid et al. 2002;Nord and Nilsson 2011;DuRant et al. 2013;Mueller et al. 2019a).Parents may alter incubation demands by nesting earlier or later within seasons, building a more or less well-insulated nest, or varying incubation constancy and time spent off the nest.However, a variety of other factors may shape costs of incubation, including individual quality, environmental conditions, and social interactions (Fitzpatrick 1996;Creswell et al. 2003;Parker et al. 2014).
Theoretically, fitness maximization requires optimization of reproductive effort (Stearns 1992) from egg production and incubation through offspring independence.Thus, if costs of incubation depend on other components of maternal investment and quality, then individual optimization of reproductive strategies, including those associated with parental condition or resource acquisition (van Noordwijk and de Jong 1986; Pettifor et al. 1988;Vézina et al. 2006;Ardia and Clotfelter 2007), may contribute to the lack of consensus on the costs of incubation.For example, within-population variation in income versus capital breeding may occur state dependently within species (Jaatinen et al. 2016), where females in better condition utilize a larger amount of stored resources while breeding than females in poorer condition.Thus, experiments altering incubation effort across a wide range of ecological conditions are needed.
We investigated fitness costs of incubation effort by experimentally altering these demands in a wild population of Carolina wrens (Thryothorus ludovicianus).To do so, we swapped clutches of eggs between nests that differed slightly in clutch initiation dates, thereby manipulating the energetic demands of incubation by changing its duration as opposed to other methods (e.g., adjusting food availability, clutch size, nest microclimate, or parental condition) that can alter variables other than incubation effort per se.In this species, only females incubate, and because maternal investment into all components of the reproductive cycle (including egg production) may be individually optimized, we cross-fostered clutches with the same number of eggs but that may or may not have differed in egg size, allowing us also to test whether costs of incubation depend on other components of maternal quality and investment.
We assessed consequences of the manipulated incubation period in a female's current brood under the position that increased incubation effort may create suboptimal incubation temperatures that alter nestling posthatching development and/or reduce females' ability to rear offspring after hatching.We predicted that females with prolonged incubation periods would fledge a lower proportion of the young they reared and that those young surviving to fledging would attain reduced prefledging mass.Given that life history trade-offs and reproductive costs may depend on environmental context and other components of parental investment (Stearns 1992;Nord and Nilsson 2012), we also predicted that any costs of incubation may depend on seasonal or annual variation in environmental conditions and between-female variation in initial allocation to eggs.We then investigated costs of incubation for future breeding productivity-namely, females' (i) probability of nesting again within seasons, (ii) clutch size in subsequent nests, (iii) body mass at subsequent nests, and (iv) probability of breeding the following year.If incubation is costly to subsequent reproduction, we predicted that greater incubation effort in a female's current brood would reduce the amount of time or energy she is able to allocate to self-maintenance or to subsequent broods within or between seasons.

Study Species and Site
Carolina wrens are insectivorous songbirds that are yearround residents throughout eastern North America and readily accept nest boxes for breeding (Mueller et al. 2019b).Nests typically have a dome and base composed of dried leaves, moss, oak catkins, and other soft materials; both sexes contribute to nest construction, which females complete by lining the nest cup with grasses or other fine materials before laying eggs (Haggerty and Morton 2020).Females lay one egg per day and typically begin incubation after laying the penultimate egg of the clutch (mean clutch size5SD p 4:55:7, min p 3, max p 6; Haggerty and Morton 2020).Only the female incubates eggs and broods nestlings, but both the male and female contribute to nestling provisioning and nest defense (Mueller et al. 2019a;Haggerty and Morton 2020).At our study site, egg production begins in mid-March and continues through early August.Eggs typically hatch after 13-18 days of incubation, and young fledge from the nest 11-16 days after hatching (Mueller et al. 2019a;Haggerty and Morton 2020).Manipulations were conducted from 2018 to 2020 at the Edward J. Meeman Biological Station in west Tennessee (35.3637N, 90.0177W),where a grid of nest boxes, each resting above aluminum predator baffles (51 cm in diameter), is distributed over ~170 ha of mature, secondary deciduous forest (for further detail, see Mueller et al. 2019b;Brode et al. 2021).

General Data Collection
Throughout each season, we checked nest boxes at least once every 5 days for the formation of new nests.Nearly Reproductive Costs of Incubation 255 all clutches were identified prior to the onset of incubation.We visited nests more frequently after clutch completion to monitor their status and capture, band, and measure adults midway through incubation and young on day 9 posthatching.Each individual was fitted with a uniquely numbered US Geological Survey aluminum band, and adults received three additional colored bands (two bands per leg) arranged in unique combinations for subsequent visual identification.Body mass (50.1 g) was measured with a digital scale.Starting day 11 posthatching, nests were monitored daily until fledging.We continued to monitor nesting activity and identify adult breeders throughout subsequent breeding seasons to determine effects of the manipulation on females' propensity to breed in the population in future years.

Experimental Manipulation
We manipulated incubation periods by swapping, early to midway through incubation, whole clutches of eggs between nests (N p 118; replacing real eggs with dummy eggs during the swap) matched according to clutch size, with a 0-3-day difference in clutch initiation date (ordinal date; 2018: N p 25 nests; 2019: N p 39 nests; 2020: N p 54 nests; fig.S1; figs.S1-S3 are available online).Thus, experimental females incubated for either a shorter duration (if her foster clutch was produced before her own) or a longer duration (if her foster clutch was produced after her own) than expected, with clutches initiated on the same day serving as controls (fig.S1).While these differences in when eggs were produced could have introduced an artifact of environmental differences, obscuring effects of our manipulation, this is true only when the parental ability to rear offspring declines linearly over the course of the breeding season (Verhulst and Nilsson 2008), which was not the case here, as neither the mass nor the proportion of young fledged from nondepredated broods (i.e., a reflection of parents' full ability to rear offspring) was directly affected by time of season (table S1; tables S1-S5 are available online).Moreover, the small differences between nests in the day on which eggs were swapped relative to incubation onset also had no effect on prefledging mass or fledging success (table S2), and there were no differences between the cross-fostered control nests in our experiment and unmanipulated wren nests at our study site during this time (table S3).Thus, our cross-fostering approach could not have introduced a bias to our results, and we are confident that the effects of our manipulation are a direct consequence of incubation duration.We also quantified egg size during swaps using a digital thickness gauge (Mitutoyo 700-122), measuring egg length and breadth (50.01 mm) to estimate egg volume (egg shape is generally similar between females; Hoyt 1979) and then averaging over the clutch.

Data Analysis
All analyses were performed in SAS (ver.9.4), and all tests were two tailed (a p :05).We centered and standardized data (mean5SD p 051) before analysis so that parameter estimates reflect effect sizes (akin to r correlation coefficients) and results of main effects are interpretable even when part of an interaction (Schielzeth 2010).We used linear and generalized (for binary outcomes) linear mixed models that included maternal ID as a random effect as needed to account for the nonindependence of multiple observations of the same female.
We first assessed interannual variation (always as a categorical effect) in ambient temperature and rainfall (obtained from National Oceanographic and Atmospheric Association station USW00013893) during the breeding season in addition to measures of breeding activity (breeding date, clutch size, egg size) that may reflect differences in resource abundance or environmental conditions between years (Christians 2002;Ruffino et al. 2014;Heming and Marini 2015).To test whether years differed in their distributions of breeding dates or whether the duration of any season was more truncated or prolonged than the others, we conducted Levene's test for equality of variances in clutch initiation dates in relation to year.We then tested for differences in clutch and egg size in relation to year and breeding date using linear mixed models.
We assessed the effect of our manipulation (incubation treatment) on the duration of incubation (from the day a female laid the final egg in her clutch until the day foster eggs hatched) using a linear mixed model.The manipulation by itself had a strong effect on incubation periods (see "Results"), but given the (i) tendency for ambient temperatures to increase over the course of a breeding season (Mueller et al. 2019a), (ii) thermal inertia of eggs (more eggs and larger eggs expected to cool less rapidly during off bouts; Williams 1996;Dobbs et al. 2006), and (iii) contribution of female size and brood patch area to incubation efficiency (larger females better able to warm all eggs of the clutch; Monaghan and Nager 1997;Hanssen et al. 2005), we examined the effect of our treatment while also including (i) females' clutch initiation date, (ii) year, (iii) clutch size (analogous to the number of eggs incubated), (iv) foster egg size (the size of eggs females incubated until hatching), and (v) female mass as covariates.
We examined potential costs of incubation for females' current breeding attempts by analyzing (i) prefledging mass of foster nestlings and (ii) fledging success.Nestling prefledging mass was analyzed using a linear mixed model assuming normally distributed error.For fledging success, we used "events/trials" syntax (number fledged/number of eggs incubated) in a generalized linear mixed model with a binomial distribution.We included the incubation treatment as a continuous independent variable (23 to 3; using the actual number of days females spent incubating yields similar results), along with females' clutch initiation date, year, and two-way interactions between these effects and treatment to test for context-dependent costs.Given the potential effects of egg size and sibling rivalry on nestling growth (Williams 1994;Krist 2011), we also included the size of eggs from which nestlings hatched and brood size on day 4 posthatching (corresponding to a period of rapid growth) as covariates in our analyses of nestling mass.We included a female's initial egg size as a covariate in our analysis of fledging success under the position that initial investment in eggs may shape incubation costs and that females investing more into their clutch should be expected to fledge a larger portion of their young.Whether egg size or brood size is included or excluded from any of these models does not influence the significance of effects of incubation treatment.We subsequently removed nonsignificant (P 1 :1) treatment#year and treatment#date interactions to simplify models.Although this process of simplifying models can affect type I error rates (Forstmeier and Schielzeth 2011), the effects we observed in simplified models were still apparent and qualitatively similar to those detected from full models (available in table S4).
We then evaluated consequences of the incubation treatment for future breeding attempts by analyzing (i) whether females bred again within seasons as a binary outcome and, if they nested again, their (ii) body mass and (iii) the number of eggs produced in their subsequent brood.Whether current nests were or were not successful did not influence any of these variables (all P 1 :25), so we included all nests in these analyses while controlling for the number fledged from a female's current brood to avoid analyzing a nonrandom sample consisting of only successful females (restricting analyses to only successful females yields similar results).Finally, we analyzed females' probability of breeding again on the study site the next year; however, a series of severe winter storms before the 2021 season extirpated a substantial portion of the study population (no females survived to breed in 2021), so we had to omit 2020 nests (zero variance).For each test, we included the incubation treatment as a continuous independent variable, along with (i) clutch initiation date, (ii) year, and (iii) the number of young fledged.We also included two-way interactions between treatment and year and between treatment and clutch initiation date, as described above for current broods (full models in table S5).

Effects on Incubation Duration
The manipulation worked as expected, as incubation duration increased linearly across treatments (fig.1), where females tending clutches produced before and after their own incubated for shorter and longer durations, respectively, than controls (effect 5 SE of the manipulation  Prolonged incubation negatively affected fledging success for females' current broods (with treatment as the lone independent variable; standardized effect5SE p 20:17250:079, F 1, 113 p 4:71, P p :032).However, this negative effect of increasing incubation duration interacted with both clutch initiation date and females' initial egg size (table 2; fig.2B), as the deleterious effect of incubation duration occurred primarily later within seasons and among females initially producing smaller-than-average eggs (fig.2B).

Effects on Future Broods
A female's probability of nesting again following her current brood was not affected directly by her current incubation treatment but was shaped by a treatment # time of year interaction (table 3; fig.3).Later within seasons, females became less likely to produce a subsequent clutch, but this seasonal effect was steepest for females with prolonged incubation periods, while those with shortened incubation periods were more likely to nest again later in the season (fig.3).Females' propensity to produce a subsequent brood was independent of the number of young fledged (table 3).
For females that initiated a subsequent clutch, the number of eggs produced at that time was also impacted by an interaction between the incubation treatment of their first clutch and time of year.While clutch sizes declined over the course of the season, the decline in females' subsequent clutch size was mitigated by a shorter current incubation period (table 3; fig.4).These subsequent clutch sizes also varied between years, with those produced in 2020 being smaller than in prior years (mean number of eggs 5 SE; 2018 p 4:4350:19; 2019 p 4:7650:19; 2020 p 4:19 5 0:12).Additionally, female mass at subsequent nests was impacted, albeit to a nonsignificant extent, by incubation duration, as females experiencing a longer incubation period tended to weigh more on their next breeding attempt (table 3).There was no effect of our treatment on the probability of a female breeding the following year (table 3).

Discussion
Exchanging clutches of eggs between nests with different clutch initiation dates altered the number of days females spent incubating as expected.Incubation periods also shortened as breeding seasons advanced and with increases in the size of clutches, eggs, and the females incubating them.These latter effects on the length of the incubation period were not unexpected given results of previous studies describing their impact on incubation thermodynamics (Williams 1996;Monaghan and Nager 1997;Hanssen et al. 2005;Dobbs et al. 2006;Mueller et al. 2019a).Females may take advantage of these thermodynamics to optimize the necessary level of investment per clutch, with our experimental changes to incubation duration generating several important costs to female breeding productivity within seasons, principally in a manner that depended on variables related to the current environment and initial maternal investment into reproduction (i.e., year, time of year, and egg size).Experimentally elongated incubation periods had no effect on prefledging mass of foster young in 2018, a positive effect in 2019, and a negative effect in 2020, despite similar distributions of breeding dates across the 3 years of our experiment.Any number of factors might differ between years to affect the conditionality of this potential life history trade-off (Stearns 1992;Martin 2002;Ricklefs and Wikelski 2002;Lima 2009).For example, total precipitation was greatest in 2019, when incubation duration had a positive effect on prefledging mass.In contrast, increasing incubation duration had a strongly negative effect on prefledging mass in 2020, a year in which clutch and egg sizes were smaller and in which daily temperatures increased most sharply from May to July (fig.S2), when most nests are initiated.Differences between years may therefore reflect environmental heterogeneity that unmasked otherwise hidden trade-offs between incubation effort and a female's ability to rear young posthatching.In addition, rates of nest depredation and infestation with hematophagous blowfly larvae (Protocaliphora spp.) also vary widely at our study site within and between years (K.D. Miller, R. D. Pell, and E. K. Bowers, unpublished data), potentially contributing to life history costs associated with a prolonged or shortened nesting cycle (e.g., Bosque and Bosque 1995;Conway and Martin 2000;Fitze et al. 2004;Lima 2009;Kovařík and Pavel 2011).We also detected a strong, positive effect of egg size on nestling prefledging mass after young were cross-fostered and reared by an unrelated female, consistent with previous studies finding that females in better condition typically produce larger eggs (Wiebe and Bortolotti 1995;Ramsay and Houston 1997;Moreno et al. 2006) and larger eggs result in larger nestlings posthatching (Williams 1994;Styrsky et al. 1999Styrsky et al. , 2000;;Krist 2009Krist , 2011)).
Prolonged incubation periods had no effect on fledging success early within seasons but became costly to fledging Reproductive Costs of Incubation 259 success later.In temperate climates, abundance of insect prey typically declines throughout the breeding season, influencing parental investment strategies (e.g., Perrins 1965;Kendeigh 1979;Siikamäki 1998;Styrsky et al. 1999Styrsky et al. , 2000;;Krist 2011) and potentially contributing to shifts in the costliness of incubation we observed both within years and between years.Our incubation treatment also interacted with the size of eggs a female originally produced to influence fledging success, suggesting that individual optimization of reproductive effort begins with egg production and that females producing larger eggs are generally better prepared for the increased demands of producing and caring for high-quality young (Amundsen and Stokland 1990; Reid and Boersma 1990;Bolton 1991; see also Styrsky et al. 1999;Krist 2011;Krist and Munclinger 2015).In addition to the contribution of egg size-whether a reflection of maternal quality or the relative level of effort needed to incubate eggs of varying size-manipulations of clutch size sim-ilarly suggest that the cost of incubation is shaped by egg production (e.g., Visser and Lessells 2001;de Heij et al. 2006;Nord and Nilsson 2012).Interestingly however, studies that experimentally increased the number of eggs incubated to explore costs of incubation revealed that larger clutch sizes typically result in prolonged incubation periods (Smith 1989;Dobbs et al. 2006;Nord and Nilsson 2012;Hope et al. 2021), whereas in this experiment, where the number of eggs incubated was unchanged from what females originally produced, larger clutch sizes were associated with shortened incubation periods, similar to patterns reported in studies in which females incubated their natural clutch sizes (e.g., Erikstad and Tveraa 1995;Both and Visser 2005;Bowers et al. 2016; for discussion on the potential perceived value to parents of experimentally manipulated clutch sizes, see also Hanssen et al. 2023).
In addition to effects on a female's current nesting cycle, the incubation treatment also affected future breeding attempts within years, largely through interactions with time of season.The likelihood of breeding again generally declined as seasons advanced, but this was especially the case for females incubating for longer durations (fig.3).Females' subsequent clutch sizes were also influenced by the incubation treatment, where females that experienced a longer incubation period produced smaller clutches in their subsequent brood (table 3).If a longer incubation period caused females to expend greater amounts of energy and resources, it may have diminished their ability to invest in a subsequent clutch (de Heij et al. 2006).However, females also tended to weigh more at their subsequent nest after experiencing a longer incubation period (table 3), which might suggest a potential benefit to having more time between egg production and nestling rearing, partic-ularly within income breeders where females must recuperate resources during incubation.Alternatively, after a longer incubation period, which reduced fledging success and thus the number of fledglings to care for before nesting again, females that produced a smaller subsequent clutch may have weighed more at that time because increased incubation demands induced a change in the prioritization of their own body condition in preparation for the end of the breeding season and forthcoming winter.Further studies are necessary to disentangle the causative nature of an extended incubation duration on subsequent female mass.While individual optimization of clutch size is well documented (Pettifor et al. 1988(Pettifor et al. , 2001;;Rowe et al. 1994;Both et al. 1998;Török et al. 2004), less attention has been

Reproductive Costs of Incubation 261
given to an individual's ability to optimize incubation effort, especially in songbirds.Previous studies on waterfowl that manipulated incubation effort by altering clutch sizes revealed differences in parents' ability to compensate for increased investment into incubation depending on their body condition (Heaney and Monaghan 1996) or initial investment in eggs (Hanssen et al. 2002).In studies that manipulated incubation duration itself, female common eiders (Somateria mollissima) lost more body mass during elongated incubation periods but at no cost to future perfor-mance (Bottitta et al. 2003;Hanssen et al. 2023; but for long-term costs of incubating enlarged clutches, see also Hanssen et al. 2005).Additionally, European coots (Fulica atra) exhibited a higher probability of survival to the next breeding season after a longer incubation period (Brinkhof et al. 2002).While long-lived capital breeders in which energy allocated toward reproduction is derived entirely from stored reserves have shed critical light on our understanding of individual optimization and life history trade-offs, whether such patterns occur in income breeders like passerines rearing altricial young has been comparatively understudied.This is especially the case for individual optimization of breeding effort, which varies not just between species but also between individuals within species and with changing environmental conditions (Garant et al. 2007;Jaatinen et al. 2016).
Even between closely related species, differing life history strategies likely play a decisive role in shaping the optimization of reproductive effort and costs of incubation.For example, in contrast with the current study, a similar experiment on the confamiliar house wren (Troglodytes aedon) found no evidence of fitness costs associated with incubation, despite creating a broader range of incubation durations (Sakaluk et al. 2018).Several key differences in the life history of these otherwise related species might contribute to the existence of costs of incubation, or lack thereof, including migratory status, breeding season length, and breeding phenology.Temperate house wrens are migratory and have a shorter breeding season than nonmigratory Carolina wrens, and their arrival on the breeding grounds in spring and surge in breeding activity in May generally coincide with peak availability of arthropod prey (Bowers et al. 2016).On the other hand, Carolina wrens in our study population begin breeding in March, a time of year when food abundance and ambient temperatures are considerably lower and more variable (sometimes to catastrophic effect, as evidenced by a population crash before the 2021 season; see "Methods") than experienced later in the season (fig.S2; Haggerty and Morton 2020).Indeed, apparent costs of incubation revealed in the current study were almost entirely year and date dependent.As income breeders, female wrens must feed during incubation, so incubating at times of varied food availability or ambient temperature may play a critical role in revealing trade-offs between current incubation effort and subsequent productivity that are otherwise masked in relatively benign environmental conditions (van Noordwijk and de Jong 1986;Stearns 1992;Nord et al. 2010).
Ultimately, an ability to adjust the duration of incubation in response to maternal condition and environmental variation should be selected for if these adjustments adaptively mitigate the costs of reproduction.Despite the diverse findings of previous studies investigating the cost of incubation, our findings suggest that these mixed results may be due to individual optimization of reproductive investment based on both internal and external conditions and that conflicting results on the costs of incubation might be attributable, at least in part, to methodological differences between studies and the breadth of variation in ecological conditions over which breeders are observed.Indeed, we found that the effect of incubation duration on offspring quality was negative in one year (consistent with a cost of incubation), positive in another, and neutral in the third year.Thus, had we conducted this experiment within only one year, our results about the existence of any costs associated with incubation would paint a drastically different picture from what is apparent after conducting our experiment across multiple seasons, thereby demonstrating the importance of utilizing a wide range of environmental settings for robust inference of the costs of reproduction and evolution of life histories.

Figure 1 :
Figure 1: Effect of the manipulation on incubation duration.Overlapping data are jittered, and the regression line is bound by 95% confidence limits while holding other covariates in the model at their average value.

Figure 2 :
Figure2: Effects on nestling prefledging body mass (g) and fledging success from females' current broods.A, For nestling mass, brood averages are plotted while adjusting for the size of eggs from which nestlings hatched (e.g., positive and negative values are heavier and lighter nestlings, respectively, than expected given the size of eggs from which they hatched; see table 2), and the regression lines are bound by 95% confidence limits.B, For fledging success, broods that succeeded in producing at least one fledgling and failed broods are plotted as black and white circles, respectively, and contours depict the model-predicted proportion of young fledged, with darker and lighter areas showing increased and decreased fledging success, respectively, while holding other covariates in the model at their average value.Data points are included to show data spread, and the location of points within the contour plot does not reflect the fledging success of that nest.

Figure 3 :Figure 4 :
Figure3: Interactive effect of incubation treatment and current clutch initiation date on the probability of a female attempting a subsequent nesting cycle within the year of the manipulation.Jittered points indicate females that did or did not produce a subsequent brood, and the regression lines are bound by 95% confidence limits while holding other covariates in the model at their average value.Incubation treatment is depicted here as a categorical variable grouping all incubation-shortened and incubation-lengthened females for graphical purposes only.
Reproductive Costs of Incubation 257 in the absence of any covariates p 0:82850:049, F 1, 105 p 283:2, P !:001).The range of incubation durations created by the manipulation (10-19 days) still fell within the breadth of what is typically observed in natural nests of this species.In addition to the effect of treatment, incubation duration also declined over the course of each season and with increases in the size of clutches, eggs, and the females incubating them (table1).

Table 1 :
Effects on incubation duration (N p 107 clutches hatched) a Positive and negative values represent prolonged and shortened incubation periods, respectively, as expected from differences in clutch initiation date.bOf the female's original clutch of eggs.c Relative to 2020.

Table 2 :
Effects of incubation duration on current nesting cycle a Of the female's original clutch of eggs.b Relative to 2020.c Size of eggs from which foster offspring hatched.d Size of eggs a female initially produced (i.e., her original eggs, not foster eggs).

Table 3 :
Effects of incubation duration on subsequent nesting cycles a Of the female's original clutch of eggs.b Relative to 2020.c Relative to 2019; note that females breeding in 2020 were omitted because of winter storms (see "Methods").