The effects of environmental conditions on eggshell thickness of two North American secondary cavity-nesting birds

Climate change is a threat to biodiversity and directly impacts ecosystems and populations. Examples of these impacts are found across taxa and have become more severe in recent decades. Many species can shift or expand their geographic ranges to track preferred climate niches or adapt to novel conditions (Johansson et al. 2014). However, there are species that cannot adapt or are not adapting quickly enough, which leads to population declines and potential local extinction (Møller et al. 2008). Latitudinal shifts and upward changes in elevation to combat climate change have been recorded in mammals, reptiles, and plants (Kelly and Goulden 2008, Chen et al. 2011, Moreno-Rueda et al. 2012, Williams and Blois 2018). Species that do not shift their ranges may adapt to novel environments by altering their food sources, breeding and nesting preferences, and their phenology (Mayor et al. 2017). Indicator species can help scientists monitor ecosystem changes, at both small and large scales, to track how species do or do not adapt.

Passerines are good indicators of ecosystem health due to the high rate of monitoring in many parts of the world and their sensitivity to environmental conditions (Gregory et al. 2003, Sousa et al. 2008, Gregory and van Strien 2010, Smits and Fernie 2013). In September 2020, there was a mass mortality event involving hundreds of thousands of migratory songbirds in the United States after a cold front moved through the Southwest (Stanek et al. 2022). These birds likely succumbed to emaciation after the combined effects of drought, wildfires, and lack of food resources along their migration corridors (Yang et al. 2021). The ability to assess ecosystem health through passerine monitoring can bring attention to possible threats.

Reproductive success in birds is directly affected by eggshell integrity. Eggs are vital for protection of the developing embryo from the environment and regulate gas and water exchange. The thickness of eggshells, among other eggshell qualities, determines how much water is lost from inside the egg and is a good indicator of population health. The negative effects of insecticides on eggshells are well-documented threats to avian populations (Ratcliffe 1970). For example, dichlorodiphenyltrichloroethane (DDT) and its metabolites dichlorodiphenyldichloroethylene (DDE) and dichlorodiphenyldichloroethane (DDD) are known to negatively impact Bald Eagle (Haliaeetus leucocephalus) and other raptor populations by weakening their eggshells (Grier 1982, Falk et al. 2019). The weakening eggshells broke easily and caused significant population declines.

Eggshell quality can be influenced by environmental factors. For example, higher temperatures result in thinner eggshells because of a decrease in available calcium in the blood (Warren and Schnepel 1940, Ruzal et al. 2011, Ebeid et al. 2012). This pattern of thinning eggshells has been documented mainly during the summer months in chickens (Warren and Schnepel 1940, Roberts 2004). Evidence suggests that humidity may play a more important role than temperature in eggshell integrity. For example, eggshells may become thicker than normal during times of high precipitation and very wet conditions to prevent oversaturation in the egg, which can lead to chick mortality through suffocation and drowning (Veldsman et al. 2020, Attard and Portugal 2022). Alternatively, drought and dry conditions may result in thinner eggshells than normal through reduced food availability (Hernández et al. 2018).

A recent review compared the thickness of eggshells across more than 4,000 avian species and found that eggshell thickness was mainly driven by phylogenetics (Attard and Portugal 2022). However, the natural variation in eggshell thickness within species and populations is still unclear and understudied, especially in relation to environmental conditions and how thickness changes over time. This makes it difficult to predict how climate change will impact reproductive success and hatchability of specific populations. Our goal is to determine variation in eggshell thickness over time and to test if environmental conditions are correlated with eggshell thickness and under what circumstances. Understanding changes in eggshell thickness will allow us to identify important selective pressures influencing eggshell integrity (Attard and Portugal 2022).

Western Bluebirds (Sialia mexicana, hereinafter bluebirds) and Ash-throated Flycatchers (Myiarchus cinerascens, hereinafter flycatchers) are secondary cavity nesters. Habitats of these two bird species are changing due to climate change in the southwestern United States, which has resulted in changes to bird communities and populations. Fair et al. (2018) documented significant long-term losses in avian species diversity and overall bird numbers in response to a piñon pine (Pinus edulis) mortality event that occurred between 2000 and 2002 due to climate-induced drought and a resulting bark beetle outbreak in northern New Mexico. Data suggest that some species moved to higher elevations on the plateau and the surrounding Jemez Mountains where tree mortality was not as high (Fair et al. 2018). A subsequent study determined that bluebirds significantly increased nesting elevation over a 19 year period (Wysner et al. 2019). Flycatchers did not alter their elevation preferences, which may suggest a higher tolerance for changing environments. These studies suggest the two species may be affected by environmental changes differently.

Because environmental conditions have been rapidly changing in the Southwest, we are investigating eggshell thickness of these two secondary cavity-nesting species using data from 1997 to 2013 from a nest box network in northern New Mexico. Our goal is to understand the natural variation in eggshell thickness over time and to determine how thickness is affected by nesting elevation, temperature, drought index (measured by the Palmer Drought Severity Index [PDSI]), and precipitation in our two focal species. Our hypothesis regarding elevation and temperature is that eggshells will be thinner at lower elevations and at higher temperatures because high temperatures could result in reduced calcium availability in the blood (Warren and Schnepel 1940, Ebeid et al. 2012). We hypothesize that eggshells will be thinner during years with low precipitation and dry conditions because of reduced food availability (Hernández et al. 2018). Using long-term datasets in a current and future hotspot of climate change allows us to identify factors important for eggshell traits and hatchability.

Western Bluebirds and Ash-throated Flycatchers are secondary cavity-nesting passerines and share life history traits, and they can provide insight into the traits that allow species to adjust to a changing climate. The Western Bluebird is a monogamous species that is widely distributed and sexually dichromatic in its plumage coloration (Guinan et al. 2020). Flycatchers are also monogamous but are not as widely distributed as bluebirds, nor are they sexually dichromatic. Both species share behavioral characteristics, such as using nest boxes and feeding on insects during breeding seasons (Cardiff and Dittmann 2020). They have slight differences in their developmental stages. For example, flycatchers fledge at about 16–17 d after hatching while bluebirds fledge at about 20–21 d of age (Mock et al. 2021, Solt and Hu 2021).

This study took place at Los Alamos National Laboratory (LANL) and its surrounding areas in Los Alamos, New Mexico, USA. LANL is located on the Pajarito Plateau and occupies approximately 111 km². The Pajarito Plateau contains large canyon systems on the eastern flanks of the Jemez Mountains. Its smaller canyon systems extend east toward the Rio Grande River or merge into larger canyons. The surrounding area can be described as narrow mesas separated by steep and deep canyons from the Jemez Mountains down to the Rio Grande River, which includes both canyon and mesa-top habitats. There are 12 different watersheds that are geographically separated by these canyons, whose floors are usually flat with rocky sides and partial tree coverage (Bartlow et al. 2021).

The habitat primarily found in this study area is piñon-juniper and ponderosa pine (Pinus ponderosa) forests. Piñon-juniper forests consist of one-seed juniper (Juniper monosperma) and piñon pine. Piñon-juniper forests generally occupy elevations between 1,900 and 2,100 m in elevation and cover south-facing slopes and mesa tops. Ponderosa pine forests occupy slightly higher elevations at 2,100 to 2,300 m. Other habitat types where data were located were in riparian areas—typically consisting of broadleaf cattail (Typha latifolia), lanceleaf cottonwood (Populus acuminata), and willow species—and mixed conifer forests. Of the 12 watersheds located on the Pajarito Plateau, only 5 were included in this study based on the locations from which our data are concentrated. These 5 watersheds are named Ancho (piñon-juniper and ponderosa habitat), Los Alamos (mixed conifer habitat), Mortandad (mixed conifer, then piñon-juniper after a wildfire in 2000), Pajarito (riparian and “dry” wetlands), and Water canyons (ponderosa habitat).

Nonviable eggs have been collected within our avian nest box network beginning in 1997 through the present, although our focal data were collected between 1997 and 2013. The avian nest box network consists of 350–500 wooden nest boxes across the different watersheds. Not all sampling locations were monitored every year due to inactive sites, accessibility, or changes in staff over the years. Rather, a subset of these sites was operated each year. During the breeding season, nest boxes in the network were continuously monitored by researchers. Each nest box has a GPS and elevation associated with it. Each box was checked once every 2 weeks unless there were signs of activity such as nesting material or newly laid eggs. Active nests were then monitored until the eggs hatched and visited once more when nestlings were 10–15 d old for banding.

Occasionally, some or all eggs do not hatch, or the nests are abandoned. To determine if unhatched eggs were viable or abandoned, previously collected data for an active nest was used in the field with records of nest activity, such as nests with warm eggs. Nests were assumed abandoned if they were left for more than 2 weeks. Eggs that were recently laid or had no visit data were left for 2 weeks prior to checking and collecting. Researchers made sure not to collect eggs too early and did so only following multiple nest box checks. New eggs found during a nest check that were cold were left in the nest with the assumption that the female was still laying. However, once eggs were recorded as warm and then became cold, they were deemed nonviable or abandoned on later visits of more than 2 weeks. There were a few clutches where not all eggs hatched. Since clutches for these species hatch at approximately the same time, any unhatched eggs in a clutch with nestlings older than 10 d were collected. Cracked eggs were typically removed from nests but not collected.

After the nonviable eggs were collected, they were refrigerated until processed, and each egg was identified as a sample. We processed each egg and recorded the total weight (grams) on a digital scientific scale, and measured length (millimeters) and width (millimeters) with calipers. The contents were emptied by carefully cracking the eggshell crosswise using a micro-spatula. The 2 halves of the eggshell were gently cleaned with water, removing any membrane layers that were doubled over from being frayed along the equator, and set aside to dry completely. The eggs contained the thin membrane for the majority of the eggshells measured. Once the eggshells dried, their thickness was measured using a 1010M dial indicator pocket gauge (L.S. Starrett Company, Athol, Massachusetts, USA) to the nearest hundredth of a millimeter. We measured each half of each eggshell 6 times along the equator of the egg for a total of 12 measurements per egg, and we calculated the average thickness of each egg. Thickness was only measured along the equator where the eggshell is essentially flat, and the piece measured would break off in the dial calipers to be flat. One person (JMF) measured 90% of the eggshells and repeatability was measured and tested to train personnel who assisted periodically. All the eggshells were stored in an amber or glass vial for future research.

Temperature and weather data were gathered to analyze patterns over time. Drought severity was measured using a water balance model (Palmer Drought Severity Index; PDSI) that considers both precipitation and temperature to estimate relative dryness (NCAR Climate Data Guide 2020; https://climatedataguide.ucar.edu/). This measures drought events using precipitation and temperature data in a simple water balance model to study moisture supply and demand. Drought index, temperature, and precipitation data were found on the National Centers for Environmental Information (NOAA) websites. We looked at Climate at a Glance (https://www.ncei.noaa.gov/access/monitoring/climate-at-a-glance/), found Divisional Time Series, selected our parameter (PDSI, Average Temperature, or Precipitation), and set our Time Scale to “All Months”. Our start year was 1997 and end year was 2013 and the state was set to New Mexico with Climate Division set to 2: Northern Mountains. Then each dataset was filtered to only include April and May, and the mean was taken for the values of these 2 months. We chose to collect environmental data for these months because bluebirds tend to initiate nest construction in early April and May (Brawn 1991, Keyser et al. 2004, Kozma and Kroll 2010), while flycatchers construct nests in May after migrating to their breeding site (Hall and Morrison 2003). The spring months of April and May are the time of year when adults of both species are exposed to environmental stress at our site during the egg laying process. Breeding birds can alter their phenology to respond to environmental conditions (Borgman and Wolf 2016), which can affect individual fitness and the resources used by bluebirds and flycatchers during egg production. For example, spring precipitation levels affect foraging behavior of breeding birds to obtain essential nutrients from the available food (Brawn 1991).

The statistical software R 4.1.2 (R Core Team 2021) was used to perform all analyses. The tidyverse (version 1.3.1) package (Wickham et al. 2019) in R was used to filter out eggshell thickness samples with missing or incomplete data. The 1997–2013 analyzed dataset included 330 bluebird eggs and 113 flycatcher eggs, for a total of 443 eggshell samples. Eggshell samples were collected from study locations that were spatially organized in different watersheds. The assumptions of normality and homogeneity of variances in the ANOVA test were both violated based on the Shapiro-Wilk normality test and the Bartlett test. Therefore, we used the nonparametric Kruskal-Wallis test to determine whether the geographical separation of the 5 watersheds contributed to variations in eggshell thickness observed in the dataset for each species separately. This test was significant for bluebirds (P < 0.05) and watershed location was initially included as a random effect in the linear mixed effect models (LMM) described below.

There were also instances of several egg samples originating from the same clutch, even though the eggs were individually processed for eggshell thickness measurements. Therefore, the identification number of individual nest boxes was also included as a random effect in the LMMs to account for natural variation among eggshell samples from eggs that were collected from the same clutch. There were 76 nests over the whole time period that had more than one egg included in the study. These nests accounted for 147 eggs in the dataset, which is 33.2% of the total eggs examined (n = 443).

The lme4 (version 1.1.27.1; Bates et al. 2015) and lmerTest (version 3.1.3; Kuznetsova et al. 2017) packages in R were used to run LMMs and the package MuMIn (Bartoń 2020) was used for the model selection procedure. Eggshell thickness was the response variable in the full LMMs. The full LMMs included watershed and nest box ID as random effects while year, nesting elevation, drought index, temperature, and precipitation were included as fixed effects. We included an interaction between PDSI and temperature. The two avian species were analyzed separately and, therefore, we went through the following model selection process twice. After building the full model, we standardized the predictor variables using the standardize function in the arm package (Gelman and Su 2021). After standardizing, we tested for multicollinearity using the variance inflation method (VIF) using the vif function in the car package (Fox and Weisberg 2019). All VIF values for the bluebird data were less than 2.9, and all VIF values for the flycatcher data were less than 3.9, except for the terms in the interactions. In our full mixed model, we used maximum likelihood (ML) for model selection, instead of the default restricted maximum likelihood (REML), which is not recommended for comparing models with different fixed effects. Upon examination of the residuals of the full model using a Q-Q plot, it appeared that the residuals were normally distributed.

We used the function dredge in the MuMIn package (Bartoń 2020) for the model selection process and ranked all fixed effects combinations using AICc. The top model criteria was a model or models with the lowest AICc value and a delta AICc value greater than 2 from the next highest ranking model. If more than one model was within 2 delta AICc units of the top model, then we used model averaging to get the estimated coefficients and P-values. These estimates are based on the standardized predictors. For model averaging, we used the full average, which uses zero as the value if that particular variable is not present in one or more of the top models. We also present the 95% confidence intervals (CI) for each of the variables in the model using the confinf function. For examining and displaying any interaction terms in our models, we used the interplot function in the interplot package (Solt and Hu 2021).

For both species, year was a variable in the top models. When plotting year against eggshell thickness, it appeared that the relationships between year and eggshell thickness were nonlinear, so we decided to run an additional model selection procedure that had a second-order (quadratic) and third-order (cubic) polynomial term with the year variable. We compared these models to the univariate year model. For this second model selection process, we used the aictab function in the AICcmodavg package (Mazerolle 2020) and again used a cutoff of 2 delta AICc units as the top model criteria.

A total of 443 eggs were collected and analyzed between 1997 and 2013 during the breeding season. Eggshell thickness for bluebirds ranged from 0.124 to 0.564 mm, while eggshell thickness for flycatchers had a smaller range of 0.109 to 0.350 mm. The mean eggshell thickness for each species each year from 1997 to 2013 is shown in Table 1.

There was a significant difference in bluebird eggshell thickness among watersheds between 1997 and 2013 (Kruskal-Wallis: χ² = 36.94, df = 4, P < 0.001; Fig. 1). Pairwise comparisons with Bonferroni-corrected P-values showed that eggshells from Pajarito Canyon were significantly thicker than eggshells from Los Alamos Canyon (Dunn test: Z = −5.88, P < 0.001) and Mortandad Canyon (Dunn test: Z = −4.80, P < 0.001). The significant differences in eggshell thickness between watersheds for bluebirds prompted us to include watershed location as a random effect in the LMM. There was no significant difference in flycatcher eggshell thickness among watersheds between 1997 and 2013 (Kruskal-Wallis: χ² = 8.80, df = 4, P = 0.07; Fig. 1).

The model selection procedure to find the best predictors of eggshell thickness was done separately for each bird species. When watershed was included as a random effect in the full model for both species, it explained zero variance. Therefore, we removed watershed as a random effect for both model selection procedures, and we used nest box as the only random effect. Even though one canyon had significantly thicker eggshells, this slight difference was negligible for our models.

For predicting bluebird eggshell thickness, there were 2 top models that were within 2 delta AICc units of the top model (Table 2). Both of these top models contained year and the interaction between drought severity (PDSI) and temperature. The model averaged coefficients for these models are shown in Table 3 using the full average approach. The year, temperature, and the interaction between PDSI and temperature variables were significant variables after model averaging (Table 3). Year was positively correlated with eggshell thickness (estimate = 0.055, P < 0.0001; Table 3; Fig. 2), while temperature was negatively correlated with eggshell thickness (estimate = −0.025, P < 0.001; Table 3). The interaction between PDSI and temperature was negative (estimate = −0.072, P < 0.0001), meaning that the coefficient for temperature is negatively correlated with PDSI (Fig. 3). As PDSI values get higher (wetter conditions), the coefficient of temperature on eggshell thickness decreases (Fig. 3). At low PDSI values, temperature is positively correlated with eggshell thickness, while at high PDSI values, temperature is negatively correlated with eggshell thickness. Around normal conditions (PDSI approximately zero), there is no effect of temperature on eggshell thickness (estimate coefficient near zero; Fig. 3).

Year appeared curvilinear, so we wanted to identify the best year model that predicted eggshell thickness for bluebirds. The 2 polynomial models ranked higher than the univariate year model but were within 2 delta AICc units of each other (Fig. 2, Table 4). The model with the cubic term did not explain any more variation than the model with the quadratic term.

For predicting flycatcher eggshell thickness, there were 3 top models (Table 2). Like the results for bluebirds, the year variable was present in all 3 models. The model averaged results show that year and temperature were the only significant variables out of the ones included in the 3 top models (Table 3). Flycatcher eggshells became thicker over time (estimate = 0.052, P < 0.0001; Table 3, Fig. 4). Eggshells were also correlated with temperature, such that thicker eggshells were found at higher temperatures (estimate = 0.027, P = 0.012; Table 3, Fig. 5). The interaction between PDSI and temperature was not in any of the top models for flycatchers.

Year appeared curvilinear, so we identified the best year model that predicted eggshell thickness for flycatchers. The univariate model was the top model (Table 4). The quadratic and cubic models were greater than 2 delta AICc units from the univariate model (Table 4).

Passerines are excellent monitors of ecosystem health because of their quick responses to environmental change. Changing environmental conditions and altered nesting and food resources can affect reproductive success through changes in eggshell traits, including eggshell thickness. The main goal of our study was to use a long-term dataset to determine if there were significant changes in Western Bluebird and Ash-throated Flycatcher eggshell thickness over time and to examine if thickness was influenced by elevation and environmental variables, including spring drought conditions, precipitation, and temperature.

Data from 1997 to 2013 showed that eggshell thickness, in both species, increased over the 17 year period. It is unclear why eggshells have increased over this time period. Eggshell thickness does not appear to have steadily increased from 1997 to 2013. The relationship is nonlinear, and these results suggest that environmental conditions could influence eggshell thickness. For bluebirds, the 2 models with a polynomial term were a better fit to the data than the univariate year model. This was not the case for flycatchers; the univariate year model was the top model predicting eggshell thickness.

For bluebirds, temperature was a significant predictor in the top models, and the effect of temperature on eggshell thickness was mediated by drought severity (PDSI), based on the significant interaction between temperature and PDSI. As drought severity lessened, the effect of temperature went from a positive effect to a negative effect on eggshell thickness. For flycatchers, higher temperatures resulted in thicker eggshells, but there was no interaction between temperature and PDSI; perhaps more data are needed. Therefore, our hypothesis of eggshells becoming thinner in response to high temperatures is partially supported by our data. For flycatchers, the pattern was opposite of our prediction: eggshell thickness was positively correlated with temperature. For bluebirds, only during wet conditions were high temperatures correlated with thinner eggshells.

The interaction between temperature and PDSI in bluebirds could be explained by a recent study investigating relative humidity and temperature on water loss from eggs. Water loss is faster at high temperatures and low humidity than at high humidity (Veldsman et al. 2020). Our results show that higher PDSI values, which correspond to wetter conditions, result in higher temperatures being associated with thinner eggshells. These thinner eggshells would lose water at a slower rate than during periods of drought. During drought conditions (low PDSI values), higher temperatures are correlated with thicker eggshells. These eggs would lose water at a faster rate and could be thicker to reduce water loss. Future work in this system should consider relative humidity and the effects of temperature on eggshell thickness, since PDSI might not be strongly reflective of relative humidity.

The effects of high temperatures on eggshell thickness has mainly been reported on chickens, in which eggshells become thinner in summer months (Warren and Schnepel 1940, Ebeid et al. 2012). The temperatures experienced by these two species during the months of April and May when they are beginning to build nests and breed may not be high enough to influence the availability of calcium for egg production. Temperatures that could influence the availability of calcium may only be consistently high enough in June through August in the Southwest. Examining eggshell thickness of females who lay a second clutch in midsummer may provide additional insight into temperature effects on wild populations. The relationship between temperature and eggshell thickness was observed in Pied Flycatcher (Ficedula hypoleuca) populations, which laid thicker eggshells when the ambient temperature was higher than usual during egg laying (Morales et al. 2013). This could be a mechanism for Ash-throated Flycatchers to prevent water loss in the egg during high temperatures (Morales et al. 2013) and could be an explanation for the flycatcher results presented here.

There was no significant relationship between eggshell thickness and precipitation or PDSI, which also contradicts our hypothesis. We predicted that eggshells would become thinner in drought conditions because of reduced food availability (Hernández et al. 2018). Two studies on species of flycatchers addressed food availability and eggshell traits. Morosinotto et al. (2019) found that Pied Flycatchers supplemented with food had thicker eggshells than those not supplemented with food, suggesting that they were able to invest more resources in eggshell development. Additionally, female Collared Flycatchers (Ficedula albicollis) that were in better condition laid thicker eggshells (Hargitai et al. 2011). These studies appear to be opposite of the patterns observed here. Bluebirds and flycatchers consume insects during the spring and summer months, and we would expect there to be fewer food resources during times of drought (Bogan and Lytle 2011), both in terms of low spring precipitation and low PDSI values.

Our study does present several limitations because it only represents two secondary cavity-nesting species and data from a specific region of New Mexico. Of these two species, there were fewer flycatcher samples. More bluebird samples were collected because they are more abundant in our study area and occupy nest boxes more readily than flycatchers. Another limitation is that this study only used nonviable or abandoned eggs to avoid collecting hatchable eggs, and it is not clear if the eggshells of nonviable or abandoned eggs are completely representative of successful eggs. While there are many reasons why eggs either did not hatch or nests were abandoned, it is an important consideration for the study. Females may have nonviable eggs or abandon nests with eggs because they are in poor condition, which could have resulted in eggshell thickness differences. A final limitation concerns the climate data used in our correlations. We obtained climate data from a fairly large geographic area (Northern Mountains region in New Mexico).

Eggshell thickness has a significant effect on reproductive success, as seen by DDT in Bald Eagles (Grier 1982, Falk et al. 2019). Avian populations around the world are threatened by anthropocentric activities, including habitat encroachment and climate change. Each of these threats can cause irreparable damage to population success. Among many other traits important for adaptation and plasticity, the structure of eggs can determine whether certain species can adapt to novel conditions because of climate change or persist in new environments after range shifts.

We show that eggshell thickness has increased over time for both species, and that environmental variables are correlated with eggshell thickness. Specifically, flycatcher eggshells got thicker at higher temperatures, whereas bluebird eggshells were influenced by both temperature and drought severity. Given the arid environment in which these populations breed and the continued threat of drought, the increase in eggshell thickness could be a way to slow water loss and desiccation of the embryo within the egg. As there are few studies investigating eggshell thickness variation over time and in response to environmental conditions, we hope our study provides useful data for future studies on potential climate change impacts in the southwestern United States and in similar environments.

Numerous students and personnel completed field work for this project and assisted with the eggshell analysis in the laboratory. Funding was provided by the U.S. Department of Energy, Triad National Security, LLC, operator of Los Alamos National Laboratory, under Contract No. 89233218CNA000001. Data collectors acted in accordance with the Guidelines for the Use of Wild Birds in Research (Fair and Jones 2010), and the Institutional Animal Care and Use Committee protocol. All New Mexico State and Federal Scientific Permits were obtained for all years of the project. The authors declare that they have no conflict of interest.