Effectiveness of novel artificial seabird nest modules for reducing ambient temperature transfer in a warming climate

Artificial habitat for cavity nesting birds can provide excellent opportunities for research and conservation efforts but may expose species to the negative impacts of warming ambient temperatures with climate change. Artificial nest boxes have been successfully used to monitor the breeding activity of the Cassin's auklet ( Ptychoramphus aleuticus ), a small burrow ‐ nesting seabird, on Southeast Farallon Island (SEFI) since 1971. Mean monthly ambient air temperatures on SEFI during the summer months have increased at an annual rate of roughly 0.03°C from 1971 to 2022, along with an increase in the number of extreme heat days and average maximum temperature, confirming a warming trend at this seabird colony. Given a projected increase in global temperature, we assessed the effectiveness of traditional wooden nest boxes vs. newer ceramic modules

been successfully used to monitor the breeding activity of the Cassin's auklet (Ptychoramphus aleuticus), a small burrownesting seabird, on Southeast Farallon Island (SEFI) since 1971.Mean monthly ambient air temperatures on SEFI during the summer months have increased at an annual rate of roughly 0.03°C from 1971 to 2022, along with an increase in the number of extreme heat days and average maximum temperature, confirming a warming trend at this seabird colony.Given a projected increase in global temperature, we assessed the effectiveness of traditional wooden nest boxes vs. newer ceramic modules at buffering external ambient air temperatures in normal and extreme heat days across a gradient of microclimates on SEFI.Results from fitting linear mixed effects models indicated that, on average, internal temperatures of wooden and ceramic nests (of comparable size and shape) exhibited similar rates of deviation from ambient air temperature of approximately 0.15°C, even during extreme heat events.Ceramic modules did keep nest chambers cooler by approximately 1.2°C than wooden boxes during extreme impacts of increasing temperatures will be critical for creating safe breeding conditions for cavity-nesting birds in a warming world (Stein et al. 2014).The Cassin's auklet (Ptychoramphus aleuticus; hereafter auklet) is a small diving seabird found throughout the northeast Pacific (Manuwal 1974, Ainley andBoekelheide 1990) that lays a single egg in an excavated burrow or rock crevice.Egg laying date ranges from March to May depending on environmental conditions (Lee et al. 2007), with some individuals initiating a second breeding attempt in mid-summer when conditions are favorable (Johns et al. 2017).Researchers from Point Blue Conservation Science have used wooden nest boxes to study auklet breeding biology on Southeast Farallon Island (SEFI) since 1971 (Ainley andBoekelheide 1990, Lee et al. 2007).
Nest boxes were originally constructed out of plywood with an above ground nesting chamber and were installed throughout the breeding habitat of auklets on SEFI (Figure 1).
Long-term environmental monitoring data from SEFI suggest an increasing trend in maximum ambient air temperature on the island since the early 1970's, along with periodic spikes in the number of days when the midday air temperature was greater than or equal to 20°C, referred to here as extreme heat events.Extreme heat events at or greater than 20°C in previous years have resulted in heat stress and even death of adult auklets (Warzybok and Bradley 2008), leading researchers at Point Blue to experiment with different nest materials and designs that would be more resistant to extreme heating.As a first step, wooden shade structures were installed over the original plywood nests to reduce temperature extremes.The shades helped moderate temperature inside the nest boxes relative to unshaded boxes (Kelsey et al. 2016), but a more durable and effective solution was required.In response, we began a multi-year effort to design and test new heat-tolerant auklet nest modules.We explored an array of nontraditional materials including ceramic, cement, insulated cement, and plastic.Also, we designed features to both insulate and cool the nest chamber by adding solar reflection (white glazed sun shields) and air flow (warm air escapes through roof ventilation).After initial testing, we chose to focus on ceramic solutions over concrete and other materials because clay displayed promising thermal properties, can mimic the shapes of natural burrows, and when fired is durable and long-lasting.
The purpose of our study was to test whether the ceramic module design provided better protection from extreme ambient air temperatures compared to the shaded wooden nest box design.First, we further examined trends in ambient air temperature during the core breeding season of auklets (May through September) to quantify the degree of warming on SEFI over the past 50 years.Second, we examined whether there were significant differences in temperatures observed between the ceramic and wooden designs, while factoring in location on the island and the occurrence of extreme heat events.We hypothesized that ceramic modules would regulate burrow temperatures better than wooden boxes and provide more protection from extreme temperature days.

STUDY AREA
Southeast Farallon Island (37.6969°N, 123.0019°W) is part of the Farallon Islands National Wildlife Refuge, located 48 km west of San Francisco, California, USA.The island has an area of 0.39 km 2 with a dominant peak named Lighthouse Hill, which is centrally located and reaches 109 m in elevation (Figure 1).Lighthouse Hill blocks the southeast portion of the island from prevailing northwest winds, creating a cooler, wetter climate on the northwest side of the hill and a warmer, drier climate on the southeast side.

Trends in long-term island air temperature
Ambient air temperature data were collected at a standard, shielded weather station containing a mercury thermometer on the low-lying marine terrace (~2 m elevation) 3 times each day (0600, 1200, 1500 hours local time) since 1971.The 3 daily measurements of ambient air temperature were summarized as mean monthly ambient air temperature, monthly maximum ambient air temperature, and monthly total number of extreme heat days observed to negatively impact auklet health (defined as a day when the maximum temperature exceeded 20°C) for each year (Warzybok and Bradley 2008).
The nonparametric Mann-Kendall Trend Test (showing the measure of monotony of the slope with rank correlation coefficient Τ, which varies between −1 and 1) was implemented using R version 4.0.2(R Core Team 2020) with the package Kendall (McLeod 2022) to determine whether any of the months of the core auklet breeding season between May and September showed an increasing trend in mean monthly ambient air temperature, whether the annual mean of the monthly maximum ambient air temperature across all months of the core breeding season (May through September) showed a trend, and whether there was a trend in total number of annual extreme heat days from 1971 through 2022.The Mann-Kendall Trend Test is the preferred method for analyzing meteorological time-series data because it tests for a consistently increasing or decreasing pattern over time and is less sensitive to outliers compared to simple linear regression (Mann 1945, Kendall 1975, National Center for Atmospheric Research Staff 2014).The null hypotheses were no trend in mean monthly ambient air temperature, annual mean of the maximum ambient air temperature for the auklet breeding season, or annual number of extreme heat days, with an alpha of 0.05.If a significant trend was detected, the Sen's Slope function in the R package trend (Pohlert 2020) was used to calculate the linear rate of change.

Nest Chamber Temperature Recordings
We examined the effectiveness of 2 artificial nest designs for moderating ambient temperature: shaded wooden boxes and shaded ceramic modules.The designs were similar in size, shape, and placement above ground.The dimensions of wooden nests were 20 × 20 × 40 cm with a 10-cm diameter polyvinyl chloride (PVC) pipe as an entrance tunnel (Figure SA, see Supporting Information).The ceramic design had a nest chamber with rounded sides and overall dimension of approximately 20 × 20 × 33 cm with a separate 10-cm diameter ceramic tube as an entrance tunnel (Figure SB, see Supporting Information).The topsoil is very shallow on Southeast Farallon Island and mixed with large, weathered granite rocks, making excavation difficult.Thus, both artificial nest designs were installed on the surface, and the nest chambers lined with a thin layer of dirt.
The nest treatments differed in the building materials and some design features of the nest chambers and shade structures.The wooden nests were constructed with 1.3-cm thick CDX plywood and painted to prolong the useful life of the nest box.For the ceramic nests, stoneware type clay was shaped by hand, molds, and extruders for production, and fired at 1,180°C (cone 5).After firing, the ceramic walls were 1.3-1.5 cm thick, with 2 cm diameter holes in the ceiling to allow warm air to escape.The shade structure on top of the wooden nests was a flat sheet of plywood (61 × 1.3 × 30 cm) painted white and raised 5 cm creating an air gap between the shade and the wooden box (Figure SA).The ceramic nest chamber was protected by a sunshield 1.5-cm thick that curves down the sides and back of the nest and sits on top of ridges built to create an air insulation gap.The sun-facing side of the sunshield was fired with a glossy white glaze to maximize solar reflection (Figure SB).Ambient temperature within the nest chambers of each wooden box or ceramic module was monitored using LogTag TRIX-8 Temperature Data Recorders (microdaq.com,LTD, Contoocook, NH, USA) affixed to the ceiling of the nest chamber.The temperature loggers recorded the temperature of the nest environment every 30 minutes for the months of March through August 2017-2022.Due to restrictions associated with Covid-19 we did not collect data in 2020.Each year, a balanced sample of temperature loggers were placed in wooden boxes and clay modules in 3 different areas along a northwest to southeast gradient across the island: Cormorant Blind (cb), Lighthouse Hill (lhh), and Carpentry Shop (cs; Figure 1).These 3 areas were approximately 250 m apart.In some instances, tags became dislodged and fell onto the burrow floor, were under birds during incubation, or were kicked out of the nesting chamber, resulting in erroneous temperature readings.Tags that collected erroneous temperate readings were noted and data from these tags excluded from the analysis.Data were also visually checked for obvious signs of failed attachment (sharp spikes in temperature and unrealistic temperature readings).For a final quality control check we assigned a z-score to each 30-minute recording and removed all outlying recordings greater than 3 standard deviations from the mean recording across all wooden boxes and ceramic modules for all years.Final sample sizes after excluding failed loggers for clay modules and wooden boxes across all years were 63 and 75 loggers, respectively (treatment cb: 15 ceramic, 22 wood; lhh: 24, 26; cs: 24, 27).
We constructed a candidate set of linear mixed effects models with categorical predictors treatment type (wooden vs. ceramic), area (cb, lhh, and cs), and extreme heat event (True if daily ambient max temperature was above 20°C, otherwise False), with the difference between daily mean nest temperature and the daily mean ambient temperature of the island (Δ temp) as the response variable.Daily Δ temperatures approximated a normal distribution, and no transformations were required.A unique identifier for nest number and year was created and used as a random intercept in all models to account for repeated measures of ambient temperatures and possible biases associated with the specific conditions of each nest.The combination of year and nest ID was included given it was assumed conditions of nests used in more than 1 year likely changed over time (changes in surrounding vegetation, nest orientation, conditions of the material, etc.).Different models consisting of all possible combinations of predictors (including main effects and interactions) in the full model (3-way interaction between treatment type, area, and extreme heat event), and a null intercept only model, were fit using the R package lme4 (Bates et al. 2015) computed with maximum likelihood (ML).We ranked all candidate models with computed Akaike Information Criterion (AIC c ; Burnham and Anderson 2002) values, where models with lowest AIC c scores with ΔAIC c values less than 2 were assumed to be competitive at explaining the pattern in ambient temperatures in artificial nests.Model averaging was used to compute weighted average parameter estimates for explanatory variables retained in the best supported or set of models with competitive ΔAIC c values (Burnham and Anderson 2002), and the coefficients used to infer the direction and magnitude of the effect of the retained predictors on the response.
May had the greatest unit increase in temperature (0.03°C) per year, equating to an approximate 1.53°C increase in mean temperature for that month over a 51-year period (Figure 2).June and August were similar (0.027°C and 0.028°C respectively), whereas July showed the smallest significant increase of 0.018°C per year.There was a significant positive trend in the annual total number of extreme heat days throughout the timeseries (Τ = 0.298, P = 0.002), with a unit increase of 0.105 days per year (Figure 3A).Extreme heat events tended to occur during the auklet nesting season between the months of May and September (Figure 3B).Further demonstrating the pattern of increasing temperatures on the island, there was also a significant positive increase in the annual mean of the maximum temperature for the auklet breeding season (Τ = 0.296, P = 0.002), with a unit increase of 0.031°C per year (Figure 3C).Three models were considered competitive for explaining variation in deviation from ambient temperatures between wooden boxes and ceramic modules (ΔAIC c < 2; Table 1).All 3 models contained the interaction of area and extreme heat, with model 2 containing a 3-way interaction between treatment, extreme heat event, and area, and model 3 the addition of the main effects for treatment (Table 1).Given no single model received overwhelming support, inferences were made from model averaged coefficients and standard errors for parameters retained in the 3 competitive models (Table 2).On average, both wooden boxes and ceramic modules at the Carpentry Shop area showed a greater positive deviation from ambient air temperature compared to nests at the Cormorant Blind and Lighthouse Hill areas, regardless of whether it was an extreme heat day (Figure 4).
The estimated mean differences between wooden boxes and ceramic modules were similar across all 3 areas during non-extreme heat days, and at the Cormorant Blind and Lighthouse Hill areas during extreme heat days.
Ceramic modules showed a markedly smaller difference from ambient air temperature compared to wooden nest boxes during extreme heat days at the Carpentry Shop area, explaining the importance of the 3-way interaction between predictors.
F I G U R E 2 Trends in mean monthly ambient air temperature on Southeast Farallon Island, USA for the core Cassin's auklet nesting season from May through September across years 1971 to 2022.September showed no significant trend.

DISCUSSION
Our work confirms an increasing trend in ambient air temperatures and number of extreme heat days at this seabird colony off the California Coast, highlighting the importance for managers to develop strategies to help mitigate the impacts of climate change.Of particular concern is the increasing trend in the number of days where ambient air F I G U R E 3 Trends in warming of ambient air temperature on Southeast Farallon Island, USA from 1971 to 2022.Panel A depicts total number of extreme heat days (max temperature ≥20°C) for each year, with a positive significant slope of 0.105 (dashed line).Panel B illustrates when these extreme heat days occurred by month for each year (warmer colors = more days, cooler colors = fewer days), with a red box that define the spatial extent of this study.Panel C depicts the annual mean of monthly maximum temperatures for the auklet breeding season (±standard deviation line range), with a significant positive slope of 0.031 (dashed line).
T A B L E 1 Models representing a cumulative AIC c weight (Wt) of 1 from the candidate set, ranked from lowest ΔAIC c to highest.Predictors include treatment (wooden nest box or ceramic module), area (cb, lhh, cs), and whether the day represented an extreme heat event (extreme; TRUE or FALSE), for Cassin's auklet nest boxes on Southeast Farallon Island, USA for the months of March through August from 2017 to 2022.Number of parameters (K) and log-likelihood (LL) of each model are shown for completeness.Response was daily deviation of mean nest box temperatures from mean ambient air temperatures, and all models contained a random intercept for nest ID.See Table S1 in supporting information for the full candidate set.Both shaded ceramic modules and shaded wooden boxes exhibited similar deviation from ambient air temperature, with ceramic modules performing slightly better at the Carpentry Shop area during extreme heat days.
A lack of evidence for a major difference in the performance between the 2 materials tested does not completely rule out the hypothesis that the use of ceramic as a building material reduces thermal compared to wood.The value of ceramic modules became apparent when the combination of area on the island and whether the ambient air temperatures constituted an extreme heat event was factored into the analysis.Increased thermal buffering by clay modules compared to wooden boxes at the Carpentry Shop during extreme heat events suggests that localized temperature microclimates may be important for determining suitability of artificial habitat.The wooden boxes performed similarly to ceramic modules in locations that received cool, damp wind off the ocean, in hot dry sites during non-extreme heat days, or locations in a sheltered northern aspect.Including a shade structure to the wooden boxes beginning in 2008 had a major impact on keeping wooden boxes relatively cool compared to unshaded ones (Kelsey et al. 2016), demonstrating a simple affordable solution to help regulate internal temperatures at a similar rate to shaded ceramic modules under historically normal environmental conditions, but not during extreme heat events at warmer microclimates on the island.
Both wooden boxes and clay models showed a greater deviation from the control ambient air temperature at the Carpentry Shop area compared to other locations on the island.Given the construction and installation of boxes and modules were the same across all areas, this trend implies the Carpentry Shop is a warmer microclimate on the island.The performance of shaded wooden boxes declined at the Carpentry Shop area during extreme heat events, a location that is more arid in nature with a southern aspect that receives higher solar input throughout the day than other portions of the island.Double-layered ceramic nesting modules deployed at a colony of African penguins (Spheniscus demersus) in South Africa maintained higher levels of relative humidity compared to other habitat types (Welman and Pichegru 2022).Although not measured in our study, ceramic modules may have retained higher humidity levels compared to wooden boxes as well, explaining the ability to buffer against extreme heat even at the relatively drier sites on the island.Holes installed on the ceiling of the ceramic modules likely provided way to vent heat from the nesting chamber compared to the more sealed-off wooden boxes, contributing to the cooling effect of ceramic modules during extreme heat at the hottest location on the island.Burying the nest boxes would likely further the thermal buffering properties of either material, but given the rocky substrate of the Farallon Islands this was not an option.
Artificial nests are commonly used to facilitate monitoring, restore habitat, translocate and maintain colonies, increase breeding success, and increase adult survival of burrow-nesting seabird species (Wilson 1986, Libois et al. 2012, Carle et al. 2019).While there are challenges in the use of artificial nests in bird studies, including potential increased rates of predation (Moore and Robinson 2004), durability and longevity of materials used, and potential sampling biases (Lambrechts et al. 2010), auklets on the Farallon Islands provide an example of a species with extremely high rates of occupancy, breeding success, and quality data obtained from artificial nest boxes (Johns et al. 2022).We demonstrated how consideration of design and materials used in artificial nesting habitat, combined with knowledge of microclimate differences in breeding habitat, can alleviate the potential negative impacts of artificial breeding habitat for burrow nesting species.The results on heat transfer and retention presented here can be used to guide future efforts in developing durable artificial ceramic habitat modules.
An important practical advantage ceramic has over wood is that the material will last for hundreds of years.This durability benefits programs with limited personnel time and infrequent colony visits required to make repairs/ replacement that prevent wooden boxes from becoming hazardous to nesting adults and chicks.On Año Nuevo Island, for example, ceramic artificial nests have been observed by researchers in the field to be durable enough to withstand trampling by California sea lions (Zalophus californianus), brown pelicans (Pelecanus occidentalis), and Brandt's cormorants (Urile penicillatus), thus providing safe and stable breeding habitat on this colony with dense wildlife eroding the soil and competing for space.We suggest that ceramic modules with protective design features such as glazed shade coverings and ventilation holes are the preferred solution for artificial nesting habitat for burrow nesting species over more traditional wooden boxes, particularly as ambient air temperatures continue to rise or if locations are periodically exposed to harsh dry conditions.

F
I G U R E 1 Map of Southeast Farallon Island, USA showing clusters of Cassin's auklet nest boxes (colored points), with labels denoting the three areas used in this study.Red dot shows the location of the weather station used to measure daily ambient air temperatures.The arrow in the upper left depicts direction of the prevailing northwest winds in the spring, and color ramp with contour lines show the topographic relief of the island and Lighthouse Hill.Tan lines show the trail system and black polygons show building footprints.USE OF CERAMIC NEST MODULES IN A WARMING WORLD | 3 of 12 temperatures exceed 20°C, given even a few days of extreme heat can result in heat exhaustion and death in birds occupying artificial nesting habitat, as observed in 2008 on the Farallones(Warzybok and Bradley 2008).

F
I G U R E 4 Difference in mean daily temperature of wooden boxes (grey) vs. ceramic modules (red) from mean daily ambient air temperature for artificial nest boxes used by Cassin's auklets on Southeast Farallon Island, USA for the months of March through August from 2017 to 2022.Data are presented for Cormorant Blind (cb), Carpentry Shop (cs), and Lighthouse Hill (lhh), separated by panels for normal conditions (extreme heat day = FALSE) and extreme heat events (TRUE).Boxplots show the interquartile range bounded by the 25th and 75th percentile, horizontal line in the middle of each box shows the median, whiskers the maximum and minimum values, and diamonds the extreme outliers.Predictions from model averaged coefficients for each scenario presented as black dots, with error bars showing 95% confidence intervals.
23285540, 0, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1501,Wiley Online Library on [18/12/2023].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1501,Wiley Online Library on [18/12/2023].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License T A B L E 2 Model averaged coefficient estimates (with associated standard errors and confidence intervals) used to infer how variation in deviation from ambient temperature in artificial Cassin's auklet nest boxes on Southeast Farallon Island, USA was influenced by nest type (treatment wooden box or ceramic module), area on the island (cs: Carp Shop, lhh: Lighthouse Hill, cb: Cormorant Blind), and occurrence of extreme heat events (TRUE or FALSE).Data span the months of March through August from 2017 to 2022.Reference levels were area (cb, treatment), wooden box, and extreme heat (FALSE).Variance (Standard Deviations) for random intercept term box ID was 1.165 (1.079).