Microclimate of Natural Cavity Nests and Its Implications for a Threatened Secondary-Cavity-Nesting Passerine of New Zealand, the South Island Saddleback

Abstract.

Secondary-cavity-nesting birds occur widely throughout the world, but little information is available on the benefits of the nest's microclimate for such species, particularly for those using natural cavities. We investigated the influences of microclimate on a threatened secondary-cavity-nesting passerine, the South Island Saddleback (Philesturnus carunculatus carunculatus). Our aims were to determine whether (1) saddlebacks select tree cavities with microclimates less variable than those of other tree cavities in their surrounding territory, (2) whether structural aspects of tree cavities translate into certain microclimate characteristics, and (3) if less frequently used sites not in tree cavities (e.g., cavities in banks or in vegetation) have thermal properties similar to those of tree-cavity nests. We found that the saddleback's tree-cavity nests were more stable in temperature, more insulated against cold, and did not change temperature as rapidly as the ambient air or unused tree cavities. Regression analysis showed that of structural characteristics of tree cavities examined, only one, entrance width, was significantly associated with an aspect of microclimate (minimum temperature). Additionally, we found that regardless of cavity type the thermal properties of saddleback nest cavities were similar. These results indicate that saddlebacks likely select nest cavities with less variable thermal properties that are potentially beneficial, and future studies experimentally manipulating the variability of microclimate may be fruitful in determining the effect of microclimate on reproductive success. Nevertheless, this study is one of the first to demonstrate microclimate as a factor determining selection of natural nest cavities over available unused cavities.

Resumen.

Las especies de aves no excavadoras que nidifican en cavidades se distribuyen ampliamente en el mundo, pero se sabe muy poco sobre los beneficies del microclima del nido para estas especies, particularmente para aquellos que utilizan cavidades naturales. Investigamos las influencias del microclima sobre Philesturnus carunculatus carunculatus, una especie paseriforme no excavadora que nidifica en cavidades. Nuestros objetivos fueron determinar si (1) los individuos de P. c. carunculatus seleccionaron las cavidades en árboles con menor variación microclimática con relación a las demás cavidades existentes en el territorio circundante, (2) los aspectos estructurales de las cavidades se tradujeron en determinadas caracteristicas microclimáticas y (3) los sitios utilizados con menor frecuencia que no se encuentran en cavidades en árboles (e.g., cavidades en barrancos o la vegetación) tienen propiedades térmicas similares a las de los nidos en cavidades en árboles. Encontramos que los nidos de P. c. carunculatus en cavidades tuvieron temperaturas mas estables, tuvieron un mayor aislamiento contra el frío y no variaron de temperatura tan drásticamente como la temperatura ambiente o de las cavidades no usadas. Los análisis de regresión mostraron que de las características estructurales examinadas, sólo la entrada de las cavidades en árboles estuvo significativamente asociada con un aspecto del microclima (temperatura mínima). Además, encontramos que independientemente del tipo de cavidad, los nidos de P. c. carunculatus tuvieron propiedades térmicas similares. Nuestros resultados sugieren que esta especie selecciona cavidades con propiedades térmicas menos variables, que son potencialmente beneficiosas. Estudios futures que manipulen experimentalmente la variabilidad del microclima podrían ser útiles para determinar los efectos del microclima sobre el éxito reproductive. Sin embargo, este estudio es uno de los primeros en demostrar que el microclima es un factor importante en la selección de cavidades naturales con relación a las cavidades disponibles no utilizadas.

Introduction

For successful reproduction, avian nests have three microclimatic requirements: the appropriate temperature, humidity, and composition of respiratory gas. Temperature, however, is regarded as the component most crucial for successful reproduction in birds (Walsberg 1980, Wachob 1996, Kern and Cowie 2000, Ar and Sidis 2002, Lill and Fell 2007). Given that temperature influences reproductive success, birds are expected to take steps to select sites that provide a favorable microclimate (van Riper et al. 1993, Gloutney and Robert 1997, Burton 2006, Tieleman et al. 2008). Numerous aspects of nest sites have been shown to contribute to a favorable microclimate, including levels of solar radiation, thickness of tree walls, density of surrounding vegetation, orientation, cavity size, bark type, and exposure to rain (Wachob 1996, Hooge et al. 1999, Reid et al. 2000, Rauter et al. 2002, Radford and Du Plessis 2003, Ardia et al. 2006).

Cavities have long been known to provide energetic benefits to birds (Kendeigh 1961). Birds save substantial energy by roosting in cavities, especially during inclement weather (Caccamise and Weathers 1977, Du Plessis et al. 1994, Du Plessis and Williams 1994, Cooper 1999). The advantages of roosting within a cavity are likely to translate to breeding in a cavity as well, as reproduction is one of the periods most energetically costly for birds (Gustafsson et al. 1994). Energy balance is critical during reproduction, as birds must incubate eggs to a temperature sufficient for embryonic development and minimize thermoregulatory costs to nestlings (Webb 1987, Wachob 1996, Reid 2000, Dawson et al. 2005).

Studies have shown that cavity-nesting birds benefit reproductively from a favorable microclimate, but these studies are largely restricted to nest boxes (Blem and Blem 1994, Wachob 1996, Stamp et al. 2002), which may confound results, as nest boxes do not replicate all of the characteristics of natural nests (McComb and Noble 1981, Møller 1989). Results of previous studies indicate that microclimate can influence the reproductive success of species nesting in natural cavities, but these studies are largely limited to primary-cavity-nesting birds (Hooge et al. 1999, Wiebe 2001). Because they excavate their own nests, primary-cavity-nesting birds likely are less constrained during cavity selection than secondary-cavitynesting birds, so those studies of microclimate selection are likely less informative (Wiebe 2001). Studies of microclimate in natural cavity nests of secondary-cavity-nesting birds are lacking (Albano 1992, Wachob 1996). Natural cavities seem to vary in quality, but whether secondary-cavity-nesting birds can and do select natural nest cavities for microclimatic benefits remains uncertain (Kesler and Haig 2005, Paclík and Weidinger 2007). We address this issue by asking: Do microclimates of natural nest cavities selected by the secondary-cavity-nesting South Island Saddleback (Philesturnus carunculatus carunculatus) differ from unused cavities within their territories? Do particular characteristics of nest cavities translate into specific microclimate characteristics? And do cavity nests not in trees provide thermal properties comparable to those of tree-cavity nests?

Methods

We studied a color-banded population of South Island Saddlebacks on Ulva Island, New Zealand (259 ha, 46° 56′ S, 168° 08′ E, highest point 74 m above sea level). To facilitate the eradication of Norway Rats (Rattus norvegicus), a grid system had previously been cut through the island's understory vegetation, with tracks spaced at 100-m intervals and bait stations placed along each track at 100-m intervals. We used these tracks and bait stations to divide the island into 100-m2 sections to search for breeding pairs of saddlebacks in the forest. In the central part of the island the forest is mature temperate podocarp—hardwood with the main canopy trees consisting of totara (Podocarpus cunninghamii), miro (Prumnopitys ferruginea), rimu (Dacrydium cupressinum), and southern rata (Metrosideros umbellata). This central forest is surrounded mainly by coastal scrub dominated by leatherwood (Olearia colensoi) and inaka (Dracophyllum longifolium).

The South Island Saddleback is a threatened poor-flying passerine (family Callaeidae) endemic to New Zealand. Saddlebacks are facultative cavity nesters and nest in a variety of substrates, including tree cavities (and nest boxes), ground cavities, the crowns of tree ferns (Dicksonia squarrosa), leaf litter in banks overlooking the ocean, clumps of vegetation, or, occasionally, in open-cup nests in tree forks, but typically saddlebacks use tree cavities (Hooson and Jamieson 2003a). Like all passerines breeding in New Zealand, saddlebacks are nonmigratory and remain in pair bonds year round. Saddlebacks have a limited ability to coexist with introduced terrestrial mammalian predators and have completely disappeared from New Zealand's main islands and many near-shore islands (Hooson and Jamieson 2003b, Wilson 2004) but were reintroduced to Ulva Island in 2000 after rats had been eradicated (Lovegrove 1996, Hooson and Jamieson 2004).

For this study from September to February (austral summer) in 2006–2007 (hereafter abridged 2007) and again in 2007–2008 (hereafter abridged 2008) we located saddleback nests by following pairs in territories established in previous breeding seasons as well as those located in new territories. We usually found nests as the female entered to incubate or as both sexes returned to feed nestlings. We then marked the nests with flagging tape within several meters and checked them periodically (longest interval was 7 days) to establish approximate dates for banding nestlings. After the young had fledged, we recorded numerous measurements for each tree-cavity nest and also for the closest most suitable unused tree cavity. We located the unused cavity by searching from the ground for the nearest cavity with a nonvertical entrance >6 cm, large enough to hold nest and bird, and dry inside. We classified saddleback nests as either bank cavities (holes in the vegetation and ground of coastal bluffs), debris cavities (holes in dead vegetation above the ground), or tree cavities, but we measured only the more frequently used tree cavities. Treecavity measurements included vertical height and vertical depth (both measured from the bottom of the entrance hole), horizontal depth (measured from the bottom of entrance hole to the back of the cavity), horizontal width (distance between walls at the bottom of the entrance hole), the tree's diameters at cavity height (DCH) and at breast height (DBH), and height of the entrance hole above the forest floor. If the cavity had more than one entrance, measurements were taken at one of the holes selected at random. We identified the tree's species and whether it was dead or alive.

We recorded microclimate data with Hobo ProSeries data loggers (Onset Computer Corp., Pocasset, MA), which recorded ambient temperature and the temperature inside each tree cavity. We recorded temperatures after the completion of first clutches during both breeding seasons (8 January—13 February 2007 and 7–24 February 2008). To record ambient temperatures we placed data loggers inside simple Stevenson's screens (device to shield weather instruments from rain and direct sun) and placed them equidistant between tree-cavity nests and unused tree cavities at approximate mean nest height (∼2 m). To record cavity temperatures we first removed the nests so that temperatures would be equivalent to those birds would experience when selecting nest sites. We then positioned data loggers in both nest cavities and unused cavities at the level of the nest cup (or the inferred level of the nest cup in unused cavities). Temperatures of each type (nest, unused cavity, ambient) were simultaneously recorded every 5 min over the same 5-day period. Because we had only 10 data loggers, we placed them at nest sites as the first nesting attempts were completed and then transferred them to other available sites until data were recorded for all first-clutch cavity nests. Additionally, during the 2008 breeding season we recorded a sample of temperatures from cavity nest sites not in trees (n = 8; see classification above).

Statistical Analysis

We used a mixed (between and within subjects) ANOVA to analyze temperature differences among tree-cavity nests, unused tree cavities, and ambient air (Tabachnick and Fidell 2007). For each site, we averaged the temperature for each hour (hr 1 = 00:00–00:55, etc.) and used these hourly measures to generate single descriptive temperature variables, mean (T_mean: mean of all hourly averages), maximum (T_max: maximum of hourly average values), minimum (T_min: minimum of hourly average values), range (T_range: equals T_max — T_min) and rate (T_rate: absolute value of hr₂ — hr₁, etc.) for each 24-hr day in the 5-day period. We analyzed each of the descriptive temperature variables in parallel mixed betweenand within-subjects ANOVA with each day (n = 5) serving as a repeated measure (within-subject effect) in the model. We included site as a categorical independent between-subjects variable. We analyzed years separately and included contrast statements to compare differences among tree cavity nests, unused tree cavities, and ambient temperature for each of the five descriptive temperature variables (Table 1).

To analyze if characteristics of tree-cavity nests influenced the five descriptive temperature variables, we used a stepwise multiple regression. However, we excluded some nest-tree characters (see below) that were significantly correlated, and we also combined nests from 2007 and 2008 for regression analysis, as none of these measurements in 2007 differed from those in 2008 (t = -0.482–1.051, P > 0.05). For each nest we averaged the descriptive temperature variables over the five 24-hr periods to generate a single value to serve as the dependent variable in our analysis. We analyzed each of these descriptive temperature variables singly in our stepwise multiple regression analysis against three characteristics (entrance hole's width, cavity's horizontal width, tree's DCH), which were entered into the model simultaneously. We included variables in the model if P < 0.05 and excluded them if P > 0.05.

We examined differences in microclimate among cavity types by using repeated-measures linear mixed models (West et al. 2007). To increase the power of this longitudinal analysis we combined the various types of cavity nests that were not in trees into one category, as preliminary analysis revealed no significant difference in the five temperature variables among these nest types (P > 0.10). A cavity nest not in a tree was then selected and paired with the nearest tree-cavity nest to control for potential differences in habitat structure and exposure across the island (Michel 2006). We included terms for cavity type (1 = tree cavity, 2 = not tree cavity), day (1, 2, … 5) and their interaction (cavity × day) as fixed factors in our model and used a diagonal covariance structure for the repeated effects in the model. Day served as the repeated measure in our model, and each of the five descriptive temperature variables were analyzed separately. We used pairwise comparisons to evaluate estimated marginal means for significant differences between cavity types after performing Bonferroni adjustments for multiple comparisons, and the nominal significance level before Bonferroni adjustment was P < 0.05. We analyzed all data with SPSS 16.0 (SPSS Institute, Inc., 2007).

Results

Saddleback nests (n = 84) were in a variety of substrates including banks (15.1%), debris piles (8.6%); one was an opencup nest (1.0%). Over the 2 years of the study, however, the majority were in tree cavities (75.3%). These cavities (n = 70) were in southern rata (24.3%), kamahi (Weinmannia racemosa, 22.9%), broadleaf (Griselinia littoralis, 21.4%), totara (7.1%), lancewood (Pseudopanax crassifolius, 1.4%), rimu (1.4%), and tree fern (Dicksonia squarrosa, 1.4%). Dead trees (species unidentified, 20.0%) composed the remainder of cavity-nest sites.

The microclimates of tree-cavity nests varied less than those of the ambient air and of unused tree cavities (n = 34 nests, Table 1). The main effect comparing differences among sites was significant for most descriptive temperature variables (T_mean, T_max, T_min, T_rate) in both years (F = 1.854– 19.584, P = <0.001–0.034), except for T_range in 2007 (F = 1.352, P = 0.187, Table 1). The significance of the main effect of nest type differed by year and the five descriptive temperature variables, with nest cavities usually possessing superior thermal qualities. Compared to unused tree cavities, nest cavities had more stable temperatures (smaller range), were more insulated against the cold (higher minimum), and had lower rates of hourly change (Table 1). Structurally, used tree cavities were similar to unused tree cavities, but vertical depth, height of entrance hole, and the tree's DBH differed, as unused cavities were deeper, typically lower to the ground, and in trees with significantly smaller diameters (Table 2).

Only T_min was associated with any of the three treecavity characteristics examined in our regression analysis of nest cavities (n = 18). Width of entrance hole was negatively associated with minimum temperature (r² = 0.33, F = 7.98, P = 0.012), i.e., the minimum temperature was lower in cavities with wider entrances than in those with smaller entrances. There were no other associations between any of the remaining descriptive temperature variables (T_mean, T_max, T_range, T_rate) and the tree-cavity characteristics.

For our comparison of microclimates of tree cavities with those of other cavity types, we located 45 first-clutch saddleback nests during the breeding season (October 2007—January 2008). During this study year a majority of birds nested in tree cavities (60.0%), but nesting in other cavity types was common with 26.7% of pairs nesting in bank cavities and 13.3% nesting in debris cavities. Because our 10 data loggers were devoted primarily to recording tree-cavity microclimate, however, we collected data from only a randomly selected subset of other cavity types (n = 8). We paired each of the eight randomly selected non-tree-cavity nests with the nearest tree-cavity nest and recorded temperatures simultaneously at both locations. Our results indicate that within the 5-day periods of recording all the five temperature variables except rate of change differed significantly; we found no significant differences, however, among cavity types or the interaction of cavity type and day (Table 3). Neither did we find significant thermal differences between the two categories of substrate, as tree cavities and cavities in other substrates did not differ significantly in any of the five descriptive temperature variables (Table 4).

Discussion

Cavities appear to buffer eggs, nestlings, and attending adult saddlebacks against oscillations in ambient temperature. Tree-cavity temperature differed from ambient temperature, with ambient temperature changing more and reaching lower minima. These fluctuations in ambient temperature are potentially detrimental to reproduction, as ambient temperature usually does not reach levels suitable for development of offspring (Webb 1987), especially in the cool, temperate climate of southern New Zealand (O'Donnell 2002). The vast majority of birds use contact incubation to afford the proper temperatures for development of their offspring and many construct nests, which play a role in insulating eggs, nestlings, and adults from ambient temperatures. Even if most nest types insulate against ambient temperature, the protective nature of and variability among tree cavities may allow rigorous selection of microclimate (Sedgeley 2001).

Variability in Tree Cavities

It is not surprising that concealing a nest within a tree cavity reduces heat loss over ambient conditions. Of greater importance is whether variability in tree cavities leads to a disparity in cavity microclimate that birds are able to recognize, which our data suggest. Structural variability seems to be common among natural tree-cavity nests and may be critical in determining if cavities are used or remain unoccupied (Lõhmus and Remm 2005, Remm et al. 2006). Tree cavities may occur in abundance within forests but only small percentages are suitable for nesting (Lõhmus and Remm 2005, Blakely et al. 2008). Structural characteristics likely are important in determining this suitability, as many properties of tree cavities directly relate to factors such as cavity predation and microclimate. The ability to evaluate these factors may allow prospecting birds to assess the breeding potential of available cavities.

Numerous characteristics, ranging from the nest site to landscape level, have been shown to influence the thermal environment within tree cavities (Wiebe 2001, Radford and Du Plessis 2003, Sedgeley 2006, Paclík and Weidinger 2007). Trees with larger diameters, for example, are thought to mitigate thermal oscillations because they store heat better and they possess thicker walls, which are more able to buffer internal and external temperature differences. We did not, however, test microclimate's association with DBH because of limitations of sample size and because measurements of DCH indicated no significant differences in size between treecavity types. Others factors, such as greater exposure to solar radiation, have also been shown to enhance microclimate in tree-cavity nests, as trees are able to gather more thermal energy. The dimensions of tree cavities also are known to affect internal temperatures because tree cavities with larger volumes retain greater quantities of heat (Paclik and Weidinger 2007), although we found that tree cavities with nests were significantly smaller than unused tree cavities.

We did not examine characteristics beyond the site level; however, we did find that the width of the entrance hole of tree-cavity nests was negatively associated with minimum temperature, with minimum temperature explaining a moderate proportion of our model. The association of treecavity-entrance width and temperature agrees with both Sedgeley (2001) and Paclík and Weidinger (2007) who found that smaller entrances were better at maintaining internal temperature. This association may occur because wider entrances allow more convection, which dissipates heat quicker to facilitate cooling of the cavity. Smaller entrances also may enhance microclimate by limiting exposure to rainfall, which can create a negative thermal environment for nestlings similar to the effects of lower temperatures through dampening (Radford and Du Plessis 2003).

We failed to detect any significant association of the other nest cavity characteristics examined (horizontal width, DCH) with microclimate although some of these characteristics have been associated with microclimate in other studies (Hooge et al. 1999, Wiebe 2001, Paclík and Weidinger 2007). In some cases the nesting female's body heat may be required to detect microclimatic differences among cavities, as a higher gradient would make any differences in insulative capacity more evident (Paclík and Weidinger 2007) and would also increase the ability to detect those differences (but see Ardia et al. 2006). Additionally, we recorded temperatures later in the breeding season when temperatures were likely higher and less variable. Therefore, some associations may have been more pronounced earlier in the breeding season, when the benefits of the insulative capacity of tree cavities would be more evident. In addition, characteristics such as entrance size and height of entrance from the forest floor are known to enhance the quality of cavity nests in ways other than stabilizing microclimate. Higher cavities with smaller entrances, for example, are known to reduce predation (Albano 1992, Hooge et al. 1999).

Our results demonstrate that tree cavities selected by South Island Saddlebacks have more stable microclimates than unused tree cavities in their territories. This study is one of the first to demonstrate that a secondary-cavity-nesting bird selects natural tree cavities with potentially greater microclimatic benefits than nearby unused tree cavities. Tree cavities buffered thermal oscillations allowing saddlebacks to experience higher minimum temperatures and lower rates of temperature change, which both may minimize the demands of thermoregulation on parents and offspring. The broader implications are, as we have seen in other studies (see Sedgeley 2001, Weibe 2001, Paclík and Weidinger 2007), that certain structural aspects of tree cavities, particularly those associated with the entrance hole, may influence temperature and cavity selection and may provide a beneficial nest microclimate for saddlebacks.

Cavity Nests Not in Trees

Although the majority of saddleback nests were located in tree cavities, nests in other cavities (e.g., cavities in banks or vegetation debris) were still common. Tree-cavity nests and other cavity nests we studied did not diverge in their thermal properties, as the substrate types did not differ in five related temperature variables. Exactly what aspects of cavities in banks and vegetation debris allowed them to equal the thermal properties of tree cavities in this study is unknown, as high-quality tree cavities are thought to provide better thermal characteristics (see above). Thermal properties of the cavity types may have been similar because the cavity nests not in trees were typically concealed within vegetation, which may reduce the negative effects of moisture (increases thermoregulation costs) on eggs and nestlings and may slow the rate of heat loss caused by convection (Radford and Du Plessis 2003). Even if the thermal properties of bank- and debris-cavity nests are similar, such sites likely have other negative aspects (Camprodon et al. 2008). For example, they are typically lower than tree cavities, if not in the ground, and so may be at greater risk of nest predation (Joy 2000, Camprodon et al. 2008). Nonetheless, our data show that saddlebacks occasionally do nest in these other cavity sites with comparable thermal properties, suggesting that tree cavities may not be necessary to provide saddlebacks with a thermal environment suitable for nesting.

General Conclusions

Given the abundance of literature devoted to cavity nesting, the lack of investigations into the importance of microclimate is surprising. Our study is one of the first to demonstrate that secondary-cavity-nesting birds breeding in natural tree cavities select nest sites whose microclimate is less variable than in other tree cavities available in their territories. To examine the reproductive consequences of cavity microclimate, experimental manipulation may be necessary, as these birds may be proficient in selecting natural nest cavities with suitable microclimates and in the absence of experimental manipulation, variation in cavity microclimate may be too small for any significant difference in reproductive success to be distinguished, especially if the number of nesting pairs examined is low. Regardless, recent experimental studies have shown increases in reproductive success or nestling quality with a more beneficial (artificially induced) microclimate (Dawson et al. 2005, Pérez et al. 2008). Therefore, selection of nest sites with less microclimate variation may have long-term fitness consequences for parents and their offspring and the effects of microclimate on cavity-nesting birds may have been underestimated. Our results also suggest that other cavity types (e.g., ground cavities), which for secondary-cavity-nesting forest birds have been largely ignored, may provide nesting sites suitable at least in terms of temperature regulation. Yet the degree to which microclimate affects selection of a natural cavity remains obscure, as most likely a balance of factors contributes to the choice (Albano 1992, Fisher and Wiebe 2006, Remm et al. 2006). How selection for a less variable microclimate is associated with other factors relating to cavity choice, such as protection from predators or intra- and interspecific competition (Remm et al. 2006, Aitken and Martin 2007), remains to be explored.

Acknowledgments

We thank the Department of Zoology, University of Otago, and the Department of Conservation (research contract no. 3575 to IGJ) for providing funding for equipment and travel. We especially thank Brent Sevan and Phred Dobbins and the Department of Conservation, Stewart Island, for providing logistic support during the field season on Ulva Island. We also thank Lisa Hegg, Stephanie Hicks, Sandra Hoeder, and several volunteers for assisting with nest finding. Thanks must also go to two anonymous reviewers for comments that improved the quality of the manuscript.

Literature Cited