Wood hardness drives nest-site selection in woodpeckers of the humid Chaco

ABSTRACT

Avian excavators (woodpeckers and other species) select nest sites based on the characteristics of the nest patch, nest tree, and substrate. These characteristics could increase foraging opportunities or reduce the risk of predation, but there is also a potentially important role for wood hardness in restricting nest-site selection, a role that has been little explored and is expected to vary among species according to their ability to excavate. We examined patterns of nest-site selection in 8 woodpecker species in the humid Chaco of South America, where the dominant trees have extremely hard wood. We hypothesized that (1) wood hardness is the main factor driving the selection of excavation sites, and (2) interspecific variation in body size and foraging behavior (traits frequently related to excavation ability) explain interspecific variation in the wood hardness of excavated nest substrates. From 2016 to 2019 in well-preserved forests of the Argentine Chaco, we compared nesting cavities excavated in wood (cases; n = 42) vs. potential wood substrates without cavities (matched controls) and made 187 focal observations of foraging woodpeckers. Woodpeckers selected nesting substrates with softer heartwood than potential substrates, regardless of any other characteristics of the tree or nest patch. Wood hardness around nest cavities increased with body size and the prevalence of chiseling during foraging, traits that were positively correlated. Woodpeckers often excavated in Prosopis spp. (Algarrobos) but rarely in Schinopsis balansae (Quebracho), a tree with exceptionally hard internal wood, in which cavity adopters frequently nest in non-excavated cavities. Wood hardness is critical to assessing the role of woodpeckers as cavity facilitators, understanding the costs and benefits of excavation, and interpreting excavation patterns across species and forests globally.

RESUMEN

Las aves excavadoras de cavidades (pájaros carpinteros y otras especies) seleccionan sitios de nidificación basándose en características del parche de nidificación, el árbol nido y el sustrato. Estas características podrían aumentar las oportunidades de forrajeo o reducir el riesgo de la depredación, pero la dureza de la madera también juega un papel importante en restringir la selección de sitios, un aspecto poco explorado y que se espera varíe entre especies según su capacidad de excavación. Examinamos los patrones de selección de sitios de nidificación en ocho especies de pájaros carpinteros en el Chaco húmedo de Sudamérica, donde los árboles dominantes tienen una madera extremadamente dura. Hipotetizamos que 1) la dureza de la madera es el factor principal en la selección de sitios para excavar, y 2) la variación interespecífica en el tamaño corporal y los comportamientos de forrajeo (rasgos frecuentemente relacionados con la capacidad de excavación) explican la variación interespecífica en la dureza de la madera excavada en los sustratos de nidificación. De 2016 a 2019 en bosques bien conservados del chaco argentino, comparamos cavidades-nidos excavadas en árboles (casos; n = 42) vs. sustratos potenciales en árboles no excavados (controles) y realizamos 187 observaciones de estrategias de forrajeo de pájaros carpinteros. Los pájaros carpinteros seleccionaron sustratos nidos con duramen más blando que los sustratos potenciales no usados, independientemente de cualquier otra característica del árbol o parche de nidificación. La dureza de la madera alrededor de la cavidad nido incrementó con el tamaño corporal y la prevalencia de cincelado durante el forrajeo, rasgos positivamente correlacionados. A menudo, los pájaros carpinteros excavaron en árboles de Prosopis spp. (Algarrobos) pero raramente en Schinopsis balansae (Quebracho), árbol con madera interna excepcionalmente dura, en el cual aves adoptadoras frecuentemente nidifican en cavidades no excavadas. La dureza de la madera es fundamental para comprender el rol de los pájaros carpinteros como facilitadores de cavidades, entender los costos y beneficios de la excavación, e interpretar los patrones de excavación entre especies y entre distintos bosques del mundo.

Graphical Abstract

Open in new tabDownload slide

INTRODUCTION

Nest-site selection is generally assumed to be an adaptive process, in which birds choose sites with characteristics that confer a nesting advantage; however, birds can also be constrained by the characteristics of available nest sites (Mezquida 2004, Latif et al. 2012). Nest-site selection by avian excavators (sometimes called primary cavity nesters) directly influences the location and characteristics of cavities that will then be available to cavity adopters (sometimes called secondary cavity nesters; Daily et al. 1993, Trzcinski et al. 2022). We use “excavators” to indicate bird species that facultatively or obligately excavate in wood or other arboreal substrates, and “adopters” to indicate species that use existing arboreal cavities but do not excavate. Whereas cavity adopters may be constrained by available cavities, excavators may be constrained by substrates suitable for excavation (Daily 1993, Jackson and Jackson 2004, Martin et al. 2004). Although the ability to generate new cavities frees excavators from the constraints imposed by having to find an existing cavity, excavation is costly in terms of energy and time, requires a substrate that is sufficiently soft (e.g., wood previously colonized by wood-decaying fungi), and may be constrained by other characteristics of trees, such as their size (Ojeda et al. 2007, Wiebe et al. 2007, Sandoval 2008, Bonnot et al. 2009, Elliot et al. 2019). To understand nest-site selection by excavator birds, it is important to understand the extent to which this selection reflects an adaptive process to maximize their fitness vs. the constraints imposed by potential excavation sites.

Excavator birds may choose nest sites that optimize foraging opportunities in the vicinity, buffer ambient temperatures, or reduce nest predation. At the nest patch scale, excavators selected patches with high density and basal areas of trees that provide feeding opportunities, or sites with low canopy cover (which reduces patch use by arboreal predators; Bonnot et al. 2009, Politi et al. 2009, Saab et al. 2009, Kozma and Kroll 2012, Berl et al. 2014, 2015). At the nest tree scale, excavators selected large trees (large enough to hold the brood and protect it from predators; Kosiński and Winiecki 2004, Zhu et al. 2012, Albanesi et al. 2016, Basile et al. 2020, Jauregui et al. 2021). They also selected trees that had little contact with neighboring trees (which reduces predator access; Wiebe 2001, Aitken and Martin 2004, Cockle et al. 2011a, Albanesi et al. 2016). At the scale of the nest substrate (trunk or branch), excavators selected well-concealed sites (hiding the nest from visual predators; Berl et al. 2014, 2015), in large diameter substrates or with some cover (buffering temperature; Wiebe 2001, Jackson and Jackson 2004). Some excavators preferred certain tree species over others (Bonnot et al. 2009, Politi et al. 2009, Ćiković et al. 2014, Albanesi et al. 2016, Lammertink et al. 2020), and larger excavators required a larger-diameter substrate (Bai et al. 2005). In some studies, small and medium-sized excavators selected dead trees, whereas larger species selected live trees (Bai et al. 2005, Schaaf et al. 2020). However, none of the aforementioned studies included direct measurements of wood hardness, an important characteristic that could severely restrict the sites available for excavation.

Avian excavators require a substrate with optimum hardness: soft enough to be excavated but hard enough to resist weather and predators (Kilham 1971, Schepps et al. 1999, Tozer et al. 2009, Lorenz et al. 2015, Puverel et al. 2019). Substrates suitable for excavation are mainly produced in trees by fungal decay that weakens the internal wood (heartwood, corresponding to the cavity chamber; Conner et al. 1976, Daily 1993, Jackson and Jackson 2004, Robledo and Urcelay 2009, Cockle et al. 2012, Jusino et al. 2016). When excavating a nest cavity, birds first break through the outer sapwood (corresponding to the cavity wall) and then remove the softened heartwood (Kilham 1971, Jackson and Jackson 2004). Four studies examined the effects of wood hardness or density on nest-site selection by excavators, all in temperate woodpecker communities (2 in the Nearctic, 1 in the Palearctic, and 1 in the Neotropics; Schepps et al. 1999, Lorenz et al. 2015, Puverel et al. 2019, Jauregui et al. 2021). In 3 of these studies, wood hardness or density at excavated sites was the characteristic that most influenced nest-site selection. In radial profiles of excavated nest sites, the sapwood (wall) was harder than the heartwood (chamber; Matsuoka 2008, Lorenz et al. 2015). Although other studies have used the presence of fungal fruiting bodies or the decay class of the tree or branch as proxies for wood hardness, Lorenz et al. (2015) demonstrated that these measures are unrelated to wood hardness, which needs to be measured directly. Few studies have investigated how wood hardness constrains nest-site selection by excavators in the context of several other competing hypotheses and in systems with multiple tree and woodpecker species.

When examining the role of wood hardness in nest-site selection, it is important to consider the gradient of interspecific variation in the ability to excavate (Martin et al. 2004, Bunnell 2013, Lorenz et al. 2015). In practice, species are often categorized as “weak” (vs. “strong”) excavators if they are small-bodied, select trees with advanced decay, select small-sized branches, excavate high above the ground, or tend to reuse pre-existing cavities (Martin 1993, Dudley and Saab 2003, Martin et al. 2004, Kosiński and Kempa 2007, Edworthy et al. 2012, Bunnell 2013, Hebda et al. 2016, 2017). However, when Lorenz et al. (2015) measured wood hardness for 6 species of excavators, they found considerable overlap in the hardness of wood excavated by species previously categorized as “weak” and species categorized as “strong.” They highlighted the need to obtain quantitative wood hardness data to improve our understanding of the factors influencing interspecific variation in excavation ability, and to assess how excavation ability might influence cavity quality.

We can infer interspecific variation in excavation ability by examining the prevalence of foraging techniques, which are associated with anatomy (Bock 1999, Donatelli et al. 2014). Anatomies associated with heavy blows allow birds to access insects inside the wood, but they are negatively correlated with agility in seeking and gleaning. Frequent and forceful blows are associated with complex mandibular apparatuses, a high ratio between the width of the first cervical rib (insertion of neck and head muscles) and femur length (associated with insertion of climbing muscles), deep and wide maxillae, reduced distal leg bones, and a well-developed pterygoid protractor muscle (Spring 1965, Kirby 1980, Bock 1999, Donatelli et al. 2014, Chhaya et al. 2023). In contrast, gleaning or probing techniques require birds to climb extensively along tree trunks and branches, and climbing is supported by anatomical features (e.g., long legs) generally opposite to the features that facilitate striking hard blows (Bock and Miller 1959, Spring 1965). Furthermore, in some woodpecker communities, body mass is positively related to foraging techniques involving hard blows, and negatively to techniques involving climbing and gleaning (Lammertink 2007, Fernández et al. 2020). Thus, higher body mass, and the use of hard blows that perforate wood during foraging, probably indicate anatomies better adapted to excavation and suggest the ability to excavate cavities in harder substrates.

The Chaco, a subtropical forest with a high diversity of excavator birds (Di Giacomo 2005), represents an ideal system in which to study how wood hardness and other factors influence nest-site selection. Most trees in the Chaco are characterized by slow growth, very hard wood, and high tannin concentrations (López et al. 1987, Chave et al. 2006, Barberis et al. 2012); these characteristics help trees resist the fungal decay that facilitates the formation of tree cavities (Jackson and Jackson 2004, Puverel et al. 2019). Although there are few published studies, the nest predator community in the Chaco is diverse and includes snakes, raptors, toucans, woodcreepers, marsupials, mustelids, rodents, and felines (Di Giacomo 2005, Berkunsky et al. 2011, Salvador and Bodrati 2013, Menezes and Marini 2017, Di Sallo, personal observation). These aerial, arboreal, and terrestrial predators detect nests visually or chemically, fly or climb to the cavity, and access the brood by entering the cavity, inserting a body part, or breaking open part of the cavity (Short and Horne 2001, Di Giacomo 2005, Cockle et al. 2015, 2016). The Chaco is the second largest forested region in South America and is experiencing one of the highest rates of deforestation worldwide (Hansen et al. 2013, Kuemmerle et al. 2017), which makes understanding habitat selection important for conservation.

The present study focuses on a community of woodpeckers in the humid Chaco of Argentina and contributes to the understanding of cavity production by examining how characteristics of the patch, tree, substrate, and bird species influence the selection of nest sites for excavation. First, we characterized the nesting sites used by 8 species of excavator birds in the humid Chaco and assessed whether body mass is related to the size and decay class of excavated wood. Second, we tested whether nest-site selection responds to factors we hypothesized to influence nest survival (i.e., proximity to foraging sites, buffering environmental temperatures, protection from failure), or to factors related to substrate suitability for excavation (Table 1). We predicted that factors related to substrate suitability would be more influential in nest-site selection than factors related to nest survival. Third, we tested the hypothesis that interspecific variation in body size and foraging behavior, traits frequently related to excavation ability, explain interspecific variation in wood hardness of excavated nest substrates. We predicted that, across species, the hardness of excavated wood (for nests) would increase with body mass and prevalence of chiseling and hammering techniques in foraging, and decrease with the prevalence of searching and gleaning techniques.

METHODS

Study System

We studied excavators and their nests during 4 breeding seasons (August to January) from 2016 to 2019 in Chaco National Park (28.8204°S, 59.6204°W, 14,981 ha) and during 2 breeding seasons (2016 and 2017) in Pampa del Indio Provincial Park (26.2767°S, 59.9702°W, 8,366 ha) in the humid Chaco, Chaco province, Argentina. The humid Chaco is a mosaic of wetlands, palm groves, quebrachales, and gallery forests (Oyarzabal et al. 2018); we worked in quebrachales and gallery forests. Quebrachales are open canopy forests dominated by Schinopsis balansae (Quebracho Colorado Chaqueño), with scattered Prosopis spp. (Algarrobos), Tabebuia nodosa (Palo Cruz), and Caesalpinia paraguariensis (Guayacan) (Supplementary Material Figure 1). Gallery forest is a vertically stratified forest with a closed canopy and a diversity of trees, including Astronium balansae (Urunday), S. balansae, Gleditsia amorphoides (Espina Corona), Pisonia zapallo (Zapallo Caspí), Peltophorum dubium (Ivyrá-Pitá), Enterolobium contortisiliquum (Timbó) and Phyllostylon rhamnoides (Palo Amarillo; Morello and Adámoli 1967, Cabrera 1976, Oyarzabal et al. 2018). The humid Chaco has a mean annual temperature of 24°C with a large thermal amplitude that can range from 15°C to 40°C in a single day, and an annual rainfall of 1,200 mm concentrated between October and April (Cabrera 1976).

Our study area includes 13 cavity-excavating bird species (12 Picidae, 1 Trogonidae) with a wide range of body sizes (Dunning 2007, Birds of the World platform 2022: https://birdsoftheworld.org) and different population trends (BirdLife International 2023). Dryocopus schulzii (Black-bodied Woodpecker, 200 g) is near threatened. Picumnus cirratus (White-barred Piculet, 10 g), Melanerpes cactorum (White-fronted Woodpecker, 40 g), and Celeus lugubris (Pale-crested Woodpecker, 125 g) have potentially declining population trends. Dryobates mixtus (Checkered Woodpecker, 30 g), D. passerinus (Little Woodpecker, 37 g), Colaptes melanochloros (Green-barred Woodpecker, 135 g), and Campephilus leucopogon (Cream-backed Woodpecker, 230 g) have stable population trends and are classified as least concern. We excluded 5 additional excavator species because we did not find them nesting in wood in our study area: Piculus chrysochloros (Golden-green Woodpecker, 60 g) and Trogon surrucura (Surucua Trogon, 70 g) excavated in arboreal termite or ant nests; Colaptes campestris (Campo Flicker, 150 g) and Melanerpes candidus (White Woodpecker, 115 g) inhabit open areas and palm groves (Di Giacomo 2005); and Dryocopus lineatus (Lineated Woodpecker, 230 g) is rare in the study area (Di Giacomo 2005).

Collection of Data on Nests and Potential Substrates

To obtain sufficient nest observations for analysis, we employed a case-control study design (Keating and Cherry 2004), comparing each cavity excavated in wood (nest) with a nearby potential substrate (trunk or branch that looked similar but was not excavated). We conducted nest searches in quebrachales (~700 hr yr^–1) and gallery forests (~250 hr yr^–1). To locate nests, we inspected areas where we heard continuous tapping or observed recurrent flights. We determined that a cavity was excavated in the corresponding breeding season when we observed (1) the cavity developing progressively over days or weeks, or (2) a substantial quantity of wood chips on the ground (in the case of nests found at the egg or nestling stage).

To find a potential substrate (paired control for each nest cavity excavated in the corresponding breeding season), we first determined the decay class of the nest tree (live unhealthy vs. dead). Second, we searched for a potential excavation substrate in a random direction and at a minimum distance of 25 m from the nest tree so as not to overlap potential sites with nest sites (see the following paragraph). Our potential (unused) tree was the first tree that met the following conditions: (1) same decay class as the paired nest tree, (2) same species as the paired nest tree, and (3) no previous excavations. We did not find any newly-excavated nest cavities in healthy trees and we did not include any healthy trees as potential nest sites. If we did not find a suitable potential tree after 150 m, we returned to the nest tree, selected another random direction, and repeated the search procedure. Many trees in our study area have only a short trunk, which divides into large-diameter second-order branches about 1–5 m above ground (Supplementary Material Figure 1). Thus, within the potential tree we randomly chose between the main trunk and the second-order branches, then divided the branch or trunk into 30-cm sections and selected a random height between 1 and 7 m in height (height range used by birds in the study system; Table 2), where we designated the 30-cm section as our potential substrate. Of our 42 case-control pairs, 26 pairs had the same branch order for the potential substrate as for the nest. Of the 16 remaining pairs, 10 had the nest in a trunk but the potential substrate in a branch, and 6 had the nest in a branch but the potential substrate in the trunk. Although we did not control for tree size when selecting potential trees, there was no significant difference in diameter at breast height (DBH) between nest trees (mean ± SD = 34.4 ± 20.1 cm) and paired potential trees (29.9 ± 12.6 cm; Wilcoxon signed-rank Test V = 552, P = 0.22).

For each excavated cavity and each potential substrate, after breeding was complete, we used a 7-m ladder or a rope with harness equipment to take measurements of 15 characteristics at the scale of the patch, tree, and excavation substrate, similar to other studies on excavator birds (Politi et al. 2009, Cockle et al. 2011a, Lorenz et al. 2015, Jauregui et al. 2021). As a nest patch we considered a circular area of 11.3 m radius with the nest tree or potential tree in the center (central tree). The minimum distance of 25 m between nest tree and potential tree ensured that the patches did not overlap. For each nest patch we measured (1) the mean percent canopy cover from 4 measurements taken with a densiometer (3% resolution) at 10 m from the central tree in the 4 cardinal directions. We recorded (2) the number of trees (individuals with DBH > 10 cm) including the central tree. For each tree in the patch with DBH > 10 cm, we used a tape (0.1 cm) to measure (3) DBH, determined (4) the decay class (dead or live), and identified (5) the species. At the level of the central tree (nest or potential), we recorded (6) DBH, and (7) the tree species. We estimated (8) the percentage of the crown in contact with neighboring trees (measure related to access of arboreal predators to the nest tree). At substrate level (excavated or potential) we measured (9) the diameter of the substrate at cavity height (DCH), (10) internal and (11) external wood hardness (see below), and (12) height above ground. We used a densiometer at substrate height to estimate (13) percent cover. We determined the percent visibility of the substrate from 2 heights: (14) at ground level (0.5 m) and (15) at substrate height. To determine the percent visibility at each of these heights, we filmed or photographed the substrate from 3 points, all at a horizontal distance of 10 m, one directly in front of the substrate, one 45° to the left, and one 45° to the right. For each photo or video, we estimated the percentage of the cavity entrance that was visible, or for potential substrates the percentage of an area of similar size to a cavity entrance. For each height, we took the mean of the 3 values obtained (from the 3 directions). We did not include cardinal cavity orientation as a predictor of nest-site selection because nest cavity orientation is uniform in our data set (Ojeda et al. 2021).

We used the method developed by Matsuoka (2000) to measure wood hardness. With a drill and a 9-mm drill bit, we drilled a hole radially into the substrate (towards the center of the trunk or branch). Wood hardness is proportional to the torque required to expand this hole with an increment borer (30 cm long and 5.15 mm external diameter). We measured the torque every 1 cm along the perforation by means of a torque meter attached to the increment borer. As suggested by Matsuoka (2000), we use Newton meters (N ∙ m) as the unit of torque for statistical analyses. Following Matsuoka (2008), we drilled 5 cm above the nest entrance or, if we could not drill above the cavity (e.g., if the substrate was broken or a branch was in the way), we drilled 10 cm below the bottom of the cavity. For nests, we drilled radially until the entire horizontal depth of the cavity was included. In potential substrates, we drilled radially to the center of the tree. We divided the perforation into 2 sections: external wood (corresponding to the cavity wall) and internal wood (corresponding to the nest chamber). For potential substrates, we considered the first 3 cm to be external wood and the following 12 cm to be internal wood. For each substrate, we took the mean of torque measurements (every 1 cm) within the corresponding section of the perforation to obtain the final values for external and internal wood hardness.

To describe forest composition, we considered the DBH, species, and decay class (live/dead) of all trees that comprised the patches (of nests and potential sites; 0.04 ha each patch) of cavity excavating birds (this paper) and cavity adopter birds in the same study area and years (Di Sallo and Cockle 2022). In total, we obtained measurements for 3,884 trees in an area of 12.4 ha for quebrachales and 827 trees in an area of 4.8 ha for gallery forests.

Collection of Data on Foraging Techniques

To obtain data on the prevalence of foraging techniques in each woodpecker species, we conducted focal observations on individuals found foraging. We found foraging adults opportunistically as we searched for nests. For each focal observation, the first author used binoculars to observe the adult foraging, for 60 s beginning with the first visual contact (n = 187 focal observations). We recorded the time (s) that the bird spent in each of the following foraging activities adapted from Remsen and Robinson (1990) and Fernández et al. (2020). (1) Chiseling: The bird removes bark or layers of wood with blows at oblique angles. When it tilts its head back to make the blows, the head exceeds the vertical axis of the body. (2) Hammering: The bird performs a series of sustained pecks (>4 blows) producing deep holes. In these blows, too, the head exceeds the vertical axis of the body. (3) Pecking: The bird intermittently (<4 pecks) taps its beak against the substrate to remove external parts of the substrate (bark, mosses, lichens). In this case, when the head is tilted backwards, the head does not exceed the vertical axis of the body. (4) Probing: The bird inserts the tip of its bill into pre-existing cracks or holes. (5) Searching/gleaning: The bird searches for and captures prey as it climbs up the trunk or branch. (6) Scanning: The bird looks around. (7) Eating: The bird consumes prey. We consider chiseling and hammering as foraging techniques related to robust anatomies and searching/gleaning as a foraging technique associated with less robust anatomies (Kirby 1980, Bock 1999, Donatelli et al. 2014). To include variability among individuals and to promote independence among focal observations, we made no more than 3 focal observations of the same species within the same 30-ha area. The 8 woodpecker species we studied represent a diversity of foraging guilds (Figure 1).

Data Analysis

We report mean ± standard deviation, except where indicated. All statistical analyses were performed in R software, version 4.3.3 (R Core Team 2024).

Analysis of Nest-Site Selection

We used an information-theoretic approach to compare conditional logistic regression models that reflected a priori hypotheses about selection of nest sites (Burnham and Anderson 2002; Table 1). Conditional logistic regression is appropriate for case-control studies because the ratio of cases (excavated cavities) to controls (potential substrates) in the sample differs from the ratio of cases to controls in the population (Keating and Cherry 2004). To build the models, we used the coxphf package (Heinze et al. 2023) to implement a Firth’s penalized maximum likelihood bias reduction method for Cox regression (Heinze et al. 2023), which provides a solution in cases of monotonic likelihood (Heinze and Schemper 2001, Allison 2008). We analyzed the correlation between independent variables and excluded variables that were correlated (Pearson’s r > 0.7; Dormann et al. 2013). Each model reflected a hypothesis related to increasing foraging opportunities, decreasing predation risk, buffering temperature, or facilitating excavation (Table 1). To avoid overparameterization, in any given model we kept the number of predictors below m/10, where m is the limiting sample size (in our case, 42 case-control pairs), following Harrell (2015). To improve the interpretation of interaction terms, we standardized all continuous predictor variables to each have a mean of zero and standard deviation of 1. For each model, we calculated the Akaike information criterion corrected for small samples (AIC_c) and the Akaike weight (w_i). We evaluated the strength of the models based on their∆AIC_c (difference between the AICc of a given model and the model with the lowest AIC_c) and their Akaike weight (Burnham and Anderson 2002). We considered a model to be supported by the data if it had a∆AIC_c < 2 with respect to the best model (lowest AIC_c) and to be strongly supported as the most plausible model in the set if it had w_i > 0.8. Conditional logistic regression models do not include a parameter for the intercept and therefore a null model was not possible (Keating and Cherry 2004). We considered variables to be good predictors when the 95% confidence intervals on their odds ratios did not overlap 1.

Analysis of Interspecific Variation in Nest Site and Wood Hardness (Birds and Trees)

To evaluate the relationship between body mass of wood excavator birds (predictor) and 3 substrate characteristics (response variables: DCH, DBH, and decay class of the nest tree), we assigned each excavated cavity a body mass value according to bird species (from published literature; see Study System) and modeled each substrate characteristic (y) separately, as a function of body mass (x). For DCH and DBH, the models were, in increasing order of complexity: (1) a null (intercept-only) model (y = b₀), (2) a linear regression (y = b₀ + b₁x), and (3) an adaptation of the non-linear Beverton–Holt model, a saturating hyperbolic, in which y = ax/(b + x), where a is the value of y at the asymptote (as × approaches infinity) and b is the half-maximum (value of × when y = a/2). These models were produced using the mle2 command in the bbmle package (Bolker and R Core Team 2020). To examine the relationship between body mass (x) and nest tree decay class (live unhealthy/dead; y) we used generalized linear models (binomial family, logit link, glm command, stats package), employed AUC (area under the curve) for diagnostics (ROCR package, Sing et al. 2005), and compared the parameterized model with a null (intercept-only) model. Within each set, we considered the best-supported model to be the simplest model with ΔAIC < 2.

We used one-factor analysis of variance (ANOVA) followed by a Tukey’s a posteriori test to assess differences in the excavated wood hardness among species of excavator birds. We performed one analysis for external wood hardness and another for internal wood hardness. For these comparisons, we only included bird species with more than 4 wood hardness observations at nest sites and used α = 0.05 (without Bonferroni correction, seeking a balance between type I and II errors). We tested for normality using the Shapiro–Wilks test. We log-transformed internal wood hardness to meet the normality assumptions of the ANOVA; external wood hardness met the assumption without transformation. Our data met the assumptions of ANOVA (independent samples, normally distributed continuous response variable, homoscedasticity of variance; see diagnostic plots in Supplementary Material). To understand how the excavated wood hardness varied across the 3 most used tree species, we used a Kruskal–Wallis test by rank, a non-parametric alternative to one-way ANOVA (Ostertagová et al. 2014).

Analysis of Relationships Among Body Mass, Foraging Behavior, and Wood Hardness

We examined the relationship between hardness of external wood at nest cavities, and the foraging techniques and body mass of the birds that excavated the cavities. We only used external wood hardness because it is usually higher than internal wood hardness and exhibited differences between species in a previous study (Lorenz et al. 2015). For each focal observation of foraging, we calculated the total time the individual was actively trying to obtain food (sum of time spent chiseling, hammering, pecking, probing, and searching/ gleaning). We excluded 2 behaviors (scanning and eating) that did not directly represent attempts to obtain food. For each focal observation we calculated the percentage of time spent in each technique over the total time spent trying to obtain food. We averaged these percentages across all individuals of the same species to obtain a prevalence of each technique for each species. We assigned to each excavated cavity the prevalence values of each technique according to the bird species.

We employed LASSO regression (a type of penalized linear regression) using the R package glmnet (Friedman et al. 2010) to model external wood hardness in relation to foraging techniques and body mass. LASSO regression (known as the L1 regularization) is one of the solutions proposed to address multicollinearity and overestimation (Ranstam and Cook 2018, Chan et al. 2022), and it was appropriate for our dataset because we observed high multicollinearity among foraging techniques and body mass (Pearson r ≥ 0.65, Dormann et al. 2013). LASSO regression tends to retain one of the correlated variables and eliminate the remainder (Chan et al. 2022). LASSO regression is preferable to Ridge regression in our context because it is better at explaining the relationship between the predictor and response variables (Chan et al. 2022). We performed a cross-validation to find the λ (penalty) that minimized the mean squared error. With this λ we obtained the LASSO estimator for each predictor variable retained in the model. Then, we calculated R² as the explanatory power of the final model.

RESULTS

We recorded 71 nesting events of 8 species of excavator birds in 49 cavities (Table 2). Of the 8 species that excavated cavities in wood, 3 also reused old cavities in subsequent years. Dryocopus schulzii reused 4 times in 2 consecutive years an excavated cavity; Celeus lugubris reused in 2 consecutive years the same excavated cavity and a cavity excavated by Colaptes melanochloros. Celeus lugubris also enlarged a cavity excavated by Dryobates mixtus and used it for 2 consecutive breeding seasons. Colaptes melanochloros reused 4 excavated cavities, but we were unable to determine the species that excavated these cavities.

Nest-site selection

Of the 49 cavities, 28 were in Prosopis spp. trees, 6 in Schinopsis balansae, 5 in Tabebuia nodosa, 3 in trees not identified because of advanced decay, 2 in Patagonula americana (Guayubira), 2 in Gleditsia amorphoides, 1 in Phytolacca dioica (Ombú), 1 in S. lorentzii and 1 in Peltophorum dubium (Supplementary Material Figure 2). Prosopis spp. comprised only 8% of the trees in the forest (377 out of 4,711) but supported 57% of the nests of excavator birds (28 of 49). In contrast, S. balansae comprised 72% of the trees in the forest (3,392 of 4,711) but supported only 12% of the nests of excavator birds (6 of 49).

Most nest sites (and most potential sites) were substrates in dead trees, but excavator birds also used dead and live substrates in unhealthy trees. Of the total nest cavities (n = 49), 63% (31 cavities) were in dead trees and 37% (18) were in live unhealthy trees. Similarly, of the 42 potential substrates, 67% (28) were in dead trees and 33% (14) were in live unhealthy trees. Overall, 88% (43 cavities) of nest cavities were excavated in a dead substrate, compared to12% (6 cavities) in live substrates. For potential substrates, 86% (36 substrates) were dead, while 14% (6 substrates) were live.

Conditional logistic regression and univariate analyses indicated wood hardness as the most important factor in nest-site selection. The best-supported conditional logistic regression model included only internal wood hardness (Table 3). Across all models, the only significant predictor of nest-site selection was external wood hardness (Table 4). Likewise, in exploratory univariate analyses of substrate characteristics, only external and internal wood hardness showed significant differences between nest substrates and potential substrates (Supplementary Material Table 1). Because they were correlated (r > 0.70), internal and external wood hardness were not included in the same conditional logistic regression models. Internal wood hardness (heartwood) measured 1.5 ± 1.2 N ∙ m (range: 0.3–5.8) in bird-excavated substrates and 17.2 ± 9.8 N ∙ m (range: 0.7–35) in potential substrates (Figure 2, Supplementary Material Table 1), but the 95% CI on its odds ratio slightly overlapped 1 (Table 4). Although external wood hardness was not in the top model, it explained selection in all models in which it was included, and always had a negative effect on nest-site selection (birds selected substrates with softer external wood; Table 4 and Supplementary Material Table 1). External wood hardness (sapwood) measured 4.7 ± 1.9 N ∙ m (range: 1.7–9.7 N ∙ m) for bird-excavated substrates and 13.2 ± 6.1 N ∙ m (range: 1.0–30.2) for potential substrates (Figure 2). In our dataset, an increase in external wood hardness of 1 N ∙ m was associated with an 18% decrease in the odds of the substrate being excavated (odds ratio for scaled wood hardness = 0.05; 1 unit scaled wood hardness = 15.12 N ∙ m; change in odds = 0.05^1/15.12 = 0.82). The differences in wood hardness between used and potential substrates were maintained across dead and live substrates (Supplementary Material Figure 3). The quadratic effect of external wood hardness (optimal point, neither hard nor soft) did not prove to be a good predictor of nest-site selection (Table 3).

Interspecific Variation in Nest Sites and Wood Hardness (Birds and Trees)

We found variation among woodpecker species in substrate characteristics but not tree characteristics (Figure 3 and 4; Supplementary Material Table 2). Excavator body mass was a positive predictor of DCH (Figure 3; Supplementary Material Table 2) but it did not predict DBH or tree decay class (live unhealthy vs dead; Figure 3; Supplementary Material Table 2). ANOVAs indicated that bird species differed in the hardness of external (F = 7.2, p < 0.001, Supplementary Material Figure 4) and internal (F = 11.67, p < 0.001, Supplementary Material Figure 5) wood that they excavated. Among the species included in this study (i.e., excavators of wood), we can differentiate 2 groups. Campephilus leucopogon and Dryocopus schulzii were the strongest excavators. Colaptes melanochloros, Melanerpes cactorum, and Dryobates mixtus can be considered regular wood excavators although the maximum hardness excavated by D. mixtus was within the range of the strongest excavators (Figure 4; Supplementary Material Tables 3–5). Of the 42 potential substrates we studied in the Chaco, 90% (38) were harder than any substrate excavated by Campephilus leucopogon and 100% were harder than any substrate excavated by the other bird species.

For excavated substrates, neither external nor internal wood hardness varied among the 3 most used tree species (external: χ² = 3.98, p = 0.13; internal: χ² = 3.77, p = 0.15; Figure 5). For potential substrates, external wood hardness did not vary among the 3 tree species (χ² = 1.54, p = 0.46, Figure 5), but internal wood hardness varied (χ² = 8.55, p = 0.01, Figure 5), with higher values for Schinopsis balansae than for Prosopis spp. (p = 0.02) and Tabebuia nodosa (p = 0.01, Figure 5).

Relationships Among Body Mass, Foraging Techniques, and Wood Hardness

The LASSO penalized regression model (λ = 0.07 with SE = 0.45, R² = 0.36; Supplementary Material Figure 6 and 7) included 3 variables as predictors of hardness of the external wood excavated by birds. As we predicted, the prevalence of chiseling (s0 = 0.155) and body mass (s0 = 0.001), both associated with robust anatomies, were positive predictors of the hardness of excavated wood, whereas the prevalence of searching/gleaning (s0 = –0.017) was a negative predictor (Supplementary Material Figure 8).

DISCUSSION

Our study in the humid Chaco showed that wood hardness (1) was the only nest site characteristic influencing nest-site selection across 8 species of woodpeckers, and (2) varied predictably among woodpecker species according to their foraging habits and body mass. Our finding that external wood was much harder than internal wood at excavated sites (but not potential sites) is consistent with previous descriptions and measurements, and indicates that although excavators can chisel through hard wood to make their small cavity entrance, they seek softer interiors for excavating the nest chamber (Jackson and Jackson 2004, Matsuoka 2008, Lorenz et al. 2015). Thus, whereas cavity adopters in our study system frequently nested in non-excavated cavities in Schinopsis balansae (Di Sallo and Cockle 2022), we found that woodpeckers avoided excavating in this tree species, which had exceptionally hard internal wood. Supporting the hypothesis that robust anatomies allow nest excavation in harder wood, wood hardness increased with species-specific body mass and the prevalence of chiseling to obtain food. Taken together, these results indicate that internal and external wood hardness can strongly restrict nest-site selection by avian excavators.

Our finding that the process of nest-site selection is first constrained by wood hardness is concordant with the few other studies that have measured hardness or density of wood excavated by birds (Schepps et al. 1999, Matsuoka 2008, Lorenz et al. 2015, Puverel et al. 2019, Jauregui et al. 2021) and highlights the key role of wood-decay fungi in softening substrates for excavation. Even though 86% of our potential substrates were dead, they all appeared too hard for most of our excavator community. Critical to the wood-softening process is the activity of wood-decay fungi (which break through cell walls, creating the conditions necessary for excavation) and their symbiosis with avian excavators (which disperse fungi to new substrates; Conner et al. 1976, Jackson and Jackson 2004, Jusino et al. 2016). To sustain this symbiosis, forest management plans need to ensure an ongoing supply of dead and diseased trees. Only in situations where excavators can find multiple substrates soft enough to excavate should we expect them to select nest sites that increase foraging opportunities or reduce the risk of nest predation.

Our findings on wood excavation in the Chaco paralleled in many ways the findings of the only similar study, which took place in the northwestern USA (Lorenz et al. 2015). Hardness of excavated external wood in Chaco (3.1–7.7 N m) was within the range of excavations studied in the northwestern USA (0–16.6 N m; Lorenz et al. 2015). In both systems, woodpeckers were easily separated into groups based on the hardness of external wood they excavated; however, woodpeckers of the Chaco also segregated based on the hardness of internal wood, and 2 species (not studied here) excavated only in arboreal termitaria. Future studies could compare wood hardness among excavating bird species and forests of the world to determine whether hardness of the nest substrate primarily reflects bird phylogeny and morphology, or ecological conditions of the forest, such as the availability of softer trees.

Our finding that wood hardness at excavated cavities was positively related to body mass and foraging techniques associated with robust anatomies (chiseling) corresponds with studies relating woodpecker anatomy to foraging habits (Kirby 1980, Bock 1999, Donatelli et al 2014) and suggests a constraint of wood hardness and anatomy on nest-site selection. The correlations between foraging techniques and body mass prevented us from testing the independent relationships among these variables, and the positive relationship we found between DCH and body mass suggests that larger species might be forced to excavate in harder sites if they are limited by the availability of sufficiently large branches or trunks (Lammertink 2007, Gutzat and Dormann 2018). To better understand the relationships among body mass, anatomy, foraging and nest cavity excavation, it is important to study other excavator communities and directly relate anatomy to measures of wood hardness.

As a field study of bird ecology in a poorly studied region of the Neotropics, our work has some caveats to consider. Our nest data set is small and dominated by a few species of excavators. The case-control study design did not allow us to examine year- or species effects on nest-site selection, since each case-control pair was in the same year and would correspond to the same species of woodpecker. Our study was limited to nature reserves; future studies are needed to evaluate whether the same factors (wood hardness) drive nest-site selection of excavating birds in logged or cleared areas. Logging in the humid Chaco has focused primarily on Schinopsis balansae and Prosopis spp. trees, and degraded sites are generally dominated by trees with softer wood (e.g., Phytolacca dioica, Enterolobium contortisiliquum; Ginzburg and Adamoli 2005, Barberis et al. 2012). Nest-site selection can change with habitat modifications, according to availability of suitable excavation sites (Bonaparte et al. 2020, Maya-Elizarrarás et al. 2024), and needs to be studied across a range of environmental conditions.

Nest-site selection by excavator birds can have important impacts on vertebrate communities because it influences the location and characteristics of cavities that become available to cavity-adopting birds and other animals (Daily 1993, Trzcinski et al. 2022, Bonaparte et al. 2024). The use of excavated cavities by adopter birds varies from 0 to almost 100% among forests globally and according to poorly understood local and regional conditions (Aitken and Martin 2007, Blanc and Walters 2008, Cockle et al. 2011b, 2019; Ruggera et al. 2016). Given the evidence that wood hardness constrains excavators in their nest-site selection, in forests with very hard wood they may have few opportunities to create cavities with characteristics (e.g., height, concealment, isolated crown) that confer an advantage to their own nests and those of cavity adopters. Researchers can explore how agents of cavity formation influence cavity quality by examining variation in reproductive output of adopter birds (Cuatianquiz Lima et al. 2024). We encourage approaches that incorporate wood hardness of excavated sites, characteristics of trees and their associated fungi, and body size and excavation ability of birds, to increase our understanding of cavity production and its implications for cavity adopters, particularly in little-studied tropical and subtropical forests.

Supplementary material

Supplementary material is available at Ornithology online.

Acknowledgments

We are grateful for the valuable contributions of many volunteers who helped with the fieldwork, among them F Dosil, J Allen, A Snell, M Aranciaga Rolando, M Castano, F Casianelli, B Wilcox, K Murphy, E Beltrocco, M Berraz, P Capovilla, D Franco, L Pacheco, D Daglio, C Gomez Venninni, C Pita, C Fantinati, E Guerts. We thank B Bonaparte, A Pietrek, A Juncosa, M Lammertink and P Blendinger for valuable comments on the study design and manuscript. Three anonymous referees provided valuable insights to improving this manuscript.

Funding statement

Fieldwork and equipment were funded by a Bergstrom Award from the Association of Field Ornithologists, Francois Vuilleumier Fund from Neotropical Ornithological Society, IdeaWild and Agencia Nacional de Promoción Científica y Tecnológica (PICT-2016-0144).

Ethics statement

Work was authorized under the NEA 408 permit, granted by the National Park Administration of Argentina.

Conflict of interest statement

The authors have no conflict of interest.

Author contributions

Facundo Di Sallo conceived the idea, formulated hypothesis, study design, data collection, wrote the paper, developed methods, analyzed the data, and contributed materials, resources, and funding. Kristina Cockle conceived the idea, supervised research, formulated hypothesis, study design, substantially edited the paper, analyzed the data, and contributed materials, resources, and funding.

Data availability

Analyses reported in this article can be reproduced using the data provided by Di Sallo and Cockle (2024).

Reflexivity statement

Were local/in-country researchers or community members involved in the study design?

The study was designed in-country, by FDS and KLC, both of whom are affiliated in the region (northeastern Argentina). We also received input from FDS’ in-country thesis committee members. Community members from the Chaco were involved during the study but not in the study design phase.

How will research products be shared to address local needs?

We provide a Spanish version of the manuscript in the Supplementary Material. We will share the Spanish version and a summary of results and management recommendations with the national park administration and the environmental department of Chaco province. We will disseminate highlights of the study in Spanish through social media.

Are researchers within the region (particularly women, gender minorities, and early career researchers) included as authors?

FDS is of South American origin (Argentina) and the article represents a chapter of his PhD Thesis at the University of Tucumán (Argentina). KLC is from Canada, affiliated in Argentina.

Did the authors search for relevant publications in regional journals, including those in languages other than English?

We conducted an exhaustive literature search of regional sources. We cited 8 sources in Spanish and 15 with first authors affiliated in Latin America.

If the study includes researchers from high income countries, how has the project developed their capacity to work collaboratively and equitably with colleagues within the region of study?

KLC was trained in the Canadian academic system and this study represented her first opportunity to supervise a PhD thesis. During this project, we faced many of the common barriers to graduate students and researchers in Latin America, including difficulties importing equipment, English language hegemony, major funding delays, 200% annual inflation, insufficient institutional infrastructure, and the need to do extensive emergency care work while precariously employed. While working to dismantle some of these barriers (e.g., through advocacy), we recognize that many will persist in the medium term. To face them, we have adopted a more structured and proactive approach for mentees to document their goals and progress, built systems for multi-directional feedback into all projects, created more space for career development beyond the PhD thesis, and prioritized strengthening mentoring networks within Latin America.

How has the project influenced the means and ability of the researchers from within the region to implement their research agenda?

The article contributes to our long-term research program on cavity-nesting communities in the Neotropics. We acquired concepts, equipment, and skills to measure wood hardness and excavation ability of birds, which we are using in other projects in the region.

LITERATURE CITED