Study area
We studied behaviour, fecundity, and habitat characteristics of cavity-nesting birds in 25 mixed coniferous-deciduous forest stands on the lands of the Tŝilhqot’in, Secwépemc, and Southern Dakelh Peoples, an area surrounding Williams Lake, British Columbia, Canada (51°52’N, 122°21’W), from 2004–2008. The predominant coniferous trees were Douglas-fir (Pseudotsuga menziesii var. glauca), lodgepole pine (Pinus contorta var. latifolia; hereafter, pine), and white and Engelmann hybrid spruce (Picea glauca x engelmannii; Meidinger and Pojar 1991). The predominant broadleaf tree was trembling aspen (Populus tremuloides). Study sites ranged from 15 to 32 ha (one 7-ha site) in size and varied in composition from continuous forest to five sites that comprised a series of ‘forest groves’ (0.2 to 5 ha) within a grassland matrix.
Study system
The small tree cavity nesting community comprised two species of insectivorous excavators and one secondary cavity nester (Martin and Eadie 1999). As many cavity excavators are also insectivores, large-scale insect outbreaks can lead to dual pulses in food and nest sites, potentially influencing the competitive interactions among tree cavity-dependent insectivores. Mountain pine beetle is a bark-boring insect that feeds on the phloem of pine trees and is a common disturbance agent in temperate forests that undergoes occasional patchy outbreaks in western North American forests (Taylor and Carroll 2003). Recent mountain pine beetle outbreaks in British Columbia increased year-round food availability, and subsequently, population densities of many insectivorous birds, including many excavators (Taylor and Carroll 2003; Martin et al. 2006). mountain chickadee (Poecile gambeli), a secondary cavity nester (cavity specialist) that relies on excavators and natural decay processes for nest cavities, is primarily a foliage gleaner, but can switch to other foraging substrates depending on forest insect abundance (McCallum et al. 2020). Red-breasted nuthatch (Sitta canadensis), is a facultative tree cavity nesting excavator, and is primarily a bark forager (Ghalambor and Martin 2020).
The mountain pine beetle outbreak led to increases in population densities of both chickadees and nuthatches. Red-breasted nuthatch shifted nest site preference from areas of high nest site availability to those of high mountain pine beetle availability, where they excavated a greater proportion of nests (Norris and Martin 2008, 2012). Mountain chickadee populations showed a one-year lag in increases following increased nuthatch populations, and used a greater proportion of smaller, safer nuthatch cavities following the beetle outbreak, suggesting that chickadee populations benefited from the higher densities of nuthatches (Norris et al. 2013, 2022). However, the beetle outbreak also led to increased population densities of a common nest predator for chickadees and nuthatches, American red squirrel (Tamiasciurus hudsonicus;(Martin and Norris 2007)). Because increased predator presence can lead to reduced parental activity around the nest resulting in reduced fecundity (Fontaine and Martin 2006), high squirrel densities may diminish territory quality and impede territory defence strategies. Such changes in territory characteristics could lead to increases or decreases in agonistic behaviour within and between species.
We located nest trees of chickadees and nuthatches by checking all nesting cavities in trees used by other cavity-nesters in previous years (1995–2007) with a camera monitoring system on an extendable pole and by following individuals to their nests. We considered nests to be active if we found eggs or chicks in a cavity and monitored all nests until fledging or failure. Additional study area and nest monitoring details are given in (Martin and Eadie 1999; Aitken et al. 2002).
Territorial intrusions
We used song playbacks with intruder simulations to investigate interference competition within and between species (Martin et al. 1996) during territory establishment and before eggs were laid until after chicks fledged, between 1 May and 30 June, during (2004–2005) and after (2006–2008) the beetle outbreak (i.e., five years of measurement). To examine territorial responses of chickadees and nuthatches we simulated conspecific and heterospecific intrusions at six treatment types that represented two temporal- and two spatial-scales with respect to nesting (Table 1). We simulated intrusions within ~ 1 m of nest trees at nests that were active with either chickadees or nuthatches in the year of the presentation (1. active chickadee nest, 2. active nuthatch nest), 3. Nest cavities that were active in a previous year by either species (inactive nest), or within ~ 1 m of a tree suitable for excavation or nesting (aspen tree ≥ 15cm DBH, with or without existing tree cavities) but was not to our knowledge used for nesting by either species, and located in a random direction ~ 50m away from an active nest but still within the active territory (4. active chickadee territory, 5. active nuthatch territory) or 6. ~50m from an inactive nest within a territory not used by either chickadees or nuthatches in the year of the experiment (inactive territory). We compared responses of chickadees and nuthatches measured at each active, inactive, and suitable nest tree to those at suitable nest trees ~ 50 m from inactive nests (inactive territory) to assess the level of territorial aggression. The two species exhibit unique behaviours with respect to aggressive calls and displays, but both species exhibit the same behaviour of moving towards and supplanting intruders (Minock 1972; Grava et al. 2012). Therefore, we measured response to intruders by estimating the closest distance (m) that a respondent approached each intruder during each simulated intrusion to examine a common behaviour and compare interspecific responses (Kershner and Bollinger 1999). Although it was not possible to record data blind because our study involved focal animals in the field, we simulated intrusions of both species for each trial and presented each species in a random order. In 2004, we used song recordings from the second edition (1992) of the Peterson Field Guide audio compact disc from Cornell Lab of Ornithology, and during 2005–2008, we used recordings of songs of local chickadees and nuthatches collected ~ 20 km outside the study area. Songs were digitally manipulated so that each song was 2 min in length, and projected at similar volumes, then transferred onto a portable media player and broadcast over speakers. A taxidermic model specimen (intruder) of the appropriate species was placed on a wire stand ~ 1m above the speakers and presented with the appropriate song type for each trial, with a 5-min period of silence following each intruder species presented. For each respondent, we recorded the species, individual (if colour banded), sex, time of day, behaviour (whether the respondent: called and call type, sang, swooped, attacked, etc.), and the closest distance (m) that they approached to the model intruder. In cases where the respondent attacked the intruder, and aggression levels remained high, we waited 10 min to start the presentation of the next intruder species until the aggressive individual returned to displaying the behaviour observed before the first intruder was presented. In 2004, mountain chickadees approached conspecifics farther than in other years except 2007 (F4,251=5.17, p < 0.01), suggesting that the recordings of local chickadees elicited a stronger response from the Peterson’s recordings, so 2004 was excluded in analyses examining responses of mountain chickadees. Nuthatches approached conspecifics closer in 2004 relative to only 2008 (F4,226=5.34, p < 0.01), therefore 2004 was included in all nuthatch analyses. Where intrusions were simulated at active nest territories, we visually inspected the nest cavity using a pole-mounted video camera, and recorded fecundity characteristics (number of eggs or nestlings) and the stage of the nest to determine breeding status (pre-nest, egg-laying, incubating, chick-rearing).
Table 1
We conducted 974 territorial intrusion experiments according to six plot types (distance to nest tree, species using the cavity, and active or inactive nest status) assigned by mean distance to nest tree (m) observed to be occupied by a chickadee or nuthatch breeding pair (Territory holder species) in the same year as the experiment or in a previous year (active/inactive nest status, respectively). To test the additional hypothesis that proximity to nest affects territoriality we conducted simulations ~ 50 m from active and inactive nests at (~ 1 m from) any available tree suitable for excavation or nesting (aspen tree ≥ 15cm DBH) but which was never, to our knowledge, used by either species. We broadcasted song recordings paired with presentations of taxidermically prepared specimens of mountain chickadee and red-breasted nuthatch procured from the Beaty Biodiversity Museum at the University of British Columbia, Vancouver, to simulate intrusions at 25 study sites in interior British Columbia, Canada, from 2004–2008. We compared the responses measured for each species at each active, inactive, and suitable nest tree to those at inactive territories.
Plot type | Distance (m) | Territory holder species | Nest status |
Active chickadee nest (ACN) | 1 | chickadee | active |
Active nuthatch nest (ANN) | 1 | nuthatch | active |
Inactive nest | 1 | chickadee and/or nuthatch | inactive |
Active chickadee territory (ACT) | 50 | chickadee | active |
Active nuthatch territory (ANT) | 50 | nuthatch | active |
Inactive territory | 50 | neither | inactive |
Population densities
To determine how population densities of conspecifics and heterospecifics (including predators) influenced the behavioural responses, we conducted point count surveys to estimate population densities per ha of mountain chickadee, red-breasted nuthatch, and red squirrel at 25 sites, during 2004–2008. Point count stations were spaced within a 100 m square grid ≥ 50 m from a grassland or wetland edge (one station ha− 1) in continuous forest sites, and at least 100 m apart in forest groves. From 0500–0930 hours, we recorded the species, and number of individual birds and squirrels detected within 50-m radius 6-min point counts at each station (7–32 stations site− 1). Each station across the 25 sites was surveyed three times (rounds) in each of the 5 years. We divided the total number of individuals observed on all rounds by the total number of point counts to obtain estimates of mean individuals ha− 1 for each site and year. Further details of population monitoring methods were reported in earlier studies (Martin and Eadie 1999; Norris and Martin 2010).
Vegetation surveys
To determine whether spatial and temporal variation in food and nest site availability influenced species interactions, we established 0.04-ha circular vegetation plots centered at each point count station every year during 2004–2008. For all trees ≥12.5 cm diameter at breast height (DBH; measured at 1.3 m above ground) in each plot, we recorded tree species, DBH, general health (e.g., presence of boring insects on the bole), and decay class. Decay class 1 was a live, healthy tree, 2 a live tree with visible sign of bark boring insects or heart rot fungus, and 3–8 were standing dead trees in progressive states of decay (Thomas 1979). Before and during the study period, an outbreak of mountain pine beetle occurred across all sites, with incidence of beetle attacks on pines increasing sharply after 2002, and by 2005 over 95% of the mature lodgepole pine trees (40% of the trees on the sites) were dead (Edworthy et al. 2011). However, the onset of the beetle outbreak showed temporal and spatial variation in the number of trees showing sign of beetle attack (Drever et al. 2009). Therefore, we could examine the effects of beetle abundance at the site-year level. Beetle eggs are laid beneath the bark in late summer where they overwinter and beetle larvae complete development in the following summer before emerging as adults (Reid 1962). Thus, beetle larvae provided a rich food source throughout the winters and following breeding seasons for insectivores. We determined beetle-infected pine densities as the total number of decay class 2 pine trees with bark boring insects, which was evident by the presence of dried resin outflows, or small entry holes (~ 2mm in diameter) on the bark, expressed on a per ha basis divided by the total number of 0.04 ha vegetation plots, for each site and year. Since over 90% of chickadee and nuthatch nests were in aspen trees (Martin and Eadie 1999), we determined the densities of potential nest trees per ha from number of aspen trees divided by the total number of 0.04 ha plots, for each site and year.
Statistical analyses
We examined how territory characteristics (proximity to active and inactive nests, food and nest site availability, and population densities of conspecifics, and heterospecifics, including squirrels) influenced intra- and inter-specific aggression of chickadees and nuthatches. We used the closest distance (m) that a respondent approached each intruder during the territory intrusion simulation as the metric of aggression. For both species, the data for closest distance approached were heavily skewed toward zero and showed an uneven distribution in the number of response variables between 0–50 m. As a result of the truncated normal distribution and the high number of zero values, we applied a mixture of the binomial (for closest distance = 0 or > 0) and left-truncated normal (for closest distance > 0) distributions. We used linear mixed-effects models (Crawley 2012) to examine how variation in closest distance approached was explained by the fixed-effects variables: intruder species (conspecific or heterospecific), plot type of the experiment (active chickadee nest; ACN, active nuthatch nest; ANN, inactive nest, active chickadee territory; ACT, active nuthatch territory; ANT, or inactive territory), breeding status of territory owners (if on an active nest plot), presentation sequence of intruders, and abundance of food (beetle-infected live pine densities per ha for the corresponding site and year in which the intrusion experiment was conducted) and nest sites (nest-tree densities per ha), and densities per ha of red-breasted nuthatch, mountain chickadee, and red squirrel, and all biologically relevant secondary interaction terms. Because we conducted intrusion experiments at multiple locations (plots) within sites and at multiple sites within years, we included plot nested within site as a random effect to account for the hierarchical error structure due to the repeated measures, in all models. Since the repeated measures were not evenly spaced within and between plots, we added a continuous autoregressive correlation within plots. These hierarchical errors were assumed to have normal distributions.
We used penalized quasi-likelihood (PQL) methods to generate parameter estimates (Bolker et al. 2009). We used Wald’s t-test to eliminate fixed-effects variables from fully parameterized models, and to determine whether fixed-effect variables had a significant effect on the response variables given the other fixed-effect variables and the hierarchical error structure in the best-fit (final) model (Bolker et al. 2009). Negative signs of significant coefficients indicated that the closest distance approached decreased (I.e., the responding bird approached closer to the model, and was more aggressive) and positive signs indicated that distance approached increased (the bird approached further distances, and was less aggressive) with increases in the fixed effect variables (Crawley 2012)). Penalized quasi-likelihood estimates were obtained using the function glmmPQL in the library MASS, and all data analyses were conducted in the program R version 2023.06.1 (R Core Team 2013; Ripley et al. 2013).