Individuals may gather information about environmental conditions when deciding where to breed in order to maximize their lifetime fitness. Many studies have shown that birds can rely on social information to select their nest site. The location of active nests and the reproductive success of conspecifics and heterospecifics can provide accurate predictions about the quality of the breeding habitat. Some short-lived species can facultatively reproduce two and/or more times within a breeding season. However, few studies have focused on how multiple-brooding individuals select nest sites for their second breeding attempts. In this study, we use long-term data to test whether the Japanese tit (Parus minor) can use social information from conspecifics and/or heterospecifics to select a nest site for the second breeding attempt. Our results showed that the nest boxes occupied by tits on their second breeding attempt tended to be surrounded by more breeding conspecific nests, first-breeding successful conspecific nests, and fewer first-breeding failed conspecific nests than the nest boxes that remained unoccupied (the control group). However, the numbers of breeding heterospecific nests, successful heterospecific nests, and failed heterospecific nests did not differ between the nest boxes occupied by tits on their second breeding attempt and the unoccupied nest boxes. Furthermore, the tits with local successful breeding experience tended to choose areas with more first-breeding successful conspecific nests than those without successful breeding experience. Thus, we suggest that conspecifics’ but not heterospecifics’ social information within the same breeding season is the major factor influencing the nest site selection of Japanese tits during second breeding attempts.

The fieldwork was conducted from March-July over successive years from 2014 to 2021 in an approximately 134.5 ha mixed deciduous forest at Zuojia Nature Reserve (44°1′–45°0′N, 126°0′–126°8′E) in Jilin Province, northeastern China. The study area consisted of 9 discrete deciduous forest patches, with nearly homogeneous forest characteristics. The dominant tree species were Quercus mongolica, Tilia mandshurica, Betula davurica, Phellodendron amurense, and Sophora japonica. Since 2004, many artificial nest boxes have been provided for secondary cavity-nesting birds in this area. The Japanese tit, Eurasian nuthatch, Daurian redstart, and yellow-rumped flycatcher readily accept artificial nest boxes for breeding, allowing detailed breeding data and the identity of breeders to be collected on a large number of nests. Approximately 400 artificial nest boxes were provided per year in our study area. The nest boxes were distributed among patches and separated by 30–50 m, with 40 to 50 boxes equally distributed per patch. The density of nest boxes was 0.24 boxes/ha. All nest boxes were the same size (inner dimensions: 12 cm × 12 cm × 25 cm, entrance diameter approximately 4.5 cm) and had removable roofs. The artificial nest boxes were hung on trees approximately 2.5–3 m above the ground. The tree species on which the nest boxes were hung and the orientation of the box holes were randomly determined.

We removed old nest materials from the artificial nest boxes before the initiation of the breeding season and routinely monitored all nest boxes from late March until mid-July. We monitored nests at least once a week (and more frequently near expected hatch/fledge dates) until the nest fledged or failed. Failed nests were classified as predated (empty nest cup with all contents removed before estimated fledge date; potential predators included snakes or chipmunks) or abandoned (dead nestlings or cold eggs on two or more consecutive visits). The breeding attempts and reproductive data (e.g., the date the first egg was laid, clutch size, hatching date, brood size, and the number of fledglings) of all four bird species (Japanese tits, yellow-rumped flycatchers, Eurasian nuthatches, and Daurian redstarts) were recorded as part of routine monitoring (Yu et al. 2017). We surveyed these forest patches and found few natural holes, with almost all breeding Japanese tits occupying the nest boxes. During the breeding season, an average of 39.8% of nest boxes were occupied (egg-laid) by birds (Japanese tits 26.3%, yellow-rumped flycatchers 8.4%, Eurasian nuthatches 2.8%, and Daurian redstarts 2.2%, unpublished data). Among the species which occupy nest boxes for breeding, Japanese tits are the most defensive in our study area and rarely find their nests usurped by other species. We determined the coordinates of occupied nest boxes using handheld GPS units with an accuracy of < 10 m. All nest boxes occupied by birds that had initiated egg-laying were considered and included in the data collection.

Almost all adult Japanese tits breeding in nest boxes in the study area were captured inside boxes during the mid-to-late chick-rearing period. The second breeding attempt was defined as an individual initiating a second clutch following a successful first clutch (at least one fledging leaving the nest) in the same breeding season (Fan et al. 2021). We defined the pairs recaptured from the first breeding attempt as second-breeding pairs (i.e., pairs with successful breeding experience). In addition, we defined pairs that initiated breeding after the first brood of chicks had fledged in the study area (i.e., those with no information on the first clutch) as late-breeding pairs (i.e., pairs without local successful breeding experience).

The phenology variables were encoded as the number of days that elapsed since the first of April (with day 1 = 1 April). The mean date that Japanese tits laid their first egg was 24.14 ± 0.38 (n=576 nests) in the first broods and 63.85 ± 0.95 (n=60 nests) in the second broods. The mean dates that Eurasian nuthatches, Daurian redstarts, and yellow-rumped flycatchers laid their first egg were 22.05 ± 0.69 (n=74 nests), 37.61 ± 2.75 (n=49 nests), and 53.45 ± 0.54 (n=240 nests), respectively. The clutch sizes of Japanese tits’ first and second broods were 11.20 ± 0.07 (range: 6–16, n=530 nests) and 8.05 ± 0.19 (range: 6–12, n=60 nests), respectively. The clutch sizes of Eurasian nuthatches, Daurian redstarts, and yellow-ridged flycatchers were 6.90 ± 0.16 (range: 6–8, n=65 nests), 6.50 ± 0.14 (range: 5–10, n=46 nests) and 5.73 ± 0.06 (range: 4–8, n=230 nests), respectively. The reproductive success rates of the first breeding Japanese tits, Eurasian nuthatches, Daurian redstarts, and yellow-ridged flycatchers were approximately 81.25% (n=576 nests), 87.84% (n=74 nests), 87.86% (n=49 nests), 79.58% (n=240 nests), respectively.

Data preparation

From 2014 to 2021, a total of 63 pairs initiated second breeding attempts (i.e., were recaptured from the first breeding attempt) with the same mate (Table 1). Among the 63 second breeding pairs, 3 pairs had been recaptured in previous years. We excluded those 3 pairs to avoid duplication in subsequent analysis. In addition, 161 late-breeding pairs were monitored in the eight years of the study (2014–2021; Table 1).

We focused on available social information (i.e., the numbers of conspecifics and heterospecifics) around the nest sites that Japanese tits selected for their second breeding attempt. We measured the Euclidean distance between the GPS points of nest sites during 2014–2021. The distance between the first and second nest sites of the 60 pairs of Japanese tits recaptured within the same breeding season was < 200 m (range: 20.1–182.9 m; mean ± standard error [SE] = 78.50 ± 6.77 m). Here, we labelled the second-breeding nests and late-breeding nests of Japanese tits as the focal nest and then assessed the social information within a 200 m radius around this focal nest at the time of the first egg-laying date of second breeding Japanese tits. We counted 1) the number of breeding conspecific nests, 2) the number of breeding heterospecific nests, 3) the number of first breeding successful conspecific nests (successfully fledged at least one young), 4) the number of successful heterospecific nests, 5) the number of first breeding failed conspecific nests (no chicks have successfully fledged), and 6) the number of failed heterospecific nests.

In addition, for each pair of second-breeding and late-breeding Japanese tits, we randomly selected one unoccupied nest box (unoccupied by any birds) in the same forest patch within the corresponding year as the control nest box. These nest boxes had not been occupied by any birds in the current year, but approximately 90% of them had been occupied in other years. We also utilized the control nest boxes as focal nests and used the first egg date of this corresponding tit pair as the node time to assess the control nest’s social information. Data on the available social information around the control nest boxes were collected using the methods described above.

We used conditional logistic regression to compare social information around the second breeding nests with paired selected control (i.e., unoccupied) nests (“control nest” = 0, “occupied nest” = 1), where the nest was the grouping variable (Thomas and Taylor 2006). The explanatory variables used in the models were 1) the number of breeding conspecific nests, 2) the number of breeding heterospecific nests, 3) the number of first breeding successful conspecific nests, 4) the number of successful heterospecific nests, 5) the number of first breeding failed conspecific nests, and 6) the number of failed heterospecific nests.

We used conditional logistic regression to compare social information around the late-breeding nests with paired selected control (i.e., unoccupied) nests (“control nest” = 0, “late-breeding nest” = 1), where the nest was the grouping variable. The explanatory variables used in the models were 1) the number of breeding conspecific nests, 2) the number of breeding heterospecific nests, 3) the number of first breeding successful conspecific nests, 4) the number of successful heterospecific nests, 5) the number of first breeding failed conspecific nests, and 6) the number of failed heterospecific nests.

We used a generalized linear mixed-effects model (GLMM) with a logit link and binomial error distribution to compare social information around second-breeding pairs and late-breeding pairs. The response variable was whether the artificial nest boxes were occupied by second-breeding pairs or late-breeding pairs (“late-brooding pairs” = 0, “second-breeding pairs” = 1). The explanatory variables were 1) the number of breeding conspecific nests, 2) the number of breeding heterospecific nests, 3) the number of first breeding successful conspecific nests, 4) the number of successful heterospecific nests, 5) the number of first breeding failed conspecific nests, and 6) the number of failed heterospecific nests. Since the data were collected over a span of several years, from 2014 to 2021, “year” was added as a random variable to avoid pseudoreplication.

The significance level was set at α= 0.05. All analyses were undertaken using the statistical software R 4.1.1 (R Core Team). We calculated all P values using Wald chi-square tests with the ANOVA function in the car package (Fox and Weisberg 2019). Conditional logistic regression models were implemented using the clogit function found within the survival package (Therneau 2015). The GLMMs were implemented using the package “lme4’’ (Bates et al. 2015).