Study area
This study was conducted during 2013–2016 on a wild breeding chestnut thrush population located at the north edge of the Lianhuashan Nature Reserve, Gansu Province, in central China (34°40’N, 103°30’E). More detail information about the reserve and local climate can be found in Sun, et al. 68. In our study area, only female chestnut thrushes select nest sites, in which they build open-cup nests in trees and shrubs during early May to late June 50,60. We searched for nests of chestnut thrushes from late April to early July in each year of the study. Clutch size was recorded when female began to incubate, and nests were monitored until nest fate was determined.
Nest site characteristics
We collected seven nest site characteristics when the nest fate failed or successfully fledge. Two scales of nest site characteristics were measured: large-scale which included distance of nest to human residences (m), distance of nest to farmland (m), distance of nest to the nearest road (m); and fine-scale measurements which included nest height (to the nearest 0.1 m), nest lateral concealment, and concealment above and below the nest. To minimize variation estimated by different observers, the same person (YB Hu) measured all nest site characteristics. All measurements followed Hu, et al. 50: nest height was defined as the distance from the nest to the ground, to the nearest 0.1 m; nest lateral concealment was measured as the estimated vegetation cover from the most obvious direction at a distance of 3 m after observer checked from all directions. Concealment above and below the nest were recorded by estimating the percentage of vertical vegetation cover at 1 m above and 1 m below each nest, respectively. In addition, we recorded the location of all nests using a Global Positioning System (GPS) receiver, and then used the ArcGIS 10.2 (ESRI, Inc., Redlands, CA, USA) to measure the distances to human residences, roads and farmland habitat (to the nearest 1 m).
Measurement of personality traits
Two personality traits, activity and breathing rate, were measured in 2013–2016. Activity was measured between 0900 and 1700 by using a simple cage test based on Kluen at al. 2012 and adapted by Zhao, et al. 26 and Lou, et al. 15. We recorded the activity behaviors (such as walk, hop and fly) for 5 min after a ten-minute acclimatization in the cage (size: L 60 × W 36 × H 60 cm) using JVC (GZ-E265AC). JWatcher software was used to analyse all recorded activity data (Blumstein et al. 2006). We calculated weighted activity scores as the following formula, to account for differences in energy expenditure: activity score = walks × 1 + hops × 2 + flights × 3 15,26. Breathing rate was measured by using the handling stress test, and we recorded the total number of breast movements within 60 seconds 15,26. In general, we assumed that birds with lower breathing rates are bolder than those with higher rates 26,63.
Ethics declarations
All birds were captured under a bird ringing license issued from the China Bird Banding Center. All methods were carried out in the study were in accordance with the guidelines and regulations of the Animal Care and Use Committee of the Institute of Zoology, the Chinese Academy of Sciences. All experimental protocols were approved by Institute of Zoology, the Chinese Academy of Sciences, Permission number 2013/ 108. To reduce human disturbance during the incubation period, only one person checked each nest at a distance of over four m from the nest. We only checked nests during good weather conditions, when the ambient temperature was relatively high. Mist netting was done only under suitable weather conditions (no precipitation and low wind speed). Birds were trapped and released into the wild within 70 minutes near the trapping locations. No trapping-related mortality or nest abandon were recorded for females.
Statistical analyses
We constructed linear mixed models (LMMs) with individual identity as a random variable to calculate the repeatability of breathing rate and activity. In the model, activity was square root transformed to meet Gaussian distribution of residuals. To control for confounding factors, seven predictor variables were included in each model: date (Julian date), year, sex, test sequence of activity and breathing rate (i.e., 1: first, 2: second or 3: third test), time of day (where 1200 = 0, 1300 = 1, 1100 = -1, etc.), context (i.e., 1: before breeding, 2: incubation or 3: nestling period) and body mass. All continuous variables were mean centered and standardized. Likelihood ratio test (LRT) between the models with and without the random effect (bird ID) was used to test the statistical significance of the repeatability. We used the LMMs of breathing rate and activity to run 1000 simulations using the arm package 69, and then used the average best linear unbiased predictors (BLUPs) for the intercepts of each individual as the personality score in the following analyses 33.
Seven separate general linear mixed models (GLMMs) were used to examine the relationships between personality traits and each nest-site characteristic, including female breathing rate and activity as fixed effects. First, we used a GLMM to analysis the effects of nest-site characteristics, female body mass, activity and breathing rate on clutch size. A GLMM with binomial distribution was used to investigate whether nest success during the incubation period was affected by nest-site characteristics, females’ activity and breathing rate. Collinearity of fixed effects were examined to ensure the Variable Inflation Factor (VIF) with values < 2.5 70. In all GLMMs, we included female ID as random effect because some repeated measurements of females among years. We also added year as a random effect to reduce the effects of environmental variations among years. All LMMs and GLMMs were fitted with lmer and glmer functions from the lme4 package 71. All statistics were performed with R software (v. 3.2.2)72.