For migratory birds that specialize on particular benthic macroinvertebrate species, the timing of migration is critical since prey availability may be temporally limited and a function of local ambient temperature. Hence, variation in local ambient temperature can influence the diet composition of migrant birds, and consequently they may be constrained by which stopover and wintering sites they are able to utilize during periods of colder temperatures. Here we use faecal analysis, observer-based population counts, digital video-recordings, and temperature data to test five predictions regarding the influence of local ambient temperature on the activity and availability of mudflat crabs - a key prey resource at three staging/wintering sites in eastern China, for migratory Red-crowned cranes (Grus japonensis) and how this subsequently influences crane diet and use of wetland sites. Pearson’s correlations and generalised linear models revealed that mudflat crabs became significantly more surface active with increasing burrow ambient temperature. Piecewise regression analysis revealed that crab surface activity was largely limited to a burrow ambient temperature threshold between 12~13℃ after which activity significantly increased. Crab activity declining temporally during the crane’s autumn migration period but increased during spring migration. Crabs accounted for a significant proportion of crane diet at two of three sites, however the frequency of crab remains was significantly different between sites, and between autumn and spring migration. Analyses of crane count data revealed a degree of congruence between the migration timing of Red-crowned cranes with periods of warmer ambient temperature, and a significant, positive correlation between the percentage of crab remains in crane faeces and site ambient temperature. Collectively our data suggests that temperature-related mudflat crab activity may provide an important time window for migratory Red-crowned cranes to utilise critical stop-over sites and the crabs’ food resources.

Crab activity and ambient temperature data

The activities of mudflat crabs around their burrows were filmed using two digital cameras (Xiaoyi 4K, Shanghai, China). We randomly selected up to two crab burrows per day to film, with the camera positioned 0.5 m above the ground, to ensure an unobstructed view to the burrow entrance. Each recording lasted 8-9 h from 8:00 am to 16.00 hrs or 17.00 hrs. Recordings were made at YRD from November 7 to December 30 in 2014, and February 26 to March 28 in 2015, and at LRD from November 7 to November 29, 2014. No recordings were made during the spring of 2015 at LRD, because crabs were still in hibernation due to low ambient temperatures during Red-crowned crane migration. In addition, no recordings were made at YNR because of limited logistical resources, but previous data has revealed that the Red-crowned crane populations wintering there do feed on the crabs during the winter time (e.g. Ma et al. 1999). In total, we obtained 96 video samples of crab activity (LRD-autumn: n=12; YRD-autumn: n=36; YRD-spring: n=48). 

Burrow ambient temperature was recorded at the entrance of each crab burrow at a depth of 2 cm using the Tinytag Plus 2 temperature data logger (TGP-4520; Gemini Data Loggers, UK). We used one data logger per site, with LRD temperature sampling conducted during September 16 to December 24, 2014 (this logger failed to work during the following spring season) and YRD sampling conducted during November 20 to April 4, 2015. We selected one burrow for temperature data recording and did not move the data logger to a different burrow each day as we suspected there would be minimal variation in ambient temperature between different burrows. Crab activity at the burrow entrance (e.g. onset of activity and the percentage of activity time) were extracted from the video recordings using the Baofeng 5.0 digital player. Daily maximum and mean burrow ambient temperatures for each day were calculated to examine the influence of temperature on crab activity. Daily maximum ambient temperature data for the time period October 1 2014 to April 30 2015 for each site were downloaded from the website (http://www.tianqihoubao.com). These site-related daily maximum ambient temperatures were used to explore the relationship between diet composition and migration timing of Red-crowned cranes.

At YRD we conducted additional monitoring of crab activity at burrow entrances. Every 2-3 days we randomly selected a sampling area measuring 5 m x 10 m within the S. salsa habitat to conduct the experiment. Ten plots measuring 1 m x 1 m were randomly selected within the sample area. From each plot we randomly selected 10 crab burrows and the location of which were marked with wooden poles. No sampling plot was repeatedly sampled and the distance between plots was no less than 10 m. Before 9am on each sampling day, we plugged the entrance of the burrows with mud from the immediate environment around the burrow entrance. We then checked each burrow 24 h later to determine whether the burrow was open or closed and used these data as an indication of crab activity. 

Faecal sample collection and prey identification

We collected a total of 902 fresh faecal samples of Red-crowned cranes from their foraging or roosting sites between 2011-2015 (Table 1). Crane feces were easily distinguishable from that of other species by their large amorphous volume and always included large amounts of crab remains. On just a few occasions we found Red-crowned cranes feeding in the same habitat with Common crane (Grus grus) or Siberian crane (Leucogeranus leucogeranus), and only on these occasions we did not collect any faecal samples to avoid any error with allocating samples to the different crane species. We limited the number of faecal samples to ≤3 samples collected from each foraging or roost site to reduce the potential for pseudo-replication. Only newly defecated samples were collected from the ground using a sterilized spoon, and these were subsequently stored in a sample tube and taken back to the field station (<8 h travel time) and stored in refrigerator at -20℃ before analysis. 

Faecal analysis was conducted following the protocol of Li et al. (2014). Samples were disinfected by ultraviolet light for 30 min, then placed over a 0.3 mm sieve and scoured under tap water for 10 min to separate soil and other matter. Indigestible parts were identified using a stereomicroscope and food items were identified to the lowest possible taxonomic level, aided by comparisons with collected prey specimens. The percentage (%) of each type of prey remains for each sample and the percentage occurrence of prey remains in the total sample per season in each site were calculated to represent site and seasonal variations in crane diet composition. 

Counts of migratory Red-crowned crane

Red-crowned crane population counts were conducted during the migration seasons in the YRD (November 1 to December 26, 2014; February 26 to March 26, 2015) and LRD (October 5 to November 30, 2014; March 10-11, 2015). Counts were conducted within the coastal S. salsa saltmarshes and adjacent intertidal mudflats known to be the main foraging habitat for Red-crowned cranes (Li et al. 2017). We selected five vantage points at YRD and six vantage points at LRD, all were situated along the shoreline and from which it was easy to count all the individual cranes present in the coastal tidal flat. Each point was separated by a distance of approximately 2 km. Vantage points were visited in the same order for each count by two experienced observers (D.L. and J.Z.) spending 10 min counting cranes before moving to the next vantage point. Cranes were counted between 8:00 am to 15.00 pm every 1-3 days and only during suitable weather conditions (i.e. no rain or strong winds) using telescopes (Swarovski ATS 80HD). Subsequently we examined all count data from all points per day to exclude the possibility of double counting the same individuals from neighboring vantage points. 

Statistical analyses

Crab activity datasets from all sites were pooled for the analyses without considering inter-site differences. To test our first prediction, that mudflat crabs become more surface active with increasing burrow ambient temperature, we first used Pearson’s correlations to examine the relationships between maximum daily and mean daily burrow ambient temperature with the onset (time) of crab activity time, and with the mean percentage of active crab burrows. In addition, we fitted two generalized linear models (GLMs) with the percentage time of crab activity at burrow entrances and percentage time crab activity outside of burrows on the mudflat as the response variables, with site (YRD, LRD), season (spring, autumn), daily maximum burrow ambient temperature and mean burrow ambient temperature included as predictor variables. Since both burrow ambient temperature response variables were positively correlated (r=0.899, P<0.001) we built these variables into the different models separately. We ran the GLMs with Poisson error structure and logit link function using the glm function included in the MASS package. We examined Wald test z scores to make inferences about each parameter estimate. 

To test our second prediction (that crab surface activity would be limited to an ambient temperature threshold), we used a piecewise linear regression model to explore the relationships between percentage time of crab surface activity (combining the activity time both on the burrow entrance and outside the burrow on the mudflat) and maximum or mean burrow ambient temperature using the R package segmented (Muggeo 2008). Differences in the burrow temperature at the onset of crab activity between different crane migration seasons were examined using independent two-sample t tests. For our third prediction (mean percentage of active crab burrows would decline temporally during the crane’s autumn migration period but increase during the spring migration period with increasing burrow ambient temperature) we fitted the data using Pearson’s correlations. To test our fourth prediction (crane diet would vary across different stopover/wintering sites and between autumn and spring migration seasons), we examined seasonal site differences in the frequency of the presence of crab remains and the percentage of crab remains in crane faecal samples using Chi-square tests. We pooled all faecal sample data since we hypothesized that there was little variation in the foraging microhabitat of Red-crowned cranes across the S. salsa wetland and that crab specialisation was a function of ambient temperature. Finally, to test our fifth prediction, that cranes select and use staging/wintering sites with mudflat crabs when the site ambient temperature provides them with the opportunity to do so, we examine the percentage of crab remains in all faecal samples with daily maximum and daily mean site ambient temperatures with Pearson’s correlations. All statistical analyses were conducted using R 3.6.0 (R Core Team, 2016), with significance set at 0.05 and the results expressed as mean ± standard error (SE).

No further information.