Spatial ecology of little egret (Egretta garzetta) in Hong Kong uncovers preference for commercial fishponds

Study area

This study was carried out in the Inner Deep Bay of the Hong Kong Special Administrative Region, China (22°29′N 114°02′E). The area consists of a natural, shallow estuarine bay with extensive intertidal mudflats connected to mangroves and human-influenced wetlands, including gei wais (tidal shrimp ponds), drainage channels and commercial fishponds. The commercial fishponds form a continuous wetland habitat of approximately 460 ha. Individual fishponds are generally 1–3 hectares in size, and contain polycultures of commercial freshwater fish, including grass carp Ctenopharyngodon idellus), grey mullet (Mugil cephalus) and tilapia (Oreochromis sp.).

Bird capturing and tracking data collection

From January–December 2018, we captured nine individuals of little egret (Egretta garzetta) using clap nets (1.5 m and 2 m in diameter) with fish bait. We put each individual into a covered, large laundry hamper. They are soft enough for the birds from getting hurt, but strong enough for retaining the birds. We attached to each bird a solar-charged GPS-UHF logger [model PICA (5.5 g in weight) or HARRIER (12 g), Ecotone Telemetry, Poland]), using Teflon tape and a backpack harness. The captured birds weighed 290–495 g. The weight of the loggers and harnesses were <3% of the birds’ weights. All birds were released within two hours at the site of capture. We programmed the loggers to record data (location and speed) hourly from 5 to 7 pm local time, thus tracking movements from around sunrise (before the egrets leave their roosting sites) to after sunset (when they return to roost). Data were automatically stored on the loggers, and were remotely downloaded every two weeks using a hand-held base station with unidirectional antenna. We included data collected between 30 Jan 2018 and 22 Mar 2019 in the analysis. All procedures were approved by the Agricultural Fisheries and Conservation Department of the Hong Kong Government [permit number: (43) AF GR CON 09/51 Pt. 6, (99) AF GR CON 09/51 Pt. 6, (166) AF GR CON 09/51 Pt. 6, (79) AF GR CON 09/51 Pt. 7].

Habitat availability

To determine habitat availability in the study area, we first mapped the study area using QGIS 3.6.1 (QGIS Development Team, 2016). Next, we delineated and classified the area into six habitat types using Google Earth: commercial fishponds, gei wais (tidal shrimp ponds), mangroves, intertidal mudflat, drainage channels and human settlement. Subsequently, we conducted fieldwork to ground-truth the habitat type. Further, we collected the draining schedules of 591 commerical fishponds by interviewing their owners throughout the study period, which covers 81.5% of all commercial fishponds in the entire Deep Bay area.

Data analysis

We performed all statistical analyses in R (R Core Team, 2019), using the packages ‘BBMM’ and ‘adehabitatHR’ for home range analysis (Calenge, 2011; Nielson, Sawyer & McDonald, 2013); ‘adehabitatHS’ for habitat selection analysis (Calenge, 2011); ‘lme4’ for model fitting (Bates et al., 2014) and ‘ggplot2’ for graphic production (Wickham, 2016).

We applied three methods to calculate the home range of individuals. First, given that the location was recorded regularly each hour, we used the Brownian bridge movement model (BBMM) to report the 50% and 95% home range as the core area and overall home range, respectively (Fischer, Walter & Avery, 2013). Based on our preliminary field testing, we set 20 m as the location errors for BBMM. We also calculated home range using fixed kernel density estimation (50% and 95% kernel) (Worton, 1989) and minimum convex polygon (MCP) (Mohr, 1947). For kernel, we used the h_ref kernel density estimators (Calenge, 2011).

To determine if little egrets exhibit habitat preference, we used compositional analysis, which compares the point habitat occurrence data with habitat availability across the entire home range of each individual (Aebischer, Robertson & Kenward, 1993). We defined the 100% minimum convex polygon (MCP) of each individual as their maximum home range, and then calculated the proportion of each habitat type as the habitat availability (Whisson, Weston & Shannon, 2015). We then assigned their relocations to the corresponding habitat and calculated the proportion of used habitat. Second, we performed Wilk’s Lambda statistic to determine their overall selection of habitat. If preferences were found, we used randomization tests to conduct pair-wise comparisons of resource types (Aebischer, Robertson & Kenward, 1993). Consequently, we used the eigenanalysis of selection ratios to examine individual variations in habitat use in different seasons (Calenge & Dufour, 2006).

We tested the seasonal effect on daily home ranges, daily travel distance and proportion of daily occurrence in fishponds (i.e., the proportion of GPS fixes on fishponds among all habitats during the daylight period) using Linear Mixed Model (LMM) and Generalized Linear Mixed Model (GLMM) followed by analysis of variance (Bolker et al., 2009). For daily home range and daily travel distance, LMMs were applied and fitted with Gaussian distribution. For the proportion of the daily occurrence in fishponds, we constructed GLMMs fitted with binomial distribution and log-linked function. We could only collect data from two individuals in the summer (June to August), so we excluded summer data from this analysis. We set bird identity as a random effect and season as a fixed effect in the model (spring: March to May; autumn: September to November; winter: December to February). Daily home range was calculated as the daily 50% and 95% utilization distribution (UD) of each individual, using the fixed Kernel Density Estimation (KDE) method (Seaman & Powell, 1996; Worton, 1989). We calculated daily travel distance by summing the travel distance between each successive location on each tracking day. We excluded data collected from 351 tracking days that had missing data.

Preference of fishponds

The preference of the little egret for commercial fishponds in the inner Deep Bay agrees with other studies on ardeids (Fidorra et al., 2016). The preference of commercial fishponds is probably associated with the draining practices that enhance food availability and accessibility (Fidorra et al., 2016; Rocha et al., 2017). In our study area, the fishponds are drained for fish harvesting, between October and May (Fig. 4). In drained fishponds, a high density of prey (e.g., fish and invertebrates) become accessible to birds in shallow water (Young, 1998). Our data showed that most little egrets rely on fishponds as the major foraging habitat from autumn to spring. Other wetland birds that likely have preference for fishponds, such as the endangered black-faced spoonbill (Platalea minor), are often seen feeding alongside little egrets in fishponds (Yu & Swennen, 2004). In light of the high development pressure on fishponds, our findings reinforce the importance of preserving commercial fishponds in the Inner Deep Bay for this group of birds.

Importance of other wetlands

Despite an obvious preference of commercial fishponds, our data indicate that other wetland types are also important to little egrets as breeding and foraging habitats (>41% of point locations in other wetland types, Table 2). In Hong Kong, little egrets mainly nest in mangrove, coastal shrubs and trees (Wong et al., 1999). One individual (PIC09) shifted from using commercial fishponds in other seasons to mangroves in the summer (breeding season). Overall, the channels, gei wai, mangroves and intertidal mudflats were used considerably, as each wetland type held about 10% point locations. Thus, despite fishponds were the major habitat for feeding, other wetland types acted as an alternative feeding grounds to little egrets when there were few drained fishponds available. Although we could not directly test the impact of habitat heterogeneity on movement and habitat use, the use of variety of wetlands by individuals implies wetland heterogeneity is vital to little egrets. Similar studies on other waterbird species would help us understand the impact of habitat heterogeneity on waterbird diversity, in turn informing land managers and governments how to best integrate biodiversity conservation into sustainable development plans.

Moreover, we detected individual variation in habitat use. One individual (PIC07) had distinctive habitat use, preferring a channel over other wetlands (including commercial fishponds) across all seasons (Fig. 2). In the channel that this individual frequented (located at Nam Sang Wai), an ecological-friendly design was implemented, including an unlined earth bottom and mangrove plantation, which has attracted a high number of ardeids and ducks (Lai, Lee & Wong, 2007). Overall, our findings also suggest that the coexistence of different wetlands is crucial for accommodating the diversity of individuals and their needs across seasons.

Seasonal variation in habitat use and movement

We detected significant seasonal differences in habitat use and movement in autumn, winter and spring. Since data were only collected from two individuals in the summer, we were unable to include this period in our analysis. According to our expectation, we found draining schedule of commercial fishponds to influence birds’ spatial ecology. The largest home range, greatest movements and highest occurrence in commercial fishponds occurred in winter. We suggest this seasonal pattern to be due to the plentiful, but unpredictable and transient nature of food availability in commercial fishponds. Winter in Hong Kong (December to February) coincides with core of the drainage schedule of commercial fishponds (October to May). Drained commercial fishponds are likely preferred because they contain a large amount of accessible prey. However, food resources in drained fishponds are usually exhausted in a few days (Rocha et al., 2017). Searching for resource-rich drained fishponds is probably frequent, but unpredictable (based on fish farmer’s preference), which may explain the greatest movements and highest occurrence in commercial fishponds during winter. Conversely, food resources in natural wetlands are likely more predictable.

The monthly cumulative area of fishponds drained in spring in the study period was found comparable to that in winter (Fig. 4), yet we found lower proportion of daily occurrences in fishpond in spring than in winter (Fig. 3), which deviates from our expectation. This hints that movement and habitat use of little egrets are shaped by a balance between foraging and reproduction constraints. In Hong Kong, the reproductive season of little egrets starts in March and April (Carey et al., 2001). When they begin sitting on nests and rearing youngs in mangrove, foraging time and movements get limited (Maccarone & Brzorad, 2005), and hence they may not be able to search for and feed in drained fishponds. Tracking more little egrets in summer and areas without fishponds will help elucidate the factors influencing the temporal changes in their habitat use and movement.