The fiddler crab Minuca pugnax occupies thermally unstable mudflat habitats along the eastern United States coastline, where it uses behavioral thermoregulation, including burrow retreats, to manage body temperature (T_b). We explored the relationship between frequency of burrow use and environmental conditions, including burrow and surface temperatures, relative tidal height, and time of day, by twenty male M. pugnax in breeding areas around Flax Pond, New York, USA. We found a highly significant positive correlation between burrow use and surface temperature, with a clear shift to longer times underground above 32°C degrees. We also experimentally heated live crabs in the laboratory and allowed them to retreat into cooled artificial burrows while continuously measuring body temperatures (T_b). Laboratory data on cooling times were compared to field observations of burrow retreat durations. The median burrow stay in the field of 2.74 min was enough time for our laboratory crabs to capture over 70% of the cooling potential of artificial burrows 10 or 15 degrees below T_b. Because crab bodies in burrows experience exponential declines in T_b due to Newton’s law of cooling, there are diminishing returns to remaining in a burrow, and many crabs probably leave before coming to equilibrium. For M. pugnax, burrow retreats reduce time spent feeding and courting, activities that only occur on the surface. Current concerns about the impacts of climate change on animals include whether compensatory mechanisms, like more frequent and longer burrow retreats, will come at the cost of other behaviors necessary for survival and reproduction.

Field-based observations of burrow retreats and environmental temperatures

The burrow retreats of displaying Minuca pugnax males were observed in the marsh by the Flax Pond Marine Laboratory in Old Field, New York, USA (40.9633274,-73.1439265) on an open area of exposed sediment in a location otherwise dominated by Spartina alterniflora and Iva frutescens. During the study period, which began on a neap tide in June 2022, the area was not inundated during the daily tidal cycle. Video data were collected with a GoPro Hero5 camera mounted on a tripod with a boom arm that extended horizontally over the study area. Eleven recordings (1 to 3 per day) lasting between 54 and 130 minutes were made between June 7–12, 2022. Crabs were recorded throughout the tidal cycle during daylight hours between 08:45 and 15:40, although the durations varied between sampling days. From these recordings, 23 focal crabs (4–6 per day) were selected and scored for burrow retreat behaviors. Focal crabs, which were chosen based on a quick scan of the videos before any scoring took place, were individual burrow owners deemed likely to stay in view (e.g., not obscured by vegetation or near the edge of the field of view). We assumed that burrow owners seen in different video segments for the same day were the same crab and that different crabs were filmed each day as the camera was moved to different locations in the same general area. The assumption of consistent burrow occupancy was supported by the fact that we never saw a focal crab give up a burrow and leave the area during 870 total minutes of video segments. Two Tidbit temperature loggers (Onset model UTBI-001) were placed within 0.5 m of the area where crabs were video recorded each day. One was at the bottom of an artificial burrow made of an open-ended PVC tube with a diameter of 5 cm, hammered to a depth of 25 cm, and the other was on the surface of the substratum adjacent to the artificial burrow. Artificial burrows were used to obtain temperatures at a consistent depth, which would have been too difficult with crab-constructed burrows, which vary in pitch and morphology. Temperature measurements were collected at one-minute intervals continuously throughout the day.

For each video segment, focal crab scoring began 10 minutes into the recording to allow for a return to normal behavior following disturbances caused by a camera setup or a battery change. Using an instantaneous sampling approach, we scored whether each focal crab was in or out of its burrow at 1-min intervals following the acclimation period. Additionally, we recorded the durations of burrow retreats for each focal animal. A burrow retreat was scored if the entire crab disappeared from view, with the time of the retreat beginning when half of the crab’s body was inside the burrow and ending when half of its body had emerged. This approach was used because crabs exiting their burrows sometimes paused at the entrance, with the distal parts of some walking legs still inside. Waiting for the entire body to emerge would have artificially increased our estimate of the burrow duration.

Three types of burrow retreats were observed during this study: retreats following a chase by a con- or heterospecific male (another fiddler species, Leptuca pugilator), retreats related to courtship where the focal crab entered its burrow and a female followed, and retreats that were not initiated by chases or potential mates. We only included the latter type in the analysis (n = 229) because they were more likely related to thermoregulation. Although courtship and aggression-related retreats (n=13) might have turned into thermoregulation-related retreats, they were usually short, often less than 30 s, and appeared to be relatively shallow (none of the females we observed entering a focal male’s burrow remained there long enough for mating to have occurred). We did not observe any retreats associated with burrow construction or potential predators.  

Statistical analysis for field-based observations

To explore how environmental conditions and time of day impacted the frequency of burrow retreats, we fit our data to a binomial logistic linear mixed model using the glmmTMB package in R, where burrow occupancy (in versus out) was the binary response variable. Sampling day and individual crab were coded as random effects, while surface temperature, burrow temperature, time of day, and relative tidal height (scaled between 0 and 10) were fixed effects. Tidal height was estimated using the NOAA tide station water level data from Port Jefferson (station 8514560), with a lag time of 1.5 hours added (Flax Pond Marine Laboratory curator Stephen Abrams, pers. comm.). Relative tidal height was included as a variable even though the site remained exposed throughout the study because fiddler crabs show lunar periodicity for reproductive behaviors (Christy 1982, Morgan and Christy 1995). Hence, the timing of high and low tides might affect burrow-related behaviors.  In the original data set, each crab was recorded as either in or out of its burrow for each minute of sampling, resulting in 3727 total observations. However, including all data points led to a highly unstable model with an autocorrelation coefficient of 1.0, probably because crabs in burrows at one observation point were likely to still be there at the next one, as were crabs on the surface. To address the high serial correlation, binary data were combined into 20-min increments for each crab's time series, where the response variable became the number of times the crab was in the burrow for each increment. Time of day, tidal height, surface temperature, and burrow temperature were then averaged for each of these increments. This change resulted in a stable model with 196 observations and an autocorrelation of 0.75. The autoregression process (AR1) was applied to observations within crabs. Random intercepts were used to model the correlation among observations incurred by the hierarchical nature of the data, i.e., crabs were sampled within sampling day, and observations were clustered within crabs. Finally, we normalized all variables with the scale() function in R to avoid numerical instability when looking for interaction effects.

To determine if duration of burrow retreats increased with increasing surface temperatures, we calculated median time underground and median surface temperature experienced for each focal crab and performed a Spearman correlation analysis. Also, upon discovering a behavioral shift around 30–32 °C, above which focal crabs were more likely to be underground, we performed a post hoc analysis, comparing the median burrow retreat durations above and below this temperature with a Wilcoxon Rank Sum Test. These analyses were performed in R with the ggpubr package. Finally, crab body sizes were measured from video images with ImageJ by comparing crab carapace width to a 2 cm ruler that was placed on the substratum within the crab habitat. 

Laboratory-based cooling experiments

Around 20 Minuca pugnax males were collected from Wareham, Massachusetts, USA (41.7587424, -70.7148547) on five different days between June 3 and July 5, 2021, for the burrow cooling experiment and two additional investigations not described here. Crabs were brought to a laboratory at Mount Holyoke College (South Hadley, Massachusetts), where approximately ten individuals each were kept in covered plastic containers (27 x 20 x 13 cm) partly filled with mud and water from the collection site, in addition to other natural items like rocks, shells, and seagrass. Because crabs were observed to deposit feed on the exposed mud inside their containers, we did not supplement their food. Crabs were selected at random for each treatment of the burrow cooling experiment and used within a week of collection. 

For all treatments in the burrow cooling experiment, we heated individual crabs to a body temperature (T_b) of around 35 °C by placing them on damp fine white marine sand (Clifford W. Estes Company, NJ) in a glass container warmed by a hot plate, while simultaneously heating them from above with a 250-watt heat lamp bulb. While 35 °C is warmer than they prefer (Hews et al. 2021), we commonly find crabs in the field with this T_b, and crabs warmed to this temperature in the laboratory did not appear to be impaired. Once a T_b of around 35 °C was recorded (see below), crabs were allowed to enter an artificial burrow cooled to 30, 25, or 20 °C (n=10 for each temperature). Burrow temperatures of around 20-25 °C are within the range that we have seen for deeper burrows (25-30 cm) in the field, and 30°C could be experienced near the top of a burrow, closer to the surface. The burrow was constructed using a 15 mL plastic centrifuge tube with a 1.5 cm diameter (Falcon) to create a hole in damp fine marine sand that filled a beverage cooler (YJ Home). T_b was measured continuously (two measurements per 0.01s) using a flexible Type T microprobe connected to a Thermes USB Temperature Acquisition system (Physitemp Instruments, NJ).  The microprobe was inserted through a hole in the crab’s carapace into the branchial chamber and reversibly affixed to the carapace with a 50:50 pine rosin (Velesco) and beeswax (Stakich Inc., MI) mixture melted by a dental wax carving tool (SJK Lab, model J0801). Crabs were heated to the target temperature immediately following the probe insertion and then allowed to enter the burrow. Once crabs cooled to the target temperature, they were retrieved from the burrow, and the microprobe was removed. Crabs survived this procedure, and most were released at the collection site.

Cooling curves for each temperature showing average T_bs and 95% confidence intervals were constructed with the seaborn package version 0.12.1 in Python 3. Crab body sizes were determined by measuring carapace width with a digital caliper.

Excel