Transitional ecosystems, such as the land-sea interface, propagate nutrient flow and species interactions. Organisms spanning these boundaries act as important models for understanding of the evolution of sensory modalities that promote movement between physically distinct media, and the ecological consequences of ecosystem connectivity. Behavior is fundamentally guided by sensory processing, yet how sensory information is transmitted and collected is heavily dependent on the physical environmental medium. The flow of stimuli across the land-sea interface and the behavioral responses to stimuli are understudied. Vertebrates that span the land-sea boundary offer the opportunity to document how stimuli can be used to complete complex behaviors across transitional ecosystems. We determined that California moray eels (Gymnothorax mordax) can use chemical stimuli (odor and taste) to locate prey across intertidal boundaries on Santa Catalina Island. We tested moray responses to chemical stimuli from four prey types during high and low tidal conditions, the latter requiring emergence from the water to navigate the land-sea interface. Gymnothorax mordax can navigate to a prey source using only chemical stimuli, both when fully submerged underwater and when emerged in the intertidal. Morays showed greater discernment between prey types when exposed in the intertidal. In the intertidal eels rubbed their faces on the substrate, suggesting multimodality in using odor for detection and taste to further assist in prey location. This research broadens our understanding of ecosystem connectivity, illustrating how stimuli can cross the land-sea boundary and be used to facilitate predation through a combination of multi-sensory modalities.

Study location & sampling sites
University of Southern California’s Wrigley Marine Science Center in Big Fisherman Cove on Santa Catalina Island, California (33.4452, - 118.4847). 

Chemical stimuli
Prey Stimuli: kelp bass (Paralabrax clathratus), two-spot octopus (Octopus bimaculoides), anchovy (Engraulis sp.), and California market squid (Doryteuthis opalescens)
Control: 50 μm filtered seawater

Trial format & experimental apparatus
Our experimental apparatus was designed to administer controlled pulses of chemical stimuli into a target area within our sample site. The experimental apparatus was fixed with an underwater camera (GoPro Hero 8), recording the entire duration of the trial. Treatments were injected to the site via vinyl tubing (4 mm diameter) into a 30 cm long, 10 cm diameter white PVC tube. The PVC tube acted as a “foraging crevice” (mimicking rocky crevices in the habitat) and is henceforth referred to as the “target”. The target required morays to navigate to a specific area from where the cue was emanating. We limited our observations to a 1500 cm² framed area (50 x 30 cm) around the target tube. Trials lasted 15 minutes, with 100 ml of the treatment injected in the first minute, and 30 ml injected every two minutes with the last injection occurring at minute 13, at which point the vinyl injection hosing was flushed with seawater.

Each of the 4 prey treatments and the control were administered at high (submerged condition) and low (exposed condition) tide, with treatment order being randomized. A site would only be administered one of the four prey treatments on a given day, either at high or low tide, and was not revisited for at least 48 hours. We strived to test each site (n = 8) at both tidal levels (high/submerged and low/exposed) for each of the 5 treatments. Thus, each site was visited a total of 10 times throughout an 8-week field season. We repeated these procedures over 2 consecutive field seasons (2022 and 2023), using the same 8 sites. As such, for each treatment and tidal condition combination (e.g., control:submerged or anchovy:exposed) we have 8 trials per summer, and up to 16 total trials across two summers.

Summary of key terms
We refer to the PVC tube itself as the “target” because this is the release site of the chemical information. The immediate 1500 cm² vicinity around the target is referred to as the “frame”. Treatments refer to any of the injected liquids, with “control” referring to seawater and “stimuli” referring to any of the four prey treatments. “Submerged” refers to trials occurring fully underwater, while “exposed” refers to trials occurring at the land-sea interface, requiring morays to at least partially emerge from the water. 

Video analysis
We tallied (yes/no) if an eel entered the frame or target. An individual was considered to have “entered” the frame or target the moment their eye passed the frame edge or target entrance, as apparent chemosensory organs are anterior to the eye in morays. We documented the total duration at least one moray was present within the target (“Target Duration”), and the maximum number of individuals within the target observed at one time. For duration data, an individual was considered to have “exited” the target after the complete removal of the face, from the eye to the end of the rostrum. .

Data analysis
RStudio (Version 2024.12.0+467) was used for statistical analysis. We analyzed all data using Generalized Linear Mixed Effects Models (GLMM), with the glmer function from the lme4 package (Bates et al. 2015). This allowed us include “site” (i.e., the experimental location) and “summer” (2022 and 2023) as random effects when the addition of these parameters lowered the AIC value. 

To report differences between the control and prey treatments for trials and target visitation, we modeled the data using a GLMM with a binomial distribution. To compare the duration spent within the target we used a GLMM with a gamma distribution and log link function. If no eels were present within the trial (i.e., no eels observed within the framed area), the duration inside the target was logged as “NA”. If eels were present within the frame but failed to find the target, the duration was logged as 0 seconds in the target. We differentiated between these scenarios because we were interested in tracking the duration in the target when eels were fully known to be within the area. In cases where zero values were present in the data, a small constant value of 0.1 was added to avoid violating the assumptions of gamma distributions. For the control treatment we observed 2 morays in the submerged context and 1 moray in the exposed context throughout the two summers. As such, duration data for the control was deficient (e.g., composed of NA’s for trials without eels) and incompatible in our analysis. We excluded the control from this comparison to instead focus on the differences between the four prey treatments.

Ethical note
This research was conducted on Santa Catalina Island following California Department of Fish and Wildlife guidelines (Permit 190830002-19986-001). Morays were not handled or harmed during this experiment. 

Changes after Dec 3, 2025: 
Modified analysis code to replace absolute file paths with relative paths -- 
Data files are unchanged.

Line 81: MultiSummerData <- read.csv("../data/MultiSummer.csv")