Edited by: Lyne Morissette, M – Expertise Marine, Canada
Reviewed by: Rebecca Ruth McIntosh, Phillip Island Nature Parks, Australia; Gail Schofield, Queen Mary University of London, United Kingdom
This article was submitted to Marine Megafauna, a section of the journal Frontiers in Marine Science
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Identifying critical aggregation sites and behavioral patterns of imperiled species contributes to filling knowledge gaps essential for their conservation. Manta rays present a prominent example of such species, the populations of which are declining globally due to directed fishery, by-catch, and other anthropogenic stressors. Our goal was to explore manta ray aggregation sites in the Philippines in order to determine the factors governing the mantas’ visits – a knowledge gap essential to understand manta ecology, facilitate ecosystem-based fishery management, and promote sustainable manta-based ecotourism. Diving surveys, environmental conditions assessment, and autonomous cameras were employed to study manta behavior and visit patterns to a cleaning station cluster on a commonly fished seamount, visited by both
Knowledge of spatial and behavioral ecology is essential for effective conservation (
Manta rays present a prominent example of such wide-ranging mobile species. The largest batoid and among the largest filter-feeders in the sea, their habitat may cover vast areas (
For a management plan to be effective it must consider the ecology of the species. Mantas tend to aggregate at specific locations in order to feed and clean (
This study was carried out at a submerged seamount in the central Philippines, and sought to identify predictors of manta presence and absence at the site and to shed light on manta ecology in core habitat use areas (sensu
The study was conducted at “Manta Bowl” (locally known as Bondot Tacdogan; 12.6597°N, 123.7550°E), a submerged reef on the top of a seamount near Ticao Island, the Philippines. The seamount area is ca. 0.4 km2 and the reef flat is ∼0.06 km2, with a crest at a depth of 20 m that gently slopes down to 50 m and deeper, to a surrounding depth of ∼200 m. The Manta Bowl area experiences strong tidal currents and both upwelling and downwelling currents. The flow regime and plankton-rich water create optimal conditions for resident reef-dwelling fish, schools of semi-pelagic fish such as Jacks (family
Surveys were conducted on the northernmost part of the seamount, within a rectangular area (ca. 200 m by 150 m) at a depth of 22 m sloping to 28 m toward north, west and east. The area is characterized by a patchy occurrence of soft corals interspersed with several types of massive coral species. The fish community comprises mainly reef and semi-pelagic fish species, with several pelagic species regularly frequenting the area. The survey area has a large concentration of cleaning stations inhabited by the cleaner wrasses
Data were acquired by deployment of autonomous cameras (Midland XTC-400
All field work was carried out by the same two-man team. A total of 119 survey days were performed, constituting 28 days of preliminary surveys conducted in October 2012 and 91 survey days carried out between July and December 2014, which comprised the main portion of the study. A typical survey day comprised 3 dives, averaging 30–45 min/dive for the purpose of camera deployment near the selected cleaning stations. Cameras were positioned and turned on as quickly as possible to minimize disturbance. In situations in which stations were occupied by visiting megafauna species, the team would wait at a distance until the visitor had left the station, and then deploy the camera. Manta rays that were sighted during dives were not included in the analyses to avoid bias caused by diver presence.
Autonomous cameras were used for data acquisition in order to eliminate the risk of mantas being deterred by diver presence, as well as to provide continuous coverage of the site, enabling a “presence and absence” (rather than “presence only”) analysis. An array of cameras, each set to HD video mode and attached to a 2 kg weight, was used. Each camera was left to run for 3.5–4.3 h (until its battery was depleted), and was then retrieved. Two recording approaches were applied to account for diving safety limitations: (a) two consecutive days during which two cameras were working simultaneously at two separate sites until battery depletion, retrieved and then immediately replaced by fully charged cameras to continue recording, resulting in ∼7 h of coverage at each of the two sites; followed by (b) 1 day during which the team deployed 1–3 cameras working simultaneously at different sites, retrieved upon battery deletion, with no replacement, resulting in ∼3.5 h coverage at the sites. Cameras were deployed a total of 263 times, yielding ∼960 h of video. The time of first deployment was routinely changed between 07:00 and 10:00 to ensure sufficient coverage of the day.
Individual manta identification was performed by noting each animal’s unique ventral spot pattern (
Each survey day was divided into 1-h periods. Environmental conditions and number of manta cleaning events were noted for each period.
Events were classifieds according to their time of onset. A single event was defined as the time from the appearance of an individual manta ray until it had left the cleaning station for at least 5 min. If the same individual re-visited either of the two stations more than 20 min after it had previously left it, this was considered as a separate event.
Environmental conditions were measured, estimated, or taken from various data sources (
Environmental factor, parameter statistical behavior, source, and significance as predictors of manta presence.
Environmental factor | Units | Parameter in the model | Source | Significance as predictor |
---|---|---|---|---|
Time | Hours (8:00–17:00) | Factorial | A clock | + |
Cloud coverage | 0–8 | Factorial | + | |
Sea state | 0–6 | Factorial | Measured |
+ |
Portion of the moon illuminated | Less than half; half lit or more | Binomial | + | |
Flow speed | 0–3 (in 0.5 increments) | Continuous | Measured |
+ |
Flow direction | South to north; north to south | Binomial | Measured |
- |
Monsoon season | Amihan; habagat | Binomial | Calendar | - |
Tide | Flood; ebb; slack | Factorial | - | |
The identification of potential cleaning stations was made in accordance with
Nine cleaning stations were identified during the preliminary phase. Two stations (Tamis Rock and Banger-II) yielded the largest number of sightings, and therefore were selected as the target sites for the data-acquisition phase of the study. The two stations were located ca. 180 m apart, facing different directions, and were very different in terms of their bathymetry; therefore, they were considered to be independent. The data obtained from the stations were not pooled but analyzed independently to enable comparison between the two.
Environmental variables (
The data were analyzed using a Generalized Linear Model (GLM; constructed in the program R), using a Negative-Binomial distribution with raw sightings per hour set as the response and environmental variables as predictors (
The outcome shows the partial contribution of each statistically significant factor to the probability of manta visits to the cleaning stations. All error bars and ribbons in the graphs represent standard errors; α was set to 0.05.
Model selection process was carried out individually for each cleaning station through Backward Elimination (omitting the least statistically significant variable from the model in each step), and comparison of AIC scores (
Model selection process for Tamis Rock cleaning station.
Variable | AIC | ΔAIC | Pseudo R2 | ||
---|---|---|---|---|---|
1 | Time + cloud coverage + sea state + flow speed + Portion of the moon illuminated | 1165.9 | 0 | 0.12 | <0.0001 |
2 | Cloud coverage + sea state + flow speed + portion of the moon illuminated | 1167.00 | 1.1 | 0.9 | <0.0001 |
3 | Time + cloud coverage + sea state + flow speed + monsoon season + portion of the moon illuminated | 1168.00 | 2.1 | 0.12 | <0.0001 |
4 | Time + cloud coverage + sea state + flow direction + flow speed + monsoon season + portion of the moon illuminated | 1168.22 | 2.3 | 0.12 | <0.0001 |
Model selection process for Banger-II cleaning station.
Variable | AIC | ΔAIC | Pseudo R2 | ||
---|---|---|---|---|---|
1 | Sea state + portion of the moon illuminated | 836.45 | 0 | 0.13 | <0.0001 |
2 | Time + sea state + portion of the moon illuminated | 840.5 | 4.05 | 0.16 | <0.0001 |
3 | Portion of the moon illuminated | 841.99 | 5.54 | 0.03 | <0.0001 |
4 | Time + cloud coverage + sea state + portion of the moon illuminated | 846.62 | 10.17 | 0.19 | <0.0001 |
5 | Time + cloud coverage + sea state + flow direction + portion of the moon illuminated | 848.53 | 12.08 | 0.19 | <0.0001 |
6 | Time + cloud coverage + sea state + flow direction + flow speed + portion of the moon illuminated | 850.53 | 14.08 | 0.19 | <0.0001 |
7 | Time + cloud coverage + sea state + flow direction + flow speed + monsoon season + portion of the moon illuminated | 852.35 | 15.9 | 0.19 | <0.0001 |
A total of 876 manta visits were analyzed, and 8 environmental factors were analyzed (
The output of the analysis shows the partial contribution of each variable on the number of estimated cleaning events at each cleaning station. Although the variables themselves were statistically significant, not all states within all variables were significant. The analysis findings were as follows:
Both stations exhibited similarity in the trend of the effect, although not all levels were found significant. Manta presence at the Tamis Rock cleaning station was low during calm-to-moderate sea states (level 3,
Generalized linear model output – Partial effect of environmental factors on manta presence at the two cleaning stations (Banger-II and Tamis Rock) in the Manta Bowl seamount. Results of the analysis of the interaction between environmental factors and manta cleaning events in Tamis Rock (colored blue) and Banger-II (colored red). Each plot presents the partial contribution of the factor as a predictor of manta cleaning events. Factors in plots
Manta presence at the cleaning stations was significantly higher when the moon was less than half full at both Tamis Rock (
Manta ray presence was significantly lower at 12:00 and 16:00 (
Manta presence showed a reverse correlation to the flow speed at Tamis Rock (
Higher manta presence in the cleaning stations correlated with higher levels of cloud coverage at Tamis Rock (levels 5, 6, 7, and 8;
Five environmental factors were found to be potential predictors of manta ray presence/absence at the cleaning stations: Sea state, Moon illumination, Time of day, Flow speed, and Cloud coverage. Due to a correlation between the Tide and Moon factors, only the Moon variable was analyzed. These factors affect the marine environment mainly via (a) light characteristic in the water (
Being large planktivores, mantas consume large quantities of plankton (
Assuming that both cleaning and feeding are crucial for the mantas’ survival, there is probably a trade-off between the two activities (
Environmental conditions dictating manta ray behavior and location. A summary of the proposed mechanism of how Time of day (particularly the angle of the sun), Sea state, Cloud coverage, and Moon illumination affect plankton movement and distribution, and hence manta presence at the cleaning stations. At times of high light intensity plankton undergo mass coordinated movement, thus their density and concentration increase, facilitating prime foraging conditions and resulting in a decline in manta presence at the cleaning stations. Light intensity is reduced by moderate to high cloud coverage, reduced angle of the sun (affected by time of day), less illuminated moon phases, and moderate sea state. Not pictured but addressed in the study are the effects of cleaner wrasse presence and strong turbulence.
Manta feeding effectiveness is dictated by the density and concentration of its planktonic prey (
We found that manta ray presence at the cleaning stations was correlated with those conditions that may cause low light intensity over a short time scale (i.e., hours), affected mainly by the time of day, sea surface state, and cloud coverage; or long time scale (i.e., days), affected by the moon illumination. Under low-light conditions, plankton concentration will be low, as there will be no plankton mass coordinated movement due to a lack of environmental cues. When the plankton are scattered, foraging will be less efficient (
In contrast, mantas were notably absent from the cleaning stations at midday and when the sky was clear of clouds, as well as during periods when the moon was more than half full. Under these conditions, light intensity in the water is high, resulting in the formation of plankton aggregations, triggering manta foraging behavior (
Cleaning effectiveness is mostly affected by the performance of the cleaner wrasses. Cleaning at the study site was found to be performed almost solely by the diurnal blue streak cleaner wrasse
We found that manta ray presence at the cleaning stations correlated with the activity hours of the cleaner wrasses and with conditions of relatively slow water flow (
This study integrates a wide range of findings regarding both biotic and abiotic aspects of the marine environment, with new observations based on presence and absence datasets rather than presence only datasets. This innovative approach provides insights into the manta rays’ ecology and behavior: notably, their decision-making regarding energy-efficient feeding or effective cleaning. Five environmental factors, which affect either the mantas’ planktonic prey or the cleaner wrasses via phototactic and hydrodynamic induced processes (
Identifying the critical habitats of manta rays, such as off-shore cleaning stations, and understanding the patterns of their visits there, are essential for the effective ecosystem-based management and conservation of manta rays. The accessible depth and rich fauna of these habitats make them attractive for fishing and, consequently, mantas visiting there are more vulnerable to fishing, bycatch and, incidental injury. However, the same sites can also be harnessed to the design of sustainable fishery management and protection.
Our findings suggest that several easily measured environmental factors can serve as predictors of manta ray behavior and habitat use, i.e., their presence at cleaning stations. The use of these predictors can be applied as an efficient ecosystem-based management tool to minimize manta ray bycatch or incidental injury, by, for instance, limiting fishing in the vicinity of the cleaning stations during times of high likelihood of manta presence. Likewise, the same prediction-based tools may be used to improve manta-based ecotourism, by embarking on manta encounter dives at sites and times where manta presence is more likely, and thereby increasing the chances of a manta encounter, and improving ecotourism. Such applications may promote socio-ecological setups that enable manta population recovery while also maintaining coastal communities’ livelihood. In contrast to total fishery bans, which may drive entire populations and towns to hunger and unemployment, or to the emergence of illegal “underground markets,” such setups have a higher chance of successful implementation.
The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.
YB designed the study concept, performed the fieldwork and analyses, and drafted the manuscript. AA supervised the scientific integrity, provided insights into the study’s scientific and ecological significance, and revised the manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors wish to thank Mr. Medel Silvosa and Mr. Marvin Mondrano for their role in the study’s fieldwork, Dr. Simon Oliver of TSRPC, Liat Shenhav and Or Givan of Tel-Aviv University, and Dotan Shalev of