Fishing and Tagging Protocol
To assess dolphinfish behavior and habitat preferences in the EPO waters of Mexico, fish were tagged in two regions: offshore of west coast of Baja California Peninsula (WBC: 20 - 35° N, 110 - 118.0° W) and farther south in the coastal waters of Oaxaca, Mexico (OAX: 10 - 18° N, 85 - 100° W; Figure 1, Table 1). Dolphinfish tagging in WBC occurred in conjunction with the Monterey Bay Aquarium’s Animal Care Division field collection trips. Although adult dolphinfish were targeted in both regions, fishing methodology differed as a result of available fishing vessels and local expertise. Tag models and tag applications were consistent in both regions – with the exception of modifications to improve tag retention in the fall of 2014.
Dolphinfish were caught either by rod and reel (WBC) or longline (OAX). Rod and reel fishing targeted dolphinfish associated with floating kelp. The fishing rigs included SeaguarÒ 25-40 lb. test monofilament and OwnerÒ circle-style hooks (size 1/0) baited with live sardines (Sardinops sagax). Longlines were set according to local practices by participating artisanal fishermen aboard small pangas, spanned approximately 5 km in total length and soaked for 7 hours with checks every 2 hours. Hooks were baited with a mix of live jacks (4-6 cm; Family Carangidae) or pieces of black skipjack (Euthynnus lineatus).
Upon capture, dolphinfish selected for tagging were either landed onboard the vessel with a nylon sling (WBC) or held tightly alongside the panga (OAX). To minimize stress in the landed fish, their eyes were covered with a wet towel, and gills irrigated with oxygenated seawater to minimize stress. In both areas, the tagging procedure lasted 3-5 minutes.
Fish were tagged with conventional, plastic dart tags (Floy Tag Inc., Seattle, Washington, USA) and/or with electronic pop-up satellite tags (MiniPAT, Wildlife Computers Inc., Redmond, WA, USA), generally according to size with fish larger than 90 cm considered for electronic tagging. For the majority of the study, conventional and electronic tags were inserted into the dorsal musculature of smaller fish to the depth of pterygiophores. While conventional tag application was unaltered through the study, electronic tag application in WBC in fall 2014 was changed from previous deployment years to improve tag retention and, therefore, deployment duration.
Prior to fall 2014, the electronic tags were attached to nylon anchors with numbered, shrink-wrapped 300 lb. (136 kg) test monofilament leaders and inserted into the dorsal musculature as described above. However, in fall 2014, this design was altered (following Merten et al.4). The new method included a 300 lb (136 kg) monofilament leader (MomoiÒ extra hard), 40 cm in length, anchored to each tag with an appropriately sized stainless-steel crimp. The unattached end of the monofilament leader was inserted into a 15.2 cm long hollow stainless steel, cannula (.125 cm OD, .093 ID) with 45° beveled sharp edge. After a small incision was made midline above the spine and below the dorsal ridge with a surgical scalpel, the applicator tube was used to push the leader entirely through the fish anteriorly from the point of insertion at an angle of 45°, parallel to the midline of the fish. After the monofilament leader passed through the fish, it was anchored to the tag leader with another stainless-steel crimp and two affixed ½ inch diameter, circular laminated vinyl backing plates (Floy TagÒ). This resulted in a small loop around the dorsal musculature of the dolphinfish with approximately ~4 cm of monofilament leader between the electronic tag and fish.
Electronic Tag Programming Protocol
Electronic tags were programmed to release from the fish by corroding a sacrificial link after either 60 or 90 days. During deployment, electronic tags recorded depth, ocean temperature and light at 3 second intervals. However, complete archival records are only available if a tag is physically recovered and capable of being downloaded. During this study, no electronic tags were physically recovered. Therefore, archived data onboard tags, either subsampled or summarized for post-release transmission via satellite, were accessed via the Wildlife ComputersÒ data portal (mywildlifecomputers.com). Data strings of temperature and depth time-series were subsampled from the archival data and transmitted in packets of ten data points per 150 seconds; binned summaries of time-at-depth [0,5,10,15,20,30,40,50,75,100,150,>150 meters] and time-at-temperature [4,16,18,20,22,24,26,27,28,29,30,>30 °C] used 12 bins to bound and center fish depth and temperature utilization, respectively. Light and mixed layer temperature daily summaries were transmitted using default settings. Transmission priorities were set to favor the light data necessary for geolocation (WC-GPE3). Statistical analyses reported here are restricted to the transmitted time-series data and were conducted in the MATLAB computing environment (MATLAB, 2016).
To assess survivorship and tag performance, recoveries, reporting rates, and early releases were calculated as a function of region, sex, and size. Tags that did not report may have failed due to one of several undiscernible events – e.g., tag/battery failure, user error upon deployment, recovery by fishermen with subsequent intentional or unintentional disabling of transmitting capability, growth of fouling organisms that impact antenna or wet/dry sensor orientation, or animal mortality causing rapid sinking of the tag to destructive depths in deep ocean waters. Tags reporting early may be the result of “tag-shedding”, animal or tag predation resulting in free floating tags, or behavior-based triggering of early release logic (e.g. extended surface orientation). Extensive analysis of factors impacting tag reporting success are interesting, but beyond the scope of this report.
Behavioral Analysis
Semi-daily estimated geographic positions (“geolocations”) for electronic tags were calculated using proprietary software (“WC-GPE3”) from tag manufacturer, Wildlife ComputersÒ. WC-GPE3 employs a state-space model to incorporate information from tags (i.e., light curves and surface temperature), satellites (i.e., surface temperature), and an animal movement (speed) model to create a grid of position likelihoods at a spatial resolution of 0.25 degrees44. Based on the high-performance swimming of dolphinfish45,46, WC-GPE3 was run with animal speeds from 1 m/s to 5 m/s at increments of 0.25 m/s. Best model runs were chosen as those model that (i) produced the lowest areas of uncertainty and (ii) had overall scores above cut-off criteria (s_crit:)
s_crit = .95 (s_max - s_min) + s_min
where s_max is the maximum score and s_min is the minimum score (Figure 8). Area of uncertainty was calculated as the area of the grid cells that contained 95% of the cumulative uncertainty per observation, excluding the first and last days of deployment. Scores were generated internally by WC-GPE3 and represent the average fit of the tag’s observations (tag light curves and sea surface temperature) with those of the model-derived locations (pers. comm., Suzanne Kohin).
Trends in overall horizontal movements were examined as both total displacements and estimated speeds. Daily displacement rates (km d-1) were calculated according to geodesic straight-line (great circle) distance and days at liberty between release and recapture/pop-up locations for conventional and electronic tags, respectively. For geolocations, daily speeds (km d-1) were estimated using the centroid of the daily area of uncertainty. These speeds were summarized and presented as median [25th percentile, 75th percentile] unless otherwise stated. Patterns in habitat extent or range – measured by range of latitude (or longitude) as a function of days at liberty – were examined for potential of migration or mixing between the two regions.
Maximum and median depth were calculated per day observed. Depth range of the vertical habitat was also calculated as the interquartile range of daily depths. Proportion of time basking was estimated as the proportion of daytime observations in the upper 5 m. Because dolphinfish have been shown to exhibit diel diving behavior in the western central Atlantic4, diving behavior was also examined diurnally. Nighttime and daytime hours were determined using tag recorded and transmitted light curves, and the hour around sunrise and sunset were excluded. Significance of diurnal comparisons was evaluated using the Wilcoxon Sign-Rank test.
Analyses were conducted per fish and patterns in fish dive behavior were examined within the context of tagging region. As there are gaps in the temperature and depth time-series, diurnal behavior was only examined in 24-hour periods where there were depth data for greater than 50% of each day and night. Values are presented as median [25th quartile, 75th quartile] unless otherwise stated. When comparing across regions, daily values were used to capture variability both among and within fish. Significance of differences between regions was determined using the Wilcoxon Rank-Sum test unless otherwise stated.
Analysis of Thermal Habitat
A main focus of this study is to place observations of movements within the context of the animal’s utilization of the available thermal habitat. Thermal habitat utilization was characterized by daily median, minimum and sea surface temperatures determined via electronic tagging data. The percent time in the isothermal layer – ILD, defined as the maximum depth at which the water temperature remains within 0.8 °C of the sea surface temperature (average temperature of the top 5 meters47) – was calculated per fish for both day and night. Thermal habitat utilization of OAX fish and WBC fish was then compared using the Wilcoxon Sign-Rank test. Thermal habitat utilization for a day was only determined if data for greater than 50% of the day was reported (as with depth data).
To assess thermal habitat availability, sea surface temperatures were compiled for the time period of this study (2010 – 2014). The custom SST climatology for the time period and region of this study was built from the NOAA high-resolution (0.25 ° x 0.25°) blended analysis of daily SST data product (OISSTv2,48). The annual climatology presented here is the median and interquartile range of sea surface temperature observations in each of the deployment regions per day of year.