Response to SERO sea turtle density analysis from 2007 aerial surveys of the eastern Gulf of Mexico: June 9, 2009

Transcription

1 Response to SERO sea turtle density analysis from 27 aerial surveys of the eastern Gulf of Mexico: June 9, 29 Lance P. Garrison Protected Species and Biodiversity Division Southeast Fisheries Science Center 75 Virginia Beach Drive Miami, FL Protected Species and Biodiversity Division Report #PRBD-8/9-11

2 9 June 29 From: Lance Garrison NMFS, Southeast Fisheries Science Center To: Jennifer Lee NMFS, Southeast Regional Office CC: Dennis Klemm, David Bernhardt RE: Comparison of loggerhead turtle density across depth strata in the eastern Gulf of Mexico. Previous summaries have been provided showing the distribution of loggerhead and unidentified hardshell turtles as a function of water depth over continental shelf waters of the Gulf of Mexico east of 85º 3 W longitude. These summaries were based upon the encounter rates (number of turtles per km of survey trackline) from aerial surveys conducted during the winter (January-February) and summer (July-August) of 27 (Appendix A, B). Generally, these data suggest that loggerhead turtle densities decline with increasing water depth, with steep declines in encounter rates in waters greater than 3 fathoms depth during both winter and summer months. Subsequent to that analysis, management options have been suggested to reduce the likelihood of loggerhead turtle bycatch in reef-fish fisheries employing bottom longlines. One option being considered is to limit fishing to waters exceeding 35 fathoms in bottom depth. The fishery currently operates in waters as shallow as 2 fathoms depth and as deep as 5 fathoms. The goal of the current analysis is to quantify the difference in sea turtle density (number of animals per km 2 ) between depth ranges in the eastern Gulf of Mexico. The presumption is that if sea turtle density declines with depth, then displacing fishing activities into waters deeper than 35 fathoms will reduce the spatial overlap between the fishery and loggerhead turtles thereby reducing the likelihood of bycatch. In this analysis, loggerhead sea turtle density was estimated in four depth strata: -2 fathoms, 2-35 fathoms, 35-5 fathoms, and 5-1 fathoms (Figures 1 and 2). These strata were covered by the 27 aerial surveys, though the majority of effort was expended in the shallowest stratum in waters less than 1 fathoms (Tables 1 and 2). Minimum sea turtle densities were estimated using standard line-transect distance analysis (Buckland et al. 21). Briefly, for each sea turtle group sighted, the perpendicular sighting distance (PSD) from the trackline was determined by measuring the angle to the turtle group when the airplane was perpendicular to the group location. Typically, a group of turtles consisted of one animal, though occasionally 2-3 turtles were seen on the surface in close proximity to one another. The probability distribution of these PSDs provides an estimate of the probability of observing a group of turtles conditional on two assumptions 1) that the turtles were on the surface and available for observation and 2) that all turtles occurring on the survey trackline were observed. Both of these assumptions are clearly violated, and therefore estimated densities are known to be negatively biased. Given the long dive times of sea turtles, this negative bias is likely to be significant. With the further assumption that sea turtle dive behavior is consistent across the depth gradient considered here, it

3 is reasonable to compare estimated densities as an indicator of the relative density of turtles across depth strata. Separate density estimates for each depth stratum were developed for identified loggerhead turtles and un-identified hardshell turtles. The proportion of loggerhead turtles in each depth stratum (for each survey) was calculated as: (1) p lh N = lh N lh + N g + N, kr where N is the number of turtles observed for loggerhead (lh), green (g) and Kemp s Ridley (kr) turtles (Table 1 and Table 2). The total estimated density of loggerhead turtles (D lh ) for each seasonal survey and stratum is thus: (2) D lh' = Dlh + plhdun, where D is density (number of animals per km 2 ) for identified loggerheads (lh) and unidentified hardshell turtles (un). The variance of this estimate is the sum of the variances of the two densities. The mean densities for loggerhead turtles and unidentified turtles by depth stratum are given in tables 3 and 4 for the winter aerial survey and in tables 5 and 6 for the summer aerial survey. It should be noted that the precision of the estimates in the deeper strata is generally very poor owing to the relatively small amount of survey effort in each stratum and the small number of turtle sightings. In particular, the 35-5 fathom depth stratum is relatively narrow, and thus there was a small amount of survey effort for this stratum in both seasons (223 km in winter and 242 km in summer, Tables 1 and 2). The combined estimates including both identified loggerheads and a proportion of unidentified hardshells are given in tables 7 and 8. The ratio between the estimated density in each stratum and that in the 2-35 fathom stratum is also given. As shown in previous analyses, the estimated density of loggerhead turtles declined with increasing water depth in both the winter and summer. During the winter, no turtles were observed in the 35-5 fathom depth range, and the estimated density in the 5-1 fathom depth ranges was approximately 64% (95% confidence interval: 32% - 128%) of that in the 2-35 fathom range. During summer months, the density in the 35-5 fathom depth range was 3% (95% CI: 13%-66%) of that in the 2-35 fathom range and further declined to 16% (95% CI: 7% - 32%) in the 5-1 fathom depth range. These ratios have a great deal of uncertainty due to the high degree of uncertainty in the associated estimates and the relatively low amount of survey effort in the deeper strata. The estimated loggerhead turtle densities from the 27 aerial surveys indicate that turtle density declines with increasing water depths, and that turtle densities in waters deeper than 35 fathoms are lower than those in waters shallower than 35 fathoms.

4 Literature Cited Buckland, S.T., D.R. Anderson, K.P. Burnham, J.L. Laake, D.L. Borchers, and L. Thomas. 21. Introduction to distance sampling: Estimating abundance of biological populations. Oxford University Press, 432 pp.

5 Table 1. Effort and turtle groups sighted by stratum during the winter 27 aerial survey. Group counts are on effort sightings with perpendicular sighting distances less than 3 m from the trackline. Stratum Area (km 2) Survey Effort (km) Loggerhead Turtle Groups Unid. Hardshell Turtle Groups Loggerheads / Total Id d Hardshells 2 81,384 4, , , , Table 2. Effort and turtle groups sighted by stratum during the summer 27 aerial survey. Group counts are on effort sightings with perpendicular sighting distances less than 3 m from the trackline. Stratum Area (km 2) Survey Effort (km) Loggerhead Turtle Groups Unid. Hardshell Turtle Groups Loggerheads / Total Id d Hardshells 2 81,384 4, , , ,

6 Table 3. Density and abundance estimates for loggerhead turtles during the winter 27 aerial survey. The coefficient of variation (CV) for parameter values is shown in parentheses. The mean group size across all strata was used. (mean = 1.1, CV =.32). Stratum Encounter Rate (Groups / km) Group Density (Groups / km 2 ) Animal Density (N / km 2 ) 95% Confidence Interval 2.21 (.11).629 (.136).695 (.14) (.659).259 (.664).286 (.665) (.626).142 (.631).157 (.632) Table 4. Density and abundance estimates for unid. hardshell turtles during the winter 27 aerial survey. The cofficient of variation (CV) for parameter values is shown in parentheses. The mean group size across all strata was used. (mean = 1.1, CV =.25). Stratum Encounter Rate (Groups / km) Group Density (Groups / km 2 ) Animal Density (N / km 2 ) 95% Confidence Interval (.13).975 (.121).145 (.123) (.291).591 (.298).633 (.299) (.365).373 (.371).399 (.371)

7 Table 5. Density and abundance estimates for loggerhead turtles during the summer 27 aerial survey. The cofficient of variation (CV) for parameter values is shown in parentheses. The mean group size across all strata was used. (mean = 1.3, CV =.81). Stratum Encounter Rate (Groups / km) Group Density (Groups / km 2 ) Animal Density (N / km 2 ) 95% Confidence Interval (.13).92 (.17).1227 (.134) (.266).451 (.268).62 (.28) (.627).173 (.627).231 (.633) (.862).44 (.862).59 (.866) Table 6. Density and abundance estimates for unid. hardshell turtles during the summer 27 aerial survey. The cofficient of variation (CV) for parameter values is shown in parentheses. The mean group size across all strata was used. (mean = 1.2, CV =.48). Stratum Encounter Rate (Groups / km) Group Density (Groups / km 2 ) Animal Density (N / km 2 ) 95% Confidence Interval 2.32 (.19).66 (.116).747 (.126) (.466).231 (.469).284 (.471) (.63).85 (.632).14 (.634)

8 Table 7. Combined density (animals per km 2 ) of loggerhead turtles and a proportion of unidentified hardshell turtles for each stratum during the winter 27 aerial survey. The coefficient of variaion (CV) for the ratio of densities is shown in parentheses. Stratum Combined Density CV of Combined Density 95% Confidence Limit Ratio with 2-35 fathom stratum (.426) (-) (.583) Table 8. Combined density (animals per km 2 ) of loggerhead turtles and a proportion of unidentified hardshell turtles for each stratum during the summer 27 aerial survey. The coefficient of variaion (CV) for the ratio of densities is shown in parentheses. Stratum Combined Density CV of Combined Density 95% Confidence Limit Ratio with 2-35 fathom stratum (.297) (.69).164 (.76)

9 Figure 1. Survey effort and turtle sightings east of 85.5º W longitude during the winter 27 aerial sruvey. The boundaries of strata defined by the 2, 35, 5, and 1 fathom depth contours are indicated.

10 Figure 2. Survey effort and turtle sightings east of 85.5º W longitude during the summer 27 aerial sruvey. The boundaries of strata defined by the 2, 35, 5, and 1 fathom depth contours are indicated.

11 Appendix A: Summary of Winter and Summer Eastern Gulf of Mexico Aerial Survey Data for Loggerhead Turtle Distribution Lance Garrison Southeast Fisheries Science Center 15 January 29 Aerial surveys were conducted in nearshore and continental shelf waters of the eastern U.S. Gulf of Mexico from 16 January to 28 February and from 17 July to 8 August 27 to estimate abundance and assess spatial distribution of bottlenose dolphins and sea turtles. A modified DeHavilland DHC-6 Twin Otter was used to conduct the line-transect surveys. Transects were flown at an altitude of 23 m (75 ft) at an airspeed of 185 km/hr (1 knots). The survey team consisted of an observer stationed at each of two forward bubble windows and a third observer stationed at a belly window that monitored the trackline. A fourth observer recorded data using a computerized data acquisition system interfaced with a GPS. Trackline data were recorded continuously at 1 second intervals including time, date, latitude, longitude, speed, and heading of the aircraft. In addition, environmental parameters were recorded including weather condition, visibility, water color, water turbidity, sea state, and glare conditions. Surveys were only flown during favorable sighting conditions, generally at sea states less than or equal to 3 (surface winds typically <1 knots). Survey transects were oriented perpendicular to the shoreline and bathymetry, and the majority of survey effort was expended in nearshore waters with depths between -2 m (Figures 1 and 2). Forward observers searched for marine mammals and turtles from directly beneath the aircraft out to a perpendicular distance of approximately 6m. Due to the bubble configuration of the observing windows and the position of the belly window observer, the survey trackline could be continuously and reliably visualized. Upon sighting marine mammals or turtles, the observer measured the angle to the animal (or group) using an inclinometer or estimated the angle based upon markings on the windows indicating 1-degree intervals. This sighting angle, θ, is converted to the perpendicular distance from the trackline (PSD) by PSD = tan(θ) x Altitude. In the case of sea turtles, species identifications were based upon shell shape, head shape, and size. Leatherback turtles are easily distinguished from hardshell turtles; however, it can be problematic to distinguish clearly between the four hardshell species observed in these surveys (loggerhead turtles, green turtles, hawksbill turtles, and Kemp s Ridley turtles). In particular, the distinction between green and hawksbill turtles is difficult from the air, and these two species generally overlap in the southern portion of the survey habitat. Hence, turtles were identified into leatherback, loggerheads, Kemp s Ridley, and a combined Green/Hawksbill category. In addition, many turtles could not be clearly identified and were classified as unidentified hardshell turtles. The winter survey covered a total of 8,9 km of trackline on effort while the summer survey covered 9, km. For sea turtles, the survey observed 27 loggerhead turtles, 19 Kemp s Ridley turtles, 35 green/hawksbill turtles, and 34 leatherback turtles (Figure 1). In addition, there were 414 un-identified hardshell turtles observed. In the summer survey, turtle sightings included 37 loggerhead turtles, 64 Kemp s Ridley turtles, 152 green/hawksbill turtles,

12 and 36 leatherback turtles. In addition, there were 227 un-identified hardshell turtle sightings. (Figure 2). The relative density of loggerhead turtles in surface waters is reflected in the encounter rate (number observed per kilometer of trackline) of turtles during the survey. This assumes that both the probability of seeing a turtle at the surface and the probability of turtles being on the surface are constant throughout the spatial and temporal range of a given survey. In addition, there may be behavioral differences in surface-dive intervals between seasons, and hence the relative density (or abundance) estimated by the surveys may not be comparable across seasons. Basic summaries of turtle distribution in the winter (Figure 3) and summer (Figure 4) by depth range and latitude grossly represent the spatial distribution of loggerhead turtles observed during the survey accounting for the spatial distribution of survey effort. The latitudinal distribution plots are most representative of nearshore waters (<2 m) depth, as this is where the majority of the survey effort occurred. In the winter survey, the encounter rate of turtles was highest in nearshore waters with depths less than 6 fathoms. The encounter rate drops off rapidly as waters are deeper than 6 fathoms, though localized high densities of turtles were observed in deeper water (Figure 3a). The highest encounter rates of turtles were observed in the northern and southern portion of the range with relatively low encounter rates between 26. and 28. degrees North (Figure 3b). During summer months, the encounter rates of loggerhead turtles were much higher in nearshore waters with encounters decreasing rapidly with increasing water depth. Encounter rates were very low in waters deeper than 4 fathoms (Figure 4a). The latitudinal distribution was generally uniform with peak encounters in the southern portion of the survey range and a localized high encounter rate between degrees N (Figure 4b). This latter peak may be an artifact of relatively low survey effort in this latitude range. In particular, there was no offshore data collected within this latitude range in either the summer or winter. Since nearshore densities were quite high in the summer, this lack of offshore data likely is responsible for this apparent peak which should not be interpreted as a localized area of high density.

13 Figure A1. Loggerhead turtles sightings and survey effort during winter 27 (16 January to 28 February)

14 Figure A2. Loggerhead turtles sightings and survey effort during summer 27 (17 July to 8 August)

15 Figure A3. Loggerhead turtle encounter rate (number of turtles per km of trackline) summarized by (A) depth and (B) latitude from the winter 27 aerial survey. A B

16 Figure A4. Loggerhead turtle encounter rate (number of turtles per km of trackline) summarized by (A) depth and (B) latitude from the summer 27 aerial survey. A B

17 Appendix B: Updated Summary of Winter and Summer Eastern Gulf of Mexico Aerial Survey Data for Loggerhead Turtle Distribution UPDATE: Data summary restricted to areas east of degrees longitude Lance Garrison Southeast Fisheries Science Center 25 February 29 In this document, I provide summary information on the spatial distribution of sea turtles in the eatern Gulf of Mexico observed during aerial line transect surveys conducted in the winter (January February) and summer (July August) of 27. The aerial survey design and execution was provided in a previous memo (dated 15 January 29). The previous summary described the relative encounter rate, as a proxy for density, for loggerhead turtles throughout the range of the survey extending from the mouth of the Mississippi River to the Dry Tortugas. However, the area of interest for turtle interactions with bottom longline fisheries is restricted to the eastern portion of that range along the West Florida continental shelf. This summary, therefore, evaluates the aerial survey data east of 85º3 W longitude. During the aerial survey, leatherback turtles are easily distinguished from hardshell turtles; however, it can be problematic to distinguish clearly between the four hardshell species observed in these surveys (loggerhead turtles, green turtles, hawksbill turtles, and Kemp s Ridley turtles). In particular, the distinction between green and hawksbill turtles is difficult from the air, and these two species generally overlap in the southern portion of the survey habitat. Hence, turtles were identified into leatherback, loggerheads, Kemp s Ridley, and a combined Green/Hawksbill category. In addition, many turtles could not be clearly identified and were classified as unidentified hardshell turtles. There were a relatively large number of unidentified hardshell turtles during these surveys. To address this problem, a neighborhood averaging approach was used to apportion the unidentified turtles to a particular species. For a given 4x4 km 2 spatial cell, a proportion of observed unidentified hardshell turtles was allocated to each observed species (loggerhead, Kemp s Ridley, and green/hawksbill) based upon the species composition of the identified hardshell turtles in its immediate neighborhood. Due to the relatively sparse nature of the turtle data, a 6x6 cell neighborhood around the target cell. As an example, for a cell containing 3 unidentified hardshell turtles, the species composition of the surrounding neighborhood was evaluated. If in this neighborhood, the observed turtles included 3% loggerhead turtles and 7% Kemp s Ridley turtles, then the unidentified hardshells were apportioned into.9 loggerheads and 2.1 Kemp s Ridleys. This apportioned number was then added to any observed and identified turtles of each species. Using this approach the relative encountere rates account for the relatively high number of unidentified turtles and used the observed spatial patterns of identified turtles to partition these sightings. Data summaries by depth (1 fathom bins) and latitude (.5 degree bins) are presented here for loggerhead turtles. Similar summaries are provided in a companion file for leatherbacks, Kemp s Ridely, and Green/Hawksbill turtles. In winter months, there was an apparent bimodal distribution of loggerhead turtles with higher densities in nearshore waters (- 3 fathoms) and a region of lower density in deeper waters between 6-8 fathoms. The

18 inclusion of hardshell turtles slightly expands this offshore range to 5-8 fathoms (Figure 1). The latitudinal distribution of loggerheads in winter is also bimodal, with higher densities south of 26 degrees latitude and north of 28.5 degrees (Figure 2). The inclusion of hardshells has little effect on these broad spatial patterns. In summer months, loggerhead turtles occurred in high densities slightly further away from shore compared to winter with the region of higher encounter rates extending to 4 fathoms. As with winter, there was another area of lower density offshore in waters of 7-9 fathoms (Figure 3). The latitudinal distribution is generally uniform during summer with slightly higher densities in the southern portion of the range (Figure 4). The peak in density between is an artifact of low survey effort in deep/offshore waters where densities are lower within this latitude range.

19 Figure B1. Loggerhead turtle encounter rate (number of turtles per km of trackline) as a function of depth during the winter survey. Plots include (A) identified loggerheads, (B) loggerheads + all unidentified hardshells, and (C) loggerheads with apportioned hardshells based on neighborhood averaging. Loggerhead Turtles.3.25 A < >12 Depth (fathoms) Loggerhead+ Unid. Hardshell Turtles B.1 < >12 Depth (fathoms) Loggerhead+ Apportioned Hardshell Turtles.7.6 C < >12 Depth (fathoms)

20 Figure B2. Loggerhead turtle encounter rate (number of turtles per km of trackline) as a function of latitude during the winter survey. Plots include (A) identified loggerheads, (B) loggerheads + all unidentified hardshells, and (C) loggerheads with apportioned hardshells based on neighborhood averaging. Loggerhead Turtles A Latitude (degrees N) Loggerhead+ Unid. Hardshell Turtles B Latitude (degrees N) Loggerhead+ Apportioned Hardshell Turtles C Latitude (degrees N)

21 Figure B3. Loggerhead turtle encounter rate (number of turtles per km of trackline) as a function of depth during the summer survey. Plots include (A) identified loggerheads, (B) loggerheads + all unidentified hardshells, and (C) loggerheads with apportioned hardshells based on neighborhood averaging. Loggerhead Turtles A.1 < >14 Depth (fathoms) Loggerhead+ Unid. Hardshell Turtles B.2 < >14 Depth (fathoms) Loggerhead+ Apportioned Hardshell Turtles C.2 < >14 Depth (fathoms)

22 Figure B4. Loggerhead turtle encounter rate (number of turtles per km of trackline) as a function of latitude during the summer survey. Plots include (A) identified loggerheads, (B) loggerheads + all unidentified hardshells, and (C) loggerheads with apportioned hardshells based on neighborhood averaging. Loggerhead Turtles A Latitide (degrees N) Loggerhead+ Unid. Hardshell Turtles B Latitude (degrees N) Loggerhead+ Apportioned Hardshell Turtles C Latitude (degrees N)