694 



Fishery Bulletin 98(4) 



and corresponding national programs have worked on 

 board tuna vessels. Since 1991, the PNAAPD has placed 

 observers on board 50% of the trips by Mexican tuna 

 boats (the remaining 50% is covered by lATTC obsei-vers). 

 Observers record data during dolphin sets, using standard- 

 ized data forms and instructions provided by the PNAAPD 

 (which are very similar to those used by the lATTC). A 

 detailed description of data collection and procedures can 

 be found in Perrin et al. (1983) and Polachek.-' In summary, 

 in the "Daily Activity Record" (DAE), obsei-vers keep a daily 

 log on events (date, time of day, departure, arrival, sight- 

 ings, sets, geographic position, aerial assistance during a 

 set, etc.), weather conditions (cloud cover, sea state, vis- 

 ibility, water temperature), and tuna catch. In other forms 

 they also record the vessel's features and all data concern- 

 ing marine mammal sightings and sets, school and log sets, 

 sea turtle sightings, and more recently, bycatch. 



When a marine mammal group is sighted (which might 

 lead to a net set for tuna catch), observers fill out the 

 "Record of Marine Mammal Obsei-vations and Set Data" 

 (RMMOSD). Herd size and composition (percent and spe- 

 cies or stock identification) is annotated as estimated by 

 the observer (usually with 7x50 or 10x50 binoculars), a 

 crew member on board (with 20x or 25x binoculars), and 

 another crew member from the helicopter. At the end of a 

 set, the observer has to decide on his "best estimate" of the 

 herd size and composition before the actual set occurred. 

 During a set, the observer estimates the number of dol- 

 phins (identified by species or stock) that actively evade 

 the boats and net before the chase, during the chase, and 

 during the encirclement, as well as the number of groups 

 within the herd during each of the three set stages. The 

 number of dolphins deliberately cut out by the skipper, 

 those that evade by diving under the net or escape by leap- 

 ing out of the net are also estimated and identified by 

 the observer All these data were of interest for our study 

 (except dolphins deliberately cut out because this action 

 was not considered evasion actively achieved by dolphins). 

 Other data recorded by the observer are chase and set 

 times, number of speedboats used, whether or not explo- 

 sives were used, the number and composition of captured 

 marine mammals (those that were encircled), number of 

 animals rescued and the manner of their rescue, start and 

 end of backdown, number of dolphins hurt or killed, and 

 further details about the fishing procedure. 



The observers' estimates of herd size and number of dol- 

 phins evading capture may be biased owing to differing 

 experience and estimation between observers. However, 

 they attend a complete training course at the institution 

 they work for (lATTC, PNAAPD, or other national pro- 

 grams), where they learn to identify fauna species they may 

 encounter at sea (marine mammals, fi.sh, birds, and sea 

 turtles), how the fishing operation proceeds, and how to col- 



lect data. At the end of each trip, experienced editors at the 

 institutions review the obsei-vers" collection of data. These 

 procedures give credence to the data collected. Moreover, 

 reactions of dolphins to approaching sui-vey vessels have 

 been studied previously by means of field experiments (Au 

 and Perryman, 1982; Hewitt, 1985). Their results might be 

 compared to the dolphins" behavior "before chase" because 

 sighting distance in those studies (as during the fishing 

 operation) was usually between 2 and 6 or 7 nmi (nautical 

 miles), aided with 20x or 25x binoculars. Dolphins started 

 to react (by changing their swimming speed and course) 

 between 1.5 to 3 nmi distance to the ship (Hewitt, 1985) 

 with one exception, where the herd reacted at 6 nmi (Au 

 and Perryman, 1982). Almost half of the herds observed 

 by Hewitt ( 1985) did not react at all. Therefore, the observ- 

 er"s estimate of initial herd size (which is also compared to 

 the estimate by a crew member on board and by the crew 

 member in the helicopter) was assumed to be reliable. We 

 also relied on the obsei'ver"s estimation of number of dol- 

 phins evading capture during all set stages (before chase, 

 during chase, and during encirclement) because of their 

 training and experience. 



The logbooks with these observations are collected by 

 the institution for which the observer works. To improve 

 the accuracy and reliability of the observers" records, 

 careful debriefing and editing is accomplished by skilled 

 technicians (with considerable experience at sea) at the 

 institutions (in our study, the PNAAPD). The observa- 

 tions are then entered into the corresponding comput- 

 erized databases which are also checked for errors that 

 might have occurred during "capture" (entry) of the data 

 into the database (Perrin et al., 1983; Polacheck^). 



Confidence in the databases of the LATTC and national 

 programmes (PNAAPD and the U.S National Marine Fish- 

 eries Sei-vice) is acknowledged with the publication of 

 studies that have analyzed some of these data. A few exam- 

 ples of such studies are the following: dolphin distribution 

 (Perrin et al., 1983; Perrin et al., 1985), dolphin abundance 

 estimation (Anganuzzi et al., 1992; Polachek-'), incidental 

 kill of dolphins in tuna fishing nets (Hall and Lennert, 

 1997; Wade^), dolphin life history (Chivers and De Master, 

 1994), tuna-dolphin association (Edwards, 1992; Scott and 

 Cattanach, 1998), blue whale distribution in the eastern 

 tropical Pacific (Reilly and Thayer, 1990). and aspects of 

 the Mexican tuna fishery (Compean and Dreyfus. 1996). 



In our study, spatial patterns in evasive behavior of the 

 northeastern offshore spotted dolphin were described, as 

 well as differences in evasive behavior and strategies (herd 

 dispersion,"' evasion under and over the net) between 



' Polachek, T. 1984. Documonlation of the lime .sequential 

 file created from the tuna boat observer data base.s for ana- 

 lyzing the relative abundance of dolphins in the eastern tropi- 

 cal Pacific. Southwest Fisheries Science Center Admin. Report 

 LJ-84-33, 26 p. National Marine Kishi^ries Service. 



'" Herd dispersion: to establish if a herd dispersed during a set, 

 the grouping codes applied by the observers in paragraph 3 of 

 the RMMOSD were used. The codes are the following: 1, herd 

 is in one group: 2, herd has divided into two or three groups: 

 3, herd consists of more than three groups. These codes are 

 recorded during three set stages: before chase, during chase, 

 and during encirclement. For each species, the number of sets 

 where the specified grouping code occurred was counted during 

 each set stage. If codes were in ascending order, this was inter- 

 preted as herd dispersion during the fishing operation. 



