Assessing evacuation rates and spawning abundance of marine fishes using coupled telemetric and acoustic surveys

David Robichaud 0 George A. Rose 0 0 D. Robichaud and G. A. Rose: Fisheries Conservation Chair, Marine Institute of Memorial University of Newfoundland , P.O. Box 4920, St John's NF , Canada A1C 5R3. Sidney, BC , Canada , V8L 3Y8; tel: 250-656-0127 ext. 241 Assessing the spawning abundance of marine fishes is difficult if spawning periods exceed the residency of individual fish on the spawning grounds. For Atlantic cod (Gadus morhua), which has a protracted spawning period, we use biotelemetric surveys to estimate the rate at which individual fish vacate the spawning ground and develop a method to adjust multiple acoustic-survey results to account for spawner turnover. Two acoustic surveys conducted one month apart (May and June 1998) on a cod-spawning ground in Placentia Bay, Newfoundland, yielded abundance estimates of 220 000 and 210 000 fish of mean length 63 cm. Rates of evacuation from the spawning ground, observed over two separate spawning seasons, were modelled as logistic decay functions with good fit (r2=0.96 in 1998; r2=0.88 in 2000). Our method estimated that only 8.8% of the fish counted during the second survey were present during the first, and that between 400 976 and 420 842 fish were actually present over the full spawning season. Coupled telemetric and acoustic surveys could be used to estimate spawning abundance in many marine fishes. 1054-3139/02/040254+07 $35.00/0 - Marine fish populations are often surveyed during spawning periods when distributions are concentrated and species mixing is minimal (God, 1989; Coombs and Cordue, 1995; Kloser et al., 1996; Williamson and Traynor, 1996; Lawson and Rose, 2000a). Such surveys are sometimes repeated during a season in attempts to estimate the error around abundance estimates. However, the potential for movement of fish into and out of the spawning area calls into question the likelihood that multiple surveys are true replicates. Surveys conducted as discrete snapshots of the spawning abundance may or may not be measuring the same fish. In the extreme case, if the time of residence of individual fish on the spawning ground is short relative to the interval between surveys, abundance estimates from each survey should be summed to estimate total abundance. If the residence time is protracted, the estimates should be averaged. Intermediate cases are likely and, typically, the average residence times of individual fish are unknown. Atlantic cod (Gadus morhua) spawn in large aggregations and are surveyed acoustically in several areas of the North Atlantic (God, 1989; Ouellet et al., 1997; Rose et al., 2000a; Anderson and Rose, 2001). Spawning is often protracted over many weeks (Brander, 1993; Hutchings and Myers, 1994; Lawson and Rose, 2000a). The residence time of cod on their spawning grounds is thought to vary among gender and age groups (Marteinsd ottir and Petursd ottir, 1995; Kjesbu et al., 1996; Thorsteinsson and Marteinsd ottir, 1998). However, the impact of variation in mean residence time of cod on the spawning ground on survey estimates of abundance is unknown. Here, we use biotelemetric techniques to estimate the rate at which individual cod evacuate a spawning ground in Placentia Bay, Newfoundland. Empirical evacuation curves are applied to a set of acoustic surveys to demonstrate how multiple survey data can be adjusted to account for spawner turnover and to more accurately estimate total spawner abundance. Materials and Methods In April 1998 aggregations of spawning cod were located acoustically from the CCGS Shamook (25 m research trawler) near Bar Haven Island in the inner part of Placentia Bay, Newfoundland. This area is used consistently by cod for spawning during spring (Lawson and Rose, 2000a, b). On 19 and 20 April 1998 approximately 225 cod (total length 39102 cm) were taken from these aggregations using feather hooks. The fish were in water of near 0 C temperature, at depths between 3050 m. Larger individuals (>60 cm) observed to be in spawning condition were held in flow-through tanks and sexed by cannulation. For each fish an individually coded ultrasonic transmitter tag was surgically implanted into the peritoneal cavity and an external spaghetti tag was anchored on the left side, near the first dorsal fin. Tagged fish were held for up to 10 hours, and those that appeared to be robust and in excellent condition were released at the location where they were caught. In total 48 cod [27 females (lengths 6487 cm) and 21 males (lengths 6788 cm)] were released (Table 1). To avoid observing the period of abnormal behaviour which may follow surgery (God and Michalsen, 2000), we waited 15 days after tagging before beginning a biotelemetric survey of the Bar Haven spawning ground. Fish were monitored between 5 May and 24 June using a decoding acoustic receiver and omnidirectional hydrophone. The spawning area was surveyed using a grid of monitoring stations spaced 0.5 n mil (0.96 km) apart to correspond with the average effective range of the telemetry equipment as tested on a control tag left on the bottom throughout the study. Surveys of the spawning ground were repeated during the two subsequent spawning seasons (8 April26 May 1999 and 4 April26 June 2000). The distance between monitoring stations was reduced to 0.3 n mil (0.58 km) in 2000 to account for the observed decay in transmit power of the control tag. The surveys in all years were terminated when no tagged cod could be located for a period of several days. Coverage of the survey grid at Bar Haven was not synoptic and all stations could not be surveyed in a single day. This condition, and the probability that fish were moving about the ground during a survey, indicated that not all fish on the ground could be expected to be relocated on each survey. It was therefore assumed that an individual fish remained within the survey area from the first time it was relocated until the last, regardless of whether or not it was relocated in the days between. The period between first and last relocation is referred to as residency time. Fish on the spawning grounds are referred to as residents. In total, Bar Haven was surveyed 11 times in 1998, three times in 1999, and 14 times in 2000. There were too few surveys in 1999 to estimate residence and these data are not included in this study. Evacuation rates were calculated for 1998 and 2000 by plotting the number of resident fish during each survey of Bar Haven against the survey midpoint. Proportions of the total number of residents observed each year are used to allow comparison between years. The data were fitted with sigmoid (dose response) curves by estimating two parameters, the slope ( 0) and the day of the inflection point, 1, as: where Y is the proportion of residents observed in a given survey and X is the survey date (or midpoint, when surveys took multiple days). A sigmoid decay curve was chosen as the functional form because the proportion of fish could not drop below zer (...truncated)