Population structure and dynamics of humpback whales (Megaptera novaeangliae) in the Gulf of St Lawrence, Canada.

Christian Ramp 1,2, Martine Bérubé 3, Wilhelm Hagen 2 and Richard Sears 1

1 Mingan Island Cetacean Study, 285 rue Green, St. Lambert, Qc, J4P 1T3, Canada2 Marine Zoologie (FB2), Universität Bremen (NW2), Postfach 330 440, 28334 Bremen, Germany3 Ecosystem Science Division, Department of Environmental Science, Policy & Management, University of California at Berkeley, 151 Hilgard Hall #3110, Berkeley, CA 94720-3110, USA

Updated Abstract:Two humpback whale aggregations can be found in the Gulf of St. Lawrence in eastern Canada. The first is located mainly in the Mingan/Anticosti area (MA) and forms the St Lawrence stock while the second one is situated in the Northeast Gulf (NEGU) and is part of the Newfoundland/Labrador stock. In 26 years of research 223 animals were photo-identified in the Mingan area and 501 in the Northeast Gulf. Fifty-four animals were observed in both areas. The sex ratio was 1.4 females to 1 male. Altogether 70 females were seen with 129 calves. Mean calving interval for females with consecutive sightings was 3.5 years. Very few males observed as calves were seen a second time, while half of the females returned, suggesting a higher site fidelity for females. Only two females seen as calves returned with their own offspring, after 9 and 11 years respectively. The high number of individuals seen only once (both calves and adults) created an age structure in the mark recapture analysis. Different models (CJS, Barker’s, Pradel’s and Multi State) yielded estimates of adult survival rate ranging from 0.974 to 0.986. No significant difference in survival was observed between the two stocks or between sexes. The dataset was analyzed in increasing time spans, backwards and forwards. Results show that longer data sets provide more precise survival estimates at first, but very long datasets could potentially hide short-term trends or between-year differences.

0,80

0,82

0,84

0,86

0,88

0,90

0,92

0,94

0,96

0,98

1,00

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Years of data

surv

ival

rat

e

phi backwards

phi forwards

Methods• Photo ID and biopsy techniques were used to determine identity, sex and stock information of individual humpbacks

• Main study areas: MA (1980-2004) & NEGU (1982-2004, irregularly), other regions were sampled occasionally

• Data from both stocks were combined in multi-strata and Barker models (using primary and secondary sampling occasions), and separated in Cormack-Jolly-Seber (CJS) model (Program MARK, White & Burnham 1999)

• MA data were analyzed (CJS) in incrementally increasing time spans (5 to17 years) forward (88-92, 88-93, …, 88-04) and ‘backwards’ (00-04, 99-04, 98-04, …, 88-04)

• Models were GOF tested, adjusted for over-dispersion and results were taken from the most supported model (QAIC) or sets of models (model averaging)

Results• Pooling all inner Gulf regions caused heterogeneity and was rejected; sightings from Gaspe, Sept-Iles and Estuary were used as secondary sightings (Barker’s)

Table 1: Estimated survival rates of adult humpback whales

Figure 2. Estimates of the survival rate and 95% CI for the Mingan data set (1988-2004), broken down into 5-17 year time spans

Acknowledgements:We would like to thank all the MICS team members for their continuous effort over 26 years. GREMM, Croisieres de la Baie de Gaspe, Jacques Gelineau, Martin Champagne and Rene Roy for contribution to the catalogue. CR wants to thank Brian Kot, Thomas Doniol-Valcroze, Jean Marie Jones, Trish Nash (QLF), Robin Kelleher (Relais Nordik) for their help during the two foggy field seasons off Blanc Sablon.

data setmodel type

φ/S (adult) SE

CI 95% lower

CI 95% upper

all datamulti strata 0,9806 0,0069 0,9612 0,9904

NEGU CJS 0,9826 0,0147 0,9121 0,9960

NEGU/MING Barker 0,9741 0,0088 0,9496 0,9868

Mingan 88-04 CJS 0,9817 0,0077 0,9585 0,9920

Mingan 88-04 with NEGU Barker 0.9785 0,0080 0,9557 0,9897

Mingan 88-04 with Gulf Barker 0,9792 0,0079 0,9564 0,9902

Figure 1: Study areas and photo ID data

Caswell H, Fujiwara M, Brault S (1999) Declining survival probability threatens the North Atlantic right whale. Pro. Natl Acad Sci USA 96:3308-3313 Pradel R, Hines JE, Lebreton JD, Nichols JD (1997) Capture-Recapture Survival Models Taking Account of Transients. Biometrics 53:60-72White GC, Burnham, KP (1999) Program MARK Survival estimation from populations of marked animals. Bird Study 46 (suppl):120-139

• Analyzing more years of data results at first in higher and more realistic values, partly due to more ‘age classes’

• Estimates vary less using 10 (+) years of data, and in general the more data used the more robust are the estimates (Figure 2)

• No trend or time variation was found in the survival rate of adult animals in any data set

• Increasing effort in MA caused heterogeneity in the capture probabilities data set 1988-2004 was used.

• High number of calves and transient whales caused heterogeneity in survival among individuals and were modeled after Pradel et al. 1997, usually adult constant survival was obtained from the 4th interval on

• No differences between the 2 stocks or between sexes

• Multi strata model: more movements MA NEGU than vice versa, especially for 1st encountered animals

• All data sets yielded estimates of adult survival rate around 0.98 (Table 1)

• Barker models yielded smaller estimates; site fidelity was modelled as for the survival rate (transients)

• MA estimates were more robust than NEGU results due to more sampling occasion (years)

Contact: christianramp@web.de

Objectives• To model the adult survival rate for two stocks of humpback whales in the Gulf St. Lawrence (Fig. 1)

• To consider benefits vs. risks of obtaining & monitoring survival rates using long-term data sets

Long-term data sets – benefits vs. risks• Survival rates are used to conduct stock health assessments• Monitoring over time allows detection of changes (e.g. decline) and rapid response in management plans to counteract potential threats to a stock (e.g. Caswell et al. 1999) We need a robust estimate & the ability to detect ‘real’ changes, not caused by e.g. the probability of capture

The following questions arise: • What is a ‘good’ initial estimate and how many years of data do we need to obtain it? • Can we detect small but significant changes, or do they average out in ‘too many’ years of data? • Can we detect these changes promptly or only after further sampling?

Next steps…• Comparison with other long-term data sets (marine mammals and other species with larger annual samples)

• Simulating long-term data sets with different scenarios to determine the likelihood of detecting changes in survivorship