Search

Hopefully, this approach represents a process that leads to open and constructive debate, to progress towards a better scientific understanding and to improvements in our ability to manage wildlife. Lonergan has examined our modeling in detail, and his critique raises a series of specific points and also some general questions about how data and expert opinion should be integrated into statistical and modeling frameworks with the aim of advising on conservation methods. we agree with several of Lonergan’s dictums, such as the need for open dialogue between the field biologists collecting data and the modelers subsequently using these data. Indeed, several of his specific points could have been resolved through such dialogue. For example, he queries the lack of data on harbor seal pups after mid-July, but this situation arises because pups cannot be reliably distinguished from juveniles after this time (thompson and Rothery 1987). similarly, detailed evaluation of previous studies from the moray Firth (e.g. thompson et al. 2007) would also call into question Lonergan’s assertion that shooting is the sole driver of historical declines in this population. Lonergan highlights his concerns about the performance and sensitivity of our model to specific assumptions and priors. these are important questions. we hope that in our publication, prior justification was explicit (pp. 155–158), prior sensitivity methodology was described in detail (supplement 3) and the implications of our modeling on our predictions were discussed in full (our points 1–5, pp. 158–159; points 1, 2, p 160). work continues, and our aim is to improve and simplify our modeling—for example, by considering seasonal variation in haulout probability. we are also exploring a question about the mis-match between one datum (2009 population estimate) and model predictions. the sensitivity of our results to these two issues was In his critique of our recent paper (matthiopoulos et al. 2014) on the population dynamics of harbor seals, Lonergan (2014a) argues that although our state–space modeling approach represents an interesting academic exercise, it nevertheless risks misleading those attempting to understand and manage observed declines of this species in UK waters. we argue that approaches like ours need not be misleading as long as information is provided on the assumptions and simplifications used. when such information is available, all assumptions (such as the shape and parameters of prior distributions in a Bayesian analysis) can then be subjected to scrutiny [such as that performed by Lonergan (2014a)], inevitably leading to future improvements.

Declines in large vertebrate populations are widespread but difficult to detect from monitoring data and hard to understand due to a multiplicity of plausible biological explanations. In parts of Scotland, harbour seals (Phoca vitulina) have been in decline for 10 years. To evaluate the contributions of different proximate causes (survival, fecundity, observation artefacts) to this decline, we collated behavioural, demographic and population data from one intensively studied population in part of the Moray Firth (north-east Scotland). To these, we fit a state-space model comprising age-structured dynamics and a detailed account of observation errors. After accounting for culling (estimated by our model as 14 % of total mortality), the main driver of the historical population decline was a decreasing trend in survival of young individuals combined with (previously unrecognised) low levels of pupping success. In more recent years, the model provides evidence for considerable increases in breeding success and consistently high levels of adult survival. However, breeding success remains the most volatile demographic component of the population. Forecasts from the model indicate a slow population recovery, providing cautious support for recent management measures. Such investigations of the proximate causes of population change (survival, fecundity and observation errors) provide valuable short-term support for the management of population declines, helping to focus future data collection on those ultimate causal mechanisms that are not excluded by the demographic evidence. The contribution of specific ultimate drivers (e.g. shooting mortality or competitors) can also be quantified by including them as covariates to survival or fecundity.