Scottish Marine and Freshwater Science Volume 6 Number 10: At-Sea Turnover of Breeding Seabirds - Final Report to Marine Scotland Science

The aim of this project was to review the potential issue of "turnover‟ of individual seabirds at sea during the breeding season and to assess how this may lead abundance estimates derived from boat or aerial surveys to underestimate the total number of b

Appendix B Alternative Model-Based Approach to Simulation of Individual Time Budgets

B1. Methodology

The main method that we use for estimating the proportion of time on each day that is spent undertaking each of the four activities (flying, foraging, resting at sea, time at the colony) is based on empirical activity data (we call this the "empirical approach").

An alternative approach is based on using the daily activity budgets that are simulated by the foraging model of Searle et al. (2014). This model assumes that individuals act in concordance with optimal foraging theory, and implies that the foraging behaviour of individual seabirds was driven by prey availability, travel costs, provisioning requirements for offspring, and behaviour of con-specifics. We called this methodology for determining time activity budgets the "model-based approach", because the foraging model directly simulates the proportion of time that each bird spends undertaking each of the four main activities on each day. This is done separately for each bird on each day. The key potential advantage of this approach is that the modelled time budget for individual birds, on individual days are linked directly to the foraging locations that they are assumed to visit. This "matching" should ensure that time budgets for individuals are directly linked to their choice of foraging location - something that we are unable to do with the empirical approach, because the time budget data and foraging location data relate to different sets of birds.

In order to assess the viability of this approach we compare the overall (mean) time budgets that are obtained from the foraging model (run without site fidelity, since this was the version of the model that was originally sense-checked in Searle et al., 2014, and hence is the version that we may expect to be most closely matched to observational data) against those that are obtained from empirical data. The model-based data refer to the entire 24 hour period, whereas the empirical data can relate either to the full 24 hour period or to the more restricted period (05:00 - 20:00 hours) that is typically used for surveying.

B2 Results

Table B1

Daily time budgets for each species, based on empirical data and on the output from the optimal foraging model of Searle et al. (2014).

Empirical for 24 hour period Empirical for 05:00-20:00 study period Model-based For 24 hour period
Kittiwake Foraging 12.3 15.8 62.1
Resting 19.3 16.4 4.2
Commuting 17.3 19.7 7.5
Razorbill Foraging 13.5 14.6 54.9
Resting 27.1 19.7 4.2
Commuting 7.3 10.0 3.2
Guillemot Foraging 22.1 21.7 44.6
Resting 29.0 22.6 4.2
Commuting 3.3 3.5 3.8
Puffin Foraging 22.4 25.0 34.4
Resting 67.7 61.4 4.2
Commuting 7.8 10.3 10.3

Comparisons of model-based and empirical mean activity budgets are shown in Table B1. The results of comparing the empirical and model-based results for the full 24 hour period suggest that, for all four species, the model-based results tend to substantially over-estimate the amount of time spent foraging and underestimate the amount of time spent resting. For kittiwake and razorbill the model-based results also substantially underestimate the amount of time spent commuting, but for puffin and guillemot this is somewhat overestimated.

The differences between the empirical results for the 05:00 - 20:00 hours survey window and those for the full 24 hour period are generally much smaller than the differences between the model-based and empirical results, suggesting that the poor performance of the model-based approach is not primarily driven by the fact that it fails to capture diurnal variations in activity.

B3. Implications

The results of this comparison suggest that the model-based approach provides a poor estimate of typical time activity budgets. We, therefore, did not pursue the model-based approach further within this study.

It should be noted that the simulation model developed by Searle et al. (2014) was not validated against empirical time budgets for each species and was not intended to provide good estimates for these budgets, so the poor performance of this model in constructing empirical activity budgets is not necessarily surprising. In particular, the finding that simulation model of Searle et al. (2014) tends to underestimate the time birds spend resting at sea arises directly from the fact that the simulation model was parameterised to favour attendance at nests over time spent resting at sea. This characteristic was deliberately introduced to the simulation model to capture observed patterns of attendance at nests, and to ensure that simulated output matched observed patterns in chick production and chick mass gain over the breeding season (directly related to adult attendance at nests). Importantly, the greater confidence in the 'empirical' method for estimating turnover should not be taken as a reflection on the validity of the estimates for the effects of displacement from the simulation model presented in Searle et al. (2014); rather, it is simply an artefact of the parameterisation of the simulation model to match empirical patterns in adult mass change, adult survival, chick mass change, and chick production, the key variables of interest in estimating population consequences of displacement from wind farms.


Back to top