Population Status Assessments: Methodology
The IWC Scientific Committee’s Methods to Assess the Status of Stocks
This page provides background on the methods used by the IWC's Scientific Committee to conduct the assessments of whale populations in this table. Brief introductions to the computer modelling and statistical estimation approaches are provided, but many methods are case-specific so further details should be sought on the assessment pages for the stock concerned. Some population assessments rely only on a review of published science, rather than computer modelling. This page also explains the rating systems for the Relative Abundance and 20-Year Change measures used in this table.
Background on assessments
Assessments of cetacean populations are conducted by fitting population dynamics models to monitoring data. The structure of the population dynamics model and the types of data used for parameter estimation differ among species and regions. Assessments of many species are based on population dynamics models that track the population by age and sex, and may also take account of the spatial structure of the population. Other assessments are based on population models that aggregate population numbers over age and sex, and assume that there is a single population only in the region being assessed. The monitoring data available differ among species but the most common data sources are removals (e.g., commercial catches, bycatch and aboriginal takes), estimates of absolute abundance from surveys, and data on movement from tagging or mark-recapture studies. The parameters of a cetacean population dynamics model include: population size in the first year of the modelled period (or, sometimes, the population size at carrying capacity or prior to commercial exploitation); those that relate to productivity and mortality; and, where appropriate, those that represent movement and dispersal rates. Some assessments will estimate all of these parameters while others will pre-specify some of the values for these parameters based on auxiliary information.
The result of fitting such a model is a set of estimates and predictions. Assessments can provide estimates of many quantities, including: the numbers of animals aged 1 and older (“1+ abundance”), mature females and/or calves in various years; the annual proportion of the population dying due to anthropogenetic causes; movement rates of whales between regions; and estimates of the model parameters such as productivity or carrying capacity. However, for consistency among the various types of models used for assessment (and for consistency with the estimates provided by abundance surveys) results are most commonly expressed in terms of 1+ abundance. The Scientific Committee’s assessment of population status is based on a time series of estimated 1+ abundances.
Assessments, particularly those on which evaluation of management approaches for commercial or aboriginal whaling are based, often involve many alternative models. These models often consider alternative assumptions about the number of stocks in the region, how they mix and animals disperse among stocks, and historical removals (e.g., due to uncertainty in levels of bycatch). The relative plausibility of all alternative models will not be the same and the Scientific Committee usually divides the plausible models into (i) more likely (or base-case) models and (ii) other models used to examine sensitivity to the assumptions of the base-case models. Models of the second type are often much less plausible than the base-case models. Implausible models are rejected and their results are not reported.
Representing uncertainty is core to any assessment. The types of uncertainties commonly captured in assessments are model and estimation uncertainty. Model uncertainty relates to uncertainty about the underlying structure of the system being modelled, such as how many stocks are in the region in which the species is found and how they move and/or mix. Estimation uncertainty is quantified by the statistical analysis used to fit the model to available data.
Assessments are used by the IWC to understand historical and future trends of populations given past removals and possible strategies for setting future limits of removals. For many stocks assessments were therefore not primarily designed to estimate status but rather to determine whether a particular removal strategy would be sustainable and robust to uncertainty. Nevertheless, the results of an assessment can be used to summarize status by focusing on the base-case (or most likely) models. To generate these status assessments, the estimates from one or more base-case models are combined as described later in this document.
The detailed assessment results for each stock are accompanied with an assessment of data availability and quality. The rating system and list of factors to be taken into account when classifying assessments are shown below. Not every criterion must be met to qualify for a category.
- Good: At least one abundance estimate endorsed by the Scientific Committee within the last 10 years. Usually, several abundance estimates available for different years. The surveys covered, or can be reliably extrapolated to, most of the presumed stock range. Uncertainty and bias of these estimates is understood and not too large (corresponding to Category 1 or 2 using the system applied in the Scientific Committee’s Consolidated Table of Agreed Abundance Estimates. Stock structure is reasonably well understood or subject to a small number of competing hypotheses whose impact on assessment is not critical. The major sources of human induced mortality are understood and reasonably well quantified.
- Fair: Abundance estimates are available, but they do not meet the standards for Good. For example, survey coverage may be limited or estimation uncertainty may be high. Nevertheless, the information available is adequate to provide some general indication of abundance. Abundance estimates may correspond to Category 3 using the system applied in the Scientific Committee’s Consolidated Table of Agreed Abundance Estimates. Stock structure is not well understood, and/or competing hypotheses have important impacts on assessment results. Scientists have a general sense of the magnitude and sources of human induced mortality.
- Poor: Abundance estimates, if any, originate from surveys covering only a small portion of the range of the stock, are highly imprecise, or may have large and/or unknown bias. Abundance estimates have often not been reviewed or endorsed by the Scientific Committee; any endorsed estimates correspond to Category 3, or worse. Little information may be available about stock structure or the magnitude and sources of human induced mortality.
Summarizing status
For a model-based status assessment, the Assessment Date provided on the table here corresponds to the final year of the model fit. When a model is not used, the assessment date is chosen on a case-specific basis to indicate the best, most recent information fully considered.
The results of fitting an assessment model include the following three quantities:
- Current 1+ abundance: the estimate of 1+ abundance in the final year of the assessment
- Current 1+ relative abundance: the estimate of 1+ abundance in the final year of the assessment expressed as a proportion of the pre-exploitation number of 1+ animals, or as a proportion of current carrying capacity, depending on the assessment.
- Time-trajectory of 1+ abundances: the time-series of the numbers of 1+ animals from the first year of the assessment to the final year. This can provide information on the rate of population increase over the most recent 30 years, for example.
A key uncertainty in cetacean assessments is the extent of productivity, quantified using the MSY rate parameter (i.e., MSYR). Here, MSYR is defined as the ratio of maximum sustainable yield (MSY) to the population abundance when exploitation rate equals that corresponding to MSY. Thus, unless the assessment estimates MSYR, the three quantities above are reported for (at least) two levels of productivity: the lowest plausible value in the base-case models and MSYRmat=4%. The default value for the lowest plausible MSYR is MSYR1+=1%. MSYR1+ is the MSYR in the theoretical situation when harvesting is on animals aged 1+ without any selection, while MSYRmat is MSYR in the theoretical situation when harvest occurs on mature animals only.
The IWC Scientific Committee will select a set of model configurations with different MSYR values, from those included in the ‘base-case”, and those configurations will be fit using either Bayesian or maximum likelihood methods. Uncertainty estimation usually uses either Bayesian or bootstrap methods. The particular models, MSYR values, and other details differ in each application; the specifics can be found on the Status Details pages. A Status Summary page and a Status Details page has been created for each of the species assessed here. The Status Details are accessible from each of the Status Summary pages.
The results for each model configuration are summarized by a set of (usually 100) parameter vectors (and hence corresponding sets of values for current 1+ abundance, current 1+ depletion, and the time-series of the number of 1+ animals). Assessments provide estimates for each stock within the region, which makes summarizing results difficult because the number of stocks may differ among model configurations. Consequently, status results are provided for the Medium Areas (the IWC SC defines a Medium Area as an area that corresponds to the known or suspected range of a distinct biological stock (IWC, 1999) or the entire Ocean Basin(or “Region”) if there are no Medium Areas. In some cases, results may be presented for individual stocks if the Scientific Committee believes that the Commission needs to be informed about the specific stock (e.g., if it is subject to aboriginal subsistence hunting) because reporting results by area might be misleading. Similarly, results for particular sub-groups of interest may be computed (and reported) separately if these are of particular management interest to the Commission. For the purposes of computing the summary statistics, when a stock is found in multiple Medium Areas the estimates of abundance are pro-rated among Medium Areas.
Summary statistics are computed using the distributions of values from the selected base-case models. Medians and 90% intervals are computed. Details are given below. Insert explanation of how trajectory plots and bands are computed.
Model estimates of 1+ abundance over time can be displayed as trajectory plots, with year on the horizontal axis and numbers of whales on the vertical axis. The trajectory is comprised of pointwise medians, and the confidence bands are pointwise 90% intervals.
Relative Abundance and 20-Year Change Statistics
For each stock assessed, two statistics are displayed on the table: Relative Abundance and 20-Year Change. The Relative Abundance statistic is often termed 1+ depletion (1+ abundance in the current year divided by either 1+ abundance prior to exploitation or current carrying capacity, depending on the assessment). The 20-Year Change statistic is the ratio of current 1+ abundance to the 1+ abundance 20 years prior, expressed as a percent increase. Note that this is a cumulative total change over 20 years, not an annual rate. A 20-Year Change of -25% means that the stock size is 75% of what it was 20 years prior.
As described above, a set of values for Relative Abundance and 20-Year Change are generated for each base-case model used in the assessment, using the statistical estimation and uncertainty approach applied in that analysis. These values are then pooled, and the central 90% of values are taken to represent the range of uncertainty for each statistic. The best estimate is defined to be the average of the medians from the separate sets of base-case values, not the median of the pooled values.
Semicircular thermometer graphs on the table are used to present the results. For Relative Abundance, the left edge of the plot corresponds to zero (extirpated), and the right edge corresponds to 1.0 (1+ abundance equal to pre-exploitation level or current carrying capacity, depending on the assessment). The grey wedge spans the 90% uncertainty range. The estimated value of Relative Abundance is printed in the centre of the thermometer and indicated with a black needle. Above the graph is a label to characterize the estimated value, using the following scale: 0.01-0.19=[very low], 0.20-0.39=[low], 0.40-0.59=[moderate], 0.60-0.79=[good], 0.80+=[very good]. The colour transitions in the graph are not intended to symbolise anything.
For 20-Year Change, a similar display is used. The range of the thermometer is -50% to +50%. When the estimate or an uncertainty bound falls outside this range, this is indicated with an asterisked arrow and central note. As above, the estimated 20-Year Change value is printed in the centre of the thermometer. The labels used to characterise the estimated 20-Year Change are as follows: ≤-20%=strongly decreased, >-20% to ≤-10%=decreased, >-10% to <10%=stable, ≥10% to <20%=increased, ≥20%=strongly increased. It should be noted that a low 20-Year Change rate is not necessarily concerning, because a population near its carrying capacity (and hence having very good status) would be expected to remain nearly stable (near zero growth).
References
IWC. 1999. Report of the Scientific Committee, Annex N: The Revised Management Procedure (RMP) for Baleen Whales. J. Cetacean Res. Manage 1 (Suppl.): 251-258.
