of 45

Report. JIP Cetacean Stock Assessment RA1007OGP. Nicola Quick

0 views45 pages

Download

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Report Project Name: Reference: Project Manager: JIP Cetacean Stock Assessment RA1007OGP Nicola Quick Draft report drafted by: Draft report checked by: Draft report approved by: Nicola Quick, Kristin Kaschner,
Report Project Name: Reference: Project Manager: JIP Cetacean Stock Assessment RA1007OGP Nicola Quick Draft report drafted by: Draft report checked by: Draft report approved by: Nicola Quick, Kristin Kaschner, Rodrigo Wiff & Len Thomas Beth Mackey Gordon Hastie Date of draft report: 30 th July 2008 Reviewer comments incorporated by Final report checked by: Final report approved by: N/A Nicola Quick Beth Mackey Date of final report: 26 th June 2009 VAT reg. No. GB SMRU LIMITED is a limited company registered in Scotland, Registered Number: Registered Office: 5 Atholl Crescent, Edinburgh EH3 8EJ Contents Summary Introduction Methodology Analysis steps Data Exploration Taxonomic covariates Spatial covariates Temporal covariates Survey related covariates Determination of Weights Calculating area and precision weightings Combining weightings Global dataset models Individual area and species models Power Analysis Results Data Exploration Taxonomic covariates Spatial covariates Temporal covariates Survey related covariates Determination of Weights Global dataset models Individual Area and Species Models Frequency distributions of important covariates Individual Species Results Harbour porpoise Dwarf Minke whale Fin whale Humpback whale Sperm whale Striped dolphin & Long-finned pilot whale Power Analysis Discussion Further work and considerations References Task 2 Deliverable: Cetacean stock assessment in relation to Exploration and Production Industry sound Detecting trends in global cetacean stocks within the Areas of Relevance Summary Task 2 delivers a report outlining which covariates may be important when assessing changes in density over time for 34 cetacean species combined globally, and for seven individual species within specific areas. Results from the global model showed that the taxonomic attributes of species and family were by far the most important covariates. For the individual models, temporal and spatial covariates generally explained the highest proportion of the total deviance across both species and area. Survey-related covariates, such as method and agency, explained little of the deviance. Power analysis showed that small changes (few tens of percent) may be detectable in some rare cases, but in most of the studies analysed, the population change would need to be at least double, before it could be detected with high power. In some cases, even an order of magnitude of population change could easily be missed. The species for which population trends can be analysed will be carried forward to Task 3 when factors influencing cetacean stocks will be investigated. 1. Introduction In order to assess any potential relationships between Exploration and Production Industry (E&P) sound and trends in cetacean stocks within JIP areas of relevance, it is necessary to first determine the possibility of detecting trends in cetacean population using published estimates. Detecting trends in populations or stocks of marine mammals is a primary focus of much research. The ability to detect how numbers of animals may change over time can help inform many management decisions including those relating to conservation, and mitigation of anthropogenic impacts. Research into population numbers has been carried out on cetacean species across the globe; but determining changes in population trends is not straight forward. Many cetacean species are wideranging and not easily observed at sea. Additionally, many undertake migrations that lead to seasonal differences in distribution. These differences in distribution may be at a population level (e.g. Humpback whales migrating between feeding and breeding areas; see Clapham, 2000 for review) or may only relate to individuals of a certain sex within a population (e.g. sperm whales, where females remain in low latitudes and males migrate to feed in higher latitudes, Whitehead & Weilgart, 2000). To deal with these difficulties, researchers have developed different methods for monitoring populations and analysing data. These methods include line-transect surveys for all species present in an area (e.g. SCANS surveys; SCANS I: and SCANS II: SCANS-II 2008); individual species accounts of stock structures and movements ( photo-identification surveys of species that assess numbers within restricted geographic boundaries (e.g. Parra et al., 2006 and Smith et al., 1999) and counts of animals passing geographic points (Buckland & Breiwick, 2002 and Zeh et al., 1991). In addition, new methods are also being developed that try to estimate abundance of highly vocal species using acoustic monitoring techniques (Barlow & Taylor, 2005). Although all these methods can provide a good means of monitoring populations, a further problem of how robust and comparable the data are across years to detect changes needs to be addressed. 2 Variation across population estimates may be caused by taxonomic, spatial, temporal and methodological differences of varying scales. Taxonomic differences will be evident when comparing estimates across species. Spatial differences will occur if survey areas cover completely different parts of a species range, are across a range of suitable habitat, or if survey areas vary between different years. Temporal differences will occur if surveys of a population are conducted in different seasons or years. Methodological differences may be evident if several survey or analysis techniques are used. Differences between methodologies such as mark-recapture versus line-transect, where the former may give higher abundance estimates if sampling focuses only on known hotspots, and density from these hotspots is then extrapolated to a wider area. Or differences in how animals may response in terms of movement away from a ship, compared to an aerial surveying vessel. Additionally, some surveys may include correction for animals missed on the track line in line transect surveys (g(0) estimation), and differences in the application of the same methodologies will be present between different agencies. The aim of Task 2 was to use available published data that were collected and reviewed during Task 1, to indicate robustness of trend estimates. The data collected during Task 1 used The Environmental Risk Management Capability Project (ERMC) database as a foundation. This database contains more than 1800 regional abundance estimates from over 350 surveys and covered a total of 70 marine mammal species. The data compiled in the database are estimated to equate to roughly 90% or more of all surveys conducted globally over the past 30 years. Within this database comprehensive details about relevant associated information for each survey are held. These survey details include; the geographic location, the time period and duration, the size of the survey area, survey methodology and any potential biases. By investigating the importance of these factors as covariates of the abundance estimates, it may be possible to assess whether or not the currently available data on species abundance over time are of sufficient quality to reliably detect trends in cetacean populations, at both a global scale and at an individual species or population level. As such the Task was split into two parts: a global analysis and, a set of species-specific analyses. Initially, a total of 34 cetacean species were selected from the database, which were (a) covered by ERMC, thus ensuring a very comprehensive coverage of all available information and/or (b) associated with nine abundance estimates in the database (Table 1). Selecting all available data on a global scale for these species resulted in a subset of 1035 abundance estimates that were used in the subsequent analysis. Using these data, the principal aim of the global analysis was to determine overall which covariates had the most influence on the density estimates, both singly and in combination. The secondary aim of the global analysis was to determine how precisely temporal trends could be estimated after all other important covariates were included. The species-specific analyses focussed on individual populations of specific data-rich species within some of the areas of relevance (identified in Task 1) for which multiple estimates exist 1. The final step was to produce a power analysis to make a preliminary determination of what level of population trend would be observable with reasonable certainty, given the levels of variability about the trend estimates observed. 1 Areas and species to take forward in the Task 2 analyses were agreed via on 1 st June 2008 by Russell Tait. The decision was based on the areas and species for which most data existed, as outlined in the Task 1 review. Four areas were outlined as possible for further analysis (Area 4 (Alaska); Area 5A (West coast of US); Area 5B (East coast of US); Area 6A (Europe). Within each of these areas, a few candidate species were explored for potential analysis. The final number of species and areas could not be determined until the end of the task 2 analysis. It should be noted that not all species within these areas will be considered due to lack of data and, as such, not all of the four areas may be taken forward to the end of the project. 3 Table 1. All species names, associated codes and abundance estimates used in the global analysis. Common name Scientific name SpecID Family Species Group Number of Abundance Records Sei whale Balaenoptera borealis Babor Balaenopteridae Baleen_whales 28 Blue whale Balaenoptera musculus Bamus Balaenopteridae Baleen_whales 16 Fin whale Balaenoptera physalus Baphy Balaenopteridae Baleen_whales 95 Minke whale 2 Balaenoptera acutorostrata Balaenoptera Bonaerensis Baacu Balaenopteridae Baleen_whales 113 Humpback whale Megaptera novaeangliae Menov Balaenopteridae Baleen_whales 93 Northern right whale dolphin Lissodelphis borealis Libor Delphinidae Delphinidae 20 Short-beaked common Delphinus delphis Dedel Delphinidae Delphinidae 32 dolphin Pygmy killer whale Feresa attenuata Featt Delphinidae Delphinidae 6 Short-finned pilot whale Globicephala macrorhynchus Glmac Delphinidae Delphinidae 10 Long-finned pilot Globicephala melas Glmel Delphinidae Delphinidae 34 whale Risso's dolphin Grampus griseus Grgri Delphinidae Delphinidae 36 Atlantic white-sided Lagenorhynchus acutus Laacu Delphinidae Delphinidae 21 dolphin Pacific white-sided Lagenorhynchus obliquidens Laobl Delphinidae Delphinidae 34 dolphin False killer whale Pseudorca crassidens Pscra Delphinidae Delphinidae 11 Killer whale Orcinus orca Ororc Delphinidae Delphinidae 34 Common bottlenose Tursiops truncatus Tutru Delphinidae Delphinidae 68 dolphin Spinner dolphin Stenella longirostris Stlon Delphinidae Delphinidae 18 Melon-headed whale Peponocephala electra Peele Delphinidae Delphinidae 8 Atlantic spotted Stenella frontalis Stfro Delphinidae Delphinidae 17 dolphin Striped dolphin Stenella coeruleoalba Stcoe Delphinidae Delphinidae 36 Rough-toothed Steno bredanensis Stbre Delphinidae Delphinidae 10 dolphin Pantropical spotted dolphin Stenella attenuata Statt Delphinidae Delphinidae 27 White-beaked dolphin Lagenorhynchus albirostris Laalb Delphinidae Delphinidae 10 Beluga or white whale Delphinapterus leucas Deleu Monodontidae Toothed_whales 58 Harbour porpoise Phocoena phocoena Phpho Phocoenidae Phocoenidae 76 Finless porpoise Neophocaena phocaenoides Nepho Phocoenidae Phocoenidae 5 Dall's porpoise Phocoenoides dalli Phdal Phocoenidae Phocoenidae 84 2 For the global analysis Minke whale refers to estimates for both the Dwarf Minke whale (Balaenoptera acutorostrata) and also the Antarctic Minke whale (Balaenoptera Bonaerensis). For the individual species analysis only estimates for the Dwarf Minke whale were used. 4 Sperm whale Physeter macrocephalus Phmac Physeteridae Toothed_whales 53 Southern bottlenose Hyperoodon planifrons Hypla Ziphiidae Toothed_whales 2 whale Longman's beaked Indopacetus pacificus Inpac Ziphiidae Toothed_whales 1 whale Baird's beaked whale Berardius bairdii Bebai Ziphiidae Toothed_whales 8 Cuvier's beaked whale Ziphius cavirostris Zicav Ziphiidae Toothed_whales 12 Blainville's beaked whale Northern bottlenose whale Mesoplodon densirostris Meden Ziphiidae Toothed_whales 6 Hyperoodon ampullatus Hyamp Ziphiidae Toothed_whales Methodology 2.1. Analysis steps The primary tool used for examining trends in cetacean populations by combining multiple surveys from the literature was generalized linear and generalized additive modelling (GLM/GAM). Since abundance estimates covering different sized areas are clearly not comparable, we used density as the response variable in these models 3. There were many variables that could potentially be used as explanatory covariates for differences in density estimates which can be classified broadly into the following groups: taxonomic (e.g., species, family), spatial (e.g., ocean, Food and Agriculture Organization of the United Nations (FAO area), temporal (e.g., year, decade) and survey-related (e.g., survey method, agency). In total, five analysis steps were required: Data Exploration; Determination of Weights; Global Dataset Models; Individual Area and Species Models; Power Analysis. We treat each step in turn below. 2.2 Data Exploration The first step of data exploration was to convert reported observed abundance estimates to density estimates for each species within each survey based on the survey area. All surveys were digitized in ArcGIS 9.1, and the survey area calculated based on the produced shapefiles. Where different strata were present within one survey each stratum was entered with its associated estimates. Once density estimates had been calculated for all species within surveys, potential explanatory covariates were chosen. Graphical techniques were used to explore both potential patterns in density related to each covariate, and the number of estimates available using different covariate combinations. It was clear that many covariates had a large number of levels e.g., species, survey agency and FAO area. These levels were a direct result of the amount of data held within the ERMC database. The level of detail precluded the fitting of models for combinations of covariates, so parsimonious groupings were investigated to enable data exploration for each of the outlined covariate groups: taxonomic, spatial, temporal and survey-related. Table 2 contains a list of the potential covariates considered. The outcome of this first phase was a set of covariates that could be used in the subsequent modelling exercises, as well as information about which ones could potentially be modelled in combination. 3 An alternative, which we did not pursue, would be to use abundance with area as an offset in the model. 5 Table 2. Potential covariates considered for inclusion during exploratory data analysis. Abbreviations are those used in subsequent tables. See text for details of factor levels. Covariate Abbreviation Covariate group Type Species Species Taxonomic Factor, 87 levels Family Family Taxonomic Factor, 6 levels Species group SppGroup Taxonomic Factor, 4 levels FAO area FAO Spatial Factor, 18 levels Ocean basin Ocean Spatial Factor, 6 levels Mean latitude Lat Spatial Continuous Maximum latitude MaxLat Spatial Continuous Minimum latitude MinLat Spatial Continuous Year Year Temporal Factor (rounded mean year for multi-year estimates, 22 levels) or continuous Decade Decade Temporal Factor, 3 levels Season Season Temporal Factor, 3 levels Survey methodology Method Survey-related Factor, 6 levels Agency Agency Survey-related Factor, 33 levels Ocean basin and grouped survey agency OceanAgency Spatial and survey-related Factor, 11 levels Taxonomic covariates In total information on 63 cetacean species is held within the database, 34 of which fulfilled the criteria set for the global analysis described above. However, even within this selected group, information on abundance is variable across all species. For the global analysis the consideration of each species individually would most likely involve too many factors. Therefore, species were grouped into families (six groupings; Balaenopteridae; Delphinidae; Phocoenidae; Monodontidae; Physeteridae and Ziphiidae), containing between 1 and 20 species. An alternative taxonomic grouping combined the last three groups (two of which contained only a single species) into a single other_toothed_whales group (now consisting of 22 species). For the individual based analysis, species were considered to have sufficient data if the database contained more than 30 specific estimates roughly equally distributed between the different geographic subareas (unless they were not found in all three areas). Note that although information on stocks of cetaceans was outlined in Task 1, the data are considered here on a species basis Spatial covariates Spatial coverage of surveys varied both within and across the areas of relevance. Surveys areas could be calculated as Km 2 and hence density estimates could theoretically be reported in the same resolution. However, since species distribution is generally unlikely to be homogeneous across entire survey areas, a spatial resolution below the size of survey areas would be misleading. Unfortunately, survey areas within the same region also tend to vary considerably between different years, thus precluding the use of survey areas as the minimum spatial unit of analysis. Therefore the surveys were allocated to specific statistical reporting areas that are used by the Food and Agricultural 4 It is not possible to look at trends of cetaceans on a stock level for the global analysis because abundance estimates reported in the literature do not consistently provide stock information. Certain stock assessment reports do provide this level of detail and, as such, stock information can be addressed during the species and area specific section of the analysis. 6 Organization (FAO) (Figure 1). However, even at this relatively large spatial scale, several surveys took place over multiple FAO areas and for these surveys, datasets with one identical record for each FAO area were created, e.g. if one overall density estimate was available it would be entered for each FAO region that the survey covered. This applied to 70 surveys, turning the 219 species-specific density estimates for these surveys into 519 records, and therefore making the total number of records for the FAO-level analysis to This is a clear case of pseudoreplication, but is defensible in the context of an exploratory analysis (see discussion). FAO areas could be allocated to ocean basins reducing the number of geographic units further, to six ocean basins (Pacific, Atlantic, Indian Ocean, Mediterranean, Arctic and Antarctic). Since no survey covered more than one ocean basin the original dataset with a column for Ocean basin could be used in the model (eliminating any pseudoreplication as seen in the FAO-level approach). The result was two datasets; one including the pseudoreplication for FAO area and one with the groupings for Ocean basin. Both of these were taken forward from the exploratory analysis to the first stage of modelling. A further spatial level that was considered was average (coarse measure of shape of survey, i.e. if it is tear-shaped, or symmetrical), and minimum and maximum latitude values for the survey blocks. This approach enabled consideration of multiple abundance estimates for single species that may consist of geographically different populations Temporal covariates Two levels of temporal information exist in the database. The first is the annual level i.e., the year of the survey. For survey estimates that cover multiple years, the midyear was used. In subsequent modelling, when year was used as a covariate, this was rounded to the nearest integer, but noninteger values were left when year was used as a continuous covariate. Year was also used to assign surveys to decades. The second level of temporal information w
Advertisement
MostRelated
View more
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks