Australia: State of the Environment Second Technical Paper Series (Coasts and Oceans), Series 2
David Barratt, John Garvey and Jean Chesson
Bureau of Resource Sciences, Australia
Department of the Environment and Heritage, 2001
ISBN 0 6425 4745 9
This chapter describes the methods that apply generally across all or most of the indicators. More specific methods are given in the chapters on individual indicators and within the case studies.
Prior to this project, there was no comprehensive list of Australian fisheries and the type of fishing methods they use. A fishery is a management unit defined by geographical area or fishing method or species or a combination of any of these. Since data are collected and management decisions are made with respect to individual fisheries, it was determined early in the project that the fishery would be the basic unit of study.
All Commonwealth, State and Territory fisheries agencies were approached for a list of their fisheries. The table in Appendix I listing fisheries by method was compiled and returned to each agency for confirmation.
There are currently 144 commercial estuarine and marine fisheries in Australia. Many of these fisheries use a variety of fishing methods. Also, many fisheries overlap in geographical area and may fish common species. Although the list of fisheries will change as management agencies modify their management arrangements, the creation of this list was an important first step in tackling this project and provided the basis for subsequent work.
For all Australian fisheries relevant to Indicators 3 and 4, the data collection forms used in logbooks and fisheries returns were examined. The following information was recorded:
- Whether the form has a section for a detailed description of the gear(s) used by the fisher,
- A description of the temporal resolution of the effort data collected,
- A description of the spatial resolution of the effort data collected, and
- A description of the effort data itself.
Gear details collected by the forms usually change infrequently. They include information such as the dimensions of nets, mesh sizes, size of hooks, lengths of branch lines, headline and footrope length. These details can be important inputs to analysis of fishing effort. For example, swept area of a trawl cannot be calculated without details of headrope length.
The temporal resolution refers to the unit of time in which the effort is reported. Some forms require that fishers report the number of days that they fished in a certain month. Other forms require that fishers report details of fishing for each day. Daily logbooks can require that fishers report details of every fishing operation.
The spatial resolution refers to the area of water in which the reported fishing activity occurred. Fisheries agencies often use grids or other specified statistical areas for catch and effort reporting. Grids can be any size up to one degree of latitude and longitude square, and statistical areas or zones can be irregular regions bounded by coastline and lines drawn between headlands. Fishers report their monthly or daily activity within each zone fished. Other logbooks request that the fisher report the latitude and longitude of each operation, or the location where most catch is taken.
The effort data collected can vary considerably from simply the number of days spent fishing using a particular method, to detailed reporting of the amount of gear used in each fishing operation. Effort details can have a gear component (eg: numbers of hooks or length of net) and a time component (eg: hours fished, start and finish times for particular operations). Sometimes only one of these components is required by a form.
The waters off NSW and Victoria were selected as the basis for case studies of Indicators 1, 3 and 4. This area was chosen for the following reasons:
- choosing the same area for all indicators illustrates the issues of multiple use and makes best use of GIS analyses and presentation,
- data for all indicators exist in this region,
- the number of fishers operating in this area is large enough to minimise the impact of matters associated with data confidentiality,
- the case studies involve a limited number of fisheries agencies,
- data from different gear types are in limited formats which reduce processing time, and,
- this is the first area for which a Regional Marine Plan will be developed under Australia's Oceans Policy (EA 1998).
The quantitative components of Indicators 3 and 4 and the fisheries selected to illustrate these components in space and time are shown in Table 3.1. Also shown are the logbooks from which data were collated and the years over which fishing trends were analysed. There is very little demersal prawn trawling in the south east of Australia, so the Northern Prawn Fishery was selected as a case study for that method.
The approximate extent of GIS analyses and maps generated as part of the fisheries case studies selected to illustrate Indicators 3 and 4 are shown in Figure 3.1.
|Indicator component||Fishery||Current Logbook||Years|
|Extent of gillnetting||Southern Shark and South East Non-trawl||GN01||1998|
|Extent of droplining||South East Non-trawl||GN01||1998|
|Extent of longlining||Southern Shark and South East Non-trawl
Australian Pelagic Longline
|Extent of meshing||South East Trawl (Danish seine)||SEF1B||1989-1998|
|Area of trawling||Northern Prawn Trawl
South East Trawl (Otter trawl)
BRS has signed a Deed of Confidentiality with AFMA covering the Bureau's use of catch and effort data for research that contributes to fishery management objectives. This project is not covered by the deed and BRS sought to facilitate discussions between AFMA and Environment Australia to define the use and approved outputs for the final report. These discussions are currently near resolution. Environment Australia also contacted fisheries agencies in Victoria and NSW and agreements on data supply and use were negotiated.
Commonwealth fisheries data were acquired from AFMA in a standard format used by BRS for fisheries assessment, and loaded into an Oracle relational database. The raw data were not validated or corrected within the database with the exception of trawl and Danish Seine records from the South East Fishery. An update script was applied to these fishing operation data to correct errors in identification of fishing method. The datasets were then exported to a temporary data library pending quality assurance and translation, where required, into a preferred ArcView compatible format. As data sets were copied to the project area, basic metadata records were created for each. The fishing effort data for Victoria and New South Wales were received in a standard digital format and treated similarly to the Commonwealth data ie the data files were copied to a data library, metadata created and then loaded into ArcView files for GIS analysis.
Criteria for GIS Analysis of Indicators
The ability of the logbooks and forms to provide the appropriate spatial, temporal and effort data for IMCRA region GIS analysis was determined using the criteria described in Table 3.1.
|Spatial||Fishing location given by latitude or longitude or by reference to a grid of 15 minutes square or less|
|Temporal||Fishing operations reported individually or details of fishing reported daily|
|Effort||Gillnet: length of net used
Dropline: number of hooks used
Longline: number of hooks used
Meshing: number of operations
Trawling: Hours trawled
These criteria were determined to be the minimum for GIS analysis for indicators 3 and 4. The spatial criteria was judged to be the largest that would allow effort to be assigned to IMCRA regions without allocating attributes of a single spatial logbook grid across IMCRA boundaries. The temporal and effort criteria were the recommended units for the indicators.
The number of logbooks that met these criteria for each gear type in indicators 3 and 4 were determined. It is understood that such an analysis gives only an indication of the ability of fisheries effort data programs to provide information for GIS analysis.
Data used in analyses of the four indicators were restricted to that available up to the end of December 1998.
Analyses by Grid and IMCRA Region
Locations (latitude and longitude) of operations in the fisheries examined as case studies in this report were recorded by boat skippers using either GPS, or reference to a grid overlaid on the fishery area. To generate a smoothed gridded representation of the relative intensity of effort throughout a fishery in each year from 1989 to 1998, a nearest neighbourhood analysis was applied to the point data from each year using ArcView Spatial Analyst tools. These analyses generated 1x1 minute grids for each year, where the value of each cell in each grid was the sum of the effort data associated with point locations within a specified area around the cell. The neighbourhood analysis area chosen varied between fisheries depending on the spatial precision of the data. Intersecting the grids from each year and summing the cell values generated a smoothed gridded representation of fishing effort between 1989 and 1998. This gridded output provided an indication of the relative intensity of fishing throughout a fishery and could be likened to a contour map of fishing intensity.
Locations (latitude and longitude) of start and end points of trawl operations were available for the South-east Trawl Fishery. To generate a gridded representation of the relative intensity of operations throughout the fishery in each year from 1989 to 1998, these data had to be first assigned a unique identifier and assembled into arcs. An "awk" command was used in UNIX to fold the original lines, as shown below, in preparation for an ARC "generate line" command.
|Original lat long data:||1,-37.3,150.3,-37.1,150.3 end
|Changed to:||1,-37.3,150.3,-37.1,150.3 end
Records with identical start and finish latitudes and longitudes were removed. Vector (line) coverages of trawl tracks for each year from 1989 to 1998 were then compiled using the ARC generate line command.
2-D seismic survey data were available as vector coverages. The ARC "linestats" command was used to convert the South East Trawl fishery coverages and 2-D seismic survey coverages into 1x1 minute grids of distance trawled using neighbourhood radiuses of 0.176779 degrees and 0.011785 degrees respectively. These radiuses were ascertained to be the minimum needed to span from the centre to the corner of 15x15 minute and 1x1 minute grid cells respectively. The radiuses were chosen because they represented the poorest spatial precision of records in the South East Trawl fishery and 2-D seismic survey datasets. (NB: The ARC linestats command does not provide an option for specifying a rectangular neighbourhood, as was done using Spatial Analyst tools on point data in other case studies).
Intersecting the South East Trawl fishery grids from each year and summing the cell values generated a smoothed gridded representation of total distance trawled between 1989 and 1998. This gridded output provided an indication of the relative intensity of trawling throughout the fishery and could be likened to a contour map of trawling intensity. The final grids (South East Trawl fishing intensity and 2-D seismic survey intensity) were converted from floating-points to integer grids using the GRID "int" command.
An ARC coverage of the IMCRA regionalisations (IMCRA Technical Group 1997) was obtained from ERIN. Calculating fishing, trawling or seismic surveying intensity by IMCRA regions was done by intersecting fishing point locations or trawl lines with the IMCRA coverage and summing the effort data attributed to these point or line records in each IMCRA region. The sum total values for each IMCRA region were then divided by the area (in square nautical miles) of the region. This gave a standardised (per square nautical mile) annual fishing or trawling intensity value in each IMCRA region. Where total fishing or trawling intensity per IMCRA region was estimated for the period 1989-1998, the effort data values in each region in each year were summed before being standardised by the area (in square nautical miles) of the region.
Depicting temporal change spatially is a significant challenge. One approach is through animation. Separate maps are prepared for successive points in time and presented to the viewer as a video. This approach was impractical for this project, but could be revisited if there is a decision to provide video clips as part of the State of Environment Report. Temporal trends for relatively large spatial units (national, IMCRA regions) were displayed in the traditional way as plots against time and these appear as figures throughout the report.
In order to display temporal changes on a map, decisions need to made as to what aspects of change need to be presented. As far as we are aware, this is the first time this type of display has been attempted for fishery effort data. Absolute change (value at time 2 minus value at time 1) is the simplest and most obvious measure. It has the disadvantage, however, of highlighting areas where there has been a large absolute change (for example, an increase in the number of fishing operations from 100 to 150) at the expense of areas with a smaller absolute change but a much greater relative change (for example, an increase in the number of fishing operations from 5 to 30). It could be argued that in terms of potential ecosystem impacts, an area that has experienced a 6-fold increase in fishing operations over a given time period is of greater interest than an area which has a continuing high, but fluctuating number of fishing operations. On the other hand, large relative increases coming from a low starting value may receive unwarranted prominence when the focus is on relative rather than absolute change.
After some experimentation, the approach taken in this project was to subtract the 1989 effort values (per grid cell or IMCRA region) from the 1998 effort values and divide the result by the standard deviation of annual effort data between 1989 and 1998. The intuitive appeal of this approach is its simplicity and the notion of expressing change in terms of the number of standard deviations to provide some sort of scale to the magnitude of the change relative to the level of year-to-year fluctuations. The approach, which has some of the features of a correlation coefficient, is not entirely satisfactory because it combines a measure of absolute change with a measure of variability and tends to de-emphasise large changes. It could be improved by using the standard deviation of the data around a fitted line rather than the simple standard deviation provided by the GIS software, but this does not seem to be a sensible way to proceed. Once you have invested the effort in fitting a line, it would be more sensible to use the results of the fitting procedure directly.
We are continuing to explore the possibilities. A more rigorous approach would be to calculate the slope of the least-squares best fit line to the data and to measure the goodness of fit with a R2 value. It would then be necessary to decide whether to display both pieces of information (slope and goodness of fit) on either the same or separate maps. Note that this is a curve fitting exercise only and does not involve distributional assumptions or tests of statistical significance.
The presentation of fishing intensity on maps produced as part of this report was standardised across fisheries. Fishing intensity categories on map legends reflected percentiles of the range of fishing effort data being displayed. Where total fishing intensity was mapped, the categories in the map legend and the corresponding data percentiles were:
Very low = bottom 10% of data (effort) values
Low = 10-30%
Moderate = 30-70%
High = 70-90%
Very high = top 10% of data (effort) values
Where change in fishing intensity was mapped, the direction of change (positive or negative) with the greatest magnitude was used to determine percentile values for each category in the legend. This meant the shading and colour of increasing and decreasing fishing intensity categories would be comparable to one another. It also meant some categories in the direction of change with least magnitude may contain no data. For example, if change in effort values ranged from -50 to +150 then the "Very large increase" and the "Very large decrease" categories would be +135 to +150 and -135 to -150 respectively. Consequently, in this case the "Very large decrease" category would contain no data.
The categories in the map legend and the corresponding data percentiles were:
Very large increase/decrease = top 10% of data values in the direction of greatest change
Large increase/decrease = 70-90%
Moderate increase/decrease = 30-70%
Small increase/decrease = 10-30% of data values in the direction of greatest change
The bottom 10% of values (based on the direction of greatest change) were not mapped and were considered to represent areas in which there was no significant change in fishing effort. While the cut-points for each category are arbitrary, they provide a means of standardising the presentation of effort information between grid and polygon analytical techniques, and within and between fisheries. When interpreting this information it should be kept in mind that the descriptions of levels of fishing intensity or change in fishing effort are only relative to other parts of the fishery. They are not definitive judgements about the level of fishing intensity or degree of change in fishing effort, as they are not based on studies of impacts on marine species or any other criteria.