|Type of data used||Reliability of
length-weight conversion data used
|Average weight data||Not applicable||1|
|Common or typical weights, or normal weight ranges||Not applicable||2|
|Average, common or typical lengths, or normal length ranges||Good||3|
|Any of the above where only the upper end of the estimated mean weight range was obtained||4|
|Average weights based on maximum weights||Not applicable||5|
|Average weights based on maximum lengths||Good||5|
|Average weights based on (rank 1-3) data for a similar species||Not applicable (not used on the species in question)||7|
Ranking data by reliability
Where possible, estimated mean weights used in this study were based on cited average weights. However, such data was often not available, and a mean weight estimate was made from other types of fish size data that are likely to be less reliable than quoted average weights. On the one hand, we want to include as many references as possible. On the other hand, we donít want to use less reliable data where good data is available. In order to balance these 2 objectives, a method of ranking data according to its reliability was used.
The basic principle is that all types of data used to estimate mean weights are assigned a 'reliability ranking' ranging from 1 to 7, with '1' being the highest and most reliable. Where a species has references pertaining to different values of reliability ranking, only those for the highest ranking are used.
Sometimes length data was available for a species where weight data was not. In such cases, lengths were converted to weights using length-weight relationship data available on fishbase.org. For our purposes, length-weight relationship data is classed as being either 'good' or 'less good', according to how well it matches the lengths we are trying to convert. This is explained in more detail on the 'Help' screen accompanying for Screen 5 (Length-weight calculations) accessed from screen 4 (reference data details) where applicable. Whether data used to convert lengths to weights is 'good' or 'less good' affects the reliability of the data, and therefore its reliability ranking, as as can be seen in the table above.