Development of a Fish Classification Scheme for UK Transitional Waters Preliminary results based on an African model
Species & Community Assessment South African Classification Scheme. Research was based on a 7 year intensive field sampling programme during which 257 estuaries were visited. Using fisheries data and typological classification, biogeographic regions were identified and characterised in order to form six basic estuary types.
Approach Delineation of biogeographic regions Estuary typology classification Selection of fish community measures/metrics Development of reference conditions and metric scores Index calculation and rating The South African Team (Trevor Harrison et al) spent 7 years and surveyed over 250 estuaries. Geomorphological and fisheries biometric data was collated at each site and these datasets form the basis for the multi-variate analysis (using PRIMER) which associated each dataset to its physical estuary type. i.e associating resident fish communities to typology (sound familiar?!). PRIMER was developed by staff at the Plymouth Marine Lab and we’re in close contact with regard to our datasets and how to take the multivariate analysis a stage further by using the work they’re pioneering with ‘Taxonomic Distinctness’. We plan to evaluate our datasets using this technique.
Biogeography 3 regions (Harrison, 2002) South Africa Subtropical Kosi Bay Orange River South Africa Subtropical These are the 3 biogeographic regions that could be separated out of the multi-variate analysis. (using PRIMER). The above is supported by S. African expert opinion i.e this is how it is Port St Johns Cool-temperate Warm-temperate Cape Agulhas
(small, ephemeral, isolated) Estuary typology 6 types (Cooper) Coastal Outlet Non estuary (small, ephemeral, isolated) Estuary Normally open Normally closed Non-barred Barred The estuaries were divided up into 6 types by Andrew Cooper (University of Coleraine. N.I.). Once this was done then estuary type datasets could be grouped so that you are comparing like with like and not estuaries of a different type Large SA >150 ha Medium SA 2-150 ha Small SA <2 ha Large MAR>15 x 106m3 Small MAR<15 x 106m3
Comparison & similarity between sites, dates, estuaries, etc Comparison & similarity between sites, dates, estuaries, etc. Functional Guild approach) This is some of the work that Mike Elliott (IECS) has been doing with functional guilds. European estuaries are along the bottom e.g Tagus, Loire, Forth, Tyne, Humber, Solway, Mersey The key on the right is each functional Guild:- FW = Freshwater CA = Catadromous MS = Marine Seasonal migrant MJ = Marine Juvenile MA = Marine Adventurous ER = Estuarine Resident The percentage of each guild is down the side
Comparison & similarity between sites, dates, estuaries, etc Comparison & similarity between sites, dates, estuaries, etc. (functional approach) Using the ecological guild approach the data for each estuary is then analyzed using the ‘Bray-Curtis’ approach and is exported from the data matrix to the above graphical form. Estuaries are grouped by their percentage similarity (down the y axis)
Metric selection Review of literature Ecological relevance (e.g. Karr, 1981; Miller et al. 1988; Deegan et al., 1997; Whitfield & Elliott, 2002) Ecological relevance Ease of measurement Structure Function Qualitative (presence/absence) Quantitative (abundance) Literature review also gave examples of what were the suitable metrics (see e.g’s) and how we could apply them to fish in Transitional Waters.
Fish Community Measures Species diversity and composition 1) Species richness (number of taxa) 2) Presence of rare/threatened species 3) Presence of exotic/introduced species 4) Species composition Species abundance 5) Species relative abundance 6) Number of taxa that make up 90% of the abundance Nursery function 7) Number of estuarine resident taxa 8) Number of estuarine-dependent marine taxa 9) Relative abundance of estuarine resident taxa 10) Relative abundance of estuarine-dependent marine taxa Trophic integrity 11) Number of benthic invertebrate feeding taxa 12) Number of piscivorous taxa 13) Relative abundance of benthic invertebrate feeding taxa 14) Relative abundance of piscivorous taxa The above are examples of how we can evaluate our datasets. The first 6 (no’s 1-6) were used in the S. African state of the environment report. No’s 7-14 are some other metrics that Trevor Harrison has been thinking about with the S. African datasets They are also colour coded, in that, green text are metrics that relied upon presence/absence data and yellow involves abundance data
Reference conditions Species diversity and composition 1) Species richness (number of taxa) Remove exotic/introduced taxa Calculate estuary species richness Closed medium-sized subtropical estuaries This is how they derived reference conditions for EACH estuary type. They took all the datasets from that type and ranked them in relation to number of taxa
Reference conditions Species diversity and composition 1) Species richness (number of taxa) Remove exotic/introduced taxa Calculate estuary species richness Select upper quartile Closed medium-sized subtropical estuaries Selected an upper quartile
Reference conditions Species diversity and composition 1) Species richness (number of taxa) Remove exotic/introduced taxa Calculate estuary species richness Select upper quartile Calculate mean richness of upper quartile Closed medium-sized subtropical estuaries Mean = 23 Took an average of that quartile, and then took the mean as what could be expected in a reference estuary of that type. They then developed a reference fish species list for that estuary type
Reference conditions Species diversity and composition 1) Species richness (number of taxa) Remove exotic/introduced taxa Calculate estuary species richness Select upper quartile Calculate mean richness of upper quartile Closed medium-sized subtropical estuaries Mean = 23 The confidence intervals e.g. 90% can be calculated. Links to the debate on whether reference conditions are a fixed number or a range. 90% Confidence Intervals 19.45 - 26.55
Metric Scoring System Score 5 - deviate slightly from reference > 90% expected reference value (range) Score 3 - deviate somewhat from reference 50 - 90% expected reference value (range) Score 1 - deviate strongly from reference < 50% expected reference value (range) Trevor (Harrison) then developed a metric scores system based upon the Index of Biotic Integrity (Karr, 1981). He then applied a scoring system for each of the 14 metrics and gave:- Score 5 - deviate slightly from reference (based on >90% of reference value) Score 3 - deviate somewhat from reference (based on 50-90% of reference value) Score 1 - deviate strongly from reference (based on <50% reference value)
Index calculation and rating Sum from all 14 metric scores Total scores range from 14 to 70 Score Rating 14 - 18 Very Poor 20 - 38 Poor 40 - 44 Moderate 46 - 54 Moderate to Good 56 - 64 Good 66 - 70 Very Good The range of scores are then split into descriptive “bands” Note - there is an assumption that all of the 14 metrics have equal weighting
State of Environment The above is for just one estuary type - Closed medium-sized subtropical estuaries The above is for just one estuary type - ‘Closed medium-sized subtropical estuaries’. The red lines are from the previous slide and is the moderate band. Most of the estuaries are below moderate - which is not surprising as this part of the S. African coast, which has been impacted by ribbon development and so the estuaries are fairly impacted.
Monitoring The Sezela estuary is interesting as its one that has a lot of long-term datasets. Similar to the Mersey, in that it suffered from chronic organic pollution in the past, and in 1984 the S. Africans started a rehabilitation programme. The fisheries data reflects the improvements made and by Aug 91 the estuary was up to moderate. Though by 92 it had dropped but this was directly as a result of a massive fish mortality caused by a pollution. A good example of how fish reflect such changes
Monitoring GOOD MODERATE POOR Class boundaries to reflect WFD status could be drawn in based on the SA splits. Case for Intercalibration? Also High Status - is it = to reference condition, a max score of 70, or can we allow a slight deviation?
Taxonomic distinctness as a new way of monitoring change Taxonomic distinctness is a new way of monitoring change and has been developed by the Plymouth Marine Laboratory Recently-developed measures which are not based on abundance's of organisms, but which take into account their taxonomic relationships, have been shown to be useful for detecting spatial and temporal changes related to variations in environmental conditions. These measures have several important potential advantages over measures currently being used to assess environmental change. For example, much research effort is required to gather quantitative information about the abundance distribution of organisms in order to calculate indices traditionally used to detect change. The new measures may be used with historical information collected with uncontrolled or unquantified sampling effort. In a recent project the utility of one of these indices, as an indicator of biodiversity and environmental health, was tested with the NMMP benthic dataset. Fisheries data is to be tested ASAP. Given these advantages such novel statistical techniques present a potential opportunity to redefine the nature of environmental change, which may have implications for the requirements (such as reducing replication effort) for environmental monitoring. With the Benthic data the index, taxonomic distinctness (+) was calculated for pooled replicate samples from 67 sites, collected mainly in 1999, 2000 and 2001. Comparisons with a suite of other indices (including, among others, Species numbers (S), Shannon diversity (H’), Margelef’s d; Pielou’s J) showed that all of these varied significantly between samples from estuaries and samples from coastal sites, samples from different years, samples from different sites, samples from different regions, samples containing different numbers of replicates, and samples collected using different methods. Many of them were significantly intercorrelated (conveying the same information). Taxonomic distinctness was much less, and generally insignificantly, influenced by such confounding factors, and not intercorrelated with other measures None of the other indices was significantly correlated with a suite of measured metals or with total organic carbon, whereas + was negatively correlated with metals (and organic carbon) in estuaries and positively correlated with organic carbon in coastal waters, suggesting that the measure has greater potential as a meaningful measure of environmental health. Finally, the utility of the statistical framework associated with the measure as a means of defining ecological status was briefly explored. By deciding the levels of departure from expecation which were acceptable, cause for potential concern, or unacceptable, each sample could be assessed independantly and assigned to a staus of ‘good’, ‘moderate’ or ‘poor’.
The expected value of the index, taxonomic distinctness (+), and confidence intervals for the benthic data sets Figure: A. The expected value of the index, taxonomic distinctness (+), and confidence intervals, are calculated for subsets comprising different numbers of species. Note that the expected value does not change, although the intervals decrease with increasing numbers of species. B. Actual values for samples may be compared with expectation. C. Different degrees of departure from expected values may be defined as representing different levels of ecological status. Here values within 90 % intervals are considered good, between 90 and 95 % moderate, and outside the 95 % interval as poor. Note that samples with the same number of species, or with the same value of the index, may have different status depending on their position in the confidence funnel. D. These values may be used to map status.
Conclusions and Next Steps The South African model shows promise A final scheme will be based on multi-metrics Problem in UK is lack of data for testing We will need further validation work Aim is to have a firm recommended approach ready by the end of 2003