The abundance of ecosystem indicators under consideration has increased substantially over the last decade (see contributions in Cury and Christensen 2005). The challenge of the IndiSeas Working Group was not to develop new indicators, but rather to use specific selection criteria to choose the most representative and practically achievable and meaningful set of indicators from existing ones.

The intent of IndiSeas was to build on the work already achieved by the SCOR/IOC Working Group 119 on “Quantitative Ecosystem Indicators”, and specifically on the results of Rice and Rochet (2005) who outline some specific practical criteria for the selection of ecosystem indicators which were adopted by the SCOR-IOC Working Group:

• ecological significance (i.e. are the underlying processes essential to the understanding of the functioning and the structure of marine and aquatic ecosystems?)
• measurability: availability of the data required for calculating the indicators
• sensitivity to fishing pressure
• awareness of the general public

The last of these criteria was of particular importance to the aims of the IndiSeas WG, that is the awareness of the general public concerning the meaning (what information is communicated) of each indicator. For example, among potential size-based indicators, we preferred to select mean length rather than the slope of the size spectrum since this would be more difficult to communicate to the general public.

In addition to these practical selection criteria, the indicators were selected to address four specific management objectives: Conservation of Biodiversity (CB), ecosystem Stability and Resistance to perturbations (SR), Ecosystem structure and Functioning (EF) and Resource Potential (RP).

The most constraining criterion in the comparative framework was that of the availability of the data (from observations or models). The indicators needed to be comparable across ecosystems and not too costly (the list of indicators must be concise), so as to be easily estimated and gathered for each ecosystem considered. In some ecosystems, the data required for calculating the selected indicators have not been collected yet or are not readily available. However, the working group found that it is important in those cases, to set up sampling or modelling programs to fill the gaps. Therefore, the minimal list is not strictly the lowest common denominator of all the ecosystems represented within the group.

In the review of existing ecosystem indicators, several categories of indicators were distinguished (Cury and Christensen 2005): size-based, species-based, and trophodynamic indicators. The eight indicators outlined in Table 1 were selected based on the above criteria, and are proposed as a minimum set of indicators for diagnosing the status of an ecosystem. Six of the indicators were used to measure the state (S) of the ecosystem and six were used to measure trends (T) over time. Data for the indicators are derived primarily from fisheries independent surveys and commercial fisheries data, with auxiliary information where indicated.

In addition to the full indicator name, a shorter “headline label” was attributed to each of the indicators (Table 1) to make them more readily comprehensible. Furthermore, the indicators are all formulated positively so that a low value of an indicator means a high impact of fishing and a high value a low impact of fishing. Similarly, an increase of the indicator means an improving state, whereas a decrease means a deteriorating state.

Indicators	Headline label	Calculation, notations, units	Used for (S)tate, (T)rend	Expected Trend	Manage-ment Objectives	Management Direction
Total biomass of surveyed species	biomass	B (tons)	T	D	RP	Reduction of overall fishing effort and quotas
1/(landings /biomass)	inverse fishing pressure	B/Y retained species	T	D	RP	Reduction of overall fishing effort and quotas
Mean length of fish in the community	fish size		S,T	D	EF	Reduction of overall fishing effort and fishing effort on large fish species
TL landings	trophic level		S,T	D	EF	Decrease fishing effort on predator fish species
Proportion of under and moderately exploited stocks	% sustainable stocks	number (under+moderately exploited species)/total no. of stocks considered	S	D	CB	Decrease fishing effort on overexploited species. Diversify resource composition
Proportion of predatory fish	% predators	prop predatory fish= B predatory fish/B surveyed	S,T	D	CB	Decrease fishing effort on predator fish species
Mean life span	life span		S,T	D	SR	Decrease fishing effort on long-living species
1/Coefficient of variation of total biomass	biomass stability	mean(total B for the last 10 years) /sd(total B for the last 10 years)	S	D	SR

Total biomass of surveyed species is a conservative property of an ecosystem; as species are fished and their biomass reduced, other species increase in abundance and “replace” these species in the foodweb. With the removal of top predators lower trophic levels can be expected to increase. Thus changes in total biomass can be reflective of changes in ecosystem productivity. “Biomass” is used here as a measure of “resource potential (Table 1). “Biomass” was not used to characterise the ecosystem state since survey data does not provide absolute estimates of biomass and thus is not comparable between species or ecosystems (due to differences in species catchability and surveys). Instead, “biomass” was used to compare biomass trends over time.

1/(landings /biomass) measures the inverse level of exploitation or total fishing pressure on the ecosystem. This indicator varies in the same direction as the other indicators in the selected suite, as it decreases when fishing pressure increases. A decrease is considered negative and is a measure of “resource potential” (Table 1). As for total biomass, this indicator is only used for comparison of trends since absolute estimates of biomass are generally not available.

Mean length of fish in the community is an indicator of the impact of fishing on an ecosystem, that is, the reduction of mean length of fish in the community (Shin et al. 2005). From a single species perspective, the removal of larger fish, which are more fecund and produce more viable eggs than smaller fish (Longhurst 1999), compromises productivity. From an ecosystem perspective, the removal of larger species changes the size structure of the community and potentially ecosystem functioning. “Fish size” is thus a measure of ecosystem structure and functioning (Table 1) and is used to measure state and trend.

Trophic level of landings measures the average trophic level of species exploited by the fishery, and is expected to decrease in response to fishing, since fisheries tend to target higher trophic level species (Pauly et al. 1998). A decrease in trophic level of landings and total catch indicates “fishing down the food web” (Pauly et al. 1998), and a change in the structure of the community and potentially ecosystem functioning. “Trophic level” is a measure of ecosystem structure and functioning (Table 1) and is used to measure state and trend. Trophic level of individual species is either estimated through modelling, or taken from global database such as Fishbase.

Proportion of predatory fish is a measure of the diversity of fish in the community. Predatory fish are all surveyed fish species that are piscivorous, or feed on invertebrates that are larger than 2 cm. “% predators” is a measure of conservation of biodiversity (Table 1) and is used to measure state and trend.

Proportion of under and moderately exploited stocks represents the success (or not) of fisheries management. Ideally, in a precautionary world, all stocks should be moderately exploited to ensure sustained biodiversity and sustainable ecosystems. “% of sustainable stocks” is a measure of conservation of biodiversity (Table 1). The FAO classification of stocks as underexploited, moderately exploited, fully exploited etc (http://www.fao.org/docrep/009/y5852e/Y5852E10.htm#tbl) was used to define these categories for the stocks in each ecosystem under consideration in the current time period. Thus this indicator is used to compare the state of ecosystems.

Mean life span is a proxy for mean turnover rate of species and communities, and is meant to reflect the buffering capacity of a system. The life span or longevity is a fixed parameter per species, and therefore the mean life span of a community will reflect the relative abundances of species with differential turnover rates. Fishing affects the longevity of a given species (direct effect of fishing and genotype selection), but the purpose here is to track changes in species composition (same principle as for mean TL of catch). “Life span” is thus a measure of ecosystem stability and resistance to perturbations (Table 1) and is used to measure state and trend.

1/Coefficient of variation of total biomass measures the stability of the ecosystem, and is measured as the coefficient of variation (CV) over the last 10 years. As with “fishing pressure”, it is expressed as an inverse to make it conform with the directionality of the other indicators. Thus a low 1/CV indicates low “biomass stability”, low ecosystem Stability and Resistance to perturbations. Since this indicator is measured over a 10 year time period, it is only used to measure state.

Surveyed species
These are species sampled by researchers during routine surveys (as opposed to species sampled in catches by fishing vessels), and should include species of demersal and pelagic fish (bony and cartilaginous, small and large), as well as commercially important invertebrates (squids, crabs, shrimps…). Intertidal and subtidal crustaceans and molluscs such as abalones and mussels, mammalian and avian top predators, and turtles, should be excluded. Surveyed species are those that are considered by default in the calculation of all survey-based indicators.

Retained species (landed)
These are species caught in fishing operations, although not necessarily targeted by a fishery (i.e. include by-catch species), and which are retained because they are of commercial interest, i.e. not discarded once caught, although this does not imply that sometimes certain size classes of that species may be discarded. A non-retained species is considered to be one that would never be retained for consumptive purposes. Intertidal and subtidal crustaceans and molluscs such as abalones and mussels are to be excluded. Retained species are those that are considered by default in the calculation of all catch-based indicators.

Predatory fish species
Predatory fish are considered to be all surveyed fish species that are not largely planktivorous (i.e. phytoplankton and zooplankton feeders should be excluded). A fish species is classified as predatory if it is piscivorous, or if it feeds on invertebrates that are larger than the macrozooplankton category (> 2cm). Detritivores should not be classified as predatory fish.

Images help us make sense of who we are, what we do and how we relate to others and to things around us (Jentoft et al. 2008). They help us to describe, explain and to synthesize information. As such, they are ideal tools for conveying and synthesising the information from a suite of ecosystem indicators such as those described here. We have developed a “generic dashboard” to present the ecosystem indicators describing the state of ecosystems and the trends within them, using pie diagrams and simple bar plots. The advantage of such a representation lies in providing a multivariate view of the ecosystem.

Pie diagrams were used to present the results of the state analysis where state indicator values (Table 1) were averaged over 2003-2005 to represent the current state of the ecosystem. Each pie corresponds to a selected indicator. On this axis, the indicator is scaled between a minimum value (centre of the diagram) and a maximum value which represents an optimum or a target value. These values are constant across all the ecosystems considered in the generic dashboard and are determined by the minimum and maximum values observed in the set of ecosystems. The purpose of the boundaries is to scale the indicators for graphical representation. This approach highlights the importance of an inclusive set of case study ecosystems.

Short to medium terms trends were calculated over a 10 year period, 1996-2005 for the suite of six standardized trend indicators (Table 1). Bar plots were used to represent the trends which were significant, green indicating an increase, and red a decrease. Grey bars represent trends which are not significant statistically.

This generic dashboard provides (i) set of common indicators, (ii) common methods for estimating the indicators, (iii) common methods for evaluating ecosystems’ status and trends, (iv) common methods for representing ecosystems’ status and trends and (v) provides a platform for comparing a broad range of ecosystems.