Citizen sciences are undergoing strong growth, a fact demonstrated by the session devoted to the topic at the 3rd French-language meetings on conservation biology (Le reveil du dodo III, 17-19 March 2009 in Montpellier) and the seminar titled Citizen science and biodiversity, held in Montpellier on 22-23 October 2009.
Marion and Frédéric Gosselin, engineer and researcher at Cemagref in Nogent-sur-Vernisson discuss the topic here with Romain Julliard, researcher at the MNHN bird-ringing project and who has managed a number of Vigie-Nature programmes requiring public participation (naturalists and amateurs) to collect the necessary data. The discussion successively addresses the history of citizen sciences, their advantages and limits, focussing on the assessment of biodiversity-conservation policies.

1. Why call on the general public to monitor changes in nature (for biodiversity, but also physiological phenomena as is the case for the seasonal observatory)?

Romain Julliard. Our primary goal is to have numerous observation points for species suitable for new scientific publications that would not be possible otherwise. For example, there are many scientific publications on butterflies, but not on butterflies on private premises, such as gardens.

Our second goal is to develop a means to raise public awareness and make observers change their perception of nature and biodiversity.

2. What is the history of citizen monitoring? Where is it the most developed?

Romain Julliard. Ever since there have been naturalists, amateurs have taken part in observations and naturalist organisations subsequently kept the networks of amateur observers. The new aspect of what is called citizen monitoring is that scientists have seized the initiative and propose protocols well suited to groups of volunteers.

In the Netherlands, the U.K. and English-speaking countries in general, people are very involved in social life, more than in countries with Latin cultures. Many associations fill the role of public services in a vast array of fields. In the naturalist field and for all taxa, it is much easier to mobilise observers than in France. Associations are much stronger in these countries, for example the Royal society for the protection of birds (RSPB) has over a million members whereas the League for the protection of birds (LPO) in France has barely 45 000.

The cultural differences are also manifested in the high degree of confidence that Anglo-Saxon observers place in the protocols and in the collective monitoring project.

In Latin countries, however, observers are wary and ask three questions before committing.

  • Are they really interested in my contribution and not trying to sell something else?
  • Is this a useful project ?
  • Will I be of any use in this project? If other people can do just as well, the (potential) observer will avoid the commitment. That is perhaps one of the reasons why the Garden butterfly observatory (Observatoire des papillons de jardin, OPJ) is so successful compared to the Phenology observatory. In addition to the initial motivation to learn about butterflies, observers have the impression that their garden, a private space, will supply data not available elsewhere, whereas the data for the Seasonal observatory, e.g. the dates of bud break or when leaves fall, do not vary much from one garden to another in a given town.

Marion Gosselin. Does that mean observers are not interested in butterflies outside their garden and thus participation in monitoring does not stimulate their curiosity as naturalists?

Romain Julliard. No. It is the decision to enter or to remain in the monitoring network that is motivated by the capacity to provide original data. But participation in the network probably whets their curiosity as observers to discover and learn about new species of butterflies, outside their garden as well.

Frédéric Gosselin. Is it linked with the notion of heritage?

Romain Julliard. Certainly for the OPJ, but it is not a systematic factor. The most important factor is to feel useful. Another project, which failed, confirmed this intuition that Latin observers need to feel useful before participating. We had noticed that darker urban pigeons better resisted parasites. We wanted to map the distribution of pigeon colours in Paris by calling on amateur photographers in a citizen-science effort that would induce them to see pigeons differently. What could be easier than taking photos of pigeons? However, fewer than 20 persons participated. Most likely because no one felt indispensable, anyone could take such photographs in Paris.

Marion Gosselin. Was the negative image of urban pigeons also a factor?

Romain Julliard. Yes, it was, but not to the point of turning the project into a failure. The issue of "Am I indispensable for this project?" was certainly the main reason. That issue guided us in designing the upcoming photographic monitoring programme for pollinating insects (SPIPOLL). The protocol is to photograph, over a limited time period, species gathering nectar and pollen in a limited sector chosen by the observer, then to sort and perform morphospecies-level identification. The goal is to analyse:

  • the richness and diversity of groups, for example, detect imbalances between Diptera and Hymenoptera as a function of an urban-agricultural gradient;
  • systems of interaction between plants and pollinators (generalist or specialist species).

In the new project, contrary to the OPJ, the notion of heritage is not a factor, the observations can be made anywhere, in a garden or along a road, but we will be very demanding in terms of time and rigour. We think those demands may be the key to success, the observers will feel useful because not everyone can fulfil those demands. Plus, there is a fun aspect to the protocol (photographing, sorting, identifying) that is also an advantage.

Frédéric Gosselin. In other countries, do naturalist associations not hire more scientists to analyse and put to use their observation data, on the condition that they are gathered using a protocol? For example, the British trust for ornithology (BTO)?

Romain Julliard. BTO in the U.K. is a non-governmental organisation that receives considerable public funds. It plays the same role as CRBPO (Bird-ringing research centre) of the National museum of natural history in France, i.e. it coordinates ringing projects, manages all citizen-science monitoring efforts and carries out scientific research, works with universities, etc. In Holland, monitoring management is ensured to a large degree by naturalist associations and data analysis is carried out by a public statistics agency, similar to INSEE (National statistics institute) in France, but with a large environmental department. Given the size of the country, this organisation with a single analysis centre is sufficient.

Frédéric Gosselin. In France, Vigie-Nature is a programme of citizen-monitoring projects coordinated by the MNHN. What is the history of the programme?

Romain Julliard. The MNHN citizen-monitoring projects started with birds. CRBPO, the coordinator of bird ringing in France, always played a role in training and managing a network of volunteers (the ringers) and it was on the basis of this legitimacy and know-how that the STOC (temporal monitoring of common birds, the French breeding bird survey) project was launched in 1989. The STOC listening-stations project gathered a loyal network of observers in addition to those of the STOC ringing project. There are many amateur ornithologists and even with just a small percentage, the number of observers is sufficient. Observers were recruited individually and local associations took over management.

Other citizen-monitoring projects have been set up since 2005, some with naturalists.

  • The butterfly-monitoring project (temporal monitoring of Rhopalocera in France - STERF), based on a network of naturalist observers, an initiative of the president of the Lepidoptera association of France. This is the best monitored group in Europe after birds. The observers in this network are difficult to mobilise, there are currently about 50 observers and 100 sites are monitored. Perhaps because naturalists are, in general, sceptical as to the usefulness of monitoring common species, for which the protocol does not foresee the monitoring of flag species. We are again confronted with the three questions that hinder citizen sciences in Latin countries.
  • The bat observation project, in conjunction with SFEPM (French society for the study and protection of mammals).
  • Vigie-Flore, launched in 2009 as a partnership with the Tela-Botanica association for network management.

Others have been set up with the general public.

  •  The Garden butterfly observatory (OPJ). The association called Noé Conservation had a project called "Butterflies and gardens, linked lives", to raise awareness of amateur gardeners for the environment via butterflies. The MNHN offered to add a citizen-science project. Building on the attractiveness of butterflies, the goal of the project is to gradually expose the observers to a scientific approach, i.e. learn to identify butterflies, learn their ecology (e.g. the existence of migratory butterflies), understand the presence of butterflies as a function of how the garden is managed.
  • The snail project, a public observatory proposed to the OPJ network. The link with gardening practices (mowing, pesticides) is even clearer for snails than for butterflies and snail distribution can vary in a single garden, depending on the local micro-habitats. With snails, it is also possible to have schools participate in the monitoring project because observations are essentially from February to November, contrary to butterflies which are most numerous in the summer, during school vacation.
  • And soon the photographic monitoring programme for pollinating insects (SPIPOLL).

3. How do you select the taxa to be monitored?

Romain Julliard. Selection of the taxa monitored in Vigie-Nature is opportunistic and pragmatic. We monitor taxa for which a network of observers can be easily mobilised and a partnership is possible with an organisation that can take on network management. We do not select a species because it would be a good indicator for the rest of biodiversity. During data analysis, it turns out that some are indicators of changes in other species. But the fact is they are indicators because we monitor them and we have nothing better!

Marion Gosselin. Does citizen monitoring deal exclusively with common species?

Romain Julliard. In Vigie-Nature, yes, we intentionally address ordinary biodiversity for a number of reasons. First, because we are looking for originality in our scientific publications and rare species are heavily monitored by associations and university researchers. Also, because the functioning of ecosystems and the resulting ecosystem services depend to a large extent on common species.

Marion Gosselin. Common species also have the advantage of being easy to observe and the source of abundant data, which in turn means greater statistical power for data analysis.

Romain Julliard. Monitoring is carried out on a species when it is easy to determine or on a group of species, e.g. according to morphological criteria, if determination is difficult.

Marion Gosselin. How can data from a "group of species" be analysed when the species in a group have such different ecologies?

Romain Julliard. That is the case for the "small blue butterflies" in the OPJ, a group comprising highly diverse species. Of course, we cannot analyse the impacts of gardening practices on the group. The data for the "small blues" is used simply to calculate the number of species or the overall abundance of the butterfly community in a garden.

Marion Gosselin. It is thus necessary to be aware of what information the data can provide and what they cannot provide because they are not designed for that purpose.

Romain Julliard. We are aware of that and try to explain it, notably to naturalists who are often frustrated by certain aspects of the protocol, e.g. grouping of species that are difficult to distinguish.

4. Does monitoring elicit a real and durable interest in scientists and the public?

Romain Julliard. We observe both a strong mobilisation on the part of the public and "losses" over time. Each year, 40% of the OPJ observers leave the programme. But they are replaced by new volunteers and the total number of observers is stable from year to year. That implies management work, i.e. recruiting new observers, but also proposing new activities and tools for the project. The long term ("we will have results in ten years") does not motivate observers. What encouraged observers to count birds each year at the same place for the STOC project, as foreseen by the protocol, was the possibility of variations from one year to the next, which made them want to understand what had happened, given that nesting birds are territorial and if they are heard somewhere, it is because there are sufficient resources. It is more difficult to keep botanist observers for Vigie-Flore because there is little chance that flora will change significantly from one year to the next. Once the description has been carried out the first year, people become bored. However, to detect change over time, it is necessary to return to the same place. If the station is changed each year, the differences in floristic distribution would mask any detectable change. The real question is how to motivate observers in this situation.

5. Are there not strong "observer effects" and risk of error if observers are beginners? If so, how does one correct that and analyse the data?

Frédéric Gosselin. To fill out the question, is it possible to calibrate the data from each observer, even integrate species-detection probability by observer for data analysis?

Romain Julliard. For the OPJ, it is the large number of observations that makes it possible to statistically compare situations, e.g. the effect of practices on community diversity or the location between town and country. But the OPJ was not designed to monitor temporal changes in the abundance of butterfly communities on the national level. It was designed for synchronic comparisons. Diachronic use of the data, i.e. the temporal change, is more limited. It is limited to the observation point, the garden, with two sources of bias, i.e. improvement in practices and in the observers, which are linked. Though sources of bias, they are improvements that we encourage because one of the goals of the OPJ is greater awareness. That does create a real problem in interpreting the observed trends. In the end, the contrast between the trends observed by the OPJ and those observed by other networks, e.g. STERF, is more useful than assuming the OPJ is a reference for butterflies in France. The OPJ was not designed to measure the health of butterfly communities in France. Its purpose is to detect the relationship between gardening practices and the diversity of butterfly communities in gardens.

Frédéric Gosselin. But as you noted, there is a bias due to the fact that the more observers are aware of the need for practices favouring butterflies, the more they will pay attention to detecting butterflies. The probability of detection thus differs as a function of the practices and correspondingly biases the analysis of species diversity.

Romain Julliard. We are thinking about integrating the "observer seniority in project" parameter in the analysis, first to correct the bias, secondly because it is interesting to analyse the relationship between observer seniority and declared gardening practices or the frequency of observation. It is already clear that more long-standing observers detect more species than newcomers.

Marion Gosselin. Is there information available on the percentages of error (species not detected, incorrectly identified) in citizen-monitoring projects?

Romain Julliard. We try to estimate the error rate to reassure ourselves, but the estimation is not used to correct the data. For the 2008 OPJ survey, the photos sent in by observers of butterflies gathering nectar revealed an identification error rate of 5%. The rate is identical for false positives (species detected in abnormal places or months).

Marion Gosselin. What are the main contributions of the OPJ data?

Romain Julliard. The results concerning practices are interesting. For example, in the most urban areas, species are less and less abundant. Each species reacts in its own way, i.e. certain species are not affected by the urbanisation gradient, others are never observed in a city centre. Similarly, butterfly species react differently to the naturalness level of gardens (an index based on the declared presence of, e.g. brambles, nettles, pools of water). It is precisely the species the most sensitive to urbanisation that profit the most from good practices that enhance garden naturalness. That result was not obvious at the start of the project.

Marion Gosselin. That is an encouraging result in that the efforts of gardeners in favour of more natural practices produce results. The conservation practices are truly effective.

Romain Julliard. True. The results show that our recommendations are supported by the facts. Good practices produce a real effect only in areas with a low level of urbanisation. In highly urbanised areas, there is no effect. Practices to increase garden naturalness do not succeed in restoring diversified butterfly communities in city centres. We also invited OPJ observers to photograph butterflies while gathering nectar to study the influence of their diet on adult butterflies. We analysed the impact of flower diversity and the proportion of ornamental vs. native flowers on the diversity of the butterflies observed. Surprisingly, it is not the urban butterflies that profit most from ornamental plants. The butterflies that prefer ornamental plants are not more present in cities than in rural areas. For example, the Silver-washed Fritillary is not found in cities, but is attracted by introduced plants such as the buddleia. On the other hand, the Speckled Wood, a very urban butterfly, will often forage native flowers such as daisies. Whether or not the flowers are native does not have any real impact. For that reason, butterflies are good candidates as indicators of connectivity between natural areas in cities. Crucifers, nettle and grasses, that most butterflies consume during the caterpillar stage, are sufficiently abundant, i.e. do not constitute a limiting factor, and whether or not flowers are native is not a decisive factor. So it is the travel possibilities and the structure of the connections between habitat patches that are decisive. What is more, butterflies function as metapopulations with local extinctions caused by parasitoids, which means that exchanges between habitat patches are crucial to ensure the functioning of the metapopulation. For example, butterflies such as the Meadow Brown require a network of close habitat patches. They are absent from gardens too distant from each other in an urban setting, whatever the gardening practices employed.

Frédéric Gosselin. What is the primary factor, the presence of dispersal corridors or rather the quantity of favourable habitat in a given territory?

Romain Julliard. Both the quantity of habitat and the geographic proximity of habitats play a role. That is why we think it will be possible to determine, given the species composition of butterfly communities, whether habitats are connected or not.

Frédéric Gosselin. But keeping in mind that this concerns exclusively butterfly habitat connectivity, i.e. that of open, flowering environments. There can be no question of using butterflies as indicators of connectivity for other environments, e.g. forest environments. That is what makes ecological networks so difficult, there are as many subnetworks as there are different types of habitats. By attempting to create corridors between open environments, we risk fragmenting wooded environments. It is easy to reason in terms of quantities of favourable habitats in a given territory, but very difficult to create a network of connected habitats for different types of environments.

Romain Julliard. Using the results on the relationship between the presence of certain butterfly species and the connectivity of open environments, we are developing a protocol to monitor butterflies in green areas of urban environments, for use by professional gardeners to study the effects of differentiated management in urban parks and gardens.

Marion Gosselin. The OPJ data are not intended to monitor changes in the abundance of butterfly populations in France, but are they used to draw up species distribution maps?

Romain Julliard. That is not in fact the goal. Distribution maps are drawn up, but above all to validate data quality, to make sure that the distribution indicated by the data corresponds to what we know, i.e. there are no abnormal location data. It is a bit risky because we are accused of drawing up distribution maps with inadequate data. For the OPJ, it will be difficult to do more than the analyses on gardening practices and the urbanisation gradient with the count data alone. The next step will be to propose experiments to the observers, e.g. a change in practices. To study changes in parasitoids over the urbanisation gradient, we gave 30 OPJ volunteers in the Paris region a "caterpillar farm" that they must monitor daily, i.e. any changes and the health status, until the chrysalid stage. The results are clear with decreasing parasitism on entering towns and cities. But this is the limit of citizen science in terms of manipulating people, because we make little use of the local knowledge of observers.

Marion Gosselin. Are 30 observation points sufficient in terms of the types of environment?

Romain Julliard. We had selected volunteers whose garden was on the studied gradient, contained crucifers (indispensable for the parasitoids) and a species, the Large White, that we know reproduces in both urban and rural areas. Another effort involving OPJ volunteers required that they find Large White caterpillars to give them to MNHN colleagues studying dispersal behaviour and who needed different population sources. The OPJists responded in greater numbers (27) than the entomologists in naturalist associations, who were also contacted (3 participants).

Marion Gosselin. Did the volunteers who collected the data receive any information on the results of the subsequent, in-depth analyses?

Romain Julliard. Of course! Noé published partial data in its monthly bulletin and an annual summary of the results was made available on line.

6. How can the representativeness of samples be controlled for a given territory, when the observation points depend on where the observers live?

Marion Gosselin. If the goal of monitoring is to detect differences in diversity of species assemblages (a function of gardening practices for OPJ or depending on the type of environment, e.g. gardens, along roads or rail tracks, meadows, for the pollinator observatory), we need a balanced sample over the studied gradient, i.e. the garden-practices gradient or the urbanisation gradient. How can the representativeness of samples be controlled when the observation points depend on where the observers live (situation for the STOC) or on the willingness of the observers (situation for the pollinator observatory)?

Frédéric Gosselin. The same questions arise if the monitoring goal is to assess public policies, e.g. monitoring of species of EU interest to assess the effectiveness of the Natura 2000 network. The sampling technique must take the goal into account. Is there a type of post-stratification? Over which gradient(s)?

Romain Julliard. For the OPJ, there is post-stratification. We take advantage of the great number of observation points to create gradients a posteriori based on the urbanisation level where the gardens are located, regional differences and on the declared practices in the gardens (mowing, use of pesticides, composting, etc.). Concerning this last point, there is a bias because observer quality is certainly correlated with the person's practices, i.e. the most attentive being a priori those taking the most care in their garden. Then, similar to the STOC, we analyse how the abundance and diversity of the observed species vary as a function of the environmental gradient. For the STOC, the representativeness of samples is ensured because the observation points are indeed near the homes of the observers, but selected randomly in the neighbourhood.

Frédéric Gosselin. Random selection is important, but does not guarantee that the results are representative of the situation in France. Certain regions are clearly undersampled because the observation-point density in each region depends on the geographic distribution of the voluntary observers. How can the differences in sampling effort be corrected between regions?

Romain Julliard. Indeed. For example, the French regions Aquitaine and Centre have fewer observation points than other regions. But we can assess representativeness and take it into account. In the current analysis, we assume that observation-point density is evenly spread over the country, but we could also analyse the results region by region, weighting them according to the size of each region to calculate a national index. For the time being, we do not do it. There are two statistical rationales when calculating an average. Either we assume that each point is measuring the same thing, with a variance, in which case there is no use in weighting by the number of points in each region, or we assume that there are strong regional variations and that the observation points in different regions are not measuring the same thing, in which case weighting is necessary. For monitoring on the national level, we assume that the studied phenomenon is the same in all regions.

Frédéric Gosselin. Yet the National forest inventory (IFN), in assessing forest resources nationally, specifies for each plot the probabilistic weight according to which it was drawn at random and that is used in the analyses. They assume that there are differences between regions. Similarly, in 1994 in the U.K., ornithologists changed their observation protocol which previously focussed on the southern part of the country. But the results after 1994 revealed that changes in the North and South differed, which indicated that the prior sampling, which was not spread evenly over the country, was not representative.

7. Would it be possible to set up mixed networks, professional and amateur, where the professionals would fill in the holes in the sampling network?

Romain Julliard. That is already the case when entities want to set up STOC-compatible monitoring in a particular territory where they operate. For example, regional councils fund associations to fill out the STOC sampling network in view of setting up a regional observatory, or the National forestry agency asks its agents experienced in ornithology to set up STOC observation points and thus provide monitoring for public forests.

Marion Gosselin. But these additions are random, they depend on the willingness of local entities. There is no real organisation to improve sample representativeness on the national level.

Romain Julliard. The national parks use a protocol suited to mountain regions, where the points are selected randomly among accessible zones and systematically spread around the upper tree line.

Frédéric Gosselin. Are these points used in analyses in spite of the differences in protocols?

Romain Julliard. Yes, because the points are selected randomly and these are the only observations in mountain areas. For common birds, the risk of a bias is low. On the other hand, we do not use the listening stations in French nature reserves (RNF) because their protocol does not include random selection. They select ten points in each reserve. In other monitoring projects, sampling is based on an itinerary along roads. For bats, the protocol calls for continuous recording of a sonogram from a vehicle at night, driving slowly along a set itinerary. No special skills are required, people must simply learn to read the sonogram for a limited number of species, less than 15. The risk of error is low and it is always possible to consult the recording and correct a datum. Typical observers are either bat enthusiasts who want to learn more and raise awareness for bats, or local governments attracted by the possibility of rapidly obtaining maps and comparing the positions of bats with zoning maps, for example. In addition, this method detects bats on the hunt, i.e. the data are also indicative of the abundance of insects, the "aerial plankton" that they eat.

Frédéric Gosselin. What is the opinion in the U.S. of their bird-monitoring projects, which are also based on sampling along roads? The sample is probably not very representative of the diversity of environments where birds are found.

Romain Julliard. That is a real problem in the U.S. because roads are not distributed evenly across the territory. In nature zones, road density is very low and the environment along roads has been modified. We do not have this problem in rural France, except in mountain regions.

Frédéric Gosselin. But do itineraries for bats traverse forest regions? We know that forests play an important role for bats, both for roosting and for hunting.

Romain Julliard. The itineraries generally follow secondary roads and forest tracks if they are open. That being said, the protocol for sonogram detection is for bats in open environments that emit strong ultrasounds, not forest bats (Rhinolophus, Plecotus) that are detected only sporadically.

Frédéric Gosselin. Just as STOC does not address all birds in France, but only common, diurnal, nesting birds that can be detected by listening stations during the reproduction season.

Romain Julliard. It is for this reason that we have difficulties in setting up an amphibian observatory because there are no protocols making it possible to monitor in a simple manner all amphibian species. We would need to limit monitoring to groups with a specific protocol.

Frédéric Gosselin. The same problem exists for mammals. The National agency for hunting and wildlife (ONCFS) does monitoring, but there is a protocol for virtually each species.

8. Or would a "professional" network be possible to assess the effectiveness of public policies for biodiversity protection?

Marion Gosselin. We have seen that monitored taxa are selected opportunistically and not because there is a political will on the national level to monitor taxa because they are representative of biodiversity or because they are threatened in the current context of biodiversity erosion due to anthropogenic disturbances.

Frédéric Gosselin. But there is ambiguity in the public debates and in how the public agencies use monitoring projects. They use the data, because they are the only data available, letting people think, and perhaps thinking themselves, that the projects were designed from the start to monitor changes in biodiversity in France.

Marion Gosselin. With the risk that politicians will be satisfied, thinking that national biodiversity monitoring exists in France, when in fact many taxa and types of environment are not covered by current projects.

Romain Julliard. Yes. But we must remain realistic. An ideal protocol does not exist to monitor even a single taxonomic group such as birds on the scale of a country with such varied environments as France, there is not even one protocol that is better than the others. In addition, pragmatic aspects often outweigh the rest. For example, it would certainly be worthwhile to have a national monitoring programme for earthworms, given their importance for ecosystem functioning, but practically speaking, it is very difficult.

Marion Gosselin. Is there a chance of seeing politicians provide serious funding for a "professional" network?

Romain Julliard. I do not know. The other not very logical aspect of existing monitoring systems is the preponderance of birds. We are now working with the Agriculture ministry on a biodiversity observatory for farms. Birds would appear to be an unavoidable taxon in the observatory even though they are not organisms whose home range is suited to analysis on the scale of a farm. It would be more rational to work on taxa whose home range coincides with the zones where farm work is carried out. But their inclusion is desired because there is a European directive on birds and birds are used in the models for the farm-product approval process.

Frédéric Gosselin. Is that due to a lack of ecological culture on the part of decision-makers or a lack of willpower? On the other hand, in forestry, decision-makers generally have forestry training, e.g. in dendrometrics, and they need information on available resources, which is why they created the National forest inventory, a solid database for statistical monitoring of forest resources.

Romain Julliard. In addition, decision-makers are certainly happy to have the data produced by the monitoring programmes, but very few operational decisions are taken on the basis of the results for conservation policies.

Frédéric Gosselin. Were results, such as STOC, not used to revise the agri-environmental aid programme for the Common agricultural policy?

Romain Julliard. There is still a long road before we can modify agricultural policy or grant aid according to the results obtained on the status of biodiversity.

Marion Gosselin. Again, we need sampling designed to assess measures with observation points correctly positioned over a gradient of the practices to be evaluated.

Romain Julliard. A major issue is required to mobilise scientists for monitoring. If we ask farmers to significantly change their practices to help biodiversity, agricultural decision-makers will react. We saw the same thing when climatologists proved that climate change is real and blamed fossil fuels, the oil companies funded scientists for monitoring, hoping to prove the contrary.

Frédéric Gosselin. There are cases where rigorous monitoring was set up in conjunction with management decisions, e.g. for the spotted owl in the U.S.

9. Is monitoring seen as a major issue by decision-makers and society in general?

Romain Julliard. We must admit that biodiversity is not perceived by society as a major issue. Scientists carry out monitoring and their results are not contested. If biodiversity were to become an issue because it questioned the practices of certain groups, we would enter a more turbulent period.

Frédéric Gosselin. But there are laws and national commitments to protect biodiversity…

Romain Julliard. For now, the laws concern primarily protected species or zones.

Marion Gosselin. And we have not come up with the means to assess their effectiveness.

Romain Julliard. In the CNPN (National council for the protection of nature) commission, for example, we study impact-study files and compensation proposals for development projects (e.g. a road) and it is striking to see that there is never an assessment a posteriori of the measures taken (the compensation work). Checks are run simply to make sure the work was done, but assessments on the effectiveness are very rare and almost never contribute to our knowledge.

Marion Gosselin. France must periodically monitor conservation of the habitats (and of the species that live there) protected by the Natura 2000 network. Are there any plans to use Vigie-Nature data to that end?

Romain Julliard. Not for the time being, for organisational reasons. The absence of a shared protocol for all sites led us to align our efforts on the sites with the least data in order to merge and analyse the results.

Marion Gosselin. Even if Vigie-Nature was not originally designed to assess the Natura 2000 network, could some of its observation points not be included in the sampling project set up for the assessment?

Romain Julliard. Yes, the Vigie-Nature points could be used as reference points outside the Natura 2000 sites.

10. Are the protocols consistent from one country to another?

Romain Julliard. In order to share the data from citizen monitoring, we obviously work a great deal with other countries. In Europe, there is a highly structured network for birds and for butterflies. The protocols have not been standardised by the countries, but they are compatible. For butterflies, they are in fact standardised, but much less so for birds, e.g. transect lines in the U.K., listening stations in France.

Frédéric Gosselin. As far as I am aware, the countries share the estimators of the overall trends observed, similar to a meta-analysis based exclusively on published average data, but not the individual observation data that could be used for more in-depth analysis.

Romain Julliard. That is the case due to the available resources for database entry and export that differ from one country to another. The more in-depth analysis that you mentioned takes place case by case, when opportunities for collaboration present themselves. We have carried out analyses on birds and butterflies with five or six European countries, with which we had a shared measurement parameter on all sites, but that would not have been possible with the entire EU. Collaboration also depends on the network leaders. For butterflies, the coordinator comes from the association sector and federates the research, which means all analysis is done by researchers from the same work group. For birds, the coordination is financed by RSPB with statistical support from Statistics Netherlands. That creates tension because the research team that manages the data on the European level is in competition with the research teams managing the national data. The problem is that the coordinators are also involved in research. The ideal situation is the sharing of know-how among the national centres to produce data for collective research on the European level, rather than a European research centre managing national data collection.

Frédéric Gosselin. What links can be established between biodiversity data and the ecological variables that can explain the observed variations? I ask the question in light of the situation with the spotted owl. No efforts were made to ensure that all the monitoring sites for the owl population include standardised habitat measurements to see if it is effectively the quantity of old forests that determines owl populations. Yet two or three limited studies addressed the issue of the favourable habitat quantity and questioned the validity of a linear relation between the favourable habitat quantity and owl populations. In which case, management efforts to mix very old and younger forests could be optimal for owls, whereas today, current management targets exclusively the development of very old forests. The problem is that because habitats and population are not monitored in a coordinated manner, we do not know if the non-linear relation between the quantity of old forests and owl populations is valid for all the monitored zones or if it is a strong relation. And the result is the spotted owl continues to decline without any clear idea of the cause. I wonder if there are not similar issues that could be improved by coordinating the data between, e.g., IFN and MNHN or between Météo-France and MNHN.

Romain Julliard. Our own habitat descriptions exist, thankfully, but are managed by observers with a possible observer effect. However, a recent partnership with Teruti (annual land-use survey by the Agriculture ministry) made it possible to produce several scientific publications because the Teruti data provide truly novel data-analysis possibilities for changes in land use. We also often use the Corine Land Cover and the data from the Paris-region land-use MOS atlas to describe the habitats around each observation point and analyse the relations between these parameters and the biodiversity data. But this type of project works only if both partners really want to work together, the point is not for MNHN simply to purchase the data-usage rights. There has to be real collaboration, which requires time and effort to align the two databases, extract the useful data, use the unprocessed data to calculate the most relevant variables for analysis and for which the best recommendations can be made by the experts of each database. We are now working with the Agriculture ministry on a biodiversity observatory for agricultural areas in view of studying the effects of agricultural practices on biodiversity. The main difficulty will be to manage a database on changes in practices in parallel with the database for species observations.

Marion Gosselin. Teruti can supply data on standing crops, but information on the practices used on a plot of land requires farmers willing to participate in the observatory.

Romain Julliard. The difficulty increases with the possible consequences of the issues addressed by the observatory in terms of imposing general changes in practices.

11. What are the links and the differences between citizen monitoring and SINP (information system on nature and landscapes)?

Romain Julliard. The goal of SINP is above all to list and clarify the inventory data, presenting the metadata (description, size of data region/plot, data precision, accessibility, contact person), whereas Vigie-Nature is a system of structured data in view of analysis for monitoring purposes or to assess the effects of a practices gradient on the diversity of a given taxonomic group. Even though SINP comprises both monitoring and inventory data, efforts are made primarily on the inventory data, for which descriptions are lacking, whereas there are no problems for the monitoring data.

Marion Gosselin. Is the end goal to analyse all the listed and aligned data?

Romain Julliard. Theoretically, yes. But it is doubtful that scientists will be willing to analyse data if they were not involved in collecting or listing it.

Frédéric Gosselin. Why? Our teams at Cemagref use part of the IFN data without too many problems.

Romain Julliard. Those data are very standardised and, in fact, quite simple. For example, Onema (National agency for water and aquatic environments) has difficulties publishing its data on electric fishing even though the data are standardised and of better quality than the STOC data. They go to great efforts to build and maintain the database, yet, in comparison, they lack funding to find people to analyse the data. The strength of Vigie-Nature is its active database that our scientists know very well and that can be used to publish new results each year. We waste no time finding a database, becoming familiar with it. And after analysis and publication of the results, we are not obliged to start over with a new, external database. The presence of a permanent analysis team and the very strong partnership with the data producers ensures great efficiency in data analysis and publication of the results.

Marion Gosselin. That partnership is probably missing in SINP, which simply lists, for information documents, the inventories and monitoring projects that exist, but that all have different protocols. That is useful as a catalogue of metadata, in the same way that the catalogue of metadata on forest information sources (Ca-SIF), developed by the Ecofor professional group (GIP Ecofor), is useful. But even if SINP did provide direct access to the data, I wonder how the data from different protocols could be combined and used.

Frédéric Gosselin. It may be possible, but not without a permanent team to analyse the data. In addition, it would be worthwhile to put some thought into data quality. For example, why list databases on species presence, if this type of data does not provide information that can be used for statistical processing?

Romain Julliard. It would be a good idea to design a system where research is an integral part of data collection.

Frédéric Gosselin. In this sense, we hope that the National observatory for biodiversity, planned by the Ecology ministry, will be different than SINP, even if the current idea is to set up a network of data producers and analysts rather than put together a new monitoring system, similar to the National forest inventory that monitors forest resources, for example.

12. The Research ministry and CNRS (National scientific research centre) are thinking about observatories for environmental research. Could they be used to assess public policies for biodiversity conservation?

Frédéric Gosselin. It would be unwise to let people think that these observatories on research could be used as operational monitoring observatories, whose results would serve to make decisions on management or policies (e.g. forestry or agriculture). In my opinion, the link between the two is far from automatic and merits ample, prior study. I do not agree that the changes in biodiversity observed on LTER (Long-term experimental research) monitoring sites, which are designed to study precise biological mechanisms, can be used to assess the effects of public biodiversity policies.

Romain Julliard. We need other levels than the LTER sites. The current approach is too focussed on opposing the various systems (remote sensing, inventories, LTER sites) whereas we must enhance compatibility between them.

Frédéric Gosselin. For the National observatory for biodiversity project, the Ecology ministry is attempting precisely to improve compatibility between the existing systems and that is a good thing. However, data on certain sectors is lacking and monitoring systems must be created for those sectors. It is not enough simply to establish a network for what already exists. For example, we lack data to evaluate the role of forests in biodiversity because, for instance, citizen monitoring of bats is not suited to forest bats. We have data on forest flora, but that group is not the most original part of forests nor the most threatened. On the other hand, we have almost nothing on biodiversity in forests   and are, in some cases, threatened. The common view in France that forests and their biodiversity are not threatened does not help efforts to create an observatory for forest biodiversity, which in itself is a complex affair.

13. What is the situation for data ownership?

Romain Julliard. In the inventories, data quality depends entirely on the observer, which is why observers feel very possessive about "their" data and that is a real problem. In monitoring projects, that is not a problem because a datum by itself has no value. Its quality and value depend on the protocol with which it was created and the database where it is located. There is no issue of "ownership" for the basic data. It is the database as a whole that has value.

14. Do we have too many or too few data to inform the public and assess public policies? And what can be said about the commitment/goal to halt biodiversity erosion by 2010?

Frédéric Gosselin. What is your position on the frequent declarations that we have enough data to assess biodiversity status and set guidelines for public protection policies? My question concerns monitoring in general in France, not only citizen monitoring. Personally, I think we lack data to assess biodiversity status and trends. If we compare with climate change, it is clear that meteorological data are easier to collect and have long been very numerous and precise.

Romain Julliard. I have rarely heard it said that we have enough or too many data. As for us at MNHN, we create observatories and new systems in order to produce new scientific publications. It is important to note that we are not funded to carry out monitoring. We are funded because we do it, but not in order to do it. But in as much as the monitoring projects exist, the State administrations are satisfied.

Marion Gosselin. And not all observatories are intended to monitor biodiversity over time. Some are designed simply to explain the current distribution of richness or diversity of the observed species.

Romain Julliard. Each new research question addressed using our observatory data throws new light on issues and changes our perspective on what we know about biodiversity, with good ("there are still some places with many species") and bad ("there are even less than we thought") surprises.

15. In conclusion…

Marion Gosselin. I draw a frustrating conclusion from this discussion. I am aware that we cannot monitor everything for biodiversity. But I am sorry to see that current monitoring is guided by purely pragmatic criteria (we monitor only those taxa for which volunteers are willing to provide inexpensive and abundant data) and that there is no political will to invest in the important taxa given the current context of biodiversity erosion, i.e. taxa that are not necessarily notable or noticed, but are nonetheless threatened.

Frédéric Gosselin. Is the situation different in other countries?

Romain Julliard. One example is the U.K. where the general monitoring system is well structured.

Frédéric Gosselin. In France, biodiversity research is oriented by a movement heavily invested in modelling. Their scientific quality is certain, but I doubt whether that movement is of any great use in guiding the formulation and assessment of biodiversity policies. Another impression drawn from this discussion is that of spheres (political, scientific, the associations) that are too disjoint in the field of biodiversity, whereas they work together well in other fields.

Romain Julliard. In North America, for example, with the adaptive-management movement, management and research are much more integrated in the field of conservation biology. But that did not occur overnight, it took 20 years to reach that result. We cannot improvise with adaptive management in France because we do not have that culture of joint scientific work. The pragmatic solution that might be suitable for the European context is more in the style of evidence-based conservation, that the English are developing by trying to group the management services to gain knowledge and not lose the information drawn from management experiences. That corresponds better to our work habits, but must still be set up, taking into account the fact that the conservation-biology community in France is very small.

Pour citer cet article :

Référence électronique :
GOSSELIN, Marion ; GOSSELIN, Frédéric ; JULLIARD, Romain, Interview : Value and limits of citizen science in biodiversity monitoring, Revue Science Eaux & Territoires, Public policy and biodiversity, numéro 03bis, 2011, p. 76-82, 15/03/2011. Disponible en ligne sur <URL :> (consulté le 28/11/2021), DOI : 10.14758/SET-REVUE.2011.3BIS.15.

Ajouter un commentaire

Télécharger au format PDF
↑ Haut de page