European database of the life-history, morphological and habitat characteristics of dragonflies (Odonata)

Climate change and an ever-increasing effect of humans on the great majority of freshwater ecosystems have affected not only individual organisms but also the links between them. In order to effectively protect communities or entire freshwater ecosystems, we need to understand how individual organisms react to these changes and the functional interconnections between individual species. This can be achieved by evaluating the functioning of individual species, i.e. their traits. Dragonfl ies and damselfl ies are increasingly used as indicators of changes in freshwater ecosystems. Although this taxonomic group is the subject of many ecological studies, there is no comprehensive database of the traits of European species. This is, why we decided to lay the foundations for an European database of dragonfl y traits called dragonfl y-database.eu, which is mainly based on Web of Science (WOS) publications and several monographs, and currently includes information on 79 European species. The main advantage of our database is that it is open to downloading, uploading, and storing of data. This is extremely important, because a database’s applicability increases with the quality of the data it contains. Initially in compiling the database, we encountered several fundamental shortcomings in the knowledge of some often easily measurable traits in, some species. We believe that in combination with new statistical methods dragonfl y-database.eu will become a very useful tool for many ecological studies and more effective conservation of threatened species.


The need to study functional diversity
There are no doubts that Europe as well as the whole World face an unprecedented decline in insect diversity (Thomas et al., 2004;Potts et al., 2010;Hallmann et al., 2017). A "deadly cocktail" of water pollution, degradation of water habitats, and various phenomena associated with climate change has affected many freshwater ecosystems much more than most terrestrial ecosystems (Sala, 2000;Dudgeon et al., 2006). This trend has been successfully reversed in many countries, but a comprehensive approach is needed for long-term management (Termaat et al., 2015). Eliminating individual negative infl uences will be diffi cult, because their effects are strongly correlated and related to various anthropogenic infl uences (Clausnitzer et al., 2009;Kalkman et al., 2010). In addition, studies on endangered species of dragonfl y indicate that the main threats to individual species differ markedly (Dolný et al., 2013;Harabiš & Dolný, 2015). Many scientists, therefore, are moving away from the idea of protecting individual species and redirecting their efforts to optimizing the management of larger groups of species with similar habitat requirements (Levine, 2015).
One of the merits of our database is its comprehensiveness. It is not just the number of traits included that is important, but above all is it summarizes information on all the life stages. This is crucial, because larvae are much more closely associated with particular habitats than the adults, which have considerable ability for dispersal. This is particularly important for conservation biology, for which studies unfortunately only record the ecological requirements of one life stage, typically the imago, and cannot account for why a species or population is struggling (see a critical review of this subject by Stoks & Córdoba-Aguilar, 2012). Data on larval life histories are relatively rare in published studies, yet we believe that such data exists, but diffi cult to fi nd in primary sources or is not published. That is why we decided to create an open database with an option for uploading data.

Technical details
The dragonfl y-database.eu provides its users with comprehensive data functions, which enable them to view, export, share and edit data on dragonfl y traits. The main data application features a web interface built using such basic web technologies as HTML (HTML5), JavaScript (jQuery framework), and CSS (CSS3). On the web interface (client side), all data are represented by JSON objects. These are dynamically updated by AJAX calls to the server and displayed in a structured table. JSON objects are based on user-selected criteria and are prepared and composed at the server. The back-end (server side) is operated by PHP (PHP 7), which acquires data directly from the MySQL database.

User registration
There is no requirement to log in or register in order to view and export data from dragonfl y-database.eu. If one intends to upload new data, a new trait, or even a new species, however, that person must register and log in for the purpose of identifying and verifying the data provided. In order to preserve the identity of each record, the record will be identifi ed by "Contributor". The registration form requests a basic user's contact data, such as name and surname, email address and institutional affi liation. User data will be neither provided to third parties nor used for any other purpose. Registered users can view, edit, export and upload data. The administrators of dragonfl y-database.eu will be notifi ed about data modifi cations and verify any changes. Administrators reserve the right to contact the data provider and, in case of doubt about the accuracy of data, to not include it in the database. Only after approval will data become available to all database users.

Data search and export
A search form allows users to search for data according to taxonomic classifi cation (e.g. family, genus, or species). Therefore, a user needs to specify at least one taxonomic group and information on traits of traits or even one specifi c trait). Filtered data on odonate traits can be viewed online or exported to a CSV fi le.

Content of dragonfl y-database.eu
The current version includes information on 12 morphological, 27 life history, 6 conservation, 8 habitat features for imagoes and 7 morphological and 10 life history features for larvae (total of 70 features, Table 1, Appendix 1) of 79 Central European species. Unfortunately, information on the majority of these features are missing for many species (Table 2).
Generally, the diversity of the Odonata is infl uenced by historical (geological) and ecological factors. The greatest species diversity occurs in tropical regions. This is prob-from cell to the whole-organism level, without any reference to the environment or any other level of organization. In the fi eld of ecology, the term "functional trait" is increasingly used, but with widely different meanings (Violle et al., 2007). Similarly, ambiguous are the defi nition and measurement of individual functional traits. The fact that various authors use the same term to represent different features has led several authors to compile a clear manual for the uniform measurement of functional traits for different taxa (Cornelissen et al., 2003;Moretti et al., 2017).
Surprisingly, compared to a wide variety of other aquatic organisms, information on odonates in hydro biological databases is either very basic or completely lacking (Schmidt-Kloiber & Hering, 2015). To our knowledge, there are currently only two databases summarizing the traits of dragonfl ies and damselfl ies: the Morphological and geographical traits of the British Odonata (Powney et al., 2014) and The odonate phenotypic database, a new open data resource for comparative studies on an old insect order (Waller et al., 2019). Unfortunately, the practical use of these databases is severely limited because they are based on just a few sources and contain only the traits of imagoes. Thus, such databases are not comprehensive, and a more complex database is needed.

Database of the traits of European Odonata
It is not possible to produce a database comparable in range or quality of those currently available for plant traits (Kleyer et al., 2008;Kattge et al., 2011). Fortunately, there are trait databases for groups similar to odonates (dragonfl ies + damselfl ies), but we looked not to a database of traits for butterfl ies or solitary bees, but rather found inspiration in the trait database for European amphibians (Trochet et al., 2014). This might at fi rst seem surprising, but the ecological requirements of odonates are more similar to those of amphibians than of insect pollinators, such as, butterfl ies. Another very important source of information and inspiration are books that contain information on many traits. The "extraction" of this information is sometimes diffi cult, because individual (unrelated) values and/or data are often incorporated in these texts, but it can be potentially valuable, because the authors also present data from sources that are diffi cult to trace and not in internet databases.
We have begun, therefore, to build a completely new database, which currently includes information on the traits of Central European species and in future, hopefully, all species occurring in Europe. This database is primarily based on information in books and scientifi c articles (WOS, Google Scholar, Scopus). Our goal is to create an open database that will expand with addition of information from new and unpublished studies. We aim to record also valuable information associated with individual traits, including basic information (e.g. number of individuals, temperature, ex situ versus in situ) and variability in the data. It is clear, that average values often do not provide meaningful information about a specifi c trait. Therefore, our database provides the mean value for a given trait and its variability. For example, the percentage of traits recorded for each species is listed in Table 2.
Any additional information not only increases the accuracy of an estimate, but also maintains its uniqueness. With the increase in use of various advanced meta-analytical tools, the importance of such information will continue to increase.
ably because a greater diversity of aquatic habitats exist in the tropics and provide a more diverse range of niches. Moreover, if there are some regional refugia, they are mostly still better protected and preserved there than in other areas of the world (Kalkman et al., 2008;Dijkstra et al., 2014). Our database is presently focused on the odonatofauna of Central Europe (79 species, Table 2). Interestingly, the entire Palearctic region is relatively poor in terms of species richness compared to other regions at the same latitude. For example the entire European continent has only a few more species than the state of Texas on the other side of the world (Kalkman et al., 2018). The main causes of this disparity probably include the effect of glaciation in the Pleistocene, when diversity decreased by two thirds and Europe's several geological barriers (Alps and Mediterranean Sea) that limit further expansion of species into central parts of Europe or Asia (Kalkman et al., 2008(Kalkman et al., , 2018.

The invaluable value of ordinary data
In initially compiling the database, we realized that many authors not only do not publish their raw data or even provide essential information necessary for the interpretation of their results. To our surprise, we were unable to fi nd basic information on the numbers of individuals upon which a given value of a morphological feature is based. Especially the authors of books, often take information from other sources quite uncritically or do not mention the source of the information. Many of the values cited for basic morphological features are from other sources, for example it was not possible to trace whether these values (or range of values) are based on the measurement of 1000, 50 or even just 5 individuals. Also often unstated is which dimension is considered when measuring the length of a wing, or whether the values are for individuals from Central Europe, Asia, or elsewhere.
We are increasingly confronted with new indices, methods of estimation and measurements and defi nitions for individual traits. Meanwhile, we continue to struggle with the fact, that basic morphological features are not yet clearly defi ned, or standardized. A fi rst and truly fundamental step was taken recently in the form a handbook on the protocols for measuring such features by Moretti et al. (2017). Here, they clearly defi ne a broad scale of features and standardized procedures for sampling them, which should be one of the fi rst steps in organized sampling for collecting data. However, there remains the problem of the motivation for providing such data. Authors simply do not realize that information on individual features has little value, but by combining it with other information to create a larger whole, greatly increases its value. For the resolution of many ecological issues we need large amounts of data or measurements. Often, it is very time consuming, if not impossible, to obtain such datasets. Many authors try to include as little information as possible in their publications and provide only the results of statistical tests or some basic plots. To some extent, this behaviour forces us to quote only the conclusions of studies without any knowledge of the primary data. We understand that scientists do so in good faith and the expectation that their data will be of use to others. Unfortunately, such behaviour does not contribute much to the general knowledge about a given problem.
Another diffi culty we encountered is a critical lack of information on some species while, conversely, some other species have in the past received considerable attention. This is mainly because authors tend to choose species that are easy to rear in the laboratory. For example, the damselfl y species Lestes sponsa and Ischnura elegans are typical examples of such species in having relatively short life cycles and being relatively easily reared in a laboratory. For that reason, we have a lot of information about these species. On the other hand, there are many common species, such as Orthetrum cancellatum, which has not received such attention, in which case our database only includes very general information on the morphological variability, phylogeny and ecological requirements of this species.

New methods = new opportunities
We have already mentioned meta-analytical methods, which many scientists perceive to be very powerful tools for solving very complex issues (Gurevitch et al., 2018). The greatest weakness of these methods lies in the fact that, however, sophisticated the statistical methods they are very dependent on the quality of the data (Koricheva & Gurevitch, 2014). Also very promising is the use of traits as one of the explanatory "blocks" of variables in modelbased approaches for analysing community data (Warton et al., 2015;Ovaskainen et al., 2017). None of these methods, however, can substitute for defi ciencies in primary data. Therefore, we intend to continue collecting data in order to enhance the information on morphological variability and life history traits of individual species.
Another aim is to include in the database other species of dragonfl ies and damselfl ies occurring in other regions of Europe. We greatly appreciate those studies on the differences in life history traits of individual species like those of De Block & Stoks (2004), De Block et al. (2008 and Schiel & Buchwald (2015a, b). We believe that more of these complex studies will help us to better understand such understudied phenomena as carry-over effects (Stoks & Córdoba-Aguilar, 2012). Clearly much remains to be done to improve our understanding of European odonate species and their ecology. We believe dragonfl y-database. eu can be an important step in that direction by improving the availability of information. We are very thankful to our colleagues within this fi eld for their support of this endeavour and look forward to receiving further data.