DISPERSE, a trait database to assess the dispersal potential of European aquatic macroinvertebrates

Dispersal is an essential process in population and community dynamics, but is difficult to measure in the field. In freshwater ecosystems, information on biological traits related to organisms’ morphology, life history and behaviour provides useful dispersal proxies, but information remains scattered or unpublished for many taxa. We compiled information on multiple dispersal-related biological traits of European aquatic macroinvertebrates in a unique resource, the DISPERSE database. DISPERSE includes nine dispersal-related traits subdivided into 39 trait categories for 480 taxa, including Annelida, Mollusca, Platyhelminthes, and Arthropoda such as Crustacea and Insecta, generally at the genus level. Information within DISPERSE can be used to address fundamental research questions in metapopulation ecology, metacommunity ecology, macroecology and evolutionary ecology. Information on dispersal proxies can be applied to improve predictions of ecological responses to global change, and to inform improvements to biomonitoring, conservation and management strategies. The diverse sources used in DISPERSE complement existing trait databases by providing new information on dispersal traits, most of which would not otherwise be accessible to the scientific community.


Background & Summary
Dispersal is a fundamental ecological process that affects the organization of biological diversity at multiple temporal and spatial scales 1,2 . Dispersal strongly influences metapopulation and metacommunity dynamics through the movement of individuals and species, respectively 3 . A better understanding of dispersal processes can inform biodiversity management practices 4,5 . However, dispersal is difficult to measure directly, particularly for small organisms, including most invertebrates 6 . Typically, dispersal is measured for single species 7,8 or combinations of few species within one taxonomic group [9][10][11] using methods based on mark and recapture, stable isotopes, or population genetics 5,12 . Such methods can directly assess dispersal events but are expensive, time-consuming, and thus impractical for studies conducted at the community level or at large spatial scales. In this context, taxon-specific biological traits represent a cost-effective alternative that may serve as proxies for dispersal 5,6,13,14 . These traits interact with landscape structure to determine patterns of effective dispersal 15,16 .
Aquatic macroinvertebrates inhabiting freshwater ecosystems include taxa with diverse dispersal modes and abilities (Fig. 1). For species with complex life cycles, such as some insects, this diversity is enhanced by life stages with different dispersal strategies. For example, aquatic juveniles of many insects disperse actively and/or passively in water whereas adults fly over land 17 . In all cases, dispersal is affected by multiple traits relating to the morphology 6,12 , life history and behaviour 2 of different life stages.
We compiled and the harmonized information on dispersal-related traits of freshwater macroinvertebrates from across Europe, including both aquatic and aerial (i.e. flying) stages. Although information on some dispersal-related traits such as body size, reproduction, locomotion and dispersal mode is available in online databases for European [18][19][20] and North American taxa 21 , other relevant information is scattered across published literature and unpublished data. Informed by the input of 19 experts, we built a comprehensive database containing nine dispersal-related traits subdivided into 39 trait categories for 480 European taxa. Dispersal-related traits were selected and their trait categories fuzzy-coded 22 following an approach comparable to that used to develop existing databases 23 . Our aim was to provide a single resource facilitating the incorporation of dispersal into ecological research, and to create the basis for a global dispersal database.

Methods
Dispersal-related trait selection criteria. We defined dispersal as the unidirectional movement of individuals from one location to another 1 , assuming that population-level dispersal rates depend on both the number of dispersing propagules and dispersers' ability to move across a landscape 11,24 .
We selected nine dispersal-related morphological, behavioural and life-history traits (Online-only Table 1). Selected morphological traits were maximum body size, female wing length and wing pair type, the latter two relating only to flying adult insects. Maximum body size influences organisms' dispersal 6 , especially for active dispersers 25 , with larger animals more capable of active dispersal over longer distances (e.g. flying adult dragonflies 6 , Fig. 1). Wing morphology, and in particular wing length, is related to the dispersal of flying adult insects 6,26 . Female wing length was selected because females connect and sustain populations through oviposition, thus representing adult insects' colonization capacity 27 . Females with larger wings are likely to oviposit farther from their source population 6,10,28 . We also described insect wing morphology as wing pair types, i.e. one or two pairs of wings, and the presence of halters, elytra or hemielytra, or small hind wings 12 (Fig. 1). Selected life-history traits were adult life span, life-cycle duration, annual number of reproductive cycles and lifelong fecundity. Adult life span and life-cycle duration respectively reflect the adult (i.e. reproductive) and total life duration, with longer-lived animals typically having more dispersal opportunities 13 . The annual number of reproductive cycles and lifelong fecundity assess dispersal capacity based on potential propagule production, with multiple reproductive cycles and abundant eggs typically increasing the number of dispersal events 6 . Dispersal behaviour was represented by a taxon's predominant dispersal mode (passive and/or active, aquatic and/or aerial), and by its propensity to drift, which indicates the frequency of flow-mediated passive downstream dispersal events.

Data acquisition and compilation. A taxa list was generated based on the taxonomies used in existing
European aquatic invertebrate databases 18,20 . Trait information was sourced primarily from the literature using Google Scholar searches of keywords including trait names, synonyms and taxon names (Supplementary File 1, Table S1), and by searching in existing databases 18,21 . Altogether, >300 peer-reviewed articles and book chapters were consulted. When no European studies were available, we considered information from other continents only if experts considered traits as comparable across regions. When published information was lacking, traits were coded based on authors' expert knowledge and direct measurements. Specifically, for 139 species in 69 genera of Coleoptera and Heteroptera, female wing lengths were characterized using measurements of 538 individuals in experts' reference collections, comprising organisms sampled in Finland, Greece and Hungary. The number of species measured within a genus varied between 1 and 10 in relation to the number of European species within each genus. For example, for the most species-rich genera, both common and rare species from northern and southern latitudes were included.
Fuzzy-coding approach and taxonomic resolution. Traits were coded using a 'fuzzy' approach, in which a value given to each trait category indicates if the taxon has no (0), weak (1), moderate (2) or strong (3) affinity with the category 22 . Affinities were determined based on the proportion of observations (i.e. taxon-specific information from the literature or measurements) or expert opinions that fell within each category for each trait 29 . Fuzzy coding can incorporate intra-taxon variability when trait profiles differ among e.g. species within a genus, early and late instars of one species, or individuals of one species in different environments 29 . Most traits were coded at genus level, but some Diptera and Annelida were coded at family, sub-family or tribe level because of their complex taxonomy, identification difficulties and the scarcity of reliable information about their traits.
www.nature.com/scientificdata www.nature.com/scientificdata/ Data Records DISPERSE can be downloaded as an Excel spreadsheet from the Intermittent River Biodiversity Analysis and Synthesis (IRBAS) webpage (irbas.inrae.fr) and the data repository Figshare 30 .
The database comprises three sheets: DataKey, Data and Reference list. The "Datakey" sheet summarizes the content of each column in the "Data" sheet. The "Data" sheet includes the fuzzy-coded trait categories and cites the sources used to code each trait. The first six columns list the taxa and their taxonomy (group; family; tribe/ sub-family or genus [depending on the level coded]; genus synonyms; lowest taxonomic resolution achieved) to allow users to sort and compile information. Sources are cited in chronological order by the surname of the first author and the year of publication. Expert evaluations are reported as "Unpublished" followed by the name of the expert providing the information. Direct measurements are reported as "Direct measurement from" followed by the expert's name. The "Reference list" sheet contains the references cited in the "Data" sheet, organized in alphabetical order and then by date.
In total, the database contains nine dispersal-related traits divided into 39 trait categories for 480 taxa. Most (78%) taxa are insects, principally Coleoptera and Trichoptera, as these are, together with Diptera, the most diverse orders in freshwater ecosystems 31 . DISPERSE provides complete trait information for 61% of taxa, with Wings are morphologically diverse and include various types: one wing pair, as in Diptera such as the syrphid genus Eristalis (g); one pair of wings with elytra for Coleoptera including the genus Enochrus (h) or with hemielytra for Heteroptera such as the genus Hesperocorixa (i); two wing pairs including one pair of small hind wings for Ephemeroptera including the genus Ephemera (j); and two pairs of similar-sized wings for the Trichoptera genus Polycentropus (k), the Megaloptera genus Sialis (i) and the Odonata genera Ischnura (m) and Crocothemis (n). Wings range in size from a few mm in some Diptera (g) up to more than 3 cm (l-n), with the Odonata exemplifying the large morphologies. Taxa vary in the number of eggs produced per female, ranging from tens per reproductive cycle for most Coleoptera and Heteroptera such as the genus Sigara (o) to several hundreds in the egg masses of most Ephemeroptera and Trichoptera, such as those of the genus Hydropsyche (p). Credits: Adolfo Cordero-Rivera (a-g,i,k-n), Jesús Arribas (h), Pere Bonada (j), José Antonio Carbonell (o) and Maria Alp (p).
www.nature.com/scientificdata www.nature.com/scientificdata/ 1-2 traits being incomplete for the 39% remaining taxa (Table 2, Fig. 2). The traits with the highest percentage of information across taxa were wing pair type and maximum body size, followed by dispersal strategy, life-cycle duration, potential number of reproductive cycles per year, and female wing length ( Table 2). The percentage of completed information was lower for two life-history traits: adult life span and lifelong fecundity ( Table 2).

technical Validation
Most of the trait information (88%) originated from published literature (Supplementary File 1) and the remaining traits were coded based on expert knowledge (9%) and direct measurements (3%) ( Table 2). The database states information sources for each trait and taxon, allowing users to evaluate data quality. Most traits were coded using multiple sources representing multiple species within a genus. When only one study was available, we supplemented this information with expert knowledge, to ensure that trait codes represented potential variability in the taxon.
Using insects as an example, we performed a fuzzy correspondence analysis (FCA) 22 to visualize variability in trait composition among taxa (Fig. 3). Insect orders were clearly distinguished based on their dispersal-related traits, with 32% of the variation explained by the first two FCA axes. Wing pair type and lifelong fecundity had the highest correlation with axis A1 (coefficient 0.87 and 0.63, respectively). Female wing length (0.73) and maximum body size (0.55) were most strongly correlated with axis A2 (Fig. 3 and 4). For example, female Coleoptera typically produce few eggs and have intermediate maximum body sizes and wing lengths, Odonata produce an www.nature.com/scientificdata www.nature.com/scientificdata/ intermediate number of eggs and have long wings, and Ephemeroptera produce many eggs and have short wings ( Fig. 1 and 4).
The database currently represents a Europe-wide resource which can be updated and expanded as new information becomes available, to include more taxa and traits from across and beyond Europe. For example, additional information could be collected on other measures of wing morphology 10,14 and functionality or descriptors of exogenous dispersal vectors such as wind and animals 32 . New data can be contributed by contacting the corresponding author or by completing the contact form on the IRBAS website (http://irbas.inrae.fr/contact), and the online database will be updated accordingly. DISPERSE lays the foundations for a global dispersal trait database, the lack of which is recognized as limiting research progress across multiple disciplines 33 .   www.nature.com/scientificdata www.nature.com/scientificdata/ Usage Notes DISPERSE is the first publicly available database describing the dispersal traits of aquatic macroinvertebrates and includes information on both aquatic and aerial (i.e. flying) life stages. It provides good coverage of macroinvertebrates at the genus level, which is generally considered as sufficient to capture biodiversity dynamics [34][35][36][37] . It will promote incorporation of dispersal proxies into fundamental and applied population and community ecology in freshwater ecosystems 5 . In particular, metacommunity ecology may benefit from the use of dispersal traits 15,38 , which enable classification of taxa according to their dispersal potential in greater detail. Such classification, used in combination with, for example, spatial distance measurements 39,40 , could advance our understanding of the effects of regional dispersal processes on community assembly and biodiversity patterns. Improved knowledge of taxon-specific dispersal abilities may also inform the design of more effective management practices. For example, recognizing dispersal abilities in biomonitoring methods could inform enhancements to catchment-scale management strategies that support ecosystems adapting to global change 41,42 . DISPERSE could also inform conservation strategies by establishing different priorities depending on organisms' dispersal capacities in relation to spatial connectivity 43 .
DISPERSE could also improve species distribution models (SDMs), in which dispersal has rarely been considered due to insufficient data 13 , limiting the accuracy of model predictions 44,45 . Recent trait-based approaches have begun to integrate dispersal into SDMs 45 , and information from DISPERSE could increase model accuracy 46,47 . Including dispersal in SDMs is especially relevant to assessments of biodiversity loss and species vulnerability to climate change 46,48,49 . DISPERSE could also advance understanding of eco-evolutionary relationships and biogeographical phenomena. In an evolutionary context, groups with lower dispersal abilities should be genetically and taxonomically richer due to long-term isolation 50,51 . From a biogeographical perspective, regions affected by glaciations should have species with greater dispersal abilities, enabling postglacial recolonization 52 .
By capturing different dispersal-related biological traits, DISPERSE provides information on organisms' potential ability to move between localities as well as on reproduction and recruitment 15 . Traits also facilitate www.nature.com/scientificdata www.nature.com/scientificdata/ comparison of taxa with different dispersal strategies, which could inform studies conducted at large spatial scales, independent of taxonomy 53 .
Users should note that the dispersal-related traits included in DISPERSE represent an indirect measure of dispersal, not effective dispersal. Therefore, the database is not intended to substitute population-level studies related to dispersal, but to act as a repository that collates and summarizes information from such studies. As freshwater biodiversity declines at unprecedented rates 54,55 , collecting, harmonizing and sharing dispersal-related data on freshwater organisms will underpin evidence-informed initiatives that seek to support the resilience of ecosystems adapting to global change.

Code availability
Analyses were conducted and figures were produced using the R environment 56