The core of this phase is represented by the tenets described in Benson et al. (2018). The tenets, proposed in the context of the collaboration agreement between the GOOS, the Ocean Biodiversity Information System (OBIS), and MBON, are reported in the following:

The approach we followed to outline the guidelines needed to address the tenets is to explore and examine the academic literature and to elaborate on the guidelines from the selected papers. To this purpose, the BioEco EOVs and EBVs-related literature (i.e., peer-reviewed papers, technical reports, and online documents) was searched and analyzed; in Figure 1 we illustrate the five tenets and the questions we formulated to perform this task. The guidelines, i.e., the output of Phase 1, are summarized in a synoptic comparison matrix which will be presented in section “Results.” As we aim at testing promptly reuse of marine biological measures, we take into consideration tenets 1–4, and we assess compliance with tenets 1 and 2 to achieve the policy-relevance discussed by tenet 5.

As anticipated in the Introduction, we select as case study the GOV, which is one of the LTER-Italy marine research sites. In the GOV site, meteo-oceanographic and biological data are gathered both at a fixed point observing system [i.e., Acqua Alta Oceanographic tower (Ravaioli et al., 2016)] and during oceanographic cruises. The selection of the GOV was driven by its different features, relevant to our aims. First of all, the GOV enacted a long-term measurement program of biological components, which is pre-requisite for all sites of the LTER networks. Second, it developed the Open Science (OS) initiative named “EcoNAOS” (Ecological Northern Adriatic Open-science Observatory System), where OS principles have been applied to the long-term ecological data gathered in 50 years in this area (Minelli et al., 2018, 2021; Acri et al., 2020). Third, the GOV is involved in the design and implementation of a marine ecological observatory in the Adriatic Sea. The observatory deployment is carried out within the Interreg V-A Italy-Croatia project ECOSS²² (ECOlogical observing System in the Adriatic Sea: oceanographic observations for Biodiversity), which aims at integrating ecological and oceanographic research and monitoring activities with the Natura 2000²³ conservation strategies (Manea et al., 2020). The crucial aspect of this process is to link local, country-specific initiatives in order to evolve common approaches to data acquisition and management, which need to be compliant with both the EV frameworks (Benedetti-Cecchi et al., 2018; Rogers et al., 2020) and the main EU Directives.

For analysis of data management practices of the GOV, we leverage on the informative sources listed in the following. They allow us to describe the attitude and experience of the site managers and the scientific staff in handling data with respect to the EV frameworks.

The output of Phase 2 is a comparison of the practices detected in the GOV with the guidelines derived in Phase 1; this result, depicted in a compliance matrix, is also presented in the section “Results.”

This section describes the guidelines referred to each tenet by reporting how they are derived from the literature in hand. They are classified in a synoptic comparison matrix (Table 1).

As regards tenet 1 Standardized data collection, we identify in the Technical Specification Sheets (TSSs)²⁶ published by GOOS synthetic information on how current monitoring of the BioEco EOVs is organized around the globe. The TSSs illustrate: (i) a classification of the information needed to calculate each essential variable and the best practices for collecting measurements whenever available, (ii) a categorization of the observing elements (e.g., satellites, ships, autonomous platforms, and monitoring networks) to be coordinated, and (iii) the different approaches to data creation and management enacted by the existing communities. With respect to developing a standardized approach to observe or model the EOVs, we consider the TSSs as the basis for identifying the observing requirements the local monitoring program could fulfill and where its contribution is positioned with respect to the communities distributed across continents. Particularly, we identify in the TSSs the guidelines to classify the information requirements of the GOOS framework (i.e., sub-variables, supporting variables, complementary variables, and derived products) as well as the best practices and methods to refer to Table 1–row 1, column 2.

In addition to that, the GEO BON guidelines proposed in the EBV-related literature (Costello et al., 2017) suggest to explicitly recognize the biases of the method chosen (e.g., the threshold in detection) and to use appropriate thesauri and classification systems to assure comparability among different methods along the same observed feature (Table 1–row 1, column 1). Instead of assuming conventional methods, these guidelines are useful to provide the data managers with ways to address the comparison of methods across geographical and temporal scales. From these guidelines, we derive two complementary approaches, that shall be adopted simultaneously by observing systems in contributing to EV programs.

As regards tenet 2 Addressing global reporting needs, we selected literature sources to identify the policy reporting conventions that can be supported by the EVs. Indeed, the EVs are conceived to be combined into indicators to evaluate progress toward objectives of international agreements and regulations (Miloslavich et al., 2018; UNEP CBD, SBI, 2016; Muller-Karger et al., 2018). This is gaining in importance after the 2020 CBD meeting (UNEP CBD, SBSTTA, 2020), which stimulated discussion on limits that prevent achievement of the ABTs (Mace et al., 2018; Phang et al., 2020; Hagerman et al., 2021; Xu et al., 2021). These limits can be bridged by taking advantage of the role of the EVs in delivering knowledge-based information (Anonymous, 2019; Hoban et al., 2020; Geldmann et al., 2021) and in assessing quality of the national and global indicators (Geijzendorffer et al., 2016; Vihervaara et al., 2017). In fact, the EVs have been employed as analytical frameworks to identify the informative gaps of indicators that impact negatively on the robustness of the latter. Specifically, in Geijzendorffer et al. (2016) the reporting needs of seven global and European policy instruments were identified and then linked by experts to in-depth knowledge of specific policy instruments to specific EBV classes. The analysis illustrates a sound method to reference the EBVs-primary data to the requirements of the global and national agreements concerning biodiversity (Table 1–row 2, column 1). In Miloslavich et al. (2018), a list of twenty-four international conventions is analyzed to cluster the drivers (i.e., the societal needs) and the pressures (i.e., anthropogenic stressors) that the BioEco EOVs can address by modeling the measurements collected at any scales (Table 1–row 2, column 2). Hence, the connections between policy reporting instruments and the BioEco EOVs are identified by design.

In these works, both the EBVs and the BioEco EOVs are linked to the ABTs of CBD²⁷, to the UN-SDGs²⁸, and to distinct programs of the United Nations Environment Program (UNEP)²⁹. Nevertheless, the current state of the art falls short of discussing the benefits of the BioEco EOVs monitoring program for national legislations and European Directives (e.g., the Convention on Migratory Species of Wild Animals, the Marine Strategy Framework Directive, the Water Framework Directive). It will be useful for policy makers, implementing agencies, resource managers, and scientists operating at these scales to easily cross-tabulate, through experts’ validation, the BioEco EOVs and the European binding agreements, both to capitalize on the monitoring efforts and to increase reuse of data for different purposes.

In both frameworks, dialog with researchers and stakeholders is crucial to building or disseminating data products so as to implement the recommendation of tenet 2.

As for tenet 3, Making data FAIR, we identify in the literature multiple examples of how the two EV frameworks foster the FAIR principles (Wilkinson et al., 2016) in data management, both with autonomous and coordinated initiatives.

Two informative systems are considered (Table 1–row 4) to maximize centralization of data access and distribution at global scale (Bax et al., 2019). The Global Biodiversity Information Facility (GBIF) is largely accounted (Hardisty et al., 2019a; Jetz et al., 2019) for sharing biodiversity measurements as it offers standard services to distribute the data captured by input users with various expertise along different taxa and ecosystems (marine, terrestrial, and freshwaters). Instead, OBIS is suggested (Benson et al., 2018) as a repository for registering and finding the BioEco EOVs through standard tools because of its consolidated community of users. Beside the federation function facilitated by these information systems, beyond the EV programs, their respective users and data-publishing institutions can benefit from more streamlined ways of working together. This is attested by a recent agreement³⁰ between the two facilities, which was signed on September 1st, 2020. The collaboration envisions a number of goals, among which (i) to encourage OBIS nodes and GBIF data providers to publish marine data in both facilities using a single publishing step, and (ii) to align written guidance on publication of the data to OBIS and GBIF. Nevertheless, the IT facilities already collaborate for technology issues since (i) OBIS supported the Integrated Publish Toolkit (IPT) developed by GBIF, (ii) they are working with the same data standard, and (iii) automatic harvesting between GBIF and OBIS tier two nodes is recommended³¹.

Also, advancements in the direction of integrating GBIF and OBIS have been made, so that data flow between the two facilities is enabled³². This way, both can work directly with the entire marine community and promote its standards and best practices without duplication of effort.

Even though data archiving is not compulsory in the two facilities, both the EV frameworks recommend adoption of standards for metadata and data, summarized below (Table 1–row 3). This allows scientists to contribute to EV-related initiatives by exploiting individual services (i.e., organization- or institution-based information systems, repository or national databases).

Currently, providing measurements for the two frameworks implies different choices of schemas for representing data and metadata, since each variable is likely to have its own distinct data model. Accepted community standards like the Darwin Core Archive (DwC-A)³³, which is maintained by the Biodiversity Information Standards (TDWG), is suggested (Kissling et al., 2018a; Snowden et al., 2019) for representing the structure both of the EBV and BioEco EOV occurrence reports, taxon checklists, sampling events, and measurements in order to facilitate in the future more easy and automatic integration. In fact, while the DwC was initially intended to represent occurrence observations only, recent advancements allow for expanding OBIS with other types of information, suitable to fit the requirements of the marine biodiversity community (De Pooter et al., 2017). Two DwC extensions, i.e., the MeasurementOrFact (MoF) and the ExtendedMeasurementOrFact (eMoF)³⁴, serve the purpose of representing new data types and sampling facts³⁵ in DwC-A, overcoming the more constraining schema requirements of the original DwC³⁶. In order not to impede representation of new schemas, these extensions allow for definition of new entities by means of simple texts (for defining both data types, values, and units), potentially weakening the DwC interoperability. To overcome this issue, and to maintain semantic interoperability of such less strict schematization, it is highly recommended the use of URIs instead of simple texts for linking measurement types, values and units to internationally acknowledged controlled vocabularies and thesauri (OBIS, 2021). Please refer to Bordogna et al. (2021) for an updated account of perspectives on semantics in geo-based information.

The Ecological Metadata Language (EML) (Jones et al., 2019) is recommended to structure metadata information for the purpose of data discovery, for interpretation of data, and for assessment of fitness-for-use. Also, other metadata schemas, such as ISO19115 (ISO TC 211, 2014), ISO19157 (ISO TC 211, 2013), and the INSPIRE Environmental Monitoring Facilities (Inspire Thematic Working Group on Environmental Monitoring, 2011; Zilioli et al., 2019) are suggested in the literature on EBVs (Kissling et al., 2018a) for the high definition of the temporal, geographical, and taxonomical information required for processing EBV datasets.

From an historical perspective, data interoperability for the EBVs and the BioEco EOVs was formulated and discussed in different documents, but it is crucial to both communities and it is developed through collaboration between the two organizations. Both frameworks deal with heterogeneity of providers and the fragmentation of the data sources. Interoperability was addressed to increase reuse of data by assuring scientific autonomy of providers.

At present, the EBVs and the BioEco EOVs share no prescriptive methods for management of the related data but some features to reach interoperability are essential to allow exchange of information when the individual components (i.e., the data service used by providers) are different and maintained by separate organizations.

The Bari Manifesto (Hardisty et al., 2019b) is the interoperability framework proposed for the EBVs. It encapsulates ten principles agreed upon by the representatives of biodiversity IT facilities and it outlines the current best practices in EBV-related informatics. These principles are related to data management topics (e.g., plans, metadata, workflows, ontologies/vocabularies, provenance, structure, quality, services, preservation, and accessibility of data) and provide guidance on how to achieve EBV estimation through trans-national and cross-infrastructure workflows in a computer assisted environment. We suggest these guidelines (Table 1–row 5, column 1) to be followed by the observing elements that belong to research infrastructures (e.g., the eLTER RI) or monitoring networks. Analogously, the Framework for Ocean Observing³⁷ (FOO) (Kissling et al., 2018a), rather than being organized around specific observing platforms, programs, or regions, is organized around the EOVs that are fed by several observing elements. In the analysis led by Snowden et al. (2019), interoperability among the elements of the GOOS promoted in the FOO (Miloslavich et al., 2019) is further evolved. Snowden et al. (2019) present a set of guidelines (Table 1–row 5, column 5) and considerations for each of the research domains involved in the FOO. Compared to physical and chemical oceanography, the biodiversity community demonstrates both a slower adoption of standards and difficulties related to the use of standards developed for other disciplines (particularly, the Earth science/geospatial standards). Moreover, integration of data from closely related communities (biodiversity and ecology) might be hindered by the use of different metadata models with a few common elements. The authors stress the crucial role of the Information management community in developing tools to reach compliance of data with well-established interoperability data standards. They also stress the importance of mapping the individual schemas to facilitate semantic reconciliation of attributes between the (meta)data produced by the communities involved.

As regards tenet 4 Analytical algorithms, tools, and workflow are accessible, we focus on works that describe approaches enacted in the two frameworks to organize a more rigorous and effective management of raw data by defining standard workflows. The scientific business processes for generation of the EVs and related indicators from raw data is increasingly formalized in the literature by specifying and explaining the necessary logical steps (Lehmann et al., 2020).

At present, the EBV-related literature provides examples of standard workflows; each of these is focusing on specific biological processes and thus, ultimately, on individual variables. Kissling et al. (2018a, b), and Jetz et al. (2019) propose workflows for species abundance (SA) and distribution (SD), population abundance, and species traits. They aim at improving accessibility in all phases of the data life cycle by identifying a detailed and common processing procedure to openly integrate any record in global data products. The SA/SD workflow was recently tested with data supplied by two biodiversity IT facilities (i.e., GBIF, ALA) (Hardisty et al., 2019a). In 2021, the beta version for the EBVs portal³⁸ was released: It demonstrates which metrics the EBVs will represent and which outputs will be achieved through common work streams.

Analogous workflow-derived products are outlined for oceanographic data in the work of Buck et al. (2019) and for the BioEco EOVs. Currently, the BioEco EOV workflows are organized mainly at the habitat and taxon levels. For example, due to the two decades of coordinated coral reef monitoring undertaken by the Global Coral Reef Monitoring Network (GCRMN), the BioEco EOV “Hard coral cover and composition” is one of the most globally coordinated and advanced workflow (GCRMN, 2019). The GCRMN adopted a model that maximizes submission of raw data meeting a minimum quality level (Obura et al., 2019) from all parts of the globe. It identifies three levels to describe quality of data and provides specific guidance to increase quality from level 1 to 3, by including also preparation of appropriate metadata.

The Continuous Plankton Recorder (CPR) surveys represent an invaluable source of data for the “Phytoplankton Diversity” biological EOV and already inform indicators for the EU Marine Strategy Framework Directive (McQuatters-Gollop et al., 2015). This program provides highly resolved (i.e., species-level taxonomy) data together with abundances at large spatial and temporal scale by leveraging on methodological similarities of the surveys (Batten et al., 2019). This program also provides evidence of the limits of and the solutions needed for integrating data from the five regional CPR surveys. While not fully global (Bax et al., 2019), the CPR surveys will be expanded to other key areas and integrated with other observing systems (Batten et al., 2019).

A more high-level and cross-variable categorization of the BioEco EOV data and data products is illustrated in the work of Lear et al. (2020). It allows to better understand the different types of European biology data and their products provided by EMODnet³⁹ with reference to the GOOS framework. The workflow facilitates assessment of the fitness-for-purpose of the accessed information by the community of data consumers. This is done through classification of datasets in distinct levels according to the amount of processing and/or analysis that has taken place to bring the primary measurements to a particular intermediate product.

The quality checks as well as the logical steps illustrated in these works, aimed at processing observations into more informative datasets or indicators (Kissling et al., 2018a), should be considered by local observing elements.

The second result of our approach consists of assessing compliance of the data collection and management practices adopted in the GOV with the guidelines derived, respectively, from the literature on the EBVs and the BioEco EOVs. Table 2 summarizes the result of this Phase, i.e., whether the guidelines are followed and to which extent the data management practices satisfy the Benson’s tenets.

As regards the guidelines related to tenet 1, analysis of the DEIMS-SDR metadata and of the 50-year dataset allows us to identify the primary in situ observations collected in the GOV that are useful for calculating the EVs. The observed and/or measured properties are reported in Table 3 (column 1) and are linked to EBVs (columns 10, 11) and BioEco EOVs, also by categorizing the observations according to the GOOS classification [i.e., sub-variables, supporting variables, complementary variables, and derived product (columns 1–9)].

Analysis of the DEIMS-SDR metadata allowed us to identify which EBVs the managers of the GOV declare to measure for the marine GEO BON biome. The GOV monitors 2 EBVs (i.e., species distribution and abundance) for 2 groups of organisms (phyto- and zoo-plankton). Data are obtained from campaign-based measures, with a sampling rate spanning from weekly to monthly frequencies. For abiotic measurements, the GOV also relies on the Acqua Alta Oceanographic tower (Ravaioli et al., 2016). Moreover, sub-, supporting, and complementary variables of phyto- and zoo-plankton biomass and diversity are collected; from these variables, the derived products obtained are illustrated in Table 3.

The taxonomic ranks detected through observation are heterogeneous and groupings are generally limited to class level (e.g., Diatoms, Dinoflagellates, Coccolithophores for phytoplankton). They are indicated as free text in field definitions.

For what concerns compliance with the guidelines identified in the BioEco EOV framework, the methodologies used for collecting and analyzing phyto- and zoo-plankton abundancies and biomass are detailed in the data paper describing the 50-year GOV dataset (Acri et al., 2020), which also takes into account the historical methodological changes occurred.

Phytoplankton abundance and total biomass are measured by using two different methods: Chlorophyll fluorescence (Lorenzen, 1966) and inverted microscopy (Lund et al., 1958). Both these methods are ranked as Best and Acceptable, respectively, in the related GOOS TSS and best practice (Lorenzoni and Benway, 2013).

Zooplankton is collected through plankton tows (Harris et al., 2000), ranked as Best method in Lorenzoni and Benway (2013); abundance is measured through visual counting (stereo microscopy) (Harris et al., 2000) which is not listed among the sampling and analytical specifications.

Methods used for analyzing phyto- and zoo-plankton abundancies and biomass are the most commonly used by marine ecologists, so their biases (Costello et al., 2017) are not expressively indicated neither in the metadata nor in the dataset. The taxonomic revision of the phytoplankton species has been made according to ‘‘Algaebase’’⁴⁰, the global algal database of taxonomic, nomenclatural, and distributional information. For the zooplankton species, the Marine Planktonic Copepods catalogue⁴¹ (Razouls et al., 2005-2021) has been used. In the past, for phytoplankton and zooplankton analyses, several texts and monographs were used (Peragallo, 1908; Pascher, 1913; Hustedt, 1962a, b; Hendey, 1964; Rampi and Bernhard, 1980; Heimdal, 1993; Throndsen, 1993; Tomas, 1997; Bérard-Therriault et al., 1999; Harris et al., 2000).

Information on instruments, sampling, and analytical methods for each parameter are recorded in specific fields in the dataset: They were compiled from both bibliographic references and interviews with active and retired researchers. When they are not available, this is clearly indicated in order to allow for appropriate assessment of fitness-for-use. Methods, including sensors, have been also described by separated metadata using the Open Geospatial Consortium (OGC) SensorML v2.0 standard (OGC, 2014) enriched with terms from the BODC NERC Vocabulary Server (NVS)⁴².

As regards the guidelines concerning tenet 2, neither the DEIMS-SDR metadata nor the six tasks of the EcoNAOS initiative and the description of the GOV dataset allow for relating local measurements to global and national policy reporting requirements. In fact, information on compliance with international conventions and country or community directives are not provided in metadata. EcoNAOS targeted dissemination of results to the scientific audience and it is accessible on the Web without restrictions, but the decision-makers in natural resource management are not directly involved as beneficiaries of the research and monitoring outputs.

With respect to the guidelines concerning tenet 3, the dataset was not deposited in GBIF/OBIS and the standards for data and metadata were different from those suggested by the GOOS and GEO BON communities. The Bari Manifesto principles were not followed nor activities were planned to help conversion between data and metadata formats.

Nevertheless, the six tasks of EcoNAOS can be considered FAIR-oriented practices [for Findable, Accessible, Interoperable, and Reusable management of data (Wilkinson et al., 2016)]. In fact, tasks 2 (Rescue of available metadata), 3 (Integration of historical time series in an interoperable data infrastructure), and 5 (Dynamic citation for time series) contribute at making data findable; tasks 3, 4 (Publication of ideas and tools for processing time series), and 6 (Output guidelines) focus on data accessibility while task 1 (Harmonization of measurements of time series) and 3 address interoperability of data; reuse of the recovered data is addressed by task 2.

The dataset is accessible both as a tabular structure in the CSV format and via the OGC Sensor Observation Service (SOS)⁴³ (OGC, 2012). The coordinate format of the sampling points and the name of sampling stations were harmonized along the temporal axis; units of measure and metadata of instruments, sampling, and analytical methods are provided for each parameter. In the SOS version, the abiotic and biotic observations are encoded in the OGC Observation and Measurement (O&M)⁴⁴ schema (OGC, 2011); metadata of fixed-point observing systems, laboratory instruments, and sensors are described by using the OGC SensorML v2.0 language by means of the open software GET-IT (Oggioni et al., 2017).

Also, the DEIMS-SDR is used by the GOV to make both the site information and the 50-year dataset⁴⁵ available. The DEIMS-SDR metadata of the GOV dataset are serialized in JSON format. In previous versions of the DEIMS-SDR, exporting metadata in XML following EML 2.1 and ISO19139 schema was possible through a module (Wohner et al., 2019) that is currently disabled, but details on the transformation of metadata into EML is documented in Kliment and Oggioni (2011).

As regards the guidelines associated with tenet 4, the workflows identified both in the GEO BON and GOOS frameworks are not specifically considered by the GOV. Nevertheless, the tenet 4 results satisfied by task 4 of EcoNAOS, which established releasing the code for cleaning-up, re-ordering and organizing the dataset, by offering an automatic and accessible procedure for the recovery of historical time series. The code is freely available under the GNU GPL v.3 License both on GitHub⁴⁶ and in Zenodo (Minelli, 2020). Data visualization is ensured by the open software GET-IT, compliant with OGC specifications.