*3.2. Selected Indicators: Details*

Details of our analysis of the indicator parameters follow and are summarized in Table 1. 1. *Taxonomic groups*. Among the nine indexes/indicators, we found almost all taxa of pollinators active at temperate latitudes. We considered the taxa of pollinators included in the recent EU guidelines on pollinator monitoring [16]: e.g., bees, butterflies, flies and beetles. GBI, HNV and I.19 focus on butterflies; StN and I.19 include hoverflies (the full list has not yet been completed; [33]) and I.19 also includes bees. Only a few species of beetles are recognized as pollinators, and they are currently not considered in any indicator (GrB focuses on ground-living species). Pollinators are rarely included, and not all groups are considered in the same indicator or at the same taxonomic level. The only exception could be I.19, though it only includes threatened species. The indicators incorporating butterflies and hoverflies include all species at the species level since knowledge of these groups is good.

**Table 1.** Main characteristics of the nine indicators. For STAR ICMi, *various* indicates Anellida, Arthropoda, Bryozoa, Cnidaria, Mollusca, Nematoda, Nemertea, Porifera, Rotatoria, Platyhelminthes; for QBS-ar, *various* indicates Arachnida, Chilopoda, Diplopoda, Insecta, Malacostraca, Pauropoda, Symphyla.


Birds are the only vertebrates that emerged among existing indicators. We have already mentioned their role in agroecosystems. At temperate latitudes, few cases of ornithophilous pollination have been documented [34]. There have been occasional reports of birds feeding on flower nectar [35]. The ecological importance of bird–flower visitation in Europe is still uncertain, especially for plant reproductive output; however, effective pollination has been confirmed for several native and exotic plant species [36,37].

2. *Monitoring type*. Monitoring can be: (a) based on international monitoring (FBI, STAR-ICMi and GBI), (b) based on cartographic analysis, especially the Geographical Information System—GIS (HNV, I.19, I.20), or (c) the output of local sampling (QBS-ar, StN, GrB).

FBI, GBI and STAR-ICMi are based on monitoring programs defined by the European Bird Census Council (EBCC), the European Butterfly Monitoring Scheme (eBMS) and Directive 2000/60/EC, respectively. These schemes are supervised by European agencies but carried out directly and independently by each European country. They occur at constant intervals, yearly for FBI and GBI and every 5 years for STAR ICMi (which includes repetition over the course of the year). FBI and GBI collect information on any available species, but few are used to discuss trends. Instead, STAR ICMi considers the macroinvertebrate population of given stretches of river. FBI monitoring uses specific 50 km<sup>2</sup> grids, where trained observers report all records of visual recognition or birdsong; in the absence of major impediments, monitoring will be replicated at the same site in future years. The Italian regulation for the application of STAR ICMi includes all water stands, while for GBI there is no predefined grid to follow and not all grasslands are monitored. Other indicators are based on local sampling in relation to similar situations from the literature (QBS-ar, StN, GrB). In some cases, such as StN, the data of all studies using the same methods are constantly updated, thus increasing the efficacy of comparison with an expected population. The database is not linked to any national agency and does not require external funds. Contributions to the database are voluntary, and no study is funded by StN. However, even without funding, the abundance of data contributed by volunteers helps to complete the overall geographic or habitat information. QBS-ar values are interpreted on the basis of available literature concerning a given pollutant, but no dataset including the entire literature is available as in the case of StN. In Italy, CREA, ISPRA, University of Parma and the private agency Timesis s.r.l. set up a permanent working group on soil science [32]. An aim is to improve application of the index through definition of local values. Finally, GrB

can only count on the technical reports of ISPRA that standardize monitoring methods but allow free interpretation of results by external experts.

Table 2 contains examples of various past and current monitoring schemes in Italy.

**Table 2.** Past and ongoing monitoring in Italy; distribution maps obtained from different sources as reported.

Example of distribution of monitoring sites for birds in Italy. Each dot indicates an agroecosystem site (10 × 10 km) where records have been taken for more (red) or less (yellow dots) than 11 years. The image (slightly modified) can be found in a national report under various authors (professional and volunteer) who collaborated with Lega Italiana per la Protezione degli Uccelli (LIPU) on the project MITO2000. The report is freely available at https:

//www.reterurale.it/flex/cm/pages/ ServeBLOB.php/L/IT/IDPagina/22311 (accessed on 29 July 2021). Reference: Rete Rurale Nazionale & LIPU (2021) Farmland Bird Index nazionale e andamenti di popolazione delle specie in Italia nel periodo 2000–2020. p. 11

Example of distribution of monitoring sites for butterflies in Italy. The image can be found at https:

//butterfly-monitoring.net/it/mydata (accessed on 29 July 2021) (base map and data from OpenStreetMap and OpenStreetMap Foundation) and is regularly updated. The version shown was downloaded on 29 July 2021. Each dot represents a transect set up by professional or volunteer workers; data are transferred to the site and corresponding database by Butterfly Conservation Europe and the Centre for Ecology & Hydrology.

Example of distribution of monitoring sites for wild bees in Italy. Site selection was by the ongoing project BeeNet, in which the authors are directly involved. The sites are situated in agroecosystems, intensive and seminatural, monitored once a month by experts in 11 Italian regions.

3. *Spatial and habitat context*. The spatial context is not usually part of the indicator itself but is included in the process of site selection. The most common cartographic system employed is that of the CORINE Land Cover Project (CLC). GBI and GrB use CLC to identify the areas to monitor, and likewise StN applies the CORINE European Habitat Classification System (for macrohabitats) and the EU Habitats Directive, further describing the microhabitats where syrphid larvae develop. FBI prefers a regular spatial distribution of the sampling sites, geolocated and later classified on the basis of land use and bird classes. STAR ICMi and QBS-ar use different cartographic systems, including CLC and regional ones. STAR ICMi identifies regions characterized by water/climate/rocky features. QBS-ar is especially linked to soil and pedoclimatic features: a level of detail (fourth level of CLC) would ideally be necessary, but it is not available for all areas. In a few cases or for certain studies (HNV and I.20), other cartographic systems are used, such as those of protected areas (Natura2000) or areas of special interest for certain butterfly or bird species or the threatened species cartography (I.19).

4. *Background*. Knowing the ecology and biology of target species is very important. I.19 will consider extinction risk, while STAR-ICMi and QBS-ar evaluate morphological adaptations to individual microhabitats. QBS-ar will add adaptation to soil characteristics

and sensitivity to pollutants, scored from 0 to 20 (eco-morphological index). StN associates species with their ecological status, assigning a code (blank-3, depending on non-preferred habitat); code 1 allows species association with a habitat when that habitat is linked to another. Syrphid flies, like butterflies, are linked to certain habitats, especially during their larval stage (not mobile), which makes prediction of species assemblages easier. Ground beetles are defined as generalist or specialist (GrB): Detailed knowledge of their diet and mobility can help define habitat alterations. Extensive knowledge of the ecological needs of different species allows selection of flag species that may provide information about the characteristics of an environment and the content of other more common species linked to it. An example is GBI, based on 10 generalist and seven specialist species that provide clues to the potential presence of about another 100 butterflies.

5. *Sampling effort and taxonomic identification*. Sampling effort is established by monitoring protocols, while taxonomic identification can be carried out in the field or in the laboratory. EBCC monitoring plans include a different pool of bird species in each country (233 nesting species in Italy, although we only have enough information for the indicator in the case of 99 species). eBMS investigates 435 European butterfly species, identified at the species level directly in the field. Both are coordinated and supervised by regulatory agencies through the work of thousands of trained professional and volunteer workers. An opposite situation is that of samplings that require microscope identification in the laboratory: STAR-ICMi at family level, QBS-ar at order level and StN and GrB at the species level. QBS-ar not only identifies the order but creates 29 morpho-functional groups that couple different orders or distinguish adult and larval stages. The implementation of the citizen science (the involvement of the public/volunteers—citizen scientists—in scientific surveys) is emerging as a key to successful monitoring programs [38], which develop support materials often translated into different languages and adapted to local/regional fauna. However, this may influence the level of identification that can be reached, depending on the complexity provided by the different target organisms.

Finally, another parameter that may vary is the type of data collected: abundance (FBI, GBI, STAR-ICMi), or presence/absence (occupancy) (StN, HNV, I.19, I.20, GrB, QBS-ar).

6. *Final output*. Indexes usually compare a value with a reference. GBI and FBI apply a population trend (the latter since 1980 in some European countries, since 2000 in Italy). For HNV, I.19 and I.20, the reference is the entire area covered by the administration grid or the farm. It may also be a given local population (StN, STAR ICMi, GrB, FBI). Ideally, the value of the index indicates the disturbance suffered by the environment and recorded by the sampled population. FBI and GBI consider few species, as already mentioned, while StN and GrB consider all species sampled. In some cases, expert opinion is needed to interpret rough data and estimate disturbance (GrB). In other cases, indexes transform the data into a well-defined qualitative scale (STAR-ICMi), or a set of user-friendly values, so that even non-experts can compare results on a national/European basis (QBS-ar and StN).

Some indicators are better employed in association with others that describe the habitat/environment. For example, STAR ICMi helps describe the environment when considered in association with other indicators based on algae, plants and fish; chemical and physical parameters (water, pollutants) or geomorphological features. Physical/chemical parameters are also employed for QBS-ar. FBI can be coupled with the Woodland Bird Index (WBI, evaluating 18 species), All Common Species Index (CSI, 99 species) or birds especially sensitive to phytochemicals (15 widespread species and another six in specific environments). eBMS is working to define indicators for agro-environments and forests, to be coupled with GBI currently available only for grasslands.

### **4. Discussion**

With a view to a future pollinator indicator that could integrate existing ones to survey biodiversity and to direct RDP actions, we assessed past and recent indexes/indicators used for biodiversity assessments. Indicators have become a common tool to evaluate goals, especially at government level [22]. The choice and targeting of indicators are constantly revised. Good examples are the past and current indicators used to monitor biodiversity at the European level (CAP 2014–2020 and CAP post-2020), where FBI has been retained and HNV discarded. While overall international pressure can drive the selection of some indicators, others may be employed at the national level, according to national laws or national mitigation measures to be evaluated. We therefore decided to target nine indexes/indicators applied (developed and endorsed) at the European level, at the national level (Italy) and involving other invertebrates.

Our analysis considered some index/indicator characteristics related to biological and environmental contexts, as well as their practical use. The taxonomic groups considered by the indicator, for example, may span in the entire animal kingdom. However, the indicator must consider the dimension of the employed variable: e.g., insects can be expected to interact with the environment very differently from birds. When considering pollinators, we found that some of them are included in past (HNV) or current (GBI, StN) indicators, or predicted in indicators yet to be defined (I.19). However, not all pollinators' groups are considered at once. Our goal was restricted to verifying how biodiversity is tackled and to what extent pollinator groups are currently included. We intentionally did not discuss here the complexity of pollinator, the ecosystem service they provide (pollination) and the interplay among these variables and the environmental characteristics (in natural or agricultural systems). For an interested reader, how published literature addresses the topic is clearly reviewed in a very recent paper [39]. The authors underlined the importance of clarifying definitions, the pollination studies' context and the focus element of the pollination system and concluded by highlighting the need for developing comparable indicators and standardized methods.

Some efforts in the direction of standardization have been made especially concerning cartography. Martin and colleagues [40] synthesized results from 49 studies to investigate how the spatial arrangement of crop fields and associated landscape features (e.g., field margins) impacts arthropods and their functions. Advances in landscape analysis make it possible to optimize descriptions of land/soil use, albeit at different levels of detail, depending on the database used. Proper identification of landscape can no longer be excluded from monitoring plans [41], and this is also currently supported by the number of indicators that include cartographic information (Tables 1 and 2). Similarly, standardization is important to monitoring. Concerning pollinators, the EU Pollinator Monitoring Scheme [16] is addressing the issue with the contribution of experts in different pollinator groups and from different countries, placing special emphasis on the monitoring plan and procedure through the pilot project SPRING. In Italy, the national BeeNet project replicates a fixed protocol that includes concurrent monitoring of wild bees and plants in 24 agroenvironments. Sites were selected in intensive or seminatural ecosystems by landscape analysis using first the standard CLC and then checking on-site actual conditions.

Our analysis highlights two critical points: the background knowledge on the target and the efforts related to sampling and taxonomic identification. For pollinators, the situation is evolving fast. Public interest has increased sharply in recent decades: society is alarmed by pollinator decrease and interested in initiatives to understand the current situation and to sustain pollinator conservation [42]. Our research team participates in the LIFE 4 POLLINATORS project, which applies some specific actions aimed at data collection on pollinators (bees, wasps, hoverflies, beeflies, beetles, butterflies and moths) in various environments. Direct involvement of citizens, students and farmers includes using a web-tool platform for uploading photos of pollinators visiting flowers, participation at "mini-Bioblitzes" in natural parks and application of specific observation protocols in schools, botanical gardens and farms.

Possibly the most interesting result of our survey is the inclusion of citizen science in data collection. Citizen science brings important added value that makes it possible to implement datasets for establishing trends and baselines useful for indexes/indicators of species. Some successful monitoring programs and indicators rely largely on volunteer citizen-science activity, which consists of involving the public (citizen scientists) in scientific surveys [38]. Citizen science is used widely in various fields of natural science, where the data collected are used to monitor species, trace populations, design distribution maps and define conservation and management plans [43]. As mentioned, Hymenoptera are a very complex taxonomic group and their identification may be even more difficult for citizens, due to their small size, many similar species and the lack of easy-to-use identification tools. To overcome this problem, some bee citizen-science projects focus on a single species in relation to the plant pollinated by it. An example is squash bee (*Eucera* (*Peponapis*) *pruinosa* (Say, 1867)) monitoring on cucurbit flowers for impacts of farm management on bee nesting [44]. Other projects rely on a single plant, which is the selected "site" for observations quantifying the pollinator service [45]. Likewise, projects may focus on a single kind of nesting site, such as nest boxes, to limit the number of insects observed to cavity-nesting species [46]. Low taxonomic data quality is generally considered a main limitation to volunteer biodiversity monitoring. However, such data can be highly informative too, if methods and protocols are developed to allow for the inaccuracy of data provided by volunteers in place of experts [47]. Kremen et al. [48] compared the data collected on pollinators at the rank of order and superfamily by citizen scientists and bee specialists, respectively, in 17 sites; a positive correlation was found between the two datasets with regard to the difference in abundance and richness of pollinator groups between sites. The result was consistent, although citizen scientists observed only half the bee groups detected by professional scientists. The few existing citizen-science projects that monitor pollinator biodiversity in a given natural, urban or agronomic ecosystem train citizens to identify bees at a higher taxonomic rank than species, and often pool pollinator species into few easily identifiable groups [48,49]. A way to reduce errors due to misidentification is to employ a "verified method", in which all observations collected or sent by citizens are verified by experts to increase the accuracy of the data collected [50].

Difficulties in species identification are not the same for all pollinators. Identification of syrphid flies and butterflies is very advanced, as demonstrated by the success of the indicators StN and GBI. Bees are in fact the largest group of pollinators and are also the most difficult to identify [51], due to the large number of species with very different characteristics [51,52] and difficult identification with wide variations in different countries: e.g., in Europe more species live in the Mediterranean area. Regarding identification, some help may soon come from metabarcoding techniques [53]. DNA analyses may enable us to avoid training experts in all taxa, although the support of morphological taxonomy will undoubtedly still be prominent for many years [54].

To conclude, a future pollinator indicator should include: (1) elements that deal with the reduced mobility of pollinators and thus read the landscape; (2) implementation of data collection through citizen science, thus supporting data verification and (3) spot distribution of RDP funding, considering national levels. Actual tools that incorporate information on land use into indicators need to be sharpened to include greater detail. We should also pay attention to the relationship between environmental parameters and the target taxa of pollinators. To overcome the many large gaps in our knowledge of the pollinator biology and ecology of certain species, we suggest broadening the environmental parameters, possibly by building a complex indicator based on several indexes. Among them, those more strictly linked to pollinators should be included (e.g., vegetation type, crops, agricultural practices, climatic context, etc.). In some cases, a reduced number of species could be selected. For example, species sensitive to pesticides can be the main target, or those reacting differently to certain agricultural practices. The ideal approach could be to incorporate information on the abundance and occupancy of sampled species, widening the range of endorsed methodologies.

**Author Contributions:** Conceptualization, S.A., E.M., M.G. (Manuela Giovanetti); methodology, S.A., E.M., M.G. (Manuela Giovanetti); literature analysis, S.A., E.M., S.F., E.L.Z., M.G. (Marta Galloni); resources M.G. (Marta Galloni), L.B., M.Q.; writing—original draft preparation, S.A., E.M., L.B., M.G. (Marta Galloni), M.G. (Manuela Giovanetti), M.Q.; writing—review and editing, M.G. (Manuela Giovanetti), L.B.; visualization, S.A., E.M., M.G. (Manuela Giovanetti); supervision, M.G. (Manuela

Giovanetti), L.B. funding acquisition, L.B., M.Q. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by MIPAAF (Ministry for Environment, Land and Sea Protection grants to CREA "Research Centre for Agriculture and Environment) through the project BeeNet 2019–2023 (Italian National Funds under FEASR 2014–2020).

**Institutional Review Board Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** We are indebted to Antonella Trisorio (CREA "Agricultural Policies and Bioeconomy") who provided updated information and important material. We also thank the project "LIFE 4 POLLINATORS" (LIFE18 GIE/IT/000755) for support. We finally thanks Helen Ampt, who kindly revised the English spelling and grammar, and four anonymous reviewers for the interesting points they raised and discussed with us.

**Conflicts of Interest:** The authors have no conflict of interest, and the funders had no role in the design of the study; in the collection, analysis or interpretation of data; in the writing of the manuscript or in the decision to publish the results.

### **References**

