1. Site attributes. As described in the earlier BON workshops, the individual BON Sites will have a number of attributes that will promote biodiversity research, including (but not limited to) field study sites with long-term security from unplanned land-use change or development (although urban/suburban study sites could be included in BON), access to both terrestrial and aquatic habitats, on-site facilities (field stations or other laboratory/housing buildings), vehicles for site access, Internet computer connectivity, GIS/GPS instrumentation, meteorological data (preferably on-site stations in close proximity to field sampling sites), on-site reference collections of local flora and fauna, and site personnel to assist with BON visiting researcher activities. All BON sites should be affiliated in some fashion with a research museum, which would archive and curate specimens collected from its associated BON site(s). It is also anticipated that BON sites could consist of multiple partners and collaborating personnel, organizations and/or institutions.
   
2. Site personnel. The Principal Investigator (PI) of individual BON sites will provide the coordination of all BON activities at the site, and will arrange the activities of site personnel and facilitate the visits of BON researchers. Each site will have a designated BON data manager/GIS manager who will work with all visiting BON researchers to ensure data quality assurance, quality control (QA/QC) and data sharing, and proper GPS site location collections and archival in site GISs for field sampling sites. Site PIs and data/GIS managers will coordinate across sites to ensure standardization and compatibility of all databases. Specimen data access also will be the responsibility of the BON site data/GIS manager, in conjunction with museum curatorial staff. Finally, each site will be responsible for repeated sampling of certain "core" taxa (see below); as such, a team of field biologists will be funded at each BON site to collect biodiversity data on these important taxa. The composition of the field team will be at the discretion of the Site PI, and could include faculty, research staff, graduate students, and undergraduate students. Training and supervision of the field team would be the responsibility of the Site PI and collaborating researchers.
   
3. BON GIS/GPS/Data Management. The BON research mission requires that each site be equipped with a minimum set of capabilities for data collection and management. The following list provides some details for these capabilities.
   
   1. Georeferencing. All field data should be georeferenced to a scale of spatial precision that is consistent with the sampling protocol. A template of surveyed permanent sample locations will facilitate georeferencing of some types of sample collection, whereas others will require that georeferencing data be collected simultaneously with biological data, using differentially-corrected GPS measurements or other surveying techniques. Site personnel should be available to assist biological researchers with field georeferencing.
      
   2. GIS and Informatics. Sites should have Geographic Information Systems (GIS), including digital data layers for basic site characteristics, and the capability to generate data layers on an ongoing basis with species distribution data that are collected on site. Basic site characteristics include, but are not limited to: topography (bathymetry), hydrography, soils, vegetation, human structures, land use/ownership, and the location of previous studies and manipulative experiments. Remote sensing and aerial photo collections that can provide a broader perspective of the site's relationship to its surroundings and a longer-term perspective of gross vegetation and land use changes are also desirable. Each site should have informatics and communications capabilities, Internet connections, and a data management scheme for backing up and archiving digital data.
      
   3. Metadata, QA/QC. Quality assurance/quality control (QA/QC) measures should be developed for data entry, data processing, and data custody. Metadata that describes georeferencing and GIS methods should accompany each data layer, using standard tools provided by the Field Biology Consortium.
4. BON "Core" Taxa. Conceptually, the network of 50 BON sites will support a large and diverse assemblage of researchers interested in biodiversity patterns. The NSF would provide competitive funding to scientists through question-based BON research grants, and in the aggregate these grants would cover a wide range of taxa. However, the BON concept cannot rely on a serendipitous distribution of research proposals to cover all major taxa at all sites during all years. Many of the BON research proposals would rely on certain taxa's being measured simultaneously with the "target taxa" of the research proposal; for example, an entomologist wishing to study the diversity of hymenopteran pollinators across 10 BON sites might need to have diversity estimates of the flowering plants at each site, but might not have the expertise, time and resources to sample the plants at every site each time a site visit is made. Thus, the entomologist would have to rely on other BON researchers to provide those data (this would clearly fall under the multi-disciplinary, data-sharing concept of the BON Program). However, if some sites (by chance alone) do not have research groups working on flowering plant diversity during the year the entomologist is sampling, there would be no data available for comparative purposes.
   

   The solution for this problem will be to require each NSF BON Site to support a small field team to collect annual diversity data on core taxa, and make these data available to all visiting researchers. The list of core taxa has not been determined at this time, and the formulation of such a list was beyond the scope of the current workshop. However, we strongly recommend that specified core taxa be sampled at each BON site. Routine sampling of the distribution and abundance of particular organisms will provide baseline information to contribute to and integrate with other BON research activities. Core taxa at a particular site might include organisms that are already routinely sampled at that site and organisms that have been deliberately selected to maximize overlap across multiple BON sites. Where possible, interdisciplinary teams should be used to collect data on multiple biological groups of core taxa, soils, climate, hydrology, and topography to produce multiple layers (themes) of biodiversity for the ecosystem. For example, data on native and exotic plant diversity could be collected with data on soil chemistry, pollinators, and soil organisms. In streams and ponds, data on fishes, amphibians, water quality, and primary producers could be collected simultaneously.

   Rather than dictate what types of organisms should be the common core taxa, we recommend that these taxa be chosen by either

   1. the first group of BON awardees at their initial meeting, or
   2. a separate working-group of ecologists/taxonomists during another BON-planning workshop. In these discussions, the scientists involved should consider taxa from multiple habitat types (freshwater, marine, wetland, terrestrial) and multiple trophic levels (producers, consumers, decomposers). Collaborative efforts, such as roving teams with particular taxonomic expertise, should be encouraged.

   Once the core taxa are designated, additional workshops should be held to develop common sampling protocols for each taxon. These meetings would bring together statisticians and field biologists with the dual goal of ensuring statistically valid designs and maximizing comparability of data across sites.

   Although our group believes that the development of core taxa recommendations is beyond the scope of our workshop, we strongly affirm their importance to the eventual success of BON. The future of BON and NEON programs depend on the routine and consistent collection of data for a select group of core taxa with common protocols designed and tested to ensure comparability of data across a wide range of biomes and habitats. Core taxa, identified in previous BON reports, might include vascular plants, selected aquatic or terrestrial invertebrates, mammals, fishes, birds, and soil organisms. Initial requests for proposals from NSF will likely be specifically written to maximize cross-site data comparability for a subset of core taxa. Addressing hypotheses about broad-scale patterns of biodiversity is problematic without highly comparable data sets.

5. BON Reference Collections and Research Museums. Species lists and availability of properly preserved, identified and curated specimens for each site are essential for attracting research on particular taxa, for biogeographic analyses, and for assessment of biotic responses to environmental gradients in space and time. Site PIs must have or obtain necessary permits for general collection and storage of specimens and should ensure adequate support for taxonomic services for core taxa. Collected specimens of core taxa must be properly identified, georeferenced, archived in an appropriate museum or reference collection (as arranged with museum personnel), and made accessible to researchers using the site. Site personnel are responsible for standardized taxonomy and updated nomenclature of core taxa. Site PIs should develop a policy, or work with researchers using the site, to determine numbers of specimens of each taxon that will be retained in local voucher collections or deposited in other research collections. Numbers of specimens of each taxon, their spatial and temporal referencing, and their storage location must be maintained in the site database and be accessible across the Network.
   
6. BON Sampling Designs. In an effort to make the data from all taxa in all years across all BON Sites as comparable as possible, the participants of this workshop discussed at length various strategies for implementing "standard" field sampling procedures.

During these discussions, a "top-down" (= NSF-imposed methods) versus "bottom-up"

We recommend instead that standardized sampling be accomplished primarily by giving incentives to Principal Investigators. First, we recommend that the PIs of BON sites be asked to demonstrate that their site would use standardized protocols for data management and field sampling of core taxa. Standardization of data management (data file formats, taxa nomenclature and codes, etc.), as well as the selection of core taxa and appropriate sampling methods, should be undertaken by another pre-BON workshop, or meetings of PIs and data managers after BON site proposals are funded. Site PIs would therefore be responsible for ensuring that

1. comparable data for multiple core taxa at multiple BON sites are being collected using mutually agreed-upon standardized methods at all sites,
2. data collected among various taxa within any BON site are integrated spatially and temporally, coincident with measurements of abiotic variables and processes, and

3. that the data sets are consistently managed within and among all BON sites. Site PIs also will be responsible for coordinating field site selection with visiting BON researchers, with the goal of overlapping the sampling of as many taxa as possible on core study areas within the BON Site.

Second, we recommend that the PIs of BON research proposals be asked to show that their data are collected using best available protocols, that the completeness (or sampling-effort dependence) of the data be assessed (see below), and that integration across taxa and across BON sites be advanced by the proposed study.

Third, we discussed in detail the potential spatial layout of sampling sites within a particular BON field site. We first considered a systematic grid system of sample points, that could be scaled up or down to reflect appropriate sample sizes for different taxa (e.g., soil microbes vs. birds). The workshop group felt that rigid systematic sampling carried considerable risk of bias (due to possible unknown underlying systematic dispersion patterns of certain taxa), and if a site were to utilize a systematic sampling scheme, then some additional precautionary sampling effort should be dedicated to evaluating potential bias. We next considered a hierarchical sampling strategy, in which stratification of sampling in field sites (based, for example, on different ecosystem types, soil texture/moisture gradients, topography, etc.) would be incorporated into the BON Site design. This approach was favored by the group, due in part to the greater chance of sampling "rarer" habitats containing unique species assemblages. Randomized sampling sites located within strata could then be arranged, with high-resolution GPS coordinates taken to record the exact locations for comparisons with samples from other taxa. In all cases, the BON sampling designs would be evaluated through the NSF review process for statistical rigor and appropriateness to the taxa and ecosystem being addressed.