A more precise, ecologically-based way of describing the trends of bee foraging preference is the term ‘association,’ which has previously been used , though not yet adopted as a standard term. We propose adopting the term ‘bee-to-plant association’ as the standard to describe the ecological trends of attraction by particular bees to certain plant’s flowers. This research uses bee-to-plant associations either as a binary value or as relative attraction associations . It is imperative that the relationship between bees and their foraging plants be proved and supported by scientific research so that designers can maximize bee habitat design effectivity.Characterized as “listmania,” Garbuzov and Ratnieks , reviewed 15 lists of ‘plants for bees’ in North America and Britain and found minimal overlap between the recommendations between the lists for similar geographic regions. The authors argue that the efficacy of how these plant lists function ecologically needs further study . Within the existing literature there are numerous conflicts and inconsistencies between plant lists to determine which plants are best .However, in contrast to Garbuzov and Ratnieks , we believe that habitat solutions are likely to reflect localized climates and specializations among various geographic bee populations and, as Garbuzov and Ratnieks found, lists of forage plants for bees will likely not have much overlap between world-wide geographically distant locations. This theory is part of a broader ecological theory stating that differences in sites and years may show different geographical mosaics of coevolution . Identifying inadequacies in current bee plant lists is an essential first step in understanding how to improve bee habitats. There is a need for better empirical data on bee’s use of plant resources, plastic flower bucket including the issues of locality, but also appropriateness of plantings for bees . This study utilizes plant list datasets which were derived from empirical data, published by Frankie and Xerces .

We test the strength of these Central Valley California geographically pertinent datasets on-site, to see how well they perform for bees, both naturalized and native, in California. At the time of fieldwork for this study these qualitatively tested datasets were both available to the public, designers included, and both reflect the climate locality of the Davis, California study site. In essence, this study explored the merits and limitations of pollinator plant lists which were available at the time. As Garbuzov and Ratnieks points out, the strength of a model is only as good as the dataset from which it is built. For example, if both data sets are stated to be the best for bees- why would their plant species differ? Designers must have the best possible quantified plant lists to maximize pollinator habitat effectively.Targeted, strategic habitat analysis and modifications could help to boost both habitat connectivity and native bee populations , and in doing so, protect pollination networks and services . Ultimately, conservation and stabilization of bee populations is vital for human resiliency . Due to the diversity and complexity of native bees and their habitat needs, it is vital to understand that protecting bee ecosystem services means conserving an entire suite of insects and considering their various feeding preferences in the process . For example, of the approximately 20,000 species of the world’s bees, about 4,000 of them live in North America, of which nearly 2,000 are in California . According to renowned bee entomologist Robbin Thorp there are 21-26 bee genera in Davis, California, with 58-72 species . In contrast, Frankie estimates that 17 genera and 46 species are commonly found in California. Effective conservation needs better basic information for guidance. A variety of bees should be studied in a site’s location and management should strive to simultaneously meet the needs of the most important bees to maintain pollination ecosystem services .A major autecological framework for conducting habitat analysis is the application of wildlife habitat relationships modeling .

A WHR model for any species typically consists of three life requisites defined by plant communities: feeding habitat, cover habitat, and reproductive habitat. Another component of WHR models is identifying essential ‘habitat elements’ which can beliving or non-living . Since plant communities tend to change over ecoregional spatial extents, WHR models can vary regionally. For instance, California has a well-developed WHR modelling system and Oregon and Washington have a different system . Historically, WHR models were created for predicting vertebrate animal occurrences, however, this study tests whether a WHR modeling approach, based on foraging data could be applied to California native and naturalized bees. WHR models have successfully been used for vertebrate animal conservation for many decades, but this approach has not yet been applied to study bees or other insects, to the authors’ knowledge. We believe it is an important step to approach bee conservation from this point of view to identify critical ecological shortcomings and to maximize conservation efforts using habitat models to guide best management practices.Located in California’s Central Valley, the UC Davis Arboretum and Public Garden is a unique environment to study bee-to-plant associations. Situated in a Mediterranean climate, 35 distinctly themed gardens compose the linear Arboretum landscape, which spans approximately 2.4 km in length . Garden themes and names range from geographic , to ecological , to special plant type . Some are more eclectic in planting theme; they are simply named after neighboring buildings . Importantly, each garden has a geographically defined border and is mapped to the plant species, subspecies, or cultivar level . The high-resolution Arboretum plant collection maps and ancillary aerial photography make spatial accuracy possible within two meters.Building a matrix of bee life history was the first step in creating a WHR model. Literature was searched to collect and compile existing information on bee-to-plant relationship lists for foraging associations, predominant nesting styles , and foraging distances . In this study we concentrated on developing the foraging component of the model and did not test reproductive needs .

We compiled a comprehensive matrix of bee foraging association data from four studies including: Frankie and Xerces . Most plants In the Arboretum collections are horticultural plantings, but there are also some remnant native heritage trees which are long established and contribute strongly to plant community structure. We also included any associations to food crops, since ensuring pollination of agricultural crops has extreme importance and has received much attention in recent years . It was unlikely that we would find crop plants in the Arboretum; however, plants of the same genus as food crops may be found. Quantifying bee-to-plant observations for crops and their close relatives should be a priority in future studies due to the gravity of importance. Meanwhile, with geographic juxtaposition, urban areas could help to support or subsidize pollination of crop plantings . Moreover, urban pollinators could contribute to the greater ecology and food webs of their place, helping more than with human needs. Table 1 shows the completed presence-only bee-to-plant foraging matrix, derived from literature-based observational, quantified data . All of the Frankie and Xerces datasets were compiled by observing the relative attraction of bee-to-plant associations. Both studies tried to determine which plants are best for bees based on site observations by counting which plants received the most visits by bees. As a baseline for our study, Table 1 reports the sum total number of plants utilized for each native and naturalized bee genera and the sum total of the number of plants per bee genera. Next, construction of the bee-to-plant foraging relationship models was done by first obtaining the Arboretum’s plant collection geodatabase , which has every planting mapped with geographic coordinates and supplementing those data with the CalFlora bloom time database . This was done for all Arboretum plant species and was added to the geodatabase using a table join function in ArcGIS . Approximately half of the Arboretum’s plant list was supplemented with CalFlora’s researched bloom times . As the remaining half of the list’s bloom times required further research, flower buckets wholesale they were determined on a case-by-case basis from reliable literature sources . In cases where bloom data were not available, approximations were made based on other ancillary data from scientific papers on each plant genus and/or species as needed; however, this was uncommon. Upon completion, Arboretum plants could be queried in the database by plant name, garden location and/or bloom month.Bee plant association data were collected on a weekly time interval for one calendar year . This frequency of sampling was chosen because previous trial runs with classic monthly and two-week sampling resolution was not sufficient to track rapid phenological changes of plants in this environment. Data collection was done primarily through classical non-lethal entomological field netting and foraging observation methods as described in Pardikes et al. .

Additionally, global positioning system technology was used to enhance traditional netting and observation methods with spatial location data. To study bees at the landscape scale, entomological on-site methods were adapted to meet the needs of this study extent . In particular, net collection was utilized due to its ability to reflect correlations of plant species richness,particularly in sites 100 m in diameter or less . Pan traps were not used due to concern of biased collection results, but also because they do not help to understand bee foraging patterns . In accordance with accepted methods in bee biology fieldwork, data surveys were completed on days with best weather for that week . Ideally, best weather is defined as calm wind , clear/sunny skies, and warm temperatures which are all preferred by bees . The weather application Weather Underground was used for daily climate data such as temperature and wind speed . In summer months with peak abundance of bee activity, a single survey took up to three days to complete due to the volume of data collected. Bee foraging surveys consisted of a weekly walk via the circular path loop throughout each of the 35 gardens in the UC Davis Arboretum and Public Garden in Davis. For each survey author KC randomized the starting point of this sequential circular sampling transect. Construction activity in a small portion of the gardens occurred from January through October 2017 at the east end of the Arboretum which limited site access times to those areas, but did not seem to affect bee behavior in those gardens. Due to varied start points for each weekly walk, the different gardens were visited at a variety of times of day throughout the year to avoid observational bias. This helped to ensure no garden would be favored by warmer afternoon temperatures or changes in sun and shadow.Bee foraging observations were done one garden at a time, by identifying each bee genus foraging on plant genera within the garden. A modern system for recording written notes and images with corresponding geographic coordinate data was devised for this task. A digital DSLR Canon T1i camera equipped with a high-quality Sigma macro lens captured a representative image of each foraging association. At each flowering plant genus per garden, author KC motionlessly observed for insect activity. If insect movement was detected author KC visually focused on the insect’s physical attributes, behavior, and movement patterns, such as has been shown to work with Citizen Scientists for bees . Using a single same observer throughout the study avoided the potential for observational bias of multiple observers. Netting was essential in collecting new specimens, both for ideal on-site as well as in-lab identification. Unique or unidentified bee specimens were collected and frozen, then thawed, pinned and identified with a dissection microscope—a standard protocol for bee identification . Due to practicalities of identification of both plants and bees in the field, and because the study collected bee foraging data across a relatively large site, we settled on genus levels of phylogeny. This was done to reconcile the micro-site scales at which bees forage versus the miles wide spatial extent of the Arboretum gardens and plantings. Importantly, bee-to-plant foraging associations were recorded per garden. In this way, entomological methods could be adapted to look for bee-to-plant associations across a large study site, rather than classical insect surveys.Data processing occurred post circular transect walk using a personal computer. Handwritten notes were transcribed to a data collection spreadsheet . One representative JPEG image with coordinates for each unique floral visitation per garden was then loaded into ArcGIS , using the ‘Photos to Points’ tool.