How bees are counted
Every count is of accepted species. The species list shows every name, including synonyms and misapplied names, with its status.
Checklist
A state counts the bees on its chosen checklist, using the checklist's own names and taxonomy: a state may treat a name as a synonym, or as a misidentification of another species, within its own checklist. Checklists stop at the state line, so a county counts the checklist's bees that also have GBIF occurrences there. A state with no chosen checklist can only be counted by occurrences.
Occurrences
A state counts every bee recorded there by an outside source, under the national taxonomy: published presence such as the USGS Bee-Gap database and assessments, plus every bee GBIF holds at least one occurrence of there — a presence record whose coordinates fall in the state, with no GBIF coordinate flag, not a fossil or a captive specimen, and dated 2000 or later unless All years is on. A county counts its GBIF occurrences. Occurrences never add a bee to a checklist.
Why the two differ
Many state checklists are incomplete, and some occurrences are misidentified or misplaced, so either count can be the larger. A much higher occurrence count points to bees a state's curators may want to review. The site does not add them to the checklist.
Documented presence, not occupancy
Occupancy estimates come from repeated surveys that record non-detections. Occurrences carry none, and nothing says which counties were searched, so the map is not even naïve occupancy. In modelling terms it is presence-only (positive-unlabelled) data grouped by county or state.
A blank county means "no record", not "absent"
The bee may be absent, overlooked, never looked for, held in a collection that is not in GBIF, or recorded without coordinates. Read the map as a lower bound on where each bee lives.
Some shaded counties are wrong
Misidentifications (a real risk in genera such as Lasioglossum and Nomada), coordinate errors GBIF does not flag, and iNaturalist's deliberately blurred coordinates for sensitive species can all place a bee in a county it does not live in, often a neighbouring one.
Numbers reflect effort, not abundance
Occurrence and species counts follow how much collecting and observing happened in a place. Counties with universities, cities or long-running surveys stand out. Comparing species counts between counties as richness needs an effort correction.
Since 2000 or all years
A recent occurrence does not mean the bee is there today, and a place with only older occurrences has not necessarily lost it. It may simply not have been revisited. The gap between the two views mixes real decline with changes in where people look.
Also
- Names follow GBIF's taxonomy: synonyms and subspecies count toward the species.
- GBIF groups a few cities with their counties (Baltimore, St. Louis, Richmond and Franklin, Virginia), so each pair shares one set of counts.
- Conservation status, SGCN and state rank are still the state's, not the county's.
- Occurrences are refreshed from GBIF about every six months.
- The town lines drawn when a county is chosen are Census county subdivisions (towns, townships or census divisions), for finding your way. Nothing is counted by town: GBIF places occurrences by county.