Nectar. Each flower secretes nectar at rate s up to a species cap Vmax (coneflower 4 µL, sage 2 µL). A visit removes min(V, room in the crop); the crop holds 40 µL (typical honeybee) before a hive trip. Overflow is spilled and counted so that
secreted = harvested + Δstock + spilled holds exactly (checked below). With many bees the stock is depleted: more foragers share the same supply, so intake per bee falls (exploitation competition).
Foraging. After a visit the bee considers flowers within radius R that were not visited in the last τ seconds (scent marks), weight w ∝ d−k. When both species are available it stays with the current species with probability c (flower constancy), otherwise it switches.
Pollen. A stigma holds at most 250 grains; when full, conspecific and foreign grains displace each other in proportion. Per visit, in this order: deposit a fraction δ of the carried load of each species on the stigma; groom a fraction g away; collect up to κ = 80 grains from the anthers. With no collection the deposit on the k-th flower of a row is Dk = δP0[(1−δ)(1−g)]k−1 (pollen carry-over).
seed set S = Dc/(Dc + K) · exp(−Dh/H) K = 12, H = 90 grains (illustrative)
purity = ΣDc / (ΣDc + ΣDh)
Conspecific load Dc saturates seed set; heterospecific pollen Dh (clogging the stigma) reduces it. These two saturation constants and the secretion rate are illustrative, not field measurements; hive-trip time, handling time (1 s + V/0.8 µL/s) and the 2 m/s flight speed are typical values.
Honest limits. No weather, no learning of nectar rewards, no pollen viability decay, no second pollinator guild; flowers are fixed in a 5.6 × 3.6 m patch drawn 2.5× larger, and bees ~6× larger than life. The constancy sweep answers one question: does faithful visiting protect a rare species from clogging pollen? — read it as a model result, not a field prediction.