2.2 Measurement of traits
Spur length was measured of all flowering individuals in each population, including both red- and yellow-flowered individuals. The spur lengths of 1–3 randomly selected flowers per plant were measured with a digital caliper (precision 0.01 mm), and the average length of the measured spurs was used as the spur length of that individual (Figure 1c). Corolla diameter and petal width of each individual were measured at the same time. The variation in floral traits was visualized by principal component analysis (PCA) and compared among populations. Preliminary observations showed that the three floral traits did not differ among flowers within an individual, and petal width did not differ among the five petals of each flower. We considered spur length to be the most important trait because of its relation to visitor size. Therefore, in subsequent analyses we focused on spur length. The multiple comparison Steel-Dwass test was used to compare average spur length between populations.
We also examined spatial autocorrelation (i.e., whether the variation in spur length could be explained by physical distance) by using the ”moran.test” function in the ”spdep” package in the R Software Environment ver. 4.0.2 (R Core Team, 2013) to run Moran’s I test. This analysis used the average spur length and the latitude and longitude of each population.