Practice makes the expert: The importance of training volunteers in the generation of phenological data from photographs of biodiversity observation platforms

and dry fruit) of N. glauca, the untrained group had a relatively high level of agreement (imma- ture flower: K = 0.674, z = 16.6, p < 0.001; mature flower: K = 0.597, z = 14.7, p < 0.001; dry fruit: K = 0.569, z = 14, p < 0.001), although not as high as the trained group (immature flower: K = 0.782, z = 37.5, p < 0.001; mature flower: K = 0.909, z = 43.6, p < 0.001, dry fruit: K = 0.603, z = 28.9, p < 0.001) or the expert group (immature flower: K = 0.885, z = 20, p = < 0.001; mature flower: K = 0.960, z = 21.9, p < 0.001, dry fruit: K = 0.724, z = 16.2, p < 0.001). For the reproductive phenophases of L. nepetifolia, as for the new growth phenophase, the agreement rates reached by the untrained volunteers were low (immature flower: K = 0.118, z = 0.51, p < 0.001; mature flower: K = 0.262, z = 7.73, p < 0.001; and dry fruit: K = 0.592, z = 17.6, p < 0.001). For both species, trained volunteer levels of agreement matched expert levels of agreement for reproductive phenophases (L. nepetifolia experts, immature flower: K = 0.697, z = 10.2, p < 0.001; mature flower: K = 0.798, z = 11.7, p < 0.001; dry fruit: K = 0.870, z = 12.8, p < 0.001; L. nepetifolia trained, immature flower: K = 0.701, z = 32.6, p < 0.001; mature flower: K = 0.834, z = 38.8, p <0.001; dry fruit: K = 0.813 z = 37.8, p < 0.001; see above values for N. glauca), except for the N. glauca dry fruit phenophase (Fig 4, Table 1). Raters took approximately half a minute to score each observation (experts: �x = 42±16 seconds; trained: �x

Fig 4. Agreement indices (±CI 95%) for raters in classifying records of Leonotis nepetifolia and Nicotiana glauca. Fleiss Kappa concordance indices for three groups of raters in classifying photographic records of L. nepetifolia (left) and N. glauca (right). Values of K can range from -1 to 1 and represent the proportion of agreement between groups beyond that expected by chance. The index equal to zero represents agreement expected by chance, while values greater than 0.4 are considered good agreement, and greater than 0.75 represent excellent agreement [44,49]. https://doi.org/10.1371/journal.pone.0282750.g004 PLOS ONE The training of volunteers as a promoter of phenological studies PLOS ONE | https://doi.org/10.1371/journal.pone.0282750 March 7, 2023 9 / 17