No combination of morphological, ecological or chemical characters can reliably diagnose species in the Parmelia saxatilis aggregate in Scotland

The Parmelia saxatilis aggregate is comprised of three species in Europe, proposed to differ in morphological, distributional or chemical characters. In this study, we sampled nearly 200 thalli from five sites across a steep ecological gradient in Scotland to investigate the distribution of the species in the aggregate, and we characterized all specimens by morphological, chemical and ITS sequence variation. In our sample, 191 specimens were identified to species using ITS. We confirm that a PCR length assay can be used for separation of P. saxatilis s. str. from P. ernstiae and P. serrana because across our sample, P. saxatilis s. str. consistently includes a group I intron c. 200 bp. Using sequences for specimen identification, we test previously proposed characters to diagnose specimens and use multivariate analysis to identify the most consistent features which may be used for identification among species. First, we test lobe morphology, presence and amount of pruina, distribution of isidia, lobe tip colour, and chemistry. Second, we use classification trees that quantify the contributions of 1) morphological and chemical factors, and 2) morphological and ecological factors, to a priori ITS-barcoded specimens. Parmelia saxatilis s. str., P. ernstiae and P. serrana all occur across the sampled gradient but differ in the frequency of occurrence, with P. saxatilis s. str. more frequent in the relatively drier east, and P. ernstiae more frequent in the wetter west. Parmelia serrana was collected around a third as often as the other two species, but more frequently on tree branches than expected. For all the morphological characters examined, all the species show some overlap and no morphological features are diagnostic, though trends are apparent by species. The classification tree approach holds promise for discovering the most meaningful variation for field workers to approach correct identifications. Chemical variation using TLC is perhaps the best way to distinguish most specimens but, even here, overlap in chemosyndromes exists among the species.