Appendix 1. Methods for measurement of functional traits.
Functional traits
To characterize the vegetative and regenerative phases of the species, we choose four weakly correlated plant functional traits (correlation coefficients not shown) related to vegetative and reproductive processes (specific leaf area, leaf carbon:nitrogen ratio, plant height and blooming time), in order to maximize the number of dimensions of the functional space (Laughlin, 2014). First, we measured the average specific leaf area (SLA) and the leaf carbon:nitrogen ratio (leaf C:N ratio) of each population within transects. To do so, we took two fully developed and healthy leaves per individual (n = 5 individuals per population) and placed them in an oven at 60ºC (in some cases where we found less than five individuals within a population we used as many individuals as we could find). After 72 hours, we weighed each leaf independently and measured the adaxial surface (including petiole when present) by scanning the leaves and using image-analysis software. We determined the adaxial surface of revolute leaves on the basis of the actual one-sided leaf area rather than the projected area. The average SLA for each individual plant was considered as one statistical observation. In cases where the weight of a single leaf from an individual was less than 1 mg, we pooled as many leaves as necessary from the same individual to reach at least 1 mg for weighing and scanning. For leafless plants (e.g. Ulexparviflorussubsp. parviflorus) we took the top 2 cm of a young tip (functional analogue of a leaf) and treated it as above (Cornelissen et al. 2003). Although SLA estimates should ideally be based on the one-sided area of fresh leaves, we used oven-dried leaves because of the technical difficulties involved in keeping leaves fresh in the field during the long sampling season. Nevertheless, we do not believe that differences in one-sided area reduction due to water loss across the species in the dataset are large enough to bias our SLA estimations (e.g. Schwery et al. 2015).
In order to measure the leaf C:N ratio, we pooled all leaves used for the SLA estimations of each population into a single bulk sample per population. Each sample was analyzed three times (1 mg per essay) by combustion in pure oxygen at 100-1000 °C for carbon and with an infrared cell and a thermal conductivity cell for nitrogen. We took the arithmetic mean of the three essays as the estimate of the leaf C:N ratio per population.Information about plant height and blooming duration of each species was obtained from a specialized regional flora (Blanca et al., 2009).
References
Blanca G, Cabezudo B, Cueto M, Fernández C, Morales, C. (eds) (2009) Flora vascular de Andalucía Oriental. Junta de Andalucía, Consejería de Medio Ambiente, Sevilla.
Cornelissen JHC, Lavorel S, Garnier E, et al (2003) A handbook of protocolsforstandardised and easymeasurement of plantfunctionaltraitsworldwide. Aust J Bot 51:335–380. doi:10.1071/BT02124
Laughlin DC (2014) Theintrinsicdimensionality of planttraits and itsrelevance to communityassembly. J Ecol 102:186–193. doi: 10.1111/1365-2745.12187
Schwery O, Onstein RE, Bouchenak-Khelladi Y, et al (2015) As old as themountains: theradiations of theEricaceae. New Phytol 207:355–367. doi: 10.1111/nph.13234