Estimates of epiphyllous population size dynamics
Data from the epiphyllous metacommunity sampling collected in the year 2000 (Zartman, 2003) and 2016 (Sierra et al., 2019b) were used as baseline information on demographic changes accountable to fragmentation in this study. The two focal leafy liverworts species selected,Radula flaccida Lindenb. & Gott. (Radulaceae) andCololejeunea surinamensis Tixier (Lejeuneaceae), are regionally and locally abundant across the landscape (Sierra et al., 2019b). At BDFFP, both species predominantly occur on understory leaves of young trees and shrubs (Figure S2A, E). Colonies of R. flaccida andC. surinamensis are typically found with sexual reproductive structures (male: androecia or/and female: gynoecia/sporophytes), and specialized asexual propagules (gemmae). Nevertheless, the two species differ in the mating system where R. flaccida is unisexual andC. surinamensis is bisexual (Figure S2). These species were chosen due to their: (i) comparatively higher local population densities relative to the other 74 epiphyllous species known from BDFFP (Zartman, 2003; Sierra et al., 2019b); (ii) distinct patterns in radial growth allowing for clear spatial delimitation of colonies; (iii) the negative effect of fragmentation on their local abundances (a characteristic this species shares with a vast majority of the other epiphylls at the BDFFP, see Sierra et al., 2019b); and (iv) empirical data for demographic patterns over a 15-year interval (Sierra et al., 2019b) as well as for associations between colonization and extinction with fragmented patches from transplant experiments (Zartman & Shaw, 2006).
The number of colonies was estimated for a total of 22 study plots (1-ha) distributed in 1-, 10-, 100-ha forest fragments, and in continuous forest (Table S1); see Sierra et al., (2019b) for further detail on the sampling design. The number of colonies was estimated for each study plot by calculating the percentage of occupied host leaves by the species (occupied patches) and multiplying by the total number of host plants with epiphyllous colonies within the respective study plot (suitable patches). This approach led us to accurately compare population size among plots in the fragmented and continuous forest avoiding sampling size bias. For each species, differences in the number of colony estimates were compared among small (1- and 10- ha), and large patches (100- ha fragments and continuous forests) over time from the years 2000 and 2016 using the non-parametric Wilcoxon rank sum exact test. Likewise, was compared tested if the increase colony density observed in small patches in the year 2016 reached similar values when compared to large patches.