Variation in the abundance and distribution of Spanish moss (Tillandsia usneoides) in different forest communities

Background/Question/Methods

Epiphytic plants may exhibit complex spatial patterns within forest communities that reflect a multitude of factors, including: inherent modes of reproduction, microclimatic variables (not excluding host plant characteristics) and environmental stability. However, understanding the factors that affect the spatial distribution of epiphytic species may be obscured by spatial structure itself, which can confound the assumption of sample independence inherent to most statistical tests.  In the present study, we tested the assertion that the Spanish moss (Tillandsia usneoides) has a patchy local occurrence and we compared its spatial pattern in two adjacent forest stands at the Lake Louise Research Center in Lowndes County, Georgia: 1) a pine-dominated forest community subject to controlled burns and 2) a mixed hardwood and pine secondary forest that has not been burned or cleared for 50 years. In each community, we sampled five 50 x 50 m quadrats. In each quadrat we recorded the location and species of each tree (≥ 5 cm dbh), as well as a rank measure of Spanish moss abundance [on a scale of zero (= no Spanish moss) to five (= maximum coverage)]. To assess the spatial pattern of Spanish moss within a quadrat, we constructed Moran’s I correlograms.

Results/Conclusions

Spanish moss was found in all quadrats sampled in secondary forest, but never in high abundance and its spatial distribution varied among quadrats: three of five quadrats exhibited correlogram patterns consistent with a gradient in abundance; the remaining two plots did not exhibit significant spatial structure. In pine forest, Spanish moss was found in only two of the five quadrats sampled and, where found, also exhibited a gradient in abundance. Our results indicate that strong patch structure did not exist at the spatial scale sampled; nevertheless, we cannot eliminate the possibility that patches may have occurred at a larger spatial scale than we could resolve with 50 x 50 m quadrats, and it is not clear whether the spatial gradients we detected represent global trends at the stand level or the edges of localized patches within a stand. Localized patches within our quadrats may also exist that could not be detected with a global autocorrelation coefficient. Our work supports previous observations that Spanish moss is less abundant in pine forest and that it is found frequently on some oak species [e.g., Live Oak (Quercus virginiana)]; however, its incidence on most trees species may be proportional to their relative abundance.