Lianas (woody climbing plants) are important elements of temperate and tropical forests, with widely varying abundance and impact across forests. As structural parasites, lianas generally reduce the growth and survival of their host trees, and higher liana abundances are associated with reduced forest carbon stocks and reduced rates of carbon accumulation in regrowing stands. The few studies that have compared the effects on host trees of different liana species have found wide variation, consistent with patterns in other host-parasite systems, and with theoretical models of parasite evolution. The existence of this variation suggests that an understanding of variation in the abundance and impacts of lianas across forests requires an understanding of how different conditions impose different constraints and selection on liana strategies. Here, we first applied models from disease ecology to develop hypotheses for how liana strategies and abundances would be expected to vary across environments. Our proxy for liana virulence was liana load – the proportion of the host tree crown covered with lianas – because higher liana load is associated with lower host tree fitness. We then analyzed covariation in liana loads, liana abundance, and liana functional composition within and across tropical forests in Panama to test these ideas.
Our model predicted higher liana virulence will be favored in forests with higher tree mortality rates and shorter canopies, broadly consistent with empirical patterns in liana abundance and liana load along gradients of disturbance and forest age. Liana species in central Panama exhibited wide variation in seed mass, wood density, specific leaf area, and other functional traits, with interspecific variation similar in magnitude to that found among co-occurring trees, but significantly different mean trait values. Co-occurring liana species at our study sites also varied in liana-specific traits, most notably climbing mode, with twining and tendril climbers most common, scramblers rarer, and root climbers very rare. Climbing mode was significantly associated with the liana load of trees hosting lianas of a single species, with the highest mean liana loads found for tendril climbers, suggesting this is the most virulent strategy. Sites differed in liana prevalence (proportion of trees infested), liana loads, and liana trait distributions. Our modeling and analyses showed that liana traits are associated with liana effects on host trees and ecosystems, and thus that a better understanding of the factors shaping liana trait distributions within forests is critical to understanding the effects of lianas on forest ecosystems.