<p> The Luquillo Forest Dynamics Plot (LFDP) has played a critical role in the initial discovery and subsequent investigation of many processes that govern tropical island wet forest dynamics. Previous work has identified past land use as the main factor in creating forest community compositional and structural differences across the plot. The responses of different species to past land-use intensity and to hurricane disturbances have created an evolving forest mosaic ideal for studying tropical forest successional dynamics. I revisited the interaction of land-use legacies and natural disturbance in the LFDP with new data and new approaches, with the motivation to reveal new information about the relative roles of anthropogenic disturbance and environmental-niche partitioning on tropical plant communities over time. </p><p> In the context of tropical forests and their successional dynamics, I asked how succession resulting from a history of human land use and more recent hurricanes interacts with background environmental variation to effect community structure and diversity. Community dynamics, in terms of forest structure and composition, were summarized over a twenty-one year period, noticing a decreasing trend in species richness over time and structural maturation of the forest, shown by a decline in small stems (trees &lt; 10 cm diameter) as it recovered from the compound effect of two major hurricane disturbances &ndash; Hugo, 1989 and Georges, 1998. We evaluate the magnitude of past human land use effects over time and define indicator species for areas of differing land-use pressure within the 16-Ha permanent LFDP. Using redundancy analysis, plant community-environmental relationships with respect to soils and topography are quantified. Spatial variables, computed using a principle coordinates of neighborhood matrix, explained the majority of the variability in plant community composition between areas of high and low past land-use within the LFDP, meaning environmental differences (e.g. niche differentiation among tree species) were found to be secondary to land-use legacies in determining forest community composition. </p><p> Over two decades, the effect of past land-use peaked about 15-years following the first of two hurricanes, and remained relative stable over time. Despite damaging the forest, hurricanes preserved community differences in species composition and reinforced structural asymmetries due primarily to two species; <i> Dacryodes excels</i> Vahl., a dominant primary forest tree species, and <i> Casearia arborea</i> (Rich.) Urb., an abundant secondary forest species. Abiotic environmental factors (e.g. soil resources and topographic variation) were weak at explaining differences in forest community composition. Plant community-environmental relationships were stronger in more anthropogenically-disturbed areas, suggesting long-term effects of land use on tropical forest communities on current community dynamics.</p>

<p> <i>Polylepis</i> forests are a unique high-elevation forests dominated by trees of the <i>Polylepis</i> (Rosaceae) genus found only in South America. These forests are distinctive not only for growing at some of the highest elevations in the world, but because they are distributed as isolated forest islands inside a landscape otherwise dominated by grasses. This distribution is due in part to human disturbance on the landscape and there is interest in restoring cover of <i>Polylepis</i>. Understanding of regeneration dynamics and seedling ecology of <i>Polylepis</i> can help in the management of forests and the restoration of this forest cover, but there are important gaps in knowledge regarding theses aspects in most species. For the expansion of forest cover into current grassland areas, it is important to understand the ability of seedlings to colonize, survive and grow in these areas and identify important biotic and abiotic barriers to this. In this dissertation, I contribute to our understanding of these ecological aspects for two species of <i>Polylepis</i> (<i>P. sericea </i> and <i>P. weberbaueri</i>) using a series observational studies and manipulative experiments in forests found between 4000&ndash;4500 masl in Huascaran National Park (Ancash, Peru). </p><p> In Chapter 1, I surveyed patterns of seedling dispersion across forest-grassland boundaries along the elevational gradient and identified significant associations of seedling densities to environmental covariates at local and landscape scales. I found that seedling densities of both species decrease drastically at the edge of the forest canopy, and few seedlings are found in the grassland even within meters of the forest, suggesting significant barriers to seedling establishment in the grassland. Additionally, seedling densities of each species showed patterns with elevation and solar irradiation that mirrored those of adult distributions on the landscape and were different from each other, suggesting that each species has unique environmental tolerances that manifest in the seedling stage and are important to consider in their management and restoration. </p><p> In Chapter 2, I used a series of livestock exclusion experiments to examine the influence of domestic livestock on natural regeneration inside and outside the forest and test the hypothesis that livestock are a significant barrier to seedling establishment at forest-grassland edges. After one year of livestock exclusion, I found only partial support for this hypothesis. In <i>P. sericea</i> forest stands, seedling densities increased, and recruitment and seedling growth was higher in fenced grassland areas. However, in <i> P. weberbaueri</i> stands there was little change in juvenile densities and no differences in seedling performance, except that recruitment was lower in fenced forests areas compared to unfenced forest. This showed that the influences of livestock on <i>Polylepis</i> seedling may be different depending on the habitat and species of <i>Polylepis</i>. Additionally, we found that seedling survival was surprisingly high, and similar between forest and grassland habitat. </p><p> In Chapter 3, I used sowing and &ldquo;wildling&rdquo; transplant experiments within the previous exclosure experiments to tease apart early ontogenetic filters to seedling establishment in the grassland compared to the forest. I found that the seedling emergence stage is a strong bottleneck for seedling establishment in the grassland. Furthermore, I found that <i>P. weberbueri </i> seedling density inside the forest appears to be seed limited, and can be increased by sowing. Transplanted seedlings survival was lower than found previously for natural seedlings, but was also similar between both habitats. However, it appears that livestock depresses <i>P. sericea </i> transplant survival inside the forest. Only <i>P. weberbueri </i> seedlings showed differences in growth between habitats, declining in height in the grassland, but there was only evidence that livestock suppressed growth. I concluded that seedling transplant will be the most efficient way to increase seedlings in the grassland for restoration. </p><p> Together, the results of this dissertation suggest that both unassisted and assisted expansion of natural <i>Polylepis</i> forests faces significant barriers, in particular low seedling emergence in grassland is a strong bottleneck to establishment in this habitat. It remains unclear whether livestock are generally a barrier to forest expansion, as their effects appear to be specific to each species of <i>Polylepis</i>. Here, <i> P. sericea</i> appears to be more sensitive in both forest and grassland to livestock activity than <i>P. weberbaueri</i>, but future work should address longer-term effects of livestock exclusion to confirm the population trajectories observed after one year. Finally, wilding transplants of these species can be used in restoration to overcome the barrier emergence poses to <i>Polylepis</i> establishment in grassland, although with further research, seeding schemes may prove a viable strategy</p><p>

Pitch pine-scrub oak communities, commonly called pine barrens, are scattered throughout the northeastern United States and are severely threatened by development and fire suppression. The pine barrens of Plymouth and Barnstable Counties in southeastern Massachusetts together comprise the third largest area of this imperiled natural community remaining in North America. The region is also experiencing extremely rapid human population growth making it necessary to set realistic, local-scale conservation priorities. I used the pine barrens of southeastern Massachusetts as a case study for addressing a variety of issues of central importance for local-level conservation planning. To address these questions, I conducted both community and individual species analyses of bird and moth survey data collected at the Camp Edwards Training Site in Barnstable County. First, I examined the relative impact of plot, patch, and landscape factors on the distribution and abundance of the breeding bird and rare moth communities of Camp Edwards. I used partial canonical correspondence analysis to decompose the variance explained by each of these levels of environmental factors. For the bird community as a whole, landscape factors explained slightly more variance than plot factors, and patch factors were nonsignificant. Examination of individual bird species of conservation concern revealed that the distributions of the majority of these species were best explained by the landscape model. In the rare moth community, landscape factors explained more than twice as much variance as plot or patch factors, and the presence of host plants was significant only at the landscape level. The specific variables with the greatest influence on community structure are discussed. I then used logistic regression to develop individual species models for birds and moths of conservation concern based on patch and multi-scale landscape variables. I used these models to predict bird and moth rarity hotspots in southeastern Plymouth County. Results indicated (1) there was little congruence between bird and moth rarity hotspots; (2) the few overlapping hotspots occurred in and around Myles Standish State Forest; and (3) most hotspots were inadequately protected. The results suggest that multi-taxa, multi-scale approaches may be required for comprehensive conservation planning in pine barrens communities.