Humans have transformed much of the earth’s surface through a wide range of activities of varying intensities and scales, shaping the landscape we see today. The combination of time and complex human-environment interactions within the Hakai Lúxvbálís Conservancy on the Central Coast of British Columbia has resulted in a landscape with many anthropogenically-generated modifications, such as shell middens which can be found at sites with histories of long-term habitation. Globally, shell middens (and in general, habitation sites) have been found to be factors in shaping site ecology. This thesis seeks to investigate this relationship between human activity and occupation and the landscape. The goal of this project is to examine the legacies from past land use, and subsidies from shell middens, within the present-day plant communities. I conducted an observational study to determine if species richness and overall plant communities on these habitation sites differed from sites without a history of intense occupation. To do this I selected ten habitation sites with known extensive shell middens and paired them with ten control sites with similar site conditions (but without the same site history or shell middens). I measured species abundances within 540 1 m x 1 m quadrats. I also surveyed a select group of culturally significant plant species and culturally modified trees within belt transects at each site. Data regarding the water table level and soil and foliar samples were also collected. A variety of environmental factors, along with the site history were evaluated as explanatory variables. Principal component analyses were used to describe how the gradients within the vegetation communities at three vegetation layers (ground, herb, shrub) to see if they respond differently to long-term site use. The habitation sites were found to be characterized by N-rich plant communities, which were significantly different from the plant communities on the control sites at the ground layer, herb layer, and with all layers combined, at both transect distances analysed, but the shrub layer was only significantly different when the entire transect was considered. The results show plant community composition is most strongly influenced by a combination of factors including site type, canopy cover, slope, topography, and distance from shore, with the weight of their importance depending on what vegetation layer is considered. The habitation sites had a lower average species richness at all layers, compared to the control sites, and their plant communities were shown to change differently with distance from the marine shoreline. Habitation sites also differed by having higher soil nutrient content, lower water table levels, and contained culturally important plant species that were absent on the control sites. This research highlights the influence that humans have had on landscapes in this region. This study shows how the patterns within the plant communities at the habitation sites differ from what is expected within the Coastal Western Hemlock zone. The research improves our understanding of the factors influencing vegetation patterns on the Central Coast of British Columbia with this examination of the complex intersection of historical practices and environmental changes. / Graduate

Human land use legacies have significant and long lasting impacts across landscapes. However, investigating the impacts of ancient land use legacies ( > 400 years) remains problematic due to the difficulty in detecting ancient land uses, especially those beneath dense canopies. The city of Caracol, one of the most important Maya archaeological sites in Belize, was abandoned after the collapse of the Maya civilization (ca. A.D. 900), leaving behind numerous structures, causeways, and agricultural terraces that persist beneath the dense tropical forest of western Belize. LiDAR (Light Detection and Ranging) technology enables detection of below canopy Maya archaeological features, providing an ideal opportunity to study the effects of ancient land use legacies on contemporary tropical forest composition. LiDAR also provided us with a detailed record of the 3-dimensional forest structure over the 200 km2 study area. This allowed the investigation how ancient land uses continue to impact both forest composition, in terms of tree species, and forest structure. I recorded tree species over four land use categories: 1) structures, 2) causeways, 3) terraced, and 4) non-terraced land. Using non-metric multidimensional scaling (NMS) and multiresponse permutation procedures (MRPP) to test for differences between the classes, I found significantly distinct tree communities associated with the presence of terraces and the underlying topography. Terraced slopes appear to function as micro-valleys on the side of a hill, creating an environmental "bridge" between slope and valley tree communities. Tree species composition over causeways and structures was also found to be significantly different from terraced and non-terraced plots. Forest structure was assessed by extracting LiDAR points for terraced (n=150) and nonterraced (n=150) 0.25 ha plots. I calculated average canopy height, canopy closure, and vertical diversity from the height bins of the LiDAR points, using slope, elevation, and aspect as covariates. Using PerMANOVA I determined that forest structure over terraces was significantly different from non-terraced land. Terraces appear to mediate the effect of slope, resulting in less structural variation between slope and non-sloped land. These results led to the conclusion that human land uses abandoned >1000 years ago continue to impact the contemporary forests.

L'Amazonie a connu une longue histoire d'occupation humaine. La nature et l'échelle de l'impact des populations d'Amérindiens précolombiens sur leur environnement sont encore largement débattues. Dans une approche pluridisciplinaire, la thèse vise 1) à améliorer les estimations de l'étendue des remaniements anciens des paysages par l'Homme et 2) à comprendre comment ces remaniements affectent la structure et le fonctionnement des écosystèmes contemporains par l'étude des complexes de champs surélevés précolombiens des savanes côtières de Guyane française. Pour la mise en place d'une agriculture intensive sédentaire, les agriculteurs précolombiens ont transformé ces savanes saisonnièrement inondées en édifiant des champs sous la forme de buttes de terre circulaires produisant une hétérogénéité topographique marquée et organisée. Pour répondre au premier objectif, nous avons développé deux approches permettant de distinguer les paysages de buttes d'origine anthropique de ceux, dont la physionomie est très proche, résultant de processus naturels. Pour répondre au second objectif, nous avons décrit comment la structure des communautés d'ingénieurs naturels d'écosystèmes (fourmis, termites, vers de terre et plantes) répond à l'hétérogénéité induite par l'Homme, et comment les activités de ces ingénieurs assurent l'auto-entretien des anciens champs contre l'érosion depuis leur abandon il y a environ 800 ans. Notre travail révèle que l'étude de la dynamique temporelle de la végétation permet d'inférer l'origine des complexes de buttes uniquement lorsqu'elle combine différents proxys. L'analyse de la structure spatiale des buttes montre que les complexes de buttes de Guyane présentent une orientation souvent en grille carrée, orientation qui n'a jamais été montrée ni prédite pour des paysages d'origine naturel, indiquant que la géométrie du paysage peut porter la signature de l'intervention de l'Homme. Depuis l'abandon des champs surélevés, les communautés d'ingénieurs naturels se structurent, et concentrent leurs activités, dans les buttes. Nous avons montré que ces activités contribuent à maintenir des habitats surélevés contre l'érosion mais que les rétroactions qu'ils conduisent sur le sol sont modulées par les conditions initiales du milieu. Le paysage observé actuellement dans les savanes de Guyane n'est ni entièrement façonné par l'Homme, ni entièrement naturel mais résulte de l'interaction complexe entre les composantes physiques et biotiques et de l'héritage des activités anciennes de l'Homme. Le résultat de ces interactions est reflété par une mosaïque de buttes plus ou moins érodées. Notre travail représente la première étude montrant l'impact à long terme des activités anciennes de l'Homme sur les écosystèmes de savanes en Amazonie. Nos résultats présentent des applications importantes dans le domaine de l'ingénierie écologique pour la conception de nouveaux agroécosystèmes durables. / Amazonia has a long history of human occupation. The nature and scale of the impact of pre-Columbian humans on their environment are still hotly debated. In a pluridisciplinary approach, this thesis aims 1) to improve estimations of the scale of ancient landscape transformations by humans and 2) to understand how these transformations influence the structure and the functioning of contemporary ecosystems, by studying the particular exemple of pre-Columbian raised-field complexes in coastal savannas of French Guiana. To conduct sedentary intensive agriculture, pre-Columbian farmers transformed these seasonally flooded savannas by building raised fieldsin the form of circular moundscreating a marked and organized topographic heterogeneity. To accomplish the first objective, we developed two approaches to distinguish anthropogenic mound-field landscapes from others, of similar physiognomy, resulting from natural processes. To accomplish the second objective, we described how the structure of the community of natural ecosystem engineers (ants, termites, earthworms and plants) is influenced by human-induced heterogeneity, and how feedbacks generated by these engineers can lead to self-organized maintenance of the ancient fields against erosion since their abandonment around 800 years ago. Our work reveals that the study of the temporal dynamics of vegetation can be used to infer the origin of mound complexes only when it combines different proxies. The analysis of spatial structure of mounds shows that mound complexes of Guiana are strongly oriented, often in a square lattice, an orientation that has been neither demonstrated nor predicted for natural landscapes, indicating that landscape geometry bears a diagnostic signature of human activities. Ever since raised fields were abandoned, the community of natural engineers is structured, and its activities are concentrated, on mounds. We showed that these activities cont ribute to maintaining these raised features against erosion, but that the effectiveness of engineer-feedbacks on soil in countering erosion are modulated by initial conditions of the environment. The current landscapes of French Guianan savannas are neither solely modeled by humans nor entirely natural, but result from the complex interaction between physical and biotic components and from the legacies of past human land use. The result of these interactions is reflected in a mosaic of more or less eroded mounds. Our work represents the first study showing the long-term impact of ancient human activities on Amazonian savanna ecosystems. Our results can have important applications in the framework of ecological engineering to conceptualize new durable agroecosystems.