Thesis (PhD)--Stellenbosch University, 2013. / ENGLISH ABSTRACT: Riparian vegetation communities that occur along perennial rivers are structured in lateral zones that run parallel to river flow. This dissertation investigated the structure of South African riparian vegetation communities along perennial, single-thread headwater streams. The central assumption was that lateral zones result from differential species’ responses to changing abiotic factors along a lateral gradient up the river bank. It was first necessary to establish the pattern of zones and whether this pattern occurs repetitively and predictably on different rivers in different biomes. Since the flow regime is considered to be the master variable that controls the occurrence of lateral zones, the link between flow as the major abiotic driver and the distribution of plants in zones was determined. Predictions were made with respect to how variable flow may influence phenological traits, particularly with respect to seed dispersal, and physiological tolerances to drying out and were tested. The existence of lateral zones at reference sites in the Western Cape of South Africa was explored and their vegetation characteristics were described. Plant distribution was related to bank slope, as defined by elevation and distance from the wetted channel edge during summer (dry season) low flow, indicating a direct link to river bank hydraulics. Whether or not the same zonation patterns occur in riparian communities in other parts of South Africa was explored next. The four zones described for Fynbos Riparian Vegetation were evident at all of the other rivers tested, despite major differences in geographic location, vegetation community type, climate and patterns of seasonal flow. The four lateral zones could be separated from each other using a combination of flood recurrence and inundation duration. Functional differences were investigated between three tree species that occur in Fynbos Riparian Vegetation. Functional differences were apparent with respect to timing of seed dispersal, growth in branch length versus girth and three physiological measures of tolerance to drying out; specific leaf area (cm2.g-1), wood density (g.cm-3) and levels of carbon isotopes (δ13C). In order to determine the impact of invasive alien plants and to monitor recovery after clearing, the physical rules devised to help delineate zones were used to locate lateral zones that had been obliterated after invasion and subsequent clearing. At the sites invaded by A. mearnsii plants, the zone delineations showed that invasion started in the lower dynamic zone, where adult and sapling A. mearnsii were most abundant. In un-invaded systems, this zone was the least densely vegetated of the four zones, the most varied in terms of inundation duration and the frequency of inter- and intra-annual floods, and was an area of active recruitment comprised mainly of recruiting seedlings and saplings. An understanding of the functional differences between lateral zones was a common thread at each riparian community that was linked to the annual frequency of inundation and the period, when inundated. / AFRIKAANSE OPSOMMING: Oewer plantegroei gemeenskappe wat langs standhoudende riviere voorkom is gestruktureer in laterale sones parallel met die rivier vloei. Hierdie verhandeling ondersoek die struktuur van Suid-Afrikaanse oewer plantegroei gemeenskappe langs standhoudende, enkelloop hoof strome. Die sentrale aanname was dat laterale sones vorm as gevolg van verskillende spesies se reaksie teenoor die verandering van abiotiese faktore teen 'n laterale gradiënt met die rivierbank op. Dit was eers nodig om die patroon van die gebiede vas te stel en uit te vind of hierdie patroon herhaaldelik en voorspelbaar binne verskillende riviere in verskillende biome voorkom. Aangesien die vloeiwyse beskou word as die hoof veranderlike wat die teenwoordigheid van laterale sones beheer, is die skakel tussen die vloei, as die belangrikste abiotiese bestuurder, en die verspreiding van plante in sones bepaal. Voorspellings is gemaak met betrekking tot hoe veranderlike vloei fenologiese eienskappe kan beïnvloed, veral met betrekking tot die saad verspreiding, en fisiologiese toleransie teen uitdroog, en is getoets. Die bestaan van laterale sones binne verwysings studie terreine in die Wes-Kaap van Suid- Afrika is ondersoek en hul plantegroei eienskappe is beskryf. Plant verspreiding was verwant aan bank helling, soos gedefinieer deur hoogte en afstand vanaf die nat kanaal rand gedurende somer (droë seisoen) lae vloei, en dui dus op 'n direkte skakel met die rivier bank hidroulika. Of dieselfde sonering patrone voorkom in oewer gemeenskappe in ander dele van Suid-Afrika is volgende verken. Die vier sones beskryf vir fynbos oewer plantegroei was duidelik by al die ander riviere wat ondersoek is, ten spyte van groot verskille in geografiese ligging, plantegroei gemeenskap tipe, klimaat en patrone van seisoenale vloei. Die vier laterale sones kan onderskei word van mekaar deur middel van 'n kombinasie van vloed herhaling en oorstroomde toestand duur. Funksionele verskille is ondersoek tussen drie boom spesies wat voorkom in Fynbos Oewer Plantegroei. Funksionele verskille was duidelik met betrekking tot tydsberekening van saad verspreiding, groei in tak lengte tenoor omtrek, en drie fisiologiese maatstawwe van verdraagsaamheid teenoor uitdroging; spesifieke blaar area (cm2.g-1), hout digtheid (g.cm-3) en vlakke van koolstof isotope (δ13C). Ten einde die impak van indringerplante te bepaal en die herstel na ontbossing te monitor is die fisiese reëls voorheen vasgestel om sones te help baken gebruik om laterale sones, wat vernietig is na indringing en die daaropvolgende ontbossing, te vind. Op die terreine wat deur A. mearnsii indringerplante binnegeval is, het die indeling van sones getoon dat die indringing begin het in die laer dinamiese sone, waar volwasse en klein A. mearnsii bome die volopste was. In stelsels wat nie binnegeval is deur indringerplante was hierdie sone die minste dig begroei van die vier sones, die mees verskillend in terme van oorstroomde toestand duur en die frekwensie van inter-en intra-jaarlikse vloede, en was 'n gebied van aktiewe werwing hoofsaaklik bestaande uit rekruut saailinge en boompies. 'n Begrip van die funksionele verskille tussen laterale sones was 'n algemene verskynsel by elke oewer gemeenskap wat gekoppel was aan die jaarlikse frekwensie van oorstroming en die oorstroomde toestand duur.

Plant macrofossils and pollen from six small basins in western Colorado are used to trace the history of vegetation and climate over the last 15,000 years. The late-glacial upper timberline was 2800 m, and sparse krummholz Picea grew up to 3200 m. Summer temperatures were 3° to 5°C cooler than today. The late Pleistocene climate was influenced by winter storms from the Pacific. Precipitation shifted to a summer-dominated pattern by at least 9000 yr B.P. with the development of the summer monsoon. Plant fossils from bogs and lakes located near modern ecotones track the elevations of the temperature-controlled upper timberline and the moisture-controlled lower forest through the Holocene. Between 9000 and 4000 yr B.P., the Picea engelmannii-Abies lasiocarpa forest covered a broader elevational range, with upper timberline 200-300 m higher than today. Mean annual temperatures were 1.8°C warmer, and mean summer temperatures were 2.1°C warmer, than today. Temperatures were still about 1°C warmer prior to 2000 yr B.P. The lower limits of the montane and subalpine forests were 100-200 m below their modern elevations from 9000-4000 yr B.P. Mean annual precipitation was 50-100 mm greater. By 2600 yr B.P. the modern lower forest borders were established. Modern pollen dispersal, transportation, and deposition was sampled in atmospheric collectors, moss polsters, and surface lake sediments. Annual accumulation rates range between 1000 and 5000 grains cm⁻²yr⁻¹. Modern influx (grains cm⁻²yr⁻¹)averages: 1100 in alpine tundra, 2700 in the subalpine forest, 3400 in the montane forest, and 200 in shrub steppe. Pollen spectra in atmospheric traps and moss polsters reflect local vegetation, and provide effective modern analogs for pollen accumulation in peat bogs. In forested environments 80-90% of the pollen deposition in small lakes (< 5 ha) with no inflowing streams comes from atmospheric input. Pollen spectra in open vegetation are distorted by pollen from other vegetation types. At least half of the pollen deposition in small alpine lakes comes from taxa growing up to 1500 m lower in elevation.

Alvars are globally rare rock barren ecosystems on limestone pavement. This thesis focused on the quantitative classification of vegetation of Manitoba alvars, the relationships between vegetation patterns and environmental factors and the effects of grazing on vegetation. Vegetation plots were completed across twenty sites. Cluster analysis, indicator species analysis and PCA were used to describe eight vegetation types. A RDA revealed moisture regime, soil depth, bare rock cover and disturbance (grazing and browsing) are the most important factors affecting floristic composition. Grazing effects were studied at two sites using paired plots on either side of a fenceline dividing grazed and ungrazed areas. PCA and RDA showed significant difference between vegetation compositions based on grazing. A partitioning of species richness and diversity by introduced and native species revealed that both sites experienced significant replacement by introduced species. Current grazing levels on Manitoba alvars are severely impacting the vegetation of this ecosystem. / October 2016

Over the last millennium peatlands have accumulated significant carbon stores. Drainage for agricultural use has been widespread and has altered the functioning of these mires: shifting them towards carbon release. Recently, in recognition of the range of ecosystem services derived from these landscapes peatland restoration projects have been initiated. Carbon storage is often cited amongst the aims of these projects, especially since the inclusion of rewetting wetlands in the Kyoto Protocol. However, little is known about the effects of ditch blocking on CO2 fluxes, particularly in Molinia caerulea dominated peatlands, a species common on degraded peatlands which tolerates a range of water table depths. This thesis aims firstly to quantify CO2 fluxes from a drained Molinia caerulea dominated blanket bog and to improve understanding of the temporal and spatial controls on these fluxes and secondly, to quantify the immediate effects of ditch blocking. Closed chamber measurements of net ecosystem exchange and partitioned below-ground respiration from control-restored paired sites were collected over the growing seasons immediately pre- (2012) and post-restoration (2013/2014). These flux data were coupled with remotely sensed data quantifying vegetation phenology and structure with a fine resolution (daily/cm) over large extents (annual/catchment). Although temporal variation in water table depth was not related to CO2 fluxes, the seasonal average related to vegetation composition suggesting raising water tables may promote a change in vegetation composition within these species-poor ecosystems. The distribution of water table depths, vegetation composition and CO2 fluxes did not vary with proximity to drainage ditches despite their prominence. An empirical model suggests in a drained state these peatlands are CO2 sources, indicating carbon previously accumulated is gradually being lost. Data suggest restoration does not always significantly affect water tables and consequently CO2 fluxes in the short-term. Where shallower water tables were maintained during dry conditions photosynthesis decreased and heterotrophic respiration increased: enhancing carbon release. Research undertaken during atypical weather has been unable to determine if restoration will be able to raise water tables sufficiently to protect the existing peat store and promote the vegetation change required to reinstate CO2 sequestration in the longer-term.

Studies in paleoclimate are important because they give us knowledge about how the climate system works and puts the current climate change in necessary perspective. By studying (pre)historic periods we increase our knowledge not just about these periods, but also about the processes that are important for climatic variations and changes. This thesis deals mainly with the interaction between climate and vegetation. Vegetation changes can affect climate in many different ways. These effects can be divided into two main categories: biogeochemical and biogeophysical processes. This thesis studies the biogeophysical effects of vegetation changes on climate in climate models. Climate models are a necessary tool for investigating how climate responds to changes in the climate system, as well as for making predictions of future climate. The biogeophysical processes are strongly related to characteristics of the land surface. Vegetation changes alter the land surface’s albedo (ability to reflect incoming solar radiation), roughness and evapotranspiration (the sum of evaporation and tran-spiration), which in turn affects the energy fluxes between the land surface and the atmosphere and thereby the climate. It is not, however, evident in what way; denser vegetation (e.g. forest instead of grassland) gives decreased albedo, which results in higher temperature, but also increased evapotranspiration, which contrastingly results in lower temperature. Vegetation changes are in this thesis studied in four different (pre)historic periods: two very cold periods with no human influence (c. 44,000 and 21,000 years ago), one warm period with minor human influence (c. 6,000 years ago) and a cold period with substantial human influence (c. 200 years ago). In addition to that the present climate is studied. The combination of these periods gives an estimate of the effect of both natural and anthropogenic vegetation on climate in different climatic contexts. The results show that vegetation changes can change temperature with 1–3 °C depending on season and region. The response is not the same everywhere, but depends on local properties of the land surface. During the winter half of the year, the albedo effect is usually most important as the difference in albedo between forest and open land is very large. During the summer half of the year the evapotranspiration effect is usually most important as differences in albedo between different vegetation types are smaller. A prerequisite for differences in evapotranspiration is that there is sufficient amount of water available. In dry regions, evapotranspiration does not change much with changes in vegetation, which means that the albedo effect will dominate also in summer. The conclusion of these studies is that vegetation changes can have a considerable effect on climate, comparable to the effect of increasing amounts of greenhouse gases in scenarios of future climate. Thus, it is important to have an appropriate description of the vegetation in studies of past, present and future climate. This means that vegetation has the potential to work as a feedback mechanism to natural climatic variations, but also that man can alter climate by altering the vegetation. It also means that mankind may have influenced climate before we started to use fossil fuel. Consequently, vegetation changes can be used as a means to mitigate climate change locally. / Studiet av paleoklimat är viktigt för att det ger kunskap om hur klimatsystemet fungerar samt för att det sätter nuvarande klimatförändring i ett nödvändigt perspektiv. Genom att studera (för)historiska perioder ökar vi vår kunskap om dessa perioder, men också om vilka processer som har betydelse för klimatets variationer. Denna avhandling behandlar framförallt interaktionen mellan klimat och växtlighet. Förändringar i växtligheten kan påverka klimatet på flera olika sätt. Dessa kan delas in i två huvudgrupper: biogeokemiska och biogeofysikaliska processer. Denna avhandling studerar de biogeofysikaliska effekterna på klimatet i klimatmodeller. Klimatmodeller är ett nödvändigt verktyg för att studera hur klimatet svarar på förändringar i klimatsystemet, samt för att göra förutsägelser om framtidens klimat. De biogeofysikaliska processerna är förknippade med markytans egenskaper. Förändrad växtlighet förändrar markytans albedo (förmågan att reflektera inkommande soltrålning), skrovlighet och förmågan att transportera vatten från marken till atmosfären genom evapotranspiration (summan av avdunstning och transpiration), vilket i sin tur påverkar energiflödena mellan markytan och atmosfären. Dessa förändringar påverkar sedermera klimatet. Det är emellertid inte självklart på vilket sätt; tätare växtlighet (t.ex. skog i stället för äng) ger minskat albedo vilket ger högre temperatur, men också ökad evapotranspiration vilket däremot ger lägre temperatur. Växtlighetsförändringars påverkan på klimatet studeras i denna avhandling i fyra olika (för)historiska perioder: två väldigt kalla perioder utan mänsklig påverkan (ca 44 000 och 21 000 år sedan), en varm period med liten mänsklig påverkan (ca 6 000 år sedan) och en kall period med avsevärd mänsklig påverkan (ca 200 år sedan). I tillägg till det studeras också dagens klimat. Resultaten visar att förändringar i växtlighet lokalt kan ha en signifikant effekt på klimatet. Kombinationen av dessa perioder ger en uppskattning av effekten av både naturlig och antropogen växtlighet i olika klimatsammanhang. Förändrad växtlighet kan ändra temperaturen med 1-3 °C beroende på årstid och område. Responsen är inte densamma överallt utan beror på lokala egenskaper hos markytan. Under vinterhalvåret är oftast albedoeffekten viktigast eftersom skillnaden i albedo mellan skog och öppet landskap då är mycket stor. Under sommarhalvåret är evapotranspirationen oftast viktigast eftersom skillnaden i albedo mellan olika växtlighetstyper då oftast är små. En förutsättning för det är att det finns tillräckligt med vatten tillgängligt för evapotranspiration. I torra områden förändras evapotranspirationen inte särskilt mycket när växtligheten förändras, vilket gör att albedoeffekten dominerar även på sommaren.  Slutsatsen av dessa studier blir att förändrad växtlighet kan ha en betydande effekt på klimatet, jämförbar med den effekt som ökade halter av växthusgaser har i scenarier för framtida klimat. Alltså är det viktigt att ha en korrekt beskrivning av växtligheten i studier av (för)historiskt, nutida och framtida klimat. Det betyder att växtligheten har potentialen att fungera som en återkopplingsmekanism till naturliga klimatvariationer, men också att människan kan påverka klimatet genom att förändra växtligheten. Det betyder också att mänskligheten kan ha påverkat klimatet innan vi började använda fossilt bränsle. Följaktligen kan växtlighetsförändringar användas som ett sätt att lokalt begränsa klimatförändringar. / <p>At the time of the doctoral defense, the following paper was unpublished and had a status as follows: Paper 4: Manuscript.</p>

Arctic air temperatures have increased at two to three times the global rate over the past century. As a result, abiotic and biotic responses to climate change are more rapid and pronounced in the Arctic compared to other biomes. One important change detected over the past several decades by satellite studies is a lengthening of the arctic growing season, which is due to earlier onsets and/or delayed ends to growing seasons. A handful of studies also suggest the peak green season (i.e. when the tundra is at maximum leaf-out and maximum carbon uptake potential) is starting earlier in the arctic tundra. The vast majority of studies detecting shifts in the growing season suggest this is due to increasing spring and fall air temperatures, which lead to earlier spring snowmelt and later fall snowfall. Less well understood is how indirect consequences of arctic warming, such as ongoing changes in plant community composition, may also be contributing to these satellite signals. For instance, there is mounting evidence that deciduous shrubs are expanding into previously non-shrub dominated tundra in several parts of the Arctic. Deciduous shrubs may alter tundra canopy phenology and contribute to the regional shifts in timing of phenological events being detected by satellites. Concurrently, in many areas where deciduous shrubs are expanding they are also becoming taller. As taller shrubs become increasingly dominant, arctic landscapes may retain more snow, which could lengthen spring snow cover duration, and offset advances in the start of the growing season that are expected as a result of earlier spring snowmelt. As a consequence, deeper snow and later snowmelt in taller shrub tundra could delay plant emergence, and shorten the period of annual carbon uptake. Thus greater dominance of taller stature deciduous shrubs in the Arctic may actually delay the onset of the growing season, which would suggest that increasing deciduous shrub dominance may not be contributing to satellite signals of an earlier start to the growing season. To contribute to satellite-detected shifts in the onset of the growing and peak seasons, tall deciduous shrubs would need to have accelerated leaf development to compensate for deeper snow packs and later spring snowmelt relative to surrounding tundra. Understanding the drivers of shifts in tundra phenology is important since longer (or shorter) growing and peak green seasons would increase (or decrease) productivity and the period of carbon uptake, which will have implications for landscape-level carbon exchange, and ultimately global carbon balances. Given the rate and magnitude of changes occurring in the face of acute arctic warming, there is a need to monitor, understand, and predict ecological responses over large spatial and temporal scales. However, compared to more southern environments, the arctic tundra is characterized by considerable heterogeneity in vegetation distribution, as well as a short and rapid growing season. In addition, the arctic tundra is relatively vast and inaccessible. These characteristics can make it difficult to monitor and study changes in the Arctic, and make it difficult to develop landscape-level models able to predict changes in ecosystem dynamics and tundra vegetation. The use of airborne and satellite sensors has at least partially fulfilled these needs to monitor, understand, and predict change in the Arctic. The normalized difference vegetation index (NDVI) acquired from these sensors, for instance, has become a widely adopted tool for detecting and quantifying spatial-temporal dynamics in tundra vegetation cover, productivity, and phenology. This suggests that remote sensing technology and vegetation indices may be similarly applied to characterizing patterns of primary and secondary consumers (e.g. arthropods), which would be enormously useful in a region as vast and remote as the Arctic. The research presented in this dissertation provides useful insight into the influence vegetation community composition, particularly increasing deciduous shrub dominance, has on phenology, carbon flux, and canopy arthropod biomass in the arctic foothills region of the Brooks Range, Alaska. Findings in Chapter one suggest that delayed snowmelt in areas dominated by taller shrubs may have a short-lived impact on the timing of leaf development, likely resulting in no difference in duration of peak photosynthetic period between tall and short- stature shrubs. Findings in Chapter two suggest that greater deciduous shrub dominance not only increases carbon uptake due to higher leaf area relative to surrounding tundra, but may also be causing an earlier onset of, and ultimately a net extension of, the period of maximum tundra greenness and further increasing peak season carbon sequestration. Findings in Chapter three suggest that measurements of the NDVI made from air and spaceborne sensors may be able to quantify spatial and temporal variation in canopy arthropod biomass at landscape to regional scales in the arctic tundra.

Changes in the productivity of temperate forests have important implications for atmospheric carbon dioxide (CO2) concentrations, and many efforts have focused on methods to monitor both gross and net primary productivity in temperate forests. Remotely sensed vegetation indices provide spatially extensive measures of vegetation activity, and the Enhanced Vegetation Index (EVI) has been widely linked to photosynthetic activity of vegetation. Networks of tree ring width (TRW) chronologies provide ground-based estimates of annual net carbon (C) uptake in forests, and time series of EVI and TRW may capture common productivity signals. Robust correlations between mean TRW and EVI may enhance spatial extrapolations of TRW-based productivity estimates, ultimately improving understanding of spatio-temporal variability in forest productivity. The research presented in this thesis investigates potential empirical relationships between networks of TRW chronologies and time series of Landsat EVI and Landsat-based phenological metrics in the Northeast United States. We hypothesized that mean TRW is positively correlated with mean monthly EVI during the growing season, EVI integrated over the growing season, and growing season length. Results indicate that correlations between TRW and EVI are largely not significant in this region. The complex response of tree growth to a variety of limiting climatic factors in temperate forests may decouple measures of TRW growth and canopy reflectance. However, results also indicate that there may be important lag effects in which EVI affects mean TRW during the following year. These findings may improve understanding of links between C uptake and growth of tree stems over large spatial scales.

This study is concerned with programs to conserve remnant stands of native vegetation in the agricultural regions of South Australia and concentrates on the development of explicit evaluation procedures which reflect stated conservation objectives. As botanical data are available for stands of native vegetation in most of the agricultural regions, stands in a particular region are able to be compared rather than assessed in isolation. Based on a review of conservation evaluation schemes in Australia and overseas, a hierarchical evaluation procedure using multiple criteria to compare stands was applied to stands of vegetation in the Mount Lofty Ranges. The conservation objective, of preserving samples of all plant communities in a region, led to the analysis of existing botanical data from two surveys of the Mount Lofty Ranges, to provide the basis for an inventory of regional plant communities. These surveys included 52 remnant stands of native vegetation and employed a point-centred quarter plotless sampling technique to summarise the vegetation. Numerical classificatory analysis of the raw sampling point data produced a more comprehensive floristic summary than the results from the plotless sampling. These floristic groups were correlated with physical environmental variables to produce an inventory of 45 regional vegetation types, as the first stage in the conservation evaluation of stands. Evaluation criteria of size, species richness and species rarity were quantified and used to select examples of each vegetation type on the basis of overall satisfaction of the criteria. In addition, the smallest suite of stands, in which all the vegetation types were represented, was determined, and was shown to be 24 stands. All of these were included in the 37 stands chosen using the three criteria. A third evaluation stage used stand parameters such as plant community richness to give a priority ranking of the 37 stands. A polythetic divisive classification of the vegetation types was developed to provide a means of evaluating communities in stands of native vegetation yet to be sampled in the region, and of comparing the vegetation types with communities in existing reservesr Examination of species-sampling area relationships led to recommended plot sizes for such future vegetation surveys in the Mount Lofty Ranges. The ease of collecting floristic data and the extensive time involved in quantitative measurements suggest that all perennial plant species be recorded and only estimations be made of vegetation quantity and structure for each sampling plot. This study demonstrates the usefulness of numerical classification techniques for conservation evaluation, and of continuous variables to quantify criteria of conservation value; and the application of those criteria in an explicit, hierarchical conservation evaluation procedure.

In this study, climate and vegetation interactions are examined for several periods of the geological past with (1) a dynamic global vegetation model CARAIB, and (2) an Earth system model of intermediate complexity Planet Simulator. Both models interact through an equilibrium asynchronous coupling procedure, which consists of a series of iterations of climate and vegetation equilibrium simulations. The climate and vegetation models, as well as the coupling technique developed here and some basic results are rst presented. The models are then applied to study three periods that have experienced large scale climate and vegetation changes: the Last Glacial Maximum, the Middle Pliocene and the Middle Miocene. First, the implications of changing land surface properties on the climate at the Last Glacial Maximum (LGM) are studied. A series of sensitivity experiments is carried out to evaluate the contribution of vegetation change relative to the contributions of the ice sheet expansion and elevation, and the lowering of the atmospheric CO2 on the Last Glacial Maximum climate. We find that the vegetation cover change at the LGM leads to signicant global cooling, together with a decrease in precipitation. The change in the vegetation cover also reinforces the cooling due to other surface cover changes, such as the ice-cover, when applied together with them. Secondly, the climate and vegetation models are used to investigate the Middle Pliocene and Middle Miocene climates and vegetations. Both periods are characterised by a warmer and wetter than present-day climate, and as a consequence, by a reduction of desert areas and an expansion and densification of forests especially at high latitudes. Our results show that the vegetation feedback on climate may have contributed to maintain and even to intensify the warm and humid conditions produced by the other climatic factors at the Middle Pliocene and Middle Miocene. Thus, considering the climate and vegetation interactions could help to reconcile model results with proxy-based reconstructions. This also suggests that vegetation-climate interactions could provide a complementary, if not an alternative mechanism,to the large increase of CO2 required by the models to produce the estimated warming at both periods. The results presented in this study highlight the contribution of the biosphere in past climate changes, and therefore emphasise the study of climate and vegetation interactions to better understand past, present and future climate changes. Furthermore, our results also illustrate that considering the vegetation feedback on climate helps to improve the comparison of climate modelling results to proxy-based reconstructions for the studied periods. Ce travail a pour objet l'étude des interactions entre la végétation et le climat au cours de plusieurs périodes du passé géologique de la Terre, à l'aide (1) du modèle dynamique global de végétation CARAIB, et (2) du modèle climatique de complexité intermédiaire Planet Simulator. Les modèles de végé- tation et de climat interagissent via une procédure de couplage asynchrone à l'équilibre. Dans ce travail, les modèles ont été appliqués à l'étude de trois périodes passées, caractérisées par un changement du climat et de la couverture de végétation à grande échelle : le Dernier Maximum Glaciaire, le Pliocène Moyen et le Miocène Moyen. Dans un premier temps, nous avons étudié l'impact de changements des propriétés de surface continentale sur le climat du dernier maximum glaciaire. Un ensemble de tests de sensibilité a été réalisés à l'aide du modèle climatique afin d'évaluer les contributions relatives de changements dans la couverture de végétation, d'une expansion des calottes de glace et d'une augmentation de leur élévation, et d'une diminution de la concentration de dioxyde de carbone dans l'atmosphère. Les résultats obtenus permettent de mettre en évidence l'impact non-négligeable du changement de végétation sur le climat. Ce dernier changement provoque en effet une diminution des températures, accompagnée d'une réduction des précipitations en moyenne globale. De plus, le changement de végétation renforce les refroidissements obtenus, lorsqu'il est combiné aux autres modifications de la couverture de surface. D'autre part, nous avons modélisé les climats et distributions de végétation du Pliocène Moyen et du Miocène Moyen. Les résultats obtenus témoignent d'un climat plus chaud et plus humide que le climat actuel au cours de ces deux périodes, et par conséquent, d'une réduction des écosystèmes désertiques au profit d'une expansion et densification des écosystèmes forestiers, particulièrement aux hautes latitudes. Ces résultats sont en bon accord avec les résultats de précédentes études de modélisation, ainsi qu'avec les observations. Nos résultats montrent également que la rétroaction de la végétation, en réponse au changement climatique, peut avoir contribué à maintenir, et même à intensier, les conditions chaudes et humides au Pliocène Moyen et au Miocène Moyen produites par les autres facteurs climatiques. Cela suggère également que les interactions végétation-climat pourraient constituer un mécanisme complémentaire, si pas alternatif, à l'importante augmentation de la concentration de CO2 atmosphérique requise par les modèles pour produire les réchauffements estimés aux périodes considérées, et dès lors réconcilier les résultats de modélisation et les estimations basées sur les données, notamment pour Miocène Moyen. Cette étude souligne donc l'importance du rôle joué par la biosphère dans les changements climatiques passés, ainsi que la nécessité de la prise en compte des interactions végétation-climat lors de la simulation de climats passés à l'aide de modèles climatiques et de végétation. De plus, les résultats obtenus montrent, du moins pour les périodes étudiées ici, que la prise en compte des rétroactions de la végétation sur le climat aident à améliorer la comparaison des résultats de modélisation aux reconstructions basées sur des données.

No description available.