No description available.

No description available.

Effects of climatic factors on the plant distribution were examined by means of direct gradient analysis, and the relationship of forest flora, life form and phytogeographical distribution was examined. Subsequently, leaf phenology of forest plants were analyzed to evaluate the adaptive significance in relation to the environments in forest understory. In the boreo-nemoral forest ecotone, Hokkaido Island, northern Japan, co-ccurrence of northern and southern plants in a certain forest site is more notable in the understory than in the crown, and this dates back to the late-uaternary period, where the decrease in temperature associated with the glacial period forced the understory flora to adapt their life form or leaf habits to snowcover and light conditions of the interior forests. / Hokkaido University (北海道大学) / 博士 / 環境科学

In the terrestrial biosphere forests have a significant role as a carbon sink. Under recent climate change, it is increasingly important to detect seasonal change or ‘phenology’ that can influence the global carbon cycle. Monitoring canopies using camera systems has offered an inexpensive means to quantify the phenological changes. However, the reliability is not well known. In order to examine the usefulness of cameras to observe forest phenology, we analysed canopy images taken in two deciduous forests in Japan and England and investigate which colour index is best for tracking forest phenology and predict carbon uptake by trees. A camera test using model leaves under controlled conditions has also carried out to examine sensitivity of colour indices for discriminating leaf colours. The main findings of the present study are: 1) Time courses of colour indices derived from images taken in deciduous forests showed typical patterns throughout the growing season. Although cameras are not calibrated instrument, analysis of images allowed detecting the timings of phenological events such as leaf onset and leaf fall; 2) The strength of the green channel (or chromatic coordinate of green) was useful to observe leaf expansion as well as damage by spring late frost. However, the results of the camera test using model leaves suggested that this index was not sufficiently sensitive to detect leaf senescence. Amongst colour indices, Hue was the most robust metric for different cameras, different atmospheric conditions and different distances. The test also revealed Hue was useful to track nitrogen status of leaves; 3) Modelling results using a light use efficiency model for GPP showed a strong relationship between GPP and Hue, which was stronger than the relationships using alternative traditional indices.

Red alder (Alnus rubra) is a tree species with high economic and ecological importance. It is subject to defoliation during unpredictable, episodic outbreaks of tent caterpillars (Malacosoma spp.) that result in reduced growth, decreased wood production, unsightly appearance and mortality in severe cases. Alder trees are weakened by severe and repeated tent caterpillar defoliation, and this can increase the susceptibility of the trees to other pests, diseases and drought. Repeated attack by tent caterpillars can cause decline in red alder populations, which can have potential negative impacts on the ecological and economic benefits of the species. Evidence from other species has shown that plants produce phytochemicals for defense against herbivores at a cost to growth and reproduction, but the relative magnitude of the cost of allocating available resources to defense depends on the level of the resources, and the plant genotype. The quality of a plant as food for herbivores is influenced by leaf physical and biochemical traits, and these traits change during a growing season or upon attack by herbivores. My research aimed to explore the defense mechanisms of red alder against western tent caterpillars (Malacosoma californicum) and determine the resistance variation among and within red alder populations, and to evaluate red alder available resource (nitrogen) allocation to defense and growth. Bioassay feeding trials were conducted in 2014 and 2015 with western tent caterpillars (WTC) (M. californicum) on twenty red alder clones from ten provenances. Phenology and quality of red alder leaves as food for the defoliators were analyzed to determine if budburst, leaf chemical content, water content or physical traits are major determinants of western tent caterpillars preference for red alder leaves. In another experiment, one-year-old seedlings from 100 half-sib red alder families were treated with two levels of ammonium nitrate (NH4NO3) in two growing seasons in a common garden. Growth, herbivore defense-related traits and root nodulation were measured and ranked among the plant genotypes and between the two nitrogen (N) treatments. Leaves from the two N treatments and different alder families were also used for bioassay feeding trials with WTC larvae to determine effects of N and genotype on red alder herbivory resistance. In my final experiments, I harvested and analyzed leaves from three-year-old red alder trees from five different families on eight dates from early April to mid-October 2016 to quantify oregonin and total phenolics concentrations, and wound induction experiments were conducted to determine if the concentrations of the chemicals vary during a growing season and upon attack by insects. Alder clones and families differed in percentage leaf area eaten by caterpillars and in leaf defense traits. The concentrations of foliar phenolic compounds negatively correlated with the percentage leaf area eaten by the caterpillars, but the results suggest a threshold, above which the concentration of each of the chemicals appeared to reduce WTC feeding, individually. Particularly, foliar oregonin concentration above 20 % leaf dry weight consistently appeared to reduce feeding by caterpillars. N availability had significant effects on red alder seedling total dry biomass and leaf N concentration. There was a clear trade-off between red alder seedling growth, and content of the phenolic compounds and leaf thickness, which supports the growth-differentiation balanced hypothesis in relation to resource availability. The concentration of oregonin varied during the growing season and there were no significant responses of any of the measured compounds to wounding. The results suggest that red alder foliar oregonin, condensed tannin and total phenolics are constitutive defenses and are not wound-induced. The effects of leaf total phenolic and condensed tannin concentrations on insect herbivory have been documented by past studies but the effects of oregonin concentration in red alder leaves on tent caterpillar feeding is a novel finding. / Graduate / 2020-04-18

Made available in DSpace on 2016-06-02T19:31:34Z (GMT). No. of bitstreams: 1 DissAVFJ.pdf: 842979 bytes, checksum: a746caf5bf4d903cb8d8859514a177a4 (MD5) Previous issue date: 2006-02-20 / Financiadora de Estudos e Projetos / There is growing recognition that classifying terrestrial plant species on the basis of their function (into functional types ) rather than their higher taxonomic identity is a promising way forward for tackling important ecological questions at the scale of communities, landscapes, and biomes. The aim of a plant ecology strategy scheme (PESS) is to express an understanding of important opportunities and selective forces that shape the ecologies of plants and to describe the general principles of plant-environment relations without taxonomic details, to provide a common language for comparing species and vegetation types worldwide. Westoby (1998) proposed a PESS for woody species consisted of three axes: 1) specific leaf area (SLA); 2) height of the canopy of the species; and 3) seed mass. These traits, leaf-height-seed (LHS), are correlated with a number of others and are fundamental trade-offs controlling plant strategies. The aim of this study was to test in a disjoint cerrado woodland site in southeastern Brazil whether traits of the LHS scheme are potential predictors of dispersal guilds. We tried to answer the following question: do the species dispersed by abiotic means present different ecological strategies to species dispersed by animals, considering the LHS scheme? Still, we tested in the same community whether specific leaf area and plant height are potential predictors of two phenological groups, that is, deciduous and evergreen species. We tried to answer the following question: are SLA and plant height related to leaf phenology? According to our results, neither dispersal guilds nor leaf phenological groups could be predicted by the functional traits studied. The similarity in SLA in both cases may be due to two factors: similar height of species (similar habitats in regard to light availability) and sclerophylly. Soil nutrient deficiency seems to lead cerrado woody species to convergent adaptative adjustments regarding both specific leaf area and height. Dispersal guilds were similar concerning seed mass, given that all zoochorous species studied were ornithochorous, whose seed mass is typically low, and abiotically dispersed species had higher than expected seed masses. In the cerrado, the latter may occur as support to the investment in high root-to-shoot ratio of biomass allocation at the seedling stage. Seeds of bird-dispersed species are limited on the size and mass because of the small size of most frugivorous birds. Additionally, in the cerrado, some plants associated with bird dispersers may have their diaspores collected by ants, which favours their seed germination. / 7 A classificação das plantas terrestres baseada em tipos funcionais, em vez de na identidade taxonômica, tem sido reconhecida como uma maneira promissora de lidar com questões ecológicas importantes em nível de comunidades, paisagens e biomas. Os objetivos de um esquema de estratégia ecológica vegetal (EEEV) são o entendimento das forças seletivas que moldam a ecologia das plantas e a descrição dos princípios gerais da relação entre as plantas e o ambiente sem necessidade de detalhamento taxonômico, a fim de que se construa uma linguagem comum para a comparação de espécies e tipos vegetacionais em escala mundial. Westoby (1998) propôs um EEEV para espécies arbustivo-arbóreas, constituído por três eixos: 1) área foliar específica (AFE); 2) altura da copa; e 3) massa da semente. Essas três características, folha-altura-semente (FAS), estão correlacionadas com várias outras e representam comprometimentos fundamentais que controlam as diferentes estratégias ecológicas das plantas. O objetivo deste trabalho foi testar, em uma área disjunta de cerrado sensu stricto em Itirapina SP, se os traços funcionais propostos no esquema FAS são potenciais previsores das guildas de dispersão. Procuramos responder à seguinte pergunta: com base no esquema FAS, as espécies dispersas por mecanismos abióticos apresentam estratégias ecológicas diferentes das espécies dispersas por animais? Ainda, testamos, na mesma comunidade, se a área foliar específica e a altura da planta foram potenciais previsores da fenologia foliar (espécies decíduas e sempre-verdes). Nesse caso, tentamos responder à seguinte pergunta: a área foliar específica e a altura estão relacionadas com o hábito foliar da planta? De acordo com os nossos resultados, os atributos funcionais estudados não puderam prever nem as guildas de dispersão nem a fenologia foliar. A similaridade da área foliar específica nos dois casos pode ser devida a dois fatores: altura similar das espécies (hábitats com disponibilidade de luz parecida) e esclerofilia. A deficiência nutricional do solo deve conduzir as espécies arbóreas de cerrado a ajustes adaptativos convergentes, tanto no que diz respeito à área foliar específica (baixos valores esclerofilia) quanto no que diz respeito à altura. As guildas de dispersão foram similares quanto à massa da semente, tendo em vista que todas as espécies zoocóricas amostradas eram ornitocóricas, cuja massa da semente é tipicamente reduzida, e que as espécies abioticamente dispersas tiveram sementes com massas maiores do que o esperado. No cerrado, isto pode ocorrer em suporte à elevada razão raiz-parte aérea das sementes de algumas espécies do cerrado. Por outro lado, as espécies ornitocóricas são limitadas no tamanho e na massa por causa do pequeno tamanho da maioria das aves frugívoras. Além disso, no cerrado, algumas plantas ornitocóricas podem ter seus diásporos coletados por formigas, o que favoreceria a germinação de suas sementes.

No description available.

Submitted by Ricardo Cedraz Duque Moliterno (ricardo.moliterno@uefs.br) on 2016-01-20T23:17:07Z No. of bitstreams: 1 Ana Paula Lima do Cout1.pdf: 2447904 bytes, checksum: 7c9755ccd95ffbece7c6941079071845 (MD5) / Made available in DSpace on 2016-01-20T23:17:07Z (GMT). No. of bitstreams: 1 Ana Paula Lima do Cout1.pdf: 2447904 bytes, checksum: 7c9755ccd95ffbece7c6941079071845 (MD5) Previous issue date: 2014-09-30 / Conselho Nacional de Pesquisa e Desenvolvimento Cient?fico e Tecnol?gico - CNPq / Funda??o de Amparo ? Pesquisa do Estado da Bahia - FAPEB / (Effect of edge in structure, diversity and phenology in seasonal tropical forest submontane) Forest fragmentation causes physical and biological changes with habitat loss and insularization of populations, and the edge effect the most obvious consequence of this process. Thus, this study aimed to investigate the effect of linear edge in a fragment of submontane seasonal forest in the Chapada Diamantina, Bahia, Brazil (12?28'31 "S and 41?23'14''W) in order to analyze the structure and composition vegetation, reproductive phenology and the vegetative phenology, classifying the forest as their leaf pattern, and the functional diversity. Were marked 30 plots (10 x 10 m) random and 15 at the forest edge (0 - 100m) and 15 inside (> 150m) in the power line high voltage. For structural studies were marked all individuals with diameter at breast height ?5 cm. The species were classified according to successional category and phytosociological parameters were calculated and diversity indices. The Shannon - Wiener index was compared by t test Hutcheson and the other parameters by ANOVA. From this study the species with the highest importance value and abundance for the other investigations were chosen. In the reproductive phenological study monthly observations were made over 26 months in 481 individuals (282 on the edge (61 species) and 199 inside (50 species) of forest), recording the presence and absence of flowering (and flower button) and fruiting (immature and mature fruit) for all life forms. It is estimated timing, frequency and duration of phenological events and the species were classified as the pollination and dispersal syndromes. Data were analyzed with G test, linear regression and circular statistics. In the study of functional diversity created a matrix of functional traits to the edge and interior including phenological, structural and reproductive aspects. Measurements of leaf traits (thickness, dry dough sheet per unit area, juiciness, density, leaf area ratio and leaf mass fraction), density and wood water saturation were carried out in the dry and wet seasons (September / 2012 and January / 2013) in 20 species. Regressions were made between phenophases and environmental variables (precipitation, temperature, relative humidity, photoperiod and heat stroke), circular statistics and comparisons of leaf traits and wood density between the dry and rainy season through the G test and ANOVA. The species ordination in relation to leaf and wood attributes was evaluated by Principal Component Analysis (PCA). The functional groups were defined by cluster analysis with distance Gower and calculated indexes Functional richness (FRic), Functional divergence (FDiv), Functional evenness (FEve), Functional dispersion (FDis). ANOVA and linear regression were used to evaluate the rates between areas. In vegetative phenological study followed up the budding and leaf fall 350 individual trees in the community (39 species). Increase the impact of straight edge on the floristic composition was checked, the diversity and abundance of species, while less interference was observed in the tree community structure, as phytosociological parameters and ratios between the guilds did not differ significantly between edge and interior fragment. The phenological analysis revealed that flowering and fruiting at the edge and interior were continued for long, low-intensity periods, asynchronous and assazonal being for most phenophases, with a significant difference in the individual to button number. Seasonality in flowering and fruiting were observed for bush and grass land in both areas and epiphytes to the edge. Similar proportions in the number of species for pollination and dispersal syndromes were found on the edge and inside, with a predominance of melittophilous and animal dispersed species. The vegetation was evergreen, with budding and continuous leaf fall, low intensity and low sync. The leaf bud was related to rainfall and sunshine and fall with precipitation and humidity. Significant variation between the dry and wet seasons were observed in the proportion of saturated water in the wood, leaf area ratio and leaf succulence. ACP showed greater variance for wood density (88.7%) and stored water (11.3%), gathering most species with standard perenif?lios phenological and episodic perenif?lios, other attributes showed no relevance to the phenological patterns. The main functional groups formed on the edge and inside were similar, defined based on the way of life, stratum and pollination and dispersal. Of co-occurring species (40 species), 22.5% belong to different functional groups due to different growth patterns. The functional diversity indices showed higher mean values for the edge, except that Fric was higher in the forest. These were not affected by the distance to the edge. Changes were observed inside in relation to species richness and Shannon - Wiener only for Fric. The implementation of linear infrastructure brings similar impacts to those coming from other types of edges in community structure. In other aspects analyzed the changes were subtle for the tree community, but the same was not true for shrubs, herbs and vines. The vegetation can be classified as seasonal submontane rainforest always green, with small annual variation in the physiognomy of the vegetation, and the perenif?lios and perenif?lios types episodic predominant both in number of species and in their abundance. The leaf traits showed no relevance to phenology, not being good descriptors for the seasonal evergreen forest. There is no significant difference in functional diversity, but the highest rate obtained inside for functional wealth reflects the biological loss caused by the installation on the edge of the linear edge. Appearance evidenced by the greater abundance of Eschweilera tetrapetala Mori inside, an endemic species of submontane forests of the region. This study contributes to future conservation work to note the interference the linear edge on the forest ecosystem, as well as to the classification of the Brazilian forests, to verify the occurrence of seasonal evergreen forests in the Caatinga. / (Efeito de borda na estrutura, diversidade e fenologia de floresta tropical estacional submontana) A fragmenta??o florestal causa mudan?as f?sicas e biol?gicas com perda de habitat e insulariza??o das popula??es, sendo oefeito de borda a consequ?ncia mais evidente desse processo.Assim,objetivou-se investigar o efeito da borda linear em um fragmento de floresta estacional submontana na Chapada Diamantina, Bahia,Brasil (12?28?31??S e 41?23?14??W) com o intuito de analisar a estruturae a composi??o da vegeta??o, a fenologia reprodutivae a fenologia vegetativa, classificando a floresta quanto ao seu padr?o foliar, e a diversidade funcional. Foram marcadas 30 parcelas (10 x 10 m) aleat?rias sendo 15 na borda da floresta (0 - 100m) e 15 no seu interior (>150m) em rela??o ? rede el?trica de alta tens?o. Para o estudo estrutural foram marcados todos os indiv?duos com di?metro ? altura do peito ?5 cm. As esp?cies foram classificadas quanto ? categoria sucessional e foram calculados os par?metros fitossociol?gicos e ?ndices de diversidade. O ?ndice de Shannon - Wienerfoi comparado pelo teste t de Hutcheson e os demais par?metros pela ANOVA. A partir desse estudo foram escolhidas as esp?cies com maior valor de import?ncia e abundancia para as demais investiga??es.No estudo fenol?gico reprodutivo foram realizadas observa??es mensais durante 26 meses a 481 indiv?duos (282 na borda (61 esp?cies) e 199 no interior (50 esp?cies) da floresta), registrando-se a presen?a e aus?ncia de flora??o (bot?o e flor) e frutifica??o (frutos imaturo e maturo) para todas as formas de vida. Estimou-se sincronia, frequ?ncia e dura??o dos eventos fenol?gicos e as esp?cies foram classificadas quanto ?s s?ndromes de poliniza??o e dispers?o. Os dados foram analisados com teste G, regress?o linear e estat?stica circular. No estudo da diversidade funcional criou-se uma matriz de tra?os funcionais para a borda e interior incluindo aspectos fenol?gicos, estruturais e reprodutivos.Mensura??es de atributos foliares (espessura, massa seca de folha por unidade de ?rea, sucul?ncia, densidade, raz?o de ?rea foliar e fra??o de massa foliar), densidade e ?gua de satura??o da madeira foram realizadas nas esta??es seca e chuvosa (setembro/2012 e janeiro/2013) em 20 esp?cies. Foram feitas regress?es entre as fenofases e as vari?veis ambientais (precipita??o, temperatura, umidade relativa, fotoper?odo e insola??o), estat?stica circular e compara??es dos atributos foliares e densidade de madeira entre a esta??o seca e chuvosa atrav?s do teste G e ANOVA. A ordena??o das esp?cies em rela??o aos atributos foliares e de madeira foi avaliada pela An?lise do Componente Principal (ACP). Os grupos funcionais foram definidos por Cluster analysis com dist?ncia de Gower e calculados os ?ndices Functional richness (FRic), Functional divergence (FDiv), Functional evenness (FEve), Functional dispersion (FDis). ANOVA e regress?o linear foram usadas para avaliar os ?ndices entre ?reas. No estudo fenol?gico vegetativo acompanhou-se o brotamento e queda foliar de 350 indiv?duos arb?reos na comunidade (39 esp?cies). Foi verificado um maior impacto da borda linear sobre a composi??o flor?stica,a diversidade e abund?ncia das esp?cies, enquanto menor interfer?ncia foi verificada na estrutura da comunidade arb?rea, uma vez que par?metros fitossociol?gicos e propor??es entre as categorias sucessionais n?o diferiram significativamente entre borda e interior do fragmento.As an?lises fenol?gicas revelaram que a flora??o e frutifica??o na borda e interior foram cont?nuas por per?odos longos e com baixa intensidade, sendo assincr?nico e assazonal para a maioria das fenofases, com diferen?a significativa no n?mero de indiv?duo para bot?o.Sazonalidade na flora??o e frutifica??o foi observada para arbusto e erva terrestre em ambas as ?reas e ep?fitas para a borda.Propor??es semelhantes no n?mero de esp?cies por s?ndromes de poliniza??o e dispers?o foram encontradas na borda e no interior, com predomin?ncia de esp?cies melit?filas e zooc?ricas.A vegeta??o foi perenif?lia, com brotamento e queda foliar cont?nuos, baixa intensidade e baixa sincronia. O brotamento foliar relacionou-se com precipita??o e insola??o e a queda com precipita??o e umidade. Varia??o significativa entre as esta??es seca e chuvosa foram observadas na propor??o de ?gua saturada na madeira, raz?o de ?rea foliar e sucul?ncia da folha. ACP revelou maior varian?a para densidade da madeira (88,7%) e ?gua armazenada (11,3%), agrupando a maioria das esp?cies com padr?o fenol?gico perenif?lios e perenif?lios epis?dicos, os demais atributos n?o apresentaram relev?ncia em rela??o aos padr?es fenol?gicos. Os principais grupos funcionais formados na borda e no interior foram semelhantes, definidos com base na forma de vida, estrato e s?ndrome de poliniza??o e dispers?o. Das esp?cies co-ocorrentes (40 esp?cies), 22,5% pertencem a grupos funcionais distintos devido a diferentes padr?es fenol?gicos. Os ?ndices de diversidade funcional apresentaram valores m?dios superiores para a borda, exceto FRic que foi superior no interior da floresta. Estes n?o foram alterados pela dist?ncia com a borda. Altera??es foram observadas no interior em rela??o ? riqueza de esp?cie e ao ?ndice de Shannon - Wiener apenas para FRic.A implementa??o de infraestruturas lineares traz impactos semelhantes ?queles advindos de outros tipos de bordas na estrutura da comunidade. Nos demais aspectos analisados as varia??es foram sutis para a comunidade arb?rea, mas o mesmo n?o ocorreuem rela??o a arbustos, ervas e trepadeiras. A vegeta??o p?de ser classificada como floresta tropical estacional submontana sempre verde, com pequena varia??o anual na fisionomia da vegeta??o, sendo os tipos perenif?lios e perenif?lios epis?dicos predominantes tanto em n?mero de esp?cies quanto na sua abund?ncia. Os atributos foliares n?o apresentaram relev?ncia em rela??o aos padr?es fenol?gicos, n?o sendo bons descritores para a floresta estacional sempre verde. N?o h? diferen?a significativa na diversidade funcional, mas o maior ?ndice obtido no interior para a riqueza funcional reflete a perda biol?gica ocasionada na borda pela instala??o da borda linear.Aspecto evidenciado pela maior abund?ncia deEschweilera tetrapetala Mori no interior, esp?cie end?mica das florestas submontanas da regi?o. O presente estudo p?de contribuir com futuros trabalhos de conserva??o por constatar a interfer?ncia da borda linear no ecossistema florestal, bem como com a classifica??o das florestas brasileiras, ao verificar a ocorr?ncia de florestas estacional sempre verde no Bioma Caatinga.

La télédétection des forêts est devenue de plus en plus accessible grâce à l'utilisation de véhicules aériens inoccupés (UAV) et à l'apprentissage profond, ce qui permet d'obtenir des images répétées à haute résolution et d’observer les changements phénologiques à des échelles spatiales et temporelles plus importantes. Dans les forêts tempérées, à l'automne, la sénescence des feuilles se produit lorsque les feuilles changent de couleur et tombent. Cependant, l'influence de la sénescence foliaire sur la segmentation des espèces d'arbres à l'aide d'un réseau neuronal convolutif (CNN) n'a pas encore été évaluée. Nous avons acquis de l’imagerie haute résolution par UAV au-dessus d’une forêt tempérée au Québec à sept reprises entre mai et octobre 2021. Nous avons segmenté et identifié 23 000 couronnes d'arbres de 14 classes différentes pour entraîner et valider un CNN pour chaque acquisition d'imagerie. La meilleure segmentation (F1-score le plus élevé) était au début de la coloration des feuilles (début septembre) et le F1-score le plus bas au pic de la coloration automnale (début octobre). La chronologie de la sénescence varie considérablement d’une espèce à l’autre et au sein d’une même espèce, ce qui entraîne une grande variabilité du signal télédétecté. Les espèces d'arbres à feuilles caduques et à feuilles persistantes qui présentaient des traits distinctifs et moins variables dans le temps entre les individus ont été mieux classées. Bien que la segmentation des arbres dans une forêt hétérogène demeure un défi, l'imagerie UAV et l'apprentissage profond démontrent un grand potentiel pour la cartographie des espèces d'arbres. Les résultats obtenus dans une forêt tempérée où la couleur des feuilles change fortement pendant la sénescence automnale montrent que la meilleure performance pour la segmentation des espèces d'arbres se produit au début de ce changement de couleur. / Remote sensing of forests has become increasingly accessible with the use of unoccupied aerial vehicles (UAV), along with deep learning, allowing for repeated high-resolution imagery and the capturing of phenological changes at larger spatial and temporal scales. In temperate forests during autumn, leaf senescence occurs when leaves change colour and drop. However, the influence of leaf senescence in temperate forests on tree species segmentation using a Convolutional Neural Network (CNN) has not yet been evaluated. Here, we acquired high-resolution UAV imagery over a temperate forest in Quebec, Canada on seven occasions between May and October 2021. We segmented and labelled 23,000 tree crowns from 14 different classes to train and validate a CNN for each imagery acquisition. The CNN-based segmentation showed the highest F1-score (0.72) at the start of leaf colouring in early September and the lowest F1-score (0.61) at peak fall colouring in early October. The timing of the events occurring during senescence, such as leaf colouring and leaf fall, varied substantially between and within species and according to environmental conditions, leading to higher variability in the remotely sensed signal. Deciduous and evergreen tree species that presented distinctive and less temporally-variable traits between individuals were better classified. While tree segmentation in a heterogenous forest remains challenging, UAV imagery and deep learning show high potential in mapping tree species. Our results from a temperate forest with strong leaf colour changes during autumn senescence show that the best performance for tree species segmentation occurs at the onset of this colour change.