Natural forest regeneration through fire protection is a less imminent threat for truly stable savannas than afforestation.

African bistable savannas have important biodiversity value and merit conservation. At the same time, forest restoration is a nature-based solution that can be used to increase biodiversity, carbon stocks, and human well-being. Here we describe an experiment based on natural forest regeneration through the exclusion of anthropogenic fire. We show that it is easier to let nature do its work instead of channeling it into an artificial man-made ecosystem through human-induced burning or planting. We emphasize that nature-based solutions must be biome-appropriate and the choice between restoring forests or protecting savannas requires a thorough understanding of the local context.

Historical tree phenology data reveal the seasonal rhythms of the Congo Basin rainforest.

Tropical forest phenology directly affects regional carbon cycles, but the relation between species-specific and whole-canopy phenology remains largely uncharacterized. We present a unique analysis of historical tropical tree phenology collected in the central Congo Basin, before large-scale impacts of human-induced climate change. Ground-based long-term (1937-1956) phenological observations of 140 tropical tree species are recovered, species-specific phenological patterns analyzed and related to historical meteorological records, and scaled to characterize stand-level canopy dynamics. High phenological variability within and across species and in climate-phenology relationships is observed. The onset of leaf phenophases in deciduous species was triggered by drought and light availability for a subset of species and showed a species-specific decoupling in time along a bi-modal seasonality. The majority of the species remain evergreen, although central African forests experience relatively low rainfall. Annually a maximum of 1.5% of the canopy is in leaf senescence or leaf turnover, with overall phenological variability dominated by a few deciduous species, while substantial variability is attributed to asynchronous events of large and/or abundant trees. Our results underscore the importance of accounting for constituent signals in canopy-wide scaling and the interpretation of remotely sensed phenology signals.

Protecting an artificial savanna as a nature-based solution to restore carbon and biodiversity in the Democratic Republic of the Congo.

A large share of the global forest restoration potential is situated in artificial 'unstable' mesic African savannas, which could be restored to higher carbon and biodiversity states if protected from human-induced burning. However, uncertainty on recovery rates in protected unstable savannas impedes science-informed forest restoration initiatives. Here, we quantify the forest restoration success of anthropogenic fire exclusion within an 88-ha mesic artificial savanna patch in the Kongo Central province of the Democratic Republic of the Congo (DR Congo). We found that aboveground carbon recovery after 17 years was on average 11.40 ± 0.85 Mg C ha-1 . Using a statistical model, we found that aboveground carbon stocks take 112 ± 3 years to recover to 90% of aboveground carbon stocks in old-growth forests. Assuming that this recovery trajectory would be representative for all unstable savannas, we estimate that they could have a total carbon uptake potential of 12.13 ± 2.25 Gt C by 2100 across DR Congo, Congo and Angola. Species richness recovered to 33.17% after 17 years, and we predicted a 90% recovery at 54 ± 2 years. In contrast, we predicted that species composition would recover to 90% of old-growth forest composition only after 124 ± 3 years. We conclude that the relatively simple and cost-efficient measure of fire exclusion in artificial savannas is an effective nature-based solution to climate change and biodiversity loss. However, more long-term and in situ monitoring efforts are needed to quantify variation in long-term carbon and diversity recovery pathways. Particular uncertainties are spatial variability in socio-economics and growing conditions as well as the effects of projected climate change.

Tree Core Analysis with X-ray Computed Tomography.

An X-ray computed tomography (CT) toolchain is presented to obtain tree-ring width (TRW), maximum latewood density (MXD), other density parameters, and quantitative wood anatomy (QWA) data without the need for labor-intensive surface treatment or any physical sample preparation. The focus here is on increment cores and scanning procedures at resolutions ranging from 60 µm down to 4 µm. Three scales are defined at which wood should be looked at: (i) inter-ring scale, (ii) ring scale, i.e., tree-ring analysis and densitometry scale, as well as (iii) anatomical scale, the latter approaching the conventional thin-section quality. Custom-designed sample holders for each of these scales enable high-throughput scanning of multiple increment cores. A series of software routines were specifically developed to efficiently treat three-dimensional X-ray CT images of the tree cores for TRW and densitometry. This work briefly explains the basic principles of CT, which are needed for a proper understanding of the protocol. The protocol is presented for some known species that are commonly used in dendrochronology. The combination of rough density estimates, TRW and MXD data, as well as quantitative anatomy data, allows us to broaden and deepen current analyses for climate reconstructions or tree response, as well as further develop the field of dendroecology/climatology and archeology.

Heading for a fall: The fate of old wind-thrown beech trees (Fagus sylvatica) is detectable in their growth pattern.

Common beech (Fagus sylvatica) is one of the most important deciduous tree species in European forests. However, climate-change-induced drought may threaten its dominant position. The Sonian Forest close to Brussels (Belgium) is home to some of the largest beech trees in the world. This UNESCO world heritage site is famous for its high density of very large beech trees as a result of its climatic suitability, fertile soil conditions, and past management. Here we utilized tree-ring data from increment cores to investigate the growth of these old and monumental beech trees, evaluating their growth trends, response to past climate, and the effect of mast years on 39 living and 16 recently wind-thrown trees. Our analysis reveals that the sampled trees were generally sensitive to spring and summer droughts but recovered quickly after such an extreme climatic event. The growth trend of living trees has remained high and only shows a slight, statistically insignificant, decline over the past 50 years. Although the overall growth rate remains strong (BAI 50 cm2/year), the past five decades have shown strong inter-annual growth variations due to frequent and more intense droughts combined with an increased frequency of mast years. We also found notable differences in growth patterns between the living trees and those that had recently been wind-thrown. While there were no significant differences between living and wind-thrown trees in response to droughts, heatwaves, or mast years when examining year-to-year growth changes, the wind-thrown trees did exhibit considerably lower overall growth rates and a significant downward trend in growth (BAI -0.57 cm2/year). This difference in growth trends has been apparent since at least the 1980s. Overall, the findings of this study can provide valuable insights for understanding the long-term dynamics of lowland beech forests and their responses to climate change.

Earlier onset and slower heartwood investment in faster-growing trees of African tropical species.

BACKGROUND AND AIMS: Heartwood plays an important role in maintaining the structural integrity of trees. While, its formation has long be thought to be solely driven by internal ageing processes, more recent hypotheses suggest that heartwood formation acts as a regulator of the tree water balance by modulating sapwood quantities. Testing both hypotheses would shed light on the potential ecophysiological nature of heartwood formation, a very common process in trees. METHODS: We measured heartwood and sapwood quantities, xylem conduits and growth ring width and number on 406 stems of Pericopsis elata with ages ranging from 2 to 237 years. A subset of 17 trees with similar ages but varying growth rate were sampled in a shaded (slower growth) and sun-exposed (faster growth) site. We used regression analysis and structural equations modelling to investigate the dynamics and drivers of heartwood formation. KEY RESULTS: We found a positive effect of growth rate on the probability of heartwood occurrence, suggesting an earlier heartwood onset in faster-growing stems. After this onset age, heartwood area increases with stem diameter and age. Despite the similar heartwood production per unit stem diameter increment, shaded trees produce heartwood faster than sun-exposed trees. Tree age and hydraulics showed similar direct effects on heartwood and sapwood area of sun-exposed trees, suggesting their mutual role in driving the heartwood dynamics of sun-exposed trees. However, for shaded trees, only tree hydraulics showed a direct effect, suggesting its prominent role over age in driving the heartwood dynamics in limited growing conditions. The positive relationship between growth rate and maximum stomatal conductance supports this conclusion. CONCLUSIONS: Heartwood area increases as the tree ages but at a slower rate in trees where water demand is balanced by a sufficient water supply. Our findings suggest that heartwood formation is not only a structural but also a functional process.

SmartWoodID-an image collection of large end-grain surfaces to support wood identification systems.

Wood identification is a key step in the enforcement of laws and regulations aimed at combatting illegal timber trade. Robust wood identification tools, capable of distinguishing a large number of timbers, depend on a solid database of reference material. Reference material for wood identification is typically curated in botanical collections dedicated to wood consisting of samples of secondary xylem of lignified plants. Specimens from the Tervuren Wood Collection, one of the large institutional wood collections around the world, are used as a source of tree species data with potential application as timber. Here, we present SmartWoodID, a database of high-resolution optical scans of the end-grain surfaces enriched with expert wood anatomical descriptions of macroscopic features. These can serve as annotated training data to develop interactive identification keys and artificial intelligence for computer vision-based wood identification. The first edition of the database consists of images of 1190 taxa, with a focus on potential timber species from the Democratic Republic of the Congo with at least four different specimens per species included. Database URL https://hdl.handle.net/20.500.12624/SmartWoodID_first_edition.

Whole genome duplication of wild-type and CINNAMYL ALCOHOL DEHYDROGENASE1-downregulated hybrid poplar reduces biomass yield and causes a brittle apex phenotype in field-grown wild types.

The potential of whole genome duplication to increase plant biomass yield is well-known. In Arabidopsis tetraploids, an increase in biomass yield was accompanied by a reduction in lignin content and, as a result, a higher saccharification efficiency was achieved compared with diploid controls. Here, we evaluated whether the results obtained in Arabidopsis could be translated into poplar and whether the enhanced saccharification yield upon alkaline pretreatment of hairpin-downregulated CINNAMYL ALCOHOL DEHYDROGENASE1 (hpCAD) transgenic poplar could be further improved upon a whole genome duplication. Using a colchicine treatment, wild-type (WT) Populus tremula x P. alba cv. INRA 717-1B4, a commonly used model clone in tree biotechnology research, and hpCAD tetraploids were generated and grown in the greenhouse. In parallel, WT tetraploid poplars were grown in the field. In contrast to Arabidopsis, a whole genome duplication of poplar had a negative impact on the biomass yield of both greenhouse- and field-grown trees. Strikingly, field-grown WT tetraploids developed a brittle apex phenotype, i.e., their tip broke off just below the apex. In addition, the chromosome doubling altered the biomass composition of field-grown, but not of greenhouse-grown tetraploid poplars. More specifically, the lignin content of field-grown tetraploid poplars was increased at the expense of matrix polysaccharides. This increase in lignin deposition in biomass is likely the cause of the observed brittle apex phenotype, though no major differences in stem anatomy or in mechanical properties could be found between di- and tetraploid WT poplars grown in the field. Finally, without biomass pretreatment, the saccharification efficiency of greenhouse- and field-grown WT diploids was not different from that of tetraploids, whereas that of greenhouse-grown hpCAD tetraploids was higher than that of greenhouse-grown diploids. Upon alkaline pretreatment, the saccharification yield of diploids was similar to that of tetraploids for all genotypes and growth conditions tested. This study showed that a whole genome duplication in hybrid WT and hpCAD poplar did neither result in further improvements in biomass yield, nor in improved biomass composition and, hence, saccharification performance.

Detecting thin adhesive coatings in wood fiber materials with laboratory-based dual-energy computed tomography (DECT).

The distribution and good spreading of adhesive resins is critical for the wood-based panels industry. Full 3D non-destructive characterization is necessary, but methods are limited due to the chemical similarities between the resins and the wood fibers. For X-ray microtomography ([Formula: see text]CT), the doping of the resin with a highly attenuating contrast agent is necessary to visualize the resin distribution. However, the attenuation signal remains hard to segment clearly due to partial volume effects in the image, and phase mixing in the material. To help in the identification of the doped resin, dual-energy X-ray CT (DECT) is used to exploit the contrast agent's K-edge, based on simulations which take into account the polychromatic properties of the X-ray tube and detector response. The contrast agent's identification with DECT is validated with elemental mapping using scanning electron microscopy combined with energy-dispersive spectroscopy (SEM-EDX) on the surface of a wood-based panel sample, using data fusion between DECT and SEM-EDX. Overall, DECT results here in the first 3D identification of doped resin inside wood fiberboards, guiding the industry's efforts in further improving the durability of wood-based panels.

Moisture Dynamics of Wood-Based Panels and Wood Fibre Insulation Materials.

Moisture performance is an important factor determining the resistance of wood-based building materials against fungal decay. Understanding how material porosity and chemistry affect moisture performance is necessary for their efficient use, as well as for product optimisation. In this study, three complementary techniques (X-ray computed tomography, infrared and low-field NMR spectroscopy) are applied to elucidate the influence of additives, manufacturing process and material structure on the liquid water absorption and desorption behaviour of a selection of wood-based panels, thermally modified wood and wood fibre insulation materials. Hydrophobic properties achieved by thermal treatment or hydrophobic additives such as paraffin and bitumen, had a major influence on water absorption and desorption rates. When hydrophobic additives did not play a role, pore distributions and manufacturing process had a decisive influence on the amount and rate of absorption and desorption. In that case, a higher porosity resulted in a higher water absorption rate. Our results show that there is a clear potential for tailoring materials towards specific moisture performance by better understanding the influence of different material characteristics. This is useful both for achieving desired moisture buffering as well as to increase service life of wood-based materials. From a sustainability perspective, fit-for-purpose moisture performance is often easier to achieve and preferred than wood protection by biocide preservative treatments.

Improved wood species identification based on multi-view imagery of the three anatomical planes.

BACKGROUND: The identification of tropical African wood species based on microscopic imagery is a challenging problem due to the heterogeneous nature of the composition of wood combined with the vast number of candidate species. Image classification methods that rely on machine learning can facilitate this identification, provided that sufficient training material is available. Despite the fact that the three main anatomical sections contain information that is relevant for species identification, current methods only rely on transverse sections. Additionally, commonly used procedures for evaluating the performance of these methods neglect the fact that multiple images often originate from the same tree, leading to an overly optimistic estimate of the performance. RESULTS: We introduce a new image dataset containing microscopic images of the three main anatomical sections of 77 Congolese wood species. A dedicated multi-view image classification method is developed and obtains an accuracy (computed using the naive but common approach) of 95%, outperforming the single-view methods by a large margin. An in-depth analysis shows that naive accuracy estimates can lead to a dramatic over-prediction, of up to 60%, of the accuracy. CONCLUSIONS: Additional images from non-transverse sections can boost the performance of machine-learning-based wood species identification methods. Additionally, care should be taken when evaluating the performance of machine-learning-based wood species identification methods to avoid an overestimation of the performance.

High-Resolution X-Ray Computed Tomography: A New Workflow for the Analysis of Xylogenesis and Intra-Seasonal Wood Biomass Production.

Understanding tree growth and carbon sequestration are of crucial interest to forecast the feedback of forests to climate change. To have a global understanding of the wood formation, it is necessary to develop new methodologies for xylogenesis measurements, valid across diverse wood structures and applicable to both angiosperms and gymnosperms. In this study, the authors present a new workflow to study xylogenesis using high-resolution X-ray computed tomography (HRXCT), which is generic and offers high potential for automatization. The HXRCT-based approach was benchmarked with the current classical approach (microtomy) on three tree species with contrasted wood anatomy (Pinus nigra, Fagus sylvatica, and Quercus robur). HRXCT proved to estimate the relevant xylogenesis parameters (timing, duration, and growth rates) across species with high accuracy. HRXCT showed to be an efficient avenue to investigate tree xylogenesis for a wide range of wood anatomies, structures, and species. HRXCT also showed its potential to provide quantification of intra-annual dynamics of biomass production through high-resolution 3D mapping of wood biomass within the forming growth ring.

Timing of spring xylogenesis in temperate deciduous tree species relates to tree growth characteristics and previous autumn phenology.

We explored the timing of spring xylogenesis and its potential drivers in homogeneous mature forest stands in a temperate European region. Three species with contrasting leaf development dynamics and wood anatomy were studied: European beech, silver birch and pedunculate oak. Detailed phenological observations of xylogenesis and leaf phenology were performed from summer 2017 until spring 2018. Cambium reactivation (CR) occurred before the buds of oak and birch were swollen, whereas these two phenological phases were concurrent for beech. On the other hand, initial earlywood vessels were fully differentiated (FDIEV) after leaf unfolding for all three species. Timing of CR was correlated to average ring-width of the last 10 years (2008-17), tree diameter and, partially, with tree age. In addition, the timing of FDIEV was correlated to tree age and previous year's autumn phenology, i.e., timing of wood growth cessation and onset of leaf senescence. Multivariate models could explain up to 68% of the variability of CR and 55% of the variability of FDIEV. In addition to the 'species' factor, the variability could be explained by ca 30% by tree characteristics and previous year's autumn phenology for both CR and FDIEV. These findings are important to better identify which factors (other than environment) can be driving the onset of the growing season, and highlight the influence of tree growth characteristics and previous year's phenology on spring wood phenology, wood formation and, potentially, forest production.

From leaf to label: A robust automated workflow for stomata detection.

Plant leaf stomata are the gatekeepers of the atmosphere-plant interface and are essential building blocks of land surface models as they control transpiration and photosynthesis. Although more stomatal trait data are needed to significantly reduce the error in these model predictions, recording these traits is time-consuming, and no standardized protocol is currently available. Some attempts were made to automate stomatal detection from photomicrographs; however, these approaches have the disadvantage of using classic image processing or targeting a narrow taxonomic entity which makes these technologies less robust and generalizable to other plant species. We propose an easy-to-use and adaptable workflow from leaf to label. A methodology for automatic stomata detection was developed using deep neural networks according to the state of the art and its applicability demonstrated across the phylogeny of the angiosperms.We used a patch-based approach for training/tuning three different deep learning architectures. For training, we used 431 micrographs taken from leaf prints made according to the nail polish method from herbarium specimens of 19 species. The best-performing architecture was tested on 595 images of 16 additional species spread across the angiosperm phylogeny.The nail polish method was successfully applied in 78% of the species sampled here. The VGG19 architecture slightly outperformed the basic shallow and deep architectures, with a confidence threshold equal to 0.7 resulting in an optimal trade-off between precision and recall. Applying this threshold, the VGG19 architecture obtained an average F-score of 0.87, 0.89, and 0.67 on the training, validation, and unseen test set, respectively. The average accuracy was very high (94%) for computed stomatal counts on unseen images of species used for training.The leaf-to-label pipeline is an easy-to-use workflow for researchers of different areas of expertise interested in detecting stomata more efficiently. The described methodology was based on multiple species and well-established methods so that it can serve as a reference for future work.

Inter-individual variability in spring phenology of temperate deciduous trees depends on species, tree size and previous year autumn phenology.

We explored the inter-individual variability in bud-burst and its potential drivers, in homogeneous mature stands of temperate deciduous trees. Phenological observations of leaves and wood formation were performed weekly from summer 2017 to summer 2018 for pedunculate oak, European beech and silver birch in Belgium. The variability of bud-burst was correlated to previous' year autumn phenology (i.e. the onset of leaf senescence and the cessation of wood formation) and tree size but with important differences among species. In fact, variability of bud-burst was primarily related to onset of leaf senescence, cessation of wood formation and tree height for oak, beech and birch, respectively. The inter-individual variability of onset of leaf senescence was not related to the tree characteristics considered and was much larger than the inter-individual variability in bud-burst. Multi-species multivariate models could explain up to 66% of the bud-burst variability. These findings represent an important advance in our fundamental understanding and modelling of phenology and tree functioning of deciduous tree species.

Timeline of autumn phenology in temperate deciduous trees.

Cessation of xylem formation or wood growth (CWG) and onset of foliar senescence (OFS) are key autumn phenological events in temperate deciduous trees. Their timing is fundamental for the development and survival of trees, ecosystem nutrient cycling and the seasonal exchange of matter and energy between the biosphere and atmosphere, and affects the impact and feedback of forests to global change. A large-scale experimental effort and improved observational methods have allowed us to compare the timing of CWG and OFS for different deciduous tree species in Western Europe, particularly in silver birch, a pioneer species, and European beech, a late-succession species, at stands of different latitudes, of different levels of site fertility, for 2 years with contrasting meteorological and drought conditions, i.e., the low moderately dry 2017 and the extremely dry 2018. Specifically, we tested whether foliar senescence started before, after or concurrently with CWG. Onset of foliar senescence and CWG occurred generally between late September and early November, with larger differences across species and sites for OFS. Foliar senescence started concurrently with CWG in most cases, except for the drier 2018 and, for beech, at the coldest site, where OFS occurred significantly later than CWG. The behavior of beech in Spain, the southern edge of its European distribution, was unclear, with no CWG, but very low wood growth at the time of OFS. Our study suggests that OFS is generally triggered by the same drivers of CWG or when wood growth decreases in late summer, indicating an overarching mechanism of sink limitation as a possible regulator of the timing of foliar senescence.

Advanced X-ray CT scanning can boost tree ring research for earth system sciences.

BACKGROUND AND AIMS: Tree rings, as archives of the past and biosensors of the present, offer unique opportunities to study influences of the fluctuating environment over decades to centuries. As such, tree-ring-based wood traits are capital input for global vegetation models. To contribute to earth system sciences, however, sufficient spatial coverage is required of detailed individual-based measurements, necessitating large amounts of data. X-ray computed tomography (CT) scanning is one of the few techniques that can deliver such data sets. METHODS: Increment cores of four different temperate tree species were scanned with a state-of-the-art X-ray CT system at resolutions ranging from 60 µm down to 4.5 µm, with an additional scan at a resolution of 0.8 µm of a splinter-sized sample using a second X-ray CT system to highlight the potential of cell-level scanning. Calibration-free densitometry, based on full scanner simulation of a third X-ray CT system, is illustrated on increment cores of a tropical tree species. KEY RESULTS: We show how multiscale scanning offers unprecedented potential for mapping tree rings and wood traits without sample manipulation and with limited operator intervention. Custom-designed sample holders enable simultaneous scanning of multiple increment cores at resolutions sufficient for tree ring analysis and densitometry as well as single core scanning enabling quantitative wood anatomy, thereby approaching the conventional thin section approach. Standardized X-ray CT volumes are, furthermore, ideal input imagery for automated pipelines with neural-based learning for tree ring detection and measurements of wood traits. CONCLUSIONS: Advanced X-ray CT scanning for high-throughput processing of increment cores is within reach, generating pith-to-bark ring width series, density profiles and wood trait data. This would allow contribution to large-scale monitoring and modelling efforts with sufficient global coverage.

The stability enigma of hydraulic vulnerability curves: addressing the link between hydraulic conductivity and drought-induced embolism.

Maintaining xylem water transport under drought is vital for plants, but xylem failure does occur when drought-induced embolisms form and progressively spread through the xylem. The hydraulic method is widely considered the gold standard to quantify drought-induced xylem embolism. The method determines hydraulic conductivity (Kh) in cut branch samples, dehydrated to specific drought levels, by pushing water through them. The technique is widely considered for its reliable Kh measurements, but there is some uncertainty in the literature over how to define stable Kh and how that relates to the degree of xylem embolism formation. Therefore, the most common setup for this method was extended to measure four parameters: (i) inlet Kh, (ii) outlet Kh, (iii) radial flow from xylem to surrounding living tissue and (iv) the pressure difference across the sample. From a strictly theoretical viewpoint, hydraulic steady state, where inflow equals outflow and radial flow is zero, will result in stable Kh. Application of the setup to Malus domestica Borkh. branches showed that achieving hydraulic steady state takes considerable time (up to 300 min) and that time to reach steady state increased with declining xylem water potentials. During each experimental run, Kh and xylem water potentials dynamically increased, which was supported by X-ray computed microtomography visualizations of embolism refilling under both high- (8 kPa) and low-pressure (2 kPa) heads. Supplying pressurized water can hence cause artificial refilling of vessels, which makes it difficult to achieve a truly stable Kh in partially embolized xylem.

Asynchronous leaf and cambial phenology in a tree species of the Congo Basin requires space-time conversion of wood traits.

BACKGROUND AND AIMS: Wood traits are increasingly being used to document tree performance. In the Congo Basin, however, weaker seasonality causes asynchrony of wood traits between trees. Here, we monitor growth and phenology data to date the formation of traits. METHODS: For two seasons, leaf and cambial phenology were monitored on four Terminalia superba trees (Mayombe) using cameras, cambial pinning and dendrometers. Subsequently, vessel lumen and parenchyma fractions as well as high-resolution isotopes (δ13C/δ18O) were quantified on the formed rings. All traits were dated and related to weather data. KEY RESULTS: We observed between-tree differences in green-up of 45 d, with trees flushing before and after the rainy season. The lag between green-up and onset of xylem formation was 59 ± 21 d. The xylem growing season lasted 159 ± 17 d with between-tree differences of up to 53 d. Synchronized vessel, parenchyma and δ13C profiles were related to each other. Only parenchyma fraction and δ13C were correlated to weather variables, whereas the δ18O pattern showed no trend. CONCLUSIONS: Asynchrony of leaf and cambial phenology complicates correct interpretation of environmental information recorded in wood. An integrated approach including high-resolution measurements of growth, stable isotopes and anatomical features allows exact dating of the formation of traits. This methodology offers a means to explore the asynchrony of growth in a rainforest and contribute to understanding this aspect of forest resilience.

The persistence of carbon in the African forest understory.

Quantifying carbon dynamics in forests is critical for understanding their role in long-term climate regulation1-4. Yet little is known about tree longevity in tropical forests3,5-8, a factor that is vital for estimating carbon persistence3,4. Here we calculate mean carbon age (the period that carbon is fixed in trees7) in different strata of African tropical forests using (1) growth-ring records with a unique timestamp accurately demarcating 66 years of growth in one site and (2) measurements of diameter increments from the African Tropical Rainforest Observation Network (23 sites). We find that in spite of their much smaller size, in understory trees mean carbon age (74 years) is greater than in sub-canopy (54 years) and canopy (57 years) trees and similar to carbon age in emergent trees (66 years). The remarkable carbon longevity in the understory results from slow and aperiodic growth as an adaptation to limited resource availability9-11. Our analysis also reveals that while the understory represents a small share (11%) of the carbon stock12,13, it contributes disproportionally to the forest carbon sink (20%). We conclude that accounting for the diversity of carbon age and carbon sequestration among different forest strata is critical for effective conservation management14-16 and for accurate modelling of carbon cycling4.