Ecology

Assembly and annotation of the chloroplast genomesAssembly resulted in a whole cp genome sequence of C. hirtinoda with a length of 139, 561 bp (Fig. 1), consisting of 83, 166 bp large single-copy region, 20, 811 bp small single-copy regions, and two 21,792 bp IR regions, comprising the typical quadripartite structure of terrestrial plants. The cp genome of C. hirtinoda was annotated with 130 genes, including 85 protein-coding genes, 37 tRNA genes, and 8 rRNA genes (Table 1). Most of the 15 genes in the C. hirtinoda cp genome contain introns. Of these, 13 genes contain one intron (atpF, ndhA, ndhB, petB, petD, rpl2, rpl16, rps16, trnA-UGC, trnI-GAU, trnK-UUU, trnL-UAA, trnV-UAC) and only the gene cyf3 includes two introns, and the gene clpP intron was deleted (Supplementary Table S1). The rps12 gene contained two copies, and the three exons were spliced into a trans-splicing gene18.Figure 1Chloroplast genome map of C. hirtinoda. Different colors represent different functional genes groups. Genes outside the circle indicate counterclockwise transcription, and genes inside the clockwise transcription. The thick black line on the outer circle represents the two IR regions. The GC content is the dark gray area within the ring.Full size imageTable 1 Summary of the chloroplast genome of C. hirtinoda.Full size tableThe accD, ycf1, and ycf2 genes were missing in the cp genome of C. hirtinoda, and the introns in the genes clpP and rpoC1 were lost. This phenomenon is consistent with previous systematic evolutionary studies on the genome structure of plants in the Poaceae family19. The phenomenon of missing genes is reported in other plants20,21,22,23.The total GC content in the C. hirtinoda cp genome was 38.90%, and the content for each of the four bases, A, T, G, and C, was 30.63%, 30.46%, 19.57%, and 19.33%, respectively (Table 2). The LSC region (36.98%) and SSC region (33.21%) exhibited much lower values than the IR region (44.23%), indicating a non-uniform distribution of the base contents in the cp genome, probably because of four rRNAs in the IR region, which in turn makes the GC content higher in the IR region. These values were similar to cp genome results previously reported for some Poaceae plants24,25.Table 2 Base composition in the C. hirtinoda choloroplast genome.Full size tableRepeat sequences and codon analysisSSR consists of 10-bp-long base repeats and is widely used for exploring phylogenetic evolution and genetic diversity analysis26,27,28,29.In total, 48 SSRs were detected in C. hirtinoda, including 27 mononucleotide versions, accounting for 56.25% of the total SSRs, primarily consisting of A or T. Additionally, four dinucleotide repeats consisting of AT/TA and TC/CT repeats, and 3 tri, 13 tetra, and 1penta-repeats (Fig. 2A). From the SSRs distribution perspective, the majority (79%) of SSRs (38) were observed in the LSC area, whereas 6 SSRs in the IR region (13%) and 4 SSRs in the SSC region (8%) were discovered (Fig. 2B). Previous research suggests that the distribution of SSRs numbers in each region and the differences among locations in GC content are related to the expansion or contraction of the IR boundary30.Figure 2Analysis of simple sequence repeats in C. hirtinoda cp genome. (A) The percentage distribution of 45 SSRs in LSC, SSC, and IR regions. (B).Full size imageThe REPuter program revealed that the cp genome of C. hirtinoda was identified with 61 repeats, consisting of 15 palindromic, 19 forward and no reverse and complement repeats (Fig. 3). We noticed that repeat analyses of three Chimonobambusa genus species exhibited 61–65 repeats, with only one reverse in C. hejiangensis. Most of the repeat lengths were between 30 and 100 bp, and the repeat sequences were located in either IR or LSC region31 (Supplementary Table S2).Figure 3Information of chloroplast genome repeats of Chimonobambusa genus species.Full size imageWe identified 20,180 codons in the coding region of C. hirtinoda (Fig. 4, Supplementary Table S3). The codon AUU of Ile was the most used, and the TER of UAG was the least used codon (817 and 19), excluding the termination codons. Leu was the most encoded amino acid (2,170), and TER was the lowest (85). The Relative Synonymous Codon Usage (RSCU) value greater than 1.0 means a codon is used more frequently32. The RSCU values for 31 codons exceeded 1 in the C. hirtinoda cp genome, and of these, the third most frequent codon was A/U with 29 (93.55%), and the frequency of start codons AUG and UGG used demonstrated no bias (RSCU = 1).Figure 4Amino acid frequencies in C. hirtinoda cp genome protein coding sequences. The column diagrams indicate the number of amino acid codes, and the broken line indicates the proportion of amino acid codes.Full size imageComparative analysis of genome structureThe nucleotide variability (Pi) values of the three cp genomes discovered in the Chimonobambusa genus species ranged from 0 to 0.021 with an average value of 0.000544, as demonstrated from DnaSP 5.10 software analysis. Five peaks were observed in the two single-copy regions, and the highest peak was present in the trnT-trnE-trnY region of the LSC region (Fig. 5). The Pi value for LSC and SSC is significantly higher than that of the IR region. In the IR region, highly different sequences were not observed, a highly conserved region. The sequences of these highly variable regions are reported in other plants during examinations for species identification, phylogenetic analysis, and population genetics research33,34,35.Figure 5Sliding window analysis of Chimonobambusa genus complete chloroplast genome sequences. X-axis: position of the midpoint of a window, Y-axis: nucleotide diversity of each window.Full size imageThe structural information for the complete cp genomes among three Chimonobambusa genus species revealed that the sequences in most regions were conserved (Fig. 6). The LSC and SSC regions exhibit a remarkable degree of variation, higher than the IR region, and the non-coding region demonstrates higher variability than the coding region. In the non-coding areas, 7–9 k, 28–30 k, 36 k and other gene loci differed significantly. Genes rpoC2, rps19, ndhJ and other regions differ in the protein-coding region. However, the agreement between the tRNA and rRNA regions is 100%. A similar phenomenon has also been reported by others36.Figure 6Visualization of genome alignment of three species chloroplast genome sequences using Chimonobambusa hejiangensis as reference. The vertical scale shows the percent of identity, ranging from 50 to 100%. The horizontal axis shows the coordinates within the cp genome. Those are some colors represents protein coding, intron, mRNA and conserved non-coding sequence, respectively.Full size imageIR contraction and expansion in the chloroplast genomeDue to the unique circular structure of the cp genome, there are four junctions between the LSC/IRB/SSC/IRA regions. During species evolution, the stability of the two IR regions sequences was ensured by the IR region of the chloroplast genome expanding and contracting to some degree, and this adjustment is the primary reason for chloroplast genome length variation37,38.The variations at IR/SC boundary regions in the three Chimonobambusa genus chloroplast genomes were highly similar in the organization, gene content, and gene order. The size of IR ranges from 21,797 bp (C. tumidissinoda) to 21,835 bp (C. hejiangensis). The ndhH gene spans the SSC/IRa boundary, and this gene extended 181–224 bp into the IRa region for all three Chimonobambusa genus. The gene rps19 was extended from the IRb to the LSC region with a 31–35 bp gap. The rpl12 gene was located in the LSC region of all genomes, varied from 35–36 bp apart from the LSC/IRb (Fig. 7).Figure 7Comparison of LSC, SSC and IR boundaries of chloroplast genomes among the three Chimonobambusa species. The LSC, SSC and IRs regions are represented with different colors. JLB, JSB, JSA and JLA represent the connecting sites between the corresponding regions of the genome, respectively. Genes are showed by boxes.Full size imageThree chloroplast genomes of the Chimonobambusa genus were compared using the Mauve alignment. The results showed that all sequences show perfect synteny conservation with no inversion or rearrangements (Fig. 8).Figure 8The chloroplast genomes of three Chimonobambusa species rearranged by the software MAUVE. Locally collinear blocks (LCBs) are represented by the same color blocks connected by lines. The vertical line indicates the degree of conservatism among position. The small red bar represents rRNA.Full size imagePhylogenetic analysisWe performed a phylogenetic analysis using the complete chloroplast genomes and matK gene reflecting the phylogenetic position of C. hirtinoda. The maximum likelihood (ML) analysis based on the complete chloroplast genomes indicated seven nodes with entirely branch support (100% bootstrap value). However, the three Chimonobambusa genera exhibited a moderate relationship due to fewer samples used, supporting that C. hirtinoda is closely related to C. tumidissinoda with a 62% bootstrap value more than C. hejiangensis. A phylogenetic tree based on the matK gene revealed that Chimonobambusa species clustered in one branch was consistent with the phylogenetic tree constructed by the complete cp genome tree (Fig. 9). The results show that the whole chloroplast genome identified related species better than the former, consistent with the previous study39.Figure 9Maximum likelihood phylogenetic tree based on the complete chloroplast genomes (A) and matK gene (B).Full size image More

Rasmussen, S. O. et al. A stratigraphic framework for abrupt climatic changes during the Last Glacial period based on three synchronized Greenland ice-core records: refining and extending the INTIMATE event stratigraphy. Quat. Sci. Rev. 106, 14–28 (2014).Article 

Google Scholar 
Heiri, O. et al. Validation of climate model-inferred regional temperature change for late-glacial Europe. Nat. Commun. 5, 1–7 (2014).Article 
CAS 

Google Scholar 
Muschitiello, F. et al. Fennoscandian freshwater control on Greenland hydroclimate shifts at the onset of the Younger Dryas. Nat. Commun. 6, 1–8 (2015).Article 
CAS 

Google Scholar 
Renssen, H. et al. Multiple causes of the Younger Dryas cold period. Nat. Geosci. 8, 946–949 (2015).CAS 
Article 

Google Scholar 
Mangerud, J. The discovery of the Younger Dryas, and comments on the current meaning and usage of the term. Boreas 50, 1–5 (2021).Article 

Google Scholar 
Cheng, H. et al. Timing and structure of the Younger Dryas event and its underlying climate dynamics. Proc. Natl. Acad. Sci. USA 117, 23408–23417 (2020).CAS 
Article 

Google Scholar 
van Hoesel, A., Hoek, W. Z., Pennock, G. M. & Drury, M. R. The Younger Dryas impact hypothesis: a critical review. Quat. Sci. Rev. 83, 95–114 (2014).Article 

Google Scholar 
Partin, J. W. et al. Gradual onset and recovery of the Younger Dryas abrupt climate event in the tropics. Nat. Commun. 6, 1–9 (2015).Article 

Google Scholar 
Reinig, F. et al. Precise date for the Laacher See eruption synchronizes the Younger Dryas. Nature 595, 66–69 (2021).CAS 
Article 

Google Scholar 
Ammann, B. et al. Quantification of biotic responses to rapid climatic changes around the Younger Dryas — a synthesis. Palaeogeogr. Palaeoclimatol. Palaeoecol. 159, 313–347 (2000).Article 

Google Scholar 
Hoek, W. Z. Vegetation response to the ∼ 14.7 and ∼ 11.5 ka cal. BP climate transitions: is vegetation lagging climate? Glob. Planet. Change 30, 103–115 (2001).Article 

Google Scholar 
Litt, T. et al. Correlation and synchronisation of Lateglacial continental sequences in northern central Europe based on annually laminated lacustrine sediments. Quat. Sci. Rev. 20, 1233–1249 (2001).Article 

Google Scholar 
Muschitiello, F. & Wohlfarth, B. Time-transgressive environmental shifts across Northern Europe at the onset of the Younger Dryas. Quat. Sci. Rev. 109, 49–56 (2015).Article 

Google Scholar 
Nakagawa, T. et al. The spatio-temporal structure of the Lateglacial to early Holocene transition reconstructed from the pollen record of Lake Suigetsu and its precise correlation with other key global archives: implications for palaeoclimatology and archaeology. Glob. Planet. Change 202, 103493 (2021).Article 

Google Scholar 
Ammann, B. et al. Vegetation responses to rapid warming and to minor climatic fluctuations during the Late-Glacial Interstadial (GI-1) at Gerzensee (Switzerland). Palaeogeogr. Palaeoclimatol. Palaeoecol. 391, 40–59 (2013).Article 

Google Scholar 
Engels, S. et al. Subdecadal‐scale vegetation responses to a previously unknown late‐Allerød climate fluctuation and Younger Dryas cooling at Lake Meerfelder Maar (Germany). J. Quat. Sci 31, 741–752 (2016).Article 

Google Scholar 
Van Raden, U. J. et al. High-resolution late-glacial chronology for the Gerzensee lake record (Switzerland): δ18O correlation between a Gerzensee-stack and NGRIP. Palaeogeogr. Palaeoclimatol. Palaeoecol. 391, 13–24 (2013).Article 

Google Scholar 
Blaga, C. I., Reichart, G.-J., Lotter, A. F., Anselmetti, F. S. & Sinninghe Damsté, J. S. A TEX86 lake record suggests simultaneous shifts in temperature in Central Europe and Greenland during the last deglaciation. Geophys. Res. Lett. 40, 948–953 (2013).Article 

Google Scholar 
Rach, O., Brauer, A., Wilkes, H. & Sachse, D. Delayed hydrological response to Greenland cooling at the onset of the Younger Dryas in western Europe. Nat. Geosci. 7, 109 (2014).CAS 
Article 

Google Scholar 
Strogatz, S. H. Exploring complex networks. Nature 410, 268–276 (2001).CAS 
Article 

Google Scholar 
Doncaster, C. P. et al. Early warning of critical transitions in biodiversity from compositional disorder. Ecology 97, 3079–3090 (2016).Article 

Google Scholar 
Jones, G. et al. The Lateglacial to early Holocene tephrochronological record from Lake Hämelsee, Germany: a key site within the European tephra framework. Boreas 47, 28–40 (2018).Article 

Google Scholar 
Blaga, C. I., Reichart, G.-J., Heiri, O. & Damsté, J. S. S. Tetraether membrane lipid distributions in water-column particulate matter and sediments: a study of 47 European lakes along a north–south transect. J. Paleolimnol. 41, 523–540 (2009).Article 

Google Scholar 
Bechtel, A., Smittenberg, R. H., Bernasconi, S. M. & Schubert, C. J. Distribution of branched and isoprenoid tetraether lipids in an oligotrophic and a eutrophic Swiss lake: insights into sources and GDGT-based proxies. Org. Geochem. 41, 822–832 (2010).CAS 
Article 

Google Scholar 
Lowe, J. et al. On the timing of retreat of the Loch Lomond (‘Younger Dryas’) Readvance icefield in the SW Scottish Highlands and its wider significance. Quat. Sci. Rev. 219, 171–186 (2019).Article 

Google Scholar 
Muggeo, V. M. R. Segmented: an R package to fit regression models with broken-line relationships. R news 8, 20–25 (2008).
Google Scholar 
Merkt, J. & Müller, H. Varve chronology and palynology of the Lateglacial in Northwest Germany from lacustrine sediments of Hämelsee in Lower Saxony. Quat. Int. 61, 41–59 (1999).Article 

Google Scholar 
Litt, T. & Stebich, M. Bio-and chronostratigraphy of the lateglacial in the Eifel region, Germany. Quat. Int. 61, 5–16 (1999).Article 

Google Scholar 
Reimer, P. J. et al. The IntCal20 Northern hemisphere radiocarbon age calibration curve (0–55 cal kBP). Radiocarbon 62, 725–757 (2020).CAS 
Article 

Google Scholar 
Giesecke, T. Holocene dynamics of the southern boreal forest in Sweden. The Holocene 15, 858–872 (2005).Article 

Google Scholar 
Müller, D. et al. New insights into lake responses to rapid climate change: the Younger Dryas in Lake Gościąż, central Poland. Boreas 50, 535–555 (2021).Article 

Google Scholar 
Davis, B. A. S. et al. The Eurasian Modern Pollen Database (EMPD), version 2. Earth Syst. Sci. data 12, 2423–2445 (2020).Article 

Google Scholar 
Neugebauer, I. et al. A Younger Dryas varve chronology from the Rehwiese palaeolake record in NE-Germany. Quat. Sci. Rev. 36, 91–102 (2012).Article 

Google Scholar 
Ralska-Jasiewiczowa, M. et al. Very fast environmental changes at the Pleistocene/Holocene boundary, recorded in laminated sediments of Lake Gościaż, Poland. Palaeogeogr. Palaeoclimatol. Palaeoecol. 193, 225–247 (2003).Article 

Google Scholar 
Bonk, A. et al. Varve microfacies and chronology from a new sediment record of Lake Gościąż (Poland). Quat. Sci. Rev. 251, 106715 (2021).Article 

Google Scholar 
Brauer, A., Haug, G. H., Dulski, P., Sigman, D. M. & Negendank, J. F. W. An abrupt wind shift in western Europe at the onset of the Younger Dryas cold period. Nat. Geosci. 1, 520–523 (2008).CAS 
Article 

Google Scholar 
Mekhaldi, F. et al. Radionuclide wiggle matching reveals a nonsynchronous early Holocene climate oscillation in Greenland and western Europe around a grand solar minimum. Clim. Past 16, 1145–1157 (2020).Article 

Google Scholar 
Mayfield, R. J. et al. Metrics of structural change as indicators of chironomid community stability in high latitude lakes. Quat. Sci. Rev. 249, 106594 (2020).Article 

Google Scholar 
van der Knaap, W. O. & Van Leeuwen, J. F. N. Climate-pollen relationships AD 1901–1996 in two small mires near the forest limit in the northern and central Swiss Alps. The Holocene 13, 809–828 (2003).Article 

Google Scholar 
Bazelmans, J. et al. Environmental changes in the late Allerød and early Younger Dryas in the Netherlands: a multiproxy high-resolution record from a site with two Pinus sylvestris populations. Quat. Sci. Rev. 272, 107199 (2021).Article 

Google Scholar 
Birks, H. H., Battarbee, R. W. & Birks, H. J. B. The development of the aquatic ecosystem at Kråkenes Lake, western Norway, during the late glacial and early Holocene-a synthesis. J. Paleolimnol 23, 91–114 (2000).Article 

Google Scholar 
Bronk Ramsey, C. Bayesian analysis of radiocarbon dates. Radiocarbon 51, 337–360 (2009).Article 

Google Scholar 
Lohne, Ø. S., Mangerud, J. A. N. & Birks, H. H. IntCal13 calibrated ages of the Vedde and Saksunarvatn ashes and the Younger Dryas boundaries from Kråkenes, western Norway. J. Quat. Sci 29, 506–507 (2014).Article 

Google Scholar 
Lohne, Ø. S., Mangerud, J. A. N. & Birks, H. H. Precise 14 C ages of the Vedde and Saksunarvatn ashes and the Younger Dryas boundaries from western Norway and their comparison with the Greenland Ice Core (GICC 05) chronology. J. Quat. Sci 28, 490–500 (2013).Article 

Google Scholar 
Wohlfarth, B. et al. Hässeldala–a key site for last termination climate events in northern Europe. Boreas 46, 143–161 (2017).Article 

Google Scholar 
Brauer, A. et al. High resolution sediment and vegetation responses to Younger Dryas climate change in varved lake sediments from Meerfelder Maar, Germany. Quat. Sci. Rev. 18, 321–329 (1999).Article 

Google Scholar 
Lane, C. S., Brauer, A., Blockley, S. P. E. & Dulski, P. Volcanic ash reveals time-transgressive abrupt climate change during the Younger Dryas. Geology 41, 1251–1254 (2013).Bronk Ramsey, C. et al. Improved age estimates for key Late Quaternary European tephra horizons in the RESET lattice. Quat. Sci. Rev. 118, 18–32 (2015).Rasmussen, S. O. et al. A new Greenland ice core chronology for the last glacial termination. J. Geophys. Res. Atmos. 111, https://doi.org/10.1029/2005JD006079 (2006).Brauer, A., Endres, C., Zolitschka, B. & Negendank, J. F. W. AMS radiocarbon and varve chronology from the annually laminated sediment record of Lake Meerfelder Maar, Germany. Radiocarbon 42, 355–368 (2000).CAS 
Article 

Google Scholar 
Wulf, S. et al. Tracing the Laacher See Tephra in the varved sediment record of the Trzechowskie palaeolake in central Northern Poland. Quat. Sci. Rev. 76, 129–139 (2013).Article 

Google Scholar 
Brauer, A. et al. The importance of independent chronology in integrating records of past climate change for the 60–8 ka INTIMATE time interval. Quat. Sci. Rev. 106, 47–66 (2014).Article 

Google Scholar 
Lane, C. S. et al. The Late Quaternary tephrostratigraphy of annually laminated sediments from Meerfelder Maar, Germany. Quat. Sci. Rev. 122 192–206 (2015).Article 

Google Scholar 
Adolphi, F. & Muscheler, R. Synchronizing the Greenland ice core and radiocarbon timescales over the Holocene–Bayesian wiggle-matching of cosmogenic radionuclide records. Clim. Past 12, 15–30 (2016).Article 

Google Scholar 
Muschitiello, F. et al. Deep-water circulation changes lead North Atlantic climate during deglaciation. Nat. Commun. 10, 1–10 (2019).CAS 
Article 

Google Scholar 
Adolphi, F. et al. Persistent link between solar activity and Greenland climate during the Last Glacial Maximum. Nat. Geosci. 7, 662–666 (2014).CAS 
Article 

Google Scholar 
Siegenthaler, U., Heimann, M. & Oeschger, H. 14C variations caused by changes in the global carbon cycle. Radiocarbon 22, 177–191 (1980).CAS 
Article 

Google Scholar 
Muscheler, R., Adolphi, F. & Svensson, A. Challenges in 14C dating towards the limit of the method inferred from anchoring a floating tree ring radiocarbon chronology to ice core records around the Laschamp geomagnetic field minimum. Earth Planet. Sci. Lett. 394, 209–215 (2014).CAS 
Article 

Google Scholar 
Muschitiello, F. An improved and continuous synchronization of the Greenland ice-core and Hulu Cave U-Th timescales using probabilistic inversion. Clim. Past Discuss. 1–39 https://doi.org/10.5194/cp-2021-116 (2021).Moore, P. D., Webb, J. A. & Collison, M. E. Pollen analysis. (Blackwell scientific publications, 1991).Engels, S. et al. Haemelsee: late-glacial pollen counts. PANGAEA, https://doi.org/10.1594/PANGAEA.939693 (2021).Weltje, G. J. & Tjallingii, R. Calibration of XRF core scanners for quantitative geochemical logging of sediment cores: Theory and application. Earth Planet. Sci. Lett. 274, 423–438 (2008).CAS 
Article 

Google Scholar 
Heiri, O., Lotter, A. F. & Lemcke, G. Loss on ignition as a method for estimating organic and carbonate content in sediments: reproducibility and comparability of results. J. Paleolimnol 25, 101–110 (2001).Article 

Google Scholar 
Brooks, S. J., Langdon, P. G. & Heiri, O. The identification and use of Palaearctic Chironomidae larvae in palaeoecology. Quat. Res. Assoc. Tech. Guid. i–vi. Vol. 10, 1–276 (2007).Heiri, O., Brooks, S. J., Birks, H. J. B. & Lotter, A. F. A 274-lake calibration data-set and inference model for chironomid-based summer air temperature reconstruction in Europe. Quat. Sci. Rev. 30, 3445–3456 (2011).Article 

Google Scholar 
Heiri, O. & Lotter, A. F. Effect of low count sums on quantitative environmental reconstructions: an example using subfossil chironomids. J. Paleolimnol 26, 343–350 (2001).Article 

Google Scholar 
Rach, O., Hadeen, X. & Sachse, D. An automated solid phase extraction procedure for lipid biomarker purification and stable isotope analysis. Org. Geochem. 142, 103995 (2020).CAS 
Article 

Google Scholar 
Huguet, C. et al. An improved method to determine the absolute abundance of glycerol dibiphytanyl glycerol tetraether lipids. Org. Geochem. 37, 1036–1041 (2006).CAS 
Article 

Google Scholar 
Hopmans, E. C., Schouten, S. & Damsté, J. S. S. The effect of improved chromatography on GDGT-based palaeoproxies. Org. Geochem. 93, 1–6 (2016).CAS 
Article 

Google Scholar 
Birks, H. J. B. & Birks, H. H. Biological responses to rapid climate change at the Younger Dryas—Holocene transition at Kråkenes, western Norway. The Holocene 18, 19–30 (2008).Article 

Google Scholar 
R CORE TEAM, A. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2012. URL http://www.R-project.org (2020).Engels, S., van Geel, B., Buddelmeijer, N. & Brauer, A. High-resolution palynological evidence for vegetation response to the Laacher See eruption from the varved record of Meerfelder Maar (Germany) and other central European records. Rev. Palaeobot. Palynol. 221, 160–170 (2015).Article 

Google Scholar 
Hughes, A. L. C., Gyllencreutz, R., Lohne, Ø. S., Mangerud, J. & Svendsen, J. I. The last Eurasian ice sheets–a chronological database and time‐slice reconstruction, DATED‐1. Boreas 45, 1–45 (2016).Article 

Google Scholar  More

Fisheries overexploitation is a problem in all oceans and seas globally. Authorities and administrations in charge of assigning quotas have very little fine-grained information on the fish captures, and instead use large-scale, coarse data to assess the health level of fisheries. Thus, being able to cross-match fish species and sizes, to the sea regions they were captured from, can be helpful in this regard, providing finer-grained information.Previous attempts at assembling datasets for fish detection and classification exist, ranging from fish detection or counting in underwater images and video streams1,2,3, to counting on belts on trawler ships4, to classification in laboratory conditions5,6, or in underwater preprocessed images of single fish7,8,9, or single fish in free-form pictures10, as well as simultaneous detection and classification of several fish11,12. However, none of the works found in the literature addresses the topic of simultaneous instance segmentation and species classification, along with fish size estimation, in a fish market environment, as is the aim of this paper. Instance segmentation refers to the extraction of pixel-level masks for each individual object (in this case fish specimens), rather than bounding boxes (object detection), or class label masks (e.g. a single mask for all fish specimens of the same species, also referred to as semantic segmentation). Moreover, works in the literature use pictures taken in laboratory conditions (with a single fish per image, shown from the side), or in underwater conditions. Only French et al.4 uses pictures of fish catches on a belt, for counting purposes. Table 1 shows a summary of the datasets identified in the literature, along with their characteristics, including how the proposed dataset compares.Table 1 Summary of previous datasets found in the literature, and comparison to proposed dataset.Full size tableThe DeepFish project (website: http://deepfish.dtic.ua.es/) is aimed at providing fish species classification and size estimation for fish specimens arriving at fish markets, both for the automation of fish sales, and the retrieval of fine-grained information about the health of fisheries. For a period of six months (April to September 2021), images have been captured at the fish market in El Campello (Alicante, Spain). Images of market trays show a variety of fish species, including targeted as well as accidental captures from the ‘Cabo de la Huerta’, an important site for protection and preservation of marine habitats and biodiversity as defined by the European Comission Habitats Directive (92/43/EEC). From the pictures, a total of 59 different species are identified with 12 species having more than 100 specimens and 25 with more than 10 specimens, as shown in Table 2. There is a high imbalance of species captured due to the natural variation in fish species populations according to seasonality and other ecological factors (rarity of the species, i.e. total population count, etc). Due to some species showing sexual dimorphism (i.e. Symphodus tinca), this species is split into two separate class labels, leading to a different number of species, and class labels (59 species, but 60 class labels). The dataset presents a high temporal imbalance too. As shown in Fig. 1, the capture of new fish tray images was not evenly distributed during the six month study period. Several factors contributed to this: wholesale fish market operating days (e.g. no weekend data, holidays and stop periods, etc.), fish species variability (one of the aims was to be able to capture at least 100 specimens from several species, and seasonality meant some could not be available for capture in later months), as well as the time availability of research group members to attend the fish arrival, tray preparation and auctioning in the evenings.Table 2 Distribution of fish species in the dataset.Full size tableFig. 1Temporal distribution of fish tray images captured. It can be observed that April (04) and May (05) were much more active than the rest of months. This is due to several contributing factors.Full size imageThe resulting DeepFish dataset introduced here contains annotated images from 1,291 fish market trays, with a total of 7,339 specimens (individual fish instances) which were labelled (species and mask) using a specially-adapted version of the Django labeller instance segmentation labelling tool13. Subsequently, another JSON file is generated, following the Microsoft Common Objects in Context (MS COCO) dataset format14, which can be directly fed to a neural network. This is done via a script that is also provided15. Figure 2 shows the distribution of individuals for the selected species within the dataset. Furthermore, Fig. 3 shows examples of the trays, with instance segmentation (ground truth silhouette, i.e. as an interpolation from human-provided points) along with species labelling (different colour shading).Fig. 2Graphical view of the distribution of fish species in the DeepFish dataset for species above 10 specimens. Note, Symphodus tinca is considered separately due to sexual dimorphism (211 male; 335 female samples).Full size imageFig. 3Examples of ground truth fish instance masks with class labelling, showing the 12 species (13 labels) with more than 100 specimens (in bold in Table 2).Full size imageFrom the point of view of research, this data is important for the classification of fish species, instance segmentation, as well as specimen size estimation (e.g. as a regression problem, or otherwise). From an end-results perspective, data automatically labelled with fish instance segmentation accompanied by species name and estimated size is useful to different stakeholders, namely: fishing authorities (to understand how much of each species is being caught per zone), maritime conservation (to calculate depletion of fisheries), but also managers of the markets themselves, as well as clients (digitized sales, e-commerce), etc.The usage of the provided data can be manifold, as it can be used for several problems, namely: object detection and classification, which involves finding objects (in this case fish specimens) providing a bounding box, and a class for each of these boxes; additionally, the data can also be used for semantic segmentation, which can provide a pixel-wise segmentation of the image providing labels (in this case species labels) to different pixel regions of the image; furthermore, also instance segmentation is possible, in which not just a single label for all instances of the same species is provided, but each specimen is provided with a mask (specimen segmentation), as well as a label (species). Furthermore, several measurements of each fish are provided, which can also be used to estimate their size, since they have been shown to be correlated with each other16. These are estimated from the calculated homography (given the tray size is known), given the burden of measuring each fish due to the large amount of specimens in the dataset. More

Koski, S. E. Broader horizons for animal personality research. Front. Ecol. Evol. 2, 70 (2014).Article 

Google Scholar 
Careere, C. & Eens, M. Unravelling animal personalities: How and why individuals consistently differ. Behaviour 142, 1149–1157 (2005).Article 

Google Scholar 
Dingemanse, N. J., Both, C., Drent, P. J. & Tinbergen, J. M. Fitness consequences of avian personalities in a fluctuating environment. Proc. R. Soc. Lond. B 271, 847–852 (2004).Article 

Google Scholar 
Bell, A. M. & Sih, A. Exposure to predation generates personality in three-spined sticklebacks (Gasterosteus aculeatus). Ecol. Lett. 10, 828–834 (2007).PubMed 
Article 

Google Scholar 
Cavigelli, S. A. Animal personality and health. Behaviour 142, 1223–1244 (2005).Article 

Google Scholar 
Barber, I. & Dingemanse, N. J. Parasitism and the evolutionary ecology of animal personality. Proc. R. Soc. Lond. B 365, 4077–4088 (2010).
Google Scholar 
Koprivnikar, J., Gibson, C. H. & Redfern, J. C. Infectious personalities: Behavioural syndromes and disease risk in larval amphibians. Proc. R. Soc. Lond. B 279, 1544–1550 (2012).
Google Scholar 
Turner, J. & Hughes, W. O. H. The effect of parasitism on personality in a social insect. Behav. Proc. 157, 532–539 (2018).Article 

Google Scholar 
Frost, A. J., Winrow-Giffen, A., Ashley, P. J. & Sneddon, L. U. Plasticity in animal personality traits: Does prior experience alter the degree of boldness?. Proc. Biol. Sci. 274, 333–339 (2007).PubMed 

Google Scholar 
Dingemanse, N. J. & Wolf, M. Recent models for adaptive personality differences: A review. Philos. Trans. R. Soc. B 365, 3947–3958 (2010).Article 

Google Scholar 
Müller, T. & Müller, C. Phenotype of a leaf beetle larva depends on host plant quality and previous test experience. Behav. Proc. 142, 40–45 (2017).Article 

Google Scholar 
Hart, B. L. Biological basis of behaviour in sick animals. Neurosci. Biobehav. Rev. 12, 123–137 (1988).CAS 
PubMed 
Article 

Google Scholar 
Hart, B. L. Behavioral adaptations to pathogens and parasites: Five strategies. Neurosci. Biobehav. Rev. 14, 273–294 (1990).CAS 
PubMed 
Article 

Google Scholar 
Johnson, R. W. The concept of sickness behavior: A brief chronological account of four key discoveries. Vet. Immunol. Immunopathol. 87, 443–450 (2002).CAS 
PubMed 
Article 

Google Scholar 
Klein, S. L. Parasite manipulation of the proximate mechanisms that mediate social behavior in vertebrates. Physiol. Behav. 79, 441–449 (2003).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Boyer, N., Reale, D., Marmet, J., Pisanu, B. & Chapuis, L. Personality, space use and tick load in an introduced population of Siberian chipmunks Tanias sibiricus. J. Anim. Ecol. 79, 538–547 (2010).PubMed 
Article 

Google Scholar 
Ezenwa, V. O. Host social behavior and parasitic infection: A multifactorial approach. Behav. Ecol. 15, 446–454 (2004).Article 

Google Scholar 
Finkemeier, M. A., Langbein, J. & Puppe, B. Personality research in mammalian farm animals: Concepts, measures and relationship to welfare. Front. Vet. Sci. 5, 131 (2018).PubMed 
PubMed Central 
Article 

Google Scholar 
Huntingford, F. & Adams, C. Behavioural syndromes in farmed fish: Implications for production and welfare. Behaviour 142, 1207–1221 (2005).Article 

Google Scholar 
Berggren, Å., Jansson, A. & Low, M. Approaching ecological sustainability in the emerging insects-as-food industry. Trends Ecol. Evol. 34, 132–138 (2019).PubMed 
Article 

Google Scholar 
Dochtermann, N. A. & Nelson, A. B. Multiple facets of exploratory behavior in house crickets (Acheta domesticus): Split personalities or simply different behaviors?. Ethology 120, 1110–1117 (2014).Article 

Google Scholar 
van Huis, A. & Tomberlin, J. K. Future prospects. In Insects as Food Feed: From Production to Consumption (eds van Huis, A. & Tomberlin, J. K.) 430–445 (Wageningen Academic Publishers, 2017).Szelei, J. et al. Susceptibility of North-American and European crickets to Acheta domesticus densovirus (AdDNV) and associated epizootics. J. Invert. Pathol 106, 394–399 (2011).CAS 
Article 

Google Scholar 
Eilenberg, J., Vlak, J. M., Nielsen-LeRoux, C., Cappellozza, S. & Jensen, A. B. Diseases in insects produced for food and feed. J. Insects Food Feed 1, 87–102 (2015).Article 

Google Scholar 
Raubenheimer, D. & Tucker, D. Associative learning by locusts: Pairing of visual cues with consumption of protein and carbohydrate. Anim. Behav. 54, 1449–1459 (1997).CAS 
PubMed 
Article 

Google Scholar 
Mallory, H. S., Howard, A. F. & Weiss, M. R. Timing of environmental enrichment affects memory in the house cricket, Acheta domesticus. PLoS One 11, e0152245 (2016).PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
Sih, A., Bell, A. M., Johnson, J. C. & Ziemba, R. E. Behavioral syndromes: An integrative overview. Q. Rev. Biol. 79, 241–277 (2004).PubMed 
Article 

Google Scholar 
Siva-Jothy, J. A. & Vale, P. F. Viral infection causes sex-specific changes in fruit fly social aggregation behaviour. Biol. Lett. 15, 20190344 (2019).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Van Houte, S., Ros, V. I. D. & Van Oers, M. M. Walking with insects: Molecular mechanisms behind parasitic manipulation of host behaviour. Mol. Ecol. 22, 3458–3475 (2013).PubMed 
Article 

Google Scholar 
Vale, P. F., Siva-Jothy, J. A., Morrill, A. & Forbes, M. R. The influence of parasites. In Insect Behavior: From Mechanisms to Ecological and Evolutionary Consequences (eds Córdoba-Aguilar, A., González-Tokman, D. & González-Santoyo, I) (Oxford University Press, 2018).de Roode, J. C. & Lefèvre, T. Behavioral Immunity in Insects. Insects 3, 789–820 (2012).PubMed 
PubMed Central 
Article 

Google Scholar 
Kutzer, M. A. M. & Armitage, S. A. O. Maximising fitness in the face of parasites: A review of host tolerance. Zoology 119, 281–289 (2016).PubMed 
Article 

Google Scholar 
Vossen, L. E., Roman, E. & Jansson, A. Fasting increases shelter use in house crickets (Acheta domestica). J. Insects Food Feed 8, 5–8 (2021).Article 

Google Scholar 
Schutgens, M., Cook, B., Gilbert, F. & Behnke, J. M. Behavioural changes in the flour beetle Tribolium confusum infected with the spirurid nematode Protospirura muricola. J. Helminthol. 89, 68–79 (2015).CAS 
PubMed 
Article 

Google Scholar 
Kazlauskas, N., Klappenbach, M., Depino, A. M. & Locatelli, F. F. Sickness behavior in honey bees. Front. Physiol. 7, 261 (2016).PubMed 
PubMed Central 
Article 

Google Scholar 
Stahlschmidt, Z. R. & Adamo, S. A. Context dependency and generality of fever in insects. Naturwissenschaften 100, 691–696 (2013).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Wang, S. Y. S., Tattersall, G. J. & Koprivnikar, J. Trematode parasite infection affects temperature selection in aquatic host snails. Physiol. Biochem. Zool. 92, 71–79 (2019).PubMed 
Article 

Google Scholar 
Berggren, Å., Jansson, A. & Low, M. Using current systems to inform rearing facility design in the insects-as-food industry. J. Insects Food Feed. 4, 167–170 (2018).Article 

Google Scholar 
Marshall, J. A. & Haes, E. C. M. Grasshoppers and Allied Insects of Great Britain and Ireland (Harley Books, Essex) (1988).GBIF Secretariat. Acheta domesticus (Linnaeus, 1758). GBIF Backbone Taxonomy. Checklist dataset. https://doi.org/10.15468/39omei accessed via GBIF.org on 12 Jan 2022 (2021).Holst, K. T. The Saltatoria (Bush-crickets, Crickets and Grasshoppers) of Northern Europe (E J Brill, 1986).Ingrisch, S. & Köhler, G. Die heuschrecken mitteleuropas. (Westarp Wissenschaften, 1998).Booth, D. T. & Kiddell, K. Temperature and the energetics of development in the house cricket (Acheta domesticus). J. Insect Physiol. 53, 950–953 (2007).CAS 
PubMed 
Article 

Google Scholar 
Ghouri, A. S. K. & McFarlane, J. E. Observations on the development of crickets. Can. Entomol. 90, 158–165 (1958).Article 

Google Scholar 
Semberg, E. et al. Diagnostic protocols for the detection of Acheta domesticus densovirus (AdDV) in cricket frass. J. Virol. Methods 264, 61–64 (2019).CAS 
PubMed 
Article 

Google Scholar 
Bergoin, M. & Tijssen, P. Parvoviruses of arthropods. In Encyclopedia of Virology. 76–85 (2008).Cotmore, S. F. et al. The family Parvoviridae. Arch. Virol. 159, 1239–1247 (2014).CAS 
PubMed 
Article 

Google Scholar 
Styer, E. L. & Hamm, J. J. Report of a densovirus in a commercial cricket operation in the southeastern United States. J. Invert. Pathol. 58, 283–285 (1991).Article 

Google Scholar 
Weissman, D. B., Gray, D. A., Pham, H. T. & Tijssen, P. Billions and billions sold: Pet-feeder crickets (Orthoptera: Gryllidae), commercial crickets farms, an epizootic densovirus, and government regulations make for a potential disaster. Zootaxa 3504, 67–88 (2012).Article 

Google Scholar 
Maciel-Vergara, G. & Ros, V. I. D. Viruses of insects reared for food and feed. J. Invert. Pathol. 147, 60–75 (2017).Article 

Google Scholar 
Liu, K. et al. The Acheta domesticus densovirus, isolated from the European house cricket, has evolved an expression strategy unique among parvoviruses. J. Virol. 85, 10069–10078 (2011).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Wang, Y. et al. Densovirus crosses the insect midgut by transcytosis and disturbs the epithelial barrier function. J. Virol. 87, 12380–12391 (2013).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
de Miranda, J. R. et al. Virus diversity and loads in crickets reared for feed: Implications for husbandry. Front. Vet. Sci. 8, 642085 (2021).PubMed 
PubMed Central 
Article 

Google Scholar 
de Miranda, J. R., Granberg, F., Onorati, P., Jansson, A. & Berggren, Å. Virus prospecting in crickets: discovery and strain divergence of a novel Iflavirus in wild and cultivated Acheta domesticus. Viruses 13, 364 (2021).PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
Niemelä, P., Vainikka, A., Hedrick, A. & Kortet, R. Integrating behaviour with life history: Boldness of the field cricket, Gryllus integer, during ontogeny. Funct. Ecol. 26, 450–456 (2012).Article 

Google Scholar 
Hedrick, A. V. Crickets with extravagant mating songs compensate for predation risk with extra caution. Proc. R. Soc. Lond. B 267, 671–675 (2000).CAS 
Article 

Google Scholar 
Hedrick, A. V. & Kortet, R. Hiding behaviour in two cricket populations that differ in predation pressure. Anim. Behav. 72, 1111–1118 (2006).Article 

Google Scholar 
Kortet, R. & Hedrick, A. V. A behavioural syndrome in the field cricket Gryllus integer: Intrasexual aggression is correlated with activity in a novel environment. Biol. J. Linnean Soc. 91, 475–482 (2007).Article 

Google Scholar 
Fisher, D. N., David, M., Rodríguez-Muñoz, R. & Tregenza, T. Lifespan and age, but not residual reproductive value or condition, are related to behaviour in wild field crickets. Ethology 124, 338–346 (2018).Article 

Google Scholar 
R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/ (2020).Plummer, M. JAGS: A program for analysis of Bayesian graphical models using Gibbs sampling. In: Proceedings of the 3rd International Workshop on Distributed Statistical Computing. Vienna, Austria (2003).Le Galliard, J. F., Paquet, M., Cisel, M. & Montes-Poloni, L. Personality and the pace-of-life syndrome: Variation and selection on exploration, metabolism and locomotor performances. Funct. Ecol. 27, 136–144 (2013).Article 

Google Scholar 
Roche, D. G., Careau, V. & Binning, S. A. Demystifying animal ‘personality’ (or not): Why individual variation matters to experimental biologists. J. Exp. Biol. 219, 3832–3843 (2016).PubMed 

Google Scholar 
Low, M. et al. The importance of accounting for larval detectability in mosquito habitat-association studies. Malar. J. 15, 1–9 (2016).Article 

Google Scholar  More

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.This is a summary of: Qiao, L. et al. Soil quality both increases crop production and improves resilience to climate change. Nat. Clim. Change https://doi.org/10.1038/s41558-022-01376-8 (2022). More

Pan, Y. et al. A large and persistent carbon sink in the world’s forests. Science 333, 988–993, https://doi.org/10.1126/science.1201609 (2011).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Pugh, T. A. M. et al. Role of forest regrowth in global carbon sink dynamics. Proceedings of the National Academy of Sciences 116, 4382–4387, https://doi.org/10.1073/pnas.1810512116 (2019).ADS 
CAS 
Article 

Google Scholar 
Chazdon, R. L. et al. Carbon sequestration potential of second-growth forest regeneration in the Latin American tropics. Science Advances 2, e1501639, https://doi.org/10.1126/sciadv.1501639 (2016).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Poorter, L. et al. Biomass resilience of Neotropical secondary forests. Nature 530, 211–214, https://doi.org/10.1038/nature16512 (2016).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Cook-Patton, S. C. et al. Mapping carbon accumulation potential from global natural forest regrowth. Nature 585, 545–550, https://doi.org/10.1038/s41586-020-2686-x (2020).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Lewis, S. L. et al. Increasing carbon storage in intact African tropical forests. Nature 457, 1003–1006, https://doi.org/10.1038/nature07771 (2009).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Hubau, W. et al. Asynchronous carbon sink saturation in African and Amazonian tropical forests. Nature 579, 80–87, https://doi.org/10.1038/s41586-020-2035-0 (2020).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Nabuurs, G.-J. et al. First signs of carbon sink saturation in European forest biomass. Nature Climate Change 3, 792–796, https://doi.org/10.1038/nclimate1853 (2013).ADS 
CAS 
Article 

Google Scholar 
Köhl, M. et al. Changes in forest production, biomass and carbon: results from the 2015 UN FAO Global Forest Resource Assessment. Forest Ecology and Management 352, 21–34, https://doi.org/10.1016/j.foreco.2015.05.036 (2015).Article 

Google Scholar 
Asner, G. P. et al. High-resolution forest carbon stocks and emissions in the Amazon. Proceedings of the National Academy of Sciences 107, 16738–16742, https://doi.org/10.1073/pnas.1004875107 (2010).ADS 
Article 

Google Scholar 
Asner, G. P. Tropical forest carbon assessment: integrating satellite and airborne mapping approaches. Environmental Research Letters 4, https://doi.org/10.1088/1748-9326/4/3/034009 (2009).Xu, L. et al. Changes in global terrestrial live biomass over the 21st century. Science Advances 7, eabe9829, https://doi.org/10.1126/sciadv.abe9829 (2021).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Aalde, U. et al. in IPCC Guidelines for National Greenhouse Gas Inventories Vol. 4 (eds Eggleston, S., Buendia, L., Miwa, K., Ngara, T. & Tanabe, K.) Ch. 4 (IPPC, 2006).Brown, S. Measuring carbon in forests: current status and future challenges. Environmental Pollution 116, 363–372, https://doi.org/10.1016/s0269-7491(01)00212-3 (2002).CAS 
Article 
PubMed 

Google Scholar 
Woodall, C. W., Heath, L. S., Domke, G. M. & Nichols, M. C. Methods and equations for estimating aboveground volume, biomass, and carbon for trees in the U.S. forest inventory, 2010. (U.S. Department of Agriculture, Forest Service, Northern Research Station, 2011).Saatchi, S. S. et al. Benchmark map of forest carbon stocks in tropical regions across three continents. Proceedings of the National Academy of Sciences 108, 9899–9904, https://doi.org/10.1073/pnas.1019576108 (2011).ADS 
Article 

Google Scholar 
Lamlom, S. H. & Savidge, R. A. A reassessment of carbon content in wood: variation within and between 41 North American species. Biomass Bioenergy 25, 381–388 (2003).CAS 
Article 

Google Scholar 
Van Der Werf, G. R. et al. CO2 emissions from forest loss. Nature Geoscience 2, 737–738, https://doi.org/10.1038/ngeo671 (2009).ADS 
CAS 
Article 

Google Scholar 
Martin, A. R., Doraisami, M. & Thomas, S. C. Global patterns in wood carbon concentration across the world’s trees and forests. Nature Geoscience 11, 915–920, https://doi.org/10.1038/s41561-018-0246-x (2018).ADS 
CAS 
Article 

Google Scholar 
Tavşanoğlu, Ç. & Pausas, J. G. A functional trait database for Mediterranean Basin plants. Scientific Data 5, 180135, https://doi.org/10.1038/sdata.2018.135 (2018).Article 
PubMed 
PubMed Central 

Google Scholar 
Chave, J. et al. Towards a worldwide wood economics spectrum. Ecology Letters 12, 351–366, https://doi.org/10.1111/j.1461-0248.2009.01285.x (2009).Article 
PubMed 

Google Scholar 
Martin, A. R., Domke, G. M., Doraisami, M. & Thomas, S. C. Carbon fractions in the world’s dead wood. Nature Communications 12, https://doi.org/10.1038/s41467-021-21149-9 (2021).Martin, A. R. & Thomas, S. C. A Reassessment of carbon content in tropical trees. PLoS ONE 6, e23533, https://doi.org/10.1371/journal.pone.0023533 (2011).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Martin, A. R., Gezahegn, S. & Thomas, S. C. Variation in carbon and nitrogen concentration among major woody tissue types in temperate trees. Canadian Journal of Forest Research 45, 744–757, https://doi.org/10.1139/cjfr-2015-0024 (2015).CAS 
Article 

Google Scholar 
Thomas, S. C. & Martin, A. R. Carbon content of tree tissues: a synthesis. Forests 3, 332–352, https://doi.org/10.3390/f3020332 (2012).Article 

Google Scholar 
Doraisami, M. et al. GLOWCAD: A global database of woody tissue carbon concentrations fractions. Dryad https://doi.org/10.5061/dryad.18931zcxk (2022)Fick, S. E. & Hijmans, R. J. WorldClim 2: new 1‐km spatial resolution climate surfaces for global land areas. International Journal of Climatology 37, 4302–4315, https://doi.org/10.1002/joc.5086 (2017).ADS 
Article 

Google Scholar 
Guerrero‐Ramírez, N. R. et al. Global root traits (GRooT) database. Global Ecology and Biogeography 30, 25–37, https://doi.org/10.1111/geb.13179 (2021).Article 

Google Scholar 
Kattge, J. et al. TRY plant trait database – enhanced coverage and open access. Global Change Biology 26, 119–188, https://doi.org/10.1111/gcb.14904 (2020).ADS 
Article 
PubMed 

Google Scholar 
Iversen, C. M. et al. A global Fine-Root Ecology Database to address below-ground challenges in plant ecology. New Phytologist 215, 15–26, https://doi.org/10.1111/nph.14486 (2017).Article 
PubMed 

Google Scholar 
Isaac, M. E. et al. Intraspecific trait variation and coordination: Root and Leaf Economics Spectra in coffee across environmental gradients. Frontiers in Plant Science 8, https://doi.org/10.3389/fpls.2017.01196 (2017).Liu, C. et al. Variation in the functional traits of fine roots is linked to phylogenetics in the common tree species of Chinese subtropical forests. Plant and Soil 436, 347–364, https://doi.org/10.1007/s11104-019-03934-0 (2019).CAS 
Article 

Google Scholar 
Wang, R. et al. Different phylogenetic and environmental controls of first‐order root morphological and nutrient traits: Evidence of multidimensional root traits. Functional Ecology 32, 29–39, https://doi.org/10.1111/1365-2435.12983 (2018).Article 

Google Scholar 
Minden, V. & Kleyer, M. Internal and external regulation of plant organ stoichiometry. Plant Biology 16, 897–907, https://doi.org/10.1111/plb.12155 (2014).CAS 
Article 
PubMed 

Google Scholar 
Alameda, D. & Villar, R. Linking root traits to plant physiology and growth in Fraxinus angustifolia Vahl. seedlings under soil compaction conditions. Environmental and Experimental Botany 79, 49–57, https://doi.org/10.1016/j.envexpbot.2012.01.004 (2012).Article 

Google Scholar 
Aubin, I. et al. Traits to stay, traits to move: a review of functional traits to assess sensitivity and adaptive capacity of temperate and boreal trees to climate change. Environmental Reviews 24, 164–186, https://doi.org/10.1139/er-2015-0072 (2016).Article 

Google Scholar 
Fernández-García, N. et al. Intrinsic water use efficiency controls the adaptation to high salinity in a semi-arid adapted plant, henna (Lawsonia inermis L.). Journal of Plant Physiology 171, 64–75, https://doi.org/10.1016/j.jplph.2013.11.004 (2014).CAS 
Article 
PubMed 

Google Scholar 
Grechi, I. et al. Effect of light and nitrogen supply on internal C:N balance and control of root-to-shoot biomass allocation in grapevine. Environmental and Experimental Botany 59, 139–149, https://doi.org/10.1016/j.envexpbot.2005.11.002 (2007).CAS 
Article 

Google Scholar 
Ineson, P., Cotrufo, M. F., Bol, R., Harkness, D. D. & Blum, H. Quantification of soil carbon inputs under elevated CO2: C3 plants in a C4 soil. Plant and Soil 187, 345–350, https://doi.org/10.1007/bf00017099 (1995).Article 

Google Scholar 
Pregitzer, K. S. et al. Atmospheric CO2, soil nitrogen and turnover of fine roots. New Phytologist 129, 579–585, https://doi.org/10.1111/j.1469-8137.1995.tb03025.x (1995).Article 

Google Scholar 
Rohatgi, A. WebPlotDigitizer: Version 4.5. https://automeris.io/WebPlotDigitizer (2021).Harmon, M. E., Fasth, B., Woodall, C. W. & Sexton, J. Carbon concentration of standing and downed woody detritus: effects of tree taxa, decay class, position, and tissue type. Forest Ecology and Management 291, 259–267, https://doi.org/10.1016/j.foreco.2012.11.046 (2013).Article 

Google Scholar 
Boyle, B. H. et al. The taxonomic name resolution service: an online tool for automated standardization of plant names. BMC Bioinformatics 14, 1-15, https://tnrs.biendata.org/ (2021).Krankina, O. N., Harmon, M. E. & Griazkin, A. V. Nutrient stores and dynamics of woody detritus in a boreal forest: modeling potential implications at the stand level. Canadian Journal of Forest Research 29, 20–32, https://doi.org/10.1139/x98-162 (1999).Article 

Google Scholar 
Whittaker, R. H. Classification of natural communities. Botanical Review 28, 1–239, https://doi.org/10.1007/BF02860872 (1962).Article 

Google Scholar 
Maiti, R. & Rodriguez, H. G. Wood carbon and nitrogen of 37 woody shrubs and trees in Tamaulipan thorn scrub, northeastern Mexico. Pakistan Journal of Botany 51, 979–984 (2019).CAS 
Article 

Google Scholar 
Durkaya, A. B. D., E. Makineci, I. Orhan Aboveground biomass and carbon storage relationship of Turkish Pines. Fresenius Environmental Bulletin 24, 3573–3583 (2015).CAS 

Google Scholar 
Tesfaye, M. A., Bravo-Oviedo, A., Bravo, F., Pando, V. & De Aza, C. H. Variation in carbon concentration and wood density for five most commonly grown native tree species in central highlands of Ethiopia: The case of Chilimo dry Afromontane forest. Journal of Sustainable Forestry 38, 769–790, https://doi.org/10.1080/10549811.2019.1607754 (2019).Article 

Google Scholar 
Abdallah, M. A. B., Mata-González, R., Noller, J. S. & Ochoa, C. G. Ecosystem carbon in relation to woody plant encroachment and control: Juniper systems in Oregon, USA. Agriculture, Ecosystems & Environment 290, 106762, https://doi.org/10.1016/j.agee.2019.106762 (2020).CAS 
Article 

Google Scholar 
Arias, D., Calvo-Alvarado, J., Richter, D. D. B. & Dohrenbusch, A. Productivity, aboveground biomass, nutrient uptake and carbon content in fast-growing tree plantations of native and introduced species in the Southern Region of Costa Rica. Biomass and Bioenergy 35, 1779–1788, https://doi.org/10.1016/j.biombioe.2011.01.009 (2011).CAS 
Article 

Google Scholar 
Assefa, D., Godbold, D. L., Belay, B., Abiyu, A. & Rewald, B. Fine Root Morphology, Biochemistry and Litter Quality Indices of Fast- and Slow-growing Woody Species in Ethiopian Highland Forest. Ecosystems 21, 482–494, https://doi.org/10.1007/s10021-017-0163-7 (2018).CAS 
Article 

Google Scholar 
Atkin, O. K., Schortemeyer, M., Mcfarlane, N. & Evans, J. R. The response of fast- and slow-growing Acacia species to elevated atmospheric CO2: an analysis of the underlying components of relative growth rate. Oecologia 120, 544–554, https://doi.org/10.1007/s004420050889 (1999).ADS 
Article 
PubMed 

Google Scholar 
Bardulis, A., Jansons, A., Bardule, A., Zeps, M. & LAzdins, A. Assessment of carbon content in root biomass in Scots Pine and silver birch young stands of Latvia. Baltic Forestry 23, 482–489 (2017).
Google Scholar 
Becker, G. S., Braun, D., Gliniars, R. & Dalitz, H. Relations between wood variables and how they relate to tree size variables of tropical African tree species. Trees 26, 1101–1112, https://doi.org/10.1007/s00468-012-0687-6 (2012).Article 

Google Scholar 
Bembenek, M. et al. Carbon content in Juvenile and mature wood of Scots Pine (Pinus sylyestris L.). Baltic Forestry 21, 279–284 (2015).
Google Scholar 
Bert, D. & Danjon, F. Carbon concentration variations in the roots, stem and crown of mature Pinus pinaster (Ait.). Forest Ecology and Management 222, 279–295, https://doi.org/10.1016/j.foreco.2005.10.030 (2006).Article 

Google Scholar 
Borden, K. A., Anglaaere, L. C. N., Adu-Bredu, S. & Isaac, M. E. Root biomass variation of cocoa and implications for carbon stocks in agroforestry systems. Agroforestry Systems 93, 369–381, https://doi.org/10.1007/s10457-017-0122-5 (2019).Article 

Google Scholar 
Borden, K. A., Isaac, M. E., Thevathasan, N. V., Gordon, A. M. & Thomas, S. C. Estimating coarse root biomass with ground penetrating radar in a tree-based intercropping system. Agroforestry Systems 88, 657–669, https://doi.org/10.1007/s10457-014-9722-5 (2014).Article 

Google Scholar 
Bueno-López, S. W., García-Lucas, E. & Caraballo-Rojas, L. R. Allometric equations for total aboveground dry biomass and carbon content of Pinus occidentalis trees. Madera y Bosques 25, https://doi.org/10.21829/myb.2019.2531868 (2019).Bulmer, R. H., Schwendenmann, L. & Lundquist, C. J. Allometric models for estimating aboveground biomass, carbon and nitrogen stocks in temperate Avicennia marina forests. Wetlands 36, 841–848, https://doi.org/10.1007/s13157-016-0793-0 (2016).Article 

Google Scholar 
Bütler, R., Patty, L., Le Bayon, R.-C., Guenat, C. & Schlaepfer, R. Log decay of Picea abies in the Swiss Jura Mountains of central Europe. Forest Ecology and Management 242, 791–799, https://doi.org/10.1016/j.foreco.2007.02.017 (2007).Article 

Google Scholar 
Cao, Y. & Chen, Y. Ecosystem C:N:P stoichiometry and carbon storage in plantations and a secondary forest on the Loess Plateau, China. Ecological Engineering 105, 125–132, https://doi.org/10.1016/j.ecoleng.2017.04.024 (2017).Article 

Google Scholar 
Castaño-Santamaría, J. & Bravo, F. Variation in carbon concentration and basic density along stems of sessile oak (Quercus petraea (Matt.) Liebl.) and Pyrenean oak (Quercus pyrenaica Willd.) in the Cantabrian Range (NW Spain). Annals of Forest Science 69, 663–672, https://doi.org/10.1007/s13595-012-0183-6 (2012).Article 

Google Scholar 
Chao, K.-J. et al. Carbon concentration declines with decay class in tropical forest woody debris. Forest Ecology and Management 391, 75–85, https://doi.org/10.1016/j.foreco.2017.01.020 (2017).Article 

Google Scholar 
Chen, Y. et al. Nutrient limitation of woody debris decomposition in a tropical forest: contrasting effects of N and P addition. Functional Ecology 30, 295–304, https://doi.org/10.1111/1365-2435.12471 (2016).Article 

Google Scholar 
Correia, A. C. et al. Biomass allometry and carbon factors for a Mediterranean pine (Pinus pinea L.) in Portugal. Forest Systems 19, 418, https://doi.org/10.5424/fs/2010193-9082 (2010).Article 

Google Scholar 
Cousins, S. J. M., Battles, J. J., Sanders, J. E. & York, R. A. Decay patterns and carbon density of standing dead trees in California mixed conifer forests. Forest Ecology and Management 353, 136–147, https://doi.org/10.1016/j.foreco.2015.05.030 (2015).Article 

Google Scholar 
Craven, D. et al. Seasonal variability of photosynthetic characteristics influences growth of eight tropical tree species at two sites with contrasting precipitation in Panama. Forest Ecology and Management 261, 1643–1653, https://doi.org/10.1016/j.foreco.2010.09.017 (2011).Article 

Google Scholar 
Cruz, P., Bascuñan, A., Velozo, J. & Rodriguez, M. Funciones alométricas de contenido de carbono para quillay, peumo, espino y litre. Bosque (Valdivia) 36, 375–381, https://doi.org/10.4067/s0717-92002015000300005 (2015).Article 

Google Scholar 
Currie, W. S. & Nadelhoffer, K. J. The imprint of land-use history: patterns of carbon and nitrogen in downed woody debris at the Harvard Forest. Ecosystems 5, 446–460, https://doi.org/10.1007/s10021-002-1153-x (2002).CAS 
Article 

Google Scholar 
Dong, L., Zhang, X. & Li Variation in carbon concentration and allometric equations for estimating tree carbon contents of 10 broadleaf species in natural forests in northeast China. Forests 10, 928, https://doi.org/10.3390/f10100928 (2019).Article 

Google Scholar 
Dossa, G. G. O., Paudel, E., Cao, K., Schaefer, D. & Harrison, R. D. Factors controlling bark decomposition and its role in wood decomposition in five tropical tree species. Scientific Reports 6, 34153, https://doi.org/10.1038/srep34153 (2016).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Durkaya, B., Durkaya, A., Makineci, E. & Ülküdür, M. Estimation of above-ground biomass and sequestered carbon of Taurus Cedar (Cedrus libani L.) in Antalya, Turkey. iForest – Biogeosciences and Forestry 6, 278–284, https://doi.org/10.3832/ifor0899-006 (2013).Article 

Google Scholar 
Elias, M. & Potvin, C. Assessing inter- and intra-specific variation in trunk carbon concentration for 32 neotropical tree species. Canadian Journal of Forest Research 33, 1039–1045, https://doi.org/10.1139/x03-018 (2003).Article 

Google Scholar 
Fang, S., Xue, J. & Tang, L. Biomass production and carbon sequestration potential in poplar plantations with different management patterns. Journal of Environmental Management 85, 672–679, https://doi.org/10.1016/j.jenvman.2006.09.014 (2007).CAS 
Article 
PubMed 

Google Scholar 
Fonseca, W., Alice, F. E. & Rey-Benayas, J. M. Carbon accumulation in aboveground and belowground biomass and soil of different age native forest plantations in the humid tropical lowlands of Costa Rica. New Forests 43, 197–211, https://doi.org/10.1007/s11056-011-9273-9 (2012).Article 

Google Scholar 
Frangi, J. L., Richter, L. L., Barrera, M. D. & Aloggia, M. Decomposition of Nothofagus fallen woody debris in forests of Tierra del Fuego, Argentina. Canadian Journal of Forest Research 27, 1095–1102, https://doi.org/10.1139/x97-060 (1997).Article 

Google Scholar 
Freschet, G. T., Cornelissen, J. H. C., Van Logtestijn, R. S. P. & Aerts, R. Evidence of the ‘plant economics spectrum’ in a subarctic flora. Journal of Ecology 98, 362–373, https://doi.org/10.1111/j.1365-2745.2009.01615.x (2010).Article 

Google Scholar 
Fukatsu, E., Fukuda, Y., Takahashi, M. & Nakada, R. Clonal variation of carbon content in wood of Larix kaempferi (Japanese larch). Journal of Wood Science 54, 247–251, https://doi.org/10.1007/s10086-007-0939-z (2008).CAS 
Article 

Google Scholar 
Ganamé, M., Bayen, P., Dimobe, K., Ouédraogo, I. & Thiombiano, A. Aboveground biomass allocation, additive biomass and carbon sequestration models for Pterocarpus erinaceus Poir. in Burkina Faso. Heliyon 6, e03805, https://doi.org/10.1016/j.heliyon.2020.e03805 (2020).Article 
PubMed 
PubMed Central 

Google Scholar 
Ganjegunte, G. K., Condron, L. M., Clinton, P. W., Davis, M. R. & Mahieu, N. Decomposition and nutrient release from radiata pine (Pinus radiata) coarse woody debris. Forest Ecology and Management 187, 197–211, https://doi.org/10.1016/s0378-1127(03)00332-3 (2004).Article 

Google Scholar 
Gao, B., Taylor, A. R., Chen, H. Y. H. & Wang, J. Variation in total and volatile carbon concentration among the major tree species of the boreal forest. Forest Ecology and Management 375, 191–199, https://doi.org/10.1016/j.foreco.2016.05.041 (2016).Article 

Google Scholar 
Gillerot, L. et al. Inter- and intraspecific variation in mangrove carbon fraction and wood specific gravity in Gazi Bay, Kenya. Ecosphere 9, e02306, https://doi.org/10.1002/ecs2.2306 (2018).Article 

Google Scholar 
Gómez-Brandón, M. et al. Physico-chemical and microbiological evidence of exposure effects on Picea abies – coarse woody debris at different stages of decay. Forest Ecology and Management 391, 376–389, https://doi.org/10.1016/j.foreco.2017.02.033 (2017).Article 

Google Scholar 
Guner, S. T. & Comez, A. Biomass equations and changes in carbon stock in afforested Black Pine (Pinus nigra Arnold. subsp. pallasiana (Lamb.) Holmboe) stands in Turkey. Fresenius Environmental Bulletin 26, 2368–2379 (2017).CAS 

Google Scholar 
Guo, J., Chen, G., Xie, J., Yang, Z. & Yang, Y. Patterns of mass, carbon and nitrogen in coarse woody debris in five natural forests in southern China. Annals of Forest Science 71, 585–594, https://doi.org/10.1007/s13595-014-0366-4 (2014).Article 

Google Scholar 
Hanpattanakit, P., Chidthaisong, A., Sanwangsri, M. & Lichaikul, N. Improving allometric equations to estimate biomass and carbon in secondary dry dipterocarp forest. Singapore SG 18, 208–211 (2016).
Google Scholar 
Herrero De Aza, C., Turrión, M. B., Pando, V. & Bravo, F. Carbon in heartwood, sapwood and bark along the stem profile in three Mediterranean Pinus species. Annals of Forest Science 68, 1067–1076, https://doi.org/10.1007/s13595-011-0122-y (2011).Article 

Google Scholar 
Huet, S., Forgeard, F. O. & Nys, C. Above- and belowground distribution of dry matter and carbon biomass of Atlantic beech (Fagus sylvatica L.) in a time sequence. Annals of Forest Science 61, 683–694, https://doi.org/10.1051/forest:2004063 (2004).CAS 
Article 

Google Scholar 
Jacobs, D. F., Selig, M. F. & Severeid, L. R. Aboveground carbon biomass of plantation-grown American chestnut (Castanea dentata) in absence of blight. Forest Ecology and Management 258, 288–294, https://doi.org/10.1016/j.foreco.2009.04.014 (2009).Article 

Google Scholar 
Janssens, I. A. et al. Above- and belowground phytomass and carbon storage in a Belgian Scots pine stand. Annals of forest science 56, 81–90, https://doi.org/10.1051/forest:19990201 (1999).Article 

Google Scholar 
Jiménez Pérez, J., Treviño Garza, E. J. & Yerena Yamallel, J. I. Concentración de carbono en especies del bosque de pino-encino en la Sierra Madre Oriental. Revista mexicana de ciencias forestales 4, 50–61 (2013).Article 

Google Scholar 
Jomura, M. et al. Biotic and abiotic factors controlling respiration rates of above- and belowground woody debris of Fagus crenata and Quercus crispula in Japan. PLOS ONE 10, e0145113, https://doi.org/10.1371/journal.pone.0145113 (2015).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Jones, D. & O’Hara, K. Variation in carbon fraction, density, and carbon density in conifer tree tissues. Forests 9, 430, https://doi.org/10.3390/f9070430 (2018).Article 

Google Scholar 
Jones, D. A. & O’Hara, K. L. Carbon density in managed coast redwood stands: implications for forest carbon estimation. Forestry 85, 99–110, https://doi.org/10.1093/forestry/cpr063 (2012).Article 

Google Scholar 
Jones, D. A. & O’Hara, K. L. The influence of preparation method on measured carbon fractions in tree tissues. Tree Physiology 36, 1177–1189, https://doi.org/10.1093/treephys/tpw051 (2016).CAS 
Article 
PubMed 

Google Scholar 
Joosten, R. & Schulte, A. Possible effects of altered growth behaviour of Norway Spruce (Picea abies) on carbon accounting. Climatic Change 55, 115–129, https://doi.org/10.1023/a:1020227806137 (2002).CAS 
Article 

Google Scholar 
Joosten, R., Schumacher, J., Wirth, C. & Schulte, A. Evaluating tree carbon predictions for beech (Fagus sylvatica L.) in western Germany. Forest Ecology and Management 189, 87–96, https://doi.org/10.1016/j.foreco.2003.07.037 (2004).Article 

Google Scholar 
Kim, C., Yoo, B., Jung, S. & Lee, K. Allometric equations to assess biomass, carbon and nitrogen content of black pine and red pine trees in southern Korea. iForest – Biogeosciences and Forestry 10, 483–490, https://doi.org/10.3832/ifor2164-010 (2017).Article 

Google Scholar 
Kort, J. & Turnock, R. Carbon reservoirs and biomass in Canadian prairie shelterbelts. Agroforestry Systems 44, 175-186, https://doi.org/10.1023/a:1006226006785 (1998).Köster, K., Metslaid, M., Engelhart, J. & Köster, E. Dead wood basic density, and the concentration of carbon and nitrogen for main tree species in managed hemiboreal forests. Forest Ecology and Management 354, 35–42, https://doi.org/10.1016/j.foreco.2015.06.039 (2015).Article 

Google Scholar 
Kraenzel, M., Castillo, A., Moore, T. & Potvin, C. Carbon storage of harvest-age teak (Tectona grandis) plantations, Panama. Forest Ecology and Management 173, 213–225, https://doi.org/10.1016/s0378-1127(02)00002-6 (2003).Article 

Google Scholar 
Laiho, R. & Laine, J. Tree stand biomass and carbon content in an age sequence of drained pine mires in southern Finland. Forest Ecology and Management 93, 161–169, https://doi.org/10.1016/s0378-1127(96)03916-3 (1997).Article 

Google Scholar 
Laiho, R. & Prescott, C. E. The contribution of coarse woody debris to carbon, nitrogen, and phosphorus cycles in three Rocky Mountain coniferous forests. Canadian Journal of Forest Research 29, 1592–1603, https://doi.org/10.1139/x99-132 (1999).Article 

Google Scholar 
Lambert, R. L., Lang, G. E. & Reiners, W. A. Loss of mass and chemical change in decaying boles of a subalpine Balsam Fir forest. Ecology 61, 1460–1473, https://doi.org/10.2307/1939054 (1980).Article 

Google Scholar 
Laughlin, D. C., Leppert, J. J., Moore, M. M. & Sieg, C. H. A multi-trait test of the leaf-height-seed plant strategy scheme with 133 species from a pine forest flora. Functional Ecology 24, 493–501, https://doi.org/10.1111/j.1365-2435.2009.01672.x (2010).Article 

Google Scholar 
Li, X. et al. Biomass and carbon storage in an age-sequence of Korean Pine (Pinus koraiensis) plantation forests in central Korea. Journal of Plant Biology 54, 33–42, https://doi.org/10.1007/s12374-010-9140-9 (2011).ADS 
CAS 
Article 

Google Scholar 
Lombardi, F. et al. Investigating biochemical processes to assess deadwood decay of Beech and Silver Fir in Mediterranean mountain forests. Annals of Forest Science 70, 101–111, https://doi.org/10.1007/s13595-012-0230-3 (2013).Article 

Google Scholar 
Lutter, R., Tullus, A., Kanal, A., Tullus, T. & Tullus, H. The impact of former land-use type to above- and below-ground C and N pools in short-rotation hybrid aspen (Populus tremula L. × P. tremuloides Michx.) plantations in hemiboreal conditions. Forest Ecology and Management 378, 79–90, https://doi.org/10.1016/j.foreco.2016.07.021 (2016).Article 

Google Scholar 
Mahmood, H. et al. Applicability of semi-destructive method to derive allometric model for estimating aboveground biomass and carbon stock in the Hill Zone of Bangladesh. Journal of Forestry Research 31, 1235–1245, https://doi.org/10.1007/s11676-019-00881-5 (2020).CAS 
Article 

Google Scholar 
Maiti, R., Gonzalez Rodriguez, H. & Kumari, A. Wood density of ten native trees and shrubs and its possible relation with a few wood chemical compositions. American Journal of Plant Sciences 07, 1192–1197, https://doi.org/10.4236/ajps.2016.78114 (2016).CAS 
Article 

Google Scholar 
Mäkinen, H., Hynynen, J., Siitonen, J. & Sievänen, R. Predicting The decomposition of Scots Pine, Norway Spruce, and Birch stems in Finland. Ecological Applications 16, 1865–1879, https://doi.org/10.1890/1051-0761(2006)016[1865:ptdosp]2.0.co;2 (2006).Article 
PubMed 

Google Scholar 
Manuella, S. et al. Chemical transformations in downed logs and snags of mixed boreal species during decomposition. Canadian Journal of Forest Research 43, 785–798, https://doi.org/10.1139/cjfr-2013-0086 (2013).CAS 
Article 

Google Scholar 
Martin, A. R. & Thomas, S. C. Size-dependent changes in leaf and wood chemical traits in two Caribbean rainforest trees. Tree Physiology 33, 1338–1353, https://doi.org/10.1093/treephys/tpt085 (2013).CAS 
Article 
PubMed 

Google Scholar 
Martin, A. R., Thomas, S. C. & Zhao, Y. Size-dependent changes in wood chemical traits: a comparison of neotropical saplings and large trees. AoB PLANTS 5, plt039–plt039, https://doi.org/10.1093/aobpla/plt039 (2013).CAS 
Article 
PubMed Central 

Google Scholar 
Medlyn, B. E. et al. Effects of elevated [CO2] on photosynthesis in European forest species: a meta-analysis of model parameters. Plant, Cell & Environment 22, 1475–1495, https://doi.org/10.1046/j.1365-3040.1999.00523.x (1999).CAS 
Article 

Google Scholar 
Moreira, A. B., Gregoire, T. G. & Do Couto, H. T. Z. Wood density and carbon concentration of coarse woody debris in native forests, Brazil. Forest Ecosystems 6, https://doi.org/10.1186/s40663-019-0177-z (2019).Morhart, C., Sheppard, J. P., Schuler, J. K. & Spiecker, H. Above-ground woody biomass allocation and within tree carbon and nutrient distribution of wild cherry (Prunus avium L.) – a case study. Forest Ecosystems 3, https://doi.org/10.1186/s40663-016-0063-x (2016).Northup, B. K., Zitzer, S. F., Archer, S., Mcmurtry, C. R. & Boutton, T. W. Above-ground biomass and carbon and nitrogen content of woody species in a subtropical thornscrub parkland. Journal of Arid Environments 62, 23–43, https://doi.org/10.1016/j.jaridenv.2004.09.019 (2005).ADS 
Article 

Google Scholar 
Palviainen, M. & Finér, L. Decomposition and nutrient release from Norway spruce coarse roots and stumps – a 40-year chronosequence study. Forest Ecology and Management 358, 1–11, https://doi.org/10.1016/j.foreco.2015.08.036 (2015).Article 

Google Scholar 
Peri, P. L., Gargaglione, V., Martínez Pastur, G. & Lencinas, M. V. Carbon accumulation along a stand development sequence of Nothofagus antarctica forests across a gradient in site quality in Southern Patagonia. Forest Ecology and Management 260, 229–237, https://doi.org/10.1016/j.foreco.2010.04.027 (2010).Article 

Google Scholar 
Pompa-García, M. & Jurado, E. Carbon concentration in structures of Arctostaphylos pungens HBK: an alternative CO2 sink in forests. Phyton 84, 385-389 (2016).Pompa-García, M., Sigala-Rodríguez, J., Jurado, E. & Flores, J. Tissue carbon concentration of 175 Mexican forest species. iForest – Biogeosciences and Forestry 10, 754–758, https://doi.org/10.3832/ifor2421-010 (2017).Article 

Google Scholar 
Pompa-García, M. & Yerena-Yamalliel, J. I. Concentration of carbon in Pinus cembroides Zucc: potential source of global warming mitigation. Revista Chapingo Serie Ciencias Forestales y del Ambiente 20, 169–175, https://doi.org/10.5154/r.rchscfa.2014.04.014 (2014).Article 

Google Scholar 
Preston, C. M., Trofymow, J. A. & Flanagan, L. B. Decomposition, δ13C, and the “lignin paradox”. Canadian Journal of Soil Science 86, 235–245, https://doi.org/10.4141/s05-090 (2006).CAS 
Article 

Google Scholar 
Preston, C. M., Trofymow, J. A. & Nault, J. R. Decomposition and change in N and organic composition of small-diameter Douglas-fir woody debris over 23 years. Canadian Journal of Forest Research 42, 1153-1167 (2012).CAS 
Article 

Google Scholar 
Preston, C. M., Trofymow, J. A., Niu, J. & Fyfe, C. A. PMAS-NMR spectroscopy and chemical analysis of coarse woody debris in coastal forests of Vancouver Island. Forest Ecology and Management 111, 51–68, https://doi.org/10.1016/s0378-1127(98)00307-7 (1998).Article 

Google Scholar 
Rana, R., Langenfeld-Heyser, R., Finkeldey, R. & Polle, A. FTIR spectroscopy, chemical and histochemical characterisation of wood and lignin of five tropical timber wood species of the family of Dipterocarpaceae. Wood Science and Technology 44, 225–242, https://doi.org/10.1007/s00226-009-0281-2 (2010).CAS 
Article 

Google Scholar 
Ray, R., Majumder, N., Chowdhury, C. & Jana, T. K. Wood chemistry and density: an analog for response to the change of carbon sequestration in mangroves. Carbohydrate Polymers 90, 102–108, https://doi.org/10.1016/j.carbpol.2012.05.001 (2012).CAS 
Article 
PubMed 

Google Scholar 
Rodrigues, D. P., Hamacher, C., Estrada, G. C. D. & Soares, M. L. G. Variability of carbon content in mangrove species: effect of species, compartments and tidal frequency. Aquatic Botany 120, 346–351, https://doi.org/10.1016/j.aquabot.2014.10.004 (2015).CAS 
Article 

Google Scholar 
Sakai, Y., Ugawa, S., Ishizuka, S., Takahashi, M. & Takenaka, C. Wood density and carbon and nitrogen concentrations in deadwood of Chamaecyparis obtusa and Cryptomeria japonica. Soil Science and Plant Nutrition 58, 526–537, https://doi.org/10.1080/00380768.2012.710526 (2012).CAS 
Article 

Google Scholar 
Sandström, F., Petersson, H., Kruys, N. & Ståhl, G. Biomass conversion factors (density and carbon concentration) by decay classes for dead wood of Pinus sylvestris, Picea abies and Betula spp. in boreal forests of Sweden. Forest Ecology and Management 243, 19–27, https://doi.org/10.1016/j.foreco.2007.01.081 (2007).Article 

Google Scholar 
Sanquetta, M. N. I., Sanquetta, C. R., Mognon, F., Corte, A. P. D. & Maas, G. C. B. Wood density and carbon content in young teak individuals from Pará, Brazil. Científica 44, 608, https://doi.org/10.15361/1984-5529.2016v44n4p608-614 (2016).Article 

Google Scholar 
Schwendenmann, L. & Mitchell, N. Carbon accumulation by native trees and soils in an urban park, Auckland. New Zealand Journal of Ecology 38(20), 213–220 (2014).Setälä, H., Marshall, V. G. & Trofymow, J. A. Influence of micro- and macro-habitat factors on collembolan communities in Douglas-fir stumps during forest succession. Applied Soil Ecology 2, 227–242, https://doi.org/10.1016/0929-1393(95)00053-9 (1995).Article 

Google Scholar 
Sohrabi, H., Bakhtiarvand-Bakhtiari, S. & Ahmadi, K. Above- and below-ground biomass and carbon stocks of different tree plantations in central Iran. Journal of Arid Land 8, 138–145, https://doi.org/10.1007/s40333-015-0087-z (2016).Article 

Google Scholar 
Telmo, C., Lousada, J. & Moreira, N. Proximate analysis, backwards stepwise regression between gross calorific value, ultimate and chemical analysis of wood. Bioresource Technology 101, 3808–3815, https://doi.org/10.1016/j.biortech.2010.01.021 (2010).CAS 
Article 
PubMed 

Google Scholar 
Thomas, A. L. et al. Carbon and nitrogen accumulation within four black walnut alley cropping sites across Missouri and Arkansas, USA. Agroforestry Systems 94, 1625–1638, https://doi.org/10.1007/s10457-019-00471-8 (2020).Article 

Google Scholar 
Thomas, S. C. & Malczewski, G. Wood carbon content of tree species in Eastern China: interspecific variability and the importance of the volatile fraction. Journal of Environmental Management 85, 659–662, https://doi.org/10.1016/j.jenvman.2006.04.022 (2007).CAS 
Article 
PubMed 

Google Scholar 
Tolunay, D. Carbon concentrations of tree components, forest floor and understorey in young Pinus sylvestris stands in north-western Turkey. Scandinavian Journal of Forest Research 24, 394–402, https://doi.org/10.1080/02827580903164471 (2009).Article 

Google Scholar 
Tramoy, R., Sebilo, M., Nguyen Tu, T. T. & Schnyder, J. Carbon and nitrogen dynamics in decaying wood: paleoenvironmental implications. Environmental Chemistry 14, 9, https://doi.org/10.1071/en16049 (2017).CAS 
Article 

Google Scholar 
Van Geffen, K. G., Poorter, L., Sass-Klaassen, U., Van Logtestijn, R. S. P. & Cornelissen, J. H. C. The trait contribution to wood decomposition rates of 15 Neotropical tree species. Ecology 91, 3686–3697, https://doi.org/10.1890/09-2224.1 (2010).Article 
PubMed 

Google Scholar 
Wang, G. et al. Variations in the live biomass and carbon pools of Abies georgei along an elevation gradient on the Tibetan Plateau, China. Forest Ecology and Management 329, 255–263, https://doi.org/10.1016/j.foreco.2014.06.023 (2014).Article 

Google Scholar 
Wang, X. W., Weng, Y. H., Liu, G. F., Krasowski, M. J. & Yang, C. P. Variations in carbon concentration, sequestration and partitioning among Betula platyphylla provenances. Forest Ecology and Management 358, 344–352, https://doi.org/10.1016/j.foreco.2015.08.029 (2015).Article 

Google Scholar 
Watzlawick, L. F. et al. Teores de carbono em espécies da floresta ombrófila mista e efeito do grupo ecológico. Cerne 20, 613–620, https://doi.org/10.1590/01047760201420041492 (2014).Article 

Google Scholar 
Weber, J. C. et al. Variation in growth, wood density and carbon concentration in five tree and shrub species in Niger. New Forests 49, 35–51, https://doi.org/10.1007/s11056-017-9603-7 (2018).Article 

Google Scholar 
Weggler, K., Dobbertin, M., Jüngling, E., Kaufmann, E. & Thürig, E. Dead wood volume to dead wood carbon: the issue of conversion factors. European Journal of Forest Research 131, 1423–1438, https://doi.org/10.1007/s10342-012-0610-0 (2012).Article 

Google Scholar 
Widagdo, F. R. A., Xie, L., Dong, L. & Li, F. Origin-based biomass allometric equations, biomass partitioning, and carbon concentration variations of planted and natural Larix gmelinii in northeast China. Global Ecology and Conservation 23, e01111, https://doi.org/10.1016/j.gecco.2020.e01111 (2020).Article 

Google Scholar 
Wu, H. et al. Tree functional types simplify forest carbon stock estimates induced by carbon concentration variations among species in a subtropical area. Scientific Reports 7, https://doi.org/10.1038/s41598-017-05306-z (2017).Xing, Z. et al. Carbon and biomass partitioning in balsam fir (Abies balsamea). Tree Physiology 25, 1207–1217, https://doi.org/10.1093/treephys/25.9.1207 (2005).Article 
PubMed 

Google Scholar 
Yang, F.-F. et al. Dynamics of coarse woody debris and decomposition rates in an old-growth forest in lower tropical China. Forest Ecology and Management 259, 1666–1672, https://doi.org/10.1016/j.foreco.2010.01.046 (2010).Article 

Google Scholar 
Yeboah, D., Burton, A. J., Storer, A. J. & Opuni-Frimpong, E. Variation in wood density and carbon content of tropical plantation tree species from Ghana. New Forests 45, 35–52, https://doi.org/10.1007/s11056-013-9390-8 (2014).Article 

Google Scholar 
Ying, J., Weng, Y., Oswald, B. P. & Zhang, H. Variation in carbon concentrations and allocations among Larix olgensis populations growing in three field environments. Annals of Forest Science 76, https://doi.org/10.1007/s13595-019-0877-0 (2019).Yuan, J., Cheng, F., Zhu, X., Li, J. & Zhang, S. Respiration of downed logs in pine and oak forests in the Qinling Mountains, China. Soil Biology and Biochemistry 127, 1–9, https://doi.org/10.1016/j.soilbio.2018.09.012 (2018).CAS 
Article 

Google Scholar 
Zabek, L. M. & Prescott, C. E. Biomass equations and carbon content of aboveground leafless biomass of hybrid poplar in Coastal British Columbia. Forest Ecology and Management 223, 291–302, https://doi.org/10.1016/j.foreco.2005.11.009 (2006).Article 

Google Scholar 
Zhang, H., Jiang, Y., Song, M., He, J. & Guan, D. Improving understanding of carbon stock characteristics of Eucalyptus and Acacia trees in southern China through litter layer and woody debris. Scientific Reports 10, https://doi.org/10.1038/s41598-020-61476-3 (2020).Zhang, Q., Wang, C., Wang, X. & Quan, X. Carbon concentration variability of 10 Chinese temperate tree species. Forest Ecology and Management 258, 722–727, https://doi.org/10.1016/j.foreco.2009.05.009 (2009).Article 

Google Scholar 
Zheng, H. et al. Variation of carbon storage by different reforestation types in the hilly red soil region of southern China. Forest Ecology and Management 255, 1113–1121, https://doi.org/10.1016/j.foreco.2007.10.015 (2008).Article 

Google Scholar 
Zhou, L. et al. Tissue-specific carbon concentration, carbon stock, and distribution in Cunninghamia lanceolata (Lamb.) Hook plantations at various developmental stages in subtropical China. Annals of Forest Science 76, https://doi.org/10.1007/s13595-019-0851-x (2019).Zanne, A. E. et al. Three keys to the radiation of angiosperms into freezing environments. Nature 506, 89–92, https://doi.org/10.1038/nature12872 (2014).ADS 
CAS 
Article 
PubMed 

Google Scholar 
Muscarella, R. et al. The global abundance of tree palms. Global Ecology and Biogeography 29, 1495–1514, https://doi.org/10.1111/geb.13123 (2020).Article 

Google Scholar 
Du, H. et al. Mapping global bamboo forest distribution using multisource remote sensing data. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 11, 1458–1471 (2018).ADS 
Article 

Google Scholar 
Iwashita, D. K., Litton, C. M. & Giardina, C. P. Coarse woody debris carbon storage across a mean annual temperature gradient in tropical montane wet forest. Forest Ecology and Management 291, 336–343, https://doi.org/10.1016/j.foreco.2012.11.043 (2013).Article 

Google Scholar 
R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria (2021).Wikström, N., Savolainen, V. & Chase, M. W. Evolution of the angiosperms: calibrating the family tree. Proceedings of the Royal Society of London. Series B: Biological Sciences 268, 2211–2220, https://doi.org/10.1098/rspb.2001.1782 (2001).Article 
PubMed 
PubMed Central 

Google Scholar 
Gastauer, M. & Meira Neto, J. A. A. Updated angiosperm family tree for analyzing phylogenetic diversity and community structure. Acta Botanica Brasilica 31, 191–198, https://doi.org/10.1590/0102-33062016abb0306 (2017).Article 

Google Scholar  More

cpDNA haplotype databaseDNA sequencing of the choloroplast (cp) markers produced sequences of the following lengths: 573 bp (atpB-rbcL); 487 bp (petG-trnP); 500 bp (trnL1-trnL2); and 593 bp (psbM-trnD). Alignment of the 352 individuals from the 44 populations yielded a total 28 variable sites: 11 in the atpB-rbcL spacer, seven in both the petG-trnP and psbM-trnD spacers, and three in the trnL1-trnL2 spacer (Supplementary Table S1). Based on these 28 variable sites (21 base substitutions and 7 deletions) across the combined intergenic regions, a total of 22 unique haplotypes were found (Fig. 1a).Figure 1(a) Chloroplast haplotype distribution in the Shorea leprosula populations. The pie chart colours indicate haplotype distributions; and sector areas are proportional to sample size (Map was generated by ArcGIS-ArcMap version 10.8). (b) STRUCTURE analysis identified two clusters (K = 2) corresponding to Region A and B.Full size imageSSR allele frequency databaseThe reproducibility of SSR genotyping was confirmed by achieving consistent genotypes from five independent PCR amplifications on a single individual for each of the ten SSR loci. Individual bar plots from STRUCTURE analysis are presented in Fig. 1b. At the highest Delta K likelihood scores, the best representation of the data was K = 2 suggesting that the 44 populations in Peninsular Malaysia can be divided into two main genetic clusters: Region A and Region B. The first cluster, ‘Region A’ consists of 12 populations, namely SBadak, BPerangin, BEnggang, GJerai, RTelui, GInas, GBongsu, Belum, Piah, BHijau, Korbu and Bubu. The second cluster, ‘Region B’ consists of 32 populations, namely Behrang, Ampang, HGombak, HLangat, SLalang, PPanjang, Berembun, Angsi, Kenaboi, Triang, Pasoh, BSenggeh, GLedang, Krau, TNegara, Terenggun, SBetis, USat, CTongkat, HTerengganu, Jengai, AGading, Tekam, Beserah, Jengka, Lentang, Lesong, ERompin, GArong, Labis, AHitam and Panti. Similarly, the UPGMA dendrogram analysis also divided the 44 populations into two genetic clusters (Fig. 2) corresponding to Region A and B of the STRUCTURE result.Figure 2Dendrogram showing the relationship between 44 populations of Shorea leprosula in Peninsular Malaysia based on the UPGMA cluster analysis of SSR markers.Full size imageSSR allele frequency databases were established according to Region A and B, and characterized to evaluate the relative usefulness of each SSR marker in forensic investigation. The distribution of allele frequencies for each locus is listed in Table S2 (Region A database) and Table S3 (Region B database). Forensic parameters are shown in Table 1, with a total of 143 alleles and 174 alleles detected in the Region A and B databases, respectively. The observed (Ho) and expected (He) heterozygosity ranged from 0.3570 to 0.8346 and 0.4375 to 0.8795, respectively for populations in the Region A database; and ranged from 0.3298 to 0.8356 and 0.3469 to 0.8793, respectively for populations in the Region B database. The power of discrimination (PD) for the SSR loci ranged from 0.601 to 0.972 and 0.554 to 0.975, in Region A and B databases, respectively. The most discriminating locus was Sle605 in both the Region A (PD = 0.972) and Region B (PD = 0.975) databases. Minimum allele frequency was adjusted for alleles falling below the thresholds of 0.0066 (Region A) and 0.0024 (Region B).Table 1 Genetic diversity and forensic variables (A: total number of alleles; Ho: observed heterozygosity; He: expected heterozygosity; PIC: polymorphic information content; HWE: Hardy–Weinberg equilibrium; MP: matching probability; PD: power of discrimination) for each the 10 SSR loci of Shorea leprosula in the Region A and B databases.Full size tableDeviations from HWE were detected in four of the SSR loci for Region A (SleT11, SleT15, SleT17 and Sle465) and six SSR loci in Region B (SleT01, SleT11, SleT15, SleT17, SleT29 and SleT31). We evaluated these loci in each population independently to rule out the possible presence of null alleles. There were four populations in Region A (GJerai, RTelui, GBongsu and Piah) where a single one locus deviated from HWE; whereas there were eight populations in Region B (Behrang, HGombak, SLalang, Angsi, Klau, USat, Jengka and Panti) with a single locus and a single population (GLedang) with two loci that deviated from HWE (Table S4). Observed deviation from HWE was substantially lower in each population (either absence or not more than two loci) and thus it might be due to Wahlund effect caused by population substructuring in both Region A and B. Linkage disequilibrium (LD) testing was used to evaluate the independence of frequencies for all the SSR genotypes. A total of 13.3% and 28.9% of the 45 pairwise loci were found significant evidence of LD for Region A and B, respectively. Some of the loci might be linked as a result of population substructuring and inbreeding (inbreeding coefficient = 0.0822 [Peninsular Malaysia]). These results are in line with observations in real populations, where the assumption of completely random mating and zero migration required for HWE and LD are unlikely to be met, either in humans, animals or plants 21,22,23.Mean self-assignment, the proportion of individuals correctly assigned back to their population, was 45.9% and ranged from 14.3% (Kenaboi) to 81.3% (CTongkat) between population (Table 2). At the regional level, correct assignment rate of individuals to their region of origin was higher, 87.4% for Region A and 90.0% for Region B, (average of 88.7%).Table 2 Self-assignment test outcomes for Shorea leprosula individuals at the population and regional levels.Full size tableConservativeness of the databaseThe coancestry coefficient (θ) for Peninsular Malaysia (0.0579) was higher than those of Region A (0.0454) and Region B (0.0500) (Table 3). A total of 4.54% and 5.00% of the genetic variability was distributed among populations within Region A and Region B, respectively. In terms of inbreeding coefficient (f), the value for the Region A database (f = 0.0892) was highest, followed by Peninsular Malaysia (f = 0.0822) and Region B (f = 0.0666). All the θ and f values were significantly greater than zero, demonstrated by the 95% confidence intervals not overlapping with zero. Both of the θ and f values were used to calculate the conservativeness of each database by testing the cognate database (Porigin) against the regional database (Pcombined). The databases were non-conservative at the calculated θ value. In order for both the Region databases (A and B) to be conservative, the value of θ was adjusted from 0.0454 to 0.1900 for Region A and from 0.0500 to 0.1500 for Region B. For the Region A database, the most common SSR profile frequency is 2.69 × 10–7 or 1 in 3.72 million and the rarest profile frequency is 1.84 × 10–14 or 1 in 54.3 trillion. For the Region B database, the most common SSR profile frequency is 1.06 × 10–7 or 1 in 9.43 million and the rarest profile frequency is 4.03 × 10–16 or 1 in 2.48 quadrillion.Table 3 Coancestry (θ) and inbreeding (f) coefficients for Shorea leprosula at each hierarchical level.Full size table More

Land Use Land Cover taxonomyThe classification schema or “taxonomy” for Dynamic World, shown in Table 1, was determined after a review of global LULC maps, including the USGS Anderson classification system18, ESA Land Use and Coverage Area frame Survey (LUCAS) land cover modalities19, MapBiomas classification20, and GlobeLand30 land cover types13. The Dynamic World taxonomy maintains a close semblance to the land use classes presented in the IPCC Good Practice Guidance (forest land, grassland, cropland, wetland, settlement, and other)21 to ensure easier application of the resulting data for estimating carbon stocks and greenhouse gas emissions. Unlike single-pixel labels, which are usually defined in terms of percent cover thresholds, the Dynamic World taxonomy was applied using “dense” polygon-based annotations such that LULC labels are applied to areas of relatively homogenous cover types with similar colors and textures.Table 1 Dynamic World Land Use Land Cover (LULC) classification taxonomy.Full size tableTraining dataset collectionOur modeling approach relies on semi-supervised deep learning and requires spatially dense (i.e., ideally wall-to-wall) annotations. To collect a diverse set of training and evaluation data, we divided the world into three regions: the Western Hemisphere (160°W to 20°W), Eastern Hemisphere-1 (20°W to 100°E), and Eastern Hemisphere-2 (100°E to 160°W). We further divided each region by the 14 RESOLVE Ecoregions biomes22. We collected a stratified sample of sites for each biome per region based on NASA MCD12Q1 land cover for 20174. Given the availability of higher-resolution LULC maps in the United States and Brazil, we used the NLCD 201610 and MapBiomas 201720 LULC products respectively in place of MODIS products for stratification in these two countries.At each sample location, we performed an initial selection of Sentinel-2 images from 2019 scenes based on image cloudiness metadata reported in the Sentinel-2 tile’s QA60 band. We further filtered scenes to remove images with many masked pixels. We finally extracted individual tiles of 510 × 510 pixels centered on the sample sites from random dates in 2019. Tiles were sampled in the UTM projection of the source image and we selected one tile corresponding to a single Sentinel-2 ID number and single date.Further steps were then taken to obtain an “as balanced as possible” training dataset with respect to the LULC classifications from the respective LULC products. In particular, for each Dynamic World LULC category contained within a tile, the tile was labeled to be high, medium, or low in that category. We then selected an approximately equal number of tiles with high, medium or low category labels for each category.To achieve a large dataset of labeled Sentinel-2 scenes, we worked with two groups of annotators. The first group included 25 annotators with previous photo-interpretation and/or remote sensing experience. The expert group labeled approximately 4,000 image tiles (Fig. 1a), which were then used to train and measure the performance and accuracy of a second “non-expert” group of 45 additional annotators who labeled a second set of approximately 20,000 image tiles (Fig. 1b). A final validation set of 409 image tiles were held back from the modeling effort and used for evaluation as described in the Technical Validation section. Each image tile in the validation set was annotated by three experts and one non-expert to facilitate cross-expert and expert/non-expert QA comparisons.Fig. 1Global distribution of annotated Sentinel-2 image tiles used for model training and periodic testing (neither including 409 validation tiles). (a) 4,000 tiles interpreted by a group of 25 experts (b) 20,000 tiles interpreted by a group of 45 non-experts. Hexagons represent approximately 58,500 km2 areas and shading corresponds to the count of annotated tile centroids per hexagon.Full size imageAll Dynamic World annotators used the Labelbox platform23, which provides a vector drawing tool to mark the boundaries of feature classes directly over the Sentinel-2 tile (Fig. 2). We instructed both expert and non-expert annotators to use dense markup instead of single pixel labels with a minimum mapping unit of 50 × 50 m (5 × 5 pixels). For water, trees, crops, built area, bare ground, snow & ice, and cloud, this was a fairly straightforward procedure at the Sentinel-2 10 m resolution since these feature classes tend to appear in fairly homogenous agglomerations. Shrub & scrub and flooded vegetation classes proved to be more challenging as they tended not to appear as homogenous features (e.g. mix of vegetation types) and have variable appearance. Annotators used their best discretion in these situations based on the guidance provided in our training material (i.e. descriptions and examples in Table 1). In addition to the Sentinel-2 tile, annotators had access to a matching high-resolution satellite image via Google Maps and ground photography via Google Street View from the image center point. We also provided the date and center point coordinates for each annotation. All annotators were asked to label at least 70% of a tile within 20 to 60 minutes and were allowed to skip some tiles to best balance their labeling accuracy with their efficiency.Fig. 2Sentinel-2 tile and example reference annotation provided as part of interpreter training. This example was used to illustrate the Flooded vegetation class, which is distinguished by small “mottled” areas of water mixed with vegetation near a riverbed. Also note that some areas of the tile are left unlabeled.Full size imageImage preprocessingWe prepared Sentinel-2 imagery in a number of ways to accommodate both annotation and training workflows. An overview of the preprocessing workflow is shown in Fig. 3.Fig. 3Training inputs workflow. Annotations created using Sentinel-2 Level 2 A Surface Reflectance imagery are paired with masked and normalized Sentinel-2 Level 1 C Top of Atmosphere imagery, and inputs are augmented to create training inputs used for modeling. Cloud and shadow masking involves a three-step process that combines the Sentinel-2 Cloud Probability (S2C) product with the Cloud Displacement Index (CDI), which is used to correct over-masking of bright non-cloud targets” and directional distance transform (DDT), which is used to remove the expected path of shadows based on sun-sensor geometry.Full size imageFor training data collection, we used the Sentinel-2 Level-2A (L2A) product, which provides radiometrically calibrated surface reflectance (SR) processed using the Sen2Cor software package24. This advanced level of processing was advantageous for annotation, as it attempts to remove inter-scene variability due to solar distance, zenith angle, and atmospheric conditions. However, systematically produced Sentinel-2 SR products are currently only available from 2017 onwards. Therefore, for our modeling approach, we used the Level-1C (L1C) product, which has been generated since the beginning of the Sentinel-2 program in 2015. The L1C product represents Top-of-Atmosphere (TOA) reflectance measurements and is not subject to a change in processing algorithm in the future. We note that for any L2A image, there is a corresponding L1C image, allowing us to directly map annotations performed using L2A imagery to the L1C imagery used in model training. All bands except for B1, B8A, B9, and B10 were kept, with all bands bilinearly upsampled to 10 m for both training and inference.In addition to our preliminary cloud filtering in training image selection, we adopted and applied a novel masking solution that combines several existing products and techniques. Our procedure is to first take the 10 m Sentinel-2 Cloud Probability (S2C) product available in Earth Engine25 and join it to our working set of Sentinel-2 scenes such that each image is paired with the corresponding mask. We compute a cloud mask by thresholding S2C using a cloud probability of 65% to identify pixels that are likely obscured by cloud cover. We then apply the Cloud Displacement Index (CDI) algorithm26 and threshold the result to produce a second cloud mask, which is intersected with the S2C mask to reduce errors of commission by removing bright non-cloud targets based on Sentinel-2 parallax effects. We finally intersect this sub-cirrus mask with a threshold on the Sentinel-2 cirrus band (B10) using the thresholding constants proposed for the CDI algorithm26, and take a morphological opening of this as our cloudy pixel mask. This mask is computed at 20 m resolution.In order to remove cloud shadows, we extend the cloudy pixel mask 5 km in the direction opposite the solar azimuthal angle using the scene level metadata “SOLAR_AZIMUTH_ANGLE” and a directional distance transform (DDT) operation in Earth Engine. The final cloud and shadow mask is resampled to 100 m to decrease both the data volume and processing time. The resulting mask is applied to Sentinel-2 images used for training and inference such that unmasked pixels represent observations that are likely to be cloud- and shadow-free.The distribution of Sentinel-2 reflectance values are highly compressed towards the low end of the sensor range, with the remainder mostly occupied by high return phenomena like snow and ice, bare ground, and specular reflection. To combat this imbalance, we introduce a normalization scheme that better utilizes the useful range of Sentinel-2 reflectance values for each band. We first log-transform the raw reflectance values to equalize the long tail of highly reflective surfaces, then remap percentiles of the log-transformed values to points on a sigmoid function. The latter is done to bound on (0, 1) without truncation, and condenses the extreme end members of reflectances to a smaller range.To account for an annotation skill differential between the non-expert and expert groups, we one-hot encode the labeled pixels, and smooth them according to the confidence in a binary label of the individual annotator (expert/non-expert): this is effectively linearly interpolating the distributions per-pixel from their one-hot encoding (i.e. a vector of binary variables for each class label) to uniform probability. We used 0.2 for experts, and 0.3 for non-experts (i.e. ~82% confidence on the true class for experts and ~73% confidence on the true class for the non-expert. We note that these values approximately mirror the Non-Expert to Expert Consensus agreement as discussed in the Technical Validation section). This is akin to standard label-smoothing27,28, with the addition that the degree of smoothing is associated with annotation confidence.We generate a pair of weights for each pixel in an augmented example designed to compensate for class imbalance across the training set and weight high-frequency spatial features at the inputs during “synthesis” (discussed further in the following section). We also include a weight per pixel designed to attenuate labels in the center of labeled polygons where human annotators often missed small details using a simple edge finding kernel.We finally perform a series of augmentations (random rotation and random per-band contrasting) to our input data to improve generalizability and performance of our model. These augmentations are applied four times to each example to yield our final training dataset of examples paired with class distributions, masks, and weights (Fig. 3).Model trainingOur broad approach to transferring the supervised label data to a system that could be applied globally was to train a Fully Convolutional Neural Network (FCNN)29. Conceptually, this approach transforms pre-processed Sentinel-2 optical bands to a discrete probability distribution of the classes in our taxonomy on the basis of spatial context. This is done per-image with the assumption that sufficient spatial and spectral context is available to recover one of our taxonomic labels at a pixel. There are a few notable benefits to such an approach: namely that given the generalizability of modern deep neural networks, it is possible, as we will show, to produce a single model that achieves acceptable agreement with hand-digitized expert annotations globally. Furthermore, since model outputs are generated from a single image and a single model, it is straightforward to scale as each Sentinel-2 L1C image need only be observed once.Although applying CNN modeling, including FCNN, to recover LULC is not a new idea30,31,32, we introduce a number of novel innovations that achieve state-of-the-art performance on LULC globally with a neural network architecture almost 100x smaller than architectures used for semantic segmentation or regression of ground-level camera imagery (specifically compared to U-Net33 and DeepLab v3+34 architectures). Our approach also leverages weak supervision by way of a synthesis pathway: this pathway includes a replica of the labeling model architecture that learns a mapping from estimated probabilities back to the input reflectances, in a way, a reverse LULC classifier that offers both multi-tasking and a constraint to overcome deficiencies in human labeling (Fig. 4).Fig. 4Training protocol used to recover the labeling model. The bottom row shows the progression from a normalized Sentinel-2 L1C image, to class probabilities, to synthesized Sentinel-2. The dashed red and blue arrows show how the labeling model is optimized with respect to both the class probability and synthesis pathway, and the synthesis model is optimized only with respect to the synthesized imagery. The example image is retrieved from Earth Engine using ee.Image(‘GOOGLE/DYNAMICWORLD/V1/20190517T083601_20190517T083604_T37UET’).Full size imageNear real-time inferenceUsing Earth Engine in combination with Cloud AI Platform, it is possible to handle enormous quantities of satellite data and apply custom image processing and classification methods using a simple scaling paradigm (Fig. 5). To generate our NRT products, we apply the normalization described earlier to the raw Sentinel-2 L1C imagery and pass all normalized bands except B1, B8A, B9 and B10 after bilinear upscaling to ee.Model.predictImage. This output is then masked using our cloud mask derived from the unnormalized L1C image. Creation of these images is triggered automatically when new Sentinel-2 L1C and S2C images are available. The NRT collection is continuously updated with new results. For a full Sentinel-2 tile (roughly 100 km x 100 km), predictions are completed on the order of 45 minutes. In total, we evaluate ~12,000 Sentinel-2 scenes per day, processing half on average due to a filter criteria on the CLOUDY_PIXEL_PERCENTAGE metadata of 35%. A new Dynamic World LULC image is processed approximately every 14.4 s.Fig. 5Near-Real-Time (NRT) prediction workflow. Input imagery is normalized following the same protocol used in training and the trained model is applied to generate land cover predictions. Predicted results are masked to remove cloud and cloud shadow artifacts using Sentinel-2 cloud probabilities (S2C), the Cloud Displacement Index (CDI) and a directional distance transform (DDT), then added to the Dynamic World image collection.Full size image More