Introduction
Dragon trees, i.e. arboreal Dracaena spp., are ancient monocotyledons still found naturally in tropical and subtropical habitats surrounding the Sahara Desert (Maděra et al. Reference Maděra, Attorre, Habrová and Van Damme2021). Among them, Dracaena draco (L.) L. subsp. draco is a species endemic to the Macaronesian region (i.e., the archipelagos of Madeira, Canary Islands, and Cape Verde) and the Moroccan Anti-Atlas. The largest dragon trees, which can reach ∼20 m in height and in circumference, are iconic plants characterized by large roots, which sometimes merge with the thick, erect trunk at the base. Their growth, after an initial height increase, is characterized by sympodial branching, i.e., each branch order is created from a sleeping lateral bud growing up after an apical leaf rosette blooms and the terminal bud dies. This process usually takes place after the flowering process, so that older individuals have a highly divided, fan-shaped crown (Lengálová et al. Reference Lengálová, Kalivodová, Habrová, Maděra, Tesfamariam and Šenfeldr2020). In addition, peripheral branches, which have a characteristic “sausage-shaped” form, have been used to infer plant age together with the number of growth tops either flowering or not (Adolt et al. Reference Adolt, Habrova and Madera2012; Adolt and Pavlis Reference Adolt and Pavlis2004).
Besides their botanical and conservation value, dragon trees have acquired numerous social and cultural meanings, including traditional and medicinal significance related, for example, to the valuable crimson red resin known as “dragon’s blood” (Gupta et al. Reference Gupta, Bleakley and Gupta2008). A famous dragon tree, located at Icod de los Vinos in Tenerife, reaches a height of ∼18 m and a basal stem circumference of ∼18.5 m; it is thought to be several centuries old according to the most recent estimates (Lengálová et al. Reference Lengálová, Kalivodová, Habrová, Maděra, Tesfamariam and Šenfeldr2020). Many aspects of the species’ physiology and reproductive biology are unknown (Durán et al. Reference Durán, Marrero, Msanda, Harrouni, Gruenstaeudl, Patiño, Caujapé-Castells and García-Verdugo2020), and it is unclear why in the Canary Islands dragon trees are found naturally only on the islands of Tenerife and Gran Canaria, while they are absent in their natural state on other islands such as La Gomera and La Palma. The 694 wild specimens found on Tenerife are located almost entirely in the older geological areas of the island: the massifs of Anaga, Teno and Adeje, with a total area of 21.8 km2 (Almeida Pérez Reference Almeida Pérez, Bañares, Blanca, Güemes, Moreno and Ortiz2004).
The species is considered “Endangered (EN B2ab(ii,iii,iv,v)) as a result of the restricted area of occupancy, severe fragmentation, and the ongoing decline in the quality and area of habitat” (Silva et al. Reference Silva, Caujapé-Castells, Lobo, Casimiro, Moura, Elias, Fernandes, Fontinha and Romeiras2021). As a consequence, in recent years there has been renewed research efforts aimed at understanding its germination responses (Cartereau et al. Reference Cartereau, Baumel, Leriche, Médail, Santos Guerra and Saatkamp2023), root and stem growth anatomy (Jura-Morawiec et al. Reference Jura-Morawiec, Oskolski and Simpson2021; Marcinkiewicz and Jura-Morawiec Reference Marcinkiewicz and Jura-Morawiec2024), aerial roots function (Krawczyszyn and Krawczyszyn Reference Krawczyszyn and Krawczyszyn2014), and maximum lifespan (Biondi et al. Reference Biondi, Santos, Rodríguez and Sosa2024). This latest aspect is closely linked with the analysis of plant life history and growth patterns, but age estimation models based on identifying past growth patterns from current external features are not entirely satisfactory, and a review of the scientific literature on dragon trees concluded that “we still do not understand how to estimate age reliably” (Maděra et al. Reference Maděra, Forrest, Hanáček, Vahalík, Gebauer, Plichta, Jupa, Van Rensburg, Morris, Nadezhdina, Vaníčková, Jura-Morawiec, Wiland-Szymańska, Kalivodová, Lengálová, Rejžek and Habrová2020).
A scientific ageing tool for plants with secondary growth that do not form clear rings is radiocarbon (14C) dating, which is gaining increasing popularity because of technological advances and widespread applications (Palli et al. submitted). Since the late 1970s, accelerator mass spectrometry (AMS) has made 14C dating more affordable by reducing sample sizes (∼100 times smaller than those needed for decay counting) and providing faster results (days rather than weeks). Breakthroughs over the last two decades have led to further reduction of mass requirements (Santos et al. Reference Santos, Southon, Griffin, Beaupre and Druffel2007) as well as higher reliability of the produced results (Bronk Ramsey Reference Bronk Ramsey2023), thus reducing 14C age uncertainties. But since radiocarbon values are affected by fluctuations in 14C production rates caused by solar activity or changes in the Earth’s magnetic field, calibration of radiocarbon ages is still needed.
The massive input of excess-14C due to nuclear bomb tests between 1955 and 1963 allows us to resolve calendar ages within 1–2 years. The post-1950 CE period is therefore commonly referred to as “post-bomb” while prior years become “pre-bomb”. Recent 14C dataset revisions and new additions have allowed to refine the post-bomb global compilations and extended them to 2019 (Hua et al. Reference Hua, Turnbull, Santos, Rakowski, Ancapichún, De Pol-Holz, Hammer, Lehman, Levin, Miller, Palmer and Turney2022). Therefore, the IntCal post-1950 datasets have helped with determining annual growth patterns of pantropical tree species with poorly defined anatomical features (i.e., vague, discontinuous, false, and absent tree rings). With this approach, lack of seasonality has been found in Juniperus procera from Ethiopia (Wils et al. Reference Wils, Robertson, Eshetu, Sass-Klaassen and Koprowski2009), Hymenaea courbaril from southwestern Brazilian Amazonia (Santos et al. Reference Santos, Rodriguez, Barreto, Assis-Pereira, Barbosa, Roig and Tomazello-Filho2021), Araucariaceae and Callitris species of Australia (respectively, Haines et al. Reference Haines, Olley, English and Hua2018; Pearson et al. Reference Pearson, Hua, Allen and Bowman2011). Complex growth trends, such as biannual formation in Cedrela odorata from Suriname, have also been evaluated using post-1950 14C analysis (Baker et al. Reference Baker, Santos, Gloor and Brienen2017), and 14C dating has overall favored the selection of woody species suitable for tropical tree-ring studies (Giraldo et al. Reference Giraldo, del Valle, González-Caro, David, Taylor, Tobón and Sierra2023; Groenendijk et al. Reference Groenendijk, Sass-Klaassen, Bongers and Zuidema2014; Santos et al. Reference Santos, Granato-Souza, Ancapichún, Oelkers, Haines, De Pol-Holz, Andreu-Hayles, Hua and Barbosa2024; Soliz-Gamboa et al. Reference Soliz-Gamboa, Rozendaal, Ceccantini, Angyalossy, van der Borg and Zuidema2011).
Pre-bomb calibration curves to convert 14C dates to calendar years are also available to users (Reimer et al. Reference Reimer, Austin, Bard, Bayliss, Blackwell, Bronk Ramsey, Butzin, Cheng, Edwards, Friedrich, Grootes, Guilderson, Hajdas, Heaton, Hogg, Hughen, Kromer, Manning, Muscheler, Palmer, Pearson, van der Plicht, Reimer, Richards, Scott, Southon, Turney, Wacker, Adolphi, Büntgen, Capano, Fahrni, Fogtmann-Schulz, Friedrich, Köhler, Kudsk, Miyake, Olsen, Reinig, Sakamoto, Sookdeo and Talamo2020). During pre-bomb periods, dating accuracy has been improved by matching radiocarbon dates to the ‘wiggles’ of the calibration curve when the age difference between the 14C dates is known (Bronk Ramsey et al. Reference Bronk Ramsey, van der Plicht and Weninger2001). However, when only the spatial distance between sequential radiocarbon samples is available, but not the exact time interval, a different approach needs to be used. In this case, “deposition models” (Bronk Ramsey Reference Bronk Ramsey2008) have been developed for radiocarbon calibration of sedimentary sequences and other chronologically ordered sample sequences (Hua et al. Reference Hua, McDonald, Redwood, Drysdale, Lee, Fallon and Hellstrom2012; Piotrowska et al. Reference Piotrowska, Vleeschouwer, Sikorski, Pawlyta, Fagel, Roux and Pazdur2010). Building from previous investigations (Biondi et al. Reference Biondi, Santos, Rodríguez and Sosa2024), and because of the need to clarify the maximum lifespan of dragon trees, we report here on a new set of AMS radiocarbon dates that were calibrated using wiggle-matching and three different models, namely the Poisson-process deposition model in OxCal (Bronk Ramsey Reference Bronk Ramsey2008), the linear regression age-depth model (Blaauw Reference Blaauw2010) and the Bayesian age-depth model (Blaauw and Christen Reference Blaauw and Christen2011).
Methods
Sample collection
The sampled dragon tree, which was already described by Biondi et al. (Reference Biondi, Santos, Rodríguez and Sosa2024), is located in Tenerife, Canary Islands, Spain (Figure S1). Its estimated age, according to local land managers, is ∼330 years; single-sample 14C-AMS dating had suggested an age of ∼350 years for this dragon tree, but with large errors and other, more recent, potential calibration matches. A large (∼60 cm diameter) branch had fallen to the ground and a cross-section was cut in the owner’s backyard on October 18, 2023. After cutting the branch cross-section, a smaller piece was cut in the radial direction, from the outside all the way to the inner circular cavity located inside the stem (Figure S4 in Biondi et al. Reference Biondi, Santos, Rodríguez and Sosa2024). The radial section, about 40 cm long, was then transported to the DendroLab, where it was left to dry. The wood was very moist, as shown by fungi growing inside the stem cavity; the same fungi later developed on the smaller piece when it was left out to dry (Figure S4c in Biondi et al. Reference Biondi, Santos, Rodríguez and Sosa2024).
At the DendroLab, the branch section was surfaced and polished using progressively finer sandpaper, first with a belt sander, then with an orbital sander, and finally by hand, all the way to >600 grits. Wood samples from the branch section were then manually collected using a steel chisel and at regular intervals, about 1 cm from each other, starting from the innermost portion and proceeding towards the outer, more recent growth tissues (Figure 1). Each sample was weighted to make sure it exceeded 30 mg (Table S1) to allow for successful chemical and isotopic analyses.
Photograph of the radial section that was cut from the dragon tree branch (see Figure S4 in Biondi et al. Reference Biondi, Santos, Rodríguez and Sosa2024). A total of 33 samples were collected at ∼1 cm from each other (see ruler at the bottom for scale) and numbered progressively from the innermost to the outermost portion of the section. The last sample (# 33) was approximately 4 cm from the bark.
Radiocarbon dating
The 14C analyses were performed at the Keck Carbon Cycle Accelerator Mass Spectrometry (KCCAMS) Laboratory at the University of California, Irvine (USA). Samples were reduced to particles (∼0.06 × 1 cm) and then individually processed to cellulose following previously published methods (Santos et al. Reference Santos, Komatsu, Renteria, Brandes, Leong, Collado-Fabbri and De Pol-Holz2023). To summarize, samples were extracted to holocellulose by cycling HCl and NaOH at 70°C for 30 min and then bleached with a 1:1 NaClO/1N HCl at 70°C for 4 hr. Afterwards, samples were rinsed with 1N HCl @70°C for 10 min, and brought to pH 6–7 with warm MilliQ water. Once the samples were combusted alongside of reference materials (i.e., samples of known age and blanks), fast and complete graphitization was performed by the Zn-method (Santos and Xu Reference Santos and Xu2017; Xu et al. Reference Xu, Trumbore, Zheng, Southon, McDuffee, Luttgen and Liu2007) followed by isotopic analysis. Even though efficiency of holocellulose extraction led to relatively low yields (Table S1), filamentous graphite targets produced in this study were ≥ 0.6 mg C.
14C-AMS measurement was performed on an in-house modified National Electrostatics Corporation (NEC 0.5MV 1.5SDH-21 model) spectrometer. This instrument is equipped with 13C and 12C measurements, allowing for online isotopic-fractionation corrections, and consequently produces high-precision 14C results (Beverly et al. Reference Beverly, Beaumont, Tauz, Ormsby, von Reden, Santos and Southon2010). A total of 7 Oxalic Acid I (OX-I or NIST HOxI SRM 4990B) as combustible graphite targets were used as the primary standard, followed by Oxalic acid II (OXII, NIST SRM 4990C) as secondary. For chemical blank and secondary standards, we made use of the AVR ancient wood, FIRI-H sub-fossil wood and FIRI-J modern barley. Radiocarbon results were analyzed following procedures described in Santos et al. (Reference Santos, Southon, Griffin, Beaupre and Druffel2007) and later were expressed as fraction modern carbon (F14C) and/or 14C age BP followed by uncertainties (Stuiver and Polach Reference Stuiver and Polach1977) before calibration to calendar years took place. The individual statistical error bars of the 14C data were calculated based on counting statistics and propagated uncertainties due to normalization and processing of blanks. Reproducibility was better than 0.3% using reference materials (as combustibles and/or holocellulose). Samples labeled “Modern” contained an excess of 14C from mid-20th century atmospheric thermonuclear weapons tests.
A simplified wiggle-match model of the pre-bomb radiocarbon ages (Barclay et al. Reference Barclay, Witter and Haeussler2023, Reference Barclay, Haeussler and Witter2024) against the IntCal20 curve (Reimer et al. Reference Reimer, Austin, Bard, Bayliss, Blackwell, Bronk Ramsey, Butzin, Cheng, Edwards, Friedrich, Grootes, Guilderson, Hajdas, Heaton, Hogg, Hughen, Kromer, Manning, Muscheler, Palmer, Pearson, van der Plicht, Reimer, Richards, Scott, Southon, Turney, Wacker, Adolphi, Büntgen, Capano, Fahrni, Fogtmann-Schulz, Friedrich, Köhler, Kudsk, Miyake, Olsen, Reinig, Sakamoto, Sookdeo and Talamo2020) was initially used. Since we had no information on time intervals between the sequential samples, we further analyzed the 14C data by means of deposition models. First, we applied the Poisson-process model in the OxCal software, which allows for fluctuations in the deposition rate according to the depth of samples along the sequence (Bronk Ramsey Reference Bronk Ramsey2009a; b). Then, we implemented two similar approaches commonly used for age-depth modeling of sedimentary records to reconstruct deposition histories. One was the linear regression model in the R package “clam” (Blaauw et al. Reference Blaauw, Christen, Esquivel Vazquez and Goring2025) and the other was the Bayesian model of the R package “rbacon” (Blaauw et al. Reference Blaauw, Christen, Aquino Lopez, Esquivel Vazquez, Gonzalez, Belding, Theiler, Gough and Karney2024). Age-depth (deposition) models were run with all 33 samples; pre-bomb dates were calibrated using the IntCal20 curve (Reimer et al. Reference Reimer, Austin, Bard, Bayliss, Blackwell, Bronk Ramsey, Butzin, Cheng, Edwards, Friedrich, Grootes, Guilderson, Hajdas, Heaton, Hogg, Hughen, Kromer, Manning, Muscheler, Palmer, Pearson, van der Plicht, Reimer, Richards, Scott, Southon, Turney, Wacker, Adolphi, Büntgen, Capano, Fahrni, Fogtmann-Schulz, Friedrich, Köhler, Kudsk, Miyake, Olsen, Reinig, Sakamoto, Sookdeo and Talamo2020) while post-bomb dates were calibrated using the NH zone 2 curve (Hua et al. Reference Hua, Turnbull, Santos, Rakowski, Ancapichún, De Pol-Holz, Hammer, Lehman, Levin, Miller, Palmer and Turney2022) in the CALIBomb program (Reimer and Reimer Reference Reimer and Reimer2025; Reimer et al. Reference Reimer, Brown and Reimer2004). Applying deposition models to plant growth samples provided a narrower probability distribution for dating the innermost sample, and also an estimation of growth rate changes.
Results and discussion
As previously reported for arboreal palms (Tomlinson Reference Tomlinson2006), dragon trees can store large amounts of moisture inside their stem. Based on the moist and dry weight of our branch radial section, water content was ∼50%. Macroscopically the surfaced and polished branch section seemed to show increment growth layers. When enlarged, the apparent banding was caused by varying orientation of the vascular bundles rather than cellular anatomy (Figure 2). These amphivisal secondary vascular bundles, where the xylem surrounds the phloem, are embedded in secondary, parenchymatous ground tissue, forming a peculiar type of ‘‘wood’’ (Jura-Morawiec Reference Jura-Morawiec2015; Tomlinson and Zimmermann Reference Tomlinson and Zimmermann1969).
Progressively enlarged photographs of the dragon tree section, surfaced and polished enough to show a macroscopic structure that vaguely resembles growth layers, but in reality is composed of a relatively uniform parenchyma matrix in which the amphivisal vascular bundles are embedded (Tomlinson and Zimmermann Reference Tomlinson and Zimmermann1969).
Of the 14C-AMS-dated samples, the last 11 (23–33) were initially categorized as “modern” (Table S1), i.e. post-1950, following the bomb-spike pattern in 14C concentrations that has been recognized worldwide (Hua et al. Reference Hua, Turnbull, Santos, Rakowski, Ancapichún, De Pol-Holz, Hammer, Lehman, Levin, Miller, Palmer and Turney2022). The Poisson-process deposition model in OxCal indicated consistency of 14C dates and identified a reduced probability distribution for 18 pre-bomb dates included in the model (Figure 3). The innermost sample resulted in a calibrated age (2σ) between 1776 and 1798 CE. This result is consistent with the simplified method of Barclay et al. (Reference Barclay, Witter and Haeussler2023), which provided an estimated innermost sample year of 1787 CE (Figure S2).
Output from the Poisson-process deposition model in OxCal (Bronk Ramsey Reference Bronk Ramsey2024) with Amodel = 36.78 and Aoverall = 30.73. Dates were assigned to a relative depth expressed as centimeters from the outer surface of the branch radial section and then calibrated using the IntCal20 curve (Reimer et al. Reference Reimer, Austin, Bard, Bayliss, Blackwell, Bronk Ramsey, Butzin, Cheng, Edwards, Friedrich, Grootes, Guilderson, Hajdas, Heaton, Hogg, Hughen, Kromer, Manning, Muscheler, Palmer, Pearson, van der Plicht, Reimer, Richards, Scott, Southon, Turney, Wacker, Adolphi, Büntgen, Capano, Fahrni, Fogtmann-Schulz, Friedrich, Köhler, Kudsk, Miyake, Olsen, Reinig, Sakamoto, Sookdeo and Talamo2020). Post-bomb dates were calibrated using the NH zone 2 curve (Hua et al. Reference Hua, Turnbull, Santos, Rakowski, Ancapichún, De Pol-Holz, Hammer, Lehman, Levin, Miller, Palmer and Turney2022) in the CALIBomb software, as shown in Figure S2, and then included in this model as calendar ages (C_dates).
Based on the CALIBomb results (Figure S2), sample 24 corresponds to ∼year 1962 (next to the peak in 14C) and sample 33 corresponds to ∼year 1994. Since the outside of the branch shows signs of partial erosion and possibly insect galleries, it is difficult to estimate when it stopped growing. As we have about 9 cm between samples 24 and 33, covering a period of about 33 years, we can roughly estimate that the radial growth rate in the outer portion of this branch was approximately 3–4 yr cm–1, while the preliminary wiggle-matching results suggested up to 6-12 yr cm–1 growth rates in the interior part. The linear regression model results produced by package “clam” suggest an average growth rate of 5.8 yr cm–1 (Figure 4). However, both the OxCal and the “rbacon” model indicate changes in growth rates occurred after ∼1900 CE (Figure 4).
Comparison between three different deposition models used for radiocarbon calibration. Pre-bomb dates were calibrated using the IntCal20 curve (Reimer et al. Reference Reimer, Austin, Bard, Bayliss, Blackwell, Bronk Ramsey, Butzin, Cheng, Edwards, Friedrich, Grootes, Guilderson, Hajdas, Heaton, Hogg, Hughen, Kromer, Manning, Muscheler, Palmer, Pearson, van der Plicht, Reimer, Richards, Scott, Southon, Turney, Wacker, Adolphi, Büntgen, Capano, Fahrni, Fogtmann-Schulz, Friedrich, Köhler, Kudsk, Miyake, Olsen, Reinig, Sakamoto, Sookdeo and Talamo2020); post-bomb dates were calibrated using the NH zone 2 curve (Hua et al. Reference Hua, Turnbull, Santos, Rakowski, Ancapichún, De Pol-Holz, Hammer, Lehman, Levin, Miller, Palmer and Turney2022).
Considering that this branch section does not include its central part, we can estimate that the sampled branch was starting to form ∼250 years ago. Given that such branch was already at a height of ∼3 m from the ground, our new set of 14C dating analyses suggests that this dragon tree was ∼300 years old in 2023, thereby confirming and further refining the preliminary results obtained by Biondi et al. (Reference Biondi, Santos, Rodríguez and Sosa2024). It should be mentioned that, thanks to a long sequence of 14C dates, we reached this conclusion by applying to arboreal plants with secondary growth a set of radiocarbon calibration tools usually applied to sedimentary and depositional records. We therefore propose that using our approach can facilitate dating wood samples when no clear growth layers can be identified and/or the time interval between such layers is unknown.
Monumental plants of impressive size are examples of charismatic megaflora (Hall et al. Reference Hall, James and Baird2011), whose presence usually becomes a synonym of sustainable land stewardship, prompting local communities to advertise their location and old ages. Our contribution is aimed at applying the most advanced scientific dating tools to estimate the longevity of arboreal monocotyledons so that it is properly reported in both technical and lay-person documents.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/RDC.2025.10149
Acknowledgments
We are grateful to the local landowners and managers for allowing us to sample the monumental plant described in this article. In addition, we thank Francesco Salomone, José Alberto Delgado, Marcos Díaz Bertrana, Paulino Melchor, Nayra Méndez and Jesús Navarro for their help in collecting samples. Many thanks also to the Cabildo de Tenerife for their logistical support during field work. G.M.S. thanks the valuable help of the undergraduate students from the University of California-Irvine, Lucas Duy Nguyen and June Nakachi Griffin, on sample processing for 14C-AMS dating. F.B. was funded in part by a Fulbright U.S. Senior Scholar award administered by the Commission for Cultural, Educational, and Scientific Exchange between the USA and Spain; the Experiment Station of the College of Agriculture, Biotechnology, and Natural Resources, the Ozmen Institute for Global Studies, the Office of the Provost, and the Faculty Research Travel Grant at the University of Nevada, Reno, USA.
Competing of interest
The authors declare no financial conflicts of interest.
Open access
