Leaf cuvette fluxes were obtained with a LI-6800 portable photosynthesis system (LI-COR, Lincoln, NE, USA) fitted with a 6800-01A fluorometer head (aperture 6 cm², air flow 500 µmol s⁻¹). We imposed standard conditions for emission factors (Es): leaf temperature 30 °C, PAR 1000 µmol m⁻² s⁻¹, CO₂ 400 µmol mol⁻¹, and 55 % RH. The LI-6800 maintains closed loop control and continuously records actual leaf temperature and PAR. Because measurements were made at the standard reference conditions, cuvette fluxes equal Es after correction to leaf dry mass. After 5 min stabilization of photosynthesis, 200 mL min⁻¹ of outlet air was diverted for 2 min through Tenax TA/Carbograph 5TD adsorbent cartridges (Markes International Ltd, Bridgend, UK) using a dual-channel pump (ZC-QL, Zhejiang Hengda Instrumentation Ltd., Zhejiang, China). Detailed leaf handling, leaf area and dry mass determination are provided in Section S1 and are also described in Zeng et al. (2024, 2025a, b).

A cylindrical PMMA chamber (∅ 25 cm × 28 cm, 13.7 L) internally coated with FEP film was used for branch-level measurements. The charcoal- and KI-scrubbed ambient air (9 L min⁻¹) was supplied by a mass-flow controller (Alicat Scientific, Inc., Tucson, AZ, USA) coupled with an oil-free pump (MPU2134-N920-2.08; KNF, Freiburg, Germany), then it was well mixed with PTFE-bladed fans in the chamber (Zeng et al., 2022a, 2025c). For comparability with leaf cuvette measurements, sunlit branches (3–5 m above ground) with ∼ 6–10 leaves were selected, avoiding mutual overlap so that foliage formed a single layer. Branches showing visible self-shading were not sampled. These practices follow our goal of minimizing light heterogeneity within the chamber. When the selected sunlit branches were enclosed; fluxes were allowed to stabilize for 1–2 h before sampling. Outlet air (and inlet blanks) was drawn at 200 mL min⁻¹ for 10 min by an automatic sampler (JEC921; Jectec Science and Technology, Co., Ltd, Beijing, China) onto the same adsorbent tubes as above. The chromatograms of a representative BVOC sample and its corresponding blank sample of inlet air during the field study was shown in Fig. S5. Concurrent meteorological and radiometric variables were logged continuously (Rotronic HC2A-S RH/T probes; LI-1500 PAR sensor; OMEGA/RKC thermocouples). Specifically, leaf temperature was measured by thermocouples attached on the leaf, while PAR was measured by a LI-1500 PAR sensor placed on the top of the chamber, these parameters were used to calculate the standard emission factors (Es). More details about the branch sampling are provided in Section S2 and Zeng et al. (2022a).

Dynamic branch chamber measurements revealed 12 BVOC species in 2-month-old seedlings (isoprene, 8 monoterpenes, 3 sesquiterpenes; Fig. S8) versus 17 species in 2-year-old trees (isoprene, 11 monoterpenes, 5 sesquiterpenes; Table S1). Thus, chemical richness increased by ∼ 40 % with age, consistent with the view that metabolic complexity develops as trees mature (Satake et al., 2024).

Isoprene dominated the flux from both age classes but its standardized Es declined significantly (p<0.001) from 107.7 ± 34.9 µg g⁻¹ h⁻¹ in seedlings to 69.8 ± 21.2 µg g⁻¹ h⁻¹ in two-year-old trees. When normalized by leaf area, however, the two groups were indistinguishable (Fig. S9a), indicating that the mass-based contrast mainly reflects an increase in leaf mass per area (LMA) with tree age (Fig. S10). Seedlings therefore allocated a larger fraction of assimilated carbon to isoprene (Fig. S9c) despite exhibiting lower net photosynthesis (Pn, Fig. S9b), suggesting a typical growth-defense trade-off of early ontogeny. Consistent with our findings, isoprene Es for the 1-year-old E. globulus was 5-fold higher than that for the 7-year-old individual (Street et al., 1997). Winters et al. (2009) also documented lower isoprene Es in four 10-year-old eucalyptus species compared to their seedlings measured by He et al. (2000).

Total monoterpene emissions rose nearly 6-fold with tree age (1.09 vs. 6.14 μg g⁻¹ h⁻¹), but the increase was almost entirely due to acyclic β-ocimenes, whose Es leapt from 0.133 ± 0.06 to 5.33 ± 4.61 µg g⁻¹ h⁻¹ (Fig. 2 and Table S1). By contrast, Es for cyclic α-pinene, limonene, and 1,8-cineole declined by 30 %–65 % (Fig. 2). Consequently, terpene composition shifted from 75 % cyclic (seedlings) to 85 % acyclic (2-year-old trees) (Fig. 2). Sesquiterpenes doubled in absolute terms but remained a minor (< 1 %) proportion of total BVOCs.

During the branch experiments, both seedlings and saplings were exposed to high growth temperature conditions, which could lead to significant stress on plant physiology as discussed by Zeng et al. (2025d). This was evidenced by the substantial emissions of typical stress-induced β-ocimene from saplings. However, although seedlings were also subjected to heat stress, they did not exhibit similarly high β-ocimene emissions but dominated by α-pinene and 1,8-cineole, closely resembling those under non-stress saplings measured during the warm dry season (Fig. S11). This indicates that stress-induced gene expression may require time to accumulate, and that seedlings grown in greenhouses, having not experienced environmental stresses, may not adequately represent the emission characteristics of trees in natural field conditions.

Such a cyclic-to-acyclic transition has been observed in other Eucalyputus species and might be attributed to age-dependent expression of terpene synthesis and to selectively pressures from biotic/abiotic stress in the field (Monson et al., 2021; Pollastri et al., 2021). For example, most previous studies reported that cyclic α-pinene and 1,8-cineole were the dominant MTs in stressless lab-grown seedlings (Evans et al., 1982; Guenther et al., 1991; He et al., 2000; Tsui et al., 2009; Malik et al., 2019), whereas acyclic MTs like β-ocimenes were generally low or even undetectable. In contrast, most stress-rich field-grown trees could emit large amounts of acyclic β-ocimenes (Street et al., 1997; Nunes and Pio, 2001; Sørensen et al., 2020; Purser et al., 2020, 2021; Nagalingam et al., 2023).

Both isoprene and β-ocimenes are known to play key roles in plant antioxidant defense. Isoprene helps scavenge reactive oxygen species (ROS) (Jardine et al., 2014), maintain membrane stability (Sharkey and Singsaas, 1995; Pollastri et al., 2019, 2021), and regulate antioxidative processes (Zuo et al., 2019; Monson et al., 2021), whereas the highly reactive β-ocimene may act as an even more efficient, direct antioxidant against ROS. The highly reactive β-ocimenes can quench ROS more efficiently than isoprene (Pollastri et al., 2021), providing a plausible advantage for field-grown trees exposed to stronger light, heat, drought, and herbivory.

Environmental factors, particularly growth temperature, light availability, and soil moisture/nutrients, can modulate BVOC emissions (Monson et al., 1994; Harley et al., 1994, 1996, 1997; Fall and Wildermuth, 1998; Funk et al., 2006; Guenther et al., 2006, 2012; Yuan et al., 2020). In our study, both campaigns were conducted under clear-sky, peak-summer conditions with comparable ambient temperature and high midday PAR (Figs. S1–3). A simple sensitivity using the MEGAN/Guenther acclimation scheme (Guenther et al., 2012) indicates that, a ∼ 10 % change in growth PAR would change standardized Es by ∼ 8 %, and a +1.1 °C change in growth temperature by ∼ 11 %. These effects are small relative to the age-related differences reported here, such as ∼ 6-fold increase in total monoterpenes and a > 40-fold rise in β ocimenes from seedlings to saplings (Fig. 2 and Table S1). Nevertheless, because we did not measure soil nutrient or moisture status, parallel experiments under controlled and co-located growth conditions are warranted to fully disentangle ontogeny from environment.

It should be noted that the seedlings used here were cultivated trees, where soil properties could be parallel with those in previous seedling-based studies. In contrast, the taller trees measured under field conditions reflect realistic and natural growth conditions. The substantial differences in Es and emission composition between seedlings and saplings underscore that seedling measurements are inappropriate as generic surrogates for natural forest emissions. Despite this, all age contrasts presented here are seedlings vs. saplings; extrapolation to mature trees requires additional in-situ measurements.

Notably, we counted the most complete Es database currently used by the MEGAN model (Wiedinmyer et al., 2004; Guenther et al., 2012), showing that one-third of the publications in this database were based on greenhouse seedlings. Our results show that these data tend to overestimate canopy-scale isoprene fluxes of subtropical eucalyptus plantation and under-represent atmospheric reactivity by excluding large β-ocimene emissions from these tree canopies. Incorporating age-resolved Es and prioritizing statistically robust and in-situ sampling of branches from adult trees, particularly in rapidly expanding subtropical plantations, will reduce inventory uncertainty and refine predictions of ozone and secondary organic aerosol formation.