Four lesser-known shrub species, including Lagerstroemia speciosa (Local name: Bungur), Ganophyllum falcatum (Locong), Myrsine avenis (Latte), and Mischocarpus sundaicus (Terasa) were selected as biomass feedstocks. The plant materials were taxonomically identified by botanist Nasri, and voucher specimens (Voucher No. WAN0005748/ L. speciosa, No. WAN0021952/ G. falcatum, No. WAN0017752/ M. avenis, and No. WAN0008828/ M. sundaicus) have been deposited in the Herbarium Wanariset, located in Kutai Kartanegara Regency, East Kalimantan Province, Indonesia (−0.997047, 116.976286).

Branches were collected from the Educational Forest Area of Hasanuddin University, Maros Regency, South Sulawesi, Indonesia (−5.005145, 119.767224). Sample preparation through to pretreatment was conducted at the Laboratory of Forest Products Utilization and Processing, Faculty of Forestry, Hasanuddin University, Makassar, Indonesia (−5.130221, 119.485481). Subsequent property testing was performed at the Environmental and Forestry Laboratory, Faculty of Forestry, Hasanuddin University, Makassar, Indonesia (−5.130277, 119.484054), and the gross calorific value analysis was carried out at the Chemical Service Laboratory of the Center for Poultry and Livestock Assembly and Testing, Bogor, Indonesia (−6.663145, 106.857261).

Collected shrub branches including the bark were cleaned and sun-dried to air-dry condition (moisture content <12%). The material was chopped using a flaker machine. It was then comminuted with a hammer mill to produce smaller particle, and sieved using a wire mesh to obtain particles retained between 20 and 40 mesh (0.41–0.82 mm). Only this fraction was used in subsequent experiments.

The pretreatment comprised a chemical extraction (soaking) followed by a physical extraction (torrefaction). Chemical extraction was designed to remove volatile and other low-energy constituents prior to thermal conditioning. Torrefaction then provided a uniform thermal step for all samples. A process flow is provided in the study’s method schematic ( Figure 1).

For each run, 100 g (oven-dry basis) of shrub particles were placed in an Erlenmeyer flask and contacted with solvent at a 1:7 (wt/wt) solid-to-liquid ratio. Four solvent conditions were tested, including distilled water, ethanol, hexane, and a sequential ethanol + hexane treatment. Extractions were conducted at 25 °C for 6 h (room temperature) and at 50 °C for 2 h (elevated temperature kept below solvent boiling points). To limit evaporation, especially at 50 °C, flasks were covered with aluminum foil. Suspensions were manually stirred every hour for approximately 30 s to promote mass transfer. For the sequential ethanol + hexane condition, soaking was performed consecutively in ethanol then hexane, with 3 h + 3 h at 25 °C and 1 h + 1 h at 50 °C. After extraction, slurries were vacuum-filtered and the retained solids were washed repeatedly with 80 °C distilled water until no solvent odor remained. This practical criterion was used to confirm solvent removal. The solids were then dried in a Memmert UN55 oven at 100 °C for 24 h.

After the chemical extraction, the dried-extracted samples were wrapped in aluminum foil and torrefied at 200 °C for 10 min in an oven. This temperature was aligned with that used in previous studies, ³ while the duration was shortened to improve the efficiency without significantly reducing the calorific value. All samples received the same torrefaction setting, without additional controlled variables in this step.

After pretreatment, the following properties were determined, including moisture content (MC), weight loss (WL) (separately after chemical extraction and after torrefaction), total weight loss, yield, volatile matter (VM), and gross calorific value. MC, WL, yield, and VM were assessed at the Environmental and Forestry Laboratory, Faculty of Forestry, Hasanuddin University (Makassar), while gross calorific value was measured at the Chemical Service Laboratory of the Center for Poultry and Livestock Assembly and Testing (Bogor). Summary data are reported for each shrub species under each solvent and temperature condition ( Table 1– 4 and Figure 2– 8).

Mass-based indicators were computed from recorded sample masses before extraction, after extraction, and after torrefaction. In keeping with standard mass-balance relationships reflected in the dataset structure. WL after chemical extraction ( % ) = m 0 − m 1 m 0 x 100 WL after torrefaction ( % ) = m 1 − m 2 m 1 x 100 Total WL ( % ) = m 0 − m 2 m 0 x 100 Yield ( % ) = 100 − Total WL

where m 0 is the oven-dry mass before extraction, m 1 the mass after extraction/drying, and m 2 the mass after torrefaction. The manuscript presents means with standard deviations for MC and VM, and tabulates WL and yield per condition, enabling direct cross-checks.

Dataset validation

It was noted several scope limitations, including: (i) extraction conditions were restricted to two temperatures and relatively short durations, which may not reflect broader operational envelopes; (ii) although VM was reduced for the most shrub species, values remain relatively high and do not yet meet SNI 8675:2018 quality thresholds for M. avenis; (iii) for each species–solvent–temperature combination, the extraction was performed once on a single sample (no technical or biological replicates). This lack of replication limits statistical power, precludes robust uncertainty quantification beyond the reported descriptive values, and warrants caution when generalizing treatment effects; and (iv) experiments were conducted under controlled laboratory settings, so scale-up and field variability were not represented. Additionally, the torrefaction step was fixed at a single set-point (200 °C, 10 min), and solvent removal was verified pragmatically by odor rather than by residual-solvent analytics. These choices simplify the protocol but leave limited room to quantify instrument-level uncertainty or solvent residues. Users should consider these constraints when extrapolating the dataset to industrial contexts.

For secondary analyses or replication, users should preserve the documented particle size, solid–liquid ratio, solvent sequence and durations (for ethanol + hexane), temperature controls, and the drying/torrefaction set-points to maintain comparability with the published tables and figures. Deviations in any of these parameters will directly affect WL, Yield, VM, and calorific value outcomes. The DOI-hosted files provide the necessary inputs to re-calculate all derived metrics and to propagate uncertainty using the reported SDs.