To analyze the conversion of smallholder cropland to Eucalyptus woodlots, this study is anchored in agricultural land-use decision theory and random utility theory (McFadden 1974). Under the traditional profit maximization behavior model, a smallholder selects a specific land allocation pattern to maximize expected multi-year net utility subject to household resource constraints (Barnum and Squire 1979). However, smallholder decisions in developing economies are often influenced by market imperfections and production risks rather than financial returns alone (Sadoulet and de Janvry 1995). Random utility theory posits that a farmer chooses Eucalyptus cultivation over traditional annual crops if the perceived total utility of tree planting (U_ip) exceeds the utility of annual cropping (U_ic) (McFadden 1974; Greene 2018). The choice of smallholders to transition toward tree production can be framed as a multi-criteria decision system, visually mapped in Fig. 1.

The study was conducted in Dangila district, Amhara region. Dangila is one of the 12 districts found in the Awi zone (Fig. 2). It is characterized by mid-altitude agroecology with a mixed farming system. The district has 11°16'N, 36°50'E. The total farming land area is 55,213 ha. According to the Dangila Woreda Office of Agriculture (DWOA) report (2023), the current land use pattern includes 78% cultivated land, 15.84% pastureland, 2.95% forestland, and the rest for other uses. The mean annual rainfall is 1200-1400 mm, with Belg and Meher cropping seasons. Its altitude ranges from 1980 to 2150 m above sea level, allowing a favorable opportunity for wider crop and better livestock production. Most of the farmland is allocated for annual crops, whereas cereals cover 11,572.5 ha. Pulses cover 22,667 ha. Oilseeds cover 2,761 ha; root crops cover 1,034.29 ha, and vegetables cover 3,733 ha. The major crops include maize, Teff, finger millet, and pulse crops, in order of area coverage (DWOA 2023).

This study utilizes a retrospective, mixed-methods cross-sectional research design. While analyzing a 15-year historical production horizon (2009–2024) through a cross-sectional approach introduces potential recall bias, this design was necessary because longitudinal panel records are absent at the local smallholder level in the study area. To ensure data reliability and minimize recall errors over the 15-year window, several data triangulation and quality control protocols were implemented (Deaton 2019).

First, physical farm inputs (seedlings, fertilizer, labor days) and harvest outputs were documented using local structural landmarks and critical timeline benchmarks (e.g., major national droughts, elections, and family milestones) to anchor the participants’ memories. Second, self-reported data were systematically cross-verified against official archives kept by the Dangila Woreda Office of Agriculture (DWOA), agricultural extension registries, and historical crop yield estimates. Third, non-market inputs, such as traditional mutual-aid exchange labor (Debo), were converted into monetary terms based on corresponding historical local market wage records.

A multi-stage sampling design involving purposive selection, simple random sampling, systematic random sampling, and stratification was used to select the study units. In the first stage, out of the 29 administrative kebeles (villages) in the district, 12 kebeles were purposively selected in consultation with agricultural extension experts. The primary selection criterion was the widespread presence of smallholder Eucalyptus woodlot cultivation with a history exceeding the 15-year target window of this financial analysis. In the second stage, four kebeles, namely Dubi, Kuancha, Abadra, and Gumdri, were randomly selected from the 12 targeted kebeles. The target population within the twelve kebeles was (N = 15,319 households). The total sample size was determined using Cochran’s (1963) formula, given the variability in incidence and participation in planting the Eucalyptus woodlot. N = (N z^2 PQ)/(e^2 (N – 1) + z^2 PQ), where N = the total target households, n = the required sample size, e = the precision level (±5%), the confidence level is 95%, P = the proportion of Eucalyptus planters (70%), and Q = 1 – P (the portion of non-planters, 30%). By substituting the values into the formula, it becomes n = (15319 × 1.96^2 × 0.7 × 0.3)/ (0.05^2 (15319 – 1) + 1.96^2 × 0.7 × 0.3) = 316.

Households within the four randomly selected kebeles were stratified into two distinct subgroups using official local agricultural registries: Eucalyptus planters and non-planters. Finally, from each stratum, proportional sample units were selected from each kebele using a systematic random sampling technique. The sample sizes for the study kebeles were 60, 80, 76, and 100 for Dubi, Kuancha, Abadra, and Gumdri, respectively (Table 1).

Primary data were gathered using a mixed-methods approach comprising structured household surveys, focus group discussions (FGDs), and key informant interviews (KIIs). To capture accurate cost and revenue streams over the 15-year production lifetime (2009–2024), structured questionnaires were translated into Amharic and administered by trained enumerators. Survey plots were tracked by registration and ownership duration. Farmers recorded physical input volumes (kilograms of fertilizer, seed volumes, seedling counts, and labor days for clearing, digging, planting, weeding, thinning, guarding, and harvesting) and physical outputs (crop yields in quintals and timber harvest volumes by log size).

Qualitative data were systematically collected to contextualize community land-use choice dynamics and triangulate input-output estimates. Eight FGDs were conducted across the four kebeles (2 FGDs per kebele). To avoid power dynamics and social friction, planters and non-planters were interviewed in separate sessions. Each group comprised 8 to 12 participants and maintained balanced gender representation (at least 35% female participants per session) to reflect diverse household financial perspectives. Additionally, eight KIIs were conducted with development agents, kebele administrators, local agricultural experts, and veteran community leaders. The KIIs focused on community land-use shifts, regional supply shocks, and historical input price changes over the 15 years.

Utilizing nominal historical values or unadjusted “current market prices” would distort the comparison due to the compounding effects of domestic inflation (Sinden and Adelman 2024). Because this study tracks a 15-year historical production horizon (2009–2024), it would distort the comparison due to the compounding effects of domestic inflation. To ensure financial validity, all historical costs and revenues were reconstructed using an asset-valuation approach based on real, inflation-adjusted values (Campbell and Brown 2003). This adjustment isolates real purchasing power, establishing a stable financial baseline before discounting cash flows. Information gathered from KIIs with local merchants and timber brokers was integrated into this model to build consistent price tracks for non-marketed inputs (such as exchange labor) and intermediate wood assets (such as poles and charcoal wood). To ensure financial validity, all historical costs and revenues were reconstructed using an asset-valuation approach based on real, inflation-adjusted values.

Quantitative survey data were analyzed using descriptive and inferential statistics, as well as discounted cash flow models. At the same time, qualitative transcripts from FGDs and KIIs were coded using thematic analysis to explain the economic drivers of the observed trends. Continuous socioeconomic variables were tested using independent-samples t-tests, and categorical variables were examined using Pearson’s chi-square (X^2) tests. Trend analysis was employed as a design research methodology to evaluate trends in Eucalyptus plantations in the study area. A trend is a pattern observed in a series of events or data points, indicating a general tendency or direction of change over time. Trend analysis involves collecting and examining data to identify and interpret these patterns, often to support forecasting or strategic decision-making (KV and Kavya 2016; Postma and Papp 2021).

To model the economic profitability of competing land-use types over time, a complete 15-year time series of real cash flows was constructed for each crop and woodlot system. This structure accounts for the multi-year cycle of Eucalyptus, in which major costs are front-loaded during land preparation and planting in Year 0, followed by minor management costs, and large revenues are realized only during clear-cutting harvests at the end of the rotation cycles (typically Year 5, Year 10, and Year 15). Conversely, crop cash flows follow a recurring annual pattern of input investment and harvesting.

Given the inherent uncertainties in long-term agricultural and forestry investments, a sensitivity analysis was incorporated into the financial model to test the robustness of the baseline Net Present Value (NPV) and Benefit-Cost Ratio (BCR) calculations (Boardman et al. 218). While the baseline evaluation relied on a 7% discount rate and constant 2024 prices, macroeconomic instability, market saturation, and agricultural input price volatility necessitate stress-testing these indicators to ensure reliable policy recommendations.

The financial viability of both Eucalyptus woodlots and crop production was re-evaluated under two scenarios:

Discount Rate Adjustments: The baseline 7% discount rate (r) was elevated to 12% and 18%. This upward adjustment simulates scenarios of heightened domestic inflation, an increased opportunity cost of capital, and the high subjective risk premiums that resource-constrained smallholders typically apply to delayed cash flows.

Market Price Shocks: A compound, worst-case market shock scenario was modeled by directly adjusting the cash flow variables. This involved a simultaneous 10% reduction in Eucalyptus gross revenues (B_t for woodlots multiplied by 0.9), simulating regional oversupply and depressed timber demand due to market saturation. Concurrently, a 10% escalation in crop production costs (C_t for crops multiplied by 1.1) was applied to simulate sudden surges in the costs of imported chemical fertilizer and hired labor.

The study area was predominantly male-headed (87.0%, p = 0.002), with female-headed households accounting for only 13% of the total sample (Table 2). Access to formal credit was identified as a critical institutional determinant for Eucalyptus woodlot cultivation (X^2 = 37.54, p < 0.01). 56.1% of planters used credit to finance substantial “Year 0” establishment costs, including seedlings, fencing, and land preparation. Because Eucalyptus enterprises yield no financial returns until the initial harvest (typically around Year 6), access to credit helps relax immediate liquidity constraints, allowing households to bridge the multi-year income gap (Dercon and Christiaensen 2011). This role of credit as a structural catalyst aligns with broader research in Ethiopia, highlighting its importance in buffering households against long gestation periods (Alemayehu and Melka 2022). Conversely, credit market failures often compel asset-poor households to prioritize immediate subsistence crops over more profitable long-term investments due to high subjective discount rates (Holden and Shiferaw 2004).

Furthermore, agricultural extension services significantly distinguished planters from non-planters (χ2 = 14.18, p < 0.01), with 76.0% of planters maintaining frequent contact with development agents, compared with 45.3% of non-planters reporting infrequent or absent contact. This institutional support facilitates the transfer of technical proficiency in silvicultural management, thereby minimizing seedling mortality and enhancing the perceived economic returns of commercial woodlots (Tesfaw et al. 2022).

The independent-samples t-test reveals a statistically significant age disparity between smallholder strata (t = 3.19, p < 0.01) (Table 3), where advanced age typically correlates with diminished physical capacity and internal labor deficits exacerbated by youth out-migration (Alemayehu and Melka 2022). This demographic challenge is compounded by a dependency ratio paradox in which planter households maintain larger family sizes (6.45 ± 1.35) but significantly lower active labor supply (3.12 ± 0.84 adult equivalents) than non-planters, creating high consumption pressure alongside a shortage of labor for intensive seasonal tasks such as plowing and weeding. Consequently, labor-constrained households strategically substitute intensive food crops with low-maintenance Eucalyptus woodlots to align with internal labor supplies and establish long-term biological assets against economic volatility (Holden and Shiferaw 2004), a transition further facilitated by higher formal education, which enhances managerial capacity and the ability to navigate timber value chains (Asfaw and Admassie 2004). Household asset configurations and spatial characteristics further delineate these resource boundaries: planters generally possess larger landholdings (1.84 ± 0.62 ha) but significantly lower livestock wealth (3.26 ± 1.42 TLU) than non-planters.

While households with minimal land (< 1.5 ha) must prioritize subsistence crops (Tesfaw et al. 2021) and non-planters leverage extensive livestock for liquid capital and oxen-driven draft power to sustain crop profitability (Storck et al. 1991), livestock-poor households utilize woodlots as an alternative asset-building mechanism. This allocation is reinforced by land fragmentation (4.38 ± 1.14 plots) and market isolation (5.82 ± 2.41 km), as managing scattered, remote parcels increases travel costs and management friction (Blarel et al. 1992). To mitigate these constraints, smallholders plant hardy, low-input trees on distant plots and utilize farm-gate sales to wholesale traders, effectively neutralizing the transit costs associated with geographic isolation.

Households in the study area rely on a diverse mix of income sources, with the most common being the sale of crops and crop by-products, livestock and livestock by-products, eucalyptus, daily wage labor, petty trade, and remittances. Among these, eucalyptus sales constitute the largest share (31%) of total annual income for planters, households engaged in eucalyptus cultivation, underscoring the central role of eucalyptus in rural livelihoods. This result aligns with Tesfaw et al. (2021), who confirm that eucalyptus provides the greatest portion of planters’ annual cash income. Crop sales and livestock sales account for 22% and 20% of annual household cash income, respectively, while the remaining 23% is generated through livestock by-products, petty trade, crop by-products, daily wage labor, and other sources (Fig. 3).

The primary cash income source for non-planters was the sale of the crop, which accounts for 37% of households’ total annual income (Fig. 4). This is due to the presence of a larger area of cropland and surplus crop products. This is the major income source for rural households. In addition, animals, which account for 35% of annual income, are deliberately reared as a cash income source. Other income sources, such as daily labor, petty trade, animal by-products, and land rent, contribute 7%, 6%, 4%, and 2%, respectively.

According to the Amhara region’s agroforestry practice package, the minimum woodlot size is 0.125 ha. The study adopted this threshold as the minimum woodlot size. The maximum woodlot size in the study area was 1.5 ha. According to key informants, the district comprises 29 kebeles, of which 12 kebeles account for 56% of the district’s area and are suited for Eucalyptus woodlots. Currently, 70% of households in these 12 kebeles have Eucalyptus woodlots. The average household landholding was 1.6 ha. Of this, 67% (1.1 ha) was allocated to crop production, 25% (0.4 ha) to Eucalyptus, and the remaining 8% (0.14 ha) to other land uses (Fig. 5).

Over the past fifteen years (2009–2024), the annual increase in new Eucalyptus woodlots grew at an accelerating rate for ten consecutive years (2009–2021). However, starting in 2022, the trend reversed (Fig. 6). This supports the group discussants’ statements that “due to the rise in crop product prices and the fall in Eucalyptus log prices caused by the northern Ethiopia conflict, the annual growth of new Eucalyptus woodlots has been decreasing.” Overall, the trend in new Eucalyptus woodlot plantations shows an increase in area coverage (Fig. 5). This aligns with previous studies reporting that Eucalyptus is expanding over time (Mekonnen 2010; Molla and Addisie 2023; Kerbo et al. 2024; Zeleke et al. 2024).

Remote sensing and multi-temporal image analysis indicate that a significant increase in Eucalyptus plantation coverage from 2000 to 2020, followed by a decline in the rate after 2020, primarily due to reduced market demand for Eucalyptus logs and rising crop prices (Zeleke et al. 2024). On average, over the past fifteen years, 25% of cropland in the study area was converted to Eucalyptus woodlots, corresponding to an annual conversion rate of 1.6%. This finding aligns with earlier research indicating the rapid expansion of Eucalyptus on farmlands in Mecha district and other regions of Ethiopia (Tefera et al. 2016; Tesfaw et al. 2023).

Conversion of croplands to Eucalyptus woodlots is ongoing in the study area (Fig. 7). According to Zerga et al. (2021), land-use change from crop production to monoculture Eucalyptus woodlots has continued over the past 30 years. Several factors drive this trend: Eucalyptus offers high financial returns, has strong coppicing ability, is less palatable to animals, requires low investment and management costs, and is highly valued for fuel and construction materials (Tefera et al. 2016; Molla and Addisie 2023; Tesfaw et al. 2023).