Agroforestry Transition Hub: Education and On-farm Research to Advance Agroforestry for Climate Resilience for Northeast Farmers

Project 1

Trees were pervasive elements of the working-farm landscapes we studied—present in woodlots, hedgerows, riparian buffers, windbreaks, and around barns, lanes, and other infrastructure. Farmers described actively managing these existing trees for multiple functions, including utility (shade/shelter, drainage and erosion control, boundary definition), aesthetics, timber/woodlot goals, and occasional income diversification. Across interviews, farmers consistently articulated potential benefits of adding trees for farm viability and climate resilience. However, new tree planting remained limited relative to the level of stated interest. This pattern held across farm types and management approaches, suggesting that low planting rates were not explained by biophysical constraints alone.

Farmers’ assessments of whether tree planting felt feasible clustered around differences in farm history and position. Multigenerational farmers commonly evaluated tree planting through instincts of risk, shaped by repeated cycles of economic and climatic volatility, prior program experiences, and hard-won operational efficiencies. In practice, this group favored changes that had clear returns, low added management burden, and a demonstrated fit with their system. Newer farmers, by contrast, often described simultaneously building financial stability and social legitimacy; they were attentive to how tree practices—and the language used to describe them—might bolster or undermine credibility with peers, landowners, lenders, and support providers.

Across farmer groups, enabling conditions converged around four linked themes: (1) Farmers emphasized that support was most usable when it was responsive to on-farm timelines, grounded in local conditions, and delivered through sustained relationships rather than one-off site visits. (2) Farmers frequently described the early years of tree planting as a period of concentrated costs (materials, protection, time) before benefits accrued, making the practice feel difficult to justify within tight margins. (3) Farmers described labor bottlenecks—especially during busy seasons—as a primary constraint. They were concerned that they lacked the skills for tasks like watering, protecting, pruning, and replanting after losses, nor would they want to prioritize them during a busy season. (4) Farmers reported that when support programs or technical providers approached tree planting with fixed sociotechnical ideals (e.g., the “right” practice, scale, or metrics), they often encountered misfit: long waits, bureaucratic friction, or rigid requirements that reduced trust and dampened participation.

Discussion

Across working farms, interest in tree planting was most consistently driven by immediate operational pressures—heat and extreme weather, wind and infrastructure damage, water management and erosion, livestock comfort, and the need to stabilize production amid increasing variability (Lane et al., 2019). Farmers discussed trees less as an abstract environmental intervention than as working infrastructure, evaluated through the same daily calculus as any other investment: risk, labor, and cash flow. This orientation helped explain why enthusiasm coexisted with low adoption. Farmers were not disputing that trees can help; they were weighing whether a practice with front-loaded costs and management demands could compete with projects that protect near-term viability. In line with other scholarship, tree planting was constrained by early expenses and labor, delayed and uncertain returns, and the difficulty of monetizing many benefits (Feder and Umali, 1993; Kassie et al., 2012; Liu et al., 2018; Waldman et al., 2020). 

These barriers were intensified by land tenure and time horizons: leasing, short agreements, and uncertain succession shortened planning horizons and elevated the risk of long-duration investments, making trees harder to justify when farmers could not count on controlling land long enough to realize benefits (Mutabazi et al., 2015; Piya et al., 2013; Teklewold et al., 2019; Brooks, 2022; Murken and Gornott, 2022). In some cases, trees were framed as competing with the need to secure or expand annual productive acreage—an equity issue insofar as constraints tied to land access are unevenly distributed and can be obscured by “voluntary adoption” narratives. Farm histories further shaped what counted as feasible and credible. Multigenerational farmers often drew on inherited heuristics and lessons from past volatility and programs, favoring low-burden changes with proven returns, while newer farmers described legitimacy-building as part of farm work itself, making them especially attentive to how trees—and the language around them—might signal stewardship and innovation or invite scrutiny if framed as impractical. Importantly, many constraints were produced at the farm–program interface: when support systems embedded narrow ideals of the “right” practice—fixed scales, standardized metrics, rigid timelines—farmers described misfit, waiting, and bureaucracy that eroded trust and reduced participation, whereas relationship-based technical assistance that was timely, local, and iterative reduced uncertainty and allowed tree practices to be adapted to on-farm realities.

When new planting felt infeasible—especially for livestock farmers facing immediate needs for shade and shelter—farmers reported leaning more heavily on existing wooded areas. This reliance sometimes came at the expense of patch condition, suggesting that adoption constraints could redirect pressure onto existing tree resources rather than remove it. This dynamic reframed the adoption problem so scaling agroforestry was not only about adding trees, but also about recognizing, protecting, and stewarding the woody resources farms already had.

A working-lands lens, therefore, underscored that many farms already held substantial woody resources—woodlots, riparian strips, hedgerows, and regenerating edges—that contributed to resilience but were often rendered invisible in incentive structures that privileged new planting. In some cases, establishing “new” tree systems may involve clearing or simplifying existing woody patches to meet program templates, risking ecological loss and social backlash (Cook-Patton et al., 2021; Ollinaho and Kroger, 2020). These tensions pointed to a basic gap in how tree-based interventions were framed: before we could meaningfully evaluate what additional planting could (or should) accomplish, we needed to understand the condition, composition, and functional role of the woody communities farms already had. This premise motivated our second project, which shifted the focus from whether farms could add trees to what those existing woodlots and tree patches actually looked like—their structure and regeneration, species composition (including nonnative presence), signs of stress or degradation, and how their location within the farm mosaic shaped both management pressures and resilience potential

Project 2

Results and Discussion

We found that forest patches were a common feature regardless of production system: farms in diversified, livestock, and vegetable operations had broadly similar “portfolios” of patch types (e.g., continuous woods, hedgerows, riparian strips; Fig. 1; χ² = 12.114, df = 10, p = 0.2775). This convergence was consistent with the broader history of the region, where secondary forests are recovering across a landscape shaped by repeated clearing, intensive use, and reforestation, often yielding a recognizable “second-growth template” across many ownerships and land uses (Foster, 2004). In that sense, it was not surprising that production type alone was a weak predictor of understory composition: we saw only modest separation by production type (Fig. 2; ANOSIM R = 0.072, p = 0.0355), suggesting that shared regional legacies and recurring management pressures could drive a degree of ecological homogenization across farm forests, even as farms differed in their enterprises.

Figures for Grant Report

Where farms diverged more clearly was not in whether patches existed, but in what those patches contained and how they were organized at the patch scale. When we grouped communities by patch type, separation was clearer and the effect size larger (Fig. 3; ANOSIM R = 0.2255, p = 0.0011), indicating that “what kind of patch is this?” (riparian vs. hedgerow vs. continuous woods vs. swamp, etc.) mattered more than “what kind of farm is this?” Practically, this suggested that while farm forests could share a common second-growth baseline—shaped by similar histories and, in many cases, ongoing intensive edge maintenance or disturbance—the ecological differences that mattered most for management were organized by patch context and microenvironment. For adoption and implementation, the implication was straightforward: farms already contained distinct ecological zones, and protecting, managing, or restoring wooded areas was likely to be most effective when tailored to patch type and condition, rather than applied as a uniform wooded-area prescription.

The structural results reinforced that patch-specificity. The proportional mix of vegetation structure classes differed strongly across both production and patch types (Fig. 4; χ² = 64.104, df = 16, p = 1.05e–07), indicating that patches varied not just in species lists but in vegetation architecture—an important lever for habitat, regeneration potential, and resilience. Structural diversity was positively associated with total understory abundance (Fig. 4; df = 49, p = 6.49e–07), even though it did not necessarily translate into higher understory diversity. Finally, patch size showed a pattern that benefited from being interpreted through the lens of scale and heterogeneity rather than as a bumper-sticker conclusion: larger patches were associated with lower understory richness in our plots (Fig. 4; df = 45, p = 3.13e–05). Classic species–area logic often predicts more species with more area at the patch or landscape scale, but plot-level richness can be higher in smaller, edge-dominated patches because edges increase light and microhabitat variation (and thus local richness) while also increasing exposure to disturbance and invasion risk. Work explicitly disentangling edge vs. interior species–area patterns supports this scale-dependent interpretation, in which edge processes can inflate local richness even as interior dynamics differ.

Taken together, these two projects pointed to the same design problem from different angles: working lands were heterogeneous mosaics, and both ecological dynamics and support systems failed when they assumed uniformity. Lesk and colleagues made this point in a different context—showing how “uniform” perturbations can miss the temporal complexity and spatial heterogeneity that shape on-the-ground outcomes —which offered a useful parallel for why one-size-fits-all tree programs often misfit farm realities. The combined implication was a broader “protect/manage/restore” approach (Cook-Patton et al., 2021) that treated existing woody patches as valued infrastructure (and managed them with patch-type specificity), while strategically supporting new establishment where it added function without displacing existing benefits or forcing farms to contort themselves to program templates.