Lodgepole pine (Pinus contorta var. latifolia Engelm.) ecosystems of central British Columbia face cumulative stresses, and management practices are increasingly scrutinized. We addressed trade-offs between “light-on-the-land” versus more aggressive silvicultural approaches by examining plant communities and indicator species (non-natives, berry producers, epiphytes, mycotrophs, pine rust alternate hosts) across a gradient of five or six site preparation treatments at the Bednesti trial (established 1987). We tested whether more severe site preparation (i) caused plant community composition to diverge from a 35- to 46-year-old reference forest, (ii) accelerated succession by hastening crown closure, or (iii) delayed succession by promoting seral species. Nonmetric multidimensional scaling (NMDS) ordination showed all treatments converging toward the reference forest composition. At 10 years, succession was incrementally delayed by more severe treatments; at 25 years, only burned windrows were still delayed. Mixed-effects models based on site preparation severity were better than crown closure models for 11 of 13 variables tested, suggesting that mostly belowground processes drive succession in these infertile ecosystems. Invasive hawkweeds persisted on all treatments at 25 years. Limited, contradictory data did not support using mechanical or fire treatments to reduce alternate hosts of pine stem rusts. Long-term trials such as Bednesti highlight the need for ecosystem-specific strategies and diverse approaches to accommodate conflicting benefits and risks of disturbance in forests.