Oak Savanna Restoration

Background

• Oak savannas once covered ~10% of Minnesota, but about 99% of historic midwestern savanna has been lost
• Remaining patches are degraded by woody encroachment after fire and grazing were removed
• Yantes et al. (2023) surveyed 242 plots at Sherburne NWR with 4 different treatments:

Control, Thin, Thin + Burn, Thin + Burn + Graze

Methods

• Dissimilarity: Bray-Curtis, since the data are species counts. It compares relative composition, ignores joint absences, and needs no pre-scaling.
• Shrubs and herbs analyzed separately, since their values are on very different scales (max 268 vs. 17)
• NMDS: smallest k with stress near or below 0.2
  • Shrubs: k = 2, stress = 0.183 (stress-plot correlation = 0.89)
  • Herbs: k = 3, stress = 0.201 (stress-plot correlation = 0.82)
• Species arrows: envfit(), arrows scaled by √R²; shrub arrows shown for R² > 0.3.

Shrubs separate by treatment; herbs barely do

Treatment explains more of the shrub variation

• Shrubby species are most responsible for differentiating between the different plots.

Shrub species driving the differences

How the top shrub species differ by treatment

Conclusions

1. Treatments differ: managed plots have different shrub communities than Control, and treatment explains about 27% of the variation in shrub composition.

2. Shrubs drive the differences: treatment explains more than twice as much variation in shrubs (27%) as in herbs (12.5%), and herb communities look very similar across treatments.

3. Standout Species:

   • Hazelnut: increases with each added treatment, highest under grazing

   • Blackberry/Raspberry: peaks with thinning, drops with burning and grazing

     • Their arrows are nearly perpendicular, so they respond independently