This paper argues that turbulence near the top of a vegetation canopy is similar to that in a plane mixing layer due to instabilities associated with a strong inflection in the mean velocity profile. Mixing-layer turbulence, formed between two streams of different velocities, differs from surface-layer turbulence. These differences explain many observed features of canopy turbulence, including Reynolds stress ratios, eddy diffusivity ratios, roles of ejections and sweeps, turbulent energy balance, and turbulent length scales. The paper predicts that these length scales are controlled by the shear length scale $ L_s = U(h)/U'(h) $, with the streamwise spacing of dominant canopy eddies being $ \Lambda_x = mL_s $, where m=8.1. These predictions are tested against field and wind-tunnel data. The paper proposes that canopy turbulence is modulated by larger-scale, inactive turbulence, which is quasi-horizontal on the canopy scale. The study challenges the traditional view that canopy turbulence is a perturbed version of boundary-layer turbulence, instead suggesting that the inflection in the canopy velocity profile leads to instability processes similar to those in a mixing layer, determining the coherent eddy structure. The paper focuses on near-neutral flow over a uniform vegetation canopy, excluding buoyancy effects. While buoyancy effects are significant in some conditions, the mixing-layer analogy suggests that the turbulent eddy structure near the top of the canopy has similar qualitative behavior across a wide range of buoyancy conditions.This paper argues that turbulence near the top of a vegetation canopy is similar to that in a plane mixing layer due to instabilities associated with a strong inflection in the mean velocity profile. Mixing-layer turbulence, formed between two streams of different velocities, differs from surface-layer turbulence. These differences explain many observed features of canopy turbulence, including Reynolds stress ratios, eddy diffusivity ratios, roles of ejections and sweeps, turbulent energy balance, and turbulent length scales. The paper predicts that these length scales are controlled by the shear length scale $ L_s = U(h)/U'(h) $, with the streamwise spacing of dominant canopy eddies being $ \Lambda_x = mL_s $, where m=8.1. These predictions are tested against field and wind-tunnel data. The paper proposes that canopy turbulence is modulated by larger-scale, inactive turbulence, which is quasi-horizontal on the canopy scale. The study challenges the traditional view that canopy turbulence is a perturbed version of boundary-layer turbulence, instead suggesting that the inflection in the canopy velocity profile leads to instability processes similar to those in a mixing layer, determining the coherent eddy structure. The paper focuses on near-neutral flow over a uniform vegetation canopy, excluding buoyancy effects. While buoyancy effects are significant in some conditions, the mixing-layer analogy suggests that the turbulent eddy structure near the top of the canopy has similar qualitative behavior across a wide range of buoyancy conditions.