Fig. 1. Overall average performance of organic agriculture relative to conventional agriculture (indicated by the red circle). (A) Performance per unit area and (B) performance per unit output.

Average organic performance is indicated by the red circle. (A) Performance per unit area and (B) performance per unit output. Figure includes production (brown petals), environmental (green petals), producer (red petals), and consumer (blue petals) benefits (petals that extend beyond the red circle) and costs (petals inside the red circle). Dimensions assessed include (starting at the top, going clockwise) production, biodiversity, soil quality, water quality, water quantity, climate change mitigation, farmer livelihoods, farmer and farm worker health, farm worker livelihoods, consumer health, and consumer access. Larger petals represent superior organic performance (for example, a larger petal for N loss means lower N loss in organic). In addition, note that per unit output performance is only relevant for environmental variables; other petals are unchanged relative to per unit area performance. Shading of petals represents level of uncertainty for each variable, with uncertainty determined by the number of primary studies included in each assessment and the level of agreement between different quantitative reviews (see fig. S6 for details). Variables that could not be quantified are in gray. Length of gray petals also varies slightly depending on whether the qualitative assessment of each dimension (see Table 2) is uncertain or suggests no difference (that is, petal is on the red circle) or shows higher (that is, petal extends beyond the red circle) or lower (that is, petal is inside the red circle) performance. Means used to quantify each variable (also known as petal length) were calculated as weighted means (weighted by the sample size, typically the number of observations in each quantitative review) across estimates of response ratios (organic/conventional) from different quantitative reviews (see table S1 for sources and figs. S1, S2, and S5 for values used) and are represented on a log scale to treat changes in the numerator and denominator the same [with the red circle indicating no change, that is, log(org/conv) = 0]. Note that this approach does not account for double-counting of primary studies included in multiple quantitative reviews or meta-analyses. This double-counting might affect petal size but would not alter qualitative size relationships among petals.