Fig 1.
Distribution of study sites in southern Wisconsin.
1—Belcamper; 2—Bush Clover; 3—Ipswich; 4—Lone Rock; 5—Drachenburg; 6—Monroe; 7—Brock; 8, 9—Rettenmund I and II; 10—Muralt; 11—Oliver; 12—Bong; 13—Westport; 14—Snapper; 15—Faville; 16, 17—Young I and II.
Fig 2.
Significant increases in CI with (A) δ13C, (B) SMI, and (C) soil % sand; significant decreases in CI with (D) CEC • soil depth, (E) soil % clay, and (F) penetrometer soil depth.
Lines represent linear regressions.
Fig 3.
Across sites, average leaf height increased significantly with (A) LAI, and decreased significantly with (B) Curtis’ CI (100 = wettest, 500 = driest).
Lines represent linear regressions.
Fig 4.
Principal components analysis on 11 soil variables; each arrow represents the eigenvector corresponding to an individual variable, surrounded by a circle of radius 1 to indicate maximum possible correlation of each variable to axes of compositional variation.
Axis 1 reflects primarily CEC • soil depth and soil % clay, with soil % sand and SMI having a negative effect; axis 2 reflected mainly [Mg] and [K]. Axis accounted for 44.5% of the overall variance; axis 2, 22.2%. Total variance = 11.0 by definition.
Fig 5.
Bray-Curtis ordination of 17 sites based on presence/absence data for 137 informative species.
Vectors show the strength and direction of correlations of individual environmental factors, and PCA axis 1 and 3 scores, with scores on (A) BC axes 1 and 2, and (B) and BC axes 1 and 3. Sum of squares of non-redundant distances in the original matrix = 51.084.
Fig 6.
Relationships of pairwise Sørensen distances in species composition to pairwise differences in (A) CI, (B) δ13C, (C) CEC • soil depth, and (D) SMI.
Lines represent linear regressions; significance levels are based on Mantel tests.
Fig 7.
Path analyses relating effects of soil moisture, fertility, and mechanical impedance on composition as measured by CI.
(A) Effects of % sand content and soil depth on CI via soil moisture supply (δ13C) and nutrient supply (CEC • soil depth). Numbers represent partial regression coefficients on standardized variables; the magnitude of those coefficients is proportional to arrow width, with negative effects indicated by hollow arrows. Based on the path coefficients shown, 95.9% of the effect of soil depth on CI flows through CEC • depth vs. δ13C; 87.2% of the effect of % sand on CI flows through CEC • depth. (B) Effects of SMI and soil depth on CI via moisture supply and nutrient supply. Based on the path coefficients shown, 79.2% of the effect of SMI on CI flows through nutrient supply; the magnitude of the effect of soil depth flowing through nutrient supply is 13.1 times that flowing through moisture supply and is opposite in sign to the former. (C) Effects of CEC • soil depth, SMI, and δ13C on CI.