Page 54 - Read online

Page 10 of 23 Diaz-Vallejo et al. Carbon Footprints 2026, 5, 12

Table 3. Analysis of variance testing difference of SOC among soil orders, suborders, and USDA texture classification across a
database of 586 pedons in Puerto Rico

Factors Df Sum of squares Mean of squares F value P-value
Agricultural lands
Soil order 6 14.970 2.495 15.720 < 0.0001
Residuals 78 12.380 0.159
Soil suborder 12 15.840 1.320 8.260 < 0.0001
Residuals 72 11.510 0.160
USDA texture classification 7 7.241 1.034 3.962 0.001
Residuals 77 20.104 0.261
Pasture lands
Soil order 7 5.690 0.812 2.575 0.016
Residuals 152 47.950 0.315
Soil suborder 18 15.660 0.870 3.230 < 0.0001
Residuals 141 37.980 0.269
USDA texture classification 9 19.080 2.120 9.202 <0.0001
Residuals 150 34.550 0.230
Forest lands Soil order 6 11.060 1.844 3.386 0.003
Residuals 289 157.370 0.545
Soil suborder 17 82.930 4.878 15.860 < 0.0001
Residuals 278 85.510 0.308
USDA texture classification 10 38.320 3.832 8.364 < 0.0001
Residuals 282 129.200 0.458

USDA: United States Department of Agriculture; SOC: soil organic carbon.

Soil Health Gap Benchmark: We calculated the land use gap as mean SOC in forests minus SOC in
agriculture or pastures [Figure 9], and within each soil order and climate class [Figure 10]. Forests on
Alfisols, Inceptisols, Mollisols, Oxisols, and Ultisols had greater SOC than managed lands, whereas on
Aridisols and Vertisols, forests had reduced SOC. Soil order significantly influenced the magnitude of the
gap between forests and managed land uses (P = 0.012; Table 6). The gap was larger for agriculture than
pastures (P = 0.068; Figure 9). Incorporating climatic variables sharpened these contrasts underscoring
climate’s modulation of the gap direction [Figure 10].

Together, these results highlight contrasting strengths and limitations of the two benchmark approaches. The
Scores Benchmark captured continuous SOC variability across soil orders and climatic gradients with
moderate predictive skill, supporting its use for regional-scale comparisons. In contrast, the Soil Health Gap
Benchmark revealed large variability in SOC differences between forests and managed lands that depended
strongly on soil order and climate class, indicating greater sensitivity to reference conditions and
environmental context.

DISCUSSION
This study provides a regional-scale assessment of SOC variability across a diversity of tropical soil
environments in Puerto Rico. By combining 586 pedons spanning nine soil orders, multiple climates, and
contrasting land uses, we evaluated environmental and edaphic controls on SOC and examined two
benchmark approaches, the Soil Health Gap and the Scores Benchmark, to assess soil health. Together, these
results highlight the complexity of soil carbon dynamics in tropical agroecosystems and the need for flexible