The CENTURY soil organic matter model was developed to describe nutrie
nt dynamics for soils of the North American Great Plains, but is now a
pplied to other soil types and climates. Phosphorus transformations in
CENTURY are based on empirical relationships, developed for soils whe
re N is the most limiting nutrient. This restricts the model's applica
bility in P-limited soils, particularly those that differ in reactivit
y to P and in organic matter stabilization. Moreover, nutrient ratios
in plant litter from nutrient-poor tropical soils may be much wider (e
.g., C:P ratios up to 3000) than those commonly found in litter from t
emperate zones. Equations used in the model to partition incoming litt
er into slowly and rapidly cycling organic matter may not be suited fo
r such litter. A problem in validating the model is achieving congruen
ce between nutrient fractions that can be measured in the soil and the
conceptual pools (state variables) that represent such fractions in t
he model. Work on P in tropical soils indicates that parameterizing CE
NTURY may require the use of different P fractions than those of the o
riginal version. State variables thus will change. The lack of experim
ental data on P fractionation in relation to P availability in highly
weathered P-deficient soils is a major restriction in formulating an a
dequate theoretical framework on P dynamics in tropical soils. The CEN
TURY model therefore has limited applicability to highly weathered low
-P soils. Various improvements in the P submodel of CENTURY are recomm
ended.