Introduction of cover cropping systems may be important to a stable food supply in sub‐Saharan Africa. We examined the effects of three cropping systems in a 2‐yr, four growing season study in Togo: continuous maize (Zea mays L.), maize–mucuna [Mucuna pruriens (L.) D.C.], and maize–pigeon pea (Cajanus cajan L.). Mucuna and pigeon pea were grown in 1‐ or 2‐yr cycles, and three N and two P fertilizer rates were factorially applied on maize. Use of mucuna and pigeon pea after maize in the first year reduced N and P fertilizer needs in the subsequent year. Cover crops increased maize grain yield by 37.5 and 32.1%, respectively, in the second year. Two‐year cumulative economic returns on maize production were optimal when cover crops were grown every other year (every fourth season), compared to continuous maize or annual cover crops. The April 2002 to December 2003 soil N budgets showed a gain of N (>400 kg ha−1) under all cropping systems. Initial soil nitrate (NO3)–N was reduced by 57.8% under the continuous maize system, but increased by 39 and 3.6% under the mucuna‐based and pigeon pea–based systems, respectively. Low (<20 kg ha−1) N losses occurred during the fallow period. Phosphorus losses occurred for all periods, but a mucuna‐based cropping system has potential for soil P replenishment. The relay of a mucuna cover crop into maize in one out of 2 yr was most economical and improved soil N and P status without commercial fertilizers.

Mathematical models may be applied to simulate nitrogen (N) dynamics under different types of soil and environmental conditions to assess fertilizer N needs or to predict nitrate-N (NO 3 -N) potential impact on water quality. The research version of LEACHMN was evaluated using data from lysimeters and field experiments conducted at the University of Lome´Research Farm in Togo, West Africa. The model was calibrated for the mineralization, nitrification, denitrification and volatilization rate constants with measured values of NO 3 -N leaching losses and maize (Zea mays L.) N uptake collected from the lysimeter experiment. The model was then tested against measured data of soil profile NO 3 -N distribution and maize N uptake from the field experiment and drainage water collected from the lysimeter experiment. The testing procedure involved two scenarios with increasing level of generalization for transformation rate constants (i) rate constant values for each N treatment and (ii) rate constant values averaged over N treatments. LEACHMN effectively simulated drainage water volume and rate (r 2 = 0.94 to 0.99). During the calibration efforts, the model satisfactorily simulated NO 3 -N leaching losses (r 2 = 0.98) and accurately simulated growing season cumulative maize N uptake. The variation of the calibrated rate constants among N treatments was primarily linked to the model's incapacity to accurately simulate maize N uptake throughout the growing period. When tested using calibrated rate constants for each treatment, the model was successful in simulating soil profile NO 3 -N distribution (r 2 = 0.52 to 0.94). Simulations of soil profile NO 3 -N distribution were not satisfactory (r 2 = 0.03 to 0.49) when rate constants were averaged over N treatments. Improvement of the plant N uptake routine of the model is needed to increase the model's performance. Using the LEACHMN model to predict N dynamics on the Ferralsols of southern Togo appears feasible when appropriate calibrations are performed.