Soil samples obtained from arable fields at the Crop Research Institute (Kumasi) in the semi-deciduous agro-ecological zone of Ghana were used in this study. The soils were air-dried, sieved (<2 mm) and packed to a bulk density of 1.23 g cm⁻³ in 500 mL kilner jars. The sandy clay loam soil contained 59.1% sand, 13.5% silt and 27.4% clay. Other properties included pH (H₂O) of 5.8, total N (0.2%) and organic C (1.2%). The soil is classified as Orthic Ferric Acrisol according to the FAO 1998 system. The soil was pre-incubated at 45% WFPS and 25 °C for 7 days prior to addition of the residues to re-initiate microbial activity after 1 year of cold storage (−4 °C), and to minimize changes in soil water filled pore space (WFPS) at the start of the experiment. WFPS was calculated based on soil volumetric water content, bulk density and a particle size of 2.3 g cm⁻³.

Maize and cowpea residues were chosen because of their varying N content and C:N ratios and the widespread use of these species as intercrop or multi-crop species in African farming systems. The cowpea and maize were grown in vermiculite in the greenhouse at the University of Aberdeen. The cowpea and maize plants were both harvested after 7 weeks (just before tasseling of maize and when adequate biomass could be obtained for the cowpea treatment). Subsamples of the maize and cowpea leaves were dried at 45 °C for the determination of dry matter (%), and then milled (<1 mm) for further laboratory analyses. Total C and N contents were determined on a 0.5 g dry subsample through dry combustion using a Metler-Toledo AG 2455 C/N auto analyser (Sercon Ltd., Cheshire, UK). Lignin content was determined in an Ankom 220 fibre analyser (Acid detergent fibre) and the total extractable polyphenol was measured using Folin-ciolcateau reagent in a method adapted from [27] (Table 1).

All data were analyzed using the Minitab 15 statistical package. All data were tested for normality and homogeneity and log-transformed where necessary. ANOVA was carried out using the measured values of N₂O, mineral N (NH₄⁺ and NO₃⁻) and WEC. Significant difference was tested using pooled standard error of the difference (SED) and Tukey’s HSD at 0.05 significance level. Pearson correlation analyses were carried out to establish relationships if any, between crop residue C:N ratio and total N₂O and CO₂ emissions, and between WEC and mineral N concentrations, and CO₂ and N₂O fluxes.

The daily N₂O emissions from all the treatments were similar on first sampling date (day 0) (Figure 1). On the second sampling date, the N₂O emission was highest (5.26 mg N m⁻² d⁻¹) in the 75:25 cowpea:maize treatment and was significantly different (P < 0.05) from the other treatments. The N₂O emissions from the 100:0 and 25:75 treatments were similar but significantly different (P < 0.05) from the remaining treatments while the 50:50 treatment was not significantly different from the 0:100 and the control. On the third sampling date, the N₂O emissions in the 0:100 and the control were significantly lower (P < 0.05) than all other treatments. N₂O emissions from the 0:100 treatment and the control were similar and lowest on all sampling dates. However, N₂O emissions from all the treatments returned to background value (represented by the control) by the sixth sampling date (day 7).

Cumulative N₂O emitted from the 75:25 cowpea:maize treatment was substantially greater than emissions from all the other residue mixture treatments (Figure 2). Cumulative N₂O emission from the 100:0 was higher than cumulative emissions from the remaining treatments. However, cumulative emissions from the 0:100 and control were similar. The cumulative N₂O emissions from the 50:50 and 25:75 treatments do not appear to differ substantially but are higher than those of the sole maize treatment and the control. Overall, the pattern of cumulative N₂O emissions (Figure 2) appears consistent with the total N input from the residue mixtures, which decreased from 136 kg N ha⁻¹ for the 100:0 treatment to 36.8 kg N ha⁻¹ for the 0:100 treatment (Table 2). The emission factor of the sole cowpea treatment was greater than the sole maize treatment. However, the cowpea:maize mixtures substantially increased the emission factors compared to the sole treatments, with the highest observed in the 75:25 treatment and the lowest in the 50:50 treatment (Table 2).

The CO₂ emissions from the treatments were not different on the first sampling date (day 0) (Figure 3). On the second sampling date (day 1), the 75:25 cowpea:maize treatment had the highest CO₂ emission which was significantly different (P < 0.05) from the other treatments. The 100:0, 50:50 and the 25:75 were similar but differed (P < 0.05) from the 0:100 treatment and the control. On the third sampling date (day 2), the CO₂ emission from the 75:25 treatment was different (P < 0.05) from the other treatments, while the 100:0 and the 25:75 were similar but different (P < 0.05) from the 0:100 and the control. On the fourth sampling date (day 3), the CO₂ emission from the 75:25 was still highest (P < 0.05) than all other treatments. The 100:0 and 25:75 were also similar but different (P < 0.05) from the remaining treatments. Daily CO₂ fluxes peaked on day 1 in the 100:0, 75:25, 50:50 and 25:75 treatments. CO₂ emissions from the 50:50 treatment returned to background value on day 10, but emissions from the 100:0, 75:25 and 25:75 treatments returned to background levels on day 30. Throughout the 30 days CO₂ fluxes measured from both the 0:100 cowpea:maize treatment and the control were low and not substantially different from each other.

Water extractable carbon (WEC) concentrations measured in all the treatments decreased from first to final sampling dates (Figure 4). The highest (P < 0.05) WEC concentration of 254.53 mg C kg⁻¹ soil was measured in the 0:100 cowpea:maize treatment on first sampling date (day 0). Apart from the 0:100 treatment, the other treatments did not differ substantially. WEC concentrations were not significantly correlated with NH₄⁺ and NO₃⁻ concentrations, but were positively correlated with N₂O emission (Table 3). The highest correlations were observed for the 25:75 and the 50:50 treatments. However, the correlation was not significant for the 0:100 treatment. CO₂ and N₂O emissions were also positively correlated. The strongest correlation was found in the 75:25 treatment, followed by the 25:75 treatment.

On the first sampling date (day 0), the 75:25 and 0:100 treatments had similar NH₄⁺ concentrations and were the lowest (P < 0.05) than all other treatments (Figure 5). The NH₄⁺ concentrations of the control and the 50:50 treatments were similar and greater than all other treatments. On the second sampling date, the NH₄⁺ concentration of the 50:50 treatment was highest and different (P < 0.05) from the other treatments except the 100:0 treatment. The NH₄⁺ concentrations of all the treatments were greater than those of the 0:100 treatment and the control. The NH₄⁺ concentration of the 100:0 treatment was highest (P < 0.05) on the third sampling date but declined on subsequent sampling dates. NH₄⁺ concentration in the 75:25 treatment, however, increased gradually between successive sampling dates from day 0 to day 30. NH₄⁺ concentration of the 50:50 treatment also decreased gradually from day zero to day 14 (5th sampling date) but recorded the highest (P < 0.05) NH₄⁺ concentration of 12.57 NH₄⁺ kg⁻¹ soil on day 30. NH₄⁺ concentrations of the remaining treatments decreased gradually in successive sampling dates relative to their initial NH₄⁺ concentrations.

On the day of incubation (day 0), apart from the 100:0 cowpea:maize treatment, all the treatments showed similar NO₃⁻ concentrations (Figure 6). On the second sampling date, the NO₃⁻ concentrations in the 100:0 and 75:25 treatments were significantly higher (P < 0.05) but the other treatments did not differ substantially. On the third sampling date, NO₃⁻ concentrations in the 100:0, 75:25 and the 50:50 did not differ substantially but were higher than the remaining treatments. The control and the 25:75 did not differ substantially but were also higher than the 0:100 treatment. On the fourth sampling date, the NO₃⁻ concentrations in the 100:0 and 75:25 were significantly higher (P < 0.05) than all other treatments. On subsequent sampling dates, however, the NO₃⁻ concentration in the 100:0 was significantly higher than the other treatments while the 75:25 and the 50:50 treatments were higher than the remaining treatments. The NO₃⁻ concentration in the 0:100 treatment did not vary substantially throughout the study period. However, the 100:0 cowpea:maize treatment showed gradual increase successive sampling dates but the increase was steepest from the third to the final sampling dates.

On day 0, N mineralization occurred only in the 100:0 and 50:50 cowpea:maize treatments (Figure 7), but between days 1 and 14, net mineralization occurred in all the treatments except in the 0:100 cowpea:maize which showed net N immobilization up to day 14. At the end of the experimental period, the 100:0 treatment recorded the highest (P < 0.05) net N mineralization. Net N mineralization for the 100:0 and 75:25 treatments did not appear to differ substantially on the third, fourth and fifth sampling dates (days 3, 7 and 14) as is also the case for the 75:25 and 50:50 treatments on days 1, 3 and 30. However, the difference between the 50:50 and 25:75 appears substantial.

Net N mineralized on day 30 is consistent with the total available N (Table 4). The total available N and net N mineralized for the 100:0 cowpea:maize treatment were significantly different from all other treatments. Net N mineralized in the 75:25 and the 50:50 treatments were not significantly different but both differed from the remaining treatments. However, the 25:75, 0:100 treatments and control did not differ significantly in their total available N and net N mineralized.

The N₂O emission from the sole cowpea (100:0 cowpea:maize) treatment was significantly higher (P < 0.05) than that of the sole maize (0:100 cowpea:maize) treatment, suggesting that in the sole residue amended treatments residue N availability was the main driving factor for N₂O production. This observation agrees with several authors, e.g., [17,23,31,32], who have also recorded enhanced N₂O emission from soils amended with crop residues and have attributed it to residue N availability. Previous studies [16,33] have found that in addition to their percentage residue N content and C:N ratios, N₂O emissions from soils amended with crop residue have also been influenced by their lignin and polyphenol contents, and lignin:N ratio, polyphenol:N ratio or (lignin + polyphenol):N ratio. [16,31] have also recommended that, for the purposes of controlling N release and subsequent N₂O emission, high N residues with high polyphenol content may be more useful. Lignin and polyphenol contents are known to delay N release from the residues, either by forming recalcitratnt N compounds or by binding to soil microbial enzymes, thereby lowering the substrate availability for N₂O production [34].

In this study the lignin and extractable polyphenol contents of the cowpea and maize residues used were lower than the 15% lignin and 3%–4% polyphenol threshold levels, respectively proposed in the Organic Resources Database [35] and by other authors [36,37] to retard residue N mineralization. Therefore, in this experiment, only the initial N contents and the C:N ratios of the residues were expected to affect N₂O emission and mineral N concentrations. Incorporation of sole cowpea resulted in net N mineralization but sole addition of maize residues led to a net N immobilization and low N₂O emission throughout the 30 days. This observation confirmed the initial hypothesis that incorporation of low C:N ratio cowpea residues will promote rapid net mineralization while the addition of high C:N ratio maize residue will result in initial N immobilization [38,39]. Immediate net mineralization is known to occur often after incorporation of residues with C:N ratios below approximately 20–25 [24,40]. The initial N concentration (3.4%) in the cowpea residue used in this study was higher than the threshold of 1.8%–2.5% N suggested by [35], but that of the maize residue (0.9%) was less than this threshold.

N₂O emissions measured from the mixed cowpea and maize treatments, except for the 75:25 cowpea:maize treatment, were lower than from the sole cowpea treatment, indicating an effect of the maize residue on N₂O emission and perhaps the lower N input from the residues. The N₂O and CO₂ emissions (Figure 1 and Figure 3 respectively) suggest that disproportionate mixing of cowpea:maize residues on weight basis might lead to relatively higher N₂O and CO₂ emissions shortly after the incorporation. This is also reflected in the relatively higher emission factors of the cowpea:maize mixture treatments compared to the sole cowpea or maize treatments (Table 2). While the N₂O emission pattern might reflect a decreasing N input from the residues, the substantially higher N₂O and CO₂ emissions from the 75:25 treatment is interesting and requires further investigation to explain the interactive effects. The same can be said of the 50:50 treatment which has N input lower than the 75:25 treatment but higher than that of the 25:75 treatment, yet has an emission factor lower than both treatments. Thus, the pattern of N₂O emission from the treatments does not reflect just a reduction in N input from the mixtures but an interactive effect from the imbalance between the N and C sources [40,41]. This observation, therefore, seems to suggest that incorporating sole cowpea residue in a tropical acrisol might increase N₂O emission over the control and sole maize treatments, but mixing cowpea and maize residues might potentially lead to relatively higher N₂O and CO₂ emissions in relation to the magnitude of N input from the residues.

Greater C availability from residues is reported to drive dissimilatory reduction of NO₃⁻ [41,42]. In this study, WEC significantly correlated with N₂O emissions, indicative that whilst microbial activity was increased, exemplified by increased CO₂ emissions, denitrification also potentially increased. This is further confirmed by the strongly positive correlation found between CO₂ and N₂O fluxes measured from the different treatments (Table 3).

NO₃⁻ concentration was highest in the 100:0 cowpea:maize treatment than all other treatments. This can be attributed to the higher residue N input and the absence of high C:N maize residue. The addition of high C:N maize residue probably modified the total residue-N input and C:N ratio in the cowpea:maize mixtures depending on the proportions of residues in the mixture. However, the concentration of NO₃⁻ in the 50:50 treatment appeared to differ substantially from the other cowpea:maize treatments except the 75:25 treatment on days 3, 14 and 30. The pattern of NH₄⁺ concentration seems to reflect the pattern of NO₃⁻ concentration and thus suggest the occurrence of nitrification. Thus, reducing the quantity of cowpea residue and increasing the proportion of low N maize residues could have lowered the total N substrate available for mineralization. Again, the apparent lower inorganic N concentration in the cowpea:maize mixtures could have further been exacerbated by the large supply of available C from the high C:N maize residues coupled with an insufficient N supply. This is particularly reflected by the patterns of immobilization and low mineralization in the 25:75 treatment. Unlike the other treatments which showed steep changes in net N mineralization, the 50:50 treatment showed a steady, slow increase in net N mineralization and had no immobilization. The results agree with Gartner and Cardon [43] who reported that decomposition patterns in leaf mixtures are not always easily predictable. They further argued that the characteristics of decomposition in such litter mixes might deviate from responses predicted from decomposition of single species alone.

N₂O emission in the 100:0 treatment was highest on day 2 but declined sharply from day 3 (Figure 1) but NH₄⁺ concentration peaked on day 3 and decreased steeply thereafter (Figure 5). NO₃⁻ concentration increased significantly after day 3 and continued to increase till day 30 (Figure 6). The same pattern can be observed in the 75:25 treatment. These patterns seem to suggest rapid nitrification of the organic N from the cowpea residue. Denitrification requires a C source as electron donor. The high N₂O emission from the 75:25 treatment might suggest accelerated denitrification due to the pulse of C substrate from the 25% maize residue, even though with decreased N input from the cowpea residue. This might also explain the rapid mineralization found in the 75:25 (Figure 7) and why even though the NH₄⁺ concentration in the 75:25 was higher than the 100:0 from day 7 to day 30, this did not correspond with higher NO₃⁻. Thus, the relatively lower N₂O emission in the 100:0 treatment (which had higher N input) with corresponding high mineral N concentrations and net N mineralization might be attributed to inadequate C source. Notwithstanding, it is noteworthy that, in this study, mixing maize with cowpea increases the proportion of added N that is lost as N₂O compared to adding cowpea alone (Table 2).

However, the peak N₂O emission in the 50:50 treatment was lower than the 100:0, 75:25 and 25:75 treatments (Figure 1) and the same can be said of CO₂ emission. However, the initial NH₄⁺ concentration (days 0 and 1) were relatively higher and the changes were slight except between days 14 and 30. With NO₃⁻ concentration, the 50:50 treatment did not appear to differ substantially from the 75:25 on days 0, 3 and 30. This reflects the stable or slow increase in net N mineralization and emission of N₂O albeit with a further low N input (compared to the 100:0 and 75:25 treatments) but with comparatively higher C substrate. The cumulative N₂O emission and N mineralization observed thus confirm the assertion that low N₂O emission will results from insufficient supply of N substrate in the presence of a large supply of C substrate which stimulate microbial immobilization [44]. Nevertheless, there was no immobilization in the 50:50 treatment compared to the other mixtures. Thus, these findings suggest that incorporating cowpea:maize residue mixtures in equal proportions (50:50 on weight basis) might moderate the proportion of added N that is lost as N₂O compared to disproportionate mixing of cowpea:maize (i.e., 75:25 and 25:75) and stabilize N mineralization in soils to match N supply to demand by plants. However, further studies will be required to confirm this under both laboratory and field conditions.

Thus, the finding in this study disagrees with previous report that showed low N₂O emission after mixing a higher C:N ratio rice straw residues with lower C:N ratio groundnut residues than from sole incorporation of groundnut residues [23]. The interactive effect of mixing residues on N release is due to the movement of soluble constituents between the residues incorporated [45]. This study was conducted under controlled moisture conditions and so the movement of soluble component as affected by rainfall could not be verified. In another study, it was reported that where residues were incorporated together, the rapid decomposition of the high quality residues in the mixture at the initial stages resulted in high N availability, which stimulated the decomposition of the low quality residues, leading to a more rapid decomposition of carbon substrates [46]. However, the interactive effect may either be positive or negative depending on the component species [47]. [46] showed that mixing residues can change the abundance and composition of soil fauna leading to alterations in residue decomposition and N mineralisation, but the abundance and composition of soil fauna following the addition of cowpea or maize residues was not examined in this study.

In this study, addition of sole cowpea residue with relatively higher N input increased the proportion of N lost as N₂O. The emission of N₂O seems to correspond with the magnitude of N input from the residues but incorporation of cowpea-maize residue mixtures (representing decreased N input) rather substantially increased the proportion of N lost as N₂O compared to the sole treatments (as shown by the emission factors). This indicates an interactive effect that requires further investigation. However, incorporating cowpea:maize residue mixture on a 50:50 weight basis resulted in lower N₂O and CO₂ emission and appreciable and steady N mineralization without immobilization. A ratio of 75:25 cowpea:maize mixture had higher N input than the 50:50 treatment but this led to a higher than proportionate N₂O emission. A 25:75 cowpea:maize mixture resulted in a higher initial N₂O emission but a cumulative emission profile that is slightly lower than but not substantially different from the 50:50 treatment, but with N immobilization. While incorporation of cowpea-maize residue mixture increases the proportion of added N lost as N₂O, the results for the 50:50 treatment show a scope for manipulating N release from N-rich cowpea residue by the addition of a proportionate amount of a high C:N maize residue. The NO₃⁻ concentrations measured in this treatment and the pattern of N mineralization suggest further that NO₃⁻ leaching and N loss as N₂O through denitrification can be minimized in this treatment. Thus, incorporation of cowpea residues mixed with maize residues has the potential to promote synchrony between residue N release for crop uptake and increased nitrogen use efficiency (NUE), but this requires further investigation on the field where NUE can be verified in the presence of a crop.

The results of this study indicate that increase in both NO₃⁻ availability and organic C concentration were likely to have been responsible for the difference between N₂O fluxes on day 1 from the 100:0 and the 75:25 cowpea:maize treatments. Holding WFPS constant, it appears that this observation is in contrast with [48] who stated that NO₃⁻ concentration and low oxygen concentration other than soluble C availability were believed to be the most important condition for denitrification at the bacterial cell level. In this study source partitioning was not done, therefore the contributions of different processes to N₂O fluxes could not be determined. However, the microbial source of N₂O requires further investigation for example by adopting ¹⁵N pool dilution approach for quantifying gross nitrification [29] or by the isotopomer strategy which identifies the microbial source of N based on its position on the liner N₂O structure [49].

In situ field methods of measuring N mineralization and N₂O emissions are laborious, time consuming and are subject to confounding effects of changing edaphic and climatic factors, but they reflect more realistic temperature and moisture conditions unlike laboratory incubations, which are conducted under temperature and moisture conditions different than those occurring in the field [50]. Furthermore, in the field, soil fauna including earthworms, nematodes and arthropods may interact with microbes to alter residue decomposition and N release through communition of plant material and N mineralization [51]. In this study, the experimental conditions were supposed to be optimal for microbial decomposition and N mineralization (i.e., high but not excessive soil water content, 60% WFPS, and constantly high temperature, 25 °C) [52]. Therefore, data generated under this aerobic incubation conditions may not be easily transferable or comparable to field conditions [53]. Moreover, the absence of plant N uptake in the laboratory incubation experiment implies that the NO₃⁻ accumulation found in this study may have been over-estimated. Again, in the field soluble carbon concentration can be further increased by root exudation, which could have increased N₂O production through denitrification. However, the results obtained from this study are useful for comparing N mineralization and N₂O emission from acrisols amended with mixed residues of differing C:N ratios as the incubation method minimizes the confounding effects of changing soil temperature and moisture conditions.