Introduction

The predominance of smallholder farming systems constitutes a major constraint to soil fertility improvement in Eastern Africa. Such systems are characterized by land scarcity, low soil fertility, low income, low input of fertilizers and therefore low crop production, etc. The limited use of fertilizers in such systems has stimulated research to identify low-cost alternatives. One such alternative is the use of organic resources available to farmers. These include plant residues, green manures, animal manures and organic wastes. Organic resources can play an important role as a direct supply of nutrients, provided the nutrient release is synchronized with crop demand (Myers et al. 1994). While organic inputs often have lower capacity as a direct source of nutrients compared to fertilizers, they affect many soil properties controlling nutrient cycling, availability and crop production (Palm et al. 1997). Studies have reported that application of organic amendments can change soil water retention (Ghuman and Sur 2001), soil structure and aggregation (Acosta-Martinez et al. 1999), soil pH (Tang and Yu 1999; Paul et al. 2001), microbial biomass and activity (Chander et al. 1995; Acosta-Martinez et al. 1999), soil organic matter (Wander et al. 1994) and plant nutrient availability. However, organic resources differ in their capacity to release nutrients (Cobo et al. 2002) and the quality of the organic material is an important consideration when studying the short- or long-term effects of organic resources on nutrient availability (Palm et al. 2001).

In Eastern Africa, N and P are considered as the most limiting nutrients for crop production and strategies for improving soil fertility include the incorporation into soil of leaf biomass from trees and shrubs found in a farm’s vicinity. Most field experiments have focused on the effect of such residues on crop yield and on the availability of the nutrient of interest, but little or no attention has been given to other soil indicators of nutrient cycling (Jama et al. 1997; Gachengo et al. 1999). In an experiment designed to study the contribution of organic residues in comparison to inorganic fertilizers initially on the control of a pest weed (striga, Striga hermonthica Del.) and then on P-availability improvement in a P-limiting soil in western Kenya, it was found that some organic residues reduced striga emergence (Gacheru and Rao 2001) whereas soil labile P in some residue treatments was higher than that seen with an equivalent rate of P fertilizers (Nziguheba et al. 2000). We then hypothesised that repeated incorporations of organic residues may have affected many other soil indicators of short- or long-term nutrient cycling. The main objectives of the present study were: (1) to assess changes in soil organic C (SOC), C and N mineralization, microbial biomass C (MBC) and N (MBN) as affected by organic residues and inorganic fertilizers; (2) to determine the effect of organic residue quality on these parameters; and (3) to study the distribution of C and N in soil organic matter fractions after organic residue amendments.

Materials and methods

Site description and experimental design

The field experiment was conducted in the highlands of western Kenya (latitude 0°06′N, longitude 34°34′E, altitude 1,420 m, mean annual rainfall 1,800 mm). The area has two growing seasons: a long rainy season (LR) from March to August and a short rainy season (SR) from September to January. The soil is classified as a very fine, isohyperthermic Kandiudalfic Eutrudox (USDA). The top 0.15-m layer, sampled at initiation of the experiment has pH (soil/water 1:2.5)=5.4, total organic C=14 g kg−1, total N=1.7 g kg−1, exchangeable Ca=3.9 cmol kg−1, exchangeable Mg=1.5 cmol kg−1, exchangeable K=0.13 cmol kg−1, exchangeable acidity=0.6 cmol kg−1, bicarbonate extractable inorganic P=1.7 mg kg−1, bulk density=1.1 g cm−3, 420 g kg−1 clay and 330 g kg−1 sand. P is often the primary limiting nutrient of crop production in this area (Sanchez et al. 1997).

The experiment was established during the SR of 1995. The field was tilled with hand hoes to a depth of 0.15 m at the beginning of each rainy season. Treatments consisted of leaf biomass from six tree or shrub species [Calliandra calothyrsus Meissner, Senna spectabilis (DC) H.S. Irwin and Barneby, Croton megalocarpus Hutch, Lantana camara L., Sesbania sesban (L.) Merr, and Tithonia diversifolia (Hemsley) A. Gray] (Table 1) and inorganic fertilizers. The specific tree or shrub species were selected for their availability in the area. Each rainy season, fresh leaves of each of the species were picked in the vicinity of the experiment and applied separately to the soil at a rate of 5 Mg ha−1 on a dry matter basis. The inorganic fertilizer treatments consisted of a uniform application of 120 kg N ha−1 as urea and the application of triple superphosphate in separate plots at different rates: 0 kg P (120N 0P), 10 kg P (120N 10P), 25 kg P (120N 25P), 50 kg P (120N 50P), or 150 kg P ha−1 (120N 150P). Urea was split applied with one-third of the rate applied at planting and the remaining 5 weeks later. K was added each season in all treatments as a blanket application of 100 kg K ha−1. All residues and fertilizers were broadcast and incorporated by hand hoes into the top 0.15 m, at planting, for five consecutive maize growing seasons, two per year in the two rainy seasons, in plots of 7×7.5 m. Each treatment was replicated 4 times in a randomized complete block design. Total N added in residues at each season ranged between 136 and 183 kg N ha−1, and total P between 8 and 14 kg P ha−1. With an average C content of 45% in all residues, a total amount of 11.3 Mg C ha−1 was applied by the end of the experiment, in each of the organic residue treatments. Maize (Zea mays L.) hybrid 511 was planted each short season and hybrid 612 or 614 during the LRs, in rows at a spacing of 0.75 m between rows and 0.25 m between plants in the row. Two seeds were sown per hole and later thinned to one per hole. Weeding took place whenever appropriate. The above-ground maize biomass was removed from the field at each harvest.

Table 1 Selected characteristics of organic materials applied in a field experiment in western Kenya during five consecutive maize growing seasonsa. PP Soluble polyphenols
Full size table

The amounts of rainfall received during the cropping seasons was 496, 1,287, 975, 1,086, and 1,272 mm for the SR 1995, LR 1996, SR 1996, LR 1997 and SR 1997, respectively.

Plant and soil analyses

Plant analyses

The quality parameters of the plant residues used are given in Table 1 and were determined following the methods recommended by Palm and Rowland (1997). Fresh leaves of each species were air-dried and ground to pass a 0.5-mm sieve. N and P contents were determined following a sulphuric acid Kjeldahl digestion (Anderson and Ingram 1993). Lignin was determined according to the acid detergent fibre method of Van Soest (1963), and total soluble polyphenols (PP) according to Anderson and Ingram (1993). The water-soluble fraction of the materials was determined based on the cold water method of TAPPI (1988). Total C in this filtrate was determined as a measure of soluble C following the potassium dichromate method of Nelson and Sommers (1982). All residues had a C/N ratio ranging between 12 and 17, which is lower than the general threshold of 25 used as an index for predicting N immobilization/mineralization. The incorporation of these residues would be expected to result in net N mineralization (Myers et al. 1994; Palm and Rowland 1997).

Soil analyses

Soil analyses were done on soil samples taken from the top 0.15-m layer at the harvest of maize grown during the fifth rainy season (SR97). Soil samples were air-dried, and sieved to 2 mm prior to analyses. Soil pH was measured in a 1:2.5 soil/water suspension.

Soil inorganic N was calculated as the sum of NH4+ and NO3− extracted with 2 M KCl after 1 h of shaking (Anderson and Ingram 1993). Potentially mineralizable N was determined following the anaerobic mineralization procedure described by Drinkwater et al. (1996). Soil was anaerobically incubated for 7 days and mineralizable N calculated as the difference between NH4+ extracted with 2 M KCl after incubation and NH4+ extracted before incubation.

MBC and MBN contents were determined by the chloroform fumigation–extraction method of Tate et al. (1988). Dry soils were re-wetted to 60% water filled pore space and pre-incubated for 7 days prior to chloroform fumigation. Soils were then exposed to alcohol-free chloroform for 24 h and C and N extracted thereafter with 0.5 M K2SO4 on fumigated and unfumigated samples. Chloroform released C and N was calculated as the difference between C and N extracted in fumigated and unfumigated samples and was converted to MBC and MBN using respective conversion factors of 0.35 and 0.4 (Sparling and West 1989).

Soil C mineralization was estimated by measuring the CO2 evolution (Anderson 1982) during the pre-incubation period of samples used for MBC and MBN determination. CO2 evolved from the soil after 1 day of pre-incubation (1 day) or throughout the pre-incubation (7 days) was trapped using 0.5 M NaOH and determined by back-titration with 0.15 M HCl after precipitation with excess BaCl2.

SOC was determined by wet oxidation with acidified dichromate coupled with heating at 150°C for 30 min, and total N by Kjeldahl digestion (Anderson and Ingram 1993).

Soil fractionation

To determine soil organic matter associated with soil particle size, soil samples were fractionated into different particle-size separates by wet sieving after soil dispersion with a sodic resin (Feller et al. 1991; Gavinelli et al. 1995). Air-dried soil (50 g) was passed through a 2-mm sieve and shaken end-to-end at 125 reciprocations min−1 for 16 h with 200 ml of deionized water and 30 g of cation exchange resin (Amberlite IR-120) saturated with Na. Soil size fractions were separated by wet sieving successively through 250- and 53-μm sieves using tap water. The fraction <53 μm was further separated into 0- to 5-μm and 5- to 53-μm fractions by sonication and sedimentation. All fractions >53 μm were separated into organic and mineral fractions by flotation in water. In this study, only the organic fractions >53 μm (POM) and the fractions <53 μm are presented: [250- to 2,000-μm organic fraction (O250), 53- to 250-μm organic fraction (O53), 5- to 53-μm fraction (5–53), and 0- to 5-μm fractions (0–5)]. Mineral fractions >53 μm are not presented because they were not affected by any treatment. Total C and N in the fractions were determined using a CN analyser–mass spectrometer [Roboprep (Europa Scientific) VG Micromass 622, UK]. Given that the fractionation is time consuming, only two inorganic treatments (i.e. 120N 10P, 120N 150P) were considered in addition to the control and organic residue treatments.

Statistical analysis

ANOVA was conducted using the general linear model procedure (GLM) of SAS (SAS Institute 1995). Differences among treatment means were determined by the least significant difference (LSD). Mention of significant differences refers to α=0.05 unless otherwise mentioned. Simple Pearson correlation analysis was carried out between plant characteristics and studied soil parameters.

Results and discussion

Soil pH, organic C, total N and inorganic N

Soil pH values ranged from 4.9 to 5.3 (Table 2). These values were lower than the pH value of 5.4 measured at the beginning of the experiment (see Site description and experimental design). Conyers et al. (1995) reported seasonal fluctuations in pH due to changes in soil temperature and water potential with corresponding effects on microbial activity and proton budgets. In our study, rainfall fluctuations, particularly heavy rains with periods of flooding during the SR97, could have affected some processes controlling soil pH such as N mineralization and nitrification. For example, the NH4+/NO3− ratio decreased from an average of 2.0 at the harvest of maize grown in SR95 to an average of 0.8 at the harvest in SR97 (data not shown), a decrease that would contribute to the observed decrease in soil pH.

Table 2 Soil pH, soil organic C (SOC), inorganic N and total N as affected by organic and inorganic amendments in a field trial cropped with maize in western Kenya. 120N 0P 120 kg N, 0 kg P; 120N 10P 120 kg N, 10 kg P; 120N 25P 120 kg N, 25 kg P; 120N 50P 120 kg N, 50 kg P; 120N 150P 120 kg N, 150 kg P ha−1; LSD least significant difference (P<0.05)
Full size table

Croton, Lantana and Tithonia additions resulted in higher pH than the fertilizer treatments receiving 0–50 kg P ha−1, with the Lantana treatment providing the highest soil pH, significantly higher than the values obtained using inorganic fertilizers and Senna, Calliandra, and Sesbania residues. Among the residue treatments, Sesbania resulted in the lowest pH. Increases in soil pH with some organic residue amendments have been reported in laboratory studies (Phan 2000; Paul et al. 2001). However, organic residues differ in their ability to change soil pH. Tang and Yu (1999) reported that the concentration of organic anions in residues and the degree of decomposition of plant residues were important parameters influencing the direction (increase or decrease) and magnitude of changes in the pH of amended soils. No organic anions were measured in our study. However, pH values correlated positively with the lignin/N ratio of the residues (Table 3), a residue property that affects decomposition and N release from residues (Palm and Rowland 1997).

Table 3 Correlation coefficients (r) between soil biochemical properties measured at the harvest of crop 5 and some characteristics of organic residues applied to the soil (n=6)a. C min Mineralized C; MBC microbial biomass C; O250-C C concentration in 250- to 2,000-μm organic fraction; O250-N N concentration in O250; O53 53- to 250-μm organic fraction; NS not significant
Full size table

The concentration of SOC was higher in soils treated with organic residues and fertilizers compared to the control, with the exception of 120N 50P treatments (Table 2). Few significant differences were observed in SOC among treated soils. Higher SOC was observed in Lantana treatments compared to the fertilizer treatments receiving 50 and 150 kg P ha−1, and in Calliandra treatments compared to 120N 50P treatments. There were no significant differences in SOC among residue treatments.

Differing results have been obtained on the sensitivity of SOC to changes in soil management. Powlson et al. (1987) reported a 5% increase in SOC after 18 years of annual incorporation of barley straw (Hordeum vulgare L.), representing 4–7% of the 37.8 Mg C ha−1 added. Goyal et al. (1999) observed a 54% increase in SOC after 11 years of addition of a combination of wheat straw (Triticum aestivum L.), totalling 33.5 Mg C ha−1, and fertilizers to a sandy loam soil in India. Changes in SOC in residue-treated soil, in our study, were very small and in general lower than the random error of 10% (Landon 1991). Of the 11.3 Mg C ha−1 added in 2.5 years, 2.9–24.9% was recovered in SOC. The lack of large changes in SOC in this study may be attributed to the small amount of residues added during the shorter time frame of the experiment. Although the magnitude of changes in SOC differed among residues, we did not find any significant correlation between SOC and residue quality parameters (Table 3). Continuous cropping without fertilization but with removal of crop residues in the control treatment resulted in a 17% reduction in the initial 14 g SOC kg−1. The observed increase in SOC in the fertilizer treatments compared to the control, despite removal of above-ground biomass, could be attributed to the contribution from root biomass induced by better maize growth than in the control (Nziguheba et al. 2000).

After five fertilizer N or organic residue additions, total soil N did not change significantly in amended soils compared to the control (Table 2).

All organic residues increased inorganic N compared to the control (Table 2). Inorganic N in Calliandra, Senna and Tithonia was higher than in the fertilizer treatments. Among organic residues, Sesbania and Croton treatments resulted in lower inorganic N than Senna and Tithonia. Inorganic N was increased by fertilizer treatments except 120N 10P and 120N 25P. The increase in inorganic N in all organic residues compared to the control suggests an improvement of N availability by organic residue additions. In addition, larger inorganic N concentrations in Calliandra-, Senna- and Tithonia -treated soils compared to fertilizer-treated soils would indicate a better improvement of soil N status by organic residues than by fertilizers. Difference in cumulative maize yield during the five seasons was not significantly different between fertilizers and organic residues (Nziguheba et al. 2000) and thus would not explain the differences in inorganic N status. Leaching was not measured in this study. However, slower N release is expected in organic residues than in fertilizers, that could lead to a gradual increase in soil inorganic N, whereas immediate N release from fertilizers can lead to important N losses. Although soil inorganic N differed among the residues, soil inorganic N did not correlate with any of the studied residue quality parameters.

MBC and MBN

MBC has been referred to as an early indicator of changes in soil organic matter due to changes in soil management practices (Powlson et al. 1987; Goyal et al. 1999). Croton, Lantana and Senna treatments resulted in significantly higher MBC compared to the control. MBC in Tithonia and Calliandra treatments did not statistically differ from that in the control treatment. Among residues, the Sesbania treatment resulted in much lower MBC compared to that in other organic residues. In addition, MBC in Tithonia and Calliandra was lower than that in Lantana treatments. Fertilizer treatments did not affect MBC compared to the control except for lower values measured in the 120N 0P treatment.

Only few significant differences were observed in MBN, with only the Croton treatment increasing MBN above that of the control.

Differences in MBC and MBN among cropping systems (crop rotations, organic residue applications) have been attributed to differences in substrate composition (Goyal et al. 1999; Kaur et al. 2000). In our study, MBC correlated positively and significantly with the lignin content of residues and its ratio with N [lignin/N, (lignin+PP)/N] (Table 3). However, the extremely low lignin and low MBC values in Sesbania compared to other residues might have favoured the correlation. A positive correlation implies that the higher the lignin content in a residue, the higher the MBC. However, since lignin is known to retard decomposition, a negative correlation would be expected. Most of the residues used in this study decomposed relatively quickly with a half-life ranging from 1.1 to 8.2 weeks (Kwabiah 1997). Soil samples were collected at maize harvest (about 19 weeks after incorporation of residues) which would explain the influence of organic residues with relatively high lignin contents on soil microbial biomass. On the other hand, the presence of lignin would favour the development of fungi. However, whether a positive correlation between MBC and lignin occurs would depend on the contribution of fungi to the measured MBC. In this study, however, we did not look at the microbial composition and we, therefore, cannot draw any conclusion on the determining process. Higher MBC and MBN indicate greater potential of a soil to supply nutrients through decomposition and mineralization which are regarded as important processes in low-input systems (Kaur et al. 2000).

N and C mineralization

Mineralizable N was larger in organic treatments than in the control and fertilizer treatments (Table 4). Among the fertilizer treatments, only the addition of 150 kg P ha−1 increased mineralizable N over that in the control treatment. Among the organic residue treatments, Lantana resulted in higher mineralizable N than Sesbania and Calliandra.

Table 4 Amounts of C and N mineralized, MBC and microbial biomass N (MBN) in a soil amended with organic residues and fertilizers in the highlands of western Kenyaa. For other abbreviations see Tables 2 and 3
Full size table

C mineralization was measured on day 1 and day 7 of pre-incubation and the ranking of organic residues for the amount of C mineralized changed depending on the time of CO2 determination. The soil was re-wetted prior to pre-incubation and the resulting flush of microbes would have affected the CO2 released in the first day. It becomes, therefore, difficult to distinguish between the treatment and the rewetting effects. Nevertheless, it appears from the results that the presence of a C substrate was already discernible (Table 4). After 7 days of pre-incubation, the highest C mineralized values were obtained in Lantana and Croton treatments, and were significantly higher than the values obtained in the control, 120N 10P and 120N 50P treatments (Table 4). Senna, Lantana and Croton treatments resulted in larger amounts of C mineralized than those measured in the 120N 10P and 120N 50P treatments. The only significant difference among organic residues was the higher mineralized C in the Lantana treatment than in Sesbania treatment in 7 days. Addition of fertilizers did not change significantly the C mineralized in the control treatment.

C and N mineralization have been referred to as the most sensitive indicators of soil organic matter dynamics and productivity of soils (Biederbeck et al. 1998). The important changes in these parameters between the different treatments, in our study, confirm their sensitivity to crop management. Although there were differences in the amount of mineralizable N in the different residue treatments, mineralizable N did not correlate significantly with any residue quality parameter (data not shown). The amount of C mineralized was affected by both the length of incubation and the quality of organic residues (Tables 3, 4). The similar behaviour of Croton- and Lantana-treated soils can be attributed to the similar ratio of (lignin+PP)/N in these residues. C mineralized after 7 days could be related to the (lignin+PP)/N and lignin/N ratios while none of these residue quality parameters could describe the amount of C mineralized after 1 day (Table 3). As for MBC, the positive correlation between C mineralized in 7 days and lignin can be attributed to the sampling time (19 weeks after the last application of residues) when the materials with lower lignin contents had already decomposed (Table 3).

Soil organic matter

Size separates

All organic residues increased the amount of particulate organic matter (POM) fractions (POM=O53 and O250) compared to the control (Table 5). The Calliandra, Croton, Lantana and Senna treatments resulted in more substantial POM fractions than the fertilizer treatments. Larger increases in POM fractions with organic residues than with fertilizers are consistent with earlier observations (Wander et al. 1994) and confirm the link between POM and organic amendments (Vanlauwe et al. 1998). Among the organic residues, the Calliandra treatment resulted in larger O250 fractions than any other residue treatment and in larger O53 fractions than any other residue, except for Lantana. Differences in POM between the different residue treatments may originate from the decomposition characteristics of the residues. By definition, POM consisted of less decomposed organic matter and is considered as the most labile fraction of soil organic matter (Woomer et al. 1994; Vanlauwe et al. 1998). The sequence of residues in terms of their decomposition rate is: Tithonia>Sesbania>Senna>Lantana>Croton>Calliandra (Kwabiah 1997). Calliandra residues were the slowest to decompose among the organic residues used and resulted in the largest amounts of POM, whereas Sesbania and Tithonia decomposed fastest. Hassink (1995) found a much bigger light fraction from chaff amendments than from lucerne (Medicago sativa L.) and assumed that organic matter from the faster decomposing residue (lucerne) was transferred rapidly to heavier fractions than that of slowly decomposing residues. In this study, we did not separate organic matter from the mineral fraction in the soil fractions <53 μm. Nevertheless, transfer of organic matter in the heavier fraction would be accompanied by an increase in C in these fractions. No significant difference in C content was observed between the different residue treatments in the soil fractions <53 μm. Whereas the short duration of the experiment (2.5 years) can partly explain the lack of many significant differences in POM between different residues compared to large differences observed after 25 years in Hassink’s study (1995), the difference in decomposition characteristics between the residues used in this study was also not as large as that between chaff and lucerne. Among the fertilizer treatments, only the 120N 150P treatment increased the O250 fraction compared to the control. The increase in the O250 fraction with the addition of 150 kg P ha−1 can be attributed to larger organic inputs, mainly root biomass, related to higher yields as reported by Nziguheba et al. (2002). This root biomass decomposed slowly compared to the biomass from the added organic residues, which resulted in increased amounts in the larger POM fractions. Addition of organic residues and fertilizers did not affect the size of fractions <53 μm, most probably because of the large size of the mineral fraction which can be derived from the soil texture.

Table 5 Amount of particle size fractions at the harvest of maize after five seasonal additions of organic residues and fertilizers to a soil in a maize-based cropping system in western Kenya. For abbreviations see Tables 2 and 3
Full size table

C, N and C/N ratio in SOM fractions

Despite high concentrations of C and N in POM fractions, with ranges of 29–37% C and 1.0–2.6% N (data not shown), the contribution of these fractions to total soil C and N was very small (Table 6). In the control soil, C and N in the POM fractions (POM-C, POM-N) represented respective values of 3.6 and 1.9% of the total soil C and N (Table 6). These values were increased by the addition of organic residues to 6.1–11.0% for C and to 3.7–8.2% for N. Although the largest proportions of soil C and N were found in the fractions <53 μm, differences between the treatments occurred mainly in the POM fractions, confirming the sensitivity of POM to soil management.

Table 6 Distribution of C and N in soil fractions of a soil amended with organic residues and fertilizers in a maize-based cropping system in western Kenya. For abbreviations see Table 2
Full size table

Addition of organic residues significantly increased POM-C and POM-N, except for the Sesbania and Tithonia treatments for O250-N. The only significant increase from fertilizers was observed in the POM-C for the 120N 150P treatment. Significant differences in POM-C and POM-N between organic residues occurred mainly in the O53 fraction where the amounts of C and N obtained with Lantana, Calliandra and Croton amendments were larger than the values from Tithonia and Sesbania amendments. In the O250 fraction, C and N contents in Calliandra treatments were significantly larger than those in other residue treatments. The amounts of C and N in fractions <53 μm were not significantly affected by the treatments.

The C/N ratios of the POM fractions in the unfertilized plots were 28 and 19, respectively, in the O250 and O53 fractions (Table 6). Addition of organic residues decreased the ratio in the O53 but not in the O250 except for Calliandra. No significant reduction in the C/N ratio was observed in fertilizer treatments. The differences in the C/N ratio between treatments narrowed in the 5- to 53-μm fractions and became relatively constant in the 0- to 5-μm fractions for all treatments.

Wider C/N ratios in POM, particularly in the O53 fractions in the fertilizer treatments than in organic residue treatments, suggest differences in the quality of soil organic matter resulting from the two types of amendments. This confirms our earlier assumption that POM in fertilizer treatments was mainly maize-root derived and was of low quality (N content of about 6.4 g kg−1) compared to that in organic residue treatments. Because of this, the release of POM-N would be slower in fertilizer-treated soil than in organic residue treatments (Bending and Turner 1999). The narrow C/N ratio in the Calliandra treatment would suggest a more rapid N mineralization in this treatment than in other organic residue treatments. However, this was not confirmed by the N mineralization data. The large PP concentrations in the Calliandra residue can bind N, thus constraining its mineralization (Handayanto et al. 1997).

The wider C/N ratio in POM compared to fractions <53 μm is in line with the findings of Hassink (1995) and Vanlauwe et al. (1998) who concluded that POM consisted mainly of less decomposed plant residues while organic matter in smaller fractions is more stable.

Relationships between studied soil parameters

Correlations between the C and N concentrations in POM and other measured soil properties were found to be more significant in the O53 fraction as compared to the O250 fraction (Table 7). C and N concentrations of the O53 fraction correlated significantly with the amounts of C and N mineralized in the whole soil. The C/N ratio in all POM fractions correlated negatively with SOC. In addition, the C/N ratio in O250 correlated negatively with C mineralized after 1 day of incubation whereas the C/N ratio in O53 was negatively correlated with N mineralized.

Table 7 Correlation coefficients (r) between different biochemical properties of a soil amended with plant residues and inorganic fertilizers in western Kenya (n=13)a.N min N mineralized; for other abbreviations see Tables 2 and 3
Full size table

C and N mineralization have been referred to as the best indicators of biologically active organic matter (Biederbeck et al. 1998). The significant correlations between C and N mineralized and the C and N content of the O53 fractions, in our study, suggest that the microbiological activity was mainly localized in this fraction. The lower microbiological activity in O250 compared to O53 can be attributed to the long time elapsed between the application of residues and the analyses. Easily decomposed materials would have been processed by then and only recalcitrant materials subject to slow change would remain. MBC correlated with total soil N, C and N mineralization at 7 days, N in O53 in POM but not with soil organic C. Often poor correlations between MBC and other soil properties have been reported (Hassink et al. 1993; Hassink 1995; Wang et al. 2003). Wang et al. (2003) attributed the poor relationship with soil organic C to the non-availability of most of the organic C to microbes due to chemical recalcitrance or physical protection. Such physical protection would be more pronounced in clayey soil, as used in our study, than in coarse-textured soils. Correlations between MBC and C mineralization have generated different conclusions. Poor correlations between the two parameters have been attributed to the non-reliability of MBC as an indicator of microbial activity. However, some studies have shown an influence of the time of CO2 determination on the resultant correlations. Wang et al. (2003) reported good correlations when CO2 was determined during 7 days of pre-incubation of re-wetted air-dried soil, whereas poor correlations were observed when CO2 was determined during the incubation of soil which had already been subject to pre-incubation. Hassink et al. (1993) and Hassink (1995) attributed the lack of correlation between POM and MBC to the presence of the largest proportion of microbial biomass in the clay fractions.

Conclusions

The often advocated advantage of organic inputs over fertilizers when applied to soil is the improvement of many soil physical and biochemical properties controlling nutrient cycling and hence crop production. In this study, MBC, C and N mineralization, and POM were the properties most increased by organic residue amendments. Changes in these properties by the addition of fertilizers depended on their rate of application. While it was true that most of the organic residues used in this study improved more than one soil parameter, in only a few cases, mainly in N mineralization and POM, was the effect larger than that of mineral fertilizers. Perhaps the short duration of the experiment (2.5 years) was the reason for the lack of a clear effect of the organic versus fertilizer treatments. The quality of the organic residue affects the magnitude of changes in a given soil parameter. The decomposition characteristics of organic residues expressed in terms of lignin/N or (lignin+PP)/N ratios influenced the MBC, C mineralization, and C and N in POM fractions. Among the POM fractions, the 53- to 250-μm fraction was found to be the most biologically active, correlating with most of the studied soil properties.