Introduction

In order to achieve food security and reduce poverty in the tropical American hillsides, soil fertility maintenance and improvement must take very high priority. This would go a long way in ensuring sustained and increased food crop production in these usually extensive agroecosystems. The hillside agroecosystems in the Cauca Department in Colombia are dominated by volcanic-ash soils, which are shallow, nutrient deficient and prone to severe erosion and subsequent degradation (CIAT 2001). These areas, like many others in the tropics, are characterized by increasing human and livestock populations that are exerting increased pressure on the land. This has undermined the ability of shifting cultivation, the dominant traditional agricultural system practiced by the farmers here, to maintain and restore soil fertility. Although fallows have historically played an important role in maintaining the productivity of farming systems (Loomis 1984), traditional fallowing has become insufficient in restoring soil fertility since the duration and intensity of fallowing has been drastically reduced over the years. This shortening of traditional fallows, combined with little or no use of fertilizers, has had negative consequences on agricultural productivity and agroecosystem stability because of nutrient mining (Smaling et al. 1997). Nye and Greenland (1960) have reported that natural fallows have long been used to overcome soil fertility depletion that results from continuous cropping with no external inputs. The fallow period may vary from 5 to 20 years. Losses of mineral nutrients during the cultivation phase, through runoff, erosion, leaching and crop removal, can no longer be restored by short 1–5 year periods of natural bush fallow (Brady 1996). This situation requires soil nutrient replenishment options. When access to organic and inorganic fertilizers is limited, then improved fallows with legume species deliberately planted in order to achieve the aims of natural fallow within a short time, becomes an attractive option (Prinz 1986; Barrios et al. 1997).

Planted fallows have been recognized as an appropriate agroforestry technology that can be used by many rural farmers because of their low risk and relatively low cost for establishment with proven potential to generate additional products in the short run (i.e. firewood) while improving soil fertility (Barrios et al. 2005). Fallowing plays several roles within a farming system, as it can reduce agricultural weeds and pests and improves soil physical, chemical and biological properties. Planted fallows are thus a permanent environmental production system in which natural vegetation is intentionally replaced by planted and managed, fast growing tree, shrub or herbaceous species grown in rotation with cultivated crops (van Noordwijk 1999). Therefore, planted fallows can optimize nutrient cycling and efficiently utilize external nutrient inputs (Sanchez 1994). Planted fallows can restore agricultural productivity more rapidly than natural fallows and have been suggested as a potential solution to declining soil productivity caused by shortened natural fallow periods (Tian et al. 1999). Among other factors, the beneficial effects of planted fallows depend on fallow species, fallow duration, soil type and climate (Adejuwon and Adesina 1990).

Some studies have been carried out on the residual effects of planted fallows on subsequent crops (Adejuwon and Adesina 1990; Drechsel et al. 1996; Barrios et al. 1998; Kwesiga et al. 1999). However, very few studies have investigated the residual effects of planted fallows, simultaneously, on soil nitrogen mineralization, organic matter and phosphorus fractions, especially on volcanic-ash soils. In the same experimental site as in this study, Phiri et al. (2001) showed that planted fallows, one year after establishment, can provide soil organic matter that boosts soil fertility through improved nitrogen and phosphorus recycling. It is of interest in this study to define how transient or permanent this phenomenon is, in the longer term.

The objective of our study was to assess the effect of four planted fallow systems and a crop rotation on soil nitrogen mineralization, organic matter and phosphorus fractions at 28 months after establishment in a volcanic-ash soil. We tested four hypotheses. These include:

  1. (1)

    Planted fallows would restore soil fertility more rapidly than natural fallows and observed differences in soil characteristics would decrease with depth;

  2. (2)

    Planted fallows contributing high quality residue inputs would generate a greater proportion of organic matter fractions contributing to short-term N availability through N mineralization than those providing low quality residue inputs;

  3. (3)

    Planted fallows contributing high quality residue inputs would generate a greater proportion of P fractions contributing to short-term P availability than those providing low quality residue inputs; and

  4. (4)

    The duration of increased soil fertility (residual effect) would be longer following planted fallows producing low quality residue inputs than those producing high quality residue inputs.

Materials and methods

Description of the study area

This study was conducted at two farms in Pescador village, located in the Andean hillsides of the Cauca Department, southwestern Colombia (2°48′ N, 76°33′ W) at 1500 m above sea level. They represented two typical small farms, on a common soil type, just prior to traditional fallow management following 3 years of cassava cultivation at the end of the cropping cycle (Barrios and Cobo 2004). The area has a mean temperature of 19.3°C and a mean annual rainfall of 1900 mm (bimodal). The experiments were started in November 1997 and the fallow period concluded in February 2000.

The soils in the study area are derived from volcanic-ash deposition and are classified as Oxic Dystropepts (Inceptisols) under the USDA soil classification system (Soil Survey Staff 1998), They have medium to fine textures (IGAC 1979), high fragility, low cohesion and shallow humic layers. The soil bulk density was about 0.8 Mg m−3. The soils are acidic (pH (H2O) = 5.1), have high SOM content (C = 50 mg g−1) and are low in base saturation (1.1 and 2.5 cmol kg−1 soil for Al and Ca, respectively). They have high P-sorbing capacity and soil available P is very low due to the high allophane content (52–70 g kg−1) (Gijsman and Sanz 1998).

Experimental treatments and design

In this fallow systems study, we used two leguminous tree fallows; Calliandra (Calliandra houstoniana (Mill.) Stan. var. calothyrsus (Meisn.) Barn. = CIAT 20400) and Indigofera (Indigofera zollingeriana Miq.); and a shrubby fallow, Tithonia (Tithonia diversifolia (Hemsl.) A. Gray). Calliandra and Indigofera can withstand frequent pruning, effectively fix N and are well adapted to low night temperatures, seasonal drought and soil acidity (Barrios et al. 2005). Indigofera produces high quality biomass (decomposes rapidly) whereas Calliandra produces low quality biomass (decomposes slowly). This contrast in the experimental design provides the opportunity to study the effect of quality of plant biomass on nitrogen, soil organic matter and phosphorus dynamics and soil fertility improvement potential. Tithonia was used because it grows rapidly and accumulates high concentrations of nutrients in the shoot, especially P in the leaves (Ganunga et al. 1998; Barrios and Cobo 2004).

Experiment BM1 was set up at San Isidro farm in Pescador village as a random complete block (RCB) design with four fallow system treatments and a crop rotation treatment with three replications.

Fallow treatments:

  1. (i)

    Calliandra houstoniana (Mill.) Stan. var. calothyrsus (Meisn.) Barney CIAT 20400 (CAL) and

  2. (ii)

    Indigofera zollingerana Miq. (IND) (both CAL and IND are leguminous trees); and

  3. (iii)

    Tithonia diversifolia (HemsL) Gray. (TTH) from the Asteraceae family;

    The above three treatments were compared to

  4. (iv)

    a maize/bean rotational system (ROT); and

  5. (v)

    a natural fallow system (NAT), that was left to natural regeneration of native vegetation that is the usual practice once agricultural soils become unproductive.

Experiment BM2 was set up at Benizio Velazco farm also in Pescador village. It was also established as a RCB design but due to limited space, it consisted of three fallow system treatments with three field replications. Treatments included CAL, IND and NAT with the same management as in BM1. In both experiments, plot size was 18 × 9 m2, Pruning management regimes of ‘slash and mulch’ were carried out according to local farmers concerns about the growth and productivity of the planted fallow systems. Additional experimental design details can be found in Barrios and Cobo (2004).

Soil sampling and laboratory analytical procedures

Soil samples (0–5, 5–10 and 10–20 cm depths) were collected in March 2000, 28 months after the fallows and rotational crops were planted. Soil sampling was conducted at one month after the fallow period was completed and plant biomass cut and laid on the soil surface (but prior to removal of firewood). Before collecting soil samples, soil surface plant litter was carefully removed. Due to high inherent variability in hillside soils, a composite sample consisting of 50 cores was collected in a grid pattern from each of the 18 × 9 m2 plots. Samples were air-dried and visible plant roots removed. The samples were then gently crushed to pass through a 2-mm sieve. The <2 mm fraction was used for all the analyses and fractionation procedures. All laboratory analyses were conducted in duplicate.

Total organic C was determined colorimetrically after wet oxidation with acidified potassium dichromate and external heating (Anderson and Ingram 1993). Total N and P for whole soil were determined by digestion with concentrated sulphuric acid using selenium as a catalyst, followed by colorimetric determination with an auto analyzer (Skalar Sun Plus, the Netherlands). Bray P and exchangeable K were extracted with Bray II solution followed by colorimetric and atomic absorption determination, respectively. Exchangeable Ca and Mg, and Al were extracted with 1 M KCl solution and determined as described in CIAT (1993). Nitrate and ammonium were extracted in 1 M KCl solution and determined by colorimetry with an auto analyzer (Skalar Sun Plus, The Netherlands).

Soil potential N mineralization was determined by anaerobic incubation of whole soil (Anderson and Ingram 1993). Briefly, 10 g of soil were flooded with 25 ml of deionized water and incubated anaerobically for 7 days at 40°C. After incubation, samples were transferred to 125 ml extraction bottles, extracted with 25 ml of 4 M KCl with shaking for 1 h at 150 reciprocations min−1, filtered by gravity using Whatman No. 5 paper, pre-washed with de-ionized water and analyzed for NH +4 –N. Nitrogen mineralization was calculated as the difference in NH +4 –N between incubated samples and unincubated samples extracted immediately after flooding the soil (Barrios et al. 1996b).

Size-density fractionation of SOM (in the sand-size fraction, 150–2000 μm) was conducted as described by Phiri et al. (2001), Barrios et al. (1996a) and Meijboom et al. (1995), with minor modifications and density separation was done in reverse order. In short, an air-dried soil sample (250 g) was gradually wetted, then flooded with 2 l of water, thoroughly mixed and sieved through two superimposed sieves of 250 μm (at the top) and 150 μm (at the bottom). A jet of water through the top sieve destroyed macroaggregates, and materials retained on the sieves were washed into separate buckets and swirled with water. Through repeated swirling and decantation, the floating macroorganic matter (>150 μm) was separated from mineral material. Swirling and decanting was repeated several times until no floating materials remained. The macroorganic matter on the 150 μm sieve was then placed, and density fractionated, in a silica suspension (LudoxTM, Du Pont) adjusted to 1.13 mg m−3. The floating fraction (Ludox Light fraction, LL) was separated and placed on a drying plate. The remaining fraction in the sieve was then placed in Ludox adjusted to 1.37 mg m−3. The new floating fraction was the Ludox Intermediate fraction (LM), and the non-floating fraction, the Ludox Heavy fraction (LH). All the three fractions were washed with tap water, followed by deionized water, and then dried to a constant weight at 40°C. After weighing, the SOM fractions were ground with a mortar and pestle to <0.3 mm and then analyzed for C, N and P (Carter 1993).

Phosphorus fractionation was carried out by using a reduced (excluding acid extractants) sequential P fractionation procedure as described by Phiri et al. (2001) and Tiessen and Moir (1993), with minor modifications, where 0.5 g sieved (2-mm) soil samples were used. A sequence of extractants with increasing strength (H2O, NaHCO3, NaOH and HClO4, respectively) was applied so as to subdivide the total soil-P into inorganic (Pi) and organic (Po) fractions. The following fractions were separated:

  • (a) Anion exchange resin membranes, in bicarbonate form, were used to extract freely exchangeable Pi, herein called Resin Pi. Potassium persulphate (K2S2O8) was used to digest the Po remaining in the water in the resin Pi extraction stage.

  • (b) Labile Pi and Po sorbed to the soil surface, including some microbial P, was extracted using NaHCO3 (sodium bicarbonate) at 0.5 M and pH = 8.5.

  • (c) Pi more strongly bound to Fe and Al compounds and associated with humic compounds was extracted using NaOH (sodium hydroxide) at 0.1 M.

  • (d) HClO4 (perchloric acid) was used to digest and extract the residue containing insoluble Pi and more stable Po forms (residual P).

Total P in the NaHCO3 and NaOH extracts were measured after digestion with K2S2O8, and organic P was calculated as the difference between total P and Pi in the NaHCO3 and NaOH extracts, respectively. Total soil-P was determined by the HClO4 digestion method as described by Olsen and Sommers (1982). Inorganic P concentrations in all the digests and extracts were measured colorimetrically by the molybdate-ascorbic acid method as described by Murphy and Riley (1962).

Statistical analysis

Since the design was unbalanced and some of the blocks and treatments were nonorthogonal, the SAS MIXED procedure (PROC MIXED) was used to determine the effect of planted fallow treatments on soil parameters within and between experimental sites BM1 and BM2 (SAS Institute Inc. 1999). Least-squares means (LS-means) were computed for all the treatments/fixed effects. Comparison of the LS-means was done by the BONFERRONI multiple comparison method, which gave various probability values (P-values) that were used to determine whether two treatments were significantly different or not. The letters following the LS-means for each treatment were manually generated after ranking them (from the highest to the lowest) and assigned considering the corresponding P-values between any two LS-means compared. These letters could not be generated (automatically) by the SAS program because the design was unbalanced. Mention of statistical significance in this study refers to P < 0.05.

Results and discussion

Soil chemical characteristics

In experiment BM1, the TTH system treatment generally had higher but not always significantly higher concentrations of soil total C; total P; Bray P; B; Zn; exchangeable K, Ca and Mg; and the lowest concentration of Al in all soil depths (Table 1). The order of soil nutrient contribution was TTH > IND > ROT > CAL > NAT. These results are in agreement with those of Gachengo et al. (1999), Buresh and Niang (1997), Ganunga et al. (1998), Jama et al. (2000) and Phiri et al. (2001). These researchers have shown that Tithonia green biomass has the potential to significantly improve soil fertility through increased nutrient addition and can considerably reduce the levels of exchangeable Al in soils, thereby curbing Al toxicity problems. It is suspected that Tithonia plants may have pumped K, Ca and Mg from the subsoil through its long rooting system as well as the intimate association found with arbuscular mycorrhizal fungi in this and other areas (Sharrock et al. 2004). It is presumed that this high cation uptake by TTH raises soil pH, thereby lowering exchangeable Al. Furthermore, as an organic source of nutrients, Tithonia has been found to be often more effective than urea when applied at the same nitrogen rate because it also adds other plant nutrients, particularly K and micronutrients (Sanchez 2002).

Table 1 Influence of planted fallows on some important soil chemical characteristics#
Full size table

In experiment BM2, there was no significant effect of the fallows on the soil chemical characteristics in all soil depths (Table 1). However, on average, NAT had the highest total organic carbon, total P and B contents. On the other hand, IND had the highest contents of Ca, K, Mg and the lowest content of Al. It is well known that given enough time, the natural fallow has the potential to regenerate the natural soil fertility in this relatively less fertile experimental site. At both experimental sites, there was a significant decrease of nutrients with increasing soil depth (P < 0.0001).

Soil N availability

The residual effect of planted fallows was significant for soil nitrate but not for soil total N, ammonium-N and nitrogen mineralization potential. Indigofera zollingeriana had significantly higher values for nitrate (NO −3 –N) at both sites and at all soil depths as compared to all other fallow treatments (Table 2). In experimental site BM1, the trend was IND > CAL > TTH > NAT > ROT whereas in experimental site BM2, the trend was IND > CAL > NAT. It is likely that the IND system treatment has shown potential to increase nitrate availability in volcanic-ash soils because of Indigofera’s fast decomposition and nutrient release rates as reported by Cobo et al. (2002a, b). These relatively higher values could also be partly attributed to the large populations of the endogeic earthworm Pontoscolex corethrurus, known to enhance soil nitrogen mineralization, that were observed in IND by Barrios et al. (2005). These findings are consistent with those of Barrios et al. (2005) who have observed that Indigofera and Calliandra trees can effectively fix nitrogen and are well adapted to the soil acidity of this Colombian hillside environment.

Table 2 Influence of planted fallows on soil nitrogen dynamics#
Full size table

Soil total N, as well as total soil C, change very slowly over time and thus their potential to detect short term changes in management is limited (Barrios et al. 1996a) and our results showing no significant results among system treatments studied corroborate such findings also in volcanic ash soils. Although Tithonia has consistently been known to promote higher N mineralization potential in soils (Ganunga et al. 1998; Jama et al. 2000), and this has been attributed to its high quality green leaf biomass that is rich in nutrients and decomposes rapidly, in this study no significant differences were found among system treatments. In this study, however, N mineralization was studied under laboratory conditions. This could lead to an underestimation of potential N mineralization because it does not account for the flush of N mineralization that occurs when dry soil is rewetted in the field (Cabrera 1993).

In a study on changes in soil properties following improved fallow systems in eastern Zambia, Chirwa et al. (2004) reported that nitrogen contribution and improvement in soil physical properties were significant after one year of fallow period but declined thereafter. This observation agrees with the findings in this experiment. It also agrees with the findings of Phiri et al. (2001) who found significant differences in soil N following one year of fallow establishment. However, it contrasts with findings by Barrios et al. (1997) in eastern Zambia where Sesbania sesban improved fallows still generated significant improvements in soil N after 2 and 3 years of fallow establishment. It is probable that differences in soil type, climatic conditions, tree species and tree/shrub management regimes used could account for some of these conflicting results.

There was a general significant decrease in total N, NO −3 –N, NH +4 –N and N mineralization potential with increasing soil depth (P < 0.0001). Since the soils were sampled at the beginning of the rainy season (in April), it is possible that the mobile NO −3 –N was not yet leached to the lower soil layers (Maroko et al. 1998). Organic matter being a major source of plant available N, decreasing SOM with depth could also be a possible explanation for this trend of results.

Soil phosphorus fractions

Phosphorus exists in soils in different chemical forms and pools (Fixen and Grove 1990). Phosphorus acquisition is, however, strongly related to soil moisture, temperature and texture because this nutrient is of limited mobility in the soil, moving to the root surface via diffusion in water films on particle surfaces. For P fractions and fractionation procedures to come closer to the real-world soil management systems, there is need for partitioning soil P fractions into three discrete pools i.e. the readily available (biologically available and easily mineralizable), the moderately resistant (moderately and reversibly available) and the stable residual (sparingly available and highly recalcitrant/resistant) P as described by Guo and Yost (1998). This is the basis of P transformations and cycling in soils. Phosphorus transformations in soils involve complex mineralogical, chemical and biological processes and knowledge of these transformations is essential to understand P behavior in soils (Oberson et al. 2001).

In experiment BM1, significantly higher H2O–Po in the 0–5 cm soil depth was found in CAL, whereas ROT showed significant results for resin-Pi, when compared with that for IND at the same depth. In the 5–10 cm depth, TTH showed significant results for both H2O–Po and resin-Pi (Table 3). Results for the 10–20 cm soil depth were not significant. These findings contrasts with previous research results at the same experiment by Phiri et al. (2001) where TTH clearly showed superior performance in terms of soil P availability. Tithonia diversifolia produced higher nutrient yields, especially of P, as compared to other species, 6 and 12 months after planting in western Kenya (Niang et al. 2002). The better performance of the fallows was attributed to efficient nutrient accumulation and cycling, high quality biomass and site adaptability (Barrios and Cobo 2004).

Table 3 Influence of planted fallows on the readily (biologically) available P fractions in soils#
Full size table

Surprisingly, compared with all the other fallow treatments, NAT results for NaHCO3–Pi, NaHCO3–Po and NaHCO3–Pt were significantly higher than those for TTH in the 0–5 cm soil depth, though the results for the 5–10 and 10–20 cm soil depths were not significant (Table 3). This could probably be due to the fact that young natural fallows (<3 years) usually have high rates of photosynthesis, rapid increases in leaf area and leaf biomass, and higher tissue concentrations than older secondary vegetation (Szott and Palm 1996). These characteristics result in high rates of nutrient demand and accumulation, litter production and soil cover establishment (Szott et al. 1999). Thus, they help to reduce nutrient losses by leaching and runoff and to mobilize P from non-readily available forms in the soil. In experiment BM2, there was no overall significant residual effect of planted fallows on the readily available P fractions in the soils. However, on average, NAT had higher values for H2O–Po, resin–Pi, NaHCO3–Pi, NaHCO3–Po and NaHCO3–Pt at all soil depths.

In both experiments, there was a significant decrease in H2O–Po, resin–Pi, NaHCO3–Pi, NaHCO3–Po and NaHCO3–Pt with increasing soil depth (P < 0.05). This could be attributed to the high soil organic matter content in the topsoil layers of these soils as reported by Phiri et al. (2001). Presumably, the presence of deep-rooted planted fallows resulted in P from deeper soil layers being pumped into the topsoil layers.

Table 4 shows the impact of planted fallow systems on moderately resistant and stable (residual) P fractions. The moderately resistant P fraction mostly includes NaOH–Pi and NaOH–Po, which normally become available in the medium term i.e. from a few months to years (Cross and Schlesinger 1995). In experiment BM1, results for the moderately resistant P fraction in the 0–5, 5–10 and 10–20 cm soil layers were higher for NAT, TTH and IND respectively, though the results were not statistically significant, except for the significant difference between NAT and TTH in the 0–5 cm soil layer. Barrios et al. (2004) have reported large populations of the endogeic earthworm Pontoscolex corethrurus under IND treatments, and so there is a possibility that in this study, there was movement of organic P by earthworms to the deeper soil layers. Maintenance of active earthworm populations can be favourable to nutrient cycling and crop production in low-input tropical agroecosystems (Pashanasi et al. 1996). These mixed results show that there was no clearly dominant fallow treatment as far as provision of moderately resistant P is concerned. However, the situation was slightly different in experiment BM2 where NAT had the highest values for this P fraction in all the soil layers, despite the fact that the results were not statistically significant. Tiessen et al. (1992) found a significant improvement in NaOH-extractable P after nine years of fallow, which leads us to the argument that probably the time (28 months of fallowing) was too short to get any significant results for this fraction.

Table 4 Influence of planted fallows on moderately resistant and stable (residual) P fractions in soils#
Full size table

The residual P fraction showed a similar trend to that of the moderately resistant fraction in both experiments. This fraction comprises of stable humus, highly insoluble Pi forms and the more stable Po forms as described by Hedley et al. (1982). It becomes available in the longer term, may be after several cropping cycles.

Total Pi, total Po, total P (Pi + Po) and total Po as a percentage of total soil P were generally not significantly affected by the planted fallows in both experiments (Table 4). These results are consistent with those of Maroko et al. (1999) who observed that land-use systems had no effect on extractable P fractions in tropical soils. Of special interest here is total Po because P in organic pools is better protected from losses due to fixation than P in inorganic pools in volcanic-ash soils with high P-sorbing capacity (Phiri et al. 2001). Phosphorus in organic pools interacts less with soil, thereby minimizing chances of loss and increasing P cycling. Lack of significant improvement of the organic P pool by the fallow treatments therefore suggests that significant differences observed by Phiri et al. (2001) were transient and there continues to be a high potential for P fixation in volcanic-ash soils of this agroecosystem.

Soil organic matter fractions

Three soil organic matter (SOM) fractions (LL, LM and LH) were recovered using size-density fractionation as described by Phiri et al. (2001), Barrios et al. (1996a) and Meijboom et al. (1995). These fractions have shown potential as more sensitive indicators of the impacts of change in land use or cropping management than total SOM content, especially for topsoil layers. According to Hassink (1994), total SOM content is less sensitive as an indicator of sustainability of land use systems, because it changes relatively slowly under different management regimes and has a high spatial variability.

In both experiments, significant results for the weights of the SOM fractions as affected by system treatments were obtained in most soil layers (Table 5). The weights decreased in the order LL > LM > LH. For experiment BM1, the average SOM fractions weights for the LL, LM and LH at the 0–5 cm soil depth accounted for 11.3, 7.3 and 1.4% of the sum of all the SOM fractions, respectively. For experiment BM2, they accounted for 22.7, 10 and 0.6%, respectively. The recovery of SOM size-density fractions is far lower than that obtained by Phiri et al. (2001) in the same experiment, one year after establishment of the fallow experiment. This could be explained by the fact that most SOM in this tropical environment could have been oxidized during the dry season prior to sampling of the soils. Increased soil temperatures can significantly reduce SOM fraction quantity as observed by Tiessen and Stewart (1983).

Table 5 Influence of planted allows on the weights and the carbon,nitrogen and phosphorus contents of the soil organic matter fractions#
Full size table

Barrios and Cobo (2004) have reported significantly higher leaf biomass for CAL and IND at both experimental sites BM1 and BM2, though biomass from other plant parts as well as nutrient accumulation were similar at both sites. This finding has important implications for long-term site productivity as CAL biomass may lead to a better synchronization of nutrient release and crop demand. Therefore, for sustainable agricultural productivity, CAL may be a better candidate than the faster decomposing (high quality biomass) TTH or IND.

Natural fallow produced significantly higher weights of the LL fraction in all soil layers in experiment BM1. A mixture of local plant communities could probably have led to higher SOM in the NAT fallow treatment. This finding seems to complement results by Meijboom et al. (1995) who recovered highest weights of the LL SOM fraction on natural grassland soils. For experiment BM2, significantly higher weights of the LL SOM fraction for all soil layers were found in CAL. Furthermore, CAL produced significant results for the weights of the LM and LH SOM fractions in most soil layers in both experiments. This confirms results by Cobo et al. (2002a) and Barrios et al. (2004) who have reported Calliandra as a fallow species that produces large quantities of low quality biomass. It has been consistently found that the amount of the light fraction is affected by the amount of litter production and its quality (Christensen 1992).

In experiment BM1, the impact of fallow systems on the contents of C, N and P in the SOM fractions clearly narrows down to IND showing significant results in the 0–5 cm soil layer (Table 5). On average, IND out-performed all the other fallow system treatments in the other soil layers, with NAT performing worst. The same trend of results is observed in experimental site BM2. These results contrast with those of Phiri et al. (2001) who reported that TTH gave significantly higher values for the contents of C, N and P in the SOM fractions of the 0–5 cm soil layer in the same experiment, one year after establishment of the fallows; and that CAL performed significantly better than the other fallow treatments in the 5–10 cm soil layer. These observations suggest that the impact of different fallow species can be inconsistent in some cases as fallow plants can have diverse responses to environmental conditions and system management. The use of mixed fallows may be a more effective strategy to maximize on the residual benefits in terms of nutrient contribution to the soils (Yamada and Gholz 2002).

Conclusions

In this study, we confirmed that mineral nitrogen, weight and nutrient content of SOM fractions and soil phosphorus fractions were sensitive to soil management using planted fallows on inherently variable volcanic-ash soils in the hillsides of the Colombian Andes. These methodologies using soil organic matter pools and phosphorus fractions have been an important alternative to conventional methods that determine total soil nutrient contents that do not distinguish active forms of the nutrients.

At the conclusion of this on-farm experiment, planted fallow systems studied were found to have generated slight improvements in soil N and P availability compared to the natural regeneration of native vegetation. This contrast with much higher differences found 12 months after planted fallow management in the same experiments and thus questions the need for 28-month fallow periods with species studied. However, other factors linked to the system such as economic attractiveness of planted fallows to farmers in addition to soil fertility recovery, like those associated with the production of firewood, would need to be also considered for future studies on trade-off analysis.

The choice of experimental on-farm sites with soils following two seasons (3 years) of cassava mono cropping at the end of the cropping cycle recognized the capacity of cassava to further deplete soil nutrients and the entry point for planted fallow systems for a more rapid regeneration of soil fertility than the natural fallow. Our results suggest that for effective regeneration of soil N and P availability in the volcanic-ash soils with high P-fixation, there is need for external nutrient inputs in addition to the planted fallow effect. Further studies should evaluate the soil nutrient replacement value provided by planted fallow systems and thus the reduction in amount of fertilizer expenditures by resource-limited farmers. In addition, multiple-location testing of best bet planted fallow species should be conducted to identify the biophysical and socioeconomic limits of the planted fallow technology in the Andean hillsides.