Soil Nutrients and Maize Yields Responses to Agroforestry Tree Post-fallows Management in Tanzania

Agriculture forms a backbone of many countries in sub-Saharan Africa (SSA) thus has the potential to contribute to achieving Sustainable Development Goals (SDGs). However, agriculture in the SSA is characterized by low production due to soil fertility depletion. Use of appropriate low input agricultural technologies may increase production and benet smallholder farmers through increased productivity in already degraded land. A eld experiment was established to assess tree coppice intercropping of Albizia harveyi and Albizia versicolor for soil fertility and maize yield improvements in Morogoro, Tanzania. Tree fallows of A. versicolor aged three years increased signicantly soil organic Carbon, Calcium, Magnesium and Potassium. Yields of maize grain, cobs and stover in maize elds intercropped with A. versicolor were signicantly higher than those with A. harveyi. Fields with continuous maize cropping had the least yields of grain, cobs and stover. The studied agroforestry tree species are recommended for rotational woodlots and short rotation coppice systems to enhance agricultural productivity for achieving SDGs.


Introduction
Agriculture employs about 75% of the workforce and forms a backbone of many developing economies of countries in sub-Sahara Africa (SSA) [1]. It has therefore the potential to contribute greatly to achieving Sustainable Development Goals (SDGs) particularly poverty eradication, attaining zero hunger and ensuring responsible consumption and production. Adopted unanimously by all UN member states in 2015, SDGs is a collection of global goals set to ensure a better and sustainable future for all citizens by 2030 [2]. Food production in SSA has increased during the last few decades as a result of expansion of area cultivated into forests and woodlands rather than increased productivity leading to deforestation and land degradation [3]. Another concern is the fact that population growth estimated at 2.7% in 2017 has already outpaced the increase in food production. Thus, there has been a parallel decline in per capita food production by about 2% per year [4]. Available records [e.g . 5] show that since the 1970s, Africa has been a net exporter of food but in recent times around $ 35 billion was spent for annual food imports [6]. More also, currently, two-thirds of African countries are net food importers [7].
Increasing human population coupled with use of inappropriate agricultural and other land management technologies are among the root causes of the problems of declining per capita food production in the SSA [8] with negative consequence on SDGs. The escalating population has constrained traditional shifting cultivation that used to be successful in sustaining crop yields. As a result, smallholder farmers have resorted to short fallow or continuous cropping with no external inputs such as fertilizers due to high farm gate prices [8,9]. This has resulted in declining soil fertility and crop yields as well as household incomes due to soil nutrient depletion through repeated crop harvest and soil erosion [10].
Soil fertility depletion is the major cause of declining per capita food production and the concomitant widespread food shortage, low income and perpetual poverty in Africa [6,11]. This is a consequence of breakdown of traditional shifting cultivation or bush fallow system that was used successfully in the past to replenish and maintain soil fertility [12]. Shifting cultivation is currently not feasible for most farmers due to dwindling landholdings as a result of increasing population. Farmers are forced to intensify land use by reducing fallow periods to a level that is below the minimum required to maintain soil fertility or practice continuous cropping [13]. As a result, nutrients are being depleted through nutrient mining via repeated harvesting. In such a situation, the only viable options for replenishment and sustaining soil fertility appear to be application of mineral fertilizers and green manuring [14] but these are constrained too. Most of the farmers lack the required capital to use adequate and appropriate speci cation of mineral fertilizers that can counter the effect of short or no fallow on soils [10,13]. While organic manuring could be an option, but the practice is limited by bulky nature of the organic matter (OM) and unavailability [13].
Agroforestry (AF) practices have been identi ed as one of appropriate sustainable solutions to the problems of soil fertility depletion, low crop productivity and woodfuel scarcity [15,16] and can reduce wood harvesting pressure on natural forests and woodlands thereby reducing the rate of deforestation. However, despite the potential of AF practices to x Nitrogen (N), utilization of indigenous tree species in AF has received little attention [17,18,19] due to low growth rate and inadequate silvicultural knowledge. Where woodfuel scarcity is acute, crop residues and livestock manure are used as supplementary energy sources [20], which means little or no OM is available to be returned to the soil in farms. This aggravates the vicious cycle of low agricultural productivity, food shortage, low income and perpetual poverty. Therefore, alternative land use strategies to improve soil fertility and woodfuel supplies are urgently needed to sustain agricultural productivity, conserve the remaining forests and woodlands and ensure achievement of the SDGs.
With few exceptions, most AF studies on evaluation and selection of AF trees/shrubs in Tanzania have focused on exotic multipurpose tree/shrub species [15,21]. This is notwithstanding the fact that the indigenous tree/shrub species are more adapted to local environmental conditions and can meet local requirements better than exotics [19]. In addition, there is increasing realization that indigenous tree species can be more e cient in improving soil fertility and less competitive to the companion crops, especially for soil moisture [22]. This study was therefore initiated with the overall objective of assessing the potential for utilizing AF tree species (Albizia harveyi and Albizia versicolor) and management of their fallows for increasing soil nutrients and yield of maize (Zea mays). Maize was used in the study because it is the most widely cultivated cereal of great importance to food security and livelihoods in SSA.

Study area
The study area (i.e. Maseyu) is located about 50 km east of Morogoro and 150 km west of Dar es Salaam, Tanzania (Fig. 1). The area experiences tropical and sub-humid climate [23). It has a bimodal rainfall pattern with annual mean of 900 mm and seasonally distributed on wet (November to May) and dry (June to October) seasons. The mean annual temperature is 24.3°C while the minimum and maximum annual temperatures are 18.6 and 28.8°C, respectively.
Because of favourable rainfall and temperature small holder farmers practice rain-fed agriculture where maize (Zea mays L.), sorghum (Sorghum bicolour L. Moench) and garden peas (Pisum sativum L.) are mainly grown. Recently a small portion of local community members have started keeping domestic animals including cattle, sheep and goats. Charcoal production outside agricultural area takes place and contributes signi cantly to the household income. The area is surrounded by miombo woodland dominated by trees of Julbernardia globi ora (Benth.) Troupin and Combretum spp. On the northern side there is Kitulanghalo Forest Reserve, while on the southern side there is a General Land. Precambrian Usagaran metasedimentory rocks consisting of garnet biotite gneiss dominate the area. Thus, mixed alluvial and colluvial deposits tend to occur in low-lying areas [24]. The soils of the area are well drained, red, acid-neutral, sandy clay loams with brown friable top soil and are generally nutrient poor.

Experimental design and establishment
A 2 x 2 x 4 x 2 factorial experiment with three replications was established in a randomized arrangement to assess tree coppice intercropping of Albizia harveyi (Ah) and Albizia versicolor (Av) for soil fertility and maize yield improvements ( Table 1). The experiment was established using four-years plantations of Ah and Av planted separately on the same site. Prior to establishment of the experiment, the site was prepared by clearing all vegetation followed by plowing by a farm tractor and pitting using hand hoe where the pit size was 20 cm x 30 cm x 30 cm. The plantations were kept clean weeded all the time for four years prior to establishment of the cropping system experiment. Plots were rectangular (12 m x 6 m) and plots and blocks were separated by 2.5 and 3 m-unplanted buffer strips, respectively. For each study species, there were a total of 18 such plots of which 14 were used for this study.  (Table 1). Land was prepared using a hand hoe, and maize seeds of TMV-1 variety were sown for two cropping seasons at a spacing of 30 cm within row and 75 cm between rows giving a population of 44,444 maize plants per ha. Thinning was applied when coppices reached 10 cm height, which was achieved after tree cutting. Subsequent coppice thinning was maintained at the same interval of four months. During the rst coppice thinning occasion, the two tallest coppices were left per thinned stump and any coppices that sprouted were subsequently removed.

Soil fertility, maize growth and yield assessments
Just before maize sowing in each season, soil samples were taken from 4 randomly selected points in each intercropped plot at a depth of 0-15 cm using soil auger leaving 1 m border strip on both sides of each plot.
These samples were mixed thoroughly and sub-sampled to get a composite sample. The composite soil samples for each plot were transported to the laboratory for soil organic matter (SOM) and soil fertility analysis. All soil sub-samples were oven-dried (70 0 C) to constant weight and sun dried, respectively. The dried samples were then ground to pass through a 1 mm sieve and analysed for Total N (TN), extractable Phosphorus (P), Potassium (K), Magnesium (Mg), Calcium (Ca) and total soil organic Carbon (OC).
Extractable P, K, Ca and Mg were determined by simultaneous Inductive Coupled Plasma (ICP) Emission Spectroscopy technique in acid digested samples while Total N was determined by the Kjeldahl procedure.
Soil OC was determined by wet calorimetric method [25].
Maize plants were measured for diameter at 30 cm above ground and height at maturity. Height was measured to the nearest 0.01 m using a graduated pole while basal diameter at 10 cm (D10) was measured to the nearest 0.01 cm using a veneer caliper. At maturity, a total of 20 maize plants in the middle two rows were harvested, stems cut at ground level, weighed fresh and sub-sampled for dry weight determination. Also, maize cobs were harvested, shelled and both shaft and grain weighed fresh and sub-sampled for dry weight determination. Assessment of coppice growth was done at the ages of 6, 12, 18 and 24 months. During each assessment, all surviving coppices were measured for D10 and total height. The tally of height and D10 provided the number of coppice stems per stump.

Statistical analyses
Visual inspection and signi cance test were used to test for standard parametric statistical assumptions of normality and constant variance of residuals. Visual inspection was done by plotting the residuals against normal scores and predicted values while signi cance test was executed using Shapiro-Wilks's test [26]. Soil Sodium (Na) content, maize plant survival, coppice mortality and stump survival data sets violated the requirements for normality and homoscedasticity. Thus, soil Na content data were square-root transformed; and maize plant survival, coppice mortality, and stump survival data sets were arcsine transformed to correct for deviations from parametric statistical assumptions. However, only non-transformed data are presented for clarity [27]. Plot means were subjected to Analysis of Variance (ANOVA) using the General Linear Model procedure in Statistical Analysis System (SAS) at 5% level of statistical signi cance. In the study, block and block-by-treatment interactions were error terms in the model. The ANOVA for maize and soil data tested the effects of two study tree species, three cropping systems, four tree cutting heights and two coppice thinning regimes (a 2 x 3 x 4 x 2 factorial experiment) replicated three times in a RCBD (Table 1). For the tree biomass data, the analysis was carried out as a 2 x 2 x 4 x 2 factorial experiment since one level of cropping system did not have tree biomass data. All statistical analyses were done by using SAS version 8 [28].
3 Results Figure 2 shows results for soil chemical properties from tree fallows of A. harveyi and A. versicolor at three years old just before cutting to establish the experiment to test the effects of A. harveyi and A. versicolor and management of their fallows on maize yield and soil nutrients. At that age of fallow TN of top soil at a depth of 0-20 cm did not differ signi cantly (p > 0.05) from that recorded in continuous cropping plots ( Fig. 2-a).

Effects of tree fallow on soils
No test statistics (i.e., F-ratios and probabilities) for replication and replication-by-treatment interactions because these were speci ed in the error terms of the General Linear Model of SAS for testing main and interaction effects of tree species, coppice cutting height and coppice thinning.
There were signi cant effects (p < 0.05) of tree species, stump cutting height and coppice thinning on the assessed soil chemical properties (Tables 2 and 3). (0.04) †Means for each individual factor are averaged over all other treatments; §Mean of three replicates with standard error in parentheses; within each category means in the same column followed by the same letters are not statistically different at p < 0.05 according to DMRT.
Signi cant effects of tree species on soil chemical properties were detected for TN, OC and soil cations (Ca, Mg and Na). On the other hand, coppice thinning had signi cant effects on TN, OC, ECEC as well as cations (Mg, K and Na). The overlaps of the effects of these two factors are revealing but it is worth to emphasize that there were no any signi cant interactions between these two factors.
Results showed no signi cant differences (p > 0.05) between the two levels of coppice thinning for TN and OC but both recorded similar and signi cantly lower amounts of TN and OC compared to continuous cropping as well as tree fallows. The same pattern was observed for effects of thinning on amounts of Mg, K and Na.
However, the pattern was reversed for ECEC, which was signi cantly highest (p < 0.05) in thinned and unthinned coppices compared to tree fallow and maize cropping. The superiority of the tree fallows over both thinned and not thinned coppice plots with regard to soil TN and OC amounts is conceivable but that of continuous cropping calls for further elaboration.

Effects on maize growth and yield
For the rst cropping season, ANOVA revealed 2-way (stump cutting height x coppice thinning) and 3-way (tree species x stump cutting height x coppice thinning) interactions for maize diameter growth and survival respectively. During the rst cropping season of 2008, results showed signi cant 2-way interaction between stump cutting height and coppice thinning on maize plant diameter growth (p = 0.0443, Table 5), and 3-way interaction between coppice tree species, coppice stump cutting height and coppice thinning (p = 0.0075, Table 4) but they were not signi cant for yields of maize grain, cobs and stovers (Table 5). However, during the second cropping season of 2009, effects of interactions between the factors on maize growth and yield were no longer signi cant (Table 5). Table 4 Summary of ANOVA (p > F) testing the effects of tree species, coppice cutting height and coppice thinning on maize plant growth and survival for the rst and second cropping seasons at Maseyu, Morogoro, Tanzania.    (Tables 4 and 5). Figure 3 shows the main effects of coppice tree species on maize growth and yield variables for the rst and second cropping seasons.
During the rst cropping season, yields of grain (1.26 Mg ha − 1 ), cobs (0.3 Mg ha − 1 ) and stover (2.43 Mg ha − 1 ) in maize intercropped with A. versicolor were signi cantly (p < 0.05) higher compared to that of maize intercropped with A. harveyi as well as continuous cropping treatment (Fig. 3). Corresponding values for continuous cropping treatment were 0.29 Mg ha − 1 , 0.06 Mg ha − 1 and 1.06 Mg ha − 1 for yield of maize grain, cobs and stover respectively. This is equivalent to yield gain in maize grain, cobs and stover by 334.5%, 129.2% and 400% relative to continuous cropping treatment as a result of rst season intercropping with A. versicolor coppices respectively. The analogous increase in yield, relative to continuous cropping treatment, due to intercropping with A. harveyi coppices were 155.2%, 40.6% and 233%. The results indicate superiority of A. versicolor coppices over that of A. harveyi in improving yields of the intercropped maize. However, this trend was reversed in the second cropping season (Fig. 3) where, though not statistically signi cant, intercropping with coppices of any of the studied tree species tended to suppress maize growth and yield.
During the second cropping season, there was no maize grain in any of the treatments due to sporadic rainfall. Despite the fact that there were no statistically signi cant differences, coppices of both tree species tended to suppress maize growth and yield in the second cropping season and the effect was similar for both tree species. Yields of maize stover, height and diameter growth were lower in intercropped maize relative to continuous cropping by 98-98.7%, 14.8-15.3% and 46.4-81.0% respectively.
Coppice stump height and coppice thinning had no signi cant effects (p > 0.5) on growth and yield of intercropped maize during the rst cropping season, whereas their effects became signi cant (p < 0.05) in the second cropping season ( Table 5). Effects of coppice stump cutting height on growth and yield of intercropped maize for two consecutive cropping seasons are presented (Fig. 4).
Though not signi cant, during the rst cropping season, maize yields and growth tended to be higher in maize intercropped with coppices compared to continuous cropping treatment. The general trend was highest maize grain yield in coppices grown from stumps cut at the ground level. A similar pattern was observed for maize cob yields, whereas the pattern for stover yields was not clearly de ned being highest in stumps cut at 90 cm above the ground, intermediate in stumps cut at the ground level and lowest in sumps cut at 30 cm from the ground level. During the rst cropping season, maize grain yield ranged from 0.98 Mg ha − 1 for coppice stumps cut at 90 cm from the ground to 1.25 Mg ha − 1 for stumps cut at the ground level. This is in contrast to maize grain yield of 0.29 Mg ha − 1 recorded in continuous cropping treatment. These results translate into an increase of maize grain yields ranging from 237.9% for maize intercropped with coppices from stumps cut at 90 cm from the ground to 331% for stumps cut at the ground level.
During the second cropping season, growth and yield of intercropped maize were reduced compared to the rst cropping season but similar for all coppice stump height treatments; whereas they become signi cantly (p < 0.05) lower compared to continuous cropping treatment (Fig. 4). In that season, stover yields ranged from 0.02 Mg ha − 1 for both stumps cut at 30 cm and 90 cm from the ground to 0.04 Mg ha − 1 for stumps cut at the ground level. This was in comparison to stover yield of 1.04 Mg ha − 1 in continuous cropping treatment. Figure 5 shows effects of coppice thinning on growth and yield of intercropped maize for two successive cropping seasons. During the rst cropping season, maize grain yield for maize intercropped with thinned coppice treatment (0.97 Mg ha − 1 ) was slightly lower than that from no coppice thinning treatment (1.17 Mg ha − 1 ) but about twice as much as grain yield of 0.29 Mg ha − 1 from continuous cropping treatment (Fig. 5). A similar pattern was observed for cobs and stover yields as well as maize plant growth. The similarities in maize growth and yield between coppice thinning and no thinning treatments continued during the second cropping season but became signi cantly (p < 0.05) lower than continuous cropping treatment (Table 5; Fig. 5).
In the second cropping season, maize stover yield was reduced in coppice thinning and no coppice thinning treatments at exactly the same rate of about 85% to 0.3 Mg ha − 1 compared to stover yield of 2.03-2.09 Mg ha − 1 in rst cropping season. This was signi cantly (p < 0.05) lower than 1.04 Mg ha − 1 recorded in continuous cropping treatment.

Effects of tree fallow on soils
The OC content in 0-20 cm soil depth found in this study is within the range of 0.41-3.14% reported by [24] in the same area and 0.1 to 3.8% for the general miombo ecoregion [29]. Soil assessment prior to establishment of coppice experiments demonstrated the potential of AF in utilizing the tested tree species to increase OC and OM within a short period of four years. These results corroborate well with results reported by [15] for rotational woodlots in Tanzania. Other studies elsewhere found no signi cant increase in OC and OM within three to ve years of AF due to the fact that processes to increase soil OC, OM and soil fertility in general occurred slowly taking several years to detect [20]. The plausible explanation for the lack of signi cant AF trees effects on OC, OM and soil fertility in some studies could be nutrient removals associated with intensive fodder, fuel wood and poles extraction during tree fallow phase [30]. In contrast, in this study and studies by [15] and [22], AF trees were not harvested until the canopy closure. Interim intensive harvesting of fodder and fuel wood from AF systems are likely to in uence OC and OM build-up since they tend to expose the soils to high temperature leading to loss of C through oxidation [31] as well as limiting foliar mass deposits on the ground surface.
The low base (Ca, K, Mg and Na) content found in this study is a characteristic of highly weathered soils [32] typical of miombo woodlands [29]. [29] found a signi cant relationship between ECEC of the soil and the amounts of clay and OC in the top soils. Thus, the low ECEC recorded in this study corresponds well to the low amounts of OM in the area. However, this study has shown the potential of the tested tree fallows in improving both OC and OM. After one cropping season, the amount of TN and OC recorded in coppice plots were signi cantly lower compared to continuous cropping plots. The most plausible explanation for this phenomenon is C loss to the atmosphere from the coppice plots. The magnitude of changes of SOM depends on the quantity and quality of prunings, soil type, system management, climate and duration of practice of the system [33,34]. Although the quantity of prunings added in coppice plots was higher than the continuous cropping plots, it is possible that the studied tree species produce prunings that are of high quality, in terms of low carbon -to -nitrogen (C:N) and lignin-to-nitrogen (L:N) ratios. Materials of this nature are likely to have negligible or little effects on soil C build up because C is returned to atmosphere via C evolution process. This proposition is supported by [35] who found that addition of plant materials of high quality to soil led to C loss rather than accumulation. This was probably aggravated by soil exposure to high temperatures when coppices were still young leading to oxidation [31] and hence loss of C to the atmosphere. This can also serve to explain the high amounts of TN and OC in the tree fallow plots associated with differences in microclimate compared to coppiced plots. However, it is important to note that the quality of prunings from the studied tree species was not assessed due to unforeseen budget constraint, thus this aspect calls for further investigation.
Lack of signi cant effects of tree coppice cutting height and coppice thinning treatments on soil chemical properties is probably due to the fact that foliar biomass as a result of these treatments were on the lower side. Improvement in soil nutrient status in AF is mainly through nutrients released from mineralization of prunings [16,36]. According to [11], nutrient contributions from AF systems are positively correlated to the amount of prunings added to the soil. In this study, foliar biomass ranged from 2.18 Mg ha -1 in A. versicolor to 2.86 Mg ha -1 in A. harveyi, which is lower compared to other AF studies that reported improved soil chemical properties. [22] reported improved soil chemical properties in AF system in which foliage yield ranged from 6.3 to 20.2 Mg ha -1 . Besides yields of prunings, the effects of AF system on soil chemical properties can be in uenced by nutrient contents and overall quality of the prunings such as C: N ratio that affects mineralization of nutrients [10,37]. This study did not assess these factors thus they require further investigation.

Effects on maize growth and yield
This study has demonstrated an increase in maize grain yield of intercropped maize of up to 100% during the rst cropping season as compared to continuous cropping. This could be attributed to fertility improvement as a result of AF tree coppices related to various mechanisms such as biological N xation, pumping up or retrieval of nutrients from lower soil horizons and interception of nutrients that would otherwise be lost through leaching and surface runoff and release of nutrients during litter and root decomposition [9,34,36].
Probably the increased maize yield in intercropped maize could be attributed to these mechanisms.
Signi cant reduction in maize stover yields for maize intercropped with coppices could be attributed to combination of competition for light as result of shading, and competition for water resulting from developed tree root system [22]. These results are consistent with other simultaneous AF system studies [15,16,21]. In a semiarid area of Tanzania, [22] reported progressive reduction of yields of maize intercropped with Australian Acacias despite the fact that trees improved soil chemical properties. In Kenya, [38] reported decreased yield of maize intercropped with Grevillea robusta after three years of intercropping. Similar results have been reported by [39] in Kenya. It is important to note that the pattern of the effects of tree age on the nature and magnitude of competitiveness of AF trees varies with planting density (widely spaced trees taking longer to reduce crop yields) and is in uenced by climate and species of crop involved [34,40,41]. These aspects were not investigated in the present study, thus require further investigation.

Conclusion
Intercropping coppices of the studied trees with maize increased maize grain yield by 100% compared to continuous cropping treatment in the rst cropping season suggesting soil amelioration effects of these tree species and the potential to boost livelihoods and achieve SDGs. However, in the second-year reduction in yields of maize intercropped with coppices of the studied tree species was observed indicating increasing competitive effects of the coppices of these tree species with age. Based on their positive effects on soil fertility, the studied tree species are recommended for on-farm planting for soil improvement. Planting these tree species in sequential AF systems such as improved fallow, intensive pruning of coppices or wider spacing in intercropping systems may reduce their competitive effects on the companion crops. The effects of AF system on soil chemical properties can be in uenced by nutrient contents and overall quality of the prunings such as C: N ratio that affects mineralization of nutrients. Thus, these aspects require further investigation.

Declaration of competing interest
There are no con icts of interest    for tree species and n = 3 for continuous cropping).

Figure 5
The effects of coppice thinning on yields of maize grain (a), stover (b), cob (c), maize plant height growth (d), diameter growth (e) and survival for the rst and second cropping seasons at Maseyu, Morogoro, Tanzania.
Treatments were TH0 = no coppice thinning, TH1 = coppice thinned to leave two coppice stems per stump, TH∞ = Tree fallow and CC = continuous cropping. For each gure and within each cropping season, means marked by the same letter are not statistically different at p < 0.05 according to DMRT. Vertical bars indicate standard errors of means (n = 7 for tree species and n = 3 for continuous cropping).