i. Clustering of soils in the Mwea Irrigation Scheme (MIS) using principal component analysis
In this study, principal component analysis was used to identify important soil properties that could describe variation in the MIS. Principal components (PC) 1 ─ 4 accounted for over 72% of the variation in soil chemical properties in the MIS (Table 2). All the twelve soil properties that were analyzed, except magnesium, were significantly correlated with either PC1 or 2 or both (Table 3). This suggests that most of the variation within the MIS for the measured soil properties can be attributed to the first two principal components. PC1 had a positive and significant correlation with soil pH, total organic carbon, macro nutrients (nitrogen, phosphorus and potassium), and exchangeable cations (Ca2+ and Na+) (Table 3). This suggests that these parameters vary together such that when one category increases the other would tend to also increase (Da Piedade et al. 2019). A scatter plot for PC1 and 2 showed that the Mwea (cluster MW), Tebere (cluster TB), Wamumu (cluster WU) sections were clearly distinct while Karaba and Thiba (cluster KT) appeared to overlap and hence were grouped together (Fig. 5). Moreover, significant differences were found among clusters in all the soil properties evaluated except for sodium (Tables 4 and 5). This indicates principal component analysis was adequate in the clustering of the sections of the MIS.
Variation in chemical properties among clusters and recommendations for agronomic management
a. Soil pH and total organic carbon
Soil pH had a coefficient of variation value of 8.5% (Table 4) indicating that very low variability exists for soil pH in MIS. These results are in agreement with findings by Kundu et al. (2016) who reported that variability for soil pH in MIS was low with CV values of 12% for water and 15% for KCI. Principal component analysis showed that soil pH had the largest positive effect on PC1 and therefore considered the key contributor of the variation found in soil chemical properties in this study (Fig. 3 and Table 3). The optimum soil pH for lowland rice ranges between 5.5 and 7.0 (IIagan et al. 2014). In clusters TB, WU and KT, the soil pH tended to be acidic in some areas and optimum in others while in cluster MW, which was mainly comprised of units in the Mwea section, pH values were below the optimum levels for the growth of rice under flooded field conditions (Table 4). Evidence in literature indicates that fertilizer management practices could have an impact on soil pH levels. In irrigated lowland rice systems urea and ammonium sulphate are the two common sources of nitrogen (Fageria et al. 2003). Findings from a survey conducted in the Mwea irrigation scheme showed that about 85% of the farmers applied ammonium sulphate during the first and second top-dressing (Onderi and Danga 2022). In studies conducted to evaluate the effect of the two sources of nitrogen on soil pH, ammonium sulphate was found to be more acidifying than urea (Fageria et al. 2010). Hence the low pH particularly in cluster MW (Table 4), where rice cultivation has been practised since 1954 (NIA 2023), could partly be attributed to continuous use of highly acidifying sources of nitrogen for rice cultivation for a prolonged period of time. Low soil pH is caused by free release of hydrogen, aluminum, manganese, and iron to toxic levels (Das 1996). Soil acidity hampers plant growth primarily by causing the inhibition of root development as well as root retardation (Rout et al. 2021). Application of lime in acids soils is one of the strategies that are commonly used to change soil pH and also improve soil fertility (Reddy and Subramanian 2016). However, excessive addition of lime can cause a significant decrease in the availability of micronutrients, especially iron, zinc and copper leading to deficiencies of those plant nutrients (Haynes 1982). The application of biochar as an alternative method of soil amendment has been reported to confer additional advantageous effects to acidic soils such as creating a carbon sink to mitigate global warming, increasing soil water holding capacity, reducing greenhouse gas emissions and stabilizing mobile heavy metals, and other organic pollutants in soil (Lehmann et al. 2006; Van Zwieten et al. 2010; Inyang et al. 2011; Abdul-Halim et al. 2018; Berek et al. 2018). In this aspect, the use of carbonated rice husk as biochar could be low-cost option in the MIS considering that the product is a freely available by-product from a number of mills nearby the scheme and is mostly found dumped along the road or openly burnt in the fields within the scheme (Bogonko 2020; Ndirangu and Oyange 2019).
Organic carbon plays a key role in determining the physical, chemical, and biological properties of soil (Yang et al. 2004; Reeves et al. 1997). In this study the coefficient of variation for total organic carbon was 13.3% indicating very low variability for organic carbon in the MIS (Table 4). The critical level for total organic carbon is 2% (Musinguzi et al. 2013). In the four clusters the total organic carbon was greater than the critical levels recording a mean of 3% (Table 4). Nonetheless, variation was found among the clusters with cluster TB followed by MW having higher levels that the other two clusters. Land topography is one of the several factors that have been reported to impact levels of soil organic carbon (Hao and Kravchenko 2007; Wang et al. 2012). In this study, total organic carbon was higher in upstream clusters MW and TB than WU and KT that are located downstream (Table 4 and Fig. 1). Due to the close proximity to the water intake, the upstream clusters seldom experience water shortages and hence soils here tend to be under flooded water conditions during the rice growing season (Mohammed et al. 2003). Findings from previous studies indicate that decomposition of soil organic matter tends to be low under anaerobic than aerobic paddy soils (Katoh 2003). This suggests that the higher total organic carbon in clusters TB and MW than in WU and KT could be as result of differences in water management practices among clusters. Still, there is need to take precaution to avoid deficiency particularly in clusters situated downstream of the MIS. Application of farmyard manure is one of the most cost effective ways of improving organic carbon in the soil (Yang et al. 2004).
b. Macro-nutrients
In rice-based systems, nitrogen is typically the most limiting nutrient to crop productivity. In the absence of inorganic fertilizer application, total nitrogen accumulation depends of indigenous sources such as soil, irrigation water, crop residues and manure application (Nguyen et al. 2008). Variation for total nitrogen amounts was very low and above the critical level of 0.2% (Olaleye et al. 2009) in all the four clusters (Table 4). Nonetheless, total nitrogen tended to be higher in upstream clusters of MW and TB than in clusters downstream the MIS. Again, in contrast to upstream clusters, downstream clusters are frequently exposed to cycles of dry and wet soil conditions that are more pronounced during peak irrigation water requirement (Mohammed et al. 2003). The inorganic fertilizers that farmers mainly use in the MIS release mineral nitrogen in the form of NH4+, which is stable in flooded anaerobic environments. The nitrogen is usually applied in three splits whereby about 30% of the total nitrogen is applied as basal with the remaining amount applied in two equal splits at the active tillering and panicle initiation stage (Oyange et al. 2019). Soil aeration due to draining of water cause changes in nitrogen from NH4+ to NO3-; the latter nitrogen form is unstable under flooded anaerobic conditions and can quickly be lost through denitrification (Ishii et al. 2011). The lower total nitrogen levels in clusters downstream of the MIS (Table 4) could be attributed to nitrogen losses caused by fluctuations in soil water conditions leading to the formation and loss of unstable forms of nitrogen. Adequate supply of water particularly during the timing of fertilizer application could mitigate against such nitrogen losses and also enhance nitrogen use efficiency.
Phosphorus availability for rice is influenced to a greater extent by soil pH dynamics as compared to other nutrients (Rakotoson 2022). Soil pH levels < 5.5 limit phosphorus availability due to fixation by iron and aluminium while those > 7.0 cause deficiency for the rice crop as a result of the phosphorus being bound to calcium. In this study, soil phosphorus in cluster MW was substantially lower than that of the other three clusters (Table 4). In contrast, Fe levels were the highest in cluster MW. These findings suggest that the low pH levels in cluster MW resulted in most of phosphorus to be fixed to iron (Abou-Seeda et al. 2020). When using the Mehlich III procedure for the determination of available phosphorus, as was the case in this study, phosphorus levels can be categorized as low (< 7 mg/kg), medium (7 ─ 15mg/kg) and high (> 15 mg/kg) (Nwilene et al. 2000). Consequently, the recommended phosphorus amounts are 30 ─ 60, 15 ─ 30, and 0 ─ 15 P2O5 kg/ha for low, moderate, and high phosphorus levels, respectively. Based on this classification the soil phosphorus levels for cluster MW and cluster WU were categorized as moderate while those in clusters TB and KT were high (Table 4). According to Oyange et al. (2019) phosphorus application rate of 60 kg P2O5 ha-1 is recommended for rice cultivation in the Mwea irrigation scheme. This shows that there is excess application of phosphorus across the scheme and there is need to review the current rates for sustainable rice production to avoid causing detrimental impact to the environment.
Potassium is an essential nutrient in several plant processes that have a direct influence on the quality of seed such as protein synthesis, carbon assimilation, photosynthesis, and enzyme activation (Marschner 2012). The extractable potassium level in the soil has to be above 0.2 cmol+/kg of soil for rice to obtain adequate amounts to sustain plant growth processes (Olaleye et al. 2009). The findings of this study indicate that there was potassium deficiency across all the clusters in the scheme, but this was more severe in clusters MW, WU, and KT (Table 4). Use of inadequate potassium fertilizer amounts coupled with removal of rice straw after harvest in the MIS have been cited as key factors that have contributed to the widespread deficiency (Kundu et al. 2016). Findings from previous research show that 10 kg K ha-1 is required per ton grain rice harvested (Dobermann and Fairhurst 2000). Within the MIS, farmers cultivate different varieties with different yield potentials (Samejima et al. 2020). Thus, the potassium application amounts need to be adjusted to meet the specific crop requirements. Moreover, for efficient utilization of applied potassium split application at early tillering and at panicle initiation have been reported to enhance rice yields than single dose basal application (Manzoor et al. 2008). Rice straw contains 1.4 to 2.0% K2O (Dobermann and Fairhurst 2002), hence long term potassium management strategies within the scheme should also include incorporation of rice straw to the paddy fields.
c. Micronutrients
Micro nutrients play an important role in the growth and development of rice (Kundu et al. 2019). Principal component analysis revealed that among micro nutrients zinc had the largest loading values for the first principal component (Fig. 3 and Table 3). This indicates that zinc is the most important soil micro nutrient in the MIS. Similar findings have been reported in studies conducted in lowland rice growing areas that found zinc to be the most wide spread nutrient disorder after nitrogen and phosphorus deficiency (Quijano-Guerta et al. 2002; Singh et al. 2003). Zinc concentration of 2.0 mg/kg (i.e. 2 ppm) 0.1N HCl is considered the critical level for deficiency in rice production (Dobermann and Fairhurst 2000; Fairhurst et al. 2007). The findings of this study showed that zinc levels in cluster WU (i.e. units mainly in the Wamumu section of the Mwea irrigation scheme) were below the critical level while those in clusters TB and KT were only slightly above deficiency levels (Table 5). In zinc deficient soils, an application rate of 25 kg/ha in the form of zinc sulphate before flooding or after transplanting has been recommended in irrigated lowland rice (Dobermann and Fairhurst 2000; Kalala et al. 2017). However, at present, only a few farmers in the MIS use zinc fertilizers because these have not been included in the recommendation package for rice cultivation (Onderi and Danga 2022). On the other hand, zinc absorption in rice has also been reported to be enhanced at sufficient nitrogen and phosphorus application (Erenoglu et al. 2011; Lal et al. 2000). A number of previous studies have reported the functional role of nitrogen application in promoting lateral root growth and development in both lowland and upland rice ecosystems (Tran et al. 2014; Menge et al. 2019). Moreover, Saneoka et al. (1990) showed that plant roots were distributed in deeper soil layers at higher than lower phosphorus levels. These findings indicate that the use of appropriate zinc fertilizers particularly in clusters TB and KT, should be coupled with proper management of nitrogen and phosphorus application could be an efficient strategy to alleviate these deficiencies in the MIS.
In cluster MW zinc levels were more than threefold of those in the other three clusters (Table 5). Qi (1987) reported that the available zinc in soils correlated negatively with soil pH. In the MIS, few incidences of continuous water logging due to poor drainage infrastructure has been reported especially in places with black cotton soils that have high clay content (Wamari et al. 2016). Perennial flooding are among the key factors that are often associated with low soil pH (Qadar 2002; Quijano-Guerta et al. 2002). The high zinc levels in cluster MW could be attributed to acidic soil conditions prevalent in this cluster (Table 4). However, these high zinc levels in cluster MW may not be beneficial to rice at low soil pH levels due to occurrence of aluminium and iron toxicity that affect root development and hence nutrient uptake (Rout et al. 2021).
Iron is an important constituent of porphyrins and ferredoxins, both of which are critical components of the light phase of photosynthesis (Dobermann and Fairhurst 2000). For normal growth and function of rice, soil iron levels of 2─300 mg/kg are required (Dobermann and Fairhurst 2000). Cluster MW had significantly higher iron levels than those found in the other three clusters (Table 5). Moreover, the iron levels in cluster MW were above the recommended range (Table 7).
Table 7 Key for classifying the various soil parameters evaluated in the Mwea irrigation scheme.
Category
|
Parameter
|
|
Critical level
|
Reference
|
|
pH
|
|
5.5 ─ 7.0
|
Ilagan et al., 2014
|
|
ToC
|
<
|
2%
|
Musinguzi et al., 2013
|
|
|
|
|
|
Macro-nutrients
|
TN
|
|
0.2%
|
Olaleye et al., 2009
|
|
P
|
|
7 mg/kg
|
(Nwilene et al., 2000)
|
|
K
|
>
|
0.2 cmol+/kg
|
Olaleye et al., 2009
|
|
|
|
|
|
Micro-nutrients
|
Cu
|
|
0.1 mg/kg
|
Dobermann and Fairhurst, 2000
|
|
Fe
|
|
2 ─ 300 mg/kg
|
Dobermann and Fairhurst, 2000
|
|
Zn
|
|
2 mg/kg
|
Dobermann &Fairhurst, 2000; Fairhurst et al., 2007
|
|
Mn
|
|
3-30 mg / kg
|
Dobermann and Fairhurst, 2000
|
|
|
|
|
|
Exchangeable cations
|
Ca
|
|
1 cmol+/kg
|
Dobermann and Fairhurst, 2000
|
|
Mg
|
|
3 cmol+/kg
|
Dobermann and Fairhurst, 2000
|
|
|
|
|
|
Salinity
|
SAR
|
|
13
|
Richards, 1954
|
TOC: Total organic carbon, TN: total nitrogen, SAR: sodium adsorption ratio
This suggests that there could be nutritional toxicity as a result of excessive iron amounts with the potential of limiting rice growth and yield in cluster MW. Accumulation of iron to toxic levels is influenced by low pH and continuous water logging. Adoption of alternate wet and drying water saving technology that is currently being disseminated in the scheme (RiceMAPP 2016); and use tolerant rice varieties are among the strategies that can be adopted in the affected zone to mitigate the effects of iron toxicity.
The rice crop removes approximately 8 grams of copper for the production of one ton of rough rice including straw (Choudhury et al. 2009). Copper deficiency in the soil increases sterility in rice grain resulting in a decrease in the yield (Ambak and Tadano 1991). Soil copper levels were above the critical levels in all the clusters within the MIS (Tables 5 and 7). Copper levels in clusters TB and KT tended to be lower than those in clusters MW and WU (Table 5). Conversely, the former two clusters were found to have high levels of available phosphorus (Table 4). There is accumulated evidence in the literature of antagonist effects between copper and phosphorus in soil and plant tissues (Zhang et al. 2020). This suggests that the high phosphorus levels in clusters TB and KT, in part, negatively affected the availability of copper in the soils of the MIS. Farmers should be advised to apply phosphorus fertilizers based on soil test results to avoid the accumulation of this nutrient to levels detrimental to availability of other nutrients. Although there were varying levels of manganese among the clusters, no deficiency was detected across the irrigation scheme.
d. Exchangeable cations
There were significant variations among clusters on the levels of magnesium and calcium ions (Table 5). Regardless of cluster however, the concentrations of these ions were higher than the critical levels i.e. 3 and 1 cmol+/kg for magnesium and calcium respectively (Dobbermann and Fairhurst 2000). These findings are in agreement with those reported by Kundu et al. (2016) who reported that deficiency of these nutrients is rare in irrigated lowland rice. There were no significant differences in sodium concentration in the soils among the four clusters (Table 5). The criterion for assessing the impact of different sodium levels in the soil on plant growth is generally based on the sodium adsorption ration (SAR). In soils with SAR values higher than 13, sodium is the main cation contributing to saline conditions (Richards 1954). In the Mwea irrigation scheme, the SAR values were significantly lower than the critical levels suggesting that the sodium amounts in these soils are less likely to contribute to occurrence of salinity based on the current soil status.
iii. Correlation among soil properties
Among the parameters analyzed only soil pH, phosphorus, potassium, iron and Zn amounts showed wide variations extending to levels not suitable for rice cultivation (Tables 4, 5 and 7). Therefore, we focused on correlation between these five and other soil parameters evaluated in this study. Correlation analysis revealed that there was a significant negative correlation between soil pH and Fe (Table 6). This suggests that the low pH in MW could partly be attributed to high concentration of Fe mainly in the form of Fe3+. Frequent cycles of wet and dry soil conditions have been reported to cause a reduction of Fe3+ and thus favoring rice growth (Dobermann 2004). A shift in water management practices from the conventional continuous flooding to alternate wetting and drying (AWD) is a key strategy in ameliorating Fe3+ accumulation to toxic levels in the MIS.
In this study, we found a positive significant correlation between phosphorus and Ca2+ (Table 6). In soils with pH ≥ 7 applied phosphorus forms phosphate complexes with Ca2+ and hence becomes unavailable to the rice plant (Abou-Seeda et al. 2020). Based on the current soil pH levels all the clusters in the MIS had pH less than 7 and hence there is no immediate risk of phosphorus deficiency as a result of this nutrient being bound to Ca2+. However, in a survey conducted by Onderi and Danga (2022) to access the most commonly used fertilizers in the MIS, it was reported that few farmers applied calcium rich fertilizers such as calcium ammonium nitrate (CAN). This coupled with long periods of water shortage particularly in KT and WM (Mohammed et al. 2003) could cause increases in soil pH and Ca2+ levels limiting P availability. In contrast, zinc showed negative correlation with calcium (Table 6). Although we found Ca2+ to be within normal ranges (Table 5), the significant negative correlation with zinc could suggest presence of few isolated cases where Ca2+ levels were high to the extent that bioavailability of zinc was hampered. Since the soil environment is highly heterogeneous as a result of different farmer practices, site specific soil testing would be able to identify such soil and give appropriate recommendations. Soil potassium showed a positive and significant relationship with total nitrogen, total organic carbon, phosphorus and calcium (Table 6). The correlation coefficient for the relationship between potassium and nitrogen (0.59) was stronger than phosphorus (0.27) (Table 6). This suggests that factors that affect potassium availability would also affect nitrogen to a greater extent than phosphorus. We have emphasized before that rice straw removal is likely to have serious implications of soil health status in the MIS. Dobernmann and Fairhurst (2002), reported that for every 1 ton of rice straw removed results in the export of 5 – 8 kg N, 14 – 20 kgs of K2O and 1.6 – 2.7 kgs of P. This indicates that the continuous straw removal in the MIS is likely to have a greater impact on nitrogen and potassium in the short term and phosphorus in the long term. Mitigation measures could include among others the application of organic manure. This is supported by the strong relationship that was found between total organic carbon and total nitrogen (1.00), and total organic carbon and potassium (0.60) (Table 6).