Assessment of soil fertility and potato crop nutrient status in central and eastern highlands of Kenya

Inherent low soil fertility remains a hindrance to potato production in Kenya and continues to pose a threat to food security. A study was conducted in Nyandarua and Meru counties to assess the soil fertility status in smallholder potato farms. Soil and plant tissue samples were collected and analysed for selected nutrients (pH, OC, N, P, K, S, Ca, Mg, Zn, B and Cu) from 198 farms. Critical nutrient levels were used to assess the sufficiency levels of nutrients for potato growth. Soils in the sampled farms were weakly to strongly acidic (pH-CaCl2 3.9–6.6) and had low to high soil organic matter content (1.5–97.5 g Kg−1). The percent of farms in Meru and Nyandarua with nutrient contents below critical levels were 66% and 20% for N, 46% and 85% for P, 67% and 31% for S, 9% and 51% for Cu, and 87% and 80% for B, respectively. Low tissue nutrient concentrations were observed for N, P, K, and S irrespective of the sites. Soil pH correlated strongly with majority of the analyzed soil and tissue nutrients. These results affirm the need to design integrative soil fertility management strategies to correct the impoverished soil fertility status in the study area.


Methodology
Sampling area. This study was conducted in Meru and Nyandarua counties of Kenya (Fig. 1). The two counties are representative of the major potato growing areas of eastern and central Kenya. Sampling in each county was done within a two-kilometer wide transect. The transect in Meru was layed along coordinate range of 0°07′ N, 37°48′ E to 0° 13′ N, 37°53′ E, and along 0°38′ S, 36°50′ E to 0°37′ S, 36°45′ E in Nyandarua. The Meru transect cut across agro-ecological zones; Upper highlands (UH3) (2100-2450 meters above sea level (masl)) and Lower highlands (LH4) (1850-2000 masl) whereas in Nyandarua it covered UH2 (2000-2150 masl), UH3 (2100-2450 masl) and LH3 (1900-2000 masl) 34 . Nitisols is the predominant soils type in Meru while Nyandarua is dominated by Andosols and Planosols 34 . Farms were selected purposively from a list of farmers generated with the assistance of the area agricultural extension officers. The overall sampling targeted farms with potatoes that had attained mid flowering stage as this is the time when peak nutrient uptake by potato occurs. A total of 100 and 98 farms were sampled in Meru and Nyandarua counties, respectively. Soil and tissue sampling. Soils were sampled from each target farms in zigzag pattern and at an interval of approximately 10 m. On average, ten (10) soil replicates were taken within rows and inter rows of each farm with a 15 mm diameter soil auger at 0-30 cm depth. The samples were mixed into a composite for each farm and a sub sample of 100 g fresh weight was taken to the laboratory and frozen at 4 o C until analysis. Extraction of soil samples for analysis of Ca, P, K, Mg, B, Zn, and Cu was done using Mehlich 1 procedures 35 and determined using inductively coupled plasma-optical emission spectrometry (ICP-OES) 36 . Calcium chloride (0.0125 M) was used to extract soil mineral N and S 37 .
Tissue samples were extracted at mid flowering stage from the youngest fully expanded potato leaves (4 th leaf from the top of the growing tip). This leaf was considered the critical part for tissue nutrient analyses because younger or older tissue have varying nutrient concentrations which can lead to erroneous interpretations 16 . All leaflets were put into khaki paper bags embedded with silica gel to absorb the moisture. Foliar samples were oven dried at 60 °C, ground using Wiley mill, wet digested, and analysed for mineral nitrogen calorimetrically using Skalar, and for Ca, P, S, K, Mg, B, Zn, S and Cu using ICP-OES after dissolving the ashes (550 °C, 4 h) in dilute HCl 36,38 .
In addition to plant and soil samples, information on the fertilizer type, amounts and manure used was collected using structured questionnaire in order to determine soil nutrient replenishment of the sampled farms. Data analysis. Descriptive statistical analyses including means, range and standard deviations were used to explore the data using STATA software and graphs plotted in MS Excel. Nutrient sufficiency ranges (low, adequate and excess) ( Table 1) were used to group the farms depending on respective nutrients levels 24,27 . The proposed leaf nutrient sufficiency were adopted from elsewhere since no documented limits exist for potato in Kenya thus are used for general interpretation while critical soil nutrient levels were extracted from a manual published by  Soil chemical characteristics and tissue nutrient content. In Nyandarua, the soil pH (CaCl 2 ) ranged from 3.9 to 6.2, with an overall mean of 5.0 ± 0.5, with 86% of sampled farms having values below the critical level of 5.5 (Table 2). Similar results were observed in Meru which showed a soil pH range of 4.2-6.6 with 55% of the sampled farms showing pH values below the critical levels for potato production. The SOC contents ranged between 1.4-91.5 g kg −1 in Meru and 19.7-65.2 g kg −1 in Nyandarua, and were below the critical levels for potato growth in 9% of sampled farms in Nyandarua and 30% of sampled farms in Meru. Total N varied widely between the sampled farms, both in Meru and Nyandarua, with 67% and 20% of the sampled farms recording values below critical levels, respectively. Soil P showed the highest variations between the sampled farms in Meru and Nyandarua with 47% and 85% of the sampled farms having this element at concentrations below the critical levels. Sulphur was below critical levels in 68% and 32% of sampled farms in Meru and Nyandarua respectively. Over 80% of the sampled farms in both regions had B below the critical levels. Zinc content showed similarity in Meru  www.nature.com/scientificreports www.nature.com/scientificreports/ (5.1 ± 3.5) and Nyandarua (5.6 ± 5.3). Calcium and Mg were generally optimal in Meru and Nyandarua. Ca-Mg ratio was higher in Meru (3.63 ± 0.88) compared to Nyandarua (1.98 ± 0.77).
Tissue N concentrations exhibited wide variations between the sampled farms with a mean of 40.27 ± 5.32 g kg −1 in Meru and 47.39 ± 4.98 g kg −1 in Nyandarua (Table 3). About 73% of the sampled farms had N below optimal levels in Meru compared to 23% in Nyandarua. Potassium variability was similarly high between the sampled farms averaging 47.33 ± 10.92 g kg −1 in Meru and 51.64 ± 12.52 g kg −1 in Nyandarua. The sampled farms with K levels below the optimal levels were 22% and 15% respectively in Meru and Nyandarua. Zinc concentrations varied between 18.3-78.9 mg kg −1 and 22.5-101.8 mg kg −1 which is within the optimal levels for potato growth. Boron and Cu similarly showed wide variations between the sampled farms averaging 31.45 ± 3.08 mg kg −1 and 23.70 ± 8.99 mg kg −1 , and 9.79 ± 3.08 and 11.62 ± 4.46 mg kg −1 respectively in Meru and Nyandarua. The sampled farms with B levels below critical limits were 17% and 55% respectively in Meru and Nyandarua.
Relationship between soil chemical properties and plant nutrient content. Tissue phosphorus was significantly positively correlated to pH (r = 0.37), soil P and negatively correlated to soil Cu (r = −0.40) ( Table 4). A significant positive correlation between tissue Ca and soil N (r = 0.49) and tissue and soil Ca (r = 0.49) was observed. Similarly, tissue K concentrations and soil K contents (r = 0.32) correlated positively. Soil Cu concentrations correlated significantly and negatively with tissue P (r = −0.40) and N (r = −0.40).
In Nyandarua, N uptake correlated negatively with concentration of other nutrients in the soil ( Table 5).    Principal soil nutrients influencing soil fertility. Factor analysis for soil chemical properties in Meru retained 7 principal factors with the first three factors accounting for 91% proportion of the variance. Zinc and S were the most unique soil chemical properties in Meru ( Table 7). The first factor was mainly weighted by soil pH, Ca, Mg and B whereas the second one was due to N and SOC. The Third factor was influenced by K and B. Soil Cu had a negative influence on the first factor (soil pH). Factor analysis results for the soil chemical properties in Nyandarua showed that the first three factors explained 90% of the variance. All the soil chemical properties (except Cu) which had major influence on the first factor in Meru had greater influence in the first factor in Nyandarua as well. The second factor in Nyandarua was dominated by SOC and N. Zinc and S were unique chemical properties in Nyandarua. In addition, Cu was found to be unique in Nyandarua.

Discussion
The differences in soil chemical properties between Meru and Nyandarua were due to the differences in soil types and fertilizer and manure use. The farms examined in Meru were mainly dominated by Nitisols, which are acidic with soil organic matter content ranging from low to high 39 . Soils in the sampled farms in Nyandarua were dominated by Planosols which are formed from volcanic ash and are regarded as degraded with the surface soil being acidic 40 . Planosols due to their imperfect drainage and waterlogging, have poor soil organic carbon build-up capacity compared to the well-drained Nitisols in Meru. Organically bound nutrients, especially N and P are not available to plants because it cannot be absorbed into root cells without first being released from the organic molecule through mineralization 41 . This process is regulated by soil micro-organisms that work at optimal aeration that is not effectively provided in Planosols. Differences in soil nutrient levels and therefore tissue nutrient concentrations between Meru and Nyandarua suggest differences in cropping systems. Where the previous cropping system has caused a depletion of soil organic matter, the soils are more likely to be acidic with limited capacity to hold N, P, K, Ca and some essential micronutrients. This coupled with the low fertilizer and manure use, meant increased soil degradation. Long term fertilization regime and manure use affects nutrient concentration, SOM  Table 5. Correlations coefficients (r) between pH, soil chemical properties and plant tissue nutrients content in Nyandarua region. *Level of significant P < 0.05. and microbial life in the soil 42 . It can therefore be concluded that the farms in the two study areas had a slight nutrient differences. The low soil pH in both Meru and Nyandarua relates to the high rainfall amounts that probably caused cation leaching. Oxidation of DAP fertilizer commonly used by the farmers in these areas results in the formation of strong inorganic acids such as nitric acid which further lower the soil pH 10 . The volcanic parent material also adds up to the low acidity as the two soil types in the study area are described as acidic soils 39 . Low soil pH in turn affects the uptake of most macro and secondary plant nutrients by either its effects on microbial activity and dissolution of Al/Fe ions. Nitrogen, K and S are less affected by pH but P is greatly affected, while micro nutrients are mostly available in slightly acidic soil 43 .
The relatively low soil N content in the sampled farms in Meru was reflected in the tissue N levels and could be attributed to the low soil pH and coupled with the low soil organic carbon contents. Soil organic matter retains soil N and prevents it from leaching beyond the active rooting zones 3,44 . Microbial activity that releases N from organic matter and certain fertilizers are particularly affected by the low soil pH since microbial activities occurs best at soil pH range of 5.5 to 7.0 45 . In a study conducted to compare the yield response of potatoes to increasing levels of N, a larger response was exhibited in soils with higher soil organic matter content irrespective of the amount of N applied, and the recovery of the applied N was greater where the soil pH was slightly above 6.0 46 . The large number of farms with N nutrient below optimum levels could thus be as a result of nutrient leaching, low amount of N replenishment and high nutrient mining as a result of continuous farming 3,47 . According to Westerman 16 , potato takes up to 235 kg N ha −1 , which is mined out of the soil in a growing season.
The low soil P content was reflected in tissue P content suggesting reduced uptake. This observation could be ascribed to the predominating clay soils in these study areas 39 as well as to the low soil pH. Both Nitisol and  Table 7. Factor loading for the first 3 factors and unique variances of soil chemical properties in Meru and Nyandarua.
Planosols examined in this study have clay content >30% 39 . Clay soils and acidic soils have high Al and Fe contents which besides fixing the available soil P are associated with increased soil acidity, thus leading to inconsistent response to soil P by potato leaf uptake 48,49 . The observed low soil phosphorus could be associated with the leaching of P, especially when the soil's sorption capacity is saturated 50 . There is recent evidence that higher P concentrations are found in the soil water moving in the bypass flow pores within the agricultural farms 51,52 . Poor P recovery through fertilizer application under acidic conditions is due to the fact that the P applied in the form of fertilizers is mainly adsorbed by the soil 48 . Furthermore, P is largely transported offsite attached to the sediment to be later released via dissolution or made available when anoxic conditions are present 13 . The reduction of transport would also mean reduced uptake. Soil K in Meru was adequate however this was not the case with the tissue K concertation. The differences between soil K and its uptake in Meru could be as a result of cation balances. A study has shown Mg induced K deficiency on Nitisol 53,54 . Soil moisture also affects the uptake of K since it helps in mass flow movement 55 . Lower soil moisture in Nitisol would thus mean reduced K uptake. Further to this, low soil pH indirectly affects K uptake. At low soil pH, Al becomes soluble thus dominating CEC hence lowering the soil capacity to hold K 56 . Potassium is of high importance in potato growing as it affects tuber yield and qualities and it is the highest absorbed nutrient 16 .
Leaf S concentration was appreciably lower than the critical concentration. This was an indication that S deficiency is a problem in the potato growing farms in Meru and Nyandarua. This observation is consistent with other findings that have established that potatoes do not respond to sulphur applications except in extremely deficient soils 57,58 . If soil test S is less than 7 ppm and/or tissue S is less than 0.18%, then 20 kg S per hectare should be banded at planting 8,27,59 . Factors contributing to this increased incidence of S deficiency among the examined farms may be related to the use of S-free fertilizers such as urea and di-ammonium phosphate 9,60 . This is despite the fact that potato tubers remove a high amount of S; typically 1 tonne of potato tubers will remove 4.5 kg of S 8 . The low soil organic matter could also lead to low S as 95% of S is known to be associated with organic matter 8 .
Boron and Cu were generally limiting in the soils tested, an observation which was reflected in the potato tissues. This observation could be asserted to the limited supply of these nutrients in the soil and to the overall low soil organic matter contents of the sampled soils 61 . Low soil pH would also be a contributing factor as a result of Al solubility that interferes with uptake and transport of other nutrients 8 . The high level of B deficiency in the sampled farms in Meru could be associated with nutrient leaching coupled with low soil pH which hinders B uptake by potato 62 . Elsewhere, B deficiency has been related to the low soil organic matter contents, especially under prevailing cold wet weather and in periods of drought 8,61 .
The significant positive correlation between leaf P and soil P relates to the low P ions in the soil solution at the root surface. This is in accordance with Morgan & Connolly 63 who observed that potato responds to nutrient deficiency by changing its root structure so as to increase the overall nutrient acquisition. Burke 8 similarly argued that the ability of potato crop to absorb P will depend on the concentration of P ions in the soil solution at the root surface. The significant interaction between tissue P concentrations and soil P contents indicates that this element was limiting in majority of the sampled farms. Soil pH and tissue P content similarly showed significant associations implying their interactive effects. These relations could be greatly attributed to the low content of active P forms in the highly weathered clay dominating soils in this study sites 39 , since in acidic soils, P may be adsorbed by Fe or Al oxides, and various clay minerals 3 .
The significant relationship between tissue K concentrations and soil K contents reflects soil K deficiency in several of the sampled farms. Though the tropical soils are generally considered to be sufficient in K, deficiency of this element was evident in this study. Potatoes take up more K than many other arable crops 16,64 . During peak vegetative growth, potatoes may require 10 kg K 2 O ha −1 per day from the soil 65 . At about 80 days after emergence high yielding potato crop may remove more than 500 kg K 2 O ha −1 8 . At harvest, more than 75% of the K uptake is found in the tubers, which typically contain around 5.8 kg K 2 O per tonne of tubers 8 . As the potato crop is harvested and the tubers are removed from the field, so K is taken away in that crop material. This must be replaced otherwise future crops will be grown in soil with a reduced K level, resulting in low yields.
Significant negative relations were exhibited between tissue Cu nutrient content and soil N, P, Ca, B, and Zn indicating that tissue Cu concentrations seems to level off at certain concentrations of these elements. Similar results have been reported elsewhere 66 . This would suggest that Cu toxicity is becoming a greater concern in Nitisols and Planosols. The significant correlation between tissue Cu concentrations and soil Cu content however indicated that Cu was a limiting nutrient requiring corrective measures. The positive association between tissue magnesium and soil Cu content showed a synergy between these two nutrient elements. This would suggest that Mg uptake by potato increases with increasing level of Cu concentrations in the soil. The positive correlation between tissue Ca and soil N, S, and B would mean Ca uptake by potato is more efficient in soils with optimum N, S, Ca and B levels. The role of Ca and B on cell wall formation and auxin transport thus explains the correlation 67 .
Soil pH is the single most factor affecting soil nutrients due to its effects on microbial activity and nutrients dissolution 8 . High concentrations of Ca 2+ in the soil solution increases the soil pH which reduces the solubility of Al and Fe ions and enhances solubility and availability of certain micro-nutrient elements. Boron levels are associated with soil organic matter thus the low soil organic matter could be the reason for B correlations with other soil nutrients. This implies that the trend of soil nutrient elements depletion is similar and that with continuous mining, other nutrients will become limiting. The negative effect of Cu levels to most of other soil nutrients has been reported in other studies 68,69 . Copper has a negative effect on soil bacteria which would in turn affect other soil nutrients concentrations 70 .
Factor analysis results showed that soil pH and CEC related to soil chemical properties (K, Ca, Mg) influenced the first factor. The higher the bases (K, Ca, Mg) the higher the soil pH. This would thus reinforce that soil pH is a major factor to consider for sustainable nutrient management 61 . The second factor affecting soil fertility in the two regions was soil organic matter. Soil organic matter retains nutrients especially N, P and S. Soil organic Scientific RepoRtS | (2020) 10:7779 | https://doi.org/10.1038/s41598-020-64036-x www.nature.com/scientificreports www.nature.com/scientificreports/ matter is subjected to change by cropping system and management practices and its influence on soil fertility is thus important 8,61 . Management of the soil pH and soil organic matter factors would thus improve soil fertility in the studied region.
The low soil nutrient contents measured for N, P, K, S and B were reflected in the tissue nutrient concentrations which were consistently lower indicating that nutrient uptake by potato crop was influenced by soil nutrient concentrations. The correlations were however not very strong which could be explained by other confounding factors influencing nutrient uptake such as crop variety, farm management, climatic factors and crop health status 16,71,72 . Similar relations have elsewhere been reported 73,74 . Nutrient interaction in the soil and within the plant also affects crop nutrient uptake 66,75 . conclusions This study shows that N, P, K, S and B are the key limiting nutrients to potato production in the major potato growing areas of Kenya. These nutrients need to be integrated in nutrient management programs for major potato growing areas of Kenya. Soil pH was found to be low in all the sampled farms and was the major factor that influenced nutrient uptake by potato crop. Improved cropping systems and soil management practices are required to adjust the soil pH for enhanced soil fertility in the study areas.

Data availability
Data is available from corresponding author.