Chapter 3
Crop productivity assessment -
results at national level

This chapter illustrates the suitability assessment and calculation of rainfed production potential with an example, followed by tabulated results of crop productivity assessment

3.1 Patting It All Together - An Example

We pick a record from the land resource inventory, an agro-ecological cell in Meru district, and look at maize productivity. The cell data record reads:

3 6 11113224 33034 0 2 35 00 000     1150
-+-+-+-+-+--+--++-+-+-+--+-++-+++-------+

The line underneath the data record indicates the individual fields, the cross marking the end of each field. The record contains the following information:

A few remarks may be helpful: Mapping unit Pn1 belongs to the landform of non-dissected erosional plains. In the legend of the soil map it is described as: ‘well drained, very deep, dark reddish brown to dusky red, friable clay; in places bouldery (nito-rhodic FERRALSOLS)’. The land extent under consideration falls into thermal zone 1, i.e. a mean daily temperature >25 Centigrade applies, corresponding to an altitude below 800 m.

In the mapping unit composition table there is only one entry for mapping unit Pn1, i.e. only one soil type (nito-rhodic Ferralsols) is identified, texture and slope class apply to the whole unit. No phase is indicated.

The attached slope class code is 2, i.e. slope class AB, representing slopes in the range of 0 – 5%. According to the slope composition table, the mapping unit must be split into two entries, half the cell relating to a slope range of 0 – 2%, the other half relating to a slope range 2 – 5%. The land resource inventory record that we have chosen refers to the latter with an average slope gradient of 3.5%.

The inventoried mean total dominant length of growing period for the cell, located in the north-east of Mount Kenya, is LGP code 11, i.e. sufficient moisture supply for a total growing period of 270 – 299 days, quite favorable conditions.

LGP-Pattern code 13 indicates that there are usually two (shorter) growing periods, in seven out of ten years according to historical profiles, and one growing period in about 1/3 of the years. The reference table relating the mean total dominant LGP to the corresponding mean total associated LGPs (see Technical Annex 1.1) lists the following for the bimodal case: the first associated LGP, LGP₂₁ with code 7, is 150 – 179 days, the second associated LGP, LGP₂₂ with code 5, lasts 90 – 119 days.

The list of crop types considered in the AEZ assessment for Kenya (see Appendix A), includes nine maize crop types, with crop sequence numbers 4 to 12. Maize types with crop codes 031 to 036, i.e. crop sequence numbers 7 to 12, are highland maize types adapted to cooler conditions and have therefore been rated as not suitable in the thermal zone suitability screen. This leaves us with three lowland maize types, crop codes 021 to 023, crop sequence numbers 4 to 6. Agronomically attainable yield potentials for the relevant LGPs (in kg dry weight per hectare), i.e. on suitable soils and terrain, at intermediate level of inputs, and the corresponding crop cycle lengths are as follows:

The maximum unconstrained yields are 2.6, 3.5 and 4.5 t/ha, respectively. For all three crop types, the highest yields can be achieved in the LGP with 150 to 179 days (LGP code 7). In shorter LGPs water stress constrains attainable yields, in longer LGPs pests, weeds and workability constraints are assumed to increasingly limit attainable maize yields. From the LGP-Pattern weights we derive average attainable yields of 2.1 t/ha (min 1.0 t/ha, max 2.6 t/ha) for the shortest maize type (code 021), 2.9 t/ha (min 1.4 t/ha, max 3.5 t/ha) for the second maize type (code 022), and 3.6 t/ha (min 1.7 t/ha, max 4.5 t/ha) for the third lowland maize type (code 023).

The soil unit rating of nito-rhodic Ferralsols for maize is S2, like with most Ferralsols, i.e. suitable with some limitations depressing yields by 25 percent. The clay texture does not affect the rating. Combining agro-climatic and agro-edaphic suitability rating we arrive at average yields of 1.7, 2.4 and 3.0 t/ha respectively. The modest average slope gradient of 3.5 % passes the slope-cultivation association screen which tolerates dry land crops on terrain with slope gradients of up to 30 %.

In a relatively long growing season, as we are considering here, additional yields from multicropping must be considered. The intercropping increment, as explained earlier, depends on the level of inputs, the length of the growing period and the overall crop suitability. At the intermediate level of inputs, with moisture availability well above 120 days the intercropping increment for maize is estimated at around 10 percent, i.e. a LER (land equivalent ratio) of 1.1.

The length of the growing period allows also for two or even three crops to be grown each year. Since we consider only lowland maize in this example, the optimal sequential crop combination with two crops grown in sequence involves crop code 023, the maize type with the longest cycle, as the first crop, followed by the medium maize type, crop code 022, with a growth cycle of 90 to 110 days. The maize crop with the shortest growth cycle, lowland maize crop code 021, requires 80 days on average (growth cycle 70 – 90 days). Assuming a 10 day turn-around time, three sequential crops can be grown within 270 days. In the dominantly bimodal LGP-Pattern (code 13), two maize crops (crop codes 022 and 021) could be grown in the first growing season, 150 – 179 days, a third maize crop (code 021) in the shorter second growing season of 90 – 119 days. Table 3.1 summarizes the best performing sequential maize crop combinations in the given cell, as also for LGPs of 240 – 269 and 210 – 239 days, when one, two or three crops are grown.The results have to be understood as an illustration of the crop combination algorithm rather than a likely or even viable option. In fact, the crop combination requirement constraints imposed to avoid continuous mono-cropping would normally rule out the sequential cropping patterns described in Table 3.1. For the quantification of maximum potential crop production, including multiple mono-cropping, as in the results presented in Appendix C and D, these combinations would however be acceptable.

Fallow requirements to maintain soil fertility and ensure sustainable production, under given conditions and input level, are set at 25 percent, i.e. 5 out of 20 years the land would not be permitted to be under maize cultivation.

TABLE 3.1
Maize production potential (t/ha) - an example

The soil erosion hazard is quantified by means of a modified Universal Soil Loss Equation (USLE), involving estimates of rain erosivity, soil erodibility, slope effect (slope length factor), crop cover factor and protective management practices. An average growing period of 285 days implies a mean average rainfall of about 1550 mm, resulting in an estimated rainfall erosivity factor of 600. Nito-rhodic Ferralsols with clay texture are considered to be of low erodibility. Hence, they are grouped into the second (out of seven) soil erodibility class with an average soil erodibility factor of 0.11. With a slope length factor of 0.8 and a crop cover factor of 0.4, adjusted to 0.36 for an assumed crop/fallow cycle - the fallow requirement is 25 percent) - finally results in an estimated annual soil loss of 19 t/ha/year, corresponding to about 1.6 mm topsoil loss per year. The regeneration capacity of topsoil, a function of thermal zone and length of growing period, stipulates an annual addition of topsoil of 1.7 mm, making up for the estimated erosion losses. Therefore, maize productivity is not assumed to be adversely influenced by water erosion in the given agro-ecological cell.

3.2 National Level Results

The calculation procedures outlined above have been performed for all crop types, pastures and fuelwood species in all agro-ecological cells of Kenya. As a result, the extent and quality of arable land and production potential of each crop type was determined in all locations. The information was then aggregated to district level, province level and national total. The results are also aggregated over crop types to indicate production potential of crop species, e.g. arable land and production potential for maize rather than nine individual maize types.

Three sets of results are presented. The first set, Assumption Set A, refers to single crop suitability and production potential on all land that is not indicated as forest zone, game park, or as belonging to an irrigation scheme.

The second round of computations, Assumption Set B, looks at productivity allowing sequential (mono) crop combinations, where feasible. Also, crop types can be combined, e.g. a first maize crop with a growth cycle of 110–130 days and a second maize crop of 70–90 days. The resulting crop production is included with the crop type that comes first in the crop combination, the primary crop.

Finally, in Assumption Set C, it has been attempted to determine the utmost crop potential by including all rainfed land in the assessment, regardless of any other land use indicated in the land resources inventory.

Six suitability classes have been defined, relating average single-crop suitability in a cell to maximum attainable yield. The classes C1 to C5 relate to average attainable yields of >80 %, 60–80%, 40–60%, 20–40%, and 5–20% compared to maximum yields. Note, that extents in suitability class C5 are usually not considered among the viable crop options, but have been included here to indicate the scope of production in very marginal areas. A sixth class accounts for areas that are entirely unsuitable or allow for only less than 5 percent of maximum yield. Data for the non-suitable class are not explicitly included in the result tables. Production potential is calculated from land extents in suitability classes C1 to C4 only. Average, minimum and maximum production potential and yields are determined according to the LGP-Pattern and associated probabilities. Multiple land use in time (sequential mono-cropping) is indicated by means of a multicropping index (MCI).

Each table shows an estimate of arable land by productivity class. The algorithm used to determine arable extents in an agro-ecological cell comprises of a two stage procedure. In a first step, the crop or (mono) crop combination that performs best under worst climatic conditions is determined. Then, all crop combinations which meet the production stability constraint, i.e. fall within a tolerable neighborhood of the best performing crop, are considered in the final selection. Finally, among all qualifying crops, the crop type that maximizes the weighted sum of extents in land suitability classes C1 to C4 is selected as describing the cell's arable land potential. Suitable extents of the primary crop type in the chosen crop combination, i.e. the first crop to be grown in the sequential cropping pattern, are recorded in the relevant totals of arable land resources.

The estimates of arable land have been grouped according to four broad climatic zones: the arid zone, areas inventoried with mean total dominant length of growing periods less than 120 days, the semi-arid zone, comprising of areas with LGPs of 120 to 179 days, the sub-humid zone, referring to LGPs in the range of 180 to 270 days, and a humid zone, where growing periods exceed 270 days. For information, the sum total of all land resources inventory records processed is also included in the table. Total arable land is then expressed as percentage of total zone area, as a relative measure of quality of land in the zone.

From the coded information contained in a land inventory record it is possible to identify the respective grid raster points in the GIS to which an entry relates. Because of the disaggregation implied by the mapping unit composition table and the slope composition table, usually more than one land inventory record will refer to the same set of grid points. Therefore, the results to be transferred to the GIS must be aggregated to average values per raster point. Aggregation of crop suitability assessment by agro-ecological cell to GIS raster points (100 ha each) has produced the results shown in Figure 3.1, for crops, Figure 3.2, for pastures, and Figure 3.3, for fuelwood species. Similar maps can be produced for each crop type in the study.

FIGURE 3.1
Generalized land suitability for rainfed crop production

FIGURE 3.2
Generalized land suitability for rainfed pastures

FIGURE 3.3
Generalized land suitability for rainfed fuelwood production

The national results in table form, for each of the three assumption sets (A, B and C), are shown on the following pages, Table 3.2 to 3.4. A breakdown of the same kind of information by province and at district level can be found in Appendixes B to D.

Permitting single crops only and setting aside land marked in the resources inventory as forest zone and/or game park, assumption set A, the estimated arable land resources in Kenya, suitability classes C1 to C4, amount to some 7 million ha. 2.3 million ha are adjudged very good or good potential, classes C1 and C2, 2.1 million ha are rated moderately suitable. The balance, around 2.7 million ha, is of low potential. Arable extents in the very marginal category, class C5, amount to 4 million ha (this would be in addition to the 7 million ha in suitability classes C1 to C4. Hence, the arable extents in classes C1 to C4 account for only little more than 12 percent of Kenya's total land area. In the sub-humid and humid zone more than SO percent of the land is suitable for rainfed crop production, about 1/4 in the semi-arid zone, and only less than 2 percent in the arid areas.

In terms of individual crop species, the largest potentially arable extents are as follows: maize 4.0 million ha, sorghum 3.9 million ha, millet 3.6 million ha, barley 3.0 million ha, and beans 2.9 million ha. Arable extents of coffee and tea are estimated at 1.6 and 2.S million ha, respectively, of which 0.7 and 0.9 million ha are rated as marginal.

When agro-ecological cells are included in the assessment which are indicated as belonging to a forest zone and/or game park, as in assumption set C, the estimate of total arable land increases to almost 8 million ha, a 12 percent increase. Much of this increase is in the humid and sub-humid zone. For example, arable land suitable for rainfed maize cultivation increases to 4.4 million ha. More than 2/3 of all the land in the (small though) humid zone is estimated to be suitable for crops. For details of crop specific performance, attainable production levels and variability of yields, the reader is referred to the tabulated results, below, containing information by crop species as also crop type.

TABLE 3.2
Results of crop productivity assessment - assumption set A

NATIONAL TOTAL: KENYA (Assumption Set A)

Arable Land by Productivity Classes (100 ha):

Potential Crop Production:

Potential Crop Production by Crop Type:

TABLE 3.3
Results of crop productivity assessment - assumption set B

NATIONAL TOTAL: KENYA (Assumption Set B)

Arable Land by Productivity Classes (100 ha):

Potential Crop Production:

TABLE 3.4
Results of crop productivity assessment - assumption set C

NATIONAL TOTAL: KENYA (Assumption Set C)

Arable Land by Productivity Classes (100 ha):