3.1 Socio-demographic profile of farmers interviewed
Table 2 shows the socio-demographic characteristics of the farmers who participated in the on-farm survey. Across all the study communities, most of the farmers had no formal education. Although males dominated farming, we purposively sampled female farmers, as shown by the high level of female respondents across all study communities. Regarding the age of respondents, in Kpalgun and Daboashie communities, 62% of the total respondents were between ages 20 and 50, whereas the remaining 38% were aged above 50years. Also, in Fihini, 63% of the respondents were aged between 20 and 50years, while 37% were above 50years. Also, in Zagua, 72% of the respondents were aged between 20 and 50years, whereas 28% were above 50years. In Yoggu, 71% of the respondents were aged between 20 and 50years, while 29% were above 50years. Cheshegu, on the other hand, recorded the highest (86%) of respondents between 20 and 50 years, whereas the remaining 14% were specifically aged above 60years.
Table 2
Demographic Profile of Respondents
Characteristics | Kpalgun | Daboashie | Cheshegu | Fihini | Zagua | Yoggu |
Household Characteristics | | | | | | |
No. Household | 111 | 32 | 26 | 38 | 48 | 216 |
No. Sampled Farmers from unique households | 56 | 17 | 14 | 20 | 25 | 115 |
Age of respondents | | | | | | |
20–50 years | 62 | 62 | 86 | 63 | 72 | 71 |
Above 50 years | 38 | 38 | 14 | 37 | 28 | 29 |
Gender of respondent | | | | | | |
Male (%) | 53.6 | 47.1 | 57.1 | 50.0 | 48.0 | 47.8 |
Female (%) | 46.4 | 52.9 | 42.9 | 50.0 | 52.0 | 52.2 |
Education Level | | | | | | |
Proportion Without Formal education (%) | 80.4 | 94.1 | 78.6 | 85.0 | 84.0 | 92.2 |
Proportion with Formal education (%) | 19.6 | 5.9 | 21.4 | 15.0 | 16.0 | 7.8 |
Primary Occupations | | | | | | |
Farming (%) | 98 | 88 | 100 | 100 | 100 | 99 |
Farming & Livestock (%) | 2 | 13 | - | - | - | 1 |
Source: Author’s field survey, 2015 |
3.2 Land use and land-cover distribution
We assessed the land use and land-cover distribution of the communities, focusing on agricultural land use (ALU) distribution and crop types. This was done in two main ways. First, we combined the satellite data and ground control points to understand the spatial distribution of ALU. Our findings showed nine (9) land use/landcover classes for the six (6) communities covering a total land size of over 2,440 ha (Fig. 4 and Table 3). Across the communities, Yoggu occupied the highest land area of around 800 ha, whereas Cheshegu (203 ha) had the lowest land area. Due to its large land area, Yoggu recorded the highest land use/landcover areas of all the nine classes of land use studied relative to the other communities. Of the 2,440 ha of land area, maize occupied the most extensive land coverage of about 16%, followed by the built-up/open land category (about 15%) and then yam and pepper farms occupying about 14% each. Waterbody was the lowest with less than a percentage coverage (Fig. 4 and Table 2).
In terms of relative coverages by percentages of the land use/landcover across the six communities, Dabogshei recorded the highest maize and pepper acreages of about 29% and 22%, respectively, while Zagua was observed to have recorded the lowest for these two crops with roughly 9% and 6% respectively (Table 3). On the contrary, Zagua has the largest land use/landcover areas for built-up/open land (29.4%) and rice cultivation (15.4%), whereas Dabogshei was seen to have the lowest coverages for both classes (Table 3). Land use activities for cowpea were highest in Dabogshei (13.1%) and lowest in Fihini (6.1%). Cheshegu recorded the most extensive area for groundnut with around 15% while being lowest for Yoggu and Fihini of about 8% each. Fihini had the largest yam (15.7%) land use while Dabogshei recorded the lowest with about 12%. Also, the fallow field's highest coverage was about 17% for Yoggu and around 4% least fallow area for Dabogshei. Effectively, this could mean that more lands in Yoggu were not in use during the assessment period compared to the other communities. Our analysis also showed that waterbodies were present only in Fihini (0.5%) and Yoggu (0.4%). In northern Ghana, water bodies are very important for household and on-farm activities, including support to dry-season farming.
Table 3
Area of Land Uses in the Study Communities 2015 (Unit: Ha; %)
Areas in ha |
Landuse | Fihini | Dabogshei | Kpalgun | Yoggu | Cheshegu | Zagua | Total (Ha) |
Maize | 77.31 (21.41) | 69.48 (28.56) | 66.87 (13.93) | 118.35 (14.49) | 37.17 (18.31) | 29.43 (8.83) | 398.61 |
Fallow Field | 48.78 (13.51) | 10.53 (4.33) | 67.14 (13.45) | 138.33 (17.29) | 22.23 (10.95) | 22.05 (6.61) | 309.06 |
Pepper | 51.57 (14.28) | 54.27 (22.31) | 55.8 (11.18) | 118.62 (14.82) | 30.78 (15.17) | 19.71 (5.91) | 330.75 |
Rice | 45 (12.46) | 9.54 (3.92) | 32.76 (6.56) | 63.54 (7.94) | 22.32 (11.00) | 51.3 (15.38) | 224.46 |
Groundnut | 30.33 (8.40) | 22.95 (9.43) | 56.7 (11.36) | 67.23 (8.40) | 29.7 (14.63) | 30.69 (9.20) | 237.6 |
Cowpea | 22.05 (6.11) | 31.77 (13.06) | 58.23 (11.66) | 70.56 (8.82) | 18.72 (9.22) | 33.75 (10.12) | 235.08 |
Waterbody | 1.89 (0.52) | 0.00 (0.00) | 0.00 (0.00) | 3.15 (0.39) | 0.00 (0.00) | 0.00 (0.00) | 5.04 |
Built Up / Openland | 27.45 (7.60) | 16.2 (6.66) | 99.81 (19.99) | 106.83 (13.35) | 15.48 (7.63) | 98.1 (29.42) | 363.87 |
Yam | 56.7 (15.70) | 28.53 (11.73) | 61.92 (12.40) | 113.67 (14.20) | 26.55 (13.08) | 48.42 (14.52) | 335.79 |
Total | 361.08 (100.00) | 243.27 (100.00) | 499.23 (100.00) | 800.28 (100.00) | 202.95 (100.00) | 333.45 (100.00) | |
Crops cultivated in the communities have been reported in several studies (see Boakye-Danquah et L., 2014; Antwi et al., 2018 a, b). The most dominant crops cultivated in the communities are maize, rice, yam, groundnut, pepper, soya bean, cotton, millet, cowpea, tobacco, sorghum, and cassava. Among these, maize is the most dominant crop. In practice, most farmlands are intercropped with the major crops (e.g., maize, rice, yam) combined with minor crops (cowpea, tobacco, sorghum, and cassava). Cereal and legumes are the most rotated crops. Earlier studies have shown that in the Tolon area, crop rotation is a common soil conservation method because of its ease of integrating into the farming system (Okorley et al., 2002). Alternating legumes with cereals observed through the landscape monitoring at the community level is important for nitrogen fertilization in smallholder cultivation systems where organic amendments are low (Bationo et al., 2006).
Although crops traditionally grown in the communities have not changed for several decades, findings from a field survey showed that hybrid crop varieties, especially maize, peanuts, cowpea, and rice, are becoming common in the study communities. Agricultural marketing companies, research units, and extension officers from the main channels through which these crop varieties are introduced to the communities. Such new crop varieties have early maturity, more resistant against droughts and diseases, and yields are better (Quaye, 2008). However, challenges such as perceived better taste for local varieties, seasonal seed purchase, and higher use of fertilizer, insecticides, and pesticides for new crop varieties have affected the adoption of new crop varieties (FAO, 1997).
3.3 Farm management practices
Farm management is the process by which resources and situations are manipulated by the farm manager in trying, with less than full information, to achieve their goals (FAO, 1997) which is often that of maximum production returns. Earlier field work conducted by the authors (Antwi et al., 2018 a,b) observed farm management practices in the communities often involves two main activities: land preparation prior to cultivation and soil and land management. Land preparation typically begins after the harvest of the previous crop, that is if the land was cultivated in the previous year. Farmers normally begin by clearing and gathering remaining crop residues (if any) as well as cutting re-emerging shrubs and weeds, which are later burnt under control. Across the communities, controlled burning of weeds is a widespread practice for land preparation. Where farm residues that could not decompose from the previous harvest before the next farming season is burned. In such small-scale farming systems, burning is an inexpensive and time-saving way of controlling weeds, insects, diseases, and excess crop residue. However, burning can deprive the soil of its protective layer against erosion, reduces the amount of organic matter received through mulching, and destroys soil organisms that play vital roles in the formation of soil structure and composition (Ringius, 2002).
Besides burning, ploughing is also an important land preparation practice which allows soils to be easily worked within dry conditions where soil compaction is high (Farage et al., 2003). In the communities, most farmers rent mechanical loughs from wealthy farmers or use cow-powered ploughs, where possible. We also observed a widespread use of weedicides by some farmers as part of the land preparation process. The use of weedicides can increase the acidity or alkalinity of the soil and destroy micro-organisms if it is not applied properly. Moreover, for some farmers, land preparation can commence even a year before planting begins depending on the resources available to the farmer. On the field, we observed that some farmer had kept their cattle on their fallow land to feed and add manure to the land before land preparation begins the following season. However, most of the farmers we talked to indicated that fallow periods have consistently declined – the maximum number of years a farm can be allowed to fallow is two years. Similar observations on the reduction of fallow periods in northern Ghana have been reported in other studies (Boakye-Danquah et al.,2014; Songsore, 2011; Quaye, 2008).
After land preparation, farm management practices commonly adopted by farmers are diverse including composting, animal manure, chemical fertilizer, turning weeds under, mechanized ploughing, animal traction and crop rotation. Across the communities, the use of tractor ploughing was most widespread farm management practice, followed by chemical fertilizer application and crop rotation. Most farmers indicated they practice crop rotation to preserve and restore the fertility of the soils. Reza (2016) notes that crop rotation that efficiently combines a mix of nutrient-fixing crops and crops with different root structures improves the physical and chemical condition of soils leading to the overall improvement in the fertility of the soil (Reza, 2016). The use of organic fertilizers, namely animal manure was also a very common farm management practice although its use is limited due to its low availability, competing uses with other livelihood activities, and bulkiness in transporting to farms. It is important to emphasize that manure and chemical fertilizer (e.g., NPK 15 15 15, Ammonium sulphate, and urea) are mostly combined or alternated in most maize farms.
The application of chemical fertilizer to boost production is essential as soils in the northern savanna under continuous cropping, and inappropriate farm management practices have declined in fertility (Bationo et al., 2018). Although there are recommended amounts per acre or hector, our findings showed that farmers are often unable to use the recommended amounts of fertilizers or were often not applied due to financial constraints, unavailability of the product, and lack of access to government-subsidized fertilizers. Across the communities, farmers complained about the high cost of chemical fertilizer. In a related study (Arthur, 2014) found that availability and application of chemical fertilizers are dependants on factors such as basic price factors, risk aversion, and price control.
3.3 Crop and Land Suitability Determining Crop Suitability
Before assessing the land suitability index, the various relationship between the yield of each crop and soil properties was examined using the GAM approach. Table 4 lists the model variable and p-value.
Table 4
List predictor variables that gave significant value for each crop type
Soil properties | Maize | Rice | Pepper | Yam | Groundnut | Cowpea |
p-value | p-value | p-value | p-value | p-value | p-value |
s(Clay) | 0.0581 | 0.3633 | 0.0484* | 0.6095 | 0.0000* | 0.1614 |
s(EC) | 0.0974 | 0.9940 | 0.4677 | 0.0550 | 0.0145* | 0.4936 |
s(N) | 0.0026* | 0.0049* | 0.0035* | 0.0124* | 0.1218 | 0.0003* |
s(NDWI) | 0.1156 | 0.2562 | 0.2933 | 0.2747 | 0.0000* | 0.0087 |
s(OC) | 0.0005* | 0.0192* | 0.8133 | 0.0423* | 0.3961 | 0.8437 |
s(OM) | 0.0005* | 0.0072* | 0.0410* | 0.0720 | 0.1931 | 0.0000* |
s(PH) | 0.5408 | 0.0000* | 0.0006* | 0.0551 | 0.1283 | 0.0120* |
s(Sand) | 0.0025* | 0.0042* | 0.0007* | 0.8187 | 0.0000* | 0.4318 |
s(Silt) | 0.0001* | 0.1265 | 0.1031 | 0.0662 | 0.0010* | 0.4332 |
s(Slope) | 0.1123 | 0.5020 | 0.4364 | 0.4143 | 0.4007 | 0.8097 |
The predictor variables were significant when the P-value is less than 0.05. Based on Table 4, only some of the soil properties in the study area had a considerable impact on the yield of each food crop. For example, the soil properties that significantly impact maize were nitrogen, organic carbon, organic matter, sand, and silt. Therefore, we only used those predictor variables to estimate the optimum concentration and calculate the land suitability index. Combining the significant variable was considered the best GAM model for specific food crops; then, the optimum concentration can be evaluated. To avoid subjectivity while combining several layers of soil properties, the PCA approach was used. Table 5 shows the optimum concentration and weight in each soil property. Based on Tables 6 and 7, we can deduce that each crop type has its preference for soil properties. Thus, identifying the land suitability index for specific food crops was extremely important.
Table 5
Optimum concentration of significant soil properties and its weight.
Crop type | Soil properties | Optimum concentration | Weight |
Maize | s(N) | 0.01–0.2 | 0.23 |
s(OC) | > 0.54 | 0.28 |
s(OM) | < 0.85 | 0.28 |
s(Sand) | 45–65 | 0.25 |
s(Silt) | < 36 | 0.24 |
Rice | s(N) | < 0.082 | 0.29 |
s(OC) | > 0.77 | 0.35 |
s(OM) | > 0.9 | 0.35 |
s(PH) | > 5.37 | 0.22 |
s(Sand) | > 58 | 0.31 |
Pepper | s(Clay) | > 6.2.25 | 0.12 |
s(N) | 0.01–0.16 | 0.24 |
s(OM) | OM < 1.3 | 0.29 |
s(PH) | 5.5–6.8 | 0.14 |
s(Sand) | > 59 | 0.16 |
Yam | s(N) | < 0.075 | 0.27 |
s(OC) | > 0.5 | 0.28 |
Groundnut | s(Clay) | < 8 | 0.22 |
s(EC) | 45–300 | 0.19 |
s(NDWI) | 0.2–0.6 | 0.08 |
s(Sand) | < 63.5 | 0.21 |
s(Silt) | < 34 | 0.20 |
Cowpea | s(N) | 0.052–0.08 | 0.24 |
s(NDWI) | < 0.39 | 0.06 |
s(OM) | 1-1.75 | 0.28 |
s(PH) | > 5.8 | 0.14 |
The land suitable index maps for the six communities are shown in Fig. 5–10. On the map green represents not suitable and red represents highly suitable for a particular food crop. In the Kpalgun community (Fig. 5), pepper was the most suitable commodity, followed by maize and rice. More than two-thirds of the community land. Particularly from the central portions to the north was highly suitable for pepper. However, cowpea and yam were the least suitable crop in Kpalgun.
In Fihini, maize and pepper were the most suitable crops covering over 90% of the landscape. Rice was highly suitable in the southernmost part of the community, where there are river valleys. However, most of the community land was least suitable for rice, groundnut, and cowpea.
The cultivated lands in Zagua were suitable for pepper, rice, and maize in descending order. The central portions of the landscape were moderately suitable for groundnut and yam. However, cowpea was the least suitable crop for Zagua.
In Cheshegu, the land was most suitable for rice, yam, and maize, but not for cowpea, while crops like maize, pepper, and groundnut thrived better on lands in Yoggu, but not rice, yam, and cowpea.
In Daboashie, the land was suitable for rice, pepper, yam, and maize but not for cowpea.
Based on previous research by (Avornyo et al., 2014), Fihini and Yoggu had the least number of households cultivating rice within the six communities in northern Ghana. This finding corresponds with the outcome of our model assessment, where agricultural lands in Fihinni and Yoggu were not suitable for rice. In general, all the study areas were not suitable for cowpea but highly suitable for maize and pepper. Thus, based on the ALU distribution in 2015 (Table 3), maize occupied most agricultural lands; this agrees with our land suitability index map.
3.4 Model Validation
Qualitative approach was used for model validation. We employed stratified random sampling to decide the sample size and data for each stratification. Out of 5013 samples, we randomly selected 485 samples point of rice, 2841 samples point of maize, 1055 samples point of groundnut, 164 samples point of yam, 274 samples point of pepper, and 194 sample points of cowpea (Table 6).
Table 6
Contingency table for rice, maize, groundnut, yam, pepper and cowpea
Crop Type | In-situ data | Model |
Yes | No |
Rice N = 485 | Yes | 454 | 31 |
No | 0 | 0 |
Maize N = 2841 | Yes | 2444 | 176 |
No | 204 | 17 |
Groundnut N = 1055 | Yes | 969 | 55 |
No | 20 | 11 |
Yam N = 164 | Yes | 150 | 3 |
No | 11 | 0 |
Pepper N = 274 | Yes | 201 | 35 |
No | 38 | 0 |
Cowpea | Yes | 120 | 67 |
N = 194 | No | 7 | 0 |
The data were stratified according to the proportion of crop type in the study area. We employed the quantitative method for validation which was based on the contingency table shown in Table 6. Based on Table 1, POD, FAR, and Bias was calculated, and the results are described in Table 7.
Table 7
Crop Type | POD | FAR | BIAS |
(%) | (%) |
Rice | 100 | 6.4 | 1.07 |
Maize | 93 | 7.7 | 0.99 |
Groundnut | 98 | 3 | 1.04 |
Yam | 93.17 | 20.26 | 0.95 |
Pepper | 84.1 | 13.13 | 0.99 |
Cowpea | 94.5 | 16.6 | 1.5 |
For example, the value of POD, FAR, Bias of Maize are 93%, 7.7%, and 0.99, respectively. It means 93% of the maize crop model areas were correctly identified, 7.7% of the model's predictions proved to be incorrect and the model has a very high similarity with the observation data. Overall, the POD, FAR, and Bias for rice, maize, groundnut, yam, pepper and cowpea type ranges were 84–100%, 6–21%, and 0.95-15, respectively. The above result implies that our model could be considered excellent for predicting land suitability for each crop.
3.5 Interventions towards Sustainable Land Utilization and Crop Production
Land suitability assessment is critical to optimizing crop yield and ensuring sustainable land management (Mazahreh, 1998). In the small-scale farming systems of northern Ghana, where soil fertility decline and land degradation are a significant challenge to food crop production (Songsore, 2011), suitability assessment of existing agricultural land use has become important. This study is the first to provide knowledge on crop-soil suitability to support adaptation to land degradation programs in northern Ghana.
By comparing the current cropping patterns with the soil and crop suitability model across six communities, our findings revealed that most farmers (in five of the six communities) across the landscape are cultivating the crops that are less suitable to the soil types identified in the communities; an outcome that necessitated the community validation and demonstration workshop. During the workshop, most farmers were surprised at the discrepancy between what the model showed and what they practiced on the farm. While some of the participants said they could consider the model in the farm decisions in the future when given more information, others thought a lot more goes into the decision (e.g., household food needs, market demands, cost of planting, etc.) on what to plant other than just the soil characteristics.
According to Sharifi, 2003, although agricultural decisions are evaluated based on social, environmental, and economic factors, economic viability often dominates these considerations. Thus, the trend observed in the study communities could also be due to economic and policy considerations that make cultivation of certain crops preferred even though available land suitability information may not support the crops grown. Cultivating crops on less suitable soils may lead to reduced yields and thus require more inputs for maximum returns. A study in East Africa with similar climatic characteristics as northern Ghana by (Fermont et al., 2010) showed that low soil fertility limited cassava response to N, P, or K fertilizers. Therefore, it is not surprising that most farmers now apply chemical fertilizer as a key farm management practice. The high use of chemical fertilizer is probable due to the low soil fertility and the quest for quick results. Government agricultural policies and programs should develop low-cost and effective means of helping farmers with information on the suitability of soils for specific crops to ensure that farmlands are not overwhelmed and that nutrients are not depleted in the long term (Sharifi, 2003).
It should be noted that soil structure and content are dynamic and change with environmental conditions and human activities changes. Thus, the soil structures in the study communities today may not have been the same 10 years ago and may not be the same 10 years to come. Therefore, information from land suitability analysis is not meant to restrict the crop selection choices of farmers but rather to provide them with needed information on what to do to ensure the soils are suitable for the crops they intend to cultivate. However, information from the crop suitability model can help to alter the level of organic matter or nitrogen in the soil or adjust the soil pH or electrical conductivity to suit specific crops. A study done by (Lal, 2006) showed that with every 1 Mg ha− 1 increase in soil organic carbon pool in the root zone, crop yields for wheat, rice, and maize increased significantly.
To resolve conflicts that may exist between increased farm productivity, income, and environmental concerns, systems and technologies that make maximum use of external inputs and natural resources and avoid degradation are put in place (Sharifi, 2003). This implies that, for one, the farmer must be provided with adequate training on up-to-date farm management practices that ensure that the right amount of fertilizers, be it organic or chemical, are applied on the farmlands, and the right choice of and combination of land preparation and management practices are observed. This is very much needed in the study communities as interviews with farmers revealed that some resorted to wrong farming practices due to financial and technological challenges and a lack of proper understanding and training. Future training sessions must focus on practices farmers can engage in to restore and keep deficient soil nutrients in the soil and how crop yields can be optimized.