Smallholder farmers’ perceptions and adaptive response to climate variability and climate change in southern rural Ghana

Abstract This paper assesses the perception and adaptation strategies employed by smallholder farmers to abate the effect of climate variability. Using a mixed method cross-sectional research approach on smallholder farmers in the Bosomtwe district, 152 farmer respondents were sampled from 12 communities by a simple random sampling technique. The quantitative data gathered were subjected to the tools of binary logistic regression analysis, multiple regression analysis, contingency tables and frequencies, embedded in the Statistical Package for Social Sciences (SPSS) v. 17. Farmers’ observations were consistent with meteorological data which revealed that temperature has been increasing since 1981 and rainfall pattern has been erratic. Perception of deforestation (p < 0.05) was a significant and positive predictor of climate change. A binary logistic regression analysis performed on disposition of farmers persistently practicing a particular adaptation strategy indicated that irrigation, mulch, use of agrochemicals and changing the type of crop cultivated were positive and significant at p = .027, p = .000, p = .020 and p = .035, respectively. Policy on adaptation must make room for and capitalize on local knowledge for sustainable climate action. Agroforestry should be introduced to farmers to curb such maladaptive practices of clearing of forests and help sustain forest and agriculture.


PUBLIC INTEREST STATEMENT
Local farmers have been responding to environmental change since time immemorial with Traditional Ecological Knowledge informing the trajectory of response to sustain livelihoods. Their understanding of what climate change is, its cause and how man contributes to it will influence their response to climate policies. This is especially so as current discussions on the subject tilts towards climate action. This paper sought to find out if the observations of local people about climate in their locale are in synch with scientific observation. It also sought to find out their understanding of factors that affect local climate and how such knowledge leads to adaptation. The results showed that their observations of changes in climate and its cause are partly in synch with scientific observations. However, local knowledge of the benefits of forest was not effectively practiced due to the perceived challenge of negatively affecting the sustainability of agriculture as a primary livelihood.

Introduction
Climate change has become the focal point of most global deliberation due to its impact on various biophysical systems disturbing prime sources of food and water, through droughts, temperature surges and variable rainfall among others (Bendell, 2018;Berg & Lidskog, 2018;USGCRP, 2017). Crop production in Sub-Saharan Africa is affected by many aspects of climate change resulting primarily from average temperature increase, changes in rainfall amounts and patterns, extreme events and sea level rise (Di Falco & Veronesi, 2013;Pereira, 2017;Serdeczny et al., 2016). Agriculture provides livelihood for 80% of the African labour force and food for its populace. In the wake of current trends of climate change and climate variability, sub-Saharan Africa's agriculture has become very vulnerable and thus needs to adapt to the projected or actual changes in climate (Assan, Suvedi, Schmitt Olabisi, & Allen, 2018). The effectiveness of these adaptation strategies depends on a myriad of factors including the knowledge base of farmers, capital, technology, the physical environment, social support systems as well as provision of farmers' support services that make such adjustments possible (Assan et al., 2018;Zizinga et al., 2017). This also implies that climate change action must be informed by Traditional Ecological Knowledge (TEK) in areas of trend of environmental change experienced, adaptations options, land use, and natural resource management (Aswani, Lemahieu, & Sauer, 2018;Saylor, Alsharif, & Torres, 2017: Vinyeta & Lynn, 2013. Uddin, Bokelmann, and Entsminger (2014) in their study in Bangladesh purport that adaptation is in itself a process occurring within a particular context with peculiar socio-cultural structures and knowledge for responding to and managing such environmental change. A completely alien and external approach will not apply, as an in-depth understanding of the processes involved is essential. Hence, there is the need to appreciate the various differences in the various climatic stimuli of change experienced and the place of local knowledge in such context (Pearce, Ford, Willox, & Smit, 2015: Vinyeta & Lynn, 2013. Such knowledge does not only influence how local people natively perceive and deal with biophysical and environmental discrepancies (Pearce et al., 2015) but also determines how they respond to policies and programmes that are either in synch with or disparate from their knowledge.
Adaptation strategies are usually inspired by government regulations, socio-economic conditions and personal motivations (Bendell, 2018: de la Poterie, Burchfield, & Carrico, 2018. In Ethiopia, Nzuma, Waithaka, Mulwa, Kyotalimye, and Nelson (2010), found off-farm adaptation practices and climate change adaptation practices include reducing consumption levels, rationing amount of food during droughts, collection of wild foods to supplement household food supply, engaging in petty trading, migration, sale of labour and assets. Crop insurance has also played a key role in the response of households in rural areas to climate change (Romero & Molina, 2015). However, such viable prospects for farmers have not been employed at a large scale in the African continent. Africa needs to take advantage of its national and regional experience to adapt effectively with respect to its farming systems (Chemnitz & Hoeffler, 2011;Welborn, 2018). This is key in ensuring the long-term sustainability of various climate change adaptation and mitigation interventions. The place of local-level actors must be emphasized and properly instituted into national adaptation policies.
The adoption of western technologies and approaches to address climate change in developing countries has faced challenges because they appeared to be alien and disparate from TEK enshrined in their socio-cultural fibre of indigenous societies (Nawrotzki & Kadatska, 2010). It is important for developing countries to recognise that adaptation does not just require advanced technology and research but more crucially the ability of vulnerable groups to respond appropriately through behavioural change and adapt their livelihoods (Ayeb-Karlsson, van der Geest, Ahmed, Huq, & Warner, 2016). Simon (2013) for instance found that farmers in response to drought adopted new improved and drought tolerant sorghum varieties. Thus, the impact of new crop varieties is affected by both local conditions and household characteristics (Simon, 2013). They were also more likely to diversify into non-farming activities.
Climate change is manifest in Ghana through increasing temperatures, declining rainfall, variation in crop growing season, sea level rise and increased extreme weather events (Ahenkan & Boon, 2010;Asamoah, 2018). This has implication for agriculture and related livelihood dependent households (Yaro, 2013). To adapt to these weather extremes, key on-farm adaptation practices that have prevailed include creating or improving drainage system, early planting, stoppage of farming in low-lying areas and waterway, irrigation or dry season farming, planting more trees and cover crops, planting drought resistant and early yielding crops, early planting, improved farm management practice, diversification into livestock rearing and increasing fertilizer (Antwi-Agyei, Fraser, Dougill, & Simelton, 2013;Asante & Amuakwa-Mensah, 2014;Mohammed, Kwaghe, Abdulsalam, Aliyu., & Dahiru, 2014). Climatic conditions in southern Ghana vary significantly from the northern part. The forested southern ecological zones have a wetter climate than the rest of the country with significant variations in rainfall and temperature (Yaro, 2013). The forest ecological zones experience a bi-modal rainfall regime commencing in April to July and then September to November with a respite in August (Asamoah, 2018). Effective adaptation in the forest zones must address illiteracy, alternative livelihood activities, easy access to information, sustainable farming systems and sustainable natural resource management (Obeng et al., 2011).
Farmers in the Bosomtwe District have the potentials for on-farm adaptation against climate variability and change. One plausible factor to promote effective on-farm adaptation is the location of farmers within the lake Bosomtwe catchment area. Drought has been part of the issues that farmers in Northern Ghana have had to deal with for decades. However, access to water for irrigation in the southern Ghana is much easier and more crucial as climate change and climate variability increasingly affect water supply. Adaptation strategies identified by Fosu-Mensah, Vlek, and Manschadi (2010) does not include irrigation although soil conservation practices are mentioned. However, one cannot clearly tell whether these are effective and dislodges the need for irrigation. Other studies in southern Ghana by Acquah (2011), andNdamani andWatanabe (2015) mention irrigation but does not expatiate on the types and the challenges associated at the communal level. This paper in this regard seeks to consider irrigation in the Bosomtwe District in light of the sources of water for irrigation, types of irrigation and the challenges associated. Knowledge on how farmers responding to climate change based on their TEK of the benefits of forests is also assessed. The study explores how traditional ecological knowledge is brought to bear in the adaptation process in southern Ghana. Again the study sought to find out to what extent local observations of climate variability is in synch with meteorological data and their perception of the causes of climate change. This research examines the on-farm adaptation practices by smallholder farmers against climate variability in the Bosomtwe District of the Ashanti Region of Ghana and their perceptions of climate change.

Profile of the study area
Bosomtwe District lies within Latitude 6º28ʹN-Latitude 6º40ʹN and Longitudes 1º 20ʹW-Longitude 1º 37ʹW in the Ashanti Region of Ghana. It covers an area of 330 km 2 with Kuntenase as its district capital (Figure 1). The district shares its northern border with the Atwima Nwabiagya district and Kumasi Metropolis. The Ejisu Juabeng Municipal Assembly is to the east and in the south, it shares its border with Amansie West and East Districts (Ghana Statistical Service, 2010). Lake Bosomtwe-the biggest Crater Lake in West Africa is located in the district which serves as a source of livelihood and recreation. Rivers and streams in the district flow into the lake in a dendritic pattern from the surrounding highlands. Supan, Butu, Oda, Adanbanwe and Siso are rivers found in the district (Appiah, Bugri, Forkuor, & Boateng, 2014).
The Bosomtwe District is in the equatorial climatic zone experiencing a double maxima regime per annum. The first rainfall regime which is the major season begins from March to July while the second and minor season starts from September to November. The mean annual rainfall is about 1,900 mm and mean monthly temperature is 36°C. The relative humidity of the district ranges between 60% and 85% (Appiah et al., 2014).

Sampling design, instruments and data analysis
This paper is an extract from a broader study on farmers' mitigation of and adaptation to climate change and climate variability in the Bosomtwe District of Ashanti Region, Ghana. The paper is particularly focused on the various on-farm adaptive strategies of smallholder farmers in light of their perceptions of change and local knowledge of climate amidst the current trend of climate variability in the Bosomtwe District. The study adopted a mixed method cross-sectional research approach on smallholder farmers in the Bosomtwe district. The study employed the mixed method approach to data collection and analysis using quantitative and qualitative data with their respective analytical methods. This approach primarily requires the collection and analysis of quantitative and qualitative data, resulting in a deeper understanding of research problem under study than either approaches would have yielded individually (McKim, 2017). This serves not only to provide data that is amenable to wide and varied range of statistical analysis but also provides in-depth and insightful understanding of the problem being researched from the analysis of the qualitative data. Consequently, a convergent parallel approach of the mixed method was adopted. This led to the simultaneous collection of qualitative and quantitative data to be subject to their, respectively, appropriate analytical procedures (McKim, 2017) with the results complementing each other.
Primary and secondary data and information were adopted for this study. Secondary data and information were acquired from publications while secondary data on rainfall and temperature of the district were acquired from the Kumasi Meteorological office. Primary data were acquired through the administration of partially pre-coded questionnaires to a cross section of smallholder farmers from 12 communities and interviews were held with the key informants who were District Extension Officer, District Crop Research Officer and the Director for Women in Agricultural Development. Primary data focused on farmers' perception of climate variability and their adaptive response to the prevailing situation of climate variability. The questionnaires were administered by the researcher and assisted by a graduate who is abreast with research procedures and Source: Author's construct (2015) requirements in a one-month period (January to February). The researcher and his assistant thoroughly discussed the question for in order to be abreast with that questions and also for clarity.
Key areas of concern included farmers' observations of local climate change vis a vis meteorological data on temperature and rainfall in the Bosomtwe district. It also sought to find out their perception of the causes of climate change, the role of forests in climate and traditional ecological knowledge on forests and climate and how such knowledge manifest in their adaptive response. Finally, their adaptive strategies employed in the face of climate change and climate variability is also assessed. Secondary quantitative data on temperature and rainfall were acquired from the Kumasi office of the Ghana Meteorological Agency (GMA). The statistical anomalies approach was used to analyze the rainfall and temperature trends in the Bosomtwe district for a 30year period from 1981 to 2011 focusing on annual and inter annual variations in rainfall and temperature patterns. Average values for rainfall and temperature for the 30-year period (climatological normal) served as the basis for the assessment for the stipulated period. The outcomes were graphically presented in charts and graphs which were exported to excel for editing. Twelve (12) communities were selected for the study which were Amankwadei, Dedesua, Kokodei, Ayuom, Aduampong, Brodekwano, Abaase, Aborodwom, Pipie Old Town, Anyinatiase, Nkowi and Obo. Yamane's formula for determining a representative sample size expressed as: n = N/1 + N (e) 2 (where "n" is the sample size, "N" is the sample frame and "e" is the margin of error) (Yamane, 1967) was used to arrive at a sample size of 152. The respective sample for each community was allocated by quota using the proportionate sampling procedure. After obtaining a list of farmers from the district office of the Ministry of Food and Agriculture (as the sample frame), the lottery method of the simple random sampling technique was adopted to hand-pick farmers from the various communities. This was to give the farmers an equal opportunity of being selected for the study. The primary quantitative data gathered were subjected to binary logistic regression analysis, contingency tables for multivariate distribution of variables and frequencies for counts using the Statistical Package for Social Scientists (SPSS) version 17 for windows application. The outcomes were graphically presented in tables, charts and graphs being exported to excel for editing. The qualitative data were gathered using an interview schedule. This was conducted in English as all the key informants could read and write. Their responses were transcribed and integrated into their respective thematic areas.
Because of a peculiar challenge of access to information for informed seasonal response to climate variability, a binary logistic regression was used to assess the likelihood of farmers persistently practicing any particular adaptation strategy. The disposition of a farmer persistently practicing a particular adaptation as a strategy against variable weather patterns and its repercussions on crops was assessed with a binary logistic regression model using the ODD ratios for the possible prolongation of an adaptation practice. Employing the Odd ratios for the probability of persistent use of an adaptive strategy, this modelling was to ascertain whether farmers are more likely to dedicate more time and resources to a particular adaptation strategy should the prevailing conditions of inconsistent and unpredictable rainfall pattern continue. To this end, farmers considering themselves vulnerable to the current vagaries of the climate was the dependent variable while the various adaptive strategies employed by the farmers were independent variables.
Respondents were thereby required to respond 1 = Yes (vulnerable due to inconsistent and unpredictable climate) and 0 = No (not vulnerable to inconsistent and unpredictable climate). To this end, a step-wise logistic regression in the Statistical Package for Social Scientist was run. The model devoid of the predictive variables was the null hypothesis stating that there is no significant difference between the dummies (Yes and No for effectiveness of adaptation strategies adopted) which are the response variable devoid of independent variables. Thus, the model will better predict the outcome if the independent variables are excluded. The alternative hypothesis states that the model would not better predict the outcome without the independent variables indicating a significant difference in the dummies in the response variables. The first model devoid of the dummy variables revealed that the total percentage correctness of the prediction was 91.4% having a significance value p < .000, with B (EXP) of 0.94. In the alternative model where the predictive variables are included in the equation, the overall correctness of the prediction increased to 96.1% at a significant value of p < .000. This had a Chi-square value χ 2 = 52.546 at 5 degrees of freedom. The results indicate that the dummy adaptation variables explain an additional 4.7% predictive capacity. The Nigelkerke R 2 shows that the model explains 66.1% of the variation in the dependent variable as attributed to changes in the independent variables. Ten independent variables were entered into the alternative hypothesis equation as non-categorized. Only four were however significant in predicting farmers' likelihood of persistently practicing a particular adaptation strategy. These independent variables were irrigation, mulching, use of agrochemicals and change of crop type.

Socio-economic characteristics of respondents
Analysis of the socio-economic characteristics of farmers revealed a trend that brings to the fore the imperative need of farmers in the Bosomtwe district to adapt. Majority of the respondents were of the age group 46-55 years and a total of 79% of respondents were less than 56 years vividly shown in Table 1. This implies that they were young and energetic.
Also, most of the respondents (82%) were married and therefore had household responsibilities on their shoulders. Hence, there was the need to ensure continuous supply of income and food to household upkeep. Also, 75% of respondents were of households with male heads while and 25% were females. However, 51% of respondents were household heads while 46% were spouses of household heads and the remaining 3% were other relations.
The communal nature of the societies is such that information is not restricted to household heads only. Frequent migration among males especially in lean and dry season, coupled with the fact that much of household spending is done usually by the women makes them adequately informed on household spending patterns. Sellers (2016) purports that households that are headed by females are more responsive to climate change than male-headed households. This was however not the case in the Bosomtwe District as both male and female heads made efforts to adapt. The analysis also showed that 88% of the respondents had formal educational level below senior high school while only 12% had senior high school and tertiary level education (Table 2). A low level of education is generally observed amongst farmers and this could affect their adaptation to climate change.
Majority of the farmers (90%) had household size of 1-10 members implying there were many mouths to feed and decrease in crop yields might lead to hunger and starvation. Majority of the farmers had farming experience of 16-20 years (38%) and 21 years and more (41%). This confirms that most of the respondents had been farming long enough to have experienced changes in climate. The average temperature for the Bosomtwe District from 1981 to 2011 was 32.2°C which served as the baseline for the analysis of the average annual temperature. Temperature values in this 30 year were erratic as indicated by the red line in Figure 2. The lowest average temperature for the three decades was 30.9°C which was recorded in 1985 while the highest for the period was 33.5°C in 2008. The yearly averages from 1981 to 1994 were all below the average for the 30-year period under study, indicated by the yellow line in Figure 2.
The year to year average temperature for the district was below the average for the period from 1981 until 1995 which recorded an average temperature value of 32.5°C. It dropped again only in 1996 and rose above the average value of 32.2°C from 1997 to 2010. The average annual linear trend for the three decades indicated by the black line in Figure 2 vividly shows that microtemperature of the district has been increasing steadily indicating change in the Bosomtwe district's micro-climate rather than variation above and below the baseline average from 1981 to 2011. According to Dinse (2011) an analysis of annual temperature and rainfall values for at least a thirty-year period produces average values not affected by year to year variability.
The average rainfall amount for the Bosomtwe District from 1981 to 2011 was 117.4 mm which served as the baseline for analysis of the rainfall pattern for the period under study. In the first decade, only 1984 recorded an average rainfall above the average for the entire period. This improved slightly in the second decade with 1991, 1993, 1995, 1999 and 2001 recording average annual rainfall amounts above the average baseline of 117.4 mm. Rainfall amount received in the last decade was equally an improvement on the second decade. This was not only in terms of the number of years with average annual rainfall amounts above the average baseline for the period under discussion but also the extent of increase above the average baseline. Average annual rainfall amount consecutively  1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011   Source: Author's construct using district data from regional meteorological office. increased from 2002 to 2009. Hence, the first decade was particularly the driest of all three decades while the third decade was the wettest. Again, the lowest average annual rainfall for the period under study was recorded in 1983 within the first decade while the highest was 2009 towards the tail end of the third decade. Generally, the annual average rainfall amounts have been below the average baseline. Figure 3 further discloses a varied inter annual variability in rainfall amounts from 1981 to 2011 with a minimum deviation of −38.7 mm in 1983 and a maximum deviation of 54.3 in 2009 from the average baseline for the period under study.
This trend of increasing temperature and variability of rainfall threatens agriculture especially among smallholder farmers who are already handicapped by the use of rudimentary equipment, difficult marketing situations and land use challenges. The future of agriculture as an integral part of subsistent life may be oblique unless adaptation and mitigation options at that level are enable farming systems to evolve and be sustainable. Majority of the respondents indicated that the local climate has become capricious manifest in increased temperature, variable rainfall pattern and increased incidence of flooding ( Figure 4). When asked about the pattern of change in rainfall observed, majority of the respondents said the rainfall frequency had increased (67% of response) while its intensity has decreased (79% response). About 91% believed that rainfall had become inconsistent and unpredictable while 9% believed that the rains were consistent and predictable. Meteorological data on rainfall for the district revealed that rainfall amount had decreased and become inconsistent just as the respondents observed.  1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999    When questioned about their views on temperature, more than three quarters of the respondents (87%) indicated that temperature has been increasing in the last 10 years while 13% said otherwise. This observation of farmers is in synch with the meteorological data for the district which shows that micro-temperature of the district has been increasing steadily as seen in Figure 1. Again, 77% claimed that the incidence of flooding had increased whereas 10% and 13%, respectively, said it had decreased and remained the same (not noticed any change). An increase in temperature and increase in rainfall variability and amount has generally been predicted by various models for the tropics (Hansen, Sato, & Ruedy, 2013;Ma et al., 2017) as confirmed by this study. Consequently, 71%, 84% and 84% perceived an increase in the incidence of drought, crop pests and diseases, and flooding, respectively, while 29%, 16% and 16%, respectively, said that these incidences were normal. In the event of crop failure, farmers resort to coping strategies such as borrowing, engaging in alternative livelihood activities, migration and changing of eating patterns.

Perceptions of causes of climate change
The effect of these human activities was generally seen as not having a direct impact on climate as 18%, 25%, and 26% respective said smoke emission of (vehicles, burning of tyres, and firewood) and industrial emissions do not affect climate. A majority of respondents (61%) however said that deforestation contributes to climate change. In a similar study in urban Ghana, Codjoe, Owusu, and Burkett (2014) identified smoke-related causes such as burning of tyres and firewood, industrial emission together with chlorofluorocarbon (CFC) emissions, burning of copper wires, refuse dumps and copper wires as supposed human causes of climate change. They also identified deforestation to be widely held as a key contributor to climate change (Codjoe et al., 2014). Let "M i " (Binary variable) represent the response on human causes of climate variability ("ith" observation). Therefore, M i = 1 for human causes of climate variability and M i = 0 for no human causes of climate variability. It follows that: where "g" is the functional relationship between the response on human causes of climate variability (M i ) and the random variable (Y i ) which determines the probability of human causes of climate variability. The equation to be estimated is given as: The probability for the occurrence of human causes of climate variability among the sampled respondents is given by: where βX = the mean of the predicted Y i values.
If P % 1, there is a high perception for human causes of climate variability among the sampled respondents. On the other hand, if P % 0, there is low perception for human causes of climate variability among the sampled respondents.

Analysis of logit regression results
The regression results of the logit model: where βX = the mean of the predicted Y values.
βX : Since P = 0.625 is closer to one than zero, there is a high perception of human causes of climate variability among the sampled respondents. In other words, with the sample evidence at hand, about 62% of the sampled respondents have perception of human causes of climate variability. The implication is that, there is high perception of human causes of climate variability at the study area.
The above logit regression results using SPSS shows that most of the coefficients are consistent with hypothesized relations and their test of significance (P values) help to indicate their importance in explaining the perception of human causes of climate variability among the sampled respondents, although there are no policy variables in the empirical model. Vehicular fumes causing CV as a dummy variable has met the expected sign in the empirical model (Table 3). The positive sign of the coefficient (0.035) means that there is more perception of vehicular fumes causing CV than otherwise (control group), holding all other variables constant. But the P-value (0.433) indicates that the coefficient of vehicular fumes causing CV is statistically insignificant even at 30% in the empirical model. The explanation is that, there is no significant perception about emission of vehicular fumes as the cause of CV in the study area. Deforestation causing CV as a dummy variable has met the expected sign in the empirical model (Table 3). The positive sign of the coefficient (0.868) means that there is more perception of deforestation causing CV than otherwise (control group), holding all other variables constant. The P-value (0.000) indicates that, the coefficient of deforestation causing CV is statistically significant at 5% in the empirical model. The implication is that, there is high perception of deforestation as human cause of climate variability in the study area. Consequently, climate increasingly becomes variable as forests are cleared. Although the increased use of fossil fuel is said to be the fundamental source of increased concentration of atmospheric greenhouse gas, land use changes including deforestation is partly also responsible (Metz et al., 2001).
Bush fires causing CV as a dummy variable has met the expected sign in the empirical model ( Table 3). The positive sign of the coefficient (0.051) means that there is more perception of bush fires causing CV than otherwise (control group), holding all other variables constant. But the P-value (0.258) indicates that the coefficient of bush fires causing CV is statistically insignificant even at 20% in the empirical model. The implication is that, there is no significant perception about bush fires as the cause of CV among the sampled respondents.
Industrial emissions causing CV as a dummy variable have met the expected sign in the empirical model (Table 3). The positive sign of the coefficient (0.041) means that there is more perception of industrial emissions causing CV than otherwise (control group), holding all other variables constant. But the P-value (0.385) indicates that the coefficient of industrial emissions causing CV is statistically insignificant even at 30% in the empirical model. The explanation is that, there is no significant perception about industrial emissions as the cause of CV according to the sample evidence at hand. Together, all the independent dummy variables have significant impact on the perception of human causes of climate variability because the F statistic (202.061) has P-value (0.00) which is statistically significant at 5% level.

Traditional ecological knowledge on forests and climate
Local knowledge and interpretation of climate change and climate variability is based on their own experience. It is however not entirely disparate from scientific discourse although the educational level among the respondents was generally low. About 83% acknowledged that land use patterns affect weather explaining that forest cover has been lost to agricultural land use and eventually residential in parts of the district. This has led to stronger winds with greater impact on farms and property. Respondents also indicated that land cover influences temperatures, explaining that forested areas are less warm compared with places with little or no forest cover. To this Ellison et al., (2017) and Wolff, Masuda, Meijaard, Wells, and Game (2018) emphasize that reforestation (or avoided deforestation) in the tropics results in a cooling effect through sequestration of atmospheric carbon, increased evaporation and cloud cover.
As high as 92%, 99% and 80% of respondents, respectively, held that forest serves as watersheds, prevents intense surface heating and also creates favourable local climate. Also, 38% and 32% said that forest/vegetation have benefits of enhancing cloud formation (hence rainfall) and sequesters carbon dioxide, respectively ( Figure 5). According to Ellison et al., (2017), land use patterns that reassure forest conservation and expansion also increases evaporation and consequently cloud formartion. This leads to a cooling effect by shielding the earth from incoming solar radiation. Also, there are myriads of benefits associated with forests for which reasons they cannot be excluded from discussion of climate change mitigation and adaptation. Tropical forests have a significant dampening effect on to global climate via their high transpirations and cloud formation and also absorb large amounts of carbon dioxide, resulting in a dampening. (Astorga, Moreno, & Reid, 2018;Deb, Phinn, Butt, & McAlpine, 2018).
Although farmers had a fair knowledge of the benefits of trees, the practice of tree planting was not popular. Most (43%) of the farmers had not planted any tree on their farms while just 38% planted between 1 and 4 trees on their farms in the last 10 years. Only 15.6% had planted between 5 and 10 trees and 3.4% had planted more than 10 trees on their farms. Discussion with the district crop research officer revealed that a Non-Governmental Organization in collaboration with the district office of the Ministry of Food and Agriculture distributed tree seedling to farmers but most of the farmers did not plant them. Knowledge of the benefits of forests does not commensurate with practice. There were a number of concerns raised by farmers that should be mentioned. Farmers believe that trees and their root systems take up space and make tilling of the farmland cumbersome. Making the most of agricultural land to ensure sustained household food supply amidst current vagaries of the climate was key to farmers. Also, farmers had notions that some crops do not do well under trees. Also, the demographic analysis revealed that about 90% of respondents had household sizes of six or more people implying that securing household food supply was very crucial and farmers were not willing to trade that off for the other benefits associated with forest conservations in the short term.
The study further revealed that local perception of human-induced climate change is quite high (63% as against 37%). This reflected in a generally high affirmative response for the question "is human cause of climate variability and climate change prominent" observed across all levels of education (Table 4).
A similar trend is revealed in a cross-tabulation of age of respondents and perceptions of human-induced climate change. However, with respect to the age of respondents, the 34-45 and 46-55 age cohort particularly recorded a high affirmative response accounting for 52% of the total response (Table 5). These trends are an amalgamated outcome of a strong traditional ecological knowledge base and formal education. It is important in light of this finding to drum home the need to critically look at policy approach with respect to dealing with climate change at  the local scale. This is because local knowledge is an outcome of experience and exposure and also determines how they respond to policies and programmes that are either in synch with or disparate from their knowledge as asserted in literature (Aswani et al., 2018: Pearce et al., 2015Saylor et al., 2017).

Adaptation strategies against climate variability and climate change
This study sought to assess the various adaptive strategies of smallholder farmers amidst the current trends of climate change and climate variability in Bosomtwe District. This is very critical as the district is predominantly agrarian in nature. Agricultural activities, dependent on such climate variables as rainfall and temperature and adversely affected by weather extremes viz floods and droughts is highly vulnerable amidst persistent climate change and climate variability. The results revealed that 93% of smallholder farmers in the Bosomtwe District have adapted to climate change while 7% claim to have not. Adaptation strategies adopted by the farmers included mulching, use of agrochemicals, irrigation, change of crop type and change of crop variety, expansion of farms and clearing of virgin forest for new farmlands. Berry, Yassin, Belcher, and Lindenschmidt (2017) and Rey, Holman, and Knox (2017) indicate that irrigation as an important adaptation measure helps to buffer for moisture deficits induced by changes in climate. Variations in the onset of rains and seasonal rainfall and temperature variations can exacerbate water availability (Dahal, Shrestha, Tuitui, & Ojha, 2019). Sources of water for irrigation in the district include Lake Bosomtwe, River Oda, streams and boreholes. The study indicated that although 50% of farmers have farms located near these sources of water that can be harnessed for irrigation to make up for moisture deficits due to climate variability, only 28% of them actually practice irrigation. This pattern is particularly so because irrigation in the district is done either with a water-pumping machine which only a few can afford or by using watering cans. Although there are no cultural, policy or legal barriers that inhibit the use of water from Lake Bosomtwe and other waterbodies for irrigation, this strategy is surprisingly very minimal in the district. This is due to the fact that the landscape in some parts of the District is undulating, making non-mechanized irrigation as currently practiced in the district a very difficult endeavour.
Although irrigation is predominantly done in the off-season, it is increasingly being used in the event of delayed rains until the onset of the rainy season. This is confirmed by the district extension officer attesting that: "Some farmers have increasingly resorted to the use of irrigation to cushion them against the reduction and irregularity of the rainfall".
This finding agrees with that of Berry et al. (2017) where irrigation and water conservation techniques were seen as an important source of additional water that is used to lengthen the growing period of crops. Hence, it is a critical adaptation strategy employed to make up for shifts in the growing season. Owusu and Klutse (2013) in their studies found that maize yield in general was high in good years of rainfall. This was attributed to among other factors, the application of fertilizer and agrochemicals. Use of agrochemicals as an adaptation strategy is important as a method of dealing with loss of rich topsoil through run-off and loss of soil nutrient through leaching resulting from increased precipitation. Among smallholder farmers in the Bosomtwe District, use of agrochemicals such as fertilizer to help increase crop yield and pesticides and weedicides to control pest is minimal. Only 11% of respondents indicated that they use agrochemicals while 88% do not. A little more than half (thus 57%) of respondents asserted that the cost of agrochemicals was high and as such they could not afford. The district agricultural officer in response to the use of agrochemicals explained said: "Some farmers have also resorted to the use of other un-recommended chemicals in their farming practices because they are cheaper".
A similar observation was made by Fagariba and Soule (2018) in the Sissala West District of Northern Ghana. The minimal usage is also attributed to the fact that 84% of the respondents have not experienced increase in the incidence of diseases and pest with the variable climate while only 16% have. Farmers indicated that agrochemicals would have come in handy to complement crop loss due to strong winds.
Subsistent farmers in the Bosomtwe District also practice mulching as a key adaptation strategy. While 92% of respondents pointed out that they retain crop residue after harvest on the surface of the soil, only 8% indicated using it to re-enforce farm boundaries. Around 91% (≈92%) of those who practiced mulching said it has been effective pointing out that mulch helps to improve soil nutrient level and retain soil moisture and 9% said otherwise. Destructive traditional farming practices as slush and burn have been discouraged while mulching has been encouraged by agricultural extension officers among smallholder farmers in the Bosomtwe District. There are other benefits associated with this practice that lessens the impact of climate change at the onfarm level. Nambiza (2013) also found that in Tanzania, farmers use mulching as one of the soil protection methods. He also highlighted that it reduces soil and wind erosion, control weeds and preserves soil nutrient and moisture Mulching also prevents soil erosion and preserves soil carbon stock.
Although mulching has been a traditional agricultural practice among subsistent farmers in the district, farmers have found the practice to be very helpful amidst current trends of climate change and climate variability. This adaptation strategy is also prevalent because beyond the challenge of water deficit associated with climate change, farmers with farmlands closer to the communities can't expand their farms and hence only maximize the current farmland. The dominance of this practice can also be attributed to the fact that mixed cropping is the predominant farming system among subsistent farmers in the Bosomtwe District. Hence after the harvest of one crop, the residue cannot be burned but at best only removed from the farm. Farmers asserted that mixed farming helps to better maintain the soil moisture and nutrient contents as compared to other farming systems and also keeps pests at bay. Improved crop varieties lessen farmers' vulnerability in that they mature much faster and are hence less likely to be adversely affected by climate change compared with the traditional varieties (AGRA, 2013). Changing the type of crops grown is another adaptation strategy that subsistent farmers in the Bosomtwe District adopt to help them adapt to climate change and climate variability. Climate change according to subsistent farmers in the Bosomtwe District is made manifest through droughts, flooding, shift in crop growing season and variations in rainfall and temperature patterns. The impact of these has been loss of crops prematurely and crop failure. Although only 30% of respondents have changed crop type while 70% have maintained the crops grown, this change is skewed towards crops that require less water, are more drought tolerant and adaptive to the precarious nature of the growing season such as cassava and plantain rather than maize. Also 43% and 57% of respondents indicated changing the variety of crops grown and otherwise, respectively. The crops were particularly maize and cassava. This represents the gradual emergence of a shift from the traditional crop variety to hybrids that mature much early. The district officer in charge of crop research added that: "we have been supplying them with new crop varieties particularly the staples. These mature earlier and are more resistant and this has been very helpful. However, our challenge is that we sometimes have transport challenges. Some NGOs have also held training programmes for farmers and provided them with tree and crop seedlings." This trend was equally revealed in Cameroon and Ethiopia (Chia, Somorin, Sonwa, & Tiani, 2013;Sime & Aune, 2018). In the Limpopo Basin in South Africa, the key adaptation strategy to increasing temperature is changing crop cultivars (Di Falco & Veronesi, 2013). This trend was equally seen in Tanzania where farmers bought variety of sorghum that is known to mature early (Nambiza, 2013). Farmers are increasingly planting improved and new crop varieties as a form of security against the vagaries of the climate (Rahmanian, Batello, & Calles, 2018). On-farm adaptation practices among smallholder farmers in the Bosomtwe District are shown in Figure 6.
A critical look at the various adaptation practices and the age of respondents (Figure 7) sheds light on how farmers of different age groups respond to climate change. In Figure 7, the reference point for each adaptation practice in each age cohort is the total affirmative response of that cohort. Hence, the variation in each adaptation strategy among various cohorts can be seen in comparison with the cohort total. With respect to irrigation, it can be seen from Figure 7 that use of irrigation was generally low in all age cohorts and the practice of irrigating farms decreased as age increased.
The age group that practiced irrigation most was 20-35 years where 4 out of 9 respondents in that group practiced irrigation representing 44% out of 100%. The practice among cohorts 36-45 years, 46-55 years, and 56-65 years, respectively, had 32%, 27% and 28% of their totals practicing irrigation. This is because farms were manually irrigated (as opposed to mechanized irrigation) using watering cans. This coupled with the undulating nature of the terrain made it difficult for aged farmers to adopt irrigation. Farmers above 65 years did not practice irrigation at all. Mulching was high among all age cohorts. This on-farm adaptive response to climate change is rather less tedious and easy to practice with well-known benefits. The age group that least practiced it was the 36-45 cohort recording 43 (86%) out of 50 (100%) while 100% of respondents in the 20-35 and 66-75 cohorts practiced it. The use of agrochemicals was generally low among all cohorts and decreased as the age bracket increased. It was highest (3 out of 9 indicating 33%) among farmers aged between 20 and 35 years due to high prices of agrochemicals as indicated by the respondents.  Figure 6. On-farm adaptation practices among subsistent farmers in the Bosomtwe District.

0%
Change of crop variety was found to increase as the age group increases (Figure 7). Staples were mostly cultivated since they are crucial for household food supply. This finding is in agreement with that of Reincke et al. (2018) in Tanzania where farmers cultivated crops that were primarily needed for household consumption. The ability of farmers to be mobile and engage in other economic activities reduced as they advance in years. While younger farmers can resort to options that make them less vulnerable to the vagaries of the climate such as migrating, sale of labour, and learning a trade among others, the adoption of improved crop variety proves more useful for the aged. Other studies have found that subsistent farmers find it necessary to migrate in search of alternative livelihood activities elsewhere as farming as a primary economic activity becomes more unsustainable (Ayamga, Das, & Banerjee, 2019: Food and Agricultural Organization, 2017c: Grecequet & Hellman, 2018: Munoz, 2018. The Food and Agricultural Organization (2017c) for instance explains that a combination of socio-economic irritants and global environmental change leads to an increased vulnerability and loss of livelihoods. This increases the tendency of migration among rural farmers either by necessity or choice. The caveat however is based on the adaptive capacity of the people involved such that migration is high among those with less capacity to adjust or respond to the situation. Quite contrary to this Grecequet and Hellman (2018) mention that migration as an adaptive strategy to climate change is not available to all as people who are underprivileged and economically and physically vulnerable will not be able to migrate. In the case of the Bosomtwe district however, migration is more of a choice rather than a necessity since a number of adaptive strategies are being employed that sustain productive agricultural activities in the district. Hence, adopting new varieties of crops cultivated is a prudent way of sustaining farming and subsequently household food supply among ageing farmers. Respondents who changed the type of crops cultivated as an adaptation strategy were generally spread across all age cohorts except for those 65-75 where it is absent.
Respondents with Junior High School/Middle School (JHS/Middle Sch.) qualification adopted more of all the adaptation practices than those of other educational qualification (Figure 8). Quite interestingly, respondents with no formal education practiced irrigation, mulching, used agrochemicals and changed the type of crop cultivated more than those with Primary, Senior High/Technical/Vocational and Tertiary education. This is attributed to the activities of agricultural extension services within the district. The district office of the Ministry of Food and Agriculture has assigned extension and crop research officers to all communities in the district who work directly with farmers helping to reduce vulnerability of farmers to various biophysical shocks. Farmer with no formal education has been less mobile as they have had no need to commute or relocate from time to time to attend school as is the case for those with various educational qualifications. Again, the adoption of these adaptive strategies in the Bosomtwe district was generally low among respondents with higher educational qualification because education gives them exposure and training, making them economically versatile ( Figure 8). The opportunity to engage in alternative livelihood activities affects the need to reduce agricultural vulnerability through the adoption of these strategies. Fagariba and Soule (2018), in a similar study in Northern Ghana conversely found that educational level enabled farmers to make rapid decisions in response to climate change. Those with formal education adapted better to climate change than those without formal education. They also found that extension services helped to strengthen the resilience of farmers.
One other practice that came up during the study is the clearing of virgin forest to open up new farmlands. This study reveals that 31% of farmers have cleared virgin forest for new subsistent farmlands although they are the minority, compared to 69% of respondent who did not. They mentioned that the output of new farmlands was much better compared with that of older ones. This practice could not be categorized as an entirely subjective response to current climate change as it is an age-old practice among smallholder farmers. They could not however attribute this practice entirely from other factors necessitating increased. However, according to Chia et al. (2013), farmers in Ca in Cameroon asserted that newly open fields in forest frontiers were more fertile, resulting in increased yield. Nambiza (2013)   Tanzania. To Duguma et al. (2019), his practice is an investment decision that is underpinned by livelihood security and must be critically ascertained. This practice in the Bosomtwe district was however constrained by the fact that unfarmed lands belonged either to other adjourning communities or other people and much of the forested terrain was difficult to manoeuvre and work in due to the undulating nature of the landscape. This practice is however a maladaptive strategy as the clearing of virgin forest released carbon in these sinks into the atmosphere, further contributing to the global climate menace (Wolff et al., 2018).
3.6. Likelihood of farmers' persistently practicing any adaptation strategy Access to information on climate change and weather forecasts is constrained and institutional outreach programmes that actually reach the farmers basically constitute the activities of extension officers both at the district and local level. The farmers indicated that forecasts are usually not reliable and this has dire implications for their adaptation. When asked about how farmers' access information on climate change, the district extension officer indicated that access to information is generally a challenge and issues of dependability were hampering adaptation among farmers: "Information from the meteorological service on monthly and seasonal forecasts are not reliable as events have mostly proven otherwise. This has curtailed trust in the forecast and farmers therefore do not depend on these to make inform decisions on seasonal variation in their adaptive response to the vagaries of the climate".
Adaptation must not be static but dynamic. Adaptation does not mean permanence: flexibility and capacity to adjust and transform are far more relevant for farmers to adapt to climate change, rather than their capacity of using year after year the same practice. Consequently, the dynamics of adaptation practices adopted in the Bosomtwe district depends on farmers' access to seasonal weather forecasts and other relevant information. However, in situations such as this when access to such critical information is constrained, their adaptation is also constrained.
The binary logistic regression model was used to test the likelihood of farmers insistently practicing any adaptation strategy and the results indicated that farmers are more likely to continuously engage in irrigation, mulching, use of agrochemicals and change of crop type. Irrigation, mulch, use of agrochemicals and changing the type of crop cultivated were respectively significant at p < .027, p < .000, p < .020 and p < .035. At a 95% confidence interval (CI), these variables had lower to upper CIs for each of the EXP (B), respectively, at CI = .009 -.746, CI = 19.287-8.328E3, CI = 1.798-981.453 and CI = 1.359-4.0380E3. Furthermore, five out of the eleven independent variables had (EXP)B > 1 of likelihood to predict the outcome of the dependent variable of changing farming as a primary occupation when those independent variables were used on the response variable at; B(EXP) = 4.513-400.787 times (Table 5).
This means the possibility of farmers considering themselves vulnerable to the vagaries of the climate is predicted by the independent variables 400 odd likelihoods. Thus, farmers are 400 times likely to continuously engage in these practices as long as they perceive their local climate to be unpredictable and inconsistent and hence making them vulnerable.
The strong B(EXP) values obtained indicate that the odd likelihood of prediction is more than 400 times likely that farmers will continue to engage in these adaptation strategies as avenues for lessening the adverse impact of climate change and climate variability on agriculture as a primary occupation. Since the timing and amount of rains have varied significantly in recent times, utilizing irrigation from hand-dug well, boreholes, streams, lakes and rivers have become a necessity if farming is to persist. Another predictive variable is the practice of using mulch. Most farmers used crop residue as mulch to help conserve soil moisture and improve soil nutrients. This practice according to farmers also controlled weeds and therefore lessened their work in preparing the land at the onset of rains. Use of agrochemicals was also significant as a long-term adaptation strategy. This is meant to make up for reduced crop yield due to destruction of crops by adverse climate events such as strong winds and flooding. Additionally, changing the type of crop cultivation is likely to be a long-term strategy as it enables the farmers to choose which crops have proven to be more adaptive and resilient, and economically prudent to cultivate amidst current climatic trends in the district.

Conclusions and recommendations
Smallholder farmers in the Bosomtwe district have observed an inconsistency in rainfall pattern and an increase in temperature which is in tandem with meteorological data. They identified deforestation as a key contributor to climate change. Although knowledge of the benefits of forests in relation to climate was high, practices in tandem with this knowledge was low due to livelihood security reasons. However, their Traditional Ecological Knowledge has played a key role in their adaptation process. Smallholder farmers in the Bosomtwe District in adapting to the current vagaries of the climate have resorted to such activities as irrigation, mulching, planting different crops, cultivation of improved crops varieties and the use of agrochemicals. However, the practice of clearing of virgin forest to open up new farmlands is maladaptive as this release carbon dioxide sequestered in trees into the atmosphere. Although farmers in the district have the lake and other sources of surface water they have not effectively utilized these opportunities. Though there are no cultural or legal constrains, financial shenanigans and the undulating nature of the terrain in some parts of the Bosomtwe District have handicapped farmers' ability to effectively practice irrigation. Traditional ecological knowledge, the activities of agricultural officers, the extent of mobility among farmers of different age groups and the ability to diversity economic activities are key factors that influence perception and adaptation patterns among farmers of different age and educational backgrounds. Accordingly, policy on climate action should not be a one size fits all but malleable to so that it can easily be adapted to meet peculiar micro socio-economic conditions and disparities. A strong institutional environment to support adaptation with a focus on farmer-led participation and the security of farmers' livelihood is needed. It is imperative to implement and/or broaden policies that seek to directly or indirectly secure farmers' income such as crop insurance, subsidies, access to credit and other incentives that motivate farmers to adopt improved crop varieties. In order to curb such maladaptive practices of clearing of forests, government through the Forestry Commission and the Ministry of Food and Agriculture should introduce farmers to practices such as agroforestry that will sustain both forest and agriculture. There is the need to invest in intermediate technology that addresses challenges hindering the adoption of strategies such as efficient irrigation. Local knowledge should form the basis for the formulation and introduction of adaptation and mitigation activities and climate advocacy in the rural communities. This knowledge base is critical in determining how realistic adaptation practices are and the willingness of local farmers to adopt them in order to adapt to changing climatic conditions.