Introduction

Climate change poses challenges to agricultural systems and livelihoods (Alam et al., 2017; Challinor et al., 2014; Eshetu et al., 2016; Porter et al., 2014). Its unpredictable scale and pace impact the frequencies of drought, floods, new pests and diseases, reduced agricultural yields, crop failures, poverty, farm abandonment, migration, food and nutrition uncertainty across the globe (Connolly-Boutin & Smit, 2016; FAO, 2008; McCarthy et al., 2001; Porter et al., 2014). Climate change is not accurately detectable by human reasoning, but with the aid of appropriate climatological tools (Weber, 2010). Hence, the questions seeking to explore the farmers’ understanding of climate change indicators in this study were not exclusively definitive (increase or decrease) for climate variables such as rainfall and temperature. This is because farmers are likely to remember climate events, which are in the interim rather than climate variability, which is fairly long term (Hasan & Kumar, 2019). Although it is easier to measure the awareness of an individual with a dichotomous response with respect to climate change (Abidoye et al., 2017), perception is a more complex phenomenon that requires further expressivity in terms of the effects and causative factors (Alam et al., 2017). Moreover, the awareness and accurate perception of climate change among smallholder farmers is imperative to scale their ability to combat and mitigate the effects of climate change on farming activities (Fadina & Barjolle, 2018; Fankhauser & Tol, 1997; Maddison, 2007).

Awareness of climate change among smallholder farmers

From the findings of Ayanlade et al. (2018); Amusa et al. (2015); Arimi (2014); Oluwatayo and Ojo (2016) and Sofoluwe et al. (2011), most farmer respondents who participated in their studies on the effects of climate change in selected rural areas of Southwest Nigeria had heard about climate change. Across the agroecological divide, Onyekuru and Marchant (2017) and Idrisa et al. (2012), in their discoveries on climate change awareness among farmers found that a large proportion of the respondents in the sampled AEZs of Nigeria were aware of climate change. In other parts of Africa, there are reports of a high level of climate change awareness among smallholder farmers (Abera & Tesema, 2019; Ajuang et al., 2016; Elia, 2017; Elum et al., 2017) and agro-pastoralists (Ado et al., 2019).

Perception of climate change among smallholder farmers

Farmers across AEZs in Nigeria have viewed climate change as an erratic rainfall distribution and fluctuating temperatures (Ayanlade et al., 2017; Nnadi et al., 2019; Tambo & Abdoulaye, 2013). Some farmers experienced a reduction in crop yields, crop failures, pest incidence and disease occurrence, and drought as the most visible effects of climate change (Adebayo et al., 2012; Falola & Achem, 2017; Ihenacho et al., 2019; Issa et al., 2015; Ologeh et al., 2016; Tologbonse et al., 2010). Others opined that climate change induced poverty and ill-health as indirect effects (Ologeh et al., 2016; Tologbonse et al., 2010). They attributed their perceptions to varying causes, including deforestation, overgrazing, bush-burning activities, God, and industrial emissions (Idrisa et al., 2012).

Studies conducted in different AEZs in Kenya reported that farmers noticed increasing temperatures and declining precipitation levels, drying of rivers, hunger, a greater incidence of crop diseases, crop failures, dry spells, droughts and a decline in agricultural productivity are induced by climate change (Abera & Tesema, 2019; Asayehegn et al., 2017; Bryan et al., 2013; Evelyn et al., 2017; Kalungu & Harris, 2013; Ndamani & Watanabe, 2016; Ogalleh et al., 2012; Simotwo et al., 2018). In South Africa, Gbetibouo (2009) discovered that a large percentage of the study respondents perceived a decrease in the amount of rain in the past 20 years as evidence of climate change. Ubisi et al. (2017) reported that delayed rains, erratic rainfall, pest attacks, high temperatures, and longer dry spells, which subsequently resulted in poor crop yields, were consequent to climate change. Elum et al. (2017) found that farmers perceived climate change as the incidence of extreme temperatures, drought, storms, reductions in crop yields, and the increased occurrence of pests and diseases.

Determinants of the awareness and perception of climate change

Studies have identified several factors, such as the age of the farmers (Asrat & Simane, 2018), gender (Ayal & Leal Filho, 2017), farming experience (Gbetibouo, 2009; Okonya et al., 2013), level of education (Ajuang et al., 2016; Deressa et al., 2008), farm size (Abid et al., 2015; Ado et al., 2019), and land tenure (Abid et al., 2019; Roco et al., 2015), as the drivers of climate change awareness and perception among smallholder farmers. Measuring smallholder farmers’ awareness and perception of climate change indicators were the highlights of this study. The measurement of the understanding of climate change across AEZs provides a certain novelty in the information generated as climate change occurrence varies across climes (Otto et al., 2015). The results of this study are expected to generate new information and inspire consultations among the concerned stakeholders and policy makers in the agricultural and associated sectors. These consultations will hinge on how accurate climate change information and weather advisory services can be scaled-up and made available to smallholder farmers in different AEZs. This is because there is a high chance of distortions in the scientifically validated facts about climate change passed on to farmers (Weber, 2010).

Materials and methods

Study area

Southwest Nigeria, where Oyo state is located, consists of two main AEZs, namely, savannah and rainforest (Amujoyegbe & Alabi, 2012). Oyo state is a largely agrarian state comprising 33 local government areas with 28 suitable for agriculture (OYSADEP, 2017). It has a land area of 28,454 km2 (Abegunde et al., 2017; Tijani & Nurudeen, 2012). According to NPC (2006), it has a population of 5,580,894 people, with 2,802,432 and 2,778,462 females. Oyo state is rich in alluvial soils and notable for the production of food and tree crops. As illustrated in Fig. 1, Oyo state is largely covered by rainforest in the south and savannah up north (Akinniran et al., 2013; Sangotegbe et al., 2015).

Fig. 1
figure 1

Oyo State and its agroecological zones

Full size image

The savannah of Oyo state runs from Eruwa through Igana, Okeho to Saki block. The rainfall amount in the savannah ranges from 1040 to 1340 mm in a predominantly bimodal pattern. The savannah zone, is a transitional zone between the rainforest and the dry savannah of the north and is sometimes referred to as the food basket of Southwest Nigeria (Sangotegbe et al., 2015). It is suitable for the cultivation of tree crops such as shea butter, cashew, mango and arable crops such as maize, sorghum, cowpea, soybean and tomato (Abegunde et al., 2017).The rainforest zone largely covers the southern Ibadan area and extends up to parts of the Iddo/Eruwa axis. The rainfall pattern is bimodal with annual rainfall above 1340 mm (Oni et al., 2013). The rainforest is favourable for the cultivation of tree crops such as oil palm, cocoa, citrus, kola, mango, bush mango, and arable crops such as cassava, yam, banana, plantain, maize and pepper. Both AEZs are widely suited to poultry and livestock production (Babayemi et al., 2014).

Sampling procedure

A multistage sampling technique was used to select the representative smallholders during the study. The first stage was to purposively select Oyo state as the study area due to its abundance of agrarian local government areas. The second stage was to identify the AEZs for sampling, namely, the savannah and rainforest. The third stage involved selecting eight local government areas with high agricultural productivity: four in the savannah (Irepo, Iseyin, Saki East and Saki West) and four in the rainforest (Akinyele, Egbeda Lagelu, Oluyole). The fourth stage involved a random selection of 50 smallholder farmers in each local government area to give 400 respondents, 200 each in the savannah and rainforest. The random selection of 50 farmers from designated communities allowed for a fair representation of the respondents due to the unavailability of data on the population of registered farmers at the local government areas.

Data collection

Primary data were obtained through a structured questionnaire that was delivered in written format (English language). The questions were interpreted in the local languages (Catered for by the enumerators as the need arose, the respondents who did not understand English language understood either Yoruba or Pidgin English) where necessary, and it was ensured that all the respondents understood the purpose of the survey. The socioeconomic characteristics of the farmers and the level of awareness and perception of climate change were documented. The level of awareness among respondents was measured as aware (2) and not aware (1), and perception was measured on a 3-point scale of favourable (3), neutral (2), and unfavourable (1). The responses on the understanding climate change indicators were categorised on a 4- point scale of very low (0–25%), low to medium (26–50%), medium to high (51–75%), and high to very high (76–100%). The categorisation was tailored according to procedures indicated in Taneja et al. (2014); Khatri-Chhetri et al. (2017) and Ayanlade et al. (2018). Secondary data on climate parameters were obtained from the Nigerian Meteorological Agency (NIMET) Saki station, latitude 8.35°N, longitude 3.47°E representing the climatic conditions peculiar to the savannah, and the International Institute of Tropical Agriculture (IITA) Ibadan station, for the location latitude 7.57°N, longitude 3.9°E representing the climatic condition for the rainforest AEZ.

Description of empirical models

Principal components analysis (PCA)

Principal components analysis-PCA was used to generate the composite variables for the awareness and perception of climate change. Principal components analysis as specified by Koutsoyiannis (2001) is presented as:

Given variables (Xs…original variables of the composite awareness and perception of climate change).

X1…Xp measured in ‘n’ farmers.

P1… Pp: the principal components, which are uncorrelated linear combinations of the original variable, X1…Xp, given as:

$$\begin{aligned} &P_{1} = \alpha_{11} X_{1} + \alpha_{12} X_{2} + \cdots + \alpha_{1p} X_{p} \hfill \\ &P_{2} = \alpha_{21} X_{1} + \alpha_{22} X_{2} + \cdots + \alpha_{2p} X_{p} \hfill \\ &\vdots \quad \vdots \quad \vdots \quad \vdots \hfill \\ &P_{p} = \alpha_{p1} X_{1} + \alpha_{p2} X_{2} + \cdots + \alpha_{1pp} X_{pz} \hfill \\ \end{aligned}$$
(1)

where α = are the factor loadings, which fulfil the condition that the first component P1 represents the maximum proportion of the total variation in the independent variables. The variables with loadings greater than 0.3 are significant at 1%, where the sample size is more than 50 (Koutsoyiannis, 2001).

Tobit regression model

The Tobit regression model was used to determine the factors influencing the awareness and perception of climate change among the farmers using a PCA-generated vector as the dependent variable. The Tobit model ensures unbiased estimates by allowing censoring due to the discrete and continuous characteristics of the PCA-generated index which ordinary least square regression does not account for (McDonald & Moffitt, 1980).

The Tobit model is expressed as;

$$Y^{*} = Xi\beta + \varepsilon$$
(2)

Y* = dependent variable and a latent variable for the awareness and perception of climate change, β = vector of coefficients, Xi = vector of explanatory variables, ε = error term.

In the model, Y* is a latent variable and observed as a continuous variable when the vector generated for the latent variable above the limiting factor equals zero. The expression is stated as;

$$Y = Y^{*} \quad {\text{if}}\;Y^{*} > Y0$$
$$Y = 0\quad {\text{if}}\;Y^{*} \le Y0$$

where Y0 = limiting factor. The equations denote a censored distribution of the data.

Ey = value expected for the awareness and perception of climate change denoted as;

$$Ey = Xi\beta F\left( z \right) + \sigma f\left( z \right)$$

Xi = vector of explanatory variables.

F(z) = cumulative normal distribution of \(z\)where z = Xβ/σ, β = vector of Tobit maximum likelihood estimates, σ = standard error.

Description of the independent and latent variables in the Tobit model

The dependent and independent variables fitted in the Tobit regression model are highlighted in Table 1. It elicits the description and coding of the socioeconomic characteristics of the farmers in the study and the factors influencing the understanding of climate change. The dependent latent variables, i.e., the awareness and perception of climate change were generated using PCA (Hitayezu et al., 2017), and fitted in the Tobit model.

Table 1 Description of independent and dependent variables in the Tobit model
Full size table

Test for multicollinearity of the independent variables

Multicollinearity between the explanatory variables used in Tobit regression was measured using the variance inflation factor (VIF). The VIF was computed for each variable, which involved the estimation of the regression model. A VIF score of less than 5 indicates that the variables do not exhibit multicollinearity (Maddala, 2000).

The VIF is calculated as;

$${\text{VIF}}_{j} = \frac{1}{{1 - R_{j}^{2} }}$$
(3)

where VIFj is the variance inflation factor for the jth explanatory variable, \(R_{j}^{2}\) denotes the coefficient of determination with the jth independent variable, \(1 - R_{j}^{2}\) = tolerance.

Independent t-test

The independent t-test was used to compare the score of the awareness and perception of climate change indicators across the savannah and rainforest AEZs. The assumption that the variance in the sample mean will be equal to zero due to the same number of respondents (n = 200) in each AEZ is considered in the model (Kim, 2015).

$$t = \frac{{\overline{X}_{1} - \overline{X}_{2} }}{{\sqrt[{S\left( {1 + 2} \right)}]{{\frac{1}{{n_{1} }} + \frac{1}{{n_{2} }}}}}}$$
(4)

where \(\overline{X}_{1}\) = score of the awareness/perception of climate change in the savannah, \(\overline{X}_{2}\) = score of the awareness/perception of climate change in the rainforest, \(S\) 1 = variance of variable X1, S2 = variance of variable X2, n1 = sample size of respondents in the savannah, n2 = sample size of respondents in the rainforest.

Mann–Kendall trend test

The Mann–Kendall (M–K) trend test is a nonparametric test for trend detection and spatial variations with the ability to evade the problem of data skewness. The Mann–Kendall statistic (S) is commonly given according to procedures in Gilbert (1987);

$$S = \mathop \sum \limits_{i = 1}^{n - 1} \mathop \sum \limits_{j = i + 1}^{n} Sgn\left( {x_{j} - x_{i} } \right)$$
(5)

where Xi = time series ranked from 1, 2,…n − I, Xj = time series ranked from j = i + 1, 2,…n.

Each data point Xi is a reference point compared with the other data reference points Xj and expressed as follows:

$$Sgn\left( {x_{j} - x_{i} } \right) = \left\{ {\begin{array}{*{20}l} { + 1,} \hfill & { > \left( {x_{j} - x_{i} } \right)} \hfill \\ {0,} \hfill & { = \left( {x_{j} - x_{i} } \right)} \hfill \\ { - 1,} \hfill & { < \left( {x_{j} - x_{i} } \right)} \hfill \\ \end{array} } \right\}$$

The mean of S is E(S) = 0, and the variance of the S given as:

$$Var\left( {\text{S}} \right) = \frac{{n\left( {n - 1} \right)\left( {2n + 5} \right) - \mathop \sum \nolimits_{i = 1}^{m} t_{i} \left( i \right)\left( {i - 1} \right)\left( {2i + 5} \right)}}{18}$$

where m = the number of tied groups in the data, ti = number of data points in the ith-tied group.

The S statistic is normally distributed with the Z transformation;

$$Z = \left\{ {\begin{array}{*{20}c} {\frac{S - 1}{{\sqrt {Var\left( S \right)} }}} \\ {0. S = 0} \\ { \frac{S + 1}{{\sqrt {Var\left( S \right)} }},s < 0} \\ \end{array} } \right\}$$

The statistic S is also associated with Kendall’s tau (τ)

$$\tau = \frac{S}{D}$$

where

$$D = \left[ {\frac{1}{2}n \left( {n - 1} \right) - \frac{1}{2}\mathop \sum \limits_{j = 1}^{p} t_{j } \left( {t_{j} - 1} \right)} \right] ^{\frac{1}{2}} \left[ {\frac{1}{2}n\left( {n - 1} \right)} \right] ^{\frac{1}{2}}$$

The Sen slope estimator, which is used to determine the magnitude of trend (Sen, 1968), is expressed as:

$$T_{i} = \frac{{x_{j} - x_{k} }}{J - K}$$

where,

i = 1,2,…N.

Xj and Xk = data values at time j and k (j > k), respectively.

Kolmogorov–Smirnov test

The Kolmogorov–Smirnov (K–S) statistic is suitable for comparing two continuous and one-dimensional data sets. The two-sample variant used in this study employs the maximum absolute difference between two cumulative distribution functions (Dimitrova et al., 2017; Lopes et al., 2007). It compares whether the two data sets share identical distributions or otherwise. It is simply expressed as;

$$D_{ks } = max\left| {F\left( x \right) - G\left( x \right)} \right|$$
(6)

To further express the model;

$$D_{n1,n2} = \mathop {\max }\limits_{Z} \left| {F_{X,n1} \left( x \right) - F_{Y,n2} \left( x \right)} \right|$$

where \(\mathop {\max }\limits_{z}\) = maximum set of distances, n1 = sample size of data for X, n2 = sample size of data for Y, FX (x) = cumulative distribution function of X, FX (y) = cumulative distribution function of Y.

Coefficient of variation

The coefficient of variation-CV measures the change of a specific distribution in a set of variables and depends on the arithmetic mean values. It is given as;

$$V = \frac{S }{X}*100\%$$
(7)

where S = standard deviation, X = arithmetic mean of variables.

A CV value of < 20 shows low variability in rainfall (Hare, 2003) and < 2 for temperature.

Table 1 highlights the salient sociodemographic characteristics of the respondents. The farmers differed in their years of farming experience, farm size and land tenure system across AEZs. In the savannah, more respondents (35.0%) had 11–20 years of farming experience, while those in the rainforest (31.5%) had 21–30 years. More farmers (43.5%) in the savannah cultivated 2–5 ha of land, while a majority (88.0%) in the rainforest cultivated less than 2 ha. More farmers (44.0%) in the savannah inherited their farmland, while 41.5% in the rainforest purchased theirs. Across AEZs, respondents were predominantly male, Christian, Yoruba, married, completed primary education, employed less than 5 farm-hands, had household sizes of more than 5 and were older than 50 years.

Ethics approval and consent to participate

An ethical certificate (Ethics number—0031918A9), which affirmed that the study satisfied the requisites of ethical guidelines for research, was appraised by the Research Ethics Regulatory Committee (NWU-RERC) and approved by the Health Science Ethics Committee—FAST-HSEC at North-West University, South Africa. The respondents were duly informed of the objective of the research before the interviews were conducted. The anonymity of the respondents was ensured, with no indication of personal identifiers being shown on the questionnaire. Information gathered from the respondent was treated with a high level of confidentiality. The welfare of the respondents in the process of data collection was ensured, and no harm or injury was recorded. Interviews were conducted only after affirmative consent was given by the respondent; hence, participation was solely voluntary. Informed consent was obtained before and throughout the data collection process.

Results and discussion

Mann–Kendall trend test for variability in climate parameters within AEZs

No trend was observed in the rainfall distribution in either AEZ. However, significant trends were detected for minimum (P < 0.002) and maximum (P < 0.0001) temperatures in the savannah (Table 2 and Fig. 2). The results indicate a larger shift in the trends of temperature in the savannah compared to the rainforest AEZ.

Table 2 Mann–Kendall trend test showing the climatic trends across AEZs in Oyo state (1984–2016)
Full size table
Fig. 2
figure 2figure 2figure 2

Mann–Kendall trend test showing the climatic trends and Sen slope across agro-ecologies in Oyo state (1984–2016). a, b Rainfall c, d Maximum temperature e, f Minimum temperature. S = savannah agro-ecology, RF = rainforest agro-ecology

Full size image

Kolmogorov–Smirnov test for differences in climatic trends between AEZs

We found a significant difference in the distribution (Table 3 and Fig. 3) of rainfall (D = 0.455, P < 0.002) and minimum temperature (D = 0.727, P < 0.0001). The findings indicate that the trend of maximum temperature across the AEZs is not significantly different.

Table 3 Kolmogorov–Smirnov test for differences in climate trends between AEZs in Oyo state (1984–2016)
Full size table
Fig. 3
figure 3figure 3

Kolmogorov–Smirnov test showing the cumulative distribution of climatic trends across agro-ecologies in Oyo state (1984–2016). a Rainfall b Maximum temperature c Minimum temperature. S denotes savannah agro-ecology, RF denotes rainforest agro-ecology

Full size image

Further variations and trends of climate variables across AEZs

Further tests on the variations in the trend of climate data across AEZs are contained in Table 4. Low coefficients of variation (CV) were observed for rainfall in the savannah (15.1%) and rainforest (16.8%). The coefficients of variation for temperature across AEZs were also minimal.

Table 4 Further trend and variability in climate across AEZs (1984–2016)
Full size table

Smallholders’ awareness of climate change across AEZs

The respondents were asked whether they were aware of changes in climatic indicators (Fig. 4). The responses across the AEZs were found to be significantly different (t = -8.777, P = 0.000). To add context to the logic employed in the study and ensure a clear understanding of what was asked and the corresponding response, Table 5 highlights the definition of terms for variables used in measuring the understanding of climate change by farmers.

Fig. 4
figure 4

Awareness of climate change among smallholders across agroecological zones

Full size image
Table 5 Table explaining the variables for measuring the awareness and perception of climate change.
Full size table

Changes in rainfall pattern and intensity, changes in temperature, changes in heat levels

In the savannah, all the respondents (100%) were aware of changes in rainfall pattern and intensity. Nearly all of them (95.5%) were aware of changes in temperature and 78.5% of changes in heat levels. Similarly, in the rainforest, a large proportion of the respondents (92.5%) were aware of changes in rainfall patterns and 78.5% of changes in rainfall intensity. The majority were aware of changes in temperature (87.5%) and 67.5% of changes in heat levels. The high level of awareness of the indicators is connected with the temperature fluctuation observed from the trend analysis. Related findings are stated in the studies of Arimi (2014); Elum et al. (2017); Abera and Tesema (2019) and Hundera et al. (2019).

Increase in pest and disease incidence and resistance to control

Moreover, in the savannah, 91.5% of the respondents were aware of increasing crop and animal pest incidence and disease occurrence and their increasing resistance to control measures (83.5%). The majority of the farmers (74.0%) in the rainforest were aware of increasing pest and disease incidence and 69.0% of resistance to control. The respondents in the savannah were more aware of this set of climate change indicators. Increasing pest and disease incidence, such as the fall armyworm are indicators of climate change (Amusan & Olalekan, 2018).

Crop failure, reduction in crop yields, less fertile soils and reduction in organic matter

In the savannah, a larger proportion were aware of crop failures (93.0%), reduced crop yields (90.0%), less fertile soils (98%) and reduced organic matter in soils (89.0%). In the rainforest, 79.5% were aware of frequent crop failures, reduction in crop yields (80.5%), low soil fertility (81.5%), and loss of soil organic matter (76.5%). The harsh growing conditions meted out to crops as a result of erratic weather patterns are evidence of climate change observed by farmers (Issa et al., 2015; Vani, 2016).

Changes in seasons

In the savannah, the majority of the farmers (87.0%) viewed the prolonged dry season as an effect of climate change, 42.5% attested to extended periods of the wet season, and 56.0% were aware of a decline in the Harmattan season and warmer Harmattan season (46.0%). Similarly, among the respondents in the rainforest, 70.0% were aware of a prolonged dry season, 39.0% of a prolonged wet season, 51.0% of a reduced Harmattan season and 56.5% of a less dry/cold Harmattan season. Views on changes in seasonality are defining indicators of variability in climatic patterns among smallholder farmers (Amusa et al., 2015; Ihenacho et al., 2019; Mustafa et al., 2019).

Drier wetlands, groundwater shortages and receding rivers

Most respondents in the savannah (76.5%) were aware of wetlands (fadama/akuro) getting drier than usual, 93.0% of shortages in groundwater and river banks reducing by the day (72.0%). More (54.0%) of the farmers in the rainforest were aware of drier wetlands, groundwater shortages (69.5%) and receding rivers (62.5%). Previous studies establish that shortages in available water for farmers and animal and crop use signify a changing climate (Issa et al., 2015).

Torrential rainfall, flooding and excessive run-off

The respondents in the savannah were highly aware of the occurrence of torrential rainfall (74.5%), flooding (81.0%) and excessive run-off after downpours (91.0%) as signs of climate change. However, in the rainforest, 57.0% were aware of torrential rainfall occurrence, few (30.5%) flooding, and excessive run-off (39.0%). Similar reports of the awareness of excessive rainfall and flood occurrence as a result of climate change are reported by Mubaya et al. (2012) and Ihenacho et al. (2019).

Increase in strong winds, drought and desertification

Approximately half (53.0%) of the farmers in the savannah were aware of increasing scenarios of strong winds carrying dust particles, drought (75.5%), and desertification (85.5%). In the rainforest, 54.0% were aware of increasing strong winds, 49.5% of drought and 36.0% of desertification. Awareness of low precipitation and dry conditions as indicators of climate change among smallholders has been documented (Hundera et al., 2019; Mubaya et al., 2012).

A categorisation of the awareness of climate change across AEZs

In the savannah, none of the respondents had a very low awareness level of all climate change indicators. Very few respondents (0.5%) had a low to medium awareness of climate change, while 38.0% had a medium to high awareness. Furthermore, 61.5% of the respondents had a very high awareness of climate change. However, in the rainforest, few respondents (7.5%) had a very low awareness level, 18.0% had a low to medium awareness, 41.5% had a medium to high awareness and 33.0% had a very high awareness of the indicators of climate change (Fig. 5). The respondents in the savannah were more aware of climate change indicators than their counterparts in the rainforest. This trend can be attributed to climate variability being more evident in the savannah AEZ, as indicated in Tables 2, 3 and 4.

Fig. 5
figure 5

Categorisation of awareness of climate change across agroecological zones

Full size image

Smallholder’s perception of climate change across AEZ

A significant difference was observed in the responses by the farmers across AEZs (t = -9.338, P = 0.000), which posits that climate variability and its corresponding views by the respondents were different across the AEZs (Fig. 6).

Fig. 6
figure 6

Perception of climate change among smallholders across agroecological zones

Full size image

Changes in rainfall pattern, rainfall intensity, temperature and heat levels

In the savannah, all the respondents (100.0%) favoured the perception that changes in rainfall pattern and intensity were due to climate change. A large proportion of the farmers (73.5%) had highly perceived that changes in temperature and heat levels (74.5%) were the results of a changing climate. However, in the rainforest, 73.5% of them perceived changes in rainfall pattern, changes in rainfall intensity (62.5%), changes in temperature (72.0%) and changes in heat levels (53.5%) were the consequences of climate change. Studies have indicated similar reports on farmers’ perceptions of climate change (Ayal & Leal Filho, 2017; Ayanlade et al., 2018; Fosu-Mensah et al., 2012; Mandleni & Anim, 2011; Simotwo et al., 2018; Sofoluwe et al., 2011; Tambo & Abdoulaye, 2013; Tologbonse et al., 2010; Zizinga et al., 2017).

Increase in pest and disease occurrence/resistance

In the savannah, 89.0% had a favourable perception of increasing pest occurrence and disease incidence, and 82.0% perceived increasing resistance to control measures. Moreover, 72.5% of the respondents in the rainforest perceived the increasing pest occurrence and disease incidence, and their resistance to control measures (62.0%) was a result of climate change. Previous reports are in agreement with the results of this study (Anyoha et al., 2013; Asrat & Simane, 2018; Callo-Concha, 2018; Ogalleh et al., 2012).

Crop failure, reduced crop yields, less fertile soil and loss of organic matter

A large proportion of the respondents (95.0%) in the savannah favourably perceived climate change as crop failure, a reduction in crop yields (91.5%), less fertile soils (97.5%) and loss of soil organic matter (89.0%). In the rainforest, 71.0% had a favourable perception of crop failure, a reduction in crop yields (70.0%), soil fertility loss (67.5%) and loss of soil organic matter (61.5%). Similar results were obtained by Anyoha et al. (2013); Roco et al. (2015); Mbah et al. (2016); Ndamani and Watanabe (2016); Falola and Achem (2017); Callo-Concha (2018) and Simotwo et al. (2018).

Changes in seasons

The respondents in the savannah perceived a prolonged dry season (85.5%), a prolonged rainy season (47.0%), a reduced Harmattan season (57.0%) and a warmer Harmattan season (45.0%) as indicators of climate change. In the rainforest, 52.0% had a favourable perception of a prolonged dry season, a prolonged rainy season (35.5%), a reduced Harmattan season (39.0%) and a warmer Harmattan season (40.0%). These findings are consistent with those of prior studies (Asrat & Simane, 2018; Enete et al., 2011; Fadina & Barjolle, 2018; Iheke & Agodike, 2016; Menike & Arachchi, 2016).

Drier wetlands, groundwater shortages and receding rivers

In the savannah, 73.0% had a favourable perception of drier wetlands, shortages in groundwater (89.0%), and receding rivers (66.5%). In the rainforest, 52.0% favourably perceived drier wetlands, groundwater shortages (53.5%) and receding rivers (58.5%) as being caused by climate change. Similar reports from Abera and Tesema (2019); Mugari et al. (2016); Mbah et al. (2016); Iheke and Agodike (2016); Alam et al. (2017); Jiri et al. (2017); Ansari et al. (2018) and Tiwari et al. (2010) are in tandem with these results.

Torrential rainfall, flooding and excessive run-off

Torrential rainfall (65.0%), flooding (76.0%) and excessive run-off (91.5%) were considered the effects of climate change by farmers in the savannah. On the other hand, 47.0% of those in the rainforest had a favourable perception of torrential rainfall, flooding (30.0%) and excessive run-off (31.5%). Several studies have reported similar findings (Alam et al., 2017; Ansari et al., 2018; Callo-Concha, 2018; Mbah et al., 2016; Mkonda et al., 2018; Nnadi et al., 2019; Tiwari et al., 2010).

Increase in strong winds, drought and desertification

Approximately half of the respondents in the savannah (50.5%) attest to the increase in strong winds carrying dust particles, drought (73.5%) and desertification (82.0%) to be a result of climate change. In the rainforest, 43.5% perceived an increase in strong winds, drought (44.5%) and desertification (28.0%). Supporting these results are the findings of Apata et al. (2009); Nnadi et al. (2019); Mugari et al. (2016); Iheke and Agodike (2016); Menike and Arachchi (2016); Falola and Achem (2017) and Ogalleh et al. (2012).

A categorisation of the favourable perception of climate change across the AEZs

In the savannah, none of the respondents had a low level of favourable perception of the indicators of climate change. Very few respondents (1.0%) had a low to medium perception, 35.5% had a medium to high perception and 63.5% had a very high and favourable perception. In the rainforest, none of the respondents had a very low perception level. However, 10.5% of the respondents had a low to medium perception, 55.5% had a medium to high favourable perception and 34.0% had a very high and favourable perception of the indicators of climate change. More farmers in the savannah had a favourable perception of climate change indicators than those in the rainforest (Fig. 7). Similar findings were posited by Asrat and Simane (2018).

Fig. 7
figure 7

Categorisation of the favourable perception of climate change across agroecological zones

Full size image

Principal components analysis of smallholders’ awareness of climate change

The matrices in the analysis were extracted using a correlation method (Krzanowski, 1987). To indicate a positive degree of relationship, only values in loadings greater than 0.30 were selected (Koutsoyiannis, 2001). Communality, which shows the association between variables was also indicated (Table 6). An overall reliability score of 0.86 (Cronbach’s alpha) was recorded, which indicates that the questionnaire instrument has high internal consistency and that submissions by the respondents were reliable and suitable for PCA. This submission was confirmed by the results of Bartlett’s Test of Sphericity with a value of X2 = 3325.115, P < 0.01. The Kaiser–Meyer–Olkin measure for sampling adequacy value of 0.809 was obtained. This shows that the sample size used in the study was adequate. The Kaiser criterion (Kaiser, 1960) was used to select factors relevant to the data. Furthermore, seven factors with eigenvalues greater than 1 were retained in the model.

Table 6 Principal component analysis for the determinants of climate change awareness among smallholders
Full size table

Factor 1, accounting for 26.212% of the variance, represents changes in precipitation patterns, water availability and soil fertility. All the coefficients observed were positive, with only changes in rainfall pattern and temperature scoring less than 0.30 for the factors loaded. Factor 2, accounting for 10.406% of the variance, represents the vector of changes in precipitation pattern, pest and disease incidence. It had positive values loaded with respect to changes in rainfall pattern, the increased occurrence of pests and diseases, their resistance to control measures, and less-fertile soils. Factor 3, representing the vectors of climate change awareness among the respondents in terms of changes in temperature and water availability, accounted for 7.470% of the variance. Factor 4 accounted for 5.781% of the variance, with positive factor loadings for prolonged rainy season, less-fertile soils, and excessive-run off. It represents the vector of water availability and soil fertility. Factor 5 accounted for 5.205% of the variance, with changes in rainfall pattern, flooding and desertification representing changes in precipitation and water availability. Factor 6 accounted for 4.731% of the variance, representing pests and disease incidence, while Factor 7 accounted for 4.470% of the variance representing changes in precipitation pattern and temperature. Largely, the seven factors that were retained accounted for 64.275% of the variation in the matrix, signifying the smallholders’ awareness of climate change in the study area.

Changes in rainfall intensity, crop failures and groundwater shortages were distinguishable factors in assessing climate change awareness among respondents and loaded under only one component with values greater than 0.3. Due to their heavy dependence on rain-fed agriculture (Cooper et al., 2008; IFAD, 2011), occasioning severe effects on their rural livelihoods and farm incomes, smallholders have been vulnerable to fluctuations in rainfall. In relation to the consequences of changes related to rainfall intensity, projected increases in Nigeria’s population, coupled with the resultant escalation in the demand for food and climatic uncertainties, will deepen food insecurity. This scenario will play out unless proper adaptations and mitigation of agricultural practices to climate change are borne out of a clear-sighted awareness of change in climate and its effects. That indigenous farmers in crop production should understand rainfall trends is imperative for them to adapt to and effectively deal with climate change (Hartter et al., 2012). However, another relevant factor signaling smallholders’ awareness of climate change in the matrix is crop failures. Climate-induced crop failures are caused by incessant rainfall, unexpected dry spells and temperature fluctuations (Morton, 2007). Closely aligned to crop failures is the issue of the availability of water. The unavailability of water is a foremost limiting factor to smallholder farming systems in Sub-Saharan Africa (IAC, 2004). Farmers in parts of Nigeria have discovered that the water retained in the pores and spaces in the soils that is essential for crop growth is rapidly declining, even after rainfall events, thus indicating the effects of climate change (Enete et al., 2011).

Principal components analysis of smallholders’ perception of climate change

Bartlett’s test of sphericity value was X2 = 3749.906, P ≤ 0.01, and the Kaiser–Meyer–Olkin measure of sampling adequacy was 0.791. An overall reliability score of 0.83 (Cronbach’s alpha) was recorded (Table 7). Factor 1 accounted for 23.131% of the variation, and all coefficients were represented in the factors loaded, except for changes in heat levels. Factors 2 and 3 accounted for 13.466% and 9.570% of the variance, respectively. Factor 2 represents the vector of seasonality and water shortage, while Factor 3 denotes changes in temperature and water availability. Factor 4 accounted for 6.83% of the variance. It represents the vector of warmer Harmattan season. Factor 5 accounted for 5.905%, with high positive coefficients for pest and disease incidence and resistance to control measures. Factor 6 accounted for 5.09% of the variance, with a positive coefficient loading of excessive runoff, which denotes changes in water availability. Factor 7 accounted for 4.47% of the variance, loading changes in temperature, and flooding.

Table 7 Principal component analysis for the determinants of climate change perceptions among smallholders
Full size table

The seven components that were retained accounted for 68.45% of the variation in the smallholders’ perception of climate change. The outstanding indicators of climate change perception were prolonged dry season and groundwater shortages. Increasing frequencies in the late onset of rainfall and extended dry periods were found to be characteristic features that indicate the effects of climate change, with drastic impacts expected on the timing of farming activities and agricultural yields across the rain-fed farming systems. The adoption of water-smart practices such as irrigation schedules and water harvesting, which are generally poorly used among smallholders (Evelyn et al., 2017), possesses the potential to assist farmers in accessing effective adaptation measures to apply to deal with extended dry periods.

Determinants of awareness and perception of climate change among smallholders

The results show that there was a significant influence of the farmer’s age, AEZ, ethnic group, religion, educational level, farm-labour size, off-farm activities, farming experience, climate variables, disease incidence and food security on the farmers’ understanding of climate change (Table 8). We obtained mean VIFs of 1.48 and 1.49 for awareness and perception of climate change, respectively. These connotes that no collinearity was observed between the independent variables used in the model (Maddala, 2000). The high level of tolerance also indicates the absence of multicollinearity. The probability chi-square value of 0.0000 for determinants of the awareness and perception of climate change was significant at P < 0.01, showing that the model was well fitted. The pseudo R squared values were 0.097 and 0.268, respectively, and connotes that 9.7% and 27% of the variation in the model was due to the independent variables.

Table 8 Determinants of climate change awareness and perception among smallholder farmers
Full size table

The coefficient representing the age of an individual farmer (0.026, P < 0.01) indicates that it has a positive influence on his perception of climate across AEZs. This result connotes that farmers’ perceptions of climate change indicators are more accurate as they advance in age. This deduction is due to reasons that such a farmer may have experienced significant variations in climatic trends over several years. Similar results were presented by Ayal and Leal Filho (2017); Asrat and Simane (2018) and Ishaya and Abaje (2008). However, the corresponding years of farming experience (−0.025, P < 0.01) exhibited a negative relationship with the perception of climate change. This implies that the longer years spent in farming system operations did not necessarily translate into favourable perception of climate change indicators across AEZs. This result indicates that the younger farmers were better able to perceive climate change. This result is in line with Okonya et al. (2013), who reported a contrasting relationship between farming experience and the awareness of climate change. However, Maddison (2007); Gbetibouo (2009) and Mustafa et al. (2019) reported a positive relationship between farmers’ years of farming experience, and awareness and perception of climate change.

The gender distribution of respondents (−0.533, P < 0.10) showed a negative relationship with awareness of climate change. This implies that female farmers were advantaged at perceiving climate change indicators compared to their male counterparts. This result is supported by the findings of Ayal and Leal Filho (2017), who stated that females are more aware and perceptive of climate change in their locality, despite the absence of gender differences in perception abilities (Swai et al., 2012). The coefficient of AEZ shows a highly significant and positive influence on the awareness (1.415, P < 0.01) and perception (0.988, P < 0.01) of climate change by smallholders. This result shows that farmers in the savannah have a better awareness and perception level of climate change than those in the rainforest. This result is further explained by the findings detailing the trend in the climate across AEZs (Tables 2, 3, 4). The result is corroborated by the findings of Addisu et al. (2016) and Hitayezu et al. (2017). The authors stated that location and AEZ influenced the perception of climate change by farmers.

The religious affiliation of the farmers revealed a positive influence on their awareness (0.492, P < 0.05) and perception (0.262, P < 0.10) of climate change. The information available to farmers during religious gatherings and sensitisations on crucial issues can increase their chances of being aware of the indicators of climate change in their respective AEZs. Few authors have reported the influence of religion and belief systems on issues about climate change (Karki et al., 2020; Morrison et al., 2015; Schuman et al., 2018; Spear et al., 2019; Wang & Kim, 2018). The coefficient of ethnic group (0.515, P < 0.01) was positively correlated with the perception of climate change. This result is explained by the majority of the respondents across both AEZs being of the same ethnic group. Farmers of the same ethnic group can be advantageous with regard to their ability to perceive climate change indicators. This is plausible through the ease of understanding media messages, agro-weather services and extension communique, and discussions on climate change among fellow farmers, family members, acquaintances and the larger society. Some reports have been presented on the influence of ethnicity on topical issues, including climate change (Dhungana & Khand, 2019; Tripathi & Mishra, 2017).

Furthermore, the perception of climate change was negatively influenced by smaller farm sizes cultivated by farmers (−0.051, P < 0.01). The difficulties and financial stress associated with effective climate change adaptation can force farmers to reduce the farmland they cultivate as a coping strategy for climate vagaries. The scenario is therefore that “the smaller the farm plot, the higher the perception of climate change”. Previous reports indicate that farmers use a reduction in farm size and farm abandonment to cope with climate change in Nigeria (Onyeneke et al., 2018; Popoola, 2019). However, this trend portends food insecurity and loss of agricultural livelihood for smallholder farmers who cultivate small farm sizes and practise subsistence agriculture, which are largely rain-fed. The reports of Ado et al. (2019) support our findings, where it was stated that farmers with smaller farm sizes are more aware of climate change. However, Abid et al. (2015); Abid et al. (2019) and Mustafa et al. (2019) reported otherwise, stating that farmers with larger farm holdings perceive climate change better than those with smaller farm sizes.

The educational attainment of the respondents (−0.651, P < 0.01) showed that there was an indirect relationship with the perception of climate change. This means that being educated did not influence the perception of climate change indicators among the farmers. Mandleni and Anim (2011) reported similar findings. However, Deressa et al. (2008); Gbetibouo (2009); Ajuang et al. (2016) and Mustafa et al. (2019) reported that the educational level of farmers significantly influenced their perception of climate change. The need for more farm hands increased the farmers’ perception of climate change (0.147, P < 0.01). This submission is due to reasons that there would be a need to increase the farm labour pool for the deployment of climate change adaptation and mitigation practices with an increase in perception levels. The strategy of increasing the farm labour pool by farmers is geared towards cultivating more expanded land to increase the number of climate-smart agricultural practices used to cushion the effects of climate change. The lack of farm hands among smallholder farmers has hampered agricultural production activities (Odoemenem & Obinne, 2010; Oluwasusi & Tijani, 2013). The land tenure systems among the farmers were found to exhibit casual support for their awareness (−0.426, P < 0.10) and perception (−0.467, P < 0.01) of climate change. In this context, the ownership of land did not translate into the ability to be aware of and perceive climate change. Abid et al. (2019) and Roco et al. (2015) reported that land ownership among smallholder farmers shaped their perception of climate change.

Temperature (1.121, P < 0.10) and rainfall distribution (1.901, P < 0.01) across AEZs were defining variables that shaped the respondents’ awareness of climate change. Soaring temperatures and erratic rainfall significantly influence farmers’ views of climate change (Ayal & Leal Filho, 2017; Hundera et al., 2019; Zizinga et al., 2017). This result is further corroborated by the variation observed in the rainfall distribution across both AEZs of the state in Table 4. A shift in trend for temperature figures observed in the savannah and rainforest further validates the respondent’s awareness levels of climate change (Tables 2, 3, 4).

The increasing incidence of crop and animal pests (−0.740, P < 0.01) and diseases (1.709, P < 0.01) was found to be a highly significant determinant of their perception and awareness of climate change, respectively. A decrease in pest occurrence did not increase the perception of climate change, while an increase in disease incidence increased the awareness of climate variability by the farmers. This result is further corroborated by the incidence of fall armyworm and locust menace and other new pests and diseases ravaging smallholder farms across SSA (Amusan & Olalekan, 2018). Additionally, the need to be food secure had a significant influence on the perception (0.637, P < 0.05) of climate change among the farmers. The more availability of, accessibility to, utilisation and stability of food among smallholder households, the more farmers’ ability to perceive the effects of climate change on food and farming systems. Similarly, awareness of climate change (0.588, P < 0.01) was found to drive farmers’ perceptions of the same. Hence, farmers aware of climate change indicators also have a high perception of the same phenomenon.

Matching climate variability with farmers’ awareness and perception of climate change

The awareness and perception of climate change among the farmers in the two AEZs were influenced by the variability in climate parameters (rainfall and temperature). Hence, farmers largely indicated in their awareness and perception levels that changes in rainfall pattern and intensity, changes in temperature and heat levels, drought, prolonged dry season, warmer Harmattan season and groundwater shortages were consequences of climate change. The wider scale of a statistically significant difference in the trends observed among multiple climate variables for the savannah (Tables 2, 3, 4; Figs. 2, 3) clarifies the larger proportion of respondents being aware of and having favourable perception of more climate change indicators than the farmers in the rainforest. Being a farmer in the savannah influenced the ability to be aware of and perceive climate variability, as discovered from the Tobit regression output (Table 8). Moreover, the Tobit regression results detailed the influence of temperature and rainfall as factors that influenced the perception of climate change by farmers.

Similar reports on the relationship between farmers’ perception of climate change and variability in climate factors are contained in the reports of studies conducted across Sub-Saharan Africa (Asare-Nuamah & Botchway, 2019; Ayanlade et al., 2017; Fosu-Mensah et al., 2012; Hundera et al., 2019; Maddison, 2007; Mkonda et al., 2018). The perception of climate change as opined by Maddison (2007), influences adaptation efforts and the activities put in place to cushion the associated climate vagaries. The understanding of climate change among farmers as occasioned by variability in different AEZs significantly influences the climate-smart agricultural practices (Fadina & Barjolle, 2018) used as adaptive and mitigation measures by farmers.

Conclusions

This study revealed a variability in climate within and across both AEZs; hence, a conclusion was drawn that climate change has occurred in Oyo state. The outcome of the study reveals that the climatic changes occurring in the savannah were more poignant and different from those in the rainforest. Changes in rainfall intensity, crop failure, prolonged dry season and groundwater shortage were important components of smallholder farmers’ awareness and perception of climate. We posit that this result is consequent to the experiences the farmers have had with observing changes in the rainfall and temperature trends in both AEZs over time. The significant difference observed for variability in the climatic data for the savannah explains why the farmers in the AEZ had a higher awareness and perception level of climate variability compared to their counterparts in the rainforest. The age of smallholder farmers, agroecological zonation, educational level, farm labour size, climate variables, disease incidence, ethnic group and religious affiliation were significant determinants of the awareness and perception of climate change among smallholder farmers in both AEZs.

Since the effective adaptation to climate change is shaped by a proper understanding of climate change (Deressa et al., 2009; Gbetibouo, 2009; Maddison, 2007; Ndamani & Watanabe, 2016), we conclude and recommend that smallholder farmers across the AEZs of Oyo state can be better advantaged to attain climate-smart agriculture. This feat is realisable if they had access to more accurate agro-weather information (especially the dynamics of variation in rainfall and temperature) in their local languages. Harping on the potential of availing the religious affiliation of smallholders to disseminate information on channelling their farming and adaptation activities to be in tandem with the proper understanding of climate change and what it portends is expedient for effective policy formulation. Last, the scaling-up of locally available agro-weather information possesses the impetus to improve the proper planning of agricultural activities and the synergy in information sharing between farmers, meteorological agencies and agro-weather advisory service providers.