Traditional practices, knowledge and perceptions of fire use in a West African savanna parkland

Understanding people’s fire practices, knowledge and perceptions of the use of fire and fire regimes can inform fire management plans that could contribute to sustainable savanna conservation and management. We investigated the frequency of fire use, control and perceptions of fire regime for selected livelihood and socio-cultural activities in six districts in the Guinea savanna of Ghana. The majority of respondents (83%) across the study districts indicated that they used fire once a year for at least one of the following activities: land preparation, weed/grass/pest control, burning stubble after harvest, bush clearing around homesteads, firebreaks, charcoal burning and hunting. The study showed a higher frequency of fire use in the dry season for land preparation for cropping. Less than a fifth of the respondents (17%) indicated that they do not use fire for any of the above activities. The majority of respondents (62%) across the districts mentioned that they controlled their use of fire to prevent destruction to property, with the remaining 3% who indicated the prevention of killing or injuring humans. The study showed a higher frequency of fire use for land preparation for cropping than for the other socio-cultural activities. However, respondents rated season of burning as the most important attribute, with little attention to the other attributes of a fire regime, contrary to what is theoretically recognized. Understanding traditional fire use practices in terms how to regulate the mix of frequency, intensity/severity, season, size and type of fire for these and other socio-cultural purposes could enhance sustainable savanna conservation and management. There is a need to unravel the specifics of fire assisted socio-cultural practices and fire regimes in West Africa.

selected cultural and livelihood activities, fire control practice and the perceptions of fire regimes as 74 defined by Bond and van Wilgen (1996) and Keeley (2009)  use fire to for other purposes including clearing bushes around their homes. 86 The study was conducted in the Guinea savanna ecological zone, Ghana (9.5439° N, 0.9057° W). The 87 climate of the region is tropical with a unimodal rainfall distribution with an annual mean of 1 100 mm 88 (Ghana Meteorological Service 2017). Thus, the region has only one cropping season (Gyasi 1995). 89 The peak of the rainy season ranges between July and September with the rainfall exceeding potential 90 evaporation over a relatively short period (Boubacar et al. 2005). The mean annual temperature is 91 27 o C. The region records comparatively higher annual potential open-water evaporation of 2 000 mm 92 than the south, which records 1 350 mm. Due to its proximity to the Sahel, the region experiences dry, 117 Among the major crops grown are maize (Zea mays), millet (Panicum miliaceum), rice (Oryza sativa), 118 yam (Dioscorea spp.), and various pulses and vegetables (Gyasi, Kranjac-Berisavljevic, Blay 2004; 119 Brookfield and Gyasi 2009). Cropping is also done around the home compounds in the rainy season 120 and in valleys or along water bodies, especially in the dry season (Gyasi et al. 2004). These compound 121 farms are usually permanent, because the soils are replenished by the continuous supply of household 122 waste and manure from livestock (Karbo and Agyare, 1997). They are also burned annually to control 123 ticks and reptiles, as well as for visibility purposes (Gyasi, Kranjac-Berisavljevic and Blay 2004). 124 There are considerable areas of parkland which have not been cultivated, either because of low soil 125 fertility or they have been left fallow for a very long period (Oppong-Anane, 2006). 126 The Northern region is leading in livestock (ruminants and poultry) production in the country. 127 Livestock is kept on both free range and semi-intensive systems by households and hired herdsman 128 (Fulani), respectively (Karbo and Agyare 1997). Other socio-economic activities are agro-processing 129 (e.g. rice and groundnut processing) and the processing of non-timber forest products (NTFPs) (e.g. 130 shea butter processing).

143
Of the 18 districts, six were selected for the survey, in which most of the communities were randomly  The open-ended questions were used to ascertain the activities for which people use fire, people's 162 practices and knowledge of the use of fire, the impacts of fire on the environment, as well as their 163 knowledge of fire management. The activities (land preparation, hunting, firebreaks, bush clearing 164 around homesteads, and burning stubble after harvest) were selected based on literature on fire use in 165 savannas (Nyongesa and Vacik 2018). The data were collected with the assistance of a research team 166 selected from the University for Development Studies in Tamale, who are fluent speakers of the 167 language spoken in the selected districts. The research team were all males and comprised of five 168 assistants for East and West Gonja Districts, four for West Mamprusi and five for Tolon, Kubungu,169 Yendi and Tamale South districts. They were trained on the objectives and some terminologies of the 170 study. Households were selected randomly; however, either a male (mostly heads of households) or a 171 female who was willing to answer the questions on fire use was interviewed in a household in each of 172 the communities. Stern (2000)  any of the questions they could discontinue answering the questionnaire without any negative 185 consequences. We also informed the data will be analysed anonymously and all the information will 186 be stored in a secure space where only the lead researchers and the supervisor have access. Participants 187 were made to understand that the responses were for an academic purpose and I showed my Identity

188
Card to the assembly persons as a proof of my status as a Student. Participants were also assured of 189 confidentiality. All the interviews were conducted in the local dialects of the selected districts as the 190 majority of the respondents could not speak and understand English. To provide an overview of socio-demographic characteristics, perceptions, knowledge and practices 193 and respondents's use of fire, data were analysed using both descriptive and inferential statistics.  The highest number of respondents (79%) indicated they were farmers, with only 6% being students; 208 and the remaining 15% engaged in other forms of activities. The mean age of respondents was 39±15 209 years, with a minimum of 18 years and maximum age of 94 years. About a quarter of the respondents, 210 which were in the majority (21%), were between 26 and 32 years and the lowest (8%) were within the 211 age brackets of 47 and 53 years. The majority of respondents (66%) have no formal education and less 212 than a quarter of the total number of respondents (15%) had education till junior high and senior 213 schools (Table 1). The mean household size was 12.2±8 persons, with a minimum of 1 and maximum of 70 persons per  The majority of respondents (83%) across the study districts indicated that they used fire for at least 221 one of the selected activities: land preparation, weed/grass/pest control, burning stubble after harvest, 222 bush clearing around homesteads, firebreaks, charcoal burning and hunting (Table 2). Less than a fifth 223 17% of said that they do not use fire in these activities. There were varied responses on how often fire 224 was used for the selected activities across the high, moderate and low fire frequency districts. There 225 was a significant association between fire frequency and land preparation for crop production amongst the three categorised districts (χ2 = 39.5, df = 9, p < 0.001). The majority of respondents in the high 227 (86%), moderate (85%) and low (75%) fire frequency districts indicated they used fire once a year for 228 land preparation, while as low as 8%, in both high and moderate and 2% in low fire frequency districts, 229 used fire twice a year for land preparation (Table 2). On average, 14% of the respondents across the 230 districts never used fire for land preparation.

231
The was a significant association between the frequency of fire used for weed control and the district 232 (χ 2 = 66.6, df = 9, p < 0.001). On average, 13% of the respondents used fire once a year for weed 233 control, while the majority (77%) never used fire for this purpose (Table 2).About a fifth (21%) of the 234 respondents in the high-frequency zone, 11% in the moderate zone and 6% in the low zone, used fire 235 once a year for weed control. In contrast, 73%, 77% and 93% of respondents in the high, moderate and 236 low fire frequency zones, respectively, never used fire for weed control.

237
The majority of respondents (79%) never used fire to burn stubble after harvest, while less than a fifth 238 (15%) used fire once a year. A little over half of the respondents in the high (52%) and low fire districts 239 indicated that they burn stubbles after harvest. There was a significant association of frequency of fire 240 use for stubble burning and districts (χ 2 = 74.8, df = 9, p < 0.001).

241
Bush clearing around homesteads also showed strong association (χ 2 = 50.3 df = 9, p < 0.001) amongst 242 the districts. However, 67% of the respondents never used fire for clearing around homesteads. Twenty 243 percent and 13% of the respondents used fire once and twice a year respectively, across the districts.

244
However, about half (46%) of respondents in high fire frequency districts and as low as 9% 245 respondents in low fire frequency district said they used fire once a year. Whereas 45% in high, 60% 246 in moderate and 80% in low fire frequency districts never used fire for the same activity (Table 2).

247
Respondents' use of fire to create firebreaks showed a strong significant association (χ 2 = 50.33 df = 248 9, p < 0.001) among the districts. On average, the majority (70%) of the respondents in the moderate 249 and low fire districts never used fire for fire breaks. Nearly three-quarters (70%) of the respondents in 250 high fire districts, a little over a fourth of respondents in moderate (27%) and low (28%) fire frequency 251 districts used fire once a year for fire breaks.

252
Similarly, there was a significant association of fire use for hunting (χ 2 = 26.88, df = 9, p < 0.003) 253 across the study districts. On average, 84% of the respondents across the districts never used fire for 254 hunting, with as low as 7% in the high, 11% in the moderate and 6% in the low fire frequency districts 255 used fire once a year (Table 2).

277
Contrary to the control of fire for land preparation, 82% of the respondents in each of the districts 278 never controlled fire for weed control, but only a fifth of respondents in the high (10%) moderate (9%), 279 and low (6%) fire frequency districts always controlled fire (χ 2 = 48.5, df = 9, p < 0.001).

280
However, 59% of the respondents in the high fire frequency districts and less than a tenth of the 281 respondents in the moderate (3%) and low (6%) districts, always controlled fire for burning stubble.

284
As much as twice the number of respondents (65%) in high fire districts always controlled fire created 285 for firebreaks than those in low fire frequency districts (31%), with only 6% in the moderate fire 286 frequency districts (χ 2 = 159.7, df = 9, p < 0.001).

287
The highest number of respondents in all the districts indicated that they never controlled fire for 288 hunting, whilst less than a fifth of the respondents in the districts, controlled fire always (Table 3).

289
During charcoal burning, respondents in the high fire (64%), moderate fire (16%) and low fire districts 290 (23%) always controlled fire, whereas 26%, 79% and 76% in the high, moderate and low fire frequency 291 districts, respectively, never controlled fire (Table 3).  There were varied perceptions of fire regimes. There was evidence of significant relationships in the 299 perceptions of fire regimes in the high, moderate and low fire frequency districts (Table 4). Half of the 300 respondents (50%) in the high fire frequency districts, 33% in the moderate and 43% in the low, were 301 of the opinion that the season of burning was very important. While less than a fifth in each of the 302 districts said the season of burning was not important.

303
The majority of respondents perceived the severity of fire as important but not very important: 45%,

304
41% and 28% in high, moderate and low respectively, indicated that the severity of fire was important.

305
However, more than half of the number of respondents (53%) in the low fire districts, 33% of the 306 respondents in high and 44% in moderate districts indicated that the severity of fire was not important.  About 77% of the respondents were of the view that fire was not good for the environment, 8% thought 318 fire use was good for the environment while the remaining 14% indicated they did not know. However, 319 thirty-five percent of the respondents indicated that burning was good for the environment, said that 320 burning increased soil fertility, followed by 28% who were of the view that it improved crop yield, 321 while less than 1% said it was good for hunting. In contrast, 42% of respondents who said that burning 322 was not good, were of the view that fire destroyed vegetation, while 33% were of the opinion that fire 323 decreased soil fertility with as few as 1% stating it exposed animals to hunting (Figure 3).  .003). However, a negative correlation was found between perception of fire effect on the environment 329 and frequency of fire use for firebreaks, bush clearing around homesteads and hunting (Table 5).

374
Prescribed fire use in Ngorongoro, Kenya was also successful in the control of ticks (Fyumagwa et al. 375 2007). However, the use of weedicides and herbicides is probably gaining ground because of their 376 ready availability in local markets and in remote areas of Ghana (Nkansah 2014). Some respondents 377 explained that it is difficult to apply fire while crops are still in the field and would prefer to use 378 chemicals and rather burn for land preparation. This could also be the reason for the relatively high 379 responses for fire use after harvesting for burning stubble, which sometimes forms part of the process 380 of land preparation and crop production (Chan and Heenan 2005).

381
Studies (Shaffer 2010;Nyongesa and Vacik 2018) have shown that the use of fire for clearing bush 382 around homesteads during the dry season is a common practice in most rural communities in Africa.

383
Most of the study districts are rural with homes having bushy surroundings which dry up every 384 Harmattan season; as a result, applying fire is much cheaper and more practical than weeding and/or 385 applying weedicides, as explained by some of the respondents. In contrast, the districts in urban areas with high human populations have more paved surroundings, thus little use of fire for this activity has 387 been observed (Archibald 2016). Nyongesa and Vacik (2018) also have shown that the use of fire in 388 crop fields is to save on labour and the cost of chemicals .

389
Homesteads are mostly used for compound gardens in the rainy season which supply vegetables to 390 supplement the nutritional needs of the household. These supplementary farms are known as 391 compound farms which are burnt to get rid of the dried stubble in time for the next rainy season.

392
Burning around homesteads is also done to get rid of ticks and reptiles as these homesteads contain 393 stables for cattle and other livestock (Mistry et al. 2018).

394
The practice of using fire for creating firebreaks was relatively high across the study area, which 395 confirms the findings of Shaffer (2010) and Nyongesa and Vacik (2018), who observed in 396 Mozambique and Kenya respectively, that using fire for firebreaks was done to reduce fire risks within 397 the communities and to protect livestock as well as crops which are yet to be harvested.

398
Hunting and charcoal burning were sensitive items in the questionnaires, particularly in communities 399 with government protected forests with more burned areas, either due to uncontrolled fires from 400 hunting or poaching, as well as charcoal production (Walters 2012  during the data collection period revealed that there is a high usage of fire for charcoal production and 408 hunting. We also observed a lot of charcoal production going on in some of the communities. The 409 respondents' reluctance in giving honest answers on their use of fire for charcoal production and 410 hunting could be attributed to people's awareness of the dangers and adverse effects that these 411 activities have on the environment, human lives and property. Also, there has been a campaign against 412 these practices on the radio, as well as local announcements made by traditional authorities at the 413 beginning of every dry season (Amoako et al. 2015).

414
In addition, the respondents seem to be aware of the suggested policy to ban charcoal production and    (Veenendaal et al. 2018). This finding agrees with other studies which found vegetation burning in 448 tropical savanna (Mistry, 1998)

476
Fire frequency was rated highly by respondents as unimportant. There was however a strong positive 477 correlation between fire effect on environment how often people used fire for land preparation.
Contrarily, the majority of respondents did not perceive fire return intervals and number of ignitions 479 as a problem. Minnich et al. (1993) claimed that human-caused fire frequency (number of ignitions) 480 is unimportant, because the onset of natural fires could be more devastating since natural fires occur 481 when there is so much fuel load for burning, resulting in high intensity fires than frequent human 482 ignited fires. However, Griffiths, Stephen and Garnett (2015) found that fire frequency has an 483 enormous effect on both plant and animal species, and therefore should not be underestimated.

484
Most respondents thought the type of fire which encompasses pattern and size of fire was unimportant.

485
However, Govender, Bond and van Wilgen (2006), have shown that fire severity is influenced by the 486 type of fire which is in turn affected by the season of fire.In contrast, Govender et al (2006) have shown 487 that fire severity is influenced by the type of fire which is in turn affected by the season of fire. This 488 suggests the need to address the knowledge gap in managing the cultural uses of fire and fire regimes 489 in savanna.

490
As shown in our study, most fires in these savannas occur annually and at specific times, thus have 491 defined the anthropogenic fire regimes in these savannas (Archibald 2016). There is a need, for further 492 studies to understand the complexities of human-driven fire regimes in Africa savannas.

494
The study confirms that the use of fire for agriculture and some socio-cultural purposes is a 495 characteristic of traditional livelihood practices, which are mostly subsistence, but could have 496 enormous impacts on the environment, as outlined by the respondents. Most of the respondents are 497 farmers who use fire once a year for land preparation for cropping. The high fire frequency districts 498 used fire for almost all the selected activities, thereby increasing the fire occurrences in these districts,

499
which confirms the fire count data obtained from the CSIR Meraka Institute.
The study revealed that fires that are set on purpose were mostly controlled, particularly for land 501 preparation for cropping, which shows that people are aware of the hazards associated with 502 uncontrolled fires. There's the need to study the ecological effects of traditional fire use.

503
The season of burning was highly rated as a very important component of fire regimes. The 504 respondents' stated that early and late season burning was dependent on the time the drought sets in.

505
Notwithstanding, the type of fire was the least rated amongst the four attributes of a fire regime. More 506 education on the impacts of different fire regimes could improve fire management in fire-prone areas.

538
The authors declare no conflict of interest.