Climate suitability predictions for the cultivation of macadamia (Macadamia integrifolia) in Malawi using climate change scenarios

Climate change is altering suitable areas of crop species worldwide, with cascading effects on people reliant upon those crop species as food sources and for income generation. Macadamia is one of Malawi’s most important and profitable crop species; however, climate change threatens its production. Thus, this study’s objective is to quantitatively examine the potential impacts of climate change on the climate suitability for macadamia in Malawi. We utilized an ensemble model approach to predict the current and future (2050s) suitability of macadamia under two Representative Concentration Pathways (RCPs). We achieved a good model fit in determining suitability classes for macadamia (AUC = 0.9). The climatic variables that strongly influence macadamia’s climatic suitability in Malawi are suggested to be the precipitation of the driest month (29.1%) and isothermality (17.3%). Under current climatic conditions, 57% (53,925 km2) of Malawi is climatically suitable for macadamia. Future projections suggest that climate change will decrease the suitable areas for macadamia by 18% (17,015 km2) and 21.6% (20,414 km2) based on RCP 4.5 and RCP 8.5, respectively, with the distribution of suitability shifting northwards in the 2050s. The southern and central regions of the country will suffer the greatest losses (≥ 8%), while the northern region will be the least impacted (4%). We conclude that our study provides critical evidence that climate change will reduce the suitable areas for macadamia production in Malawi, depending on climate drivers. Therefore area-specific adaptation strategies are required to build resilience among producers.

respectively. Of concern, suitable areas for macadamia production are predicted to shrink by −18% (17,015 25 km 2 ) and −22% (20,414 km 2 ) based on RCP 4.5 and RCP 8.5, respectively, with much of the suitability 26 shifting northwards. Although a net loss of area suitable for macadamia is predicted, some currently 27 unsuitable areas will become suitable in the future. Notably, suitable areas will increase in Malawi's central 28 and northern regions, while the southern region will lose most of its suitable areas. In conclusion, our study 29 provides critical evidence that climate change will significantly affect the macadamia sub-sector in Malawi. 30 Therefore area-specific adaptation strategies are required to build resilience. 31 Keywords: Malawi, macadamia, climate change, ensemble model. 32

Introduction. 33
Global climate change is having wide-ranging impacts on ecosystems, human health, livelihoods, food 34 security, water supply, and economic growth [1]. These impacts are expected to intensify in direct 35 proportion to the degree of warming. Warming by 2 o C, for example, is expected to increase the number of 36 people exposed to climate-related risks and poverty by several hundred million by the 2050s [1]. This 37 warming poses significant threats to many parts of Africa's current agricultural production systems, . Malawi has been identified as particularly vulnerable to climate change within SSA due to 41 high levels of poverty, limited finances and technology, and a heavy reliance on a predominantly rain-fed 42 agricultural sector for food and nutritional security, economy, and employment [7], [8]. and increases in rainfall intensity [11]. These changes are expected to threaten livelihoods, increase the risk 50 of food insecurity, and negatively affect Malawi's economic growth. 51 Increased warming and unreliable rainfall within Malawi will impact the landscape and the livelihoods of 52 many rural populations who depend on agricultural activities [9], [12]. Barrueto et al. [13] observed that in 53 Nepal's highlands, increased temperatures led to changes in cropping systems because of an upward shift in 54 perennial tree crops' suitability, including macadamia. Climate change will cause changes in agricultural 55 systems in Malawi, as it has in Nepal. Furthermore, due to Malawi's high deforestation rates, increases in 56 diurnal temperature changes are expected, making crop growth and development more difficult [14]. boundaries for crops before exploring the impacts of climate change [17]. Bock et al. [18] argue that climate 64 suitability assessment should be the first step in agricultural land use planning. For perennial crops, climate 65 suitability assessment is essential because they are long-term investments with high initial costs to establish 66 the crops [19]. Proper planning is, therefore, key to the success of perennial tree production. 67 On the other hand, Malawi's climate suitability research has primarily focused on cash and staple crops [20]. 68 Cash crops are grown mainly to be sold rather than consumed by the people who grow them [12]. Cashew, 69 coffee, cotton, pulses, sugarcane, tobacco [12], macadamia [20], and tea [21] are all important cash crops in 70 Malawi. Cash crops like macadamia support smallholder producers in two ways: they contribute to food 71 and nutritional security and economic growth since they are used to supplement cereal-based diets and 72 generate income. Staple crops are eaten regularly and in large amounts, and they make up the majority of 73 the standard diet. In Malawi, currently, the most important staple crop is maize [20]. 74 Malawi's economy and society are built around agriculture [22]. Agricultural operations provide a living 75 for nearly 85% of the country's households [23]. The agricultural sector consists of two distinct sub-sectors: 76 smallholder farmers and commercial estates. About 11% of the rural labor force works on commercial 77 estates to supplement farm income, and around 80% is engaged in the smallholder sub-sector [24]. . It is one of the most highly regarded nuts globally, with a kernel that contains more 102 than 72% oil. Because of their nutritional value, macadamia nuts fetch high market prices, owing to 103 consumer demands [32]. Historically macadamia trees were inter-cropped with coffee, tea, and tung trees 104 in the commercial estates in southern Malawi [20], [33]. Approximately one million macadamia trees have 105 been planted in the commercial estates sub-sector, with over 300,000 trees planted in the smallholder sub-106 sector, which is expected to increase to over one million in the next decade [20]. Globally, Malawi is the 107 seventh-largest producer of macadamia nuts [20]. Malawi can become the biggest producer of macadamia 108 in the world [20], [33]. This potential is due to the country's ideal altitude and climatic conditions for growth 109 and development, as well as large land pockets among smallholders that allow for expansion. 110 Macadamia is a lucrative crop among smallholder farmers in Malawi [34]. However, due to macadamia's 111 sensitivity to extreme temperatures and drought, climate change is expected to negatively impact its 112 productivity and suitability in the future [35]. Studies in South Africa found that macadamia yields were 113 lost due to spatio-temporal variability in precipitation and temperature [36]. According to Barrueto et al. 114 [37], macadamia is best suited to areas with annual mean temperatures ranging from 10-30 o C. Lower day 115 production is given in Supplementary Table S1. 119 Climate suitability for crop production is critical for developing adaptation and mitigation strategies to the 120 projected negative impacts of climate change on food security. Our study, therefore, aims to assess the 121 suitability for macadamia production under Malawi's current and future climatic conditions. First, we 122 examine the current spatial distribution of suitable areas for macadamia in Malawi. Then, utilizing 123 bioclimatic variables [39]  surfaces. Because of variations in topography (Fig 1), parent materials, and management, soil nutritional 132 status varies greatly across the country, particularly among smallholder farmers [15]. As a result, soil 133 characteristics were not taken into account in our analysis. 134 Vista ® Cx) together with altitude (Fig 3).  (Table 1). 168

Modelling approach. 169
We modelled the current and future distribution of macadamia species in Malawi based on an ensemble We used an AUC value of 0.77 as a threshold to select the best-performing algorithms. Species distribution 187 model algorithms that did not fit this criterion were not used to calculate the ensemble model's suitability 188 [52]. The SDM algorithms used in our analysis were those that can distinguish between suitable and non-189 suitable areas without needing absence locations [53]. We utilized the presence-only approach because, for 190 agricultural applications of niche models, it is inappropriate to treat areas without current production as 191 entirely unsuitable. Further, it is difficult and rare to determine whether a species is absent in a particular 192 location. Hence, absence data may not represent naturally occurring phenomena [46]. Additionally,193 presence-only models can produce reliable predictions from limited presence datasets, meaning that they 194 are robust and a cheaper option for obtaining training datasets [54]. To enhance the ensemble model's 195 predictive ability, we coupled the macadamia occurrence data with 500 randomly pseudo-absence data 196 generated throughout Malawi (Fig 3). We used the pseudo-absence data as opposed to using real absences 197 to avoid underestimation issues [55]. 198 In the second step, we retained only the algorithms that contributed at least 5% to the ensemble suitability 199 (1) Where the ensemble suitability (Se) is obtained as a weighted (w) average of suitabilities predicted by the 209 contributing algorithm (Si). 210 The third step generated the current distribution maps (probability maps and presence-absence maps) of 211 macadamia under the current climatic conditions. This was based on the weights which were generated 212 during model calibration (cross-validation). To generate the presence-absence layers, we used the maximum 213 sensitivity (true positive + ) and maximum specificity (true negative -) approach [58], where we reclassified 214 the distribution maps to binary maps (suitable and unsuitable areas). This method is one of the most reliable 215 for choosing a reclassification method [59], [55], [60]. In this analysis, sensitivity is the proportion of 216 observed presences correctly predicted and therefore is a measure of omission errors, whereas specificity 217 represents the proportion of correctly predicted absences and thus quantifies commission errors. 218 To create distribution maps for future bioclimatic conditions, we utilized the same procedure used in the 219 baseline suitability and presence-absence maps but utilized the climate information from each of the 17 220 future GCMs for RCP 4.5 and RCP 8.5. Since no criteria exist to assess which of the GCM's best predict 221 future climate [61], by incorporating all 17 GCM's, we encompassed all possible changes in the distribution 222 of the macadamia species. To integrate the results of the 17 GCM's presence-absence layers into a single 223 layer, we used the criterion of likelihood scale [59], which requires at least 66% of agreement among GCM's 224 to keep the predicted presence or absence in a given grid cell. 225

Factors determining the suitability of macadamia in Malawi. 227
Precipitation-related variables were the most important in determining suitability for macadamia production 228 in Malawi. Precipitation of the driest month (May-November) was the variable with the greatest relative 229 influence on the suitability for macadamia production. Possibly due to the sensitivity of pod growth during 230 this phase to water scarcity. Among the temperature variables, isothermality (this variable is calculated by 231 dividing mean diurnal temperature range by mean annual temperature range) was the most significant, with 232 a VIF score of 8.95 (Table 1). Our results indicate that annual means have no effect on the suitability for 233 macadamia production in Malawi.   Mwanza, Neno, Ntcheu, and Zomba districts. Marginally suitable areas were found to be in the lower 256 elevated (≤ 900 m.a.s.l) parts of Malawi. As expected, our predictions suggest that the current distribution 257 of climatically suitable areas for macadamia largely overlapped the areas where macadamia are grown in 258 Malawi. Though, these areas are also used for the production of other crops, particularly annuals. 259

Gains and losses of macadamia suitability under future projections in Malawi. 262
Under both emission scenarios used in this analysis, the extent of suitable areas for macadamia production 263 is expected to decrease in the future. Our findings show a net loss of −18% and −22% of suitable land for 264 macadamia production under RCP 4.5 and RCP 8.5, respectively (Fig 5, Table 2). This translates to 17,015 265 km 2 (RCP 4.5) and 20,414 km 2 (RCP 8.5) of Malawi's total cultivatable surface area. Lower altitudes (500-266 1000 m.a.s.l.) will experience the greatest decline in suitability due to projected heatwaves, flooding, and 267 droughts linked to the El Niño Southern Oscillation. These losses will be more pronounced in Malawi's 268 southern region areas, especially those along the Shire valley. Thyolo district, which is currently the 269 country's most productive and biggest macadamia growing area, is projected to suffer significant reductions 270 in suitable areas for macadamia production due to climate change. Therefore, investment in irrigation 271 infrastructure, both now and in the future, remains a viable option for continuing macadamia production in 272 vulnerable areas in Malawi. 273 Surface gains for macadamia suitability under future climate projections are described in Fig 5, Table 2. 276 Both scenarios show that a large fraction of suitable areas for macadamia production will remain unchanged,    variations are explained by differences in scale and geography, indicating that local and regional factors can 300 influence the potential for macadamia. Thus explaining our findings where precipitation-based factors have 301 been identified as more important in determining the suitability for macadamia, verifying zoning studies for 302 macadamia production done for the country [62]. Consequently, projections that climate change will reduce 303 rainfall amounts making its distribution unreliable in many parts of Malawi [10], will drive many areas out 304 of macadamia production. In this regard, farmers are encouraged to adopt moisture conservation measures 305 (mulching, rainwater harvesting, box ridging, and basins) and possibly develop irrigation infrastructure to 306 match the water requirements for macadamia growth and development annually. 307 Isothermality was the second most important factor determining the suitability for macadamia in Malawi. 308 Our findings suggest that the large fluctuations in day and night temperatures and increased warming are 309 responsible for the marginal suitability for macadamia, notably along the lakeshore and Shire valley. 310 Optimal suitability was observed in intermediate to upper elevated areas that experience cooler 311 temperatures, especially at night. Consequently, projections that climate change will increase the number 312 of hot (≥ 30 o C) days (30.5 days per year) and hot nights (40 days per year) will certainly reduce the number 313 of suitable areas for macadamia production in Malawi [11]. Therefore, trees currently grown in the hotter 314 areas will require sufficient water availability to cater to the water lost through evapotranspiration. High 315 day and nighttime temperatures have been linked to a reduction in suitable areas for macadamia in Australia 316 [38], Nepal [37], and South Africa [36], findings that are consistent with our current findings in Malawi. 317 Malawi is very vulnerable to climate change compared to other countries in SSA due to its geographic 318 location and socio-economic status [7], [8]. Our results show that extensive areas in Malawi under the 319 current climatic conditions are suitable for macadamia production. Moreover, our outcomes show the 320 suitability for macadamia in Malawi's south-eastern parts beyond the current reported production areas. 321 This is expected as the suitability maps capture the potential production areas, some of which have not yet 322 been translated to realized areas [46]. However, areas of Malawi already fall outside the prescribed optimal 323 range for macadamia production, which may be attributed to the increase in annual mean temperatures (0.9 324 o C) and overall drying recorded in the past five decades [6], [63], [64]. 325 Due to the projected temperature increases and changing rainfall patterns, the suitability for macadamia 326 production in Malawi is likely to decline in the 2050s and is expected to shift northwards. Areas in southern 327 Malawi will be the most affected, with the central and northern highlands even improving capacity to sustain 328 macadamia production in areas where this is impossible due to environmental constraints. Other authors 329 have predicted similar effects of climate change on the suitability for macadamia production globally. 330 Barrueto et al. [37] reported an upward shift in suitable areas for macadamia production in Nepal due to 331 climate warming. This shift resulted in losses in suitable areas for macadamia production in This shift 332 resulted in losses in suitable areas for macadamia production in Nepal's lowlands and gains in suitable areas 333 for the highlands, which is consistent with our findings. 334 Our results, therefore, show the sensitivity of macadamia to variations in environmental conditions. Thus, 335 in areas where there are no predicted changes in the suitability for macadamia, farmers can continue planting 336 their macadamia trees. However, both research and field-based evidence from discussions with farmers 337 show that climate-related changes are already occurring and affecting the suitability for macadamia 338 production in Malawi. Farmers are, therefore, encouraged to start implementing adaptation measures such 339 as the use of improved macadamia varieties, agroforestry, intercropping, water conservation, and irrigation 340 for long-term and sustainable macadamia production. Nevertheless, these suitability changes are predicted 341 to occur over the next 30 years, so these will mostly impact the next generation of macadamia farmers. 342 Therefore, there is still time for adaptation. 343

Applicability and potential limitations of this study. 344
Species distribution modelling is founded on assumptions intrinsic in the models, some of which cannot be 345 tested [65], [66]. Although this study's findings can be considered robust, several issues should be 346 considered in the interpretation and application of the results. Though we identified areas as suitable for 347 macadamia production based on environmental factors, however on the ground, this may not directly 348 translate to the size of the arable land. Other physical and socio-economic factors (including the gender and 349 age of the smallholder farmers, availability of agricultural advisory services, and market availability) that 350 determine suitability are difficult to capture in this type of modelling and should be considered in applying 351 model results. Due to these challenges, the authenticity of models in making predictions is questioned [76-352 77], but modeling remains an important tool for future planning purposes [60], [78], [79]. Therefore, there 353 is a need for a thorough evaluation of adaptation approaches suggested for smallholder macadamia farmers, 354 as these may be different from those utilized by commercial growers. 355

Conclusions. 356
This study utilized an ensemble model to identify current and future suitability for macadamia production 357 in Malawi. Three important conclusions can be drawn from the study results. First, precipitation factors 358 are the most important determinants for macadamia suitability in Malawi. Second, the current and future 359 areas identified as suitable for macadamia production exist on agricultural land currently used to grow other 360 crops. As a result, we propose promoting macadamia intercrops and agroforestry as a climate change 361 adaptation strategy. Third, the extent of suitable areas for macadamia production in Malawi is projected to 362 decrease under both emission scenarios utilized in this analysis, and the most vulnerable areas are those in 363 southern Malawi. We conclude that the macadamia sector faces production risks from climate change, but 364 there are opportunities for adaptation strategies to build a resilient sector in Malawi. 365