The ANOVA of nested models reveals that population and precipitation provide the best model in all three cases for the crops.
4.     Discussion
In the previous sections, we introduced the results of the regression of crop production onto population and precipitation in Chad from 1980 through 2011, a 32-year period. The results show that crop production increased as the population of Chad grew whereas crop production oscillated in accordance with precipitation patterns. An analysis using both SPSS and R reveal that the country’s population had significant positive associations with crop production, while precipitation showed positive but moderate associations. Statistically, SPSS shows that population explained about 70% to 90% of the variabilities in the crops’ production, whereas precipitation explained 50% to 60% of the variability. The R analysis shows that the population accounted for 50% to 73% of the variabilities of the three crops, while precipitation predicted about 32% to 62% of the variability. The standardized coefficients show that Chad’s population has the strongest contribution to the production of all three crops while the contribution of precipitation is much less. The R-squared between observed and predicted productions from the bilinear regression analysis are 58%, 60%, and 24% for sorghum, maize, and rice, respectively. Those low to moderate values of R-squared between observed and predicted productions are an indication that population and precipitation are not the only predictors of the yields of those three crops. Below, we discuss the relationships for each independent variable in greater detail.
4.1. Sorghum
Sorghum is highly significantly positively correlated with population (0.94 in SPSS and 0.93 in R) and moderately correlated with precipitation (0.53 in SPSS and 0.73 in R). The simple linear regression reveals that the variability of sorghum’s production is explained at 88% (SPSS) and 73% (R) by population alone. Precipitation alone accounts for 52% (SPSS) and 32% (R) of the variability of sorghum. The bilinear regression indicates that both population and precipitation accounts for 90% (SPSS) and 73% (R) of the variability of sorghum. The proportions forming the bilinear regression are the same as the proportions explained by population alone in the simple linear regression. This means that population had a strong effect, whereas precipitation had a lesser effect in terms of explaining the variability of sorghum, confirming the results from the simple linear regression where population alone explains more than 90% of the variability of sorghum. The interaction terms analysis run in R improves the proportion explained by the independent variables by only 2% (R2 = 75%). Ultimately, both population and precipitation made statistically significant contributions to the production of sorghum, but population accounted for more of this effect.
4.2. Maize
Maize displays a significant positive correlation with population (0.90 in SPSS and 0.87 in R) and a moderate to high correlation with precipitation (0.72 in SPSS and R). The simple linear regression analysis indicates that population alone explained 80% (SPSS) and 50% (R) of the variability in maize’s production. Furthermore, 50% (SPSS) and 37% (R) of the variability for maize is explained by precipitation alone. The results from the bilinear regression show that both population and precipitation accounted for 82% (SPSS) and 55% (R) of the variability in maize. The proportions forming the bilinear regression are in the same order of the proportions explained by population alone in the simple linear regression. These results indicate that population had a strong impact in explaining the variability in maize while precipitation had a lesser effect in explaining the variability in maize. This analysis confirms the results from the simple linear regression where population alone explained about 80% of the maize production variability. The interaction terms analysis generated in R improved the proportion explained by the independent variables by only 4% (R-squared = 59%). Therefore, both population and precipitation made statistically significant contribution in the production of maize; however, population accounted for more.
4.3. Rice
Rice has a significant positive correlation with population (0.85 in both SPSS and R) and a moderate to high correlation with precipitation (0.78 in SPSS and 0.77 in R). From the simple linear regression analysis, population alone accounted for 72% (SPSS) and 56% (R) of the variability in the production of rice. Precipitation alone accounted for 60% (SPSS) and 62% (R) of the variability in rice. From the bilinear regression analysis, results show that both population and precipitation accounted for 76% of the variability in rice using both SPSS and R. The proportions forming the bilinear regression are 4% more than the proportions explained by population alone in the simple linear regression. This small difference (4%) indicates that population had a stronger effect than precipitation in explaining the variability of rice’s production. The interaction terms analysis run in R decreased by 2% the proportion explained by both population and precipitation and their interactions (R2 = 74%). Hence, both the contributions of population and precipitation to variations in rice production were statistically significant, but the effect of precipitation was more pronounced.
To summarize, the best predictive model is the bilinear model for all three crops. Adding interaction terms to the model does not significantly improve or lessen the explanatory power of the model. The two predictive variables (population and precipitation) account for a unique and statistically significant contribution to the prediction of sorghum, maize, and rice. The fact that population accounted for more of the variability of sorghum and maize production compared to precipitation could be due to two reasons. First, growing population often leads to expansion of agricultural land in this part of the world. Additionally, our first-hand experience suggests that farmers tend to use adaptive water management methods to extend the window for crop production into dry seasons. An example is the cultivation of a flood recession or transplanted sorghum commonly called “Berebéré” in the Lake Chad region (Vara Prasad and Staggenborg 2011; Ahmed et al. 2000) that is grown in cold season from September/October to January/February. Moreover, other significant predictors such as fertilizers may exist, especially for rice. Unfortunately, data on the widespread use of fertilizers by farmers in Chad is not available. TNAU (2013) provided detailed guidance on how to use fertilizers including nitrogen, phosphorus, and potassium fertilizer (NPK) for rice production. The results obtained in this study are consistent with the findings reported in Boserup (2017) in which population growth was shown to be a major factor determining agricultural development.
5.     Conclusion
In this study, we used regression analysis to explore the relative contributions of growing population and precipitation variability on the production of sorghum, maize, and rice in Chad between 1980 and 2011. We found that population has a very significant positive association with sorghum (> 0.90), a very strong positive association with maize (≈0.90), and a strong positive association with rice (0.85). Precipitation was moderately positively associated with all three crops (0.50 to 0.70). The two independent variables, population and precipitation, were moderately positively associated (≈ 0.70), while the crops were strongly positively associated (≈ 0.80 to 0.92). We conclude that both the contributions of population and precipitation to variations in sorghum, maize, and rice production in Chad were statistically significant. However, the effect of population was more pronounced for sorghum and maize while the effect of precipitation was more pronounced for rice.
This strong effect of population can be explained by the fact that Chad’s population increased from 4,554,000 in 1980 to 11,525,000 in 2011. Meanwhile, agricultural land also increased from 48,150,000 to 49,932,000 hectares (FAOSTAT, 2013). Djimadoumngar (2014) also found a positive and moderate correlation coefficient (0.41) between population and agricultural land in Chad during the same time period. This suggests that increasing population is one of the major drivers of the expansion in agricultural land area in Chad over the last three decades. The strong effect of precipitation for rice is likely due to the fact that the cultivation of rice is mostly fluvial. We expect that further studies, especially by incorporating data on more predictors such as fertilizers and other climate variables such as evapotranspiration and air temperature, will yield additional helpful and complementary insights and those reported in this study.
References
Ahmed, MM, Sanders, JH, Nell, WT (2000) New sorghum and millet cultivar introduction in Sub-Saharan Africa: impacts and research agenda. Agricultural Systems 64(2000) 55-65.
Barrios, S, Bertinelli, L, Strobl, E (2010) Trends in rainfall and economic growth in Africa: A neglected cause of the African growth tragedy. The Review of Economics and Statistics, 92(2): 350-366.
Boserup, E (2017) The conditions of agricultural growth. The economics of agrarian change under population pressure. George Allen & Unwin LTD, Ruskin House Museum Street, London.
Djimadoumngar, K (2014) Multiscale study of the interactions between climate, land use, and agricultural productivity in western Sahel: A case study of Chad. A thesis in Environmental and Urban Geosciences, University of Missouri-Kansas City, U.S.A, 2014.
Food and Agriculture Data, FAOSTAT (2013) Food and Agriculture Data for Chad [Cited 2013 May 2]. Available from: http://faostat3.fao.org/faostat-gateway/go/to/download/O/*/E.
Food and Agriculture Data, FAOSTAT (2018) Food and Agriculture Data for Chad [Cited 2018 June 14]. Available from: http://www.fao.org/faostat/en/#data/QC.
Funk, C, Rowland, J, Adoum, A, Eilerts, G, White, L (2012) A climate trend analysis of Chad. U.S. Department of the Interior, U.S. Geological Survey, Fact Sheet 2012–3070 [Cited 2014 February 6]. Available from: http://pubs.usgs.gov/fs/2012/3070/FS2012-3070.pdf.
Gareth, J, Witten, D, Hastie, R, Tibshirani, R (2014) An introduction to statistical learning with application in R. Springer, ISBN 978-1-4614-7138-7 (eBook).
Gbetibouo, GA, Hassan, RM (2005) Measuring the economic impact of climate change on major South African field crops: a riparian approach. Global and Planetary Change 47 (2005) 143-152.
Henley, EC (2010) Sorghum: an ancient, healthy, and nutritious old world cereal. The United Sorghum Checkoff Program.
Hodgson, GM (2004) The Biographical dictionary of British Economists, Donald Rutherford, Bristol: Thoemmes Continuum.
Jayne, TS, Chamberlin, J, Heady, DD (2014) Land pressures, the evolution of farming systems, and development strategies in Africa: a synthesis. Food Policy 48 (2014) 1-17.
Malthus, RT (1803) An essay on the principles of population, 14th edition 1826. London: J.M. Dent, pp.1-24 passim.
Marquette, C (1997) Turning but not toppling Malthus: Boserupian theory on population and the environmental relationships. Working paper, Chr. Michelsen Institute, Development Studies and Human Rights, Bergen, Norway, ISBN 0804-3639.
McSweeney, C, New, M, Lizcano, G (2006) Chad. UNDP Climate Change Country Profiles [Cited 2014 May 10]. Available from: www.un-gsp.org/sites/default/files/documents/chad.oxford.report.pdf.
Meza-Pale, P, Yunez-Naude, A (2015) The effect of rainfall variation on agricultural households: evidence from Mexico. International Conference of Agricultural Economists, May 29, 2015, Milan Italy.
Mongabay. The largest cities in Chad, ranked by population [Cited 2018 June 15]. Available from: https://population.mongabay.com/population/chad.
National Centre for Research Methods, NCRM (2017) NCRM methods review papers [accessed on June 15, 2018].
National Framework for Climate Services, NFCS (2013) Report of the Workshop for Launching the National Framework for Climate Services in Chad, Bakara, April 17-19 2013.
Ndamani, F, Watanabe T (2015) Influences of rainfall on crop production and suggestion for adaptation. International Journal of Agricultural Sciences, Vol. 5(1), pp. 367-374.
Pendergrass, A, & National Centre for Atmospheric Research Staff (Eds). Last modified 02 Jul 2016. "The Climate Data Guide: GPCP (Monthly): Global Precipitation Climatology Project" [Cited 2018 June 15] Available from: https://climatedataguide.ucar.edu/climate-data/gpcp-monthly-global-precipitation-climatology-project.
Population Reference Bureau, PRB (2013) World Population Data Sheet [Cited 2014 March 10]. Available from: www.prb.org.
Population Reference Bureau, PRB (2016) World population data sheet. With a special focus on human needs and sustainable resources [Cited 2018 August 2]. Available from: www.prb.org.
ReStore, National Centre for Research Methods (2017) Using statistical regression methods in education research. Model diagnostics and checking your assumptions [Cited 2018 August 3]. Available from: http://www.restore.ac.uk/srme/www/fac/soc/wie/research-new/srme/modules/mod3/14/index.html.
The World Bank (2017) Employment in agriculture (% of total employment) [Cited 2018 July 29]. Available from: https://data.worldbank.org/indicator/SL.AGR.EMPL.ZS.
TNAU Agritech Portal (2013) Nutrient management: rice [Cited 2018 August 9]. Available from:  http://agritech.tnau.ac.in/agriculture/agri_nutrientmgt_rice.html.
Tropical Rainfall Measuring Mission, TRMM (2011) TRMM Microwave Imager Precipitation Profile L3 1 month 0.5 degree x 0.5 degree V7, Greenbelt, MD, Goddard Earth Sciences Data and Information Services Center (GES DISC) [Cited 2013 May 1]. Available from: https://disc.gsfc.nasa.gov/datacollection/TRMM_3A12_7.html.
Vara Prasad PV, Staggenborg, SA (2010) Growth and production of Sorghum and millets. EOLSS Publishers, Oxford, U.K.
Weather in the World [Cited 2015 September 5]. Available from: http://rp5.ru/Weather_in_the_world.
World Food Program, WFP, Food and Agriculture Organization, FAO, and Famine Early Warning Systems Network, FEWS NET (2011) Marché et sécurité alimentaire au Tchad [Cited 2019 January 1]. Available from: www.fao.org/fileadmin/user.../Marche_SA_Tchad_fev_2011.pdf .