SciELO - Scientific Electronic Library Online

 
vol.52 issue3Performance of dairy production, reproduction and longevity in Holstein and its crosses with CebuTypology of cattle farms in a peasant community from southwest of Holguín, Cuba author indexsubject indexarticles search
Home Pagealphabetic serial listing  

My SciELO

Services on Demand

Journal

Article

Indicators

  • Have no cited articlesCited by SciELO

Related links

  • Have no similar articlesSimilars in SciELO

Share


Cuban Journal of Agricultural Science

Print version ISSN 0864-0408On-line version ISSN 2079-3480

Cuban J. Agric. Sci. vol.52 no.3 Mayabeque July.-Sept. 2018  Epub Sep 01, 2018

 

Animal Science

Effect of climate change on milk production in productive enterprices from Jimaguayú, Camagüey province

J. Somoza1  * 

J.M. Febles2 

R. Rangel3 

E. Sedeño4 

E. Figueredo4 

Odalys Brito4 

1Facultad de Economía, Universidad de La Habana, Calle L entre 23 y 21, Vedado, Plaza de la Revolución, La Habana, Cuba, C.P. 10 400

2Centro de Investigaciones Marinas, Universidad de La Habana, Calle 16 No.114, entre 1ra y 3ra, Miramar. Playa. La Habana. Cuba.

3Instituto de Geografía Tropical del CITMA. Calle F, No.315 esquina 11, Vedado, Plaza de la Revolución, La Habana, Cuba, CP. 10 400

4Centro de Investigaciones de Medioambiente de Camagüey, Cisneros No. 105 (altos), entre Pobre y Ángeles, Camagüey, Cuba

Abstract

In the absence of bioclimatic models, and given the scarcity of information about parameters and other input data, the panel data methodology was applied to establish the magnitude and direction of expected consequences in milk production, which is the main economic activity of Jimaguayú municipality. This study presents a summary about the theory of panel data models. Several specifications are evaluated, obtaining expressions of long-term elasticity of rainfall and temperature and its effect on milk production. Applied empirical models indicate that milk production in Jimaguayú could decrease by approximately 200 thousand L per year, due to the combined effect of the decrease in rainfall and the increase in temperature, with economic losses in the order of constant 100 thousand dollars informed in 2014.

Keywords: empirical panel data models; bioclimatic models; climatic elasticity

The examination of fluctuations in the climatic variables allows to advance the hypothesis that Cuban climate transits towards a state of greenhouse effect, intensified in the terrestrial atmosphere (Centella et al., 2001), with very similar characteristics to those projected by the Intergovernmental Panel on Climate Change Group (IPCC). The evidence provided by the climatic information registered from 1960 to 1990, and from the latter to 2016, indicate that there is a tendency to increase the mean annual temperature in 0.5 °C and the mean annual minimum temperature in 1.4 °C, with significant reduction of the day range. There is also an increase in the frequency and impact of extreme events (Pérez et al. 2013), even of drought, and the event "El Niño Southern Oscillation" (ENSO).

For the future (2050-2100), it is estimated that average temperature could increase from 1.6 to 2.5 °C in relation to 1960-1990 period. The climate could be between 10 and 19 % more or less rainy than in the present, and sea level could rise between 20 and 95 cm with respect to the current, which will cause a reduction in water availability and quality. It is estimated that, by 2100, it could flood up to 35 % of the area of the country, in the most extreme climate scenario. This prognosis increases vulnerability of approximately 185 human settlements (Planos et al. 2013a). This increase in sea level could cause great affectations in mangroves and other coastal ecosystems, with losses in productive yields, biological diversity and associated climatic and ecological values (Planos et al. 2014).

The effect of climate change (CC) would also occur in grasslands of cattle rearing areas, which are predominantly of grasses in Cuba. This could benefit very harmful species such as marabu (Riverol and Riquelme 2011). All this would bring about substantial modifications in the nutritional value, digestibility and palatability of forage species (Planes et al. 2013 b).

Studies carried out by the Institute of Meteorology in Camagüey (Rivero et al. 2004), related to the effects of CC in livestock regions indicate the following:

  • Systematic increase in the duration and intensity of meteorological and agricultural drought episodes, as CC advances during this century. The areas most affected by these processes will be the northeast and south of the province (Rivero 2008).

  • Accentuation of the aridity degree of grasses and increase of the frequency of moderate and severe droughts, which will reduce the productivity of the grasses formed mainly by C4 species, even in the presence of the effect of fertilization by CO2.

  • Conditions of cattle welfare will decrease with the gradual increase in temperatures, which become totally unfavorable in the interior of the province, as a result of the effects of "continentality" (such will be the case of Jimaguayú). In coastal areas, although the increase in temperatures will occur to a lesser extent, solar radiation will be more harmful than in land (MINAG 2011).

  • The bioproductive performance of livestock will experience considerable effects on future scenarios, taking into account that there will be a decrease of birth rate and an increase of general and specific mortality. It will also increase average age at which the first birth occurs (Delgado 2010).

  • Changes that take place in these indicators will have more significance in the southern and northeastern areas of the province, with marked deterioration indexes of vital livestock aspects, mainly during daytime hours.

Due to the lack of bioclimatic models and the scarcity of information on parameters and other input data for Jimaguayú territory, added the need to establish the magnitude and direction of the expected consequences in milk production, the panel data methodology (Greene 1993) in a non-action or reference scenario (Bussiness as Usual).

The use of this type of model is quite common in studies of CC economy. A remarkable example of this, in the Latin American region, is the study of Galindo (2010) in Mexico.

Before the task of determining the existence of some statistically significant relationship between milk production in the main livestock enterprice of Jimaguayú and the available climate variables (precipitations, mean temperature and relative humidity, among others) that participate in the construction of climatic scenarios, this study had the objective of determining the impacts of CC on milk production of Jimaguayú, and its possible generalization to the best milk-producer provinces of Cuba.

Materials and Methods

Milk production data correspond to the four enterprices of the municipality (Triángulo 1, Triángulo 5, Rescate de Sanguily and Maraguán). These data comprise 60 months elapsed between March 2009 and March 2013. Climate data come from the meteorological station located in Camagüey city, the closest to the territory under study (table 1).

A dynamic panel data model was used. This is, with the delayed dependent variable, which is used as an explanatory variable to estimate long-term coefficients. The used dynamic panel models are also non-monotonic, which means that are quadratic in temperature and precipitation variables, in order to obtain an estimate of the non-constant climatic elasticity that depend on the value of temperature and precipitation variables (Pérez 2008).

Selection of panel data specification. The results of the application of panel data models provide results consistent with the characteristics of enterprices. They indicate that there is heterogeneity among the enterprices. The statistical tests of Breouch-Pagan, such as the Student's T test, reveal that random effect models and fixed effect models are better than the pooled model to explain the relationships between variables milk production and those concerning climate (Greene 2005 a).

Table 1 Descriptive statistics of climate variables 

Source: Instituto Provincial de Meteorología, 2015

y(t.i): production of the unit per month i (i = 1…4) and month in liters rr(t): mean monthly precipitation in mm of rain hrmed(t), hrmax(t) and hrmin(t): values of mean, maximum and minimum relative humidity, respectively temp(t), tempmax(t) y tempmin(t): values of mean, maximum and minimum temperature per month in degrees centigrade y1(t.i) is delayed milk production per unit and month

The application of these tests is relevant because if the random-fixed effects model and the pooled model were the same (the variance of the errors were equal to zero in both models), then, preferably, the pooled model would be used, since it offers the best estimated coefficients (Ordinary Least Squares - OLS). That is, unbiased and efficient coefficients (minimum variance). From the practical point of view, this would mean assuming that all the studied enterprices have the same production technology, which is difficult to sustain.

The results of Hausman Test (Hausman 1978 and Hausman and McFadden 1984) allow to select between random (RE) and fixed (FE) models, which better explains the relations among variables. In this case, the result demonstrates the impossibility of accepting the null hypothesis (H0). That is, the coefficients estimated by the FE and RE model are equal (βef-βea=0), so it identifies the fixed effect model as better that of random effects. Results indicate that the model of fixed effects is the most adequate to estimate parameters of milk production model (table 2).

Tabla 2 Test of Hausman, for specifications of panel data of fixed and random effects. Milk production. Results of STATA 12.0  

b = consistent under Ho and Ha; obtained from xtreg; B = inconsistent under Ha, efficient under Ho; obtained from xtreg; Test: Ho: difference in coefficients not systematic; chi2 (5) = (b-B)'[(V_b - V_B)^(-1)] (b-B) = 310,74; Prob > chi2 = 0.0000 (V_b-V_B is not positive definite).

Note: lyr, milk production logarithm; lrr and lr2, precipitation logarithm and square precipitations; ltemp, mean temperatura logarithm; and lhrmax, maximum relative humidity

About dynamic model and long-term elasticities. The quantity of milk produced in a period (Qt) and its relationship with climate variables that characterize the climatic scenarios is assumed as a non- linear or non-monotonous function of rainfall and temperature. This relation is expressed as a quadratic function to collect the "U" performance of milk production as the above mentioned climatic variables grow1 (Somoza 2008). This quadratic specification offers elasticities that are functions of the explanatory variables themselves, and they are not independent constants of the performance of the climatic variables (Brendt 1991).

Results and Discussion

This study presents two models: the fixed effect model (FE) and the robust fixed effect vs. heteroscedasticity (FE (vce robust)) model. Table 3 summarizes the results of the estimation of model parameters using both specifications. As it is observed, the value of coefficients (in this case, they represent the climatic elasticities of milk production) are the same in both cases. However, robust model in heteroscedasticity offers a higher probability level to the estimated values with respect to FE model, which translates into the values of P> t of H0: βi coefficients are significantly equal to zero (Greene 2005b).

Table 3 Specification for precipitation and mean minimum temperature 

Source: salida STATA 12.0

In both models, the importance of the temporal variable (lt) in the explanation of milk production in the municipality in the last 60 months is highlighted. This type of variable is usually used to capture the trends of technological change. In the FE model, this variable is significant with a probability level of 10 %. On the other hand, the sign of the coefficient turns out to be the expected, according to what happened in the study period where a decreasing and sustained trend of milk production is observed, which is due to technological problems in the application of the production model from the municipality.

Precipitations are significant in both models, and the signs of both coefficients were the expected ones. In this sense, milk production grows with the reduction of rainfall (marginal production decreases), up to a level of precipitation where milk production begins to be reduced by the effect of lack of rains (figure 1).

The first aspect to be highlighted is the non-monotonic relationship between the volume of precipitations and milk production. Precipitation elasticity of milk production of the municipality is a function of the level of precipitations. It decreases while the mean monthly rain does.

Thus, the reduction of precipitations from the mean maximum value recorded in the last 60 months, about 250 mm per month (from 330 to 80 mm per month), would mean a reduction in milk production of 150 liters per month, almost 2 thousand liters on average per year (figure 2).

On the other hand, the effects expected from the increase of the average monthly minimum temperature counteracts the effect of rainfall reduction. An increase of minimum temperature of 4 °C, causes milk production to increase by 1.4 %. This means that the improvement of climatic comfort of livestock has a positive impact on yield per cow.

In terms of its effect on milk production, such an increase of the minimum temperature would mean an increase of monthly milk production of about 20 thousand liters on average, which could mean production increases in the order of 280 thousand liters per year (figure 2).

Figure 1 Elasticity of precipitations on milk production in enterprises from Jimaguayú municipality, Camagüey province 

Figure 2 Milk production vs. precipitation (A) minimum temperature (B) 

In this sense, everything seems to indicate that, in a model of fixed effects that uses as explanatory variables a temporary tendency to capture technological effects, rainfall and mean minimum temperature, it would be indicating that the effects of climate change would be favorable for the increase of milk production. However, this result would be partial and limited if the effect of increases of maximum temperatures are not considered, which are estimated to have a superior impact on the reduction of cattle comfort levels, and, therefore, on the reduction of milk production, proposed by Rodríguez et al. (2003).

It is in this sense that results (table 4) of the runs of a group of models that take as explanatory variables both, precipitation and mean monthly temperatures (minimum, mean and maximum) are presented, which, on the other hand, is the quantified variable in the different available climatic scenarios, those of IPCC and those built by INSMET.

As mentioned in the case of fixed effect model using the average minimum temperature, in this case, it also happens that the value of the coefficients estimated by FE and FE (vce robust) specifications are equal, although with better significance levels of the coefficients in the latter case (Greene 2005 b).

In the evaluated specifications, one of robust fixed-effect model is used in heteroscedasticity but with a cubic term, in order to verify if this specification provides a greater evidence of the inverted "U" performance that should characterize the relationship of this type of economic variable with those of climatic type. The model gets worse at the moment of estimating the coefficients, and signs radically change with the inclusion of the quadratic term in the mean temperature, so it throws increasing values of temperature elasticity of milk production, indicating that as mean temperature increases, milk production also increases.

Table 4 Summary of results of the estimation for several specifications 

Note: ** significant at 5%; ***significant at 10%

On the other hand, FE (vce robust) model with AR (1) autoregressive fit shows more coherent results, according to significance of estimated coefficients and to expected signs. For example, in the case of rainfall, results show how the "precipitations" elasticity of milk production grows with the increase of rainfall level of up to a level that reaches an average of 1,200 mm per month. The following figures show elasticity-rainfall level and milk production -precipitation (figure 3 A and B).

In this sense, a reduction of rainfall of 29 %, compared to the recorded mean in the last 60 months, would mean a reduction of milk production of about 15 liters per month, almost 180 liters per year. On the other hand, mean temperature elasticity of the production is indicating that, as rains increase, milk production decreases (figure 4).

Figure 3 Precipitations Vs elasticity (A) and Production (B) 

Figure 4 Mean temperature vs. elasticity (A) and milk production (B) 

The increase of mean temperature in about 4 °C (from 25 to 29 °C) may cause elasticity to be reduced in some 0.5 %, which would mean a reduction of milk production in about 1,000 liters per month on average, equivalent to about 12 thousand average liters per year. Finally, the evaluation of the results of FE (vec robust) models show similar results to the model fitted by autocorrelation of the perturbations. Values of precipitation elasticity and mean temperature of milk production increase more than proportionally with the growth of average monthly precipitation volumes and decrease with the increase of temperature (figure 4).

In this model, rainfall reduction from the average levels recorded in the last 60 months at 80 mm, only mean reductions in average milk production of five liters (figure 5), about 60 liters per year, which is half of the value reported by the model fitted by autocorrelation, previously analyzed. In terms of temperature, the increase of mean temperature in 4 °C, from 25 to 29 °C, means that milk production is reduced by an average of 1,000 liters per month, similar to the AR (1) model, analyzed above, with 12 thousand liters per year.

Figure 5 Elasticity precipitation (A) and temperature (B) of milk production 

Finally, the results obtained for the population mean model are presented. Firstly, in this model, the significance of coefficients obtained for precipitations, significantly different from zero for 95 %, is highlighted, while for temperature is not the same because they are the least significant of all the studied models.

The results in terms of long-term elasticities are similar regarding trend in relation to the rest of the models. However, it shows a higher level of sensitivity, for example, reduction of precipitations from 112 to 80 mm, its incidence on the reduction of milk production is five times higher than FE models (it reduces about 50 liters per month on average, about 600 liters per year). For its part, the increase of 4 °C, means in this model, a reduction of milk production per month of about 650 liters, about 7,000 liters per year.

The results obtained using mean temperature as one of the factors that characterize climate change, are quite modest results in relation to the response of the elasticity changes resulting from variations in precipitation levels and temperature values, which suggests that a variable that captures the effect on milk production could be mean maximum temperature.

Indeed, table 5 summarizes the results of the run of a robust fixed-effect model in heteroscedasticity using the maximum temperature as an explanatory variable to replace mean temperature.

Table 5 Results of estimation of dynamic panel data model with robust fix effects in heteroscedasticity 

Source: STATA 12.0

Values of long-term coefficients for precipitations do not differ much from the models discussed above. However, impacts on milk production due to the increase of maximum temperature are significantly higher than those obtained when using mean monthly temperature values.

The reduction of the volume of precipitations from the average level of the last 60 months to about 80 mm per month would imply a very modest reduction of milk production of only 30 liters per month, about 360 liters per year, which is not very different from the results of previous models.

However, the increase of 4 °C in the maximum monthly temperatures, this is 29 to 33 °C, would have an important impact on milk production, in this case, it would be reduced by about 14 thousand liters per month, 5 % of mean monthly production, which would be equivalent to a loss of almost 170 thousand liters a year.

This way, it is verified that, according to the information available, milk production is affected by the reduction of rainfall but more limited in relation to impacts of the increase of mean temperature values and that this is much more remarkable when taking into account variations in the maximum temperature. This result is much more consistent with the comfort values for dairy cattle reported for the country (table 6).

The use of a population mean model using average maximum temperature shows similar results in terms of impact on milk production in Jimaguayú. Graphs in figure 6 compare the results of this last specification with the one mentioned in previous paragraphs. Both specifications report very similar reductions in milk production for the 4 °C increase of mean monthly maximum temperature, like 14.2 thousand liters for the population mean model versus 14.3 in the case of the fixed effect model analyzed above. Similar performances are reported by several international studies (for example, Galindo 2010, for Mexico).

Table 6 Classification of climate regime (comfort) for milk cattle 

Source: Delgado (2010)

Figure 6 Contrast between FE models and population means for precipitations and maximum temperature (B) 

Conclusions

Due to the reduction of precipitations by 32 mm, from 112 to 80 (28 % of the monthly average), milk production could be affected between just 60 liters (robust FE models) up to 1,000 liters (model of robust population average), that is, reductions in milk production between 1,000 and 12,000 liters per year.

With the increase of the average monthly temperature in 4 °C (from 24.6 to 28.6 °C), milk production will be affected in approximately a range from 8 thousand to 12 thousand liters per year (650 a 1,000 liters per month), for robust FE specifications, up to 120 thousand liters in the case of population mean model.

On the other hand, the increase of maximum temperature in 4 °C would cause a reductions of milk production of 14 thousand liters per month (approximately 170 thousand liters per year), both in the FE specification and in the population average, which is remarkably higher than the results with the use of mean monthly temperature. This reduction of milk production would be equivalent to 64 % and almost 15% of mean and maximum monthly production value, respectively, of the last 60 months of the four enterprices of the Jimaguayú municipality. In monetary terms, the amount of milk that was not produced would be in the order of 120 thousand dollars in 2012, assuming a cost of 0.05 USD/L, an extraordinary amount for the budget of the municipality.

References

Berndt, E. 1991. The practice of Econometrics. Classic and contemporary. Addison Wesley Publishing Company. Reading Massachusset, Menlo Park. ISBN 0-201-17628 - 9. New York. [ Links ]

Centella, A., Llanes, J., Paz, L., López, C., y Limia, M. 2001. Primera Comunicación Nacional a la Convención Marco de Naciones Unidas sobre Cambio Climático. República de Cuba, 166 pp. [ Links ]

Delgado, C. 2010. Condiciones de confort para la explotación de la ganadería en Cuba. Ganadería tropical. Centro de Meteorología Agrícola. Instituto de Meteorología. La Habana, Cuba. [ Links ]

Galindo, L.M. 2010. La Economía del Cambio Climático en México. Secretaría de Medio Ambiente y Recursos Naturales. Subsecretaría de Planeación y Política Ambiental. Dirección General Adjunta de Análisis de Política y Financiamiento Estratégico. Blvd. Adolfo Ruíz Cortines 4209, Col. Jardines en la Montaña, C.P. 14210, Tlalpan, México, D.F. ISBN 978-607-7908-06-7. Impresos en México, D.F. Julio 2010. [ Links ]

Greene, W.H. 1993. Econometric Analysis. Prentice Hall Inc. A. Simon & Schuster Company. Upper Saddle River, New Jersey 07458. ISBN 0-02-346602-2. [ Links ]

Greene, W.H. 2005a. Reconsidering Heterogeneity in Panel Data Estimators of the Stochastic Frontier Model. Journal of Econometrics, 126: 269-303 [ Links ]

Greene, W.H. 2005b. Fixed and random effects in stochastic frontier models. Journal of Productivity Analysis, 23(1): 7-32. [ Links ]

Hausman, J.A. 1978. Specification test in econometrics. Econometrica. 46: 1251 - 1271. [ Links ]

Hausman, J. & McFadden, C. 1984. Specification test in econometrics. Econometrica, 52, 1219 - 1240 [ Links ]

Instituto Provincial de Meteorología 2015. Base de datos climáticos por estaciones de la provincia de Camagüey. Camagüey. Cuba [ Links ]

MINAG 2011. Composición del rebaño a nivel nacional al termino del 2007. Registro Pecuario. MINAG, La Habana, Cuba, 5 pp [ Links ]

Planos, E., Rivero, R. & Guevara, V. 2013a. Impactos del Cambio Climático en las actividades evaluadas. Segunda Comunicación Nacional a la Convención Marco de Naciones Unidas sobre Cambio Climático, pp 122. Instituto de Meteorología, Agencia de Medio Ambiente. La Habana, Cuba [ Links ]

Planos, E., Rivero, R. & Guevara, V. 2013b. Impacto del Cambio Climático y medidas de adaptación en Cuba. Instituto de Meteorología, Agencia de Medio Ambiente. Ministerio de Ciencias Tecnologías y Medio Ambiente. La Habana, Cuba, 430 pp. [ Links ]

Planos, E., Rivero, R. & Guevara, V. 2014. Impacto del Cambio Climático y Medidas de Adaptación en Cuba. Instituto de Meteorología y Centro Meteorológico de Camagüey; CITMA; PNUD. La Habana [ Links ]

Pérez, C. 2008. Econometría avanzada. Técnicas y herramientas. Editorial Pearson Hall. ISBN 978-84-8322-479-3; Madrid. [ Links ]

Pérez, R., Aguilar, G., Ballester, M., Carnesoltas, M., Cutié, V., Fonseca, C., González, C., Lapinel, B., Limia, M., Planos, E. & Vega, R. 2013. Segunda Evaluación de las Variaciones y Tendencias del Clima en Cuba. Taller sobre la Vulnerabilidad y Adaptación en Cuba al Cambio Climático Habana 2013. Instituto de Meteorología. Agencia de Medio Ambiente. CITMA. [ Links ]

Rivero Jaspe, R.R. & Rivero Vega, R. E. 2004. Integración de los impactos del cambio climático en la provincia Camagüey. Centro Meteorológico de Camagüey; Camagüey. [ Links ]

Rivero, R. R. 2008. Workbook on Climate Change Impact assessment in Agriculture. Basic knowledge, methodologies and tools. In collaboration with the Caribbean Community Climate Change Centre & the Cuban Institute of Meteorology, sponsored by The Commonwealth Secretariat. ISBN 978-976-95260-1-3. [ Links ]

Riverol, M. & Richelme, W. A 2011. Opciones de mitigación en el sector agropecuario, cambio de la agricultura de altos insumos a la agricultura de conservación. Reporte del Instituto de Suelos al Producto Mitigación de la Segunda Comunicación Nacional a la Convención Marco de Naciones Unidas para el Cambio Climático. La Habana. Cuba. 38 pp [ Links ]

Rodríguez, A., Roger, E., Rivero Vega, Z. I., Rivero Jaspe, R. R., Rivero Jaspe, D., García Bargados, R., Varela Ledesma, N. & Pérez Abraham, I. 2003. Impacto sobre los pastos y el ganado vacuno. Centro Meteorológico de Camagüey. Departamento científico. [ Links ]

Somoza, J. 2008. Energía y Desarrollo. Cuba en el contexto de América Latina y Caribe. Revista Cuba: Investigación Económica. 14(2): 55-90. ISSN 1026-485X [ Links ]

STATA 12.0: Stata Corp web site at http://www.stata.comLinks ]

Received: December 11, 2017; Accepted: July 25, 2018

Creative Commons License This is an open-access article distributed under the terms of the Creative Commons Attribution License