Modelado y control hídrico del continuo suelo-planta-atmósfera en cultivos leñosos

Universidad de Sevilla Departamento de Ingeniería de Sistemas y Automática Memoria de Tesis Doctoral Modelado y control hídrico del continuo suelo-planta-atmósfera en cultivos leñosos Hydraulic modelling and control of the soilplant-atmosphere continuum in woody crops Rafael Romero Vicente Directores: José Enrique Fernández Luque Instituto de Recursos Naturales y Agrobiología (IRNAS-CSIC) José Luis Muriel Fernández IFAPA, Centro Las Torres-Tomejil David Muñoz de la Peña Sequedo Dpto. Ingeniería de Sistemas y Automática. Universidad de Sevilla A Cris y Héctor… Agradecimientos La senda que culmina en este trabajo de tesis representa mucho más que un hito profesional en mi carrera investigadora. Estos años de dedicación a esta profesión que tanto amo me ha permitido atesorar grandes, más que compañeros, amigos que me han demostrado que su valor humano supera incluso su talento profesional. Sin ellos mi trabajo no solo hubiera sido infinitamente más difícil, sino también mucho más aburrido y menos enriquecedor. Es por esto que me gustaría dedicar a todos ellos las primeras palabras de esta tesis expresando mi profundo y eterno agradecimiento: A mis compañeros y profesores de la Universidad de Sevilla donde inicié mi formación en Ingeniería en un ya lejano 1993; y en particular a Ismael Alcalá con el que compartí innumerables horas domando al viejo robot RM10, a los Manolos, Fernando Castaño, Dani Jiménez y tantos otros compañeros de laboratorio que siempre están dispuestos a ayudar a los recién llegados. Gracias a mis profesores Carlos Bordons, Javier Aracil, Eduardo Fernández y Miguel Ángel Ridao por inspirarme y mostrarme el camino de mi vocación. A Francisco Rodríguez por su paciencia y apoyo durante el desarrollo de mi proyecto de fin de carrera, y por ser mi mentor en la universidad durante tantos años…y los que están por venir. Y finalmente a mis compis Rubén, Javi, Virginia, Antonio y, cómo no, a mi querido José Luis, Jere. ¡Qué grande es! Nos presentaron durante nuestro primer día en la universidad y, desde entonces, ya nunca volvimos a separarnos. A los compañeros del Instituto de Recursos Naturales y Agrobiología de Sevilla (IRNAS). A Juan Carlos Montaño por valorarme tanto y creer siempre en mí y a Fran Moreno por ofrecerme su ayuda constante y regalarme tantas conversaciones en los laboratorios de electrónica. A Antonio por nuestros largos y fructíferos diálogos telefónicos, y porque siempre encuentra un hueco a pesar de estar tan ocupado; ojalá pronto acabemos trabajando juntos. Gracias A Vicky, Vanesa, José Manuel, Luisa Candau, Maria José, a las recepcionistas y limpiadoras y todos los demás compañeros, por recibirme siempre con una sonrisa y hacerme sentir como en casa. A los compañeros del Instituto de Investigación y Formación Agraria y Pesquera de Andalucía (IFAPA); empezando por Fernando Romero, que siempre me ha aportado buenos consejos y tratado como a un hijo. A Joaquín y Jorge, siempre generosos en ayudarme, especialmente cuando algo se torcía, y a las recepcionistas y limpiadoras que derrochan amabilidad con todos. En compañeros como ellos se cimienta sin duda el alma de la institución. Gracias también a Antonio Daza, Paco Perea, Arturo, Dulce, Pedro Cermeño, Berta, Jorge Andrada, José Luis Campos, Paco Fuentes, Víctor, Marcelino, Fernando Morillo, José Rodríguez, Miguel Ángel… y tantos otros que no podría nombrar, no por falta de merecimiento sino de espacio y tiempo. Y muy especialmente a mis queridos amigos y compañeros de departamento. A Juan Antonio, que se hace querer tanto que vaya donde vaya todo el mundo le conoce. A Jorge, Miguel Ángel y Manolo, que siempre están dispuestos a ayudar en los trabajos más difíciles y cuyo esfuerzo ha sido imprescindible para conseguir los resultados presentados en esta tesis. A los Antonios, Sara, Ainhoa, y todos cuanto pasaron en algún momento por allí y nunca olvidaremos. A Iván, con el que he compartido y aprendido lecciones de trabajo y de vida. El mundo sería un lugar más humano y justo con más personas como él. Gracias a María Antonia y Natalia, que siempre han sabido escuchar mis problemas con cariño y ayudarme. A Gonzalo, el mejor compañero de despacho que se pueda tener y a Víctor Hugo, con el que no tuve oportunidad de compartir tantos momentos como hubiera deseado. A Karl, por tener tiempo siempre para iluminarme senderos de conocimiento. Thanks to Alan Zinober for teaching me mathematics, control theory, English jokes, delicious foods and great landscapes when I was living in England. And to my dear fellow Zhivko Stoyanov, who is as good person as he is tall (2 m!). I will never forget the moments we shared when his first son was born. And thanks to Alex and Iman, my Mexican and Iranian brothers, and to all the friends that I met in Sheffield. They helped me to enjoy such a wonderful city. Thanks to my kiwi family: my father Brent, my mother Penny and my grumpy uncle Steve. I learned a lot of science knowledge during my stay in New Zealand but, above all, they showed me the meaning of hospitality and how to be a better man. I don’t need to say that I will never forget them, because hopefully they will be forever in my life. Thanks again to all those wonderful people I met in that magical country: Rosa, Stephanie, Jenny, Kim, Sam, Malin, Edward, Ian, Alistair, Carlo, Markus, Siva, and all the staff in Plant and Food. A David. Nos conocimos hace mucho siendo aun estudiantes, quien me iba a decir que acabaría teniendo el honor de tenerle como director de tesis. Espero y deseo que todo el tiempo dedicado en este trabajo sea solo el comienzo de muchas y fructíferas colaboraciones. No imagino mejor compañero que un amigo como él. Este agradecimiento es poco para compensar el tiempo que me ha dedicado a costa de su familia. Gracias a ellos por extensión. A José Enrique. No sé si se puedo llamar trabajo a dedicarme a algo que me gusta tanto y además con amigos y maestros como Enrique, que lo hacen todo aún más fácil. Me ha enseñado mucho, y no solo en mi carrera investigadora, sino también en deliciosas lecciones de vida durante tantos viajes y jornadas de trabajo. Es para mí una fuente de inspiración inagotable. Gracias a sus hijos, por prestarme a su papi en tantas (demasiadas) ocasiones. A José Luís, por su eterno apoyo y dedicación. Cuando nací él ya estabas allí, muy cerquita. ¿Imaginaría entonces que algún día dirigiría mi tesis doctoral? Siempre he sentido a sus hijos como mis hermanos, así que eso le convierte en un segundo padre para mí, y sin duda lo ha sido en estos años de dedicación profesional. Gracias por creer en mí, gracias por enseñarme tanto, y gracias por darme la oportunidad de dedicarme a la profesión que adoro. Gracias a mi otra familia. Dicen que los amigos son la familia que se elige. Si esto es así, no podría haber elegido mejor, aunque no dejo de reconocer la enorme y hermosa fortuna que he debido tener para haberlos encontrado. Gracias a los Caimanes y Caimanas por entender mis ausencias en los momentos más exigentes de trabajo, y por regalarme tantos instantes de diversión, risas y emociones. Gracias a mis abuelos, que siempre me demostraron mucho más amor del que nunca merecí. A mis tíos y primos del Sur y del Norte, a los que adoro y espero no dejar nunca de ver. A mis suegros y cuñada por permitirme secuestrar de Almería su bien más preciado y estar siempre ahí cuando les hemos necesitado. Gracias a mis padres, que siempre antepusieron la felicidad de sus hijos a la suya propia. Todo lo que soy o pueda llegar a ser algún día se lo debo a ellos. Lucharé toda mi vida por honrarles, buscando cada día ser un poco mejor persona. Ahora que yo también soy padre, no imagino mejor forma de devolverles parte de lo que me han regalado todos estos años que revelándoles lo que espero ya sepan, que soy un hijo muy feliz y se lo debo a ellos. Gracias también a ellos por regalarme a mis hermanos, que son además mis mejores amigos, los que nunca me fallan, a los que nunca fallaré. Y gracias a mis cuñados, por su colaboración en la bendita tarea de hacerme tío dentro de unos meses. Y por supuesto, a mi alma gemela, mi hermosa flor de adversidad, a mi amada esposa Cris. Tú completas mi vida y me motivas a querer superarme. Atraes las sonrisas de mis días y ahuyentas la soledad de mis noches, haces que me sienta pleno y feliz. No hay noche que no recuerde la inmensa suerte que he tenido de haberte conocido. Y si todo esto fuera poco me regalaste al ser más hermoso que la tierra haya visto jamás: a nuestro hijo Héctor. Algún día, no demasiado lejano, cuando aprenda a leer, espero que encuentre en estas líneas un atisbo del inmenso amor que sentimos por él. Para ti Cris, y para ti hijo mío, va dedicada esta tesis y mi vida entera. Table of contents i Table of contents Figures .......................................................................................................................... 7 Tables .......................................................................................................................... 15 Symbols and abbreviations ......................................................................................... 17 Abstract ....................................................................................................................... 21 Resumen ...................................................................................................................... 23 Chapter 1. Introduction................................................................ 25 1.1. The need for better water management ................................................................ 25 1.2. Background .......................................................................................................... 27 1.3. Hypothesis and objectives ................................................................................... 30 1.4. Outline ................................................................................................................. 32 1.5. Publications by the author .................................................................................... 33 1.6. References ............................................................................................................ 35 Chapter 2. Impact of sustained deficit irrigation on yield, water productivity and fruit quality in a commercial citrus orchard, cv. Salustiana ..................................................................................... 39 2.1. Introduction .......................................................................................................... 40 2.2. Material and methods ........................................................................................... 42 2.2.1. Experimental site ...................................................................................................... 42 2.2.2. Irrigation treatments and experimental design ......................................................... 43 2.2.3. Plant measurements .................................................................................................. 44 8 Figures Fig. 3.4. Soil water content in the root zone for each treatment at the end of the fruit growth period of 2007 and 2008. The line inside the box shows the median and the letters after each box-whiskler plot show statistical differences at P < 0.05 level. ....... 78 Fig. 3.5. Evolution of midday stem water potential (Ψstem) in each treatment during the fruit growth and fruit maturation periods of the experimental years. Vertical bars indicate the standard deviation in each treatment (ten repetitions per treatment). DOY = day of year. ..................................................................................................................... 80 Fig. 3.6. Evolution of stomatal conductance (gs) in each treatment during the fruit growth and fruit maturation periods of the experimental years. Vertical bars indicate the standard deviation in each treatment (ten repetitions per treatment). DOY = day of year. ........................................................................................................................................ 81 Fig. 3.7. Box and whisker plots for water stress integral (SΨ) and stomatal conductance integral (Sg) during the studied years. Vertical lines indicate the standard deviation for each treatment. ................................................................................................................ 82 Fig. 3.8. Linear relationships among irrigation amounts (IAs) and water stress integral (SΨ) and stomatal conductance integral (Sg) in control and RDI treatments during the irrigation period. ............................................................................................................. 83 Fig. 3.9. Yield vs. annual irrigation amounts (IAs). Yield was normalized to take into account the temporal variability of the results due to changing weather conditions during the studied years. ................................................................................................. 83 Fig. 3.10. Box and whisker plot for yield (years 2007 and 2008). The line inside the box shows the median and the letters after each box-whiskler plot show statistical differences at P < 0.05 level. .......................................................................................... 84 Fig. 3.11. Box and whisker plot for water productivity (years 2007 and 2008). The line inside the box shows the median and the letters after each box and whiskler plot show statistical differences at P < 0.05 level. .......................................................................... 87 Fig. 3.12. Relationships among water stress integral (SΨ, MPa) and yield and fruit quality parameters (years 2007 and 2008). PT, peel thickness; TSS, total soluble solids; MI maturity index; TA, titratable acidity; ED and PD equatorial and polar diameter respectively. .................................................................................................................... 88 Figures 9 Fig. 3.13. Relationships among stomatal conductance integral (Sg, mmol s-1 m-2) and yield and fruit quality parameters (years 2007 and 2008). PT, peel thickness; TSS, total soluble solids; MI maturity index; TA, titratable acidity; ED and PD equatorial and polar diameter respectively. ............................................................................................ 89 Fig. 4.1. Effective precipitation (Pe), crop (ETc) and potential (ETo) evapotranspiration (A), and irrigation amounts (IA) for each treatment (B) during the irrigation season. DOY = day of year. ...................................................................................................... 105 Fig. 4.2. Dynamics of soil water content (SWC) in each treatment at 30 and 60 cm depth. SDI: Sustainable deficit irrigation; LFDI: Low frequency deficit irrigation. DOY = day of year. ................................................................................................................ 106 Fig. 4.3. Dynamics for stem water potential (A), stomatal conductance (B) and maximum daily shrinkage (C) in each treatment during the irrigation period. DOY = day of year. ................................................................................................................... 108 Fig. 4.4. Temporal evolution of the daily maximum stem diameter (MXSD) in each treatment during the studied period. SDI: Sustainable deficit irrigation; LFDI: Low frequency deficit irrigation. DOY = day of year. ......................................................... 110 Fig. 5.1. Flow diagram showing the tasks of the three main physical components of the CRP: the measuring unit (MU), the control unit (CU), and the pump & electrovalve controller (PEC). See text for details. ........................................................................... 128 Fig. 5.2. Output voltage (U) from the outer pair of thermocouples of one of the probe sets monitored by the measurement unit 1, before (A) and after (B) being amplified and filtered by the CRP. Amplified sections of the graphs correspond to the calculated tz values, indicated by the dashed lines. The X-axis shows the time (t) after the firing of the heat pulse. ............................................................................................................... 133 Fig. 5.3. Daily transpiration (Ep) values calculated by the CRP from the sap flow records in two OI trees  those for which the highest (∆) and the lowest (∇) transpiration rates were estimated (A). Fig. B is the same but for the NI trees. Also shown are the potential evapotranspiration (ETo) values estimated from the FAO56 10 Figures Penman-Monteith equation and the data recorded by the weather station next to the orchard (●). DOY = day of year. .................................................................................. 134 Fig. 5.4. Daily values of the transpiration ratio between the normally irrigated and overirrigated trees (EpNI/EpOI) determined by the CRP during the field trial (B), and the derived irrigation doses, ID (C). Irrigation began on day-of-year (DOY) 128. Water supply on DOY 199 corresponds to a recovery irrigation (see text for details). Also shown are the values of stem water potential measured at midday (Ψstem) in the instrumented trees (n = 6) (A) and the relative extractable water (REW) calculated from the soil water profiles measured in the rootzone of the instrumented trees (n = 7 for the NI trees and n = 3 for the OI trees) (D). Vertical bars represent ± the standard error. The crop evapotranspiration values (ETc), calculated as explained in the Materials and Methods section, are represented in panel C. ............................................................... 135 Fig. 6.1. Location of the treatments in the IFAPA experimental farm (1 = 100% IN: 2 = 75% IN; 3 = Treatment normally carried out in the orchard, ca 70% IN).................... 154 Fig. 6.2. Spatial variability of the soil electrical conductivity determined with an EM38- DD electromagnetic induction sensor in the area of orchard where the water treatments were imposed (marked area in Fig 1). Also shown are the representative locations for the 100% IN treatment (a1) and the 75% IN treatment (a2). ....................................... 155 Fig. 6.3. Cumulated values of the water supplies made by the AIC in each treatment, as recorded by the flow meters. The irrigation needs calculated by the AIC with Eq. 1 (se text for details), are also shown. ................................................................................... 158 Fig. 6.4. Root distribution observed by the trench method in October 18th. The trench was dug in the vertical of the irrigation pipe. There was as a dripper at x = 50 cm and another at x = 150 cm. .................................................................................................. 159 Fig. 6.5. Seasonal dynamics of the volumetric soil water content measured in both treatments at the maximum root depth (0.8 m) and at 1.5 m depth. See text for details on the treatments and on the measurements. Dashed lines represent the water contents at field capacity (-0.03 MPa) and wilting point (-1.5 MPa). ............................................ 159 Figures 11 Fig. 6.6. Seasonal dynamics of the average volumetric soil water content in the top 0.5 m of soil of each treatment. See text for details on the treatments and on the measurements. Dashed lines as in Fig. 6.5. .................................................................. 161 Fig. 7.1. Analysis scheme for two new heat-pulse methods. The left panel represents the Symmetrical Gradient method (HPSG) whereby the temperature difference signal is averaged over the first 60 s ( ). The right panel shows the Symmetrical Derivative method (HPSD) where we register the maximum value of the derivative of the temperature difference signal. ...................................................................................... 168 Fig. 7.2. Computer simulations of the performance of two new heat-pulse methods. The left panel shows the Symmetrical Gradient method (HPSG) and the right panel shows the Symmetrical Derivative method (HPSD). A strong linear relationship exists between these indicators ( and and the imposed heat-pulse velocity, HPVI. ............... 168 Fig. 7.3. Location of probes used in the willow experiment. Using just three sets of conventional probes, we were able to test and compare five different heat-pulse methods (CHP, T-max, HRM, CAG, HPSG and HPSD). Here, T is the temperature difference recorded following application of the 1-2 s heat pulse. ................................................ 169 Fig. 7.4. A comparison between sap flux density measurements obtained with the new Symmetrical Gradient (TRSG) and Symmetrical Derivative (TRSD) heat-pulse methods, versus branch transpiration calculated using a big-leaf model (TRmodel). ................. 171 Fig. 7.5. Scatter plots showing the relationship between sap flux density obtained from the new Symmetrical Gradient (TRSG) and Symmetrical Derivative (TRSD) methods and branch transpiration estimated by big-leaf model (TRmodel). ........................................ 171 Fig. 7.6. Comparing the temporal evolution of the heat pulse velocity from the new Symmetrical Gradient (SG) and Symmetrical Derivative (SD) methods and from the traditional compensation (CM) and T-max (TM) methods in the willow tree. ............ 172 Fig. 7.7. Linear correlations between the heat-pulse velocity from the new Symmetrical Gradient (SG) and Symmetrical Derivative (SD) methods and from the compensation method (HPVCM) in the willow tree. .......................................................................... 173 12 Figures Fig. 8.1. Inputs and outputs of the soil-plant-atmosphere model. Variations in the soil water content are calculated as the sum of the inputs: irrigation amount (IA) + effective precipitation (Pe) minus the sum of the outputs: transpiration (TR) + evaporation (ER) + drainage below the rootzone (D) .................................................................................. 181 Fig. 8.2. Main diagram of the Simulink model to estimate soil water content from the evapotranspiration (ETo), irrigation and precipitation inputs. Thetai(n) is a vector composed of the soil water content in each layer in day n. .......................................... 182 Fig. 8.3. Soil and plant water stress models (ER vs. EP and TR vs. TP). .................... 186 Fig. 8.4. Generalized crop coefficient (Kc) curve. DOY = day of year. ...................... 190 Fig. 8.5. Identification and validation data for irrigation and no irrigation periods. DOY = day of year. SWC = soil water content. ..................................................................... 190 Fig. 8.6. Dynamics of the soil water content (SWC) simulated with CROPSYST and the proposed SPA model. ................................................................................................... 198 Fig. 8.7. Trajectories for the actual (dotted line) and the simulated (solid line) soil water content (SWC). One-step validation results. DOY = day of year. ............................... 201 Fig. 8.8. Trajectories for the actual (dotted line) and the simulated (solid line) soil water content (SWC). N-step validation results. DOY = day of year. ................................... 202 Fig. 8.9. Temporal evolution of the soil water content (SWC) when no irrigation was applied. DOY = day of year. ........................................................................................ 205 Fig. 8.10. Temporal evolution of the soil water content (SWC) when irrigation amounts are calculated as 100% of the crop evapotranspiration (ETc) in the previous day. DOY = day of year. ................................................................................................................... 205 Fig. 8.11. PID control scheme. ..................................................................................... 207 Fig. 8.12. Temporal evolution of A: the soil water content (SWC, A) and B: the irrigation amounts (IA) and potential evapotranspiration (ETo) when PID irrigation controller is applied. ..................................................................................................... 207 Fig. 8.13. Simulink model for the simulation of the proportional, integral and derivative (PID) + feedforward ETc irrigation controller. ............................................................ 208 Figures 13 Fig. 8.14. Comparison of the trajectories of the soil water content (SWC) when applying PID and when applying PID with feedforward ETc. IA = irrigation amount. DOY = day of year. .......................................................................................................................... 208 Fig. 8.15. Comparing soil water content trajectories (A) and irrigation amounts (B) of the MPC and PID controllers assuming no precipitations. DOY = day of year. .......... 209 Fig. 8.16. Comparing soil water content trajectories (A) and irrigation amounts (B) of the MPC and PID controllers assuming precipitation events. DOY = day of year. ..... 210 Fig. 8.17. Soil water content (SWC) trajectories (A) and irrigation amounts (IAs) applied (B) by the MPC with (MPCp) and without (MPCnp) the predictions of changes in the set point. ............................................................................................................. 212 Fig. 8.18. Soil water content (SWC, A) and irrigation amounts (IA, B) when PID was applied in a real field experiments, from DOY 186 to 199. Setpoint was set to 180 mm. ...................................................................................................................................... 214 Fig. 8.19. Soil water content (SWC, A) and irrigation amounts (IA, B) when PID was applied in a real field experiments, from DOY 208 to 220. Setpoint was set to 180 mm. ...................................................................................................................................... 215 Tables 15 Tables Table 2.1. Applied water and water balance characteristics in irrigation treatments. ... 47 Table 2.2. Yearly values of water stress integral (SΨ) in each treatment during the irrigation periods. ........................................................................................................... 51 Table 2.3. Average yield, water productivity and fruit quality parameters for 2005. ... 52 Table 2.4. Average yield, water productivity and fruit quality parameters for 2006. ... 52 Table 2.5. Average yield, water productivity and fruit quality parameters for 2007. ... 52 Table 2.6. Temporal variability analysis for all parameters listed by Tukey’s test. ...... 54 Table 2.7. Pearson’s correlation coefficients among studied parameters for the 3-year period. ............................................................................................................................. 54 Table 3.1. Irrigation treatments. The aimed irrigation amount at different main phenophases is shown in terms of % of the irrigation needs. ......................................... 68 Table 3.2. Main water balance components for the irrigation period at main phenophases. ................................................................................................................... 71 Table 3.3. Irrigation water applied, actual water stress ratio and water savings for each treatment at different phenophases. ................................................................................ 76 Table 3.4. Yield components and fruit quality parameters. ........................................... 85 Table 3.5. Water stress integral, stomatal conductance integral, yield and fruit quality parameters for the entire study period (2007-2008) (Normalized data). ........................ 86 Table 4.1. Total irrigation amounts (IA) and water savings (WS) as compared to the control for each treatment. The experimental period was from day of year 152 to day of year 286. See text for details on the treatments. ........................................................... 105 Table 4.2. Correlation coefficients among between the studied parameters. .............. 110 16 Tables Table 4.3. Yield and fruit quality parameters. ............................................................. 112 Table 8.1. Variables and parameters of the SPA model. ............................................. 184 Table 8.2. Variables and parameters of the SPA model. ............................................. 198 Table 8.3. Identified values the four models of the crop coefficient (Kc). ................... 200 Table 8.4. Maximum and averaged relative errors for 1-step and N-step validation .. 203 Table 8.5. Irrigation ammount (IA) and mean squared error (MSE) for the tested irrigation controllers. .................................................................................................... 214 Symbols and abbreviations 17 Symbols and abbreviations Abbreviations Description AIC automatic irrigation controller CAG calibrated gradient method CHP compensation heat pulse CRM coefficient of residual mass CRP control de riego de la plantación CU control unit DI deficit irrigation FDR frequency domain reflectometry FI full irrigation HPSD heat pulse symmetrical derivative method HPSG heat pulse symmetrical gradient method HPV heat pulse velocity HRM heat ratio method IA irrigation amount LFDI low frequency deficit irrigation LVDT linear variable displacement transducer MIMO multiple inputs multiple outputs MLD mixed logical dynamical MPC model predictive control MSE mean squared error MU measurement unit NI normally irrigated OI overirrigated PCA principal component analysis PCi i-th principal component PEC pump and electrovalve controller PID proportional-integral-derivative RDI regulated deficit irrigation RIA red de información agroclimática SISO single input single output SDI sustained deficit irrigation SDI53 treatment irrigated at 53% ET c SDI67 treatment irrigated at 67% ETc SDI77 treatment irrigated at 77% ET c SPA soil-plant-atmosphere TDV trunk diameter variations WUE water use efficiency 24 Resumen de humedad en el suelo, se evaluó en una plantación de almendros, demostrándose su utilidad para reducir las pérdidas de agua por drenaje, evaporación y escorrentía. Uno de los enfoques más innovadores y prometedores para la automatización del riego es el basado en la medida del flujo de savia en órganos conductores de la planta. Su correcta aplicación requiere de sensores capaces de medir de forma fiable rangos amplios del flujo de savia. La mayor parte de los sensores disponibles en el mercado funcionan bien en rangos más restrictivos, es decir, no son fiables para el caso de flujos de savia muy bajos o muy altos. Una de las principales contribuciones de esta tesis ha sido el desarrollo y evaluación de dos nuevos métodos de medida de flujo de savia capaces de ampliar el rango de medida respecto de los métodos actuales, permitiendo, además, medir flujos inversos, lo cual es de sumo interés para el estudio de fenómenos relacionados con la elevación hidráulica en raíces de árboles frutales. Más adelante se muestran nuestros trabajos de modelización del sistema sueloplanta-atmósfera y de diseño de estrategias de riego basadas en el uso de reguladores de la humedad en suelo, en particular del tipo proporcional-integral-derivativo y predictivo basado en modelo. El modelo matemático que presentamos se ha implementado en un entorno gráfico e intuitivo, apto para realizar simulaciones y para el desarrollo de controladores avanzados basados en modelo. Se evaluaron y compararon los resultados en simulación de diferentes estrategias de control. Uno de estos controladores fue también ensayado con éxito en una parcela de almendros, en dos periodos diferentes de una campaña de riego. En el último capítulo de esta tesis mostramos que el uso de técnicas y resultados de teoría de control puede tener un gran impacto en la mejora de los sistemas de riego y el uso eficiente de los recursos hídricos. 25 Chapter 1 Introduction 1.1. The need for better water management Water is a scarce resource1, and its rational use is compulsory. Problems derived from lack of water will likely increase, if long-term predictions on global climate change are right. Meteorological records suggest significant increases in temperature and decreases in annual precipitation, which will entail a reduction of the available water resources of the XXI century2. Industry and tourism, among others productive activities, compete for this resource increasing its profitability and productivity. Nowadays, the economic sector that most fresh water consumes is agriculture: ca. 70% of the total resources, against the 20% used by industry and the 10% for domestic use. Data for Spain and Andalusia shows a similar distribution at national and regional scales (Fig. 1.1). Irrigated agriculture is of crucial importance in the economy of Andalusia. The irrigated surface in the region accounts for only 23% of the total agricultural surface, with an average water consumption of 4.761 hm3 year-1, but it generates 57% of the total yield and 60% of the agricultural employment (Plan Nacional de Regadíos, Horizonte 2008). Considering the expected increase in world population3, it is urgent to find solutions to ensure enough food supply. This can be only achieved by increasing the world agricultural yield and water productivity, mainly from the irrigated areas as suggested by the mentioned data. 1 The world’s consumption of water is doubling every 20 years, which is more than twice the rate of our population increase (Clothier y col., 2008). 2 Climate change, water and food security. FAO Water Reports 36, 201. 3 World population will hit 7 billion by 2012 and 9 billion by 2050 (ONU, 2008). 26 Chapter 1 Fig. 1.1. World (A), national (B) and regional (C) water demand distributions by sectors. 0 10 20 30 40 50 60 70 80 Water demand (%) World water demand Domestic Industry Irrigation A 0 10 20 30 40 50 60 70 80 Water demand (%) Spanish water demand Domestic Industry Irrigation B 0 10 20 30 40 50 60 70 80 90 Water demand (%) Water demand of the Guadalquivir River Valley Domestic Industry Irrigation C Introduction 27 This justifies the need to improve our understanding of the dynamics of the water used by the plants in order to develop methods aimed at optimizing water consumption. Indeed, this is the main motivation of this doctoral thesis. 1.2. Background Archaeological discoveries have identified evidence of irrigation since ancient times. A form of water management called basin irrigation began at about the same time in Egypt and Mesopotamia ca. 8000 years ago (Taylor and Ashcroft, 1972), using the water of the flooding Nile or Tigris/Euphrates rivers. Nowadays, deficit irrigation (DI) strategies have proved to be efficient for achieving significant water savings with minimum reductions in crop performance. They are, in fact, the most advisable irrigation strategies for arid and semi-arid regions. The effects of DI in the performance of different fruit tree species are documented in a variety of publications which outline the advantages and shortfalls of this practice. Some authors reported that the effects of deficit irrigation on yield mainly depend on the growth stage of the crop (Doorenbos and Kassam, 1979; Ginestar and Castel, 1996; García-Tejero et al., 2008). On the other hand, the negative effects from yield reductions may be partially mitigated by improvements in fruit quality (Sánchez-Blanco et al., 1989; González-Altozano and Castel, 2000; Verreynne et al., 2001). Furthermore, both the irrigation strategy and the cultivar definitely affect the yield response to deficit irrigation (Ginestar and Castel 1996; Treeby et al., 2007; Pérez-Pérez et al. 2010). Different deficit irrigation strategies have been proposed, which differ mainly on the criteria for water distribution during the irrigation season. Among the most widely used are sustained deficit irrigation (SDI), which applies a constant water restriction throughout the season; regulated deficit irrigation (RDI), which involves the application of water shortages depending of crop growth stage; and low frequency deficit irrigation (LFDI), based on cycles of irrigation withholding and rewatering, designed according to a targeted crop water status, as determined from plant-based measurements (leaf or stem water potential, sap flow, trunk diameter fluctuations...). See Ruiz-Sanchez et al. (2010), for details on these and other DI strategies. 28 Chapter 1 As early as 1859, Charles Darwin made a first reference to regulated deficit irrigation, perhaps the first documented one. Thus, in his famous work "The Origin of Species" noted that: “In some few cases it has been discovered that a very trifling change, such as a little more or less water at some particular period of growth, will determine whether or not a plant will produce seeds”. The work by Fereres and Soriano (2007) summarizes the principles of and the need for deficit irrigation. Later, Ruiz-Sanchez et al. (2010) reviewed the use of DI in fruit trees growing in Spain. These references provide an extensive bibliography on SDI and RDI strategies. To the best of our knowledge, very little research has been made on the effect of the LFDI strategy on the physiological response, water productivity and yield of fruit trees. Among the very few papers on this subject, Hutton et al. (2007) addressed the question of timing irrigation to suit citrus phenology. They, however, did not use any physiological criteria to fix the intervals between irrigation supplies. A different approach for optimizing irrigation is the use of automatic irrigation controllers based on feedback, feed-forward strategies or a combination of both. Feedback is a mechanism, process or signal that is looped back to control a system within itself. Irrigation controllers use the information of the consequences of previous actions to calculate the next irrigation dose. In feed-forward strategies, the controllers use known or estimated values of future disturbances to compensate their effects in advance. In the field of automatic irrigation, measurements of soil, plant or atmosphere variables related to the plant water status provides the feedback and feed-forward signals. Feedback control can be said to have originated with the float valve regulators of the Hellenic and Arab worlds (Mayr, 1975), however it does not appear to have spread to medieval Europe. It seems rather to have been reinvented during the industrial revolution, where level, temperature and finally Watt’s centrifugal governor where developed (Dickinson and Jenkins, 1927). Since then, automatic control has been applied in almost all engineering fields with great success; see Bennett (1996) for a brief history of automatic control, although the impact in agriculture, and in particular in precision irrigation, is limited. Most of the irrigation controllers available on the market require the irrigation dose to be provided by the user. Only then, they are able to switch on/off the irrigation pump and to open or close the valves to apply the irrigation doses to every sector of the Introduction 29 orchard. Very few automatic irrigation controllers are able to calculate autonomously the irrigation dose. Most are based on measurements of soil matric potential (Luthra et al., 1997; Klein, 2004; Miranda et al., 2005). These devices are relatively inexpensive and easy to use, but ground water measurements imply certain limitations: they require a large number of sensors and do not take into account the plant status and response. Protocols for automatic irrigation controllers have been reported based on trunk diameter variation (Goldhamer and Fereres, 2004; Garcia-Orellana et al., 2007) or sap flow measurements (Fernández et al., 2001, 2008). Both methods are considered having a great potential for irrigation control (Fereres et al., 2003; Jones, 2004). Unfortunately, most of the commercially available sap flow sensors are not reliable both for very low and very high sap flows. Recently, there has been an increasing interest on developing mathematical models representing both, the dynamics of water in the soil-plant-atmosphere (SPA) system and crop performance. Among the most popular models are WAVE (Vanclooster et al., 1994), SPASMO (Green, 2001), SWAP (Van Dam, 2000; Van Dam et al., 2008), MACRO (Larsbo and Jarvis, 2003), CROPGRO (Boote et al., 1998), WOFOST (Van Diepen et al., 1989) and DSSAT (Hoogenboom et al., 2004). The utility of these models both for understanding the simulated process and for optimizing related practices has been widely proved. They are able to simulate the effect of irrigation strategies, climate conditions or diseases on crop performance. Unfortunately, the current models show certain limitations. The great variability of every component of the soil-plant-atmosphere system makes difficult having good quality input data. This issue limits the accuracy of the results, especially when extrapolating situations very different from those occurred during the identification process. Another problem is that these models usually require the identification of many parameters, which means that they are not user-friendly models. Furthermore, they are often too complex, so their use requires a specific training that can be given to a reduced number of users and are not suitable for designing advanced irrigation controllers based on these models. Recent research in modelling has provided a better understanding of the SPA system. It is now possible to test automatic irrigation controllers in computer simulations prior to their use in field experiments. These controllers can be based on 30 Chapter 1 simple laws such as on / off strategies in which irrigation is switched on or off when certain threshold values are overcome or based on more advanced laws, such as proportional-integral derivative (PID) or model predictive controllers (MPC). In a PID controller, the control signal is generated as a weighted sum of three terms: the error between the process variable and the setpoint, the integral of recent errors, and the rate by which the error has been changing. MPC is based on the use of a model to predict the mathematical evolution of the system, on the minimization of a cost function based on this prediction and on the use of a receding horizon strategy. These controllers, although successfully and extensively used in other areas of science and industry, see for example Astrom and Hägglund (2006) and Camacho and Bordons (1997), have been seldom applied in agriculture. However, we might find promising examples, especially in the management of greenhouses environmental control (Rodriguez et al., 2008; Piñón et al., 2005; El Ghoumari et al., 2005). In the next section we show the hypothesis and objectives of this thesis, which aims to cover some of the mentioned gaps of the current knowledge. The reader can find additional publications to those mentioned here in the Introduction section of each chapter of this thesis. 1.3. Hypothesis and objectives The main objective of this thesis is to provide new techniques to optimize the irrigation water consumption in woody crops as well as to get an insight of the soil-plant- atmosphere system. To this end, identification, modelling and control tools from control theory as well as deficit irrigation strategies have been applied. We present results from experiments with three important species in Andalusia, with contrasting water use behaviours: orange, olive and almond trees. Next, the hypothesis and corresponding objectives of this thesis are presented. Hypothesis 1. Different DI strategies have a different impact on water productivity and yield quality. Objective 1. To evaluate the effects of various DI strategies on fruit yield and fruit quality in citrus orchards of the Guadalquivir River Valley. Introduction 31 Hypothesis 2. Irrigation of woody crops can be controlled from sap flow measurements in trees under the orchard irrigation conditions and in control, fullyirrigated trees. Objective 2. To design, implement and test the software and hardware of an automatic irrigation controller, able to calculate and apply suitable irrigation doses to mature olive trees from sap flow measurements in the trunk of deficit and fully irrigated trees. Hypothesis 3. Precise irrigation needs can be estimated from weather data and soil water content measurements, and used to minimize water losses by evaporation, drainage and runoff. Objective 3. To design, implement and test the software and hardware of an automatic irrigation controller suitable for fruit tree orchards, able to calculate and apply daily irrigation from soil water content and weather measurements. The device was tested in an almond orchard. Hypothesis 4. Any sap flow method suitable to optimize irrigation must be able to measure low transpiration values. Objective 4. To model, simulate and test new sap flow methods with a better performance than the currently available ones, especially for estimating low sap flow values. The method was evaluated both in computer simulations and in the stem of a willow tree. Hypothesis 5. Simplified models able to design control laws could overcome the shortfall of current simulation models based on the SPA system, which are not suitable for the design and simulation of automatic irrigation controllers. Objective 5. To design and implement a SPA model in a commercial simulation platform and validate it with a model widely used in agronomy and with data from field experiments. Hypothesis 6. Irrigation controllers based on feedback of measurable variables of the SPA system such as soil water content, weather forecasts and dynamic models of the SPA system can optimize irrigation water use. 32 Chapter 1 Objective 6. To design, implement and test both in simulation and in field experiments automatic irrigation controllers based on well known techniques from control theory. 1.4. Outline Since most of the results of this thesis have already been reported by the author in scientific journals and workshops, we considered appropriate to structure the thesis into chapters corresponding to the most important publications by the author (Section 1.5), directly related to the hypothesis and objectives previously mentioned (Section 1.3). Thus, Chapters 2, 3 and 4 deal with the development of experiments aimed to study three different deficit irrigation strategies and their effect on yield and fruit quality, namely sustained deficit irrigation, regulated deficit irrigation and low frequency deficit irrigation with withholding & rewatering cycles. We evaluated the effect caused by the applied deficit irrigation strategies during a whole irrigation season in three different orchards located in the Guadalquivir River Valley. We estimated water productivity, water savings and the physiological response of the crop (stem water potential, stem diameter variations, stomatal conductance...). At the end of the season, we determined the effects on yield and in the organoleptic properties of the fruits. In Chapter 5 we designed and tested an automatic irrigation control system for fruit tree orchards, denominated CRP. Control was based on sap flow measurements in the trunk as a feedback and sap flow readings in overirrigated irrigated trees were used as a reference. The controller was successfully tested in an olive orchard (Olea europea cv. Manzanilla de Sevilla), in the experimental farm “La Hampa”, belonging to the Consejo Superior de Investigaciones Científicas (CSIC) and located in the municipality of Coria del Rio. In Chapter 6 we developed an automatic irrigation system in an almond orchard, in which we had plant physiological sensors (sap flow and dendrometers), soil moisture sensors and a nearby weather station. The controller used weather data as a measurement of the disturbance to be compensated in the system and soil moisture data as a feedback to prevent flooding situations, regulating the irrigation intervals to maintain the soil water content levels between threshold values defined in previous Introduction 33 trials. The automatic irrigation system was evaluated in an almond orchard belonging to the Centro Las Torres-Tomejil IFAPA (TM Alcalá del Río, Seville). In Chapter 7, we introduce new methods to estimate plant water consumption from sap flow measurements in trees. Sap flow monitoring is one of the most promising approaches to automate precision irrigation systems. We developed field experiments in plants, in which we compared our new methods against other heat-pulse techniques, to evaluate the proposed solutions. In Chapter 8 we show the design, identification, simulation and validation processes of a representative mathematical model of this system. We used the model to design and tune both classical proportional-integral-derivative and model-based advanced controllers. In particular, we developed model based predictive controllers. In addition, an automatic irrigation controller based on a proportional-integral-derivative control law was applied in a real case, specifically in the almond orchard described in chapter 6. Finally, in Chapter 9 we present concluding remarks and future work. 1.5. Publications by the author Most of the results presented in this thesis have already been reported by the author in scientific journals and workshops. Major publications that have contributed to the achievement of this thesis are listed below: Fernández JE, Diaz-Espejo A, Torres-Ruiz JM, Muriel JL, Romero R, Morales-Sillero A, Martín-Palomo MJ. 2009. Seasonal Changes of Hydraulic Conductance of Mature Olive Trees under Different Water Regimes. Acta Hort. (ISHS) 846:263- 270. Fernández JE, Romero R, Díaz-Espejo A, Cuevas MV, Muriel JL, Montaño JC. 2008. A device for Scheduling irrigation in fruit tree orchards from sap flow readings. Acta Horticulturae, 792: 283-290. (Chapter 5). 40 Chapter2 variables and to establish a correlation matrix. Finally, a classification tree was derived and principal-component analysis was undertaken. This allowed us to identify and evaluate the variables that better explain the effect of irrigation treatments on the crop. In conclusion, the tested SDI strategies increased water productivity and fruit quality with non significant reductions in yield. They can be considered, therefore, as advisable strategies for orange orchards (Citrus Sinensis L. Osbeck, cv. Salustiana) in our region. 2.1. Introduction From 70% to 80% of the worldwide water usage is associated with agriculture, the greater percentages corresponding to arid and semi-arid areas. This has created competition for water with other sectors, such as industry or tourism, in which water is more commercially valuable than in agriculture. However, the socio-economic value of irrigated land in Andalusia exceeds 3.5-fold the value in extensive agriculture on nonirrigated land (Berbel and Gutiérrez, 2004). The current area cropped with citrus in Andalusia (74,000 ha) has increased by 2.04% annually for the last 25 years. The expected production of citrus in the region will be close to 2.3 million tons by 2015 (30% of the national production). Of the total area of citrus in Andalusia, 51,500 ha are dedicated to sweet orange, covering the Salustiana and Valencia cultivars almost 20% of that area. The average annual irrigation water consumed in Andalusia is close to 4,761 hm3, with about 10% diverted to citrus groves (García-Tejero et al., 2008). Weather forecasts for the 21st century predict significant increments in temperature and major reductions in the annual precipitation, which may led to an estimated 17% decline in the available water resources for agriculture worldwide. An increase in potential evapotranspiration (ETo, mm) of over 20% is expected in the Guadalquivir river valley by the year 2050. The biggest increases will occur in the most western areas where the majority of arable land is concentrated (Rodríguez Díaz et al., 2007). Under such restrictive conditions, finding new cultivation strategies is becoming a priority. They must be focussed on reducing water consumption and on making more efficient the use of the available water resources. Thus, maximizing the use of water saving techniques and improving crop productivity is crucial for a rational agriculture. Impact of sustained deficit irrigation on yield, water productivity… 41 This can be achieved with deficit irrigation (DI) strategies, i.e. by applying lower amounts of irrigation water than those needed by the crop. Plants undergo stress when the water taken up by the roots fails to compensate for the transpiration driven by atmospheric conditions. When the water stored in the soil is being depleted the actual crop evapotranspiration (ETc, mm) decreases. This usually affects photosynthesis and lowers carbon assimilation (Hsiao, 1973). Consequently, vegetative development and crop yield are reduced (González-Altozano and Castel, 2000). In recent years, several contributions have documented the advantages of using deficit irrigation (DI) strategies to improve water saving and fruit quality in citrus trees (Southwick and Davenport, 1986; Ginestar and Castel, 1996; González-Altozano and Castel, 1999, 2000; Muriel et al., 2006; García-Tejero et al., 2007, 2008). Shalhevet and Bielorai (1978) and Doorenbos and Kassam (1979) reported that the response of the citrus trees to water stress depends strongly on the phenological stage of the crop. Other authors agreed with that statement, and also outlined the importance of the pedoclimatic characteristics of the orchard (Ginestar and Castel, 1996, and Treeby et al., 2007, in orange orchards; Sánchez-Blanco et al., 1989, in lemon orchards). Thus, a water deficit at the end of the fruit growing period increases juice acidity and soluble-solid contents (González-Altozano and Castel, 1999; Hutton et al., 2007). Other authors have shown that a reduction in the available soil water during flowering could decrease yield, because it may affect the fruit-setting process (Doorenbos and Kassam, 1979; Castel and Buj, 1990). Pérez-Pérez et al. (2008) reported that water stress during flowering and fruit setting reduces the number of fruits. Concerning fruit weight and fruit diameter, the response of fruit weight and fruit diameter to water stress is species-dependent. Thus, it has been reported that DI conditions are responsible for a greater number of smaller fruits, since the lack of water reduces fruit weight and diameter (Treeby et al., 2007). Peng and Rabe (1998), however, worked with ‘Satsuma’ and found that fruit size was not affected by water deficit. The term water productivity (WP, kg·L-1) refers to the crop production per unit of water applied, rather than per unit of irrigated surface (Fereres and Soriano, 2007). The effects of DI on yield and fruit quality in orange groves will depend mainly on the 42 Chapter2 plant material, and the intensity, duration, and timing of the imposed water deficits (Vaux and Pruitt, 1983; Ginestar and Castel, 1996). In any case, the correct application of any DI strategy should boost WP with a minimum yield decrease. Sustained deficit irrigation (SDI) is a widely used DI strategy. Basically, SDI consists on applying a reduced percentage of ETc all throughout the irrigation season. The aim of this work was to evaluate the impact of SDI treatments with different levels of water stress on yield and fruit quality in a commercial citrus orchard, cv. Salustiana, under non-tillage soil management. The work, which lasted for three years (2004-2007), was focused on evaluating plant-water relations in the orchard with the aim of improving WP. 2.2. Material and methods 2.2.1. Experimental site This work was made in a commercial orchard located in the Guadalquivir River Valley, SW Spain (37º 44´ 57´´ N, 5º 10´ 6´´ W), planted with 12-year-old orange trees (Citrus sinensis, L. Osbeck, cv Salustiana) grafted on Citrange Carrizo (Citrus sinensis, L. Osbeck x Poncirus trifoliata, L. Raf.). The trees, spaced 6 m x 4 m, were, on average, 3.25 m in height and 4.0 m in diameter. They were planted on ridges 0.4 m high and 4 m wide, with 2 m between ridges. The orchard was under non-tillage conditions. From October to May grass covered the row between ridges. In October, immediately after pruning the trees, the grass was mechanically cut and left on the soil orchard. Later in the year, from the beginning of the period of high atmospheric demand, herbicides were used to keep the soil free of weeds. The soil is a calcareous sandy-clay loam fluvisol (FAO, 1998) with an effective depth of 0.6 m. At deeper depths, the soil has high contents of clay and calcic carbonate. In the top 0.6 m soil layer, textural values are 56.65 ± 5.0% sand, 21.99 ± 3.0% silt, and 21.39 ± 2.5% clay. The soil is rich in calcium (75.4 cmol kg-1) and poor in both nitrogen (460 mg kg-1) and organic matter (1.0 %). The water-holding capacity is 107 mm on average and the bulk density ranges from 1.23 to 1.30 Mg m-3. Nutrients were applied twice per week from the end of March to October. Each treatment received a total of 150 kg ha-1 of N, 70 kg ha-1 of P2O5 and 110 kg ha-1 of K2O. Impact of sustained deficit irrigation on yield, water productivity… 43 The climate is typically Mediterranean, with dry, hot summers and mild, wet winters. Average annual ETo and rainfall values are 1400 mm and 534 mm, respectively (period 1971-2000). Rainfall is distributed mainly from late autumn to early spring, with November to February being the wettest months. Temperature in winter rarely falls below 0 ºC; in July and August may peak to over 40 ºC. 2.2.2. Irrigation treatments and experimental design Four irrigation treatments were applied over three consecutive years (2005-2007): 1) control, where water was applied to satisfy 100% of the irrigation needs (IN, mm); 2) SDI77, at 77% of IN; 3) SDI67, at 67% of IN; 4) SDI53, at 53% of IN. Each year, the treatments were implemented from May-June to October-November. Irrigation in each treatment was controlled automatically, with a head-unit programmer and electrohydraulic valves. The irrigation system consisted of two laterals per tree row with selfcompensating drippers 1 m apart (8 drippers per tree). The discharge rate of the drippers depended on the treatment: control, 3 L h-1; SDI77, 2.3 L h-1; SDI67, 2 L h-1; SDI53, 1.6 L h-1. Five 288 m2 plots per treatment were distributed in a randomized complete block design. Each plot consisted of three rows with four trees per row. Measurements were made in the two central trees of each plot, termed here as sample trees. Seasonal values of ETo were calculated with the FAO56 Penman-Monteith equation (Allen et al., 1998), with data from an automatic weather station nearby the orchard. Irrigation needs (IN, mm) were weekly calculated as: Eq. 2.1 where Kc is the crop coefficient, Kr is a reduction coefficient accounting for the percentage of ground surface covered by the crop and Pe is the effective precipitation. Although the use of monthly values of Kc may lead to more accurate estimations of ETc than one single Kc value for the whole season, we assumed Kc = 0.7 all throughout the 44 Chapter2 season, because this is the usual practice in the region. This will help growers to adopt our results. We considered Kr = 1 after Fereres and Castel (1981), since more than 50% of the ground surface was covered by the canopies of the trees. 2.2.3. Plant measurements Values of Ψstem were obtained with a pressure chamber (Scholander et al., 1965) following Turner (1988). Once every 10-15 days, two mature leaves were sampled from the north quadrant of each one of two sample trees of each plot. Measurements were made between 12.00 and 14.00 Greenwich Mean Time (GMT). SΨ was estimated according to a modified equation derived from that proposed by Myers (1988), which integrates the water-potential values over the period for which the trees are stressed: Eq. 2.2 Ψi and Ψi+1 are the Ψstem values measured in two sampling days (ni and ni+1). The assumption implicit in the calculation of SΨ is that stressing conditions reflected in Ψstem reduce the growth rate. SΨ is an approximation of the integral of the rate. For example, the increment of the basal area reached at the end of the growing period will be a function of SΨ calculated over the complete period. At the end of each season, the total fruit weight of each of the two sample trees of each plot was determined. For each treatment, WP was calculated by dividing the yield (kg) by the volume of water applied (L). For each SDI treatment, we also studied the relationships between the relative ratio of SΨ (SRI) and the total applied irrigation amounts (IAs, % IN), SRI was defined as: Eq. 2.3 where subindices i and j stand for treatment and year, and SΨ(control)j is the SΨ of the control treatment in each year. Impact of sustained deficit irrigation on yield, water productivity… 45 Fruit-quality characteristics were analysed after harvest. Ten fruits from each of the two sample trees of each plot were used for analysis. For each fruit we measured equatorial (ED, mm) and polar (PD, mm) diameters, peel thickness (PT, mm), fruit weight and juice content. Total soluble solids content (TSS, ºBrix) was measured with a digital refractometer PR-101. Titratable acidity (TA, g L-1) was determined by titrating the samples with NaOH 0.1 N by the colorimetric method, using phenolphthalein as indicator solution. Maturity index (MI, ºBrix L g-1) was then calculated by dividing TSS by TA. This is a key parameter to determine the optimal time for harvesting. 2.2.4. Statistical analysis Data of each year were subjected to a one-way variance analysis (ANOVA; SPSS statistical package; SPSS, Chicago, IL, USA) with four irrigation treatments and ten replicates per treatment, using Tukey’s test for mean separations (P<0.05). With this method, yearly values of yield, WP and fruit quality were compared between treatments. A similar analysis was carried out with the whole dataset (2004 to 2007) to evaluate the temporal variability by comparing the same parameters between years. An overall analysis was made for evaluating the correlations between the studied parameters. Datasets were previously normalized following Sterk and Stein (1997), whose methodology allows a dataset to be analysed when different conditions have occurred during sampling (in our case, different years differing in meteorological conditions). Then, the Pearson’s correlation coefficient was calculated for the whole period. When applied to two variables, the coefficient is defined as the covariance of the two variables divided by the product of their standard deviations. With a value within ±1, the coefficient informs on the correlation (linear dependence) between the variables. A classification tree (Rawls and Pachepsky, 2002) was derived using the R software (Team RDC, 2008) to determine the relationships between the variables and to group the complete dataset into the different irrigation strategies. The classification tree is based on the calculation of the deviations of each variable from its mean value and the corresponding deviation of each group from the mean of the group for different thresholds. The variable and threshold that give the maximum deviations are selected 46 Chapter2 and classification is continued in this manner until determining a minimum number of elements belonging to a particular group. We also used the R software to make a principal component analysis (PCA). The PCA (Ihaka and Gentleman, 1996) is a factor analysis protocol used to identify variables or underlying factors that better explain the correlation or covariance matrix of several variables (Davis, 2002). Thus, linear combinations of the factors or components may explain a large part of the found variability, thus reducing the number of variables required to explain the total variability. 2.3. Results 2.3.1. Water relations The main components of the water balance (Pe, ETc and IA) in the irrigation treatments are shown in Table 2.1 and Fig. 2.1. Rainfall values registered during each irrigation period were quite similar except for 2005, when the Pe was 89% and 84% lower than in 2006 and 2007 respectively, thus increasing the water deficit (ETc-Pe) to as much as 679 mm, 8% higher than in 2006 and 22% higher than in 2007. Rainfall was very low during the period of highest atmospheric demand from mid-June to mid-September. For all treatments, IA was quite close to the aimed amounts, except in 2007. 2.3.2. Plant-water measurements The influence of IA on Ψstem is shown in Fig. 2.2. In 2005 and 2007, lack of water at the end of August (DOY 240) caused a reduction in the IAs, which led to a decrease of Ψstem down to -1.5 MPa and -1.8 MPa in 2005 and 2007 respectively. The greatest variability of Ψstem along the season was found in the SDI53 treatment. Maximum value of -0.5 MPa for Ψstem was recorded in the control treatment in 2005. The seasonal pattern of the plant water status, as indicated by Ψstem, in the SDI treatments did not appreciably differ from control. Significant differences were found between treatments, but only during the period of highest atmospheric demand from mid-June to mid-September. Differences were more evident between the SDI53 and the control treatments. These differences were reflected in SΨ. Thus, the control treatment Impact of sustained deficit irrigation on yield, water productivity… 47 Table 2.1. Applied water and water balance characteristics in irrigation treatments. Years 2005 2006 2007 Average IP1 (DOY) 152-282 114-279 133-312 158 Pe2 (mm) During IP 13 115 81 70 Annual 296 436 347 360 ETc (mm) During IP 692 745 639 692 Annual 1026 1041 935 1001 Irrigation treatments control IA (mm) 704 653 658 672 IA 3 (% IN) 104 104 118 108 WB4 25 23 100 49 SDI77 IA (mm) 540 501 504 515 IA (% ETc) 80 80 90 83 WB (mm) -139 -129 -54 -107 SDI67 IA (mm) 469 436 438 448 IA (% ETc) 69 69 78 72 WB (mm) -210 -194 -120 -175 SDI53 IA (mm) 376 348 351 358 IA (% ETc) 55 55 63 58 WB (mm) -303 -282 -207 -264 1 Irrigation Period; 2 Effective precipitation; 3 Irrigation amount; 4 Water balance (=IA+Pe-ETc) SDI77, low deficit irrigation; SDI67, moderate deficit irrigation; SDI53 severe deficit irrigation. DOY, day of year 48 Chapter2 Fig. 2.1. Effective precipitation (Pe), crop (ETc) and potential (ETo) evapotranspiration, and irrigation amounts (IA) for each treatment during the tree experimental years. DOY = day of year. Impact of sustained deficit irrigation on yield, water productivity… 49 160 200 240 280 DOY 2006 -2 -1.6 -1.2 -0.8 -0.4 stem (MPa) B 160 200 240 280 DOY 2007 -2 -1.6 -1.2 -0.8 -0.4 C Fig. 2.2. Seasonal patterns of midday stem water potential (Ψstem) for each irrigation treatment and experimental year. Each point represents the average of 20 readings. Vertical bars represent the standard deviation. DOY = day of year. 160 200 240 280 DOY 2005 -2 -1.6 -1.2 -0.8 -0.4 A Treatments SDI53 SDI67 SDI77 Control 56 Chapter2 the water status of the trees (Tables 2.1-2.2). Despite of this significant reduction in supplied water, no appreciably decrease in fruit yield was observed. In the SDI53 and SDI67 treatments there were no significant differences in yield, although it is noteworthy that we observed differences that were appreciable and that the changes were related to the water deficit undergone by the trees in each treatment. In these two treatments we found significant differences in the water status of the trees (Table 2.2) . The SDI treatments registered the highest SRI in 2005 (Fig. 2.3). Furthermore, the differences between treatments were clearer than in other years. Vélez et al. (2007) were unable to detect any significant differences in either the final production or the fruit weight or in the number of fruits per tree in mandarin “Clementine of Nules”, in response to a regulated deficit irrigation (RDI) strategy based on daily changes in trunk diameter. According to González-Altozano and Castel (1999), SΨ Fig. 2.5. Loadings of the studied variables (yield, fruit weight and water stress integral, S Ψ ) to the first three principal components (PC1, PC2 and PC3). Impact of sustained deficit irrigation on yield, water productivity… 57 values for Ψstem must exceed a threshold of -1.3 MPa to exert a significant effect on final crop production. This might partly explain why none of the three studied years showed any statistical significant difference in yield between irrigation treatments. In fact, Ψstem values of trees grown under the most severe SDI treatments showed Ψstem levels lower than the threshold value just occasionally. Under more severe water-stress conditions, Ginestar and Castel (1996) observed important differences in yield. They evaluated a treatment in which the available water was 50% of the crop water needs during the irrigation season. The effects in parameters such as fruit weight, juice content, maturity index, fruit diameter or peel thickness showed no statistically significant differences between treatments (Tables 2.3-2.5). Under our experimental conditions, the effects of the irrigation treatments were more evident in the organoleptic characteristics of the fruit, including TA and TSS (Tables 2.3-2.5). Similar results have been reported by other authors, such us Vélez et al. (2007), Ginestar and Castel (1996), González-Altozano and Castel (1999), Hutton et al. (2007), Treeby et al. (2007) and Pérez-Pérez et al. (2008). Furthermore, Hockema and Etxeberría (2001) showed that water stress leads to increased TSS and TA, but this is not a result of fruit dehydration but rather of the osmoregulatory response caused by the lack of water (Yakushiji et al., 1998). Regarding the relations between water stress and fruit weight, significant differences were detected only in 2005, when SΨ showed the highest values. In average, however, significant correlations were not found between SΨ and fruit weight. Several significant relations were found between weight and other parameters more affected by water stress (TSS or TA). Several authors have also reported relations between the water stress and fruit size. Treeby et al. (2007) observed that water stress boosted the number of fruits while reducing fruit size. Vélez et al. (2007) reported a decrease in the weight of the fruit and a slight increase in the number of fruits per tree, although none of these differences could be considered significant. Similar results were observed in the present work. Although we observe fruit weight reductions in SDI treatments, which were compensated with an increase in the number of fruits, the differences were not significant. 58 Chapter2 2.5. Conclusions Irrigation water savings of up to 55% of IN had no significant impact on tree yield, but rather affected other key factors (TSS and TA) that have direct relevance for the final quality of the harvested product. The seasonal pattern of SΨ in different irrigation treatments was consistent with the water deficit imposed in each treatment. Any water deficit considerably boosts SΨ, which results in a strong negative correlation between SRI and IAs, indicating that SΨ and SRI can be used as reliable stress indexes. Our results indicate that the main effects of water stress are reflected in organoleptic fruit parameters, with strong correlations between TSS and SΨ and between TA and SΨ. Water stress effect was less obvious in other morphological variables such as fruit weight, ED or PT, although these also registered significant correlations. Furthermore, the main result given by the PCA analysis was that yield, fruit weight, and SΨ explain a large part of the variability found in the total dataset. The higher increase in WP was detected in SDI53 treatment, with not significant decrease in yield within the three studied years. However, the low values of Ψstem detected in this treatment suggest an excessive stress that could result in reductions of yield in the future. A longer experiment is required to confirm the long-term effect of this water stress. On the other hand, the SDI67 and SDI77 treatments did not cause significant Ψstem reductions. We demonstrate in this work the potential of these treatments to increase WP and fruit quality (TSS and TA) with non significant reductions of yield. For this reason, and considering the problem of water scarcity in Andalusia, we strongly recommend these strategies for orange orchards (Citrus Sinensis L. Osbeck, cv. Salustiana) in our region. 2.6. References Ahuja L, Saseendran S, Reddy V, Yu Q. 2008. Synthesis, actions, and further research to improve response of crop system models to water stress. Response of crops to limited water: understanding and modeling water stress effects on plant growth processes 1:411. Impact of sustained deficit irrigation on yield, water productivity… 59 Allen RG, Pereira LS, Raes D, Smith M. 1998. Crop evapotranspiration-Guidelines for computing crop water requirements-FAO Irrigation and drainage paper 56. FAO, Rome 300:6541. Berbel J, Gutiérrez C. 2004. I Estudio de sostenibilidad del regadío del Guadalquivir. FERRAGUA. Spain. Castel J, Buj A. 1990. Response of Salustiana oranges to high frequency deficit irrigation. Irrigation Science 11(2):121-127. Davis JC. 2002. Statistics and Data Analysis in Geology (Third Edition). John Wiley & Sons. p 678. Doorembos J, Kassam A. 1979. Yield response to water. Rome: FAO 33. FAO. 1998. World Reference Base for Soil Resources. Rome: Food and Agriculture Organization of the United Nations. Fereres E, Castel JR. 1981. Drip Irrigation Management. Division of Agricultural Sciences. University of California. Leaflet Nº 21259. 2,11-24. Fereres E, Soriano MA. 2007. Deficit irrigation for reducing agricultural water use. Journal of Experimental Botany 58(2):147-159. García-Tejero I, Jiménez JA, Muriel JL, Martínez G. 2007. Planificación y desarrollo de estrategias de riego deficitario en una plantación de naranjos: influencia de la variabilidad espacial de las propiedades del suelo. Book of Abstracts XXV Congreso Nacional de Riegos. Pamplona. Spain. p 63. García-Tejero I, Jiménez JA, Reyes MC, Carmona A, Pérez R, Muriel JL. 2008 Aplicación de caudales limitados de agua en plantaciones de cítricos del valle del Guadalquivir. Fruticultura Profesional 173:5-17. Ginestar C, Castel JR. 1996. Responses of young clementine citrus trees to water stress during different phenological periods. Journal of Horticultural Science 71(4):551-559. 60 Chapter2 Gonzalez-Altozano P, Castel JR. 1999. Effects of regulated deficit irrigation on 'Clementina de Nules' citrus trees I. Yield and fruit quality effect. J Am Soc Hort Sci 74:749-758. Gonzalez-Altozano P, Castel JR. 2000. Effects of regulated deficit irrigation on 'Clementina de Nules' citrus trees II. Vegetative growth. J Am Soc Hort Sci 75:388-392. Hockema BR, Etxeberria E. 2001. Metabolic contributors to drought-enhanced accumulation of sugars and acids in oranges. Journal of the American Society for Horticultural Science 126(5):599-605. Hsiao TC. 1973. Plant responses to water stress. Annu Rev Plant Physiol Plant Molec Biol 24:519-570. Hutton R, Landsberg J, Sutton B. 2007. Timing irrigation to suit citrus phenology: a means of reducing water use without compromising fruit yield and quality? Australian Journal of Experimental Agriculture 47(1):71-80. Ihaka R, Gentleman R. 1996. R: a language for data analysis and graphics. Journal of computational and graphical statistics 5:299-314. Muriel J, Jiménez J, García I, Vaquero I. 2006. Relaciones hídricas en una plantación de naranjos (Citrus sinensis, L. cv. Navelino) bajo estrategias de riego deficitario mantenido. VIII Simposium Hispano Portugués de Relaciones Hídricas en las Plantas:17–21. Myers BJ. 1988. Water stress integral—a link between short-term stress and long-term growth. Tree Physiology 4(4):315. Peng Y, Rabe E. 1998. Effect of differing irrigation regimes on fruit quality, yield, fruit size and net CO2 assimilation of Mihowase Satsuma. Journal of horticultural science & biotechnology 73(2):229-234. Pérez-Pérez J, Romero P, Navarro J, Botía P. 2008. Response of sweet orange cv ‘Lane late’ to deficit-irrigation strategy in two rootstocks. II: Flowering, fruit growth, yield and fruit quality. Irrigation Science 26(6):519-529. Impact of sustained deficit irrigation on yield, water productivity… 61 Rawls W, Pachepsky YA. 2002. Soil consistence and structure as predictors of water retention. Soil Science Society of America journal 66(4):1115-1126. Rodríguez Díaz J, Weatherhead E, Knox J, Camacho E. 2007. Climate change impacts on irrigation water requirements in the Guadalquivir river basin in Spain. Regional Environmental Change 7(3):149-159. Sánchez-Blanco MJ, Torrecillas A, León A, Delamor F. 1989. The Effect of Different Irrigation Treatments on Yield and Quality of Verna Lemon. Plant and Soil 120(2):299-302. Scholander PF, Bradstreet ED, Hemmingsen E, Hammel H. 1965. Sap pressure in vascular plants. Science 148(3668):339. Shalhevet J, Bielorai H. 1978. Crop water requirements in relation to climate and soil. Soil Science 125(4):240-247. Southwick SM, Davenport TL. 1986. Characterization of water stress and low temperature effects on flower induction in citrus. Plant Physiology 81(1):26-29. Sterk G, Stein A. 1997. Mapping wind-blown mass transport by modeling variability in space and time. Soil Science Society of America Journal 61(1):232-239. Team RDC. 2008. R: A language and environment for statistical computing. R Foundation for Statistical Computing Vienna Austria ISBN 3(10). Treeby MT, Henriod RE, Bevington KB, Milne DJ, Storey R. 2007. Irrigation management and rootstock effects on navel orange Citrus sinensis (L.) Osbeck fruit quality. Agric Water Manage 91(1-3):24-32. Turner NC. 1988. Measurement of plant water status by the pressure chamber technique. Irrigation Science 9(4):289-308. Vaux H, Pruitt WO. 1983. Crop-water production functions. Advances in irrigation 2:61-67. Velez J, Intrigliolo D, Castel J. 2007. Scheduling deficit irrigation of citrus trees with maximum daily trunk shrinkage. Agric Water Manage 90(3):197-204. 62 Chapter2 Yakushiji H, Morinaga K, Nonami H. 1998. Sugar accumulation and partitioning in Satsuma mandarin tree tissues and fruit in response to drought stress. American Society for Horticultural Science (USA) 123:719-726. 63 Chapter 3 Impact of regulated deficit irrigation on yield, water productivity and fruit quality in a commercial citrus orchard, cv. Navelina Part of this chapter is published in: García-Tejero I, Romero R, Jiménez-Bocanegra JA, Martínez G, Durán-Zuazo VH, Muriel JL. 2010. Response of citrus trees to deficit irrigation during different phenological periods in relation to yield, fruit quality, and water productivity. Agricultural Water Management Vol 97, 689-699. Abstract. In 2007 and 2008, four strategies of regulated deficit irrigation (RDI) and a control treatment were implemented in 11-year-old citrus trees (Citrus sinensis L. Osb. cv. Navelina) grafted onto carrizo citrange (Citrus sinensis L. Osb. x Poncirus trifoliata L. Osb.). Irrigation in the control treatment was aimed at guarantying 100% of the irrigation needs (IN, mm). The four RDI treatments were defined based on irrigation amounts (IA, mm) calculated as a percentage of the control (70% for moderate water stress and 56% for severe water stress), varying in each of the following growth phases: flowering (from 50% of opened flowers to fruit setting), fruit growth and fruit maturation. Midday stem water potential (Ψstem, MPa) and stomatal conductance (gs, mmol·m-2·s-1) measurements were made in each of the considered treatments. Both the water stress integral (SΨ, MPa) and the stomatal conductance integral (Sg, mmol·m-2·s-1) were calculated for all treatments and used for quantifying the water-stress levels suffered by the trees. Reference equations were formulated to quantify the correlations between the supplied irrigation amounts (IA) or SΨ and Sg and yield and fruit-quality parameters. Significant differences in yield between the control and the RDI treatments were found in the second year of the experiment. In the first year the differences were not significant. The greater differences were found in treatments with severe water stress applied during both the flowering and fruit growth phases. When severe water stress was allowed during the fruit maturation phase, the effects were especially evident in fruit-quality parameters (total soluble solids content and titratable acidity). The best results were obtained with the RDI strategy characterized by moderate deficit irrigation (70% of control) during flowering and fruit growth and severe deficit irrigation (56% of control) during fruit maturation. This strategy saved 1030 m3 ha-1 with Impact of regulated deficit irrigation on yield, water productivity… 65 65 respect to the control, with not significant effect in yield. It also improved fruit quality parameters as TSS and TA. The RDI strategy with severe deficit irrigation at flowering and fruit growth, and moderate deficit irrigation during maturation allowed water savings of up to 1375 m3 ha-1 in 2008, but yield was reduced by 22%. Water productivity, however, increased 30% as compared to the control. 3.1. Introduction Water deficits conditions are common in the Guadalquivir River Valley (SW Spain). Therefore, irrigation is needed for commercial production of crops such as citrus trees. The increasing demand for food-related agricultural products, fodder, and fuel, due to population growth, makes compulsory agricultural practices focused to a more efficient use of water. In this context, there is an increasing challenge for scientists to develop innovative crop-management practices for a use of the soil, water and agrochemicals oriented to improve the sustainability of agricultural systems (Anapalli et al., 2008). Water productivity (WP, kg of marketable yield ·L-1 of water consumed by the crop) can be improved either by increasing yield or reducing crop water consumption (Fereres et al., 2003). Deficit irrigation (DI) has been widely investigated as a valuable and sustainable production strategy in dry regions. Any rational DI strategy saves water and increases WP. Different DI strategies cause different effects on the crop. In general photosynthetic rates are lowered, reducing carbon assimilation (Hsiao, 1973) and exerting a negative impact on the crop development and production (González-Altozano and Castel, 2000a). Any DI strategy must be designed to achieve the best compromise between the negative impact of the reduced water supplies and the advantages of water and energy saving. It is known that the response of citrus trees to DI depends on a variety of factors, such as the phenological stage, intensity and duration of the water stress period, crop physiological status, irrigation-water quality, plant genotype, and the degree of stress endured by the crop (Doorenbos and Kassam, 1979; Ginestar and Castel, 1996; García-Tejero et al., 2008). It has also been described that DI may increase fruit quality (Sánchez-Blanco et al., 1989, in Verna lemon trees; González- Altozano and Castel, 2000b, in cv. Clementina de Nules; Verreynne et al., 2001, in cv. Marisol Clementines). 72 Chapter 3 Fig. 3.1. Effective precipitation (Pe), crop (ETc) and potential (ETo) evapotranspiration during the experimental years. DOY = day of year. 0 5 10 15 20 25 30 35 40 45 0 2 4 6 8 10 77 127 177 227 277 P e(mm) ET (mm) DOY 2007 (77= March 18) ETo ETc 0 5 10 15 20 25 30 35 40 45 0 2 4 6 8 10 77 108 138 169 199 230 261 291 Pe(mm) ET (mm) DOY 2008 (91= March 31) ETo ETc Impact of regulated deficit irrigation on yield, water productivity… 73 73 Fig. 3.2. Irrigation amounts (IA) in mm and as percentage of the irrigation needs (IN) during the irrigation periods on 2007 and 2008. 74 Chapter 3 The total IA in each treatment was similar in both years (Table 3.3). RDI-667 was the most restrictive treatment (55% of IN on average). The total IA in the RDI-676, RDI-677, and RDI-776 treatments was similar (64±1% of IN on average), although the water distribution according to the phenological periods differed among treatments. For the control treatment, the average IA for the two experimental years amounted to 95±3% of IN. For the RDI treatments, TDR measurements showed water depletions in the soil profile according to the reduced water supplies (Fig. 3.4). The control treatment maintained moisture contents close to 80% of field capacity on average, with slight decreases in 2008. As shown in Fig. 3.3, the IAs applied during the flowering and fruit growth periods were slightly below the actual INs of the crop. Data in the figure shows little differences in soil water contents (SWC) between the control and the RDI treatments. This is striking, since it does not agree with differences in supplied water (Table 3.3). Possibly the soil variability, together with other limitations of soil moisture measurements, reduced the reliability of our SWC measurements. Taking this into account, data in Fig. 3.4 show that SWC values in the RDI-667 treatment were especially low during the fruit growth period, being between 15 and 30% lower than those recorded in the control treatment. The highest SWC were measured in the control treatment, both in 2007 and 2008 seasons. However, due to the high variability of the data, the differences between RDI-776 and control treatments were not significant in 2007. We did not find significant differences either in SWC among the RDI-677, RDI- 776, and control treatments during 2008. 3.3.2. Plant water status and gas exchange In both seasons, Ψstem was very similar in all the treatments at the end of the flowering period (Fig. 3.5). This was expected because we were, at that time, at the end of the rainy season. This explains that the differences in SWC between treatments were not enough to cause significant differences in the water status. Both at fruit growth and fruit maturation, RDI trees showed more negative values of Ψstem than the control trees. However the differences were only significant between control and RDI-667 during the fruit growth period. In the RDI-776 treatment, in which moderate stress was applied during flowering and fruit growth (IA = 70% of control), Ψstem decreased significantly at Impact of regulated deficit irrigation on yield, water productivity… 75 75 the end of the fruit maturation period, when severe stress was applied (IA= 56% of control). Concerning gas exchange, the most restrictive treatment, RDI-667, showed the lowest gs values and the control treatment the highest ones (Fig. 3.6). Differences between treatments were not significant, likely because of the high variability of gs. Box and whisker plots of Sg and SΨ for each treatment and season are shown in Fig. 3.7. It was notable that treatments with similar water-stress levels during the fruit growth period (RDI-676, RDI-677 and RDI-776) showed similar values both of Sg and SΨ in 2007 and 2008. The differences in water-stress levels in these treatments occurred during flowering and fruit maturation. Rainfall in these periods might explain why these differences were not enough to affect Sg and SΨ. Significant linear correlations were found between IA and Sg and SΨ (Fig. 3.8) in 2007 and 2008. The coefficients of determination were particularly high between IA and Sg in 2007 (r2 = 0.751) and between IAs and SΨ in 2008 (r2 = 0.837). 3.3.3. Yield, fruit quality and water productivity We found clear relations between yield (average per treatment, normalized for each year) and the annual IA (Fig. 3.9). On the other hand, these differences were not clear in the box and whisker plots, especially in 2007 (Fig. 3.10). This was mainly due to a high variability of the yield data of each treatment. However, in 2008, significant differences were found between the control treatment and the RDI-676 and RDI-677 treatments. The biggest differences were for RDI-667, with an average yield reduction of 15% in 2007 and 26% in 2008, as compared to the control treatment (Table 3.4). In the rest of the RDI treatments yield reductions were also observed, but the differences were not always statistically significant. In 2008, when the differences were greater, severe water stress applied during the flowering period was reflected in the final fruit number per tree (Table 3.4). This is clear for treatment RDI-677, in which the fruits per tree were significant lower than in the control treatment. 76 Chapter 3 Table 3.3. Irrigation water applied, actual water stress ratio and water savings for each treatment at different phenophases. Season 2007 2008 Phenophase Phenophase Treat. Flowering Growth Maturity Total Flowering Growth Maturity Total RDI-676 IA (mm)1 42.3 144.7 20.3 207.3 18.7 178.1 19.2 216 IA (% IA control) 57.95 69.50 55.77 65.27 60.91 69.35 55.65 67.08 IA (% IN)2 58.14 57.63 91.77 59.92 44.55 61.16 - 66.20 WS (mm)3 30.7 63.5 16.1 110.3 12 78.7 15.3 106 RDI-677 IA (mm) 43.4 144.8 25.3 213.5 18.7 178.7 24 221.4 IA (% IA control) 59.45 69.55 69.51 67.22 60.91 69.59 69.57 68.76 IA (% IN) 59.65 57.67 114.38 61.71 44.55 61.36 - 67.85 WS (mm) 29.6 63.4 11.1 104.1 12 78.1 10.5 100.6 RDI-667 IA (mm) 41.7 115.9 25.3 182.9 17.1 143.5 24 184.6 IA (% IA control) 57.12 55.67 69.51 57.59 55.70 55.88 69.57 57.33 IA (% IN) 57.31 46.16 114.38 52.87 40.73 49.28 - 56.58 WS (mm) 31.3 92.3 11.1 134.7 13.6 113.3 10.5 137.4 RDI-776 IA (mm) 49.7 144.7 20.3 214.7 21.3 178.1 19.2 218.6 IA (% IA control) 68.08 69.50 55.77 67.60 69.38 69.35 55.65 67.89 IA (% IN) 68.31 57.63 91.77 62.06 50.74 61.16 - 67.00 WS (mm) 23.3 63.5 16.1 102.9 9.4 78.7 15.3 103.4 control IA (mm) 73 208.2 36.4 317.6 30.7 256.8 34.5 322 IA (% IA control) 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 IA (% IN) 100.33 82.93 164.56 91.81 73.13 88.18 - 98.69 WS (mm) 0 0 0 0 0 0 0 0 1IA, Irrigation amount; 2IN, Irrigation needs; 3WS, Water Savings (related to control treatment); *Average of water supplied to each treatment referred to control treatment during the irrigation period. Impact of regulated deficit irrigation on yield, water productivity… 77 77 Fig. 3.3. Irrigation amounts (IA) for each treatment during the experimental years. DOY = day of year. 78 Chapter 3 On the other hand, when these water restrictions were applied during the fruit growth or fruit maturation periods, i.e. in the RDI-676, RDI-667 and RDI-776 treatments, fruit weight rather than fruit number per tree was the affected variable. Particularly, RDI-667 presented significant differences in fruit weight with the control treatment. As mentioned above, this was the treatment in which yield was most affected. RDI-776 was the treatment with the highest number of fruits in both experimental years, although these differences were not significant. Table 3.4 also shows the effects on fruit morphological parameters (ED, PD, and fruit weight). Not significant relations were found between these parameters and the IA applied in the treatments. Regarding the effect of water stress on organoleptic properties, fruit quality was affected mainly in treatments with higher stressed levels during the fruit growth and fruit maturation periods (RDI-676, RDI-667, and RDI-776). In these treatments we registered increases both in TSS and TA, coupled with small decreases in MI. These effects were particularly significant in the RDI-667 treatment, in which the most severe reductions in IA were applied. Results from the RDI-776 treatment shows that a severe reduction of IA during fruit maturation improved juice quality without excessively lowering MI. Fig. 3.4. Soil water content in the root zone for each treatment at the end of the fruit growth period of 2007 and 2008. The line inside the box shows the median and the letters after each box-whiskler plot show statistical differences at P < 0.05 level. Impact of regulated deficit irrigation on yield, water productivity… 79 79 For the entire study period (2007-2008), our results show significant differences between some of the treatments for SΨ, Sg, yield, fruit weight, TSS, TA, and ED (Table 3.5). This overall analysis confirmed the results in the year-by-year analysis, especially for the second experimental year, when the rise in TSS and TA values registered in the RDI-667 and RDI-776 treatments was especially noticeable. Concerning yield parameters, only the most water limited treatment (RDI-667) showed significant differences in yield, mainly due to a significant decrease in fruit weight. The rest of the treatments did not show significant differences in yield, fruit weight or fruit number when considering the whole experimental period (2007-2008). In 2007 values of WP in the tested deficit irrigation treatments were similar, but significantly higher than in the control treatment (Fig. 3.11). In 2008 WP was reduced in all the RDI treatments, but they again were significantly higher than in the control treatment. 3.3.4. Plant water status and gas exchange versus yield and fruit quality We analysed the relationships between both yield and fruit-quality parameters and Sg and SΨ, to define which of those parameter was most related to the crop response to DI. The SΨ values showed strong linear correlation with TSS only (Fig. 3.12). Other parameters, such as PD, MI, and juice content showed no significant correlations when considering the whole dataset, but were statistically significant when considering each year separately. Therefore, in 2007 year, the juice content (r2 = 0.71) and PD (r2 = 0.80) had a significant correlation with SΨ (regressions not shown). MI also showed significant correlations with SΨ, with r2 values of 0.76 in 2007 and 0.52 in 2008. Values of Sg were closely correlated with some yield and fruit parameters (Fig. 3.13), especially with fruit weight (r2 = 0.78), TA (r2 = 0.92), PD (r2 = 0.79) and MI (r2 = 0.76). 80 Chapter 3 Fruit Growth Fruit Growth Fig. 3.5. Evolution of midday stem water potential (Ψ stem ) in each treatment during the fruit growth and fruit maturation periods of the experimental years. Vertical bars indicate the standard deviation in each treatment (ten repetitions per treatment). DOY = day of year. Impact of regulated deficit irrigation on yield, water productivity… 81 81 Fig. 3.6. Evolution of stomatal conductance (g s ) in each treatment during the fruit growth and fruit maturation periods of the experimental years. Vertical bars indicate the standard deviation in each treatment (ten repetitions per treatment). DOY = day of year. 88 Chapter 3 SΨ (MPa) Fig. 3.12. Relationships among water stress integral (S Ψ , MPa) and yield and fruit quality parameters (years 2007 and 2008). PT, peel thickness; TSS, total soluble solids; MI maturity index; TA, titratable acidity; ED and PD equatorial and polar diameter respectively. Impact of regulated deficit irrigation on yield, water productivity… 89 SΨ (MPa) Fig. 3.13. Relationships among stomatal conductance integral (Sg, mmol s-1 m-2) and yield and fruit quality parameters (years 2007 and 2008). PT, peel thickness; TSS, total soluble solids; MI maturity index; TA, titratable acidity; ED and PD equatorial and polar diameter Sg (MPa) 90 Chapter 3 3.4. Discussion In terms of the threshold value for midday Ψstem in citrus trees, the control treatment registered values ranging between -0.6 and -1.3 MPa, in periods of minimum and maximum evapotranspiration demand, respectively. These values are close to those reported by Ortuño et al. (2006a). According to De Swaef et al. (2009), Ψstem directly reflects the plant’s water status, bearing strong relationships with the sap-flow rate or daily radial-stem growth. Many authors have found similar results for Ψstem (Goldhamer et al., 1999; Naor and Cohen, 2003; Nortes et al., 2005), a parameter which usefulness for irrigation management is widely accepted (Shackel et al., 1997; Naor, 2000). In general, gs is not sensitive to the irrigation treatment until a certain threshold of Ψstem is reached. In this context, Ortuño et al. (2004), observed in well-irrigated lemon trees greater fluctuations in gs than in ΨStem. In other experiment, Ortuño et al. (2006b) observed that Ψstem was more sensitive to water stress than gs, since significant differences between treatments were found in Ψstem a week before than in gs. Our results suggest that both SΨ and Sg are good indicators of potential effects caused by water stress in some of the fruit quality parameters such as TA and TSS. Nevertheless, Sg showed better results when considering the relations obtained each year than that of the two years experimental period. In contrast, treatments with similar SΨ or Sg showed different effects on yield and fruit-quality parameters. This could have been due both to changing meteorological conditions and to the stress distribution over the phenological periods. Differences on water distribution along the irrigation season, depending on the irrigation strategy, had a greater effect on the response of the citrus tree than the annual IA applied in each treatment. Several pieces of work have shown that certain levels of water stress increases TSS and TA in citrus trees (Bielorai, 1982; Kuriyama et al., 1981; Yakushiji et al., 1998; Hockema and Etxeberría, 2001). In this context Pérez-Pérez et al. (2009) pointed out that a reduction of IA during fruit maturation period in ‘Lane late’ sweet orange significantly increased TSS and TA, without changes in MI, PD, and peel thickness. Moreover, this irrigation reduction lowered juice parameters and ED. Ginestar and Castel (1996) observed that withholding of irrigation during the fruit maturation of young clementine citrus trees in two consecutive years caused a reduction of yield, fruit weight and fruit Impact of regulated deficit irrigation on yield, water productivity… 91 number, although fruit number differences were not significant. Gonzalez-Altozano and Castel (1999) documented similar results in two treatments with reductions of 50% and 75% of irrigation during fruit maturation of 10-year-old clementine citrus trees in the first experimental year. In the second year, these authors observed reductions in yield and fruit weight and increments in fruit number. Nevertheless, only fruit weight differences were significant. Hutton et al. (2007) tested DI by applying different irrigation interval treatments during fruit growth and fruit maturation periods in ‘Valencia’ orange trees during two consecutive years. In their experiment, the development of more vegetative shoot growth by trees growing under increasing water deficit in late summer increased the number of potential fruiting sites for flowering in the following season. This was seen in the increase in fruit count per tree recorded for trees grown under longer irrigation intervals. This also could explain the results of Gonzalez-Altozano and Castel (1999) reported above. Hutton et al. (2007) also observed that the increased crop load (number of fruits per tree) and periodic water stress during fruit growth period contributed to the smaller fruit size recorded in trees irrigated at the longer intervals. Consequently, the increased fruit numbers did not result in significant increases in fruit mass per tree. They observed yield differences between years were much greater tan between irrigation treatments. The higher number of fruits in our RDI-776 treatment could be explained because the water limitations during fruit maturation. Gonzalez-Altozano and Castel (2003) also documented that prolonged moderated deficit irrigation in autumn promoted higher flowering during next spring. González-Altozano and Castel (2003) observed significant effects on yield, because of a decrease in fruit number when the crop underwent moderate to severe water stress. However, this strategy did not affect either the fruit weight or the organoleptic properties. When a severe reduction in water supplied by irrigation was applied during the fruit-growing period, the fruit number was not affected but fruit weight, and thus yield, was reduced, although only fruit weight differences were significant. Our results in 2008 suggest that any severe water stress (IA < 57% of IN) at flowering affected mainly the fruit number, thus reducing yield. Nevertheless, the differences were only significant in treatment RDI-667. Furthermore, when a similar 92 Chapter 3 reduction in IA was maintained during fruit growth the fruit weight significantly decreased. Deficit irrigation during fruit maturation affected the organoleptic fruit characteristics. DI strategies were influenced by the spring and autumn rainfall, suggesting that its application could be appropriate from mid-spring to the end of summer, coinciding with late flowering and the final fruit growth period, respectively. Increasing WP may be a mean of achieving a rational use of water in agriculture. Taking into account that the available water for irrigated land is a limiting factor in many world areas (Ali and Talukder, 2008), strategies such as DI have shown that water productivity can be enhanced (Ali et al., 2007; Jalota et al., 2006) and could be associated with acceptable commercial production. Nowadays, the low importance given to improving WP must be related to the reduced water costs in Mediterranean agricultural areas (Lorite et al., 2004), where water represents only less than 10% of the total production costs, a clear contradiction with the Common Agricultural Policy and the water Framework Directive (García-Vila et al., 2008). Our WP results were particularly good for the RDI-776 treatment during 2008, when the highest water productivity was achieved (0.025 kg L-1). Thus, this DI strategy offered the best results, with water savings greater than 30% (Table 3.3) and non-significant yield losses (Table 3.4). The total yearly IAs in the RDI-676, RDI-677 and RDI-776 treatments were similar (64%±1% of IN), but the different periods in which IA was more severely reduced had significant effects on yield, and hence in final water productivity (Tables 3.3-3.4 and Fig. 3.11). 3.5. Conclusions Although all the tested RDI treatments increased WP, our results show that RDI- 776 is the best strategy for Citrus sinensis L. Osb. cv. Navelina trees grown in the area. This treatment allowed 1200 m3 Ha-1 (37% IN) of water savings per year on average with no significant effect in yield. It also improved significantly fruit quality parameters as TSS and TA. In the rest of the tested RDI treatments, with a greater reduction of irrigation water at flowering (44% of control), mainly fruit number was affected. Furthermore, maintaining this reduction during the fruit growth period caused a significant loss in fruit weight and some changes in fruit quality parameters, such as an increase of TSS and TA. Impact of regulated deficit irrigation on yield, water productivity… 93 Our results suggest that both SΨ and Sg are good indicators of potential effects caused by water stress in some fruit quality parameters as TA and TSS, although Sg showed better results when considering the relations obtained each year than that of the two years experimental period. In contrast, treatments with similar SΨ or Sg showed different effects on yield and fruit-quality parameters, probably due to changing meteorological conditions and to the stress distribution over the phenological periods. Differences on water distribution along the irrigation season, depending on the irrigation strategy, had a greater effect on the response of the citrus tree than the annual IA applied in each treatment. 3.6. References Ali M, Hoque M, Hassan A, Khair A. 2007. Effects of deficit irrigation on yield, water productivity, and economic returns of wheat. Agric Water Manage 92(3):151-161. Ali M, Talukder M. 2008. Increasing water productivity in crop production--A synthesis. Agric Water Manage 95(11):1201-1213. Allen RG, Pereira LS, Raes D, Smith M. 1998. Crop evapotranspiration-Guidelines for computing crop water requirements-FAO Irrigation and drainage paper 56. FAO, Rome 300:6541. Anapalli SS, Ahuja, L.R., Ma, L., Timlin, D.J., Stockle, C.O., Boote, K.J., Hoogenboom, G. 2008. Current Water Deficit Stress Simulations in Selected Agricultural System Simulation Models. In: Ahuja LR, Reddy, V.R., Saseendran, S.A., Yu, Q., editor. Response of Crops to Limited Water: Understanding and Modeling Water Stress Effects on Plant Growth Processes. Madison, WI: American Society of Agronomy, Crop Science Society of America, Soil Science Society of America. p 1-38. Bielorai H. 1982. The effect of partial wetting of the root zone on yield and water use efficiency in a drip-and sprinkler-irrigated mature grapefruit grove. Irrigation Science 3(2):89-100. 94 Chapter 3 BOJA. 2000. Orden de 21 de septiembre de 2000 por la que se aprueba el Reglamento Específico de Producción Integrada de Cítricos. Sevilla: Consejería de Agricultura y Pesca. 30 de Septiembre de 2000, 15287-15297. Castel J. 1991. El riego de los cítricos. Hortofruticultura 5:41-52. De Swaef T, Steppe K, Lemeur R. 2009. Determining reference values for stem water potential and maximum daily trunk shrinkage in young apple trees based on plant responses to water deficit. Agric Water Manage 96(4):541-550. Doorembos J, Kassam A. 1974. Crop water requirements. FAO Irrigation and drainage paper 24. Doorembos J, Kassam A. 1979. Yield response to water. Rome: FAO 33. FAO. 1998. World Reference Base for Soil Resources. Rome: Food and Agriculture Organization of the United Nations. Fereres E, Goldhamer DA, Parsons LR. 2003. Irrigation water management of horticultural crops. Hortscience 38(5):1036-1042. García-Tejero I, Jiménez JA, Reyes MC, Carmona A, Pérez R, Muriel JL. 2008 Aplicación de caudales limitados de agua en plantaciones de cítricos del valle del Guadalquivir. Fruticultura Profesional 173:5-17. García-Vila M, Lorite I, Soriano M, Fereres E. 2008. Management trends and responses to water scarcity in an irrigation scheme of Southern Spain. Agric Water Manage 95(4):458-468. Ginestar C, Castel JR. 1996. Responses of young clementine citrus trees to water stress during different phenological periods. Journal of Horticultural Science 71(4):551- 559. Goldhamer DA, Fereres E, Mata M, Girona J, Cohen M. 1999. Sensitivity of continuous and discrete plant and soil water status monitoring in peach trees subjected to deficit irrigation. J Am Soc Hortic Sci 124:437-444. Impact of regulated deficit irrigation on yield, water productivity… 95 Gonzalez-Altozano P, Castel JR. 1999. Effects of regulated deficit irrigation on 'Clementina de Nules' citrus trees I. Yield and fruit quality effect. J Am Soc Hort Sci 74:749-758. Gonzalez-Altozano P, Castel JR. 2000a. Effects of regulated deficit irrigation on 'Clementina de Nules' citrus trees II. Vegetative growth. J Am Soc Hort Sci 75:388-392. González-Altozano P, Castel JR. 2000b. Effects of regulated deficit irrigation in ‘Clementina de Nules’ citrus trees growth, yield and fruit quality. Acta Hort 537:749-758. González-Altozano P, Castel JR. 2003. Riego deficitario controlado en" Clementina de Nules". I: efectos sobre la producción y la calidad de la fruta. Spanish Journal of Agricultural Research(2):81-92. Hockema BR, Etxeberria E. 2001. Metabolic contributors to drought-enhanced accumulation of sugars and acids in oranges. Journal of the American Society for Horticultural Science 126(5):599-605. Hsiao TC. 1973. Plant responses to water stress. Annu Rev Plant Physiol Plant Molec Biol 24:519-570. Hutton R, Landsberg J, Sutton B. 2007. Timing irrigation to suit citrus phenology: a means of reducing water use without compromising fruit yield and quality? Australian Journal of Experimental Agriculture 47(1):71-80. Jalota S, Sood A, Chahal G, Choudhury B. 2006. Crop water productivity of cotton (Gossypium hirsutum L.)-wheat (Triticum aestivum L.) system as influenced by deficit irrigation, soil texture and precipitation. Agric Water Manage 84(1-2):137- 146. Kuriyama T, Shimoosako M, Yoshida M, Shiraishi S. The effect of soil moisture on the fruit quality of Satsuma mandarin (Citrus unshiu Marc); 1981. International Society of Citriculture. p 524-527. 96 Chapter 3 Lorite I, Mateos L, Fereres E. 2004. Evaluating irrigation performance in a Mediterranean environment. Irrigation Science 23(2):77-84. Myers BJ. 1988. Water stress integral—a link between short-term stress and long-term growth. Tree Physiology 4(4):315. Naor A. 2000. Midday stem water potential as a plant water stress indicator for irrigation scheduling in fruit trees. Acta Hort 537:447-454. Naor A, Cohen S. 2003. Sensitivity and Variability of Maximum Trunk Shrinkage, Midday Stem Water: Potential, and Transpiration Rate in Response to Withholding Irrigation from Field-grown Apple Trees. Hortscience 38(4):547-551. Nortes PA, Pérez-Pastor A, Egea G, Conejero W, Domingo R. 2005. Comparison of changes in stem diameter and water potential values for detecting water stress in young almond trees. Agric Water Manage 77(1-3):296-307. Ortuño M, Alarcón J, Nicolás E, Torrecillas A. 2004. Comparison of continuously recorded plant-based water stress indicators for young lemon trees. Plant and Soil 267(1):263-270. Ortuño MF, García-Orellana Y, Conejero W, Ruiz-Sánchez MC, Alarcón JJ, Torrecillas A. 2006a. Stem and leaf water potentials, gas exchange, sap flow, and trunk diameter fluctuations for detecting water stress in lemon trees. Trees-Structure and Function 20(1):1-8. Ortuño MF, García-Orellana Y, Conejero W, Ruiz-Sánchez MC, Mounzer O, Alarcón JJ, Torrecillas A. 2006b. Relationships between climatic variables and sap flow, stem water potential and maximum daily trunk shrinkage in lemon trees. Plant and Soil 279(1-2):229-242. Pérez-Pérez J, Robles J, Botía P. 2009. Influence of deficit irrigation in phase III of fruit growth on fruit quality in ’lane late’ sweet orange. Agric Water Manage 96(6):969-974. Impact of regulated deficit irrigation on yield, water productivity… 97 Sanchez-Blanco MJ, Torrecillas A, Leon A, Delamor F. 1989. The Effect of Different Irrigation Treatments on Yield and Quality of Verna Lemon. Plant and Soil 120(2):299-302. Scholander PF, Bradstreet ED, Hemmingsen E, Hammel H. 1965. Sap pressure in vascular plants. Science 148(3668):339. Shackel K, Ahmadi H, Biasi W, Buchner R, Goldhamer D, Gurusinghe S, Hasey J, Kester D, Krueger B, Lampinen B and others. 1997. Plant Water Status as an Index of Irrigation Need in Deciduous Fruit Trees. HortTechnology 7(1):23-29. Steppe K, Dzikiti S, Lemeur R, Milford JR. 2006. Stomatal oscillations in orange trees under natural climatic conditions. Annals of botany 97(5):831. Sterk G, Stein A. 1997. Mapping wind-blown mass transport by modeling variability in space and time. Soil Science Society of America Journal 61(1):232-239. Thompson R, Gallardo M, Valdez L, Fernandez M. 2007. Determination of lower limits for irrigation management using in situ assessments of apparent crop water uptake made with volumetric soil water content sensors. Agric Water Manage 92(1-2):13- 28. Turner NC. 1988. Measurement of plant water status by the pressure chamber technique. Irrigation Science 9(4):289-308. Verasan VP, Ronald E. 1978. Effects of Soil Water Stress on Growth and Nutrient Accumulation in Corn. Agronomy Journal 70(4):613-618. Verreynne JS, Rabe E, Theron KI. 2001. The effect of combined deficit irrigation and summer trunk girdling on the internal fruit quality of 'Marisol' Clementines. Scientia Horticulturae 91(1-2):25-37. Yakushiji H, Morinaga K, Nonami H. 1998. Sugar accumulation and partitioning in Satsuma mandarin tree tissues and fruit in response to drought stress. American Society for Horticultural Science (USA) 123:719-726. 104 Chapter 4 LTD, UK). We used two access tubes per treatment, located in the soil of one sample tree per plot, at 0.01 m from the closer dripper and 1 m. apart from the trunk. Measurements in each tube were made at 10, 30, 60 and 90 cm depth, every 15 min. The total fruit weight of each sample tree was determined at the end of the season. For each treatment, WP was calculated as the yield (kg) divided by the volume of water applied (L). Fruit-quality characteristics were analysed after harvest. Ten fruits from each sample tree were used for analysis. For each fruit we measured equatorial (ED, mm) and polar (PD, mm) diameters, peel thickness (PT, mm), fruit weight and juice content. Total soluble solids contents (TSS, ºBrix) were measured with a digital refractometer PR-101. Titratable acidity (TA, g L-1) was determined by titrating the samples with NaOH 0.1 N by the colorimetric method, using phenolphthalein as indicator solution. The maturity index (MI, ºBrix L g-1) was then calculated by dividing TSS by TA. For evaluating the impact of the considered DI strategies, yield and fruit quality variables were subjected to analysis with one-way variance by Tukey’s test at level of P < 0.05. 4.3. Results 4.3.1. Water supplies During the studied period (DOY 152-286 in 2008), the seasonal values of ETo, ETc and Pe were 776, 299 and 47 mm, respectively. Rainfall was very low during the period of greatest atmospheric demand (from mid-June to mid-September, Fig. 4.1A). Most rainfall events were recorded from mid-September to mid-October. The irrigation amounts (IA, mm) amounted to 280 mm in control (111% IN), 145 mm in SDI (58% IN) and 165 mm in LFDI (65% IN) (Fig. 4.1B, Table 4.1). The SWC dynamics (Fig. 4.2) was consistent with the applied IA in each treatment. In the control treatment the SWC at 30 cm depth was nearly constant during the studied period (Fig. 4.2A). In the SDI treatment SWC values decreased for most of the dry period, until the first rainfall events in late summer (DOY 260). Afterwards, SWC Impact of regulated and low deficit irrigation on the physiological response… 105 Fig. 4.1. Effective precipitation (Pe), crop (ETc) and potential (ETo) evapotranspiration (A), and irrigation amounts (IA) for each treatment (B) during the irrigation season. DOY = day of year. Table 4.1. Total irrigation amounts (IA) and water savings (WS) as compared to the control for each treatment. The experimental period was from day of year 152 to day of year 286. See text for details on the treatments. Treatments Control SDI LFDI IA (mm) 280 145 165 IA (%IN) 111 58 65 WS (mm) 0 135 115 IN = irrigation needs 0 2 4 6 8 10 12 0 1 2 3 4 5 6 7 8 9 10 152 172 192 212 232 252 272 Pe (mm) ET (mm) DOY 2008 (152 = May 31) ETo ETc A 0 1 2 3 4 5 6 7 8 9 10 152 172 192 212 232 252 272 IA (mm) DOY 2008 (152 = May 31) control SDI LFDI control: 2800 m3 Ha-1 SDI: 1450 " LFDI: 1650 " B 106 Chapter 4 Fig. 4.2. Dynamics of soil water content (SWC) in each treatment at 30 and 60 cm depth. SDI: Sustainable deficit irrigation; LFDI: Low frequency deficit irrigation. DOY = day of year. Impact of regulated and low deficit irrigation on the physiological response… 107 increased up to values close to those measured in the control treatment. For the LFDI treatment the SWC values at 30 cm depth fluctuated according to the irrigation events. Thus, the SWC values decreased drastically in the first 48 h after withholding irrigation. At 60 cm depth (Fig. 4.2B), SWC in control and SDI treatments was nearly constant for most of the dry season. Fluctuations also occurred in the LFDI treatment at 60 cm depth, although less markedly than at 30 cm depth. The decreases in SWC at 60 cm were delayed 2-3 days respect to the decreases at 30 cm. 4.3.2. Plant-based measurements The time courses of the studied physiological variables ( Ψ stem, gs and MDS, Fig. 4.3) were consistent with the IAs applied in each treatment (Fig. 4.1). The gs values showed a large temporal variability (Fig. 4.3B), likely due to the variability of meteorological conditions. In the control treatment we recorded values of Ψ stem between -0.7 and -1.3 MPa throughout the studied period (Fig. 4.3A). For the SDI treatment Ψ stem followed a decreasing trend (from -1.2 MPa to -2.0 MPa during the dry period), according to the decreasing available water in the soil of this treatment (Fig. 4.2). The first rainfall events after the dry season, from DOY 262 (Fig. 4.1A), increased SWC in all plots (Fig. 4.2) and the differences between treatments on Ψ stem disappeared (Fig. 4.3). In the LFDI treatment Ψ stem values markedly depended on the withholding&rewatering cycles. For each cycle, Ψ stem values were close to those of the control treatment after the recovery irrigations, showing that these were enough for the LFDI trees to recover from water stress. For the control trees, the average value of gs for the dry period (DOY 185-246) was 121 mmol m-2 s-1 (Fig. 4.3B). The low values of gs in all the treatments on DOY 258 were probably due to the particularly low solar radiation conditions in this measurement day (data non shown). Values of gs in the SDI treatment during the dry season were significantly lower than in the control treatment. In this treatment, gs showed a decreasing trend, from 107 to 34 mmol m-2 s-1. This agrees with the cumulative stress during that period (Fig. 4.3A). In the LFDI treatment gs fluctuated, according to the 108 Chapter 4 Fig. 4.3. Dynamics for stem water potential (A), stomatal conductance (B) and maximum daily shrinkage (C) in each treatment during the irrigation period. DOY = day of year. -3 -2.5 -2 -1.5 -1 -0.5 0 185 205 225 245 265 285 Ψstem (MPa) DOY (185 = July 4) control SDI LFDI Dry period A 0 20 40 60 80 100 120 140 160 180 200 185 205 225 245 265 285 gs (mmol m-2 s-1) B 0 50 100 150 200 250 300 350 400 185 205 225 245 265 285 MDS (mm) DOY (185 = July 7) C Impact of regulated and low deficit irrigation on the physiological response… 109 withholding&rewatering cycles, from 166 to 21 mmol m-2 s-1. As for Ψ stem, differences on gs between treatments disappeared after the autumn rainfalls. Values of MDS in the control treatment varied from 100 to 200 μm (Fig. 4.3C) for most of the dry season. The greatest values corresponded to the non-irrigation days. The MDS values in the control decreased, as the ETo, from early August. In the SDI trees the MDS trend mimicked that of ETo (Fig. 4.1), and values were usually greater than in the control trees, likely because of the most severe waters stress levels reach by the SDI trees. In the LFDI trees the MDS values increased with soil drying between two recovery irrigation periods, and decreased when the trees were irrigated. After the dry season, i.e. after the occurrence of the rainfall events from DOY 262, differences on MDS between treatments disappeared. The correlation coefficients (r) of the relationships between Ψ stem, gs, MDS and SWC at 30 cm (SWC30) and 60 cm (SWC60) depth are shown in Table 4.2.The results obtained between MDS and the other parameters were especially significant (Table 4.2). Highest correlation coefficients were obtained between the Ψ stem and the SWC at 30 cm depth (r = -0.53), Ψ stem and gs (r = 0.75), and MDS and Ψ stem (r = -0.63). This suggests that these parameters (i.e. Ψ stem, gs, SWC30 and MDS) offer certain information about the crop water status when an irrigation deficit is applied. Nevertheless, the moderate values of the correlation coefficients suggest that no direct relationships should be assumed between these variables. These results show that the relationships between the considered variables should be explained with more complex multiparametric models. We also measured the maximum stem diameter (MXSD, mm), which showed different seasonal patterns, in each treatment (Fig. 4.4). The dry period coincided with the fruit growth stage. This explains the limited trunk growth showed in Fig. 4.4 for the control trees from DOY 179 to DOY 242. Only at the end of this period, coinciding with the fruit maturity phase, there was a slight increase in trunk diameter. After harvesting, from DOY 268, net trunk growth was negligible. In relation to the SDI treatment, we observed a significant decrease of the MXSD values due to the accumulated water stress of these trees. The MXSD trend in the SDI trees changed with the first rainfall events at 110 Chapter 4 the end of the dry period. This increment confirms the recovery of these trees suggested by the Ψstem and gs observations (Fig. 4.3). Finally, the MXSD values registered in LFDI trees showed decline and recovery periods according to the irrigation events in this treatment. At the end of the season the LFDI trees showed a net growth similar to that in control trees. Likely, the depletion of SWC together with high atmospheric demand led to negligible growth and progressive dehydration of tissues at the end of the withholding periods, which causes the marked decreases of MXSD recorded in the LFDI trees at the end of the periods in which irrigation was withheld. The comparison of Figs. 4.2, 4.3 and Fig. 4.4. Temporal evolution of the daily maximum stem diameter (MXSD) in each treatment duri ng the studied period. SDI: Sustainable deficit irrigation; LFDI: Low frequency deficit irrigation . DOY = day of year. Table 4.2. Correlation coefficients among between the studied parameters. MDS SWC30 SWC60 Ψ stem MDS SWC30 -0.41** SWC60 NS NS Ψ stem -0.63** -0.53** NS gS -0.25* 0.21* NS 0.75** MDS: Maximum daily shrinkage; SWC30: soil water content at 30 depth; SWC60: soil water content at 60 of depth; Ψ stem: stem water potential; gS: stomatal conductance; NS: no significant; * and ** significant correlations at P < 0.05 and P < 0.01, respectively. Impact of regulated and low deficit irrigation on the physiological response… 111 4.4 suggests that higher correlation coefficients might be obtained with MXSD versus SWC30, Ψstem and gs (data not shown) than with MDS. We also measured the maximum stem diameter (MXSD, mm), which showed different seasonal patterns, in each treatment (Fig. 4.4). The dry period coincided with the fruit growth stage. This explains the limited trunk growth showed in Fig. 4.4 for the control trees from DOY 179 to DOY 242. Only at the end of this period, coinciding with the fruit maturity phase, there was a slight increase in trunk diameter. After harvesting, from DOY 268, net trunk growth was negligible. In relation to the SDI treatment, we observed a significant decrease of the MXSD values due to the accumulated water stress of these trees. The MXSD trend in the SDI trees changed with the first rainfall events at the end of the dry period. This increment confirms the recovery of these trees suggested by the Ψ stem and gs observations (Fig. 4.3). Finally, the MXSD values registered in LFDI trees showed decline and recovery periods according to the irrigation events in this treatment. At the end of the season the LFDI trees showed a net growth similar to that in control trees. Likely, the depletion of SWC together with high atmospheric demand led to negligible growth and progressive dehydration of tissues at the end of the withholding periods, which causes the marked decreases of MXSD recorded in the LFDI trees at the end of the periods in which irrigation was withheld. The comparison of Figs. 4.2, 4.3 and 4.4 suggests that higher correlation coefficients might be obtained with MXSD versus SWC30, Ψ stem and gs (data not shown) than with MDS. 4.3.3. Yield and fruit quality response Table 4.3 shows the effects of the water treatments on yield and fruit quality parameters. Yield was lower both in LFDI and SDI than in the control treatment, the decreases being due to both lower fruit weight and reduced number of fruits per tree. Concerning juice content, decreases of 17% in SDI and 10% in LFDI, were recorded, as compared to the control. There were also significant effects on fruit organoleptic properties. Significant increases of total soluble solids and titratable acidity were registered both in the SDI and LFDI treatments. We also observed increases of maturity index in SDI and LFDI, although only in the LFDI treatment the differences were significant. 112 Chapter 4 Values of WP were 0.03 kg L-1 in LFDI, 0.025 kg L-1 in SDI and 0.024 kg L-1 in control. These results show that greatest WP values did not only depend on the amount of water supplied, but also on the irrigation strategy. 4.4. Discussion The MDS values observed in the SDI trees and during the withholding periods and in the LFDI treatment along the season were significant higher than in the control trees (Fig. 4.3C). This agrees with the levels of water stress suffered by the trees (Fig. 4.3A). This is in accordance with the results reported by Ortuño (2004a) for young lemon trees and Ginestar (1995) for mandarin trees. These authors concluded that when Ψ stem was higher than-2.5 MPa an increase in MDS was associated with a decrease in Ψ stem. On the other hand, Ortuño et al. (2004b), in an experiment with young lemon trees, found differences in MDS values the second day after the IAs were significantly restricted, i.e. soon after the increase in water stress. The water depletion was 25% lower in the deficit irrigation treatment compared with the control treatment. Steppe et al. (2006) reported fluctuations in gs measured in young orange trees, even with stable atmospheric conditions. The observed fluctuations in gs were accompanied by lower fluctuations in Ψ stem, as observed in our experiment. Stomatal Table 4.3. Yield and fruit quality parameters. Treatments Control SDI LFDI Yield (kg tree -1 ) 152.6 a 91.4 c 124.4 b Fruit weight (g) 264.7 a 228.1 b 237 b Fruits per tree 577a 495b 497b Juice weight (%) 47.5 a 39.9 c 43.5 b TSS (ºBrix) 10.3b 12.6a 13.3a TA (g L -1 ) 1.2 b 1.5 a 1.5 a MI 8.3b 8.8b 9.2a ED (mm) 79.4a 72.6b 72.7b PD (mm) 84.6a 80.3b 79.2b Peel thickness (mm) 6.2 a 8.7 b 6.4 a SDI: Sustainable deficit irrigation; LFDI: Low frequency deficit irrigation; TSS: total soluble solids; TA: titratable acidity; MI: maturity index; ED: equatorial diameter; PD: polar diameter. Impact of regulated and low deficit irrigation on the physiological response… 113 oscillations have been observed in several species, including orange trees (Dzikiti et al., 2007; Westhoff et al., 2009; Zimmermann et al., 2010). Zimmermann et al. (2010) showed a strong correlation between the frequency (and amplitude) of the oscillations and the wind speed. An increase in wind speed obviously reduces the hydration boundary layer on the leaves and, in turn, increases temporarily local transpiration causing a reduction of turgor pressure in the leaf tissues. This phenomenon may explain the fluctuation of the gs values we observed in the field (Fig. 4.3B). Our results show that MDS could be a reliable and sensitive indicator of water stress for a rational control of DI strategies in commercial citrus orchards. Similar findings were obtained in previous experimental works (Ortuño et al., 2006), showing high significant correlations between MDS and Ψ stem (r = 0.77) in young lemon tree. The same can be deduced from the review works by Ortuño et al. (2010) and Fernández and Cuevas (2010). Our results also suggest that MXSD could also be an advisable indicator of water stress for the DI trees. Other authors have found a better performance on MXSD than MDS as an indicator of water stress in fruit tree orchards. This is the case of Cuevas et al. (2010) and Fernández et al. (2011) in mature olive trees with heavy crop load. Regarding gs, the significant differences between treatments showed in Fig. 4.3B is in accordance with Gomes et al. (2004). The authors studied the effects of a LFDI treatment on gs of orange trees. They reported significant differences on the gs values of the LDFI trees after seven days without irrigation, as compared with those full irrigated. On the tenth day after rewatering the gs values of the stressed trees were not significant different from those in the non stressed trees. Yield was significant lower in the SDI trees than in the control trees, the decreases being due to a reduction in both fruit weight and No. fruits per tree. Several authors have described the agronomic effects of DI, showing important differences among species. Thus, the applications of SDI strategies in vineyards have shown a general increase in yield and, in some instances, also some beneficial effects on fruit ripening (García- Escudero et al., 1991, 1997; Rubio et al., 2004; Yuste et al., 2005; Intrigliolo and Castel, 2008). In apricot trees (Ruiz-Sánchez et al., 2000; Pérez-Pastor et al., 2009) SDI applied throughout the growing season affected productivity and limited vegetative and reproductive growth. In almond trees, Girona et al. 2005 considered more interesting a 120 Chapter 4 Ortuño MF, Alarcón J, Nicolás E, Torrecillas A. 2004. Comparison of continuously recorded plant-based water stress indicators for young lemon trees. Plant and Soil 267(1):263-270. Ortuño MF, Brito JJ, García-Orellana Y, Conejero W, Torrecillas A. 2009. Maximum daily trunk shrinkage and stem water potential reference equations for irrigation scheduling of lemon trees. Irrigation Science 27(2):121-127. Ortuño MF, Conejero W, Moreno F, Moriana A, Intrigliolo DS, Biel C, Mellisho CD, Pérez-Pastor A, Domingo R, Ruiz-Sánchez MC. 2010. Could trunk diameter sensors be used in woody crops for irrigation scheduling? A review of current knowledge and future perspectives. Agric Water Manage 97(1):1-11. Ortuño MF, García-Orellana Y, Conejero W, Ruiz-Sánchez MC, Alarcón JJ, Torrecillas A. 2006. Stem and leaf water potentials, gas exchange, sap flow, and trunk diameter fluctuations for detecting water stress in lemon trees. Trees-Structure and Function 20(1):1-8. Pérez-Pastor A, Domingo R, Torrecillas A, Ruiz-Sánchez MC. 2009. Response of apricot trees to deficit irrigation strategies. Irrigation Science 27(3):231-242. Pérez-Pérez J, Romero P, Navarro J, Botía P. 2008a. Response of sweet orange cv ‘Lane late’ to deficit irrigation in two rootstocks. I: water relations, leaf gas exchange and vegetative growth. Irrigation Science 26(5):415-425. Pérez-Pérez J, Romero P, Navarro J, Botía P. 2008b. Response of sweet orange cv ‘Lane late’ to deficit-irrigation strategy in two rootstocks. II: Flowering, fruit growth, yield and fruit quality. Irrigation Science 26(6):519-529. Pérez-Pérez J, Robles J, Botía P. 2009. Influence of deficit irrigation in phase III of fruit growth on fruit quality in ‘lane late’ sweet orange. Agric Water Manage 96(6):969-974. Romero P, Navarro J, Pérez-Pérez J, García-Sánchez F, Gómez-Gómez A, Porras I, Martínez V, Botía P. 2006. Deficit irrigation and rootstock: their effects on water relations, vegetative development, yield, fruit quality and mineral nutrition of Clemenules mandarin. Tree Physiology 26(12):1537. Impact of regulated and low deficit irrigation on the physiological response… 121 Rubio J, Alburquerque MV, Yuste J. 2004. Influence of water stress and crop load on the physiology and the productivity of Tempranillo under semiarid conditions. Acta Hort (ISHS) 640:99-106. Ruiz-Sánchez MC, Domingo Miguel R, Castel Sánchez JR. 2010. Review. Deficit irrigation in fruit trees and vines in Spain. Plant Soil 120:299-302. Ruiz-Sánchez MC, Torrecillas, A., Pérez-Pastor, A. and Domingo, R. 2000. Regulated deficit irrigation in apricot trees. Acta Horticulturae 537:759-766. Scholander PF, Bradstreet ED, Hemmingsen E, Hammel H. 1965. Sap pressure in vascular plants. Science 148(3668):339. Seckler DW. 1996. The new era of water resources management: from" dry" to" wet" water savings: Issues in Agriculture No. 8 CGIAR. 17 p. Steppe K, Dzikiti S, Lemeur R, Milford JR. 2006. Stomatal oscillations in orange trees under natural climatic conditions. Annals of botany 97(5):831. Treeby MT, Henriod RE, Bevington KB, Milne DJ, Storey R. 2007. Irrigation management and rootstock effects on navel orange Citrus sinensis (L.) Osbeck fruit quality. Agric Water Manage 91(1-3):24-32. Turner NC. 1988. Measurement of plant water status by the pressure chamber technique. Irrigation Science 9(4):289-308. Velez J, Intrigliolo D, Castel J. 2007. Scheduling deficit irrigation of citrus trees with maximum daily trunk shrinkage. Agric Water Manage 90(3):197-204. Westhoff M, Reuss R, Zimmermann D, Netzer Y, Gessner A, Geßner P, Zimmermann G, Wegner L, Bamberg E, Schwartz A. 2009. A non invasive probe for online monitoring of turgor pressure changes under field conditions. Plant Biology 11(5):701-712. Yakushiji H, Morinaga K, Nonami H. 1998. Sugar accumulation and partitioning in Satsuma mandarin tree tissues and fruit in response to drought stress. American Society for Horticultural Science (USA) 123:719-726. 122 Chapter 4 Yuste J, Asenjo J, Alburquerque M, Rubio J. 2005. Relationships among Physiology, Growth and Production as Affected by Water Regime and Vine Spacing of 'Tempranillo' Grapevines. Acta Horticulturae 689:343-348. Zimmermann U, Rüger S, Shapira O, Westhoff M, Wegner L, Reuss R, Gessner P, Zimmermann G, Israeli Y, Zhou A. 2010. Effects of environmental parameters and irrigation on the turgor pressure of banana plants measured using the non invasive, online monitoring leaf patch clamp pressure probe. Plant Biology 12(3):424-436. 123 Chapter 5 An automatic irrigation controller for fruit tree orchards, based on sap flow measurements Part of this chapter is published in: Fernández JE, Romero R, Díaz-Espejo A, Cuevas MV, Muriel JL, Montaño JC. 2008. A device for Scheduling irrigation in fruit tree orchards from sap flow readings. Acta Horticulturae, 792: 283-290. Fernández JE, Romero R, Díaz-Espejo A, Muriel JL, Cuevas MV, Montaño JC. 2008. Design and testing of an automatic irrigation controller for fruit tree orchards based on sap flow measurements. Australian Journal of Agricultural Research 59, 589–598. Abstract. We designed and tested an automatic irrigation control system for fruit tree orchards, denominated CRP. At the end of each day, the device calculates the irrigation dose (ID) from sap flow readings in the trunk of trees irrigated to replenish the crop water needs, relative to similar measurements made in overirrigated trees. It then acts on the pump and electrovalve to supply an ID enough to maintain the soil close to its field capacity during the irrigation period. Remote control of the system is possible from any computer or smartphone connected to the Internet. We tested the CRP in an olive orchard in southern Spain. The device was robust and able to filter and amplify the output voltages of the heat-pulse velocity probes and to calculate reliable sap flow data. It calculated and supplied daily irrigation amounts to the orchard according to the specified irrigation protocol. The remote control facility showed to be useful for getting real-time information both on the CRP behaviour and the applied IDs, and for changing parameters of the irrigation protocol. For our conditions, olive trees with big root systems growing in a soil with a remarkable water-holding capacity, the approach mentioned above for calculating ID had not enough resolution to replace the daily crop water consumption. The device, however, was able to react when the soil water content fell below the threshold for soil water deficit. The threshold value was identified with simultaneous measurements of stem water potential in the instrumented trees. Our results suggest a change in the irrigation 124 Chapter 5 protocol that will allow the CRP to apply a recovery irrigation whenever that threshold is reached, making the device suitable for applying a deficit irrigation strategy in the orchard. 5.1. Introduction Demands from other water-user sectors, apart from agriculture, are increasing pressure to improve irrigation practices (Fereres and Evans, 2006). As a response, new approaches and technologies to optimise irrigation scheduling are being developed. Those based on monitoring a key variable in the plant have the potential advantage of integrating, in a single measurement, not only the plant’s response to the prevailing soil and atmospheric water conditions, but also the effect of the plant’s physiological mechanisms controlling water use. Thus, canopy temperature, water content in the trunk, trunk diameter variations, and sap flow readings are currently considered promising plant-based variables for irrigation control (Fereres et al., 2003; Jones, 2004). An additional advantage of these variables for irrigation scheduling is that they can be automated. Infrared thermometry has been used since the sixties, both to monitor plant stress and to provide useful information for irrigation control (Jackson, 1982; Hatfield, 1983). Recent improvements have increased the potential of the technique as a tool for irrigation scheduling (Jones, 1999; Alves and Pereira, 2000; Lobo et al., 2004). However, the use of this technique in fruit tree orchards has to overcome the difficulty of the high variability induced by the canopies’ covering just part of the orchard floor. Monitoring water content changes in the stem of some species by automatable techniques, such as time-domain reflectometry (TDR), has long been used for detecting the onset of water stress (Constant and Murphy, 1990; Holbrook and Sinclair, 1992). It seems, however, that simpler and less expensive sensors are required before farmers adopt this approach for irrigation control (Nadler et al., 2006). Extensive research has been carried out recently to improve irrigation control based on trunk diameter variations and sap flow in the trunk. Continuous monitoring of trunk diameter by linear variable displacement transducers (LVDT sensors) for assessing tree’s response to irrigation water deficits has been studied in a variety of fruit tree species. In olive, the technique has been evaluated by Moriana and Fereres (2002) and Moreno et al. (2006). An example of irrigation control based on the use of LVDT sensors An automatic irrigation controller for fruit tree orchards, based on sap flow… 125 in almond trees was published by Goldhamer and Fereres (2004). Recently, García- Orellana et al. (2007) have evaluated the feasibility of scheduling irrigation of lemon trees from the maximum daily trunk shrinkage. Sap flow readings are being widely used to determine both water consumption and the dynamics of transpiration and water uptake by main roots. Among the different sap flow methods, the compensation heat-pulse (CHP) method has been used in various fruit tree species, including kiwifruit (Green et al., 1989), apple (Green et al., 1989, 2003), pear (Caspari et al., 1993), apricot (Alarcon et al., 2000, 2003; Nicolás et al., 2005), lemon (Ortuño et al., 2004), and olive (Moreno et al., 1996; Fernández et al., 2001, 2003, 2006a,b; Giorio and Giorio, 2003; Williams et al. 2004). Comparisons between sap flow and trunk diameter readings as water stress indicators in fruit trees, and between these two variables and more-traditional water-stress indicators such as leaf- or stem-water potential and stomatal conductance, have been carried out by Ortuño et al. (2004, 2005, 2006a, b), Conejero et al. (2007) and Intrigliolo and Castel (2006a, b), among others. These papers outline the high potential of the two variables. Most of the irrigation controllers available on the market require the irrigation dose (ID) to be provided by the user. Only then are they able to switch the irrigation pump on and off and to open and close electrovalves to apply the input ID to each sector of the orchard. In this work, we consider an automatic irrigation controller (AIC) to be a device able to calculate ID by itself, based on one or several recorded variables, and to act on the irrigation system so that the calculated ID is supplied to the crop. These processes must be carried out automatically. Very few AICs have been developed so far. Most of them are based on soil matric potential measurements (Luthra et al., 1997; Klein, 2004; Miranda et al., 2005). These are relatively inexpensive, user-friendly devices, but soil moisture monitoring implies certain limitations: a high number of sensors may be required to cope with the spatial variability within the orchard, and they do not take plant performance into account. To our knowledge, an AIC based on either LVDT or sap flow readings has not yet been developed, although irrigation protocols based on both approaches have been suggested (Fernández et al., 2001; Goldhamer and Fereres, 2004; García-Orellana et al., 2007). The ‘Pepista’ system (Pelloux et al., 1990) uses trunk diameter variations to control irrigation, but it requires criteria for data interpretation. 126 Chapter 5 Nadezhdina and Cermak (1997) reported an AIC based on what they call sap flow index (see Nadezhdina, 1999, for details), but a commercial version was not released. The aim of the present work was to design and build an AIC for fruit tree orchards, based on sap flow measurements in the trunk of representative trees, and to test the performance of the device after installing it in an olive orchard, both for amplifying and filtering output signals from the thermocouples of the sap flow sensors and for calculating the ID according to the specified irrigation protocol. Water status, both in the instrumented trees and soil, was monitored throughout the field test, and results compared with the calculated IDs. 5.2. Materials and methods 5.2.1. Irrigation controller We named the AIC developed in this work CRP, the Spanish acronym for crop irrigation controller. The CRP has been designed to adjust the ID daily to maintain the soil around field capacity. The ID values are automatically calculated from sap flow readings in the trunk of trees irrigated to supply the crop water needs, denominated normally irrigated (NI) trees, relative to similar measurements made in overirrigated (OI) trees, used as reference trees. The CRP was built in our laboratory during 2005, and tested in an olive orchard in 2006, as described below. The CRP has three main physical components: a) The measurement unit (MU). This part of the CRP is installed in the orchard, and records sap flow in one NI and one OI tree. The MU uses the heat-pulse velocity (HPV) system of Green (1998) for sap flow readings. This system employs the CHP method. The CRP tested here had three MUs, each with three sets of HPV probes (Tranzflo NZ Ltd, Palmerston North, New Zealand) per tree. These, together with the corresponding heat-pulse controllers and the power supply system, were the only components of the CRP not designed in our laboratory. The main characteristics of the HPV probes are described in Green et al. (2003). Each set of probes has two temperature probes, inserted at x´=5 mm upstream and x=10 mm downstream of the heater. Each temperature probe consists of four thermocouples, at 5, 12, 22, and 35 mm below the An automatic irrigation controller for fruit tree orchards, based on sap flow… 127 cambium. Heat pulses were fired for 1 s every 30 min throughout the testing period. After each heat pulse, output signals from each HPV probe are amplified and filtered to derive tz values, the time intervals for the coincidence of temperatures at points x’ and x (see appendixes 5.A1 and 5.A2). The four tz values from each set of probes are stored in an EEPROM. b) The control unit (CU). This is a standard PC located close to the head of the irrigation system, wired to each of the three MUs, using RS485 communication protocol. This allows up to 32 devices to communicate at half-duplex on a single pair of wires at distances up to 1200 m, more than enough for our experimental set-up. For operating, the CU runs the software described in the appendix 5.A3. Every night, at 23:45 hours, the CU gets the tz values measured by each MU that day. After saving these values, the CU uses them to calculate the sap flow values (Q, L hour-1) for each HPV probe set and time of measurement. The Q values are saved, and then used to calculate the total tree transpiration for the day (Ep, L day-1). The CU then computes the ID for the next day as described below, and sends this value to the pump & electrovalve controller. c) The pump & electrovalve controller (PEC). This device, also located close to the head of the irrigation system, is wired to the CU, from which it receives the calculated ID. The role of the PEC is to switch the irrigation pump on and off, and to open and close the electrovalve of the irrigation sector for a time sufficiently long to supply the ID. In addition, the PEC reads pulses from a flow meter installed after the electrovalve, enabling the system to control the supply of water either by volume or by time (software details in appendix 5.A4). Fig. 5.1 shows the tasks carried out by each of the main components of the CRP. The number of MUs and electrovalves controlled can be customised depending on orchard requirements. The number of HPV probe sets per tree, the depth of the thermocouples within each probe, and other parameters related to sap flow monitoring, can also be changed. We used a remote desktop of Windows XP to access the CU via the Internet from home or office, for remote control of the CRP functions, including parameter changing and program debugging. 128 Chapter 5 ID Fig. 5.1. Flow diagram showing the tasks of the three main physical components of the CRP: the measuring unit (MU), the control unit (CU), and the pump & electrovalve controller (PEC ). See text for details. tz’s G e t t z ' s f r o m U M ' s S a v e t z ' s i n . D A T f i l e s U s e t z ' s t o e s t i m a t e Q ' s S a v e Q ' s i n . S U M f i l e s E s t i m a t e E v a p o t r a n s p i r a n t i o n f o r O I & N I t r e e s ( E p O I & E p NI ) C a l c u l a t e i r r i g a t i o n d o s e ( I D ) d y e v o l u t i o n o f E p N I / E p O I I r r i g a t e C U Y e s No H o u r = 2 3 : 4 5 ? f r o m l a i An automatic irrigation controller for fruit tree orchards, based on sap flow… 129 The irrigation protocol was as follows: the soil in the orchard must be around field capacity on the first two days of the irrigation period. On these two days, the CRP does not calculate the ID; instead, its value is defined by the user. At 23:45 hours on the first day of the irrigation period, the CU collects the tz values from each MU and calculates Ep of that day, for each NI tree (EpNI) and each OI tree (EpOI). The CU then calculates the transpiration ratio between the two types of tree (EpNI/EpOI). The following day, the CU does the same, and compares the resulting value with the one calculated on the previous day. From this comparison, the CU automatically adjusts the ID for the next day (the third of the irrigation period): if (EpNI/EpOI)DOY ≅ ( EpNI/EpOI)DOY-1 (where DOY denotes ‘day- of-year’), the ID applied on the current day could have been either enough to cover the water needs of the NI trees or too high. Accordingly, the ID of the next day is reduced. If (EpNI/EpOI)DOY ≠ ( EpNI/EpOI)DOY-1, the CU assumes the ID applied on the current day was not enough to cover the demand of the NI trees, and increases the ID for the next day. This procedure is repeated each day of the irrigation period, with the following rules: 1) If the transpiration ratios of two consecutive days are considered different, that is, |(EpNI/EpOI)DOY-1 -(EpNI/EpOI)DOY|>0.05(EpNI/EpOI)DOY-1; then the CU increases the ID by 10% on the next day, and by 20% on the following days. 2) If |(EpNI/EpOI)DOY-1 -(EpNI/EpOI)DOY|≤0.05(EpNI/EpOI)DOY-1, then the CU reduces the ID by 10% on the first day, and by 20% on the following days. 5.2.2. Field trial We tested the CRP in an olive orchard (Olea europaea cv. ‘Manzanilla de Sevilla’) of La Hampa, an experimental farm of the Spanish Research Council (CSIC) close to Seville, Spain (37º 17’ N, 6º 3’ W, elevation 30 m). The trees, planted at 7 m x 5 m, were 38 years old. The soil is a sandy loam (Xerochrept) with a depth of about 2 m. The soil texture is quite homogeneous, with average values of 73.5% coarse sand, 4.7% fine sand, 14.8% clay, and 7.0% silt. Laboratory measurements showed that the volumetric soil water content (θv, m3 m-3) for -0.1 and -1.5 MPa soil matric potential was 0.33 m3 m-3 and 0.10 m3 m-3, respectively. Field measurements, however, showed that θv values close to the emitters a few hours after irrigation were rarely greater than 0.26 m3 m-3. 136 Chapter 5 the variable atmospheric demand was responsible, at least in part, for the changing EpNI/EpOI values recorded on that period. Apart from that, on the first 46 days, the time course of the daily EpNI/EpOI values showed a slope of approximately zero, with EpNI/EpOI = 0.85 on DOY 175. Consequently, the IDs calculated by the CRP decreased from DOY 130 until DOY 176, when the calculated ID was just 0.58 L tree-1 (Fig. 5.4C). This caused a decrease in the soil water content in the rootzone of the NI trees, with REW≈0.57 on DOY 176 (Fig. 5.4D). From DOY 177 to 187, differences between the EpNI/EpOI ratios of consecutive days were greater than 5% (Fig. 5.4B); from DOY 188 to 196, the EpNI/EpOI ratio decreased markedly nearly everyday, showing, for the first time since the beginning of the irrigation season, a clear response to the low water supplied to the NI trees. The daily evolution of the EpNI/EpOI ratio from DOY 177 to 196 yielded increasing IDs calculated by the CRP according to the irrigation protocol: by 10% on DOY 177 and by 20% on the following days. Due to the low ID on DOY 176, however, the ID on DOY 196 was only 6.71 L tree-1, much lower than the ETc value estimated for that day with the crop coefficient approach (106.9 L tree-1) (Fig. 5.4C). Both Ψstem (Fig. 5.4A) and the EpNI/EpOI ratio (Fig. 5.4B) decreased markedly in the NI trees from DOY 187, indicating that the soil water content was too low to prevent a significant increase in the trees’ water stress. This made us apply a recovery irrigation of 102 L tree-1 on DOY 199, corresponding to the ETc value estimated with the crop coefficient approach (Fig. 5.4C). From that day on, we left the CRP to control irrigation again as programmed. As shown in Fig. 5.4C, the ID increased for four consecutive days, reaching a maximum of 193.9 L tree-1 on DOY 203, and then decreased. The average REW in the rootzone of the NI trees went from a minimum of 0.48 on DOY 196 to a maximum of 0.80 on DOY 209. According to this change in the soil water status, Ψstem of the NI trees recovered quickly, reaching values similar to those of the OI trees on DOY 205. It took longer, however, for the EpNI/EpOI ratio to recover. This agrees with what is already known about the behaviour of the olive tree after a recovery irrigation: the water potential of stressed olive trees recovers quickly after rewatering, but it takes longer for stomatal conductance to recover, the delay being related to the level of water stress previously reached (Fereres et al. 1996; Fernández et al. 1997). Large daily fluctuations in the EpNI/EpOI ratio were often recorded from DOY 224 to 242, which led the CRP to calculate increasing IDs on nearly every day of that period (Fig. 5.4B,C). This caused a substantial increase in the soil water content of An automatic irrigation controller for fruit tree orchards, based on sap flow… 137 the NI trees, reaching similar values to those of the OI trees from DOY 234. The lack of water available for irrigation on the farm from DOY 245 made us reduce the irrigation supplied to the OI trees to about one third of ETc, and adjust the ID calculated by the CRP on DOY 248 to 30 L tree-1. No significant changes were recorded from that day to the end of the irrigation season (September 16; DOY 259), apart from the decrease in both Ψstem and REW, for all the experimental trees, caused by the mentioned reduction in the irrigation amounts. The time-course evolution both of the EpNI/EpOI ratio and Ψstem from the beginning of the irrigation season until DOY 187 (Fig. 5.4A,B), suggests that the NI trees were able to take up similar amounts of water from the soil as the OI trees until REW≈0.50, the value recorded on that day. The cumulated ETc from DOY 128 to 187 amounted to 4520 L tree-1, while the amount of water supplied in that period by the CRP to the NI trees was 1410 L tree-1 only. The difference, 3110 L tree-1, should have been provided by the soil. Fernández et al. (2006b) estimated that the available water in the soil orchard is 170 mm, which means some 85 L of water per cubic metre of soil before REW≈0.5. Therefore, some 36 m3 of rhizosphere should have been needed for each tree to be able to cover the mentioned difference. Taking into account that the average effective rootzone depth estimated from the soil water profiles measured during the irrigation period was 1.6 m, such rhizosphere could correspond to about 23 m2 ground surface, of the 35 m2 available to each tree. This is reasonable, since it agrees with observations on the root system distribution carried out in the same orchard by Fernández et al. (1991). Besides the relatively high amount of water available for each tree from the soil water reserves, the high capacity of the olive tree for taking up water from drying soils (Xiloyannis et al., 1998) probably contributed to the delay in the Ψstem and EpNI/EpOI responses to the low IDs applied until DOY 187. It seems that, for our orchard conditions, REW≈0.5 is a threshold for soil water depletion at the beginning of the irrigation period, when the soil outside the irrigation bulbs is wetted after the rainy season. Later in the season, from about DOY 210, values of Ψstem in the NI trees were lower than those recorded before DOY 187, despite of the greater REW values recorded at that time (Fig. 5.4A,D). This cannot be attributed to differences in atmospheric demand only, because average ETo values recorded in the two periods were similar (Fig. 5.3). We can assume that the volume of soil from which the roots could absorb water from DOY 210 was lower than at 138 Chapter 5 the beginning of the irrigation period, being likely restricted to the irrigation bulbs. Under these conditions, any reduction in the volumetric soil water content within the irrigation bulbs has a greater impact on the amount of water taken up by the tree. We can expect, therefore, greater threshold REW values later in the irrigation period than at the beginning. These results suggest that the threshold value of REW=0.25 usually recommended for olive (Orgaz and Fereres, 2004) is too low. The mentioned results suggests that the EpNI/EpOI ratio may not has enough resolution for the daily adjustment of ID to keep the soil around field capacity, in orchards where the soil has a medium-to-high soil water-holding capacity and the roots of trees explore large volumes of soil. The ratio, however, seems to be a sensitive indicator of a threshold value of water stress, even for those conditions, suggesting a potential for the automatic control of deficit irrigation. Depending on the soil characteristics, a careful evaluation of the level to which the OI trees must be irrigated could be required. Thus, lack of oxygen in the rootzone, nitrogen deficiency because of excessive leaching losses, and anomalous leaf area development could make overirrigated trees to become non-representative of those in the orchard (Goldhamer and Fereres, 2001; Fernández et al. 2007). In addition, there are evidences of flooding causing a reduction in the recorded value of some plant-based water status indicators, including sap flow (Ortuño et al., 2007). In this work we irrigated our OI trees to 130% of ETc, to assure non-limiting soil water conditions all throughout the experiment. On the other hand, the fact that our experimental soil has no restrictions to percolation (Palomo et al., 2002) made us to expect a negligible impact of the mentioned problems derived from overirrigating the trees. Our results show that the increase or reduction of ID by 10% on the next day and 20% on the following days according to the differences of the EpNI/EpOI ratios of two consecutive days, was not always enough to match the changes on the crop water requirements. This was the case from DOY 177 to 196, Fig. 5.4C. We have found no evidences in the literature on the physiological bases to be taken into account for adjusting properly those percentages. There are, however, articles in which other authors evaluate, for different species, the results of working with similar percentages as those used in this work. Thus, Conejero et al. (2007) evaluated the use of sap flow and trunk An automatic irrigation controller for fruit tree orchards, based on sap flow… 139 diameter readings for scheduling irrigation in young peach trees. They reduced irrigation by 10% when the signal intensities derived from the sensor readings was at or below unity on at least two or three consecutive days, and increased irrigation by 10% when the signal intensity exceeded unity. Their results show that the precision of water scheduling decreased during periods of increasing irrigation need, and that 10% irrigation increases were insufficient, or that irrigation should be scheduled more frequently than every three days. Another example is that of Velez et al. (2007), who used maximum daily trunk shrinkage (MDS) in mature ‘Clementina de Nules’ citrus trees to schedule deficit irrigation. The authors varied in ± 10 to 20% the water applied weekly, according to the evolution of the MDS ratio between overirrigated and control trees. By using this approach they managed to maintain the MDS ratio close to the target value, for most of the season. Analysis of the EpNI/EpOI values showed that reducing the 5% threshold difference between (EpNI/EpOI)DOY and (EpNI/EpOI)DOY-1 is not advisable: daily fluctuations were close to that value even on days with no significant changes in either the soil water content or the atmospheric demand. We cannot explain the unusually high fluctuations in the EpNI/EpOI ratio recorded from DOY 224 to 242. The system apparently worked well, and no significant changes in the environmental conditions occurred in that period, except for a 15 mm rainfall event on DOY 229, likely responsible for both the increase on Ψstem (Fig. 5.4A) and the decreased on ID (Fig. 5.4C) recorded on the following days. Except for period between DOY 224 and 242, the only case in which daily fluctuations of the EpNI/EpOI ratio were greater than 5% for several consecutive days was from DOY 187 (Fig. 5.4B), in agreement with the decrease in soil water content to below the threshold for soil water (Fig. 5.4D), and the marked decrease in Ψstem of the NI trees (Fig. 5.4A). This, together with the results discussed above, suggests the need for a change in the irrigation protocol: the CRP must apply a recovery irrigation when the EpNI/EpOI ratio decreases more than 5% for three consecutive days. The amount of water for the recovery irrigation must be set by the user, depending on the fruit tree species and orchard characteristics. For our experimental orchard, 100 L tree-1 would be a reasonable amount (Fernández and Moreno 1999). The effect that this change in the irrigation protocol might have on CRP performance is still to be tested. 140 Chapter 5 Monitoring the soil water status for scheduling irrigation can be recommended in some cases, especially for homogeneous, annual crops with reduced rootzones (Hoppula and Salo, 2007; Thompson et al. 2007). In the case of woody plants exploring big volumes of soil, soil water variability imposes limitations to this approach, especially because the number of required sensors may become unaffordable. In those cases, plantbased measurements may be especially useful for irrigation scheduling purposes, since they inform on the plant response to the soil and atmospheric conditions (Jones, 2004). Compared to irrigation scheduling methods based on the atmospheric demand, such as the crop coefficient method (Allen et al. 1998), plant-based measurements can increase the resolution of the calculated IDs, which is certainly an advantage for precise highfrequency irrigation. In addition, plant-based measurements such as sap flow and trunk diameter can be easily automated, which is particularly valuable for irrigation scheduling (Jones, 2007). Automatic irrigation controllers, such as the CRP, have an additional advantage: the device is able not only to measure the plant indicator automatically; it also processes the collected information and operates the irrigation system to apply the calculated ID. Still, the highly variable soil and crop conditions in many commercial orchards may limit the potential use of AICs such as the CRP. On those cases, however, the use of remote sensing techniques could inform on the variability of water stress within the orchard, helping to choose the most representative locations for the MUs, thus reducing the number of required sensors. See the work by Sepulcre-Cantó et al. (2007) for an example on the use of those techniques in olive and peach orchards. 5.4. Conclusions The CRP proved to be a robust device able to calculate and supply daily irrigation amounts to the orchard, in accord with the specified irrigation protocol. It allows the user to interact with the device from any computer, PDA, or smartphone connected to the Internet, for consulting, changing parameters, or even taking full control of the irrigation practice. This is a clear advantage for irrigating orchards in remote areas. For the case studied in this work (olive trees with big root systems growing in a soil with a high waterholding capacity) the daily values of the EpNI/EpOI ratio had not enough resolution for the desired irrigation approach, intended to replace the daily crop water consumption. The CRP, however, was able to react to a sudden increase in the tree’s water stress caused by An automatic irrigation controller for fruit tree orchards, based on sap flow… 141 the soil water content falling below the threshold for soil water deficit, suggesting the device could be suitable for applying deficit irrigation in olive orchards with similar characteristics as our experimental orchard. This would require a small change in the irrigation protocol, suggested by our field results. The resolution of the EpNI/EpOI ratio could be greater when irrigating species with a lower capacity to take up water from drying soils, especially if these have a low water-holding capacity. Whether for those species and soils the CRP will be efficient to keep the soil close to its field capacity throughout the irrigation season is still to be tested. 5.5. References Alarcon JJ, Domingo R, Green SR, Nicolas E, Torrecillas A. 2003. Estimation of hydraulic conductance within field-grown apricot using sap flow measurements. Plant and Soil 251(1):125-135. Alarcón JJ, Domingo R, Green SR, Sánchez-Blanco MJ, Rodríguez P, Torrecillas A. 2000. Sap flow as an indicator of transpiration and the water status of young apricot trees. Plant and Soil 227(1-2):77-85. Allen RG, Pereira LS, Raes D, Smith M. 1998. Crop evapotranspiration-Guidelines for computing crop water requirements-FAO Irrigation and drainage paper 56. FAO, Rome 300:6541. Alves I, Pereira LS. 2000. Non-water-stressed baselines for irrigation scheduling with infrared thermometers: A new approach. Irrigation Science 19(2):101-106. Caspari HW, Green SR, Edwards WRN. 1993. Transpiration of well-watered and waterstressed Asian pear trees as determined by lysimetry, heat-pulse, and estimated by a Penman-Monteith model. Agricultural and Forest Meteorology 67(1-2):13-27. Conejero W, Alarcón J, García-Orellana Y, Nicolás E, Torrecillas A. 2007. Evaluation of sap flow and trunk diameter sensors for irrigation scheduling in early maturing peach trees. Tree physiology 27(12):1753. 142 Chapter 5 Constantz J, Murphy F. 1990. Monitoring moisture storage in trees using time domain reflectometry. Journal of Hydrology 119(1-4):31-42. Fereres E, Evans RG. 2006. Irrigation of fruit trees and vines: an introduction. Irrigation Science 24(2):55-57. Fereres E, Goldhamer DA, Parsons LR. 2003. Irrigation water management of horticultural crops. Hortscience 38(5):1036-1042. Fereres E, Ruz C, Castro J, Gómez JA, Pastor M. 1996. Recuperación del olivo después de una sequía extrema. Proceedings of the XIV Congreso Nacional de Riegos. Aguadulce. Almería. p 89-93. Fernández JE, Díaz-Espejo A, Infante JM, Durán P, Palomo MJ, Chamorro V, Girón IF, Villagarcía L. 2006. Water relations and gas exchange in olive trees under regulated deficit irrigation and partial rootzone drying. Plant and Soil 284(1- 2):273-291. Fernández JE, Durán PJ, Palomo MJ, Díaz-Espejo A, Chamorro V, Girón IF. 2006. Calibration of sap flow estimated by the compensation heat pulse method in olive, plum and orange trees: relationships with xylem anatomy. Tree Physiology 26(6):719-728. Fernández JE, Green SR, Caspari HW, Díaz-Espejo A, Cuevas MV. 2008. The use of sap flow measurements for scheduling irrigation in olive, apple and Asian pear trees and in grapevines. Plant and Soil 305(1-2):91-104. Fernández JE, Moreno F. 1999. Water use by the olive tree. Journal of crop production 2(2):101-162. Fernández JE, Moreno F, Cabrera F, Arrue JL, Martín-Aranda J. 1991. Drip irrigation, soil characteristics and the root distribution and root activity of olive trees. Plant and Soil 133(2):239-251. Fernández JE, Moreno F, Girón IF, Blázquez OM. 1997. Stomatal control of water use in olive tree leaves. Plant and Soil 190(2):179-192. An automatic irrigation controller for fruit tree orchards, based on sap flow… 143 Fernández JE, Palomo MJ, Díaz-Espejo A, Clothier BE, Green SR, Girón IF, Moreno F. 2001. Heat-pulse measurements of sap flow in olives for automating irrigation: tests, root flow and diagnostics of water stress. Agric Water Manage 51(2):99-123. Fernández JE, Palomo MJ, Díaz-Espejo A, Girón IF. 2003. Influence of partial soil wetting on water relation parameters of the olive tree. Agronomie 23(7):545-552. García-Orellana Y, Ruiz-Sánchez MC, Alarcón JJ, Conejero W, Ortuño MF, Nicolás E, Torrecillas A. 2007. Preliminary assessment of the feasibility of using maximum daily trunk shrinkage for irrigation scheduling in lemon trees. Agric Water Manage 89(1-2):167-171. Giorio P, Giorio G. 2003. Sap flow of several olive trees estimated with the heat-pulse technique by continuous monitoring of a single gauge. Environmental and Experimental Botany 49(1):9-20. Goldhamer DA, Fereres E. 2001. Irrigation scheduling protocols using continuously recorded trunk diameter measurements. Irrigation Science 20(3):115-125. Goldhamer DA, Fereres E. 2004. Irrigation scheduling of almond trees with trunk diameter sensors. Irrigation Science 23(1):11-19. Granier A. 1987. Evaluation of transpiration in a Douglas-fir stand by means of sap flow measurements. Tree Physiology 3(4):309. Green SR. 1998. Flow by the heat-pulse method. HortResearch Internal Rep. 1998/22. HortResearch, Palmerston North, New Zealand. Green SR, Clothier B, Jardine B. 2003. Theory and practical application of heat pulse to measure sap flow. Agronomy Journal 95(6):1371-1379. Green SR, McNaughton KG, Clothier BE. 1989. Observations of night-time water use in kiwifruit vines and apple trees. Agricultural and Forest Meteorology 48(3-4):251- 261. 144 Chapter 5 Hatfield JL. 1983. The utilization of thermal infrared radiation measurements from grainsorghum crops as a method of assessing their irrigation requirements. Irrigation Science 3(4):259-268. Holbrook NM, Sinclair TR. 1992. Water-balance in the arborescent palm, sabal-palmetto .2. Transpiration and stem water storage. Plant Cell and Environment 15(4):401- 409. Hoppula KI, Salo TJ. 2007. Tensiometer-based irrigation scheduling in perennial strawberry cultivation. Irrigation Science 25(4):401-409. Intrigliolo DS, Castel JR. 2006. Performance of various water stress indicators for prediction of fruit size response to deficit irrigation in plum. Agric Water Manage 83(1-2):173-180. Intrigliolo DS, Castel JR. 2006. Usefulness of diurnal trunk shrinkage as a water stress indicator in plum trees. Tree Physiology 26(3):303-311. Jackson R. 1982. Canopy temperature and crop water stress. Advances in irrigation 1:43- 85. Jones HG. 1999. Use of infrared thermometry for estimation of stomatal conductance as a possible aid to irrigation scheduling. Agricultural and Forest Meteorology 95(3):139-149. Jones HG. 2004. Irrigation scheduling: advantages and pitfalls of plant-based methods. Journal of Experimental Botany 55(407):2427-2436. Jones HG. 2007. Monitoring plant and soil water status: established and novel methods revisited and their relevance to studies of drought tolerance. Journal of Experimental Botany 58(2):119. Klein I. 2004. Scheduling Automatic Irrigation by Threshold-Set Soil Matric Potential Increases Irrigation Efficiency While Minimizing Plant Stress. Acta Hort (ISHS) 664:361-368. An automatic irrigation controller for fruit tree orchards, based on sap flow… 145 Lobo FD, Oliva MA, Resende M, Lopes NF, Maestri M. 2004. Infrared thermometry to schedule irrigation of common bean. Pesquisa Agropecuaria Brasileira 39(2):113- 121. Luthra SK, Kaledhonkar MJ, Singh OP, Tyagi NK. 1997. Design and development of an auto irrigation system. Agric Water Manage 33(2-3):169-181. Meiresonne L, Nadezhdina N, Cermak J, Van Slycken J, Ceulemans R. 1999. Measured sap flow and simulated transpiration from a poplar stand in Flanders (Belgium). Agricultural and Forest Meteorology 96(4):165-179. Miranda FR, Yoder RE, Wilkerson JB, Odhiambo LO. 2005. An autonomous controller for site-specific management of fixed irrigation systems. Comput Electron Agric 48(3):183-197. Moreno F, Conejero W, Martin-Palomo MJ, Girón IF, Torrecillas A. 2006. Maximum daily trunk shrinkage reference values for irrigation scheduling in olive trees. Agric Water Manage 84(3):290-294. Moreno F, Fernandez JE, Clothier BE, Green SR. 1996. Transpiration and root water uptake by olive trees. Plant and Soil 184(1):85-96. Moriana A, Fereres E. 2002. Plant indicators for scheduling irrigation of young olive trees. Irrigation Science 21(2):83-90. Nadezhdina N, Cermak J. 1997. Automatic control unit for irrigation systems based on sensing the plant water status. Anais do Instituto Superior de Agronomia 46:149- 157. Nadezhdina N. 1999. Sap flow index as an indicator of plant water status. Tree Physiology 19(13):885-891. Nadler A, Raveh E, Yermiyahu U, Green S. 2006. Stress induced water content variations in mango stem by time domain reflectometry. Soil Science Society of America Journal 70(2):510-520.