Riego deficitario controlado del olivar de alta densidad: programación automática basada en la fisiología de la planta.

TESIS DOCTORAL Riego deficitario controlado del olivar de alta densidad: programación automática basada en la fisiología de la planta Memoria que presenta Dña. Carmen María Padilla Díaz Para optar al título de Doctora por la Universidad de Sevilla Sevilla, abril 2018 Universidad de Sevilla Departamento de Cristalografía, Mineralogía y Química Agrícola Programa de Doctorado Recursos Naturales y Medio Ambiente Instituto de Recursos Naturales y Agrobiología de Sevilla (IRNASCSIC) Universidad de Sevilla Departamento de Cristalografía, Mineralogía y Química Agrícola Programa de Doctorado Recursos Naturales y Medio Ambiente Tesis Doctoral Riego deficitario controlado del olivar de alta densidad: programación automática basada en la fisiología de la planta Tesis Doctoral presentada por Dña. Carmen María Padilla Díaz, en satisfacción de los requisitos necesarios para optar al grado de Doctora en Biología, dirigida por Dr. José Enrique Fernández Luque y Dr. María Victoria Cuevas Sánchez (Instituto de Recursos Naturales y Agrobiología de Sevilla) y tutorada por Dr. Isabel González Díez (Dpto. Cristalografía, Mineralogía y Química Agrícola, Universidad de Sevilla). LOS DIRECTORES Dr. José Enrique Fernández Luque Dr. María Victoria Cuevas Sánchez EL TUTOR Dr. Isabel González Díez EL DOCTORANDO Dña. Carmen María Padilla Díaz University of Seville Department of Crystallography, Mineralogy and Agricultural Chemistry Doctoral Program in Natural Resources and Environment Ph.D. Dissertation Regulated deficit irrigation in a super high density olive orchard: automatic scheduling based on the plant physiology Ph.D. Dissertation presented by Carmen María Padilla Díaz to fulfill the necessary requirements of the Doctor of Philosophy degree of Biology under the supervision of Dr. José Enrique Fernández Luque and Dr. María Victoria Cuevas Sánchez (Institute of Natural Resources and Agrobiology of Seville), and being advised by Dr. Isabel González Díez (Department of Crystallography, Mineralogy and Agricultural Chemistry, University of Seville). THESIS SUPERVISORS Dr. José Enrique Fernández Luque Dr. María Victoria Cuevas Sánchez THESIS ADVISOR Dr. Isabel González Díez Ph.D. CANDIDATE Dña. Carmen María Padilla Díaz FACULTAD DE QUÍMICA Departamento de Cristalografía, Mineralogía y Química Agrícola DRA. ISABEL GONZÁLEZ DÍEZ, DIRECTORA DEL PROGRAMA DE DOCTORADO EN RECURSOS NATURALES Y MEDIO AMBIENTE DE LA UNIVERSIDAD DE SEVILLA, Certifica: que la presente Memoria de Investigación titulada “Riego deficitario controlado del olivar de alta densidad: programación automática basada en la fisiología de la planta”, presentada por Dña. Carmen María Padilla Díaz para optar al grado de Doctora por la Universidad de Sevilla, ha sido realizada en el marco del Programa de Doctorado en Recursos Naturales y Medio Ambiente del Departamento de Cristalografía, Mineralogía y Química Agrícola. En Sevilla, a 11 de abril de 2018. OTHER WORKS PUBLISHED DURING THIS PhD De Baerdemaeker N.J.F., Ubeau, M. Szanne Degraeve, S., Padilla-Díaz, C. M., […] Steppe K. 2018. The trident of xylem vulnerability to droughtinduced cavitation: acoustic vulnerability, hydraulic capacitance and xylem anatomy. Tree Physiology. (Under review). Torres-Ruiz J.M., Cochard H. , Choat B., Jansen S. , López R. ,Tomášková I. , Padilla-Díaz C.M. , Badel E., Burlett R., King A., Lenoir N., Martin- StPaul N.K., Delzon S. ( 2017) Xylem resistance to embolism: presenting a simple diagnostic test for the open vessel artefact. New Phytologist. 215: 489–499. Doi: 10.1111/nph.14589 Koenig W.D., Alejano R., Carbonero M.D., Fernández-Rebollo P., Knops J.M.H., Marañón T., Padilla-Díaz C.M., Pearse I.S., Pérez-Ramos I.M., Vázquez-Piqué J., Pesendorfer M.B. (2016) Is the relationship between mast-seeding and weather in oaks related to their life-history or phylogeny? Ecology, 0(0), 2016, pp. 1–13 DOI: 10.1002/ecy.1490 Pérez-Ramos IM, Padilla-Díaz CM, Koenig W, Marañón T (2015) Environmental drivers of mast-seeding in Mediterranean oak species. Does leaf habit matter? Journal of Ecology. 103(3): 691–700. Doi: 10.1111/1365-2745.12400. Marañón T, Padilla-Díaz CM, Pérez-Ramos IM, Villar R (2014) Investigación sobre ecología y gestión de las especies de Quercus frente al cambio global. Ecosistemas 23(2): 124-129 [Mayo-Agosto 2014] Doi.: 10.7818/ECOS.2014.23-2.16 Pérez-Ramos IM, Aponte C, García LV, Padilla-Díaz CM, Marañón T (2014) Why is seed production so variable among individuals? A ten-year study with oaks. PLoS ONE 9(12): e115371. doi:10.1371/journal.pone.0115371 SCIENTIFIC AWARDS DERIVED FROM THIS Ph.D. THESIS Egea G, Padilla-Díaz CM, Martínez J, Fernández JE, Pérez-Ruiz M. Scientific Award Accesit, IV Premio José Humanes para el Desarrollo del Sector Oleícola y del Aceite [IV Award José Humanes for the development of the olive oil sector and industry] bestowed by Editorial Agrícola 2017, 23rd February 2017, Seville (Spain). Title of the award “Desarrollo de una metodología de Apoyo al Riego de Precisión del Olivar, basada en el Uso de la Termometría Infrarroja y Drones Agrícolas” [Development of a Methodology to Support Precision Irrigation of the Olive Orchard, based on the Use of Infrared Thermometry and Agricultural Drones]. Our work, developed during my PhD, was selected as the best cutting-edge technique for its application in a close future in irrigation scheduling of olive orchards. Photographic Award "images for Science", ASECIC award. Title of photography "Haz de luz para el agua" [Light beam for water]. University of Cordoba (Spain). 6th April, 2015 (Scientific Photography). My awarded photography consisted in a microphotography of a xylem cross section of the olive tree taken with a light microscope during my short research-stay at University of Bordeaux (France). He reducido el mundo a mi jardín, y ahora veo la inmensidad de todo lo que existe. José Ortega y Gasset A mi familia CONTENTS Chapter 1: Summary, Resumen, Introduction and Objectives ..................... 1 1. Summary ..................................................................................................... 3 2. Resumen ...................................................................................................... 7 3. Introduction ............................................................................................... 10 4. Objectives .................................................................................................. 12 Chapter 2: Scheduling regulated deficit irrigation in a hedgerow olive orchard from leaf turgor pressure related measurements .......................... 13 1. Introduction ............................................................................................... 15 2. Materials and methods Materials and methods ........................................ 17 2.1. Orchard characteristics and irrigation management ........................... 17 2.2. Soil, plant and weather measurements ............................................... 18 2.3. Statistical analysis .............................................................................. 20 3. Results ....................................................................................................... 20 4. Discussion ................................................................................................. 29 5. Conclusions ............................................................................................... 31 Chapter 3: Water status, gas exchange and crop performance in a super high density olive orchard under deficit irrigation scheduled from leaf turgor measurements ...................................................................................... 33 1. Introduction ................................................................................................. 3 2. Materials and methods .............................................................................. 37 2.1. Orchard characteristics and irrigation management ........................... 37 2.2. Soil, plant and weather measurements ............................................... 39 2.3. Growth and production measurements ............................................... 40 2.4. Statistical analysis .............................................................................. 40 3. Results ....................................................................................................... 41 3.1. Water supplies, atmospheric demand and soil water content ............. 41 3.2. Plant water status versus leaf turgor related measurements ............... 45 3.3. Gas exchange, growth and production ............................................... 49 4. Discussion ................................................................................................. 55 4.1. Impact of the scheduling irrigation approach on crop physiology and crop performance ............................................................................... 55 4.2. Suitability of the scheduling irrigation approach for commercial orchards ..................................................................................................... 57 5. Conclusions ............................................................................................... 58 Chapter 4: Effect of fruit development on turgor related measurements in olive leaves ........................................................................................................ 61 1. Introduction ............................................................................................... 63 2. Materials and methods .............................................................................. 65 2.1. Orchard characteristics and irrigation management ........................... 65 2.2. Experimental setup and maximum turgor related measurements ....... 66 2.3. Leaf-water relations ............................................................................ 67 2.4. Pressure-volume curves ..................................................................... 67 2.5. Leaf gas exchange measurements....................................................... 68 2.6. Fruit dry weight and colour index ...................................................... 68 2.7. Statistical analysis .............................................................................. 69 3. Results ....................................................................................................... 70 3.1. Maximum turgor related measurements and the influence of fruit presence ..................................................................................................... 70 3.2. Relationship between maximum turgor and environmental variables 70 3.3. Relationship between of maximum turgor and fruit development ..... 73 3.4. Leaf-water relations and the influence of fruit presence .................... 76 4. Discussion ................................................................................................. 79 4.1. Daily variations on maximum turgor: environmental components .... 79 4.2. Seasonal variations on maximum turgor: fruit influences .................. 80 4.3. Future challenges ................................................................................ 82 5. Conclusions .............................................................................................. 82 6. Supplementary material ............................................................................. 83 Chapter 5: Leaf turgor under drought stress and leaf growth in olive ...... 87 1. Introduction ............................................................................................... 89 2. Materials and methods .............................................................................. 91 2.1. Plant material and experimental setup ................................................ 91 2.2. Soil and meteorological measurements .............................................. 91 Chapter 1 5 daily Pp curve. This finding opens new possibilities for improving irrigation scheduling in periods when the olive tree is less sensitive to drought stress. This fills a gap previously identified by our own results, on the reliability of using ZIM records to schedule irrigation in periods of the growing cycle when the 45RDI strategy advices for reducing water supplies below the crop water requirements. Introduction 6 Chapter 1 7 2. RESUMEN La distribución geográfica del olivo (Olea europaea L.) se encuentra principalmente en áreas con clima Mediterráneo donde áreas áridas y semiáridas son bastante frecuentes. A pesar de que el olivo es una especie que está bien adaptada a condiciones de escasez hídrica, su respuesta ante aportaciones de riego es destacable. Esto explica por qué ca. 20% de los olivares cultivados son de regadío. Del mismo modo que ocurre para muchas otras especies frutales, son cada vez más frecuentes los marcos de plantación con una densidad de árboles elevada. Estos cambios se justifican por el valor económico de los productos derivados del olivo. Ambas estrategias agronómicas contribuyen a hacer frente al reto de obtener una mayor producción de aceitunas y aceite de oliva de cara al aumento de la población mundial. Esto es particularmente importante en las áreas de cultivo del olivo, en las cuales se prevé un incremento de las temperaturas junto a una disminución en las precipitaciones. De hecho, el incremento de la productividad del agua por unidad de área cultivada es uno de los mayores retos a los que se enfrentan los olivicultores. En este contexto, una de las respuestas más efectivas de la comunidad científica ha sido el desarrollo de nuevas estrategias de riego y de herramientas fiables tanto para la programación del riego como para la monitorización del estrés hídrico en plantas. Los olivares en seto con alta densidad de plantas (a partir de 1.500 árboles ha-1), también llamados olivares en seto de alta densidad, son considerados como los más productivos entre los distintos tipos de manejo que existen en el cultivo del olivar. Por otro lado, precisamente porque son de elevada densidad, se requieren aportes de riego para que estas plantaciones tengan una rentabilidad aceptable. Una de las estrategias de riego más prometedoras para este tipo de plantaciones es la aplicación del riego deficitario controlado (RDC), que permite ahorros de agua considerables al mismo tiempo que se evitan episodios de estrés hídrico excesivo en los momentos en los que el cultivo es más sensible a la sequía. En 2014 y 2015, los dos primeros años de esta tesis doctoral, nos centramos en el estudio de la idoneidad de la estrategia de RDC aplicando un 45% de las necesidades de riego del cultivo (45RDC; 45RDI a lo largo de esta tesis en inglés), estrategia especialmente diseñada para olivares en seto de alta densidad, basada en la forma de las curvas diarias registradas con las sondas ZIM. Estas curvas están relacionadas con la presión de turgencia de la hoja. Trabajamos en una finca comercial y totalmente productiva de la variedad Arbequina con 1667 árboles ha-1. Además de este tratamiento consistente en la aplicación de la mencionada estrategia 45RDC, tuvimos un tratamiento 45RDC programado mediante el coeficiente del cultivo junto a otro tratamiento donde las plantas se encontraban bien regadas (FI, del inglés full irrigated), el cual consistió en un riego diario que reemplazaba las necesidades de hídricas del cultivo. Nuestros resultados muestran que fuimos capaces de programar el riego tan solo con el análisis visual de las curvas derivadas de las lecturas de las sondas ZIM (i.e. valores de Pp) sin ningún tipo de tratamiento de datos posterior. Una comparativa de nuestra aproximación con la aproximación del método del coeficiente del cultivo mostró que, tanto el estado hídrico de Introduction 8 la planta, como valores máximos diarios de intercambio gaseoso, número de entrenudos en los ramos de años, área foliar, acumulación de aceite en los frutos, crecimiento y producción tanto de aceitunas como de aceite fueron similares entre tratamientos. Con nuestra estrategia 45RDC obtuvimos ahorros de agua de más del 50%, con un descartable impacto en la producción comparado con el tratamiento FI. Sin embargo, se debe tener precaución al extrapolar nuestros resultados, ya que hay evidencias en la literatura de que las relaciones entre el estado hídrico de la planta y la forma de las curvas diarias registradas por las sondas ZIM dependen de la variedad del cultivo y de las condiciones de la plantación. En los dos segundos años, 2016 y 2017, llevamos a cabo dos experimentos para profundizar en el conocimiento sobre las interacciones que se dan entre las condiciones ambientales y los mecanismos de las plantas que influyen en la turgencia y que están relacionados con las medidas de las sondas ZIM. Nuestro propósito fue desentrañar las respuestas basadas en la planta de las lecturas de las sondas ZIM, con el propósito de derivar nuevos indicadores para la programación del riego desde dichas lecturas. Exploramos la idoneidad del uso de la máxima turgencia (Ppmin) para la programación del riego, como un indicador fiable que pudiera ser interpretado independientemente de la variedad de cultivo y de las condiciones de cada finca. El experimento se llevó a cabo en la finca mencionada anteriormente, pero únicamente bajo el tratamiento FI. Básicamente, exploramos la influencia de la proximidad de frutos a hojas instrumentadas con las sondas ZIM, así como también el efecto del envejecimiento de la hoja, la disponibilidad de agua en el suelo y el déficit de presión de vapor. Nuestros datos mostraron que no hubo una influencia en la proximidad de frutos sobre las lecturas de Ppmin sugiriendo que éstas podrían ser usadas para cualquier hoja independientemente de si está cerca o lejos de un racimo de frutos. Además, el incremento del módulo de elasticidad con la edad de la hoja tampoco tuvo ninguna influencia en la lectura de Ppmin. Sin embargo, sí hubo un claro efecto de las dos principales variables motoras de la transpiración. De este modo, los patrones diarios de Ppmin estuvieron principalmente influenciados por la demanda atmosférica seguidos, en menor medida, de la disponibilidad hídrica en el suelo. El patrón de Ppmin cambió dependiendo del estadío de desarrollo del fruto. Se encontraron dos fases del desarrollo que tuvieron una influencia más evidente: la tasa máxima de endurecimiento del hueso y el de rápido crecimiento del fruto que se produce a fínales de la segunda mitad del verano. Nuestros datos apoyan al Ppmin como un indicador sensible y fiable del estrés hídrico en aquellos periodos en los que el estado hídrico de la planta puede tener una marcada influencia en la producción del fruto. Los sensores basados en medidas directas sobre la planta (plant-based methods, en inglés) tienen la ventaja de usar la propia planta como un biosensor, es decir, con sus registros informan de la respuesta de la planta al estado hídrico atmosférico y del suelo al mismo tiempo. Por tanto, estos sensores son particularmente útiles en la programación del riego. No obstante, los registros no son siempre fáciles de interpretar debido a la complejidad de las relaciones suelo-agua-planta-atmósfera y a la respuesta adaptativa de las especies al estrés hídrico. Así mismo, se necesita una mejor comprensión del Chapter 1 9 significado real de las medidas en planta para evaluar el estrés hídrico y así aumentar la aceptación del uso de este tipo de medidas para la gestión del riego en fincas comerciales El cuarto experimento de esta tesis doctoral se llevó a cabo con este propósito. Para ello usamos plantones de dos años de edad cultivados en invernadero, imitando las condiciones típicas de áreas mediterráneas. Se instalaron sensores ZIM y sensores de grosor de hoja en plantas representativas con el objeto de determinar los valores umbrales de las principales variables ambientales hídricas como motores de la turgencia y del grosor de la hoja. En nuestros experimentos, consideramos plantas sometidas a estrés hídrico con ciclos de recuperación y plantas bien regadas como control. También, estudiamos los cambios en las propiedades hídricas de la pared celular de la hoja a partir de curvas de presión-volumen. Los procesos hidráulicos que envuelven la combinación de la turgencia de la hoja, el grosor de la misma y el estado hídrico de la planta a lo largo del continuo suelo-planta- atmósfera dependieron de la capacidad de campo, siendo esta última un umbral de la disponibilidad hídrica del suelo y, en menor medida, el déficit de presión de vapor del aire. Bajo estrés hídrico severo, las propiedades de las paredes celulares de la hoja cambiaron y tuvo lugar la inversión de la curva Pp diaria. Esta última, atribuida a la acumulación de aire en los tejidos de la hoja, presentó una dinámica similar a las medidas del grosor de la hoja. Esto sugiere que la sonda ZIM mide una variable relacionada con el grosor de la hoja durante la inversión de la curva diaria de Pp. Estos resultados, por tanto, abren nuevas posibilidades para la mejora en la programación del riego en los periodos en los que el olivo es menos sensible al estrés hídrico. Esto, resuelve las carencias en el conocimiento previamente identificadas en nuestros resultados, con la fiabilidad de usar los registros por las sondas ZIM para la programación del riego a lo largo de todo el periodo de crecimiento del olivo siguiendo las recomendaciones de la estrategia 45RDC, reduciendo las cantidades de riego aplicadas por debajo de las necesidades hídricas del cultivo. Introduction 10 3. INTRODUCTION It is widely recognized that the olive tree (Olea europaea L.) is one the most symbolic species of the Mediterranean Basin due to its ecological, cultural, and economical relevance (Kaniewski et al., 2012). The native or wild distribution of olive is concentred around Mediterranean Basin. In the eastern of Mediterranean Basin, about 6,000 years ago the domestication or cultivation of the olive tree (Zohary et al.; 2012) was being developed and spreading either from Cyprus, Anatolia, Crete and Egypt towards the western of Mediterranean by Phoenicians, Greeks and later by Romans (Baldoni et al., 2006; Breton et al., 2006). Nowadays, Spain, Tunisia, Italy and Morocco are the largest areas in the world dedicated for its growing (Data from FAO, 2016; http://faostat.fao.org) and also, most of the total olive oil is being produced in such areas. However, currently it is widely spread around the world too but particularly in areas with Mediterranean climate, i.e. mild winters with torrential precipitations and hottest summers with scant and irregular precipitations, as the best growing conditions for the olive. Although olive trees are well adapted to Mediterranean drought conditions (Fernández 2014a), their response to water supply is remarkable. Nowadays there is an ongoing replacement of the traditional rainfed olive-groves by orchards with irrigation systems as well as a higher number of trees per hectare. Arguably, those changes were due to modernisation of agricultural machinery and are justified by the high economic value of the olive products. These explain that ca. 20% of the olive cropped area is irrigated. In addition, the world population is expected to increase 1:3 before 2050 (FAO 2009). In this period, rising temperatures and lower precipitation is forecasted for Mediterranean areas (IPCC, 2014). Both agronomical approaches (i.e. irrigation and high plant densities) have the advantage of contributing to facing the challenge of producing olive fruits and olive oil for the increasing global population. This is particularly important in olive growing areas, for which rising temperatures and lower precipitations are forecasted. Increasing crop water productivity per unit of cropped land is, in fact, a main challenge to olive growers. In this context, one of the most effective responses of the scientific community is the development of new irrigation strategies and reliable tools for monitoring plant water stress and scheduling irrigation (Iniesta et al., 2009; Martín- Vertedor et al., 2011; Gomez-del-Campo, 2013; Fernández 2014a). Among the tools for monitoring the plant water stress are plant-based sensors which also are widely used for irrigation scheduling. Plant-based sensors are located directly on the plant and have the advantage of using the tree as a biosensor. i.e. their records inform on the plant response to the soil and atmospheric water status. Thus, those sensors are particularly useful for irrigation scheduling. However, their records are not always easy to interpret, due to both the complexity of the soil-water-plant-atmosphere relationships and the adaptive responses of the species to water stress. Therefore, a greater understanding of the actual meaning of plant-based measurements for assessing water stress is needed, to increase the acceptance of plant-based measurements in commercial orchards. There is needed to carry on experiments under controlled climatic conditions Chapter 1 11 which allow us to comprehend the physiological mechanisms that are under the signals of the sensors for its implementation in the olive orchards. The most suitable irrigation strategies for arid and semi-arid areas are deficit irrigation (DI) strategies designed for the crop and environment conditions. These are widely used in olive, a typical crop in these areas (Rallo et al. 2016). Hedgerow olive orchards with high plant densities (from 1.500 trees ha-1), or super high density (SHD) olive orchards, are considered to be among the most productive management systems for olive. On the other hand, irrigation supplies are required for an acceptable profitability in those orchards, precisely because of the high plant densities. To control the frequency and irrigation amounts, i.e. irrigation strategies, according to plant water needs in SHD olive orchards is needed to extend the timespan at the fully productive and also to enable water savings (Fernandez 2014b). Deficit irrigation is specially required in hedgerow orchards with high plant densities, or super high density (SHD) orchards, where excessive growth must be avoided by controlling water and fertilizer supplies. For those orchards, regulated deficit irrigation (RDI) has been reported as one of the best DI strategies (Fernández et al. 2013; Fernández 2014b). After several experiments at the Sanabria orchard, a commercial SHD ‘Arbequina’ orchard near Seville, southwest Spain (Fernández et al., 2013, 2017), our group recommends the 45RDI strategy for SHD olive orchards in the area, and in many other olive growing areas of similar characteristics. The 45RDI strategy consists on irrigating daily with to replace 100% of irrigation needs (IN) in three periods of the annual growing cycle. These periods, named as Periods 1, 2 and 3, are those in which olive is most sensitive to water stress. In between those periods, one or two irrigation events per week are applied, with a total water supply, for the considered period, from 15 to 30% of IN, depending of the time of the year. The total water supply, for the whole season, in this 45RDI strategy is aimed at 45% of IN. The successful application of any RDI strategy requires effective irrigation scheduling. In our group we have tested different water stress indicators and related systems to schedule irrigation, based on sap flow, trunk diameter variations and leaf turgor related measurements (Fernández 2014a). Among them, the ZIM system, which uses the leaf patch clamp pressure probe (LPCP probe) (Zimmermann et al., 2008) has been proved to be useful to schedule the 45RDI irrigation strategy in our SHD olive orchard (Fernández et al., 2011; Fernández 2014a). To these reasons the main purpose of this PhD was to validate of the use of the indicators for irrigation scheduling derived from the turgor related measurements in a commercial olive orchard under a 45RDI along two years of experiment that will be described in the Chapters 2 and 3. Moreover, to getting deeper in the fundamental understanding of the interaction between the environmental and plant mechanisms that drive turgor measurements to implement new indicators for irrigation scheduling will be further developed in the Chapters 4 and 5. 12 4. OBJECTIVES - To validate and corroborate the usefulness of the methodology based on ZIM system of two years of the irrigation scheduling - To test the influence among two years in plant water status (water potential, stomatal conductance and net photosynthesis) as well as vegetative growth (number of internodes and leaf area) and production. We found that no differences between 45RDICC and 45RDITP for any variable studied of both years. - To unravel the role of the proximity of fruits to the leaf on the maximum daily turgor. - Due to the complex relations between leaves and fruits, acting this last one as sinks and sources of water and carbon, and their potential effect on the maximum turgor, the aim was to test the influence of the location of the fruit as well as its development on the leaf monitored by ZIM probe - To determine, from the combined use of ZIM probes and LS sensors, the threshold level of leaf turgor that is driving leaf thickness growth and changes in cell wall properties using pressure-volume curves under the influence of drought stress. - To determine the threshold levels of water-environmental drivers, and the impact of drought stress, of leaf turgor, and leaf turgor as a driver of leaf growth. Chapter 2 Scheduling regulated deficit irrigation in a hedgerow olive orchard from leaf turgor pressure related measurements Chapter 2 21 values were 93.1 mm in period 2 and 69.7 mm in period 3. Differences in IA between treatments had little impact on REW values in those two periods (Fig. 2D). Consequently, similar water stress levels were found in trees of all treatments, for both period 2 and 3, as Ψpd (Figs. 3A,C) and Ψstem (Figs. 3B,D) values show. In between those periods the trees’ water stress increased considerably in the 45RDI treatments, as expected. After period 3 the autumn rainfall (Figs. 2A,B) contributed to keeping low values of water stress (Fig. 3). Fig. 2. Seasonal courses of both the irrigation amounts (IA) supplied to each treatment and the precipitation (P) collected in the orchard (A,B), and the values of relative extractable water (REW) derived from the soil water contents (avg – SE) measured in the plots of each treatment (C,D). Measurements were made on the irrigation seasons of 2013 and 2014. Different letters indicate significant differences between treatments, at p < 0.05. Letters are not shown when no differences were found. DOY = day of year. 120 150 180 210 240 270 300 0.0 0.2 0.4 0.6 0.8 1.0 1.2 120 150 180 210 240 270 300 0 1 2 3 4 0 20 40 60 IA (mm) 0 1 2 3 4 5 P (mm) 0 20 40 60 DOY (DOY 120 = April 30) REW 0.0 0.2 0.4 0.6 0.8 1.0 1.2 A B C D 2014 2014 2013 2013 a b b a a b a b b a b b a b b a b b a b b a b a b a b a b a b FI 45RDICC 45RDITP a b b Jun Jul Aug Sep Oct NovMay Jun Jul Aug Sep Oct NovMay Testing the scheduling RDI from leaf turgor pressure 22 Fig. 3. Seasonal courses of predawn water potential (Ψpd, avg ± SE) and midday stem water potential (Ψstem, avg ± SE) measured in 2013 (A, B) and 2014 (C, D) in FI, 45RDIcc and 45RDITP trees. Different letters indicate significant differences between treatments at p < 0.05. Letters are not shown when no differences were found. DOY = Day of year. Since our irrigation scheduling approach was based on outputs of the ZIM probe, we wanted to compare those outputs with measurements with the Scholander-type pressure chamber, a widely used instrument to monitor olive water stress (Moriana et al., 2012; Naor et al., 2013). In Fig. 4 we show data from both methods, collected in May and September, i.e. before and after the highly demanding mid-summer period. The shown Pp curves correspond to a 45RDCC tree (Fig. 4A) and a FI tree (Fig. 4B). All the other instrumented trees showed a similar behaviour. Values of Ψpd and Ψstem showed that trees of both treatments had similar water stress levels in May and in September. Values of Pp, however, were greater in September than in May, for both the 45RDICC and the FI trees. Results in Fig. 4, therefore, suggest that water stress monitoring should not rely on absolute Pp values, at least for our orchard conditions. DOY (DOY 150 = May 30) 150 180 210 240 270 300 stem (MPa) -5 -4 -3 -2 -1 0 pd (MPa) -4 -3 -2 -1 0 2013 150 180 210 240 270 300 FI 45RDICC 45RDITP a b b a b b a b b a b b a b b a b b a a b a b b a ab b 2014 a b b A B C D a b a b a b a b Jun Jul Aug Sep Oct Jun Jul Aug Sep Oct Chapter 2 23 Fig. 4. Daily curves of the ZIM probe outputs (Pp) installed in a representative tree of the 45RDICC (A) and FI (B) treatments applied in 2013. The shown curves correspond to days prior (May) and after (September) the most demanding, in terms of water stress, midsummer period. Also shown are predawn (Ψpd, avg ± SE, triangles) and midday stem water potential (Ψstem, avg ± SE, circles) values measured with a Scholander-type pressure chamber in trees next to the trees instrumented with ZIM probes, also in May and September. Both for the curves and symbols, the grey and black colours mean measurements in May and the September, respectively. 0 60 120 180 May 19 May 20 May 21 Sept 11 Sept 12 Sept 13 pd, stem (MPa) -1.5 -1.0 -0.5 0.0 45RDIcc Date 2013 0 60 120 180 May 19 May 20 May 21 Sept 11 Sept 12 Sept 13 -1.0 -0.5 0.0 FI Pp (kPa) A B Testing the scheduling RDI from leaf turgor pressure 24 Our irrigation scheduling approach, however, relies on the State shown by the Pp curves, and not on absolute values. As mentioned in the Introduction, Fernández et al. (2011) and Ehrenberger et al. (2012) found that State 1 was observed in olive trees with Ψstem > −1.2 MPa, Sate 2 in trees with −1.2 MPa < Ψstem < −1.7 MPa and State 3 in trees with Ψstem < −1.7 MPa. But they both worked with young trees only. To test the suitability of our irrigation scheduling approach for olive trees of any age, from young to fully mature, highly productive trees, we made Fig. 5. In this figure Ψstem values measured in trees of all treatments are plotted against the State shown by the Pp curves collected on the same days. The figure shows the data we collected in 2013 and 2014, and also data collected by our team from 2010 in the same orchard, when we began this set of studies on scheduling regulated deficit irrigation in super-high-density olive orchards. Including data from our previous work allows for a more robust assessment on the relation between Ψstem values and the State shown by Pp curves. Measurement details for the 2010-2013 period can be seen in Fernández et al. (2011, 2013) and Diaz-Espejo et al. (2012). Figure 5 shows data collected in 4 to 8-year-old trees, under a wide range of both soil water conditions and atmospheric demand. In the 68.3% of the cases in which Pp curves were in State 1 we found Ψstem values to be > −1.2 MPa, and in the 81.8% of the cases in which Pp curves were in State 3, Ψstem < −1.7 MPa. Thus, both for States 1 and 3 we got similar results, in most cases, than those reported by Fernández et al. (2011) and Ehrenberger et al. (2012). However, about one third only of the trees in which Pp records showed State 2 had Ψstem values in between −1.2 and −1.7 MPa, as previously reported by those authors. This lack of agreement can be explained, at least in part, by State 2 being not always easy to identify. Both States 1 and 3 can be clearly identified from the shape of the Pp curves. However, at moderate levels of water stress typical of State 2, decreases in the Pp values collected at the central hours of the day were highly variable, being not always easy to identify whether the shape of the Pp curve suggested State 2 or it was just noise caused by changing atmospheric conditions. Changes among States observed in 2014 in each of the trees instruments with ZIM probes, as well as the average Ψstem values for each treatment, are shown in Fig. 6. This figure illustrates, in fact, the tree-to-tree variability of the State shown by the Pp curves collected in our orchard. Trees of the FI treatment always showed State 1, except for days of sudden increase in ETo, such as DOY 242-244. The 45RDITP trees showed State 1 for most days of periods 2 and 3. An exception was at the beginning of period 3, when the available water in the soil was very low after the mid-summer period of reduced irrigation. State 2 was also observed on some days of Period 2, likely because of the reduced IA, which amounted to 0.68% of IN only (Fig. 7). In between periods 2 and 3, all 45RDITP trees showed State 2 or 3, as expected. In the autumn irrigation was also reduced, but the total water supplied by irrigation and precipitation (Fig. 2B) was enough to avoid severe water stress, as commented when reporting findings shown in Fig. 3. The 45RDICC trees showed a similar behaviour than the 45RDITP trees, with the difference of a lower recovery from water stress after the beginning of both periods 2 and 3. This can be explained by the fact that 45RDITP trees were irrigated with IA = 120% IN on the first three days of each period, as explained in Section 2.1. Chapter 2 25 Fig. 5. Midday stem water potential (Ψstem) values measured in representative trees of all the irrigation treatments at the Sanabria orchard during the irrigation seasons of 2010 to 2014. Each data point corresponds to a single measurement with a Scholander-type pressure chamber. Dashed lines represent the midday stem water potential values values identified by Fernández et al. (2011a) and Ehrenberger et al. (2012) as typical of State 1 (Ψstem > -1.2 MPa) (A), State 2 (-1.2 MPa > Ψstem > -1.7 MPa) (B) and State 3 (Ψstem < -1.7 MPa) (C). The State is given by the shape of the daily leaf patch clamp pressure curve, as described by Ehrenberger et al. (2012). DOY = Day of year. State 1 -5.0 -4.0 -3.0 -2.0 -1.0 0.0 2.5 % 29.2 % 68.3 % stem (MPa) State 2 -5.0 -4.0 -3.0 -2.0 -1.0 43.8 % 32.8 % 23.4 % State 3 DOY 2010-2014 (DOY 120 = April 28, except for 2012 in which DOY 120 = April 29) 120 160 200 240 280 320 -5.0 -4.0 -3.0 -2.0 -1.0 1.1 % 81.8 % 17.0 % A B C Testing the scheduling RDI from leaf turgor pressure 26 160 180 200 220 240 260 280 300 320 -6 -5 -4 -3 -2 -1 0 FI 45RDICC 45RDITP State 1 State 3State 2 DOY (160 = June 9th) 160 180 200 220 240 260 280 300 320 stem (MPa) -6 -5 -4 -3 -2 -1 0 FI CC PB 45RDICC 45RDITP Period 2 Period 3 DOY 2014 (DOY 160 = June 9th) Y stem (MPa) Fig. 6. Seasonal courses of midday stem water potential (Ψstem, avg. ± SE) measured in FI, 45RDICC and 45RDITP trees of the Sanabria orchard during the whole irrigation season of 2014. Pp curves were collected by three ZIM probes per treatment and shown here through State 1 (low water stress), State 2 (moderate stress) and State 3 (severe stress) by horizontal colour bars. Periods 2 and 3 shown in Fig. 1, as observed in this year 2014, are represented. DOY = Day of year. Chapter 2 27 Figure 7 shows the results from applying our irrigation approach to schedule the 45RDITP treatment. Figures 7A and 7C show the Pp curves collected in a 45RDITP tree, for periods 2 and 3, respectively. We chose the tree in which State 2 appeared earlier. The other two trees showed similar behaviour, although signs of water stress appear one or two days later than in the tree used for Fig. 7. A certain tree-to-tree variability in fruit tree orchards can be expected, due to spatio-temporal variations of soil and plant conditions (Fernández and Cuevas, 2010). Arrows in Fig. 7 show the days on which we increased or decreased irrigation according to the State shown by the Pp curve. Figures 7B (period 2) and 7D (period 3) show differences in the daily IA values when estimated with the crop coefficient approach (45RDICC) as compared to our irrigation scheduling approach based on the ZIM system (45RDITP). Also shown are the total IA applied for both treatments in each period, expressed as a percentage of the IN calculated for the period. These data show that in period 2 the amount of water supplied by irrigation was lower in 45RDITP than in 45RDICC (Fig. 7B). In period 3, when irrigation demands were lower, IA values in 45RDITP and in 45RDICC were similar (Fig. 7D). Fig. 7. Time courses of the Pp values recorded on a 45RDITP representative tree in period 2 (A) and period 3 (C) of the 2014 irrigation season (see Fig. 1 to identify the periods). Also shown are the irrigation amounts (IA) supplied to each treatment in each period (B,D), expressed as a fraction of the calculated irrigation needs (IN) for the period. IA was increased (arrows up) or decreased (arrows down) according to changes in the State of the daily Pp curves (see the irrigation approach described in Section 2.1). DOY = Day of year. FI 45RDICC 45RDITP DOY 2014 (DOY 154 = June 3) 156 159 162 165 168 171 174 177 180 183 IA (L tree-1 day-1) 0 5 10 15 20 FI 45RDICC 45RDITP Pp (kPa) 0 30 60 90 120 150 DOY 2014 (DOY 238 = August 26) 240 243 246 249 252 255 258 0 10 20 30 0 30 60 90 120 A B Pp (kPa) C D In 45RDICC, IA = 0.81 IN In 45RDITP, IA = 0.68 IN In 45RDICC, IA = 0.99 IN In 45RDITP, IA = 1 IN IA (L tree-1 day-1) Testing the scheduling RDI from leaf turgor pressure 28 The crop response to the irrigation treatments in terms of fruit yield and irrigation water productivity (WP) is shown in Table 2. The trees were already fully productive in 2013 and 2014, as expected for their age (7 and 8 years old, respectively). Considering data of both years, and averaging results from the 45RDICC and 45RDITP treatments, the 45RDI trees, which received 53.9% of the total IA supplied to the FI trees (Table 1), had a fruit yield of 72.4% of that in FI (Table 2). Values of WP increased with RDI in 30.2%, on average. For 2013, however, WP data are not reliable, because of the problem with the irrigation pump we had at the beginning of the 2013 irrigation season (Section 3). Data from 2014 show that differences in fruit yield between 45RDITP and 45RDICC were not significant (p = 0.54 and p = 0.22, respectively). Still, data of one year only is not enough to evaluate the impact of the irrigation treatment on production. Table 1. Water supplies (IA = irrigation amounts; P = precipitation) and potential evapotranspiration (ETo) in the experimental orchard for the two irrigation seasons. All values are in millimeters. Values of IA are also expressed as percentages of irrigation needs (% IN). DOY = day of year. Data with * were affected by reduced water supply at the beginning of the irrigation season (see Section 3 for details). 2013 2014 Whole Irrigation period Whole Irrigation period year (DOY 133–301) year (DOY 116–324) ETo 1464.3 1000.3 1437.6 1110.1 P 476.4 112.5 549.4 317.6 IA in FI 367.6 (80.0% IN) 462.9 (87.1% IN) IA in 45RDICC 197.9 (43.1% IN) 238.2 (44.8% IN) IA in 45RDITP 235.9 (44.4% IN) Chapter 2 29 Table 2. Fruit yield (n = 4) and water productivity values for each treatment and experimental year. Different letters indicate significant differences between treatments at p < 0.05. Year Treatment Fruit yield (kg ha-1) Irrigation water productivity (kg ha-1 mm-1) 2013 FI 22559.8 ± 1453.1 a 61.4* ± 3.9 a 45RDICC 14952.1 ± 873.0 b 75.6* ± 4.4 b 2014 FI 19283.0 ± 2708.5 a 41.6 ± 5.9 a 45RDICC 13443.0 ± 2847.9 a 53.4 ± 12.0 a 45RDITP 17025.6 ± 2077.0 a 72.2 ± 8.8 a 4. DISCUSSION As mentioned above, our group began in 2010 a set of studies to identify both a suitable RDI strategy for hedgerow olive orchards with high tree densities (SHD olive orchards) and a reliable, user-friendly water stress indicator to schedule irrigation. In a first set of experiments made from 2010 to 2102, Fernández et al. (2013) evaluated the impact on crop performance of an earlier version of the RDI strategy, with two irrigation levels (30% and 60% of IN). Experiments were run in parallel to assess the performance of different water stress indicators, from the conventional leaf and stem water potential, and stomatal conductance, to new plant-based methods for automatic and continuous monitoring of water stress. These experiments with concomitant measurements of a wide range of variables related to the water status in the soil, plant and surrounding atmosphere, provided insight into the links between physiological processes in olive trees under water stress and outputs from sap flow, trunk diameter variations and leaf turgor related measurements (Fernández et al., 2011; Diaz-Espejo et al., 2012; Rodriguez-Dominguez et al., 2012; Cuevas et al., 2013). Those findings, together with contributions from other authors on the usefulness of those plant-based sensors to monitor water stress in fruit trees, allowed for detailed assessments on the potential of each method for monitoring water stress and schedule irrigation. The work by Fernández et al. (2008), Ramos and Santos (2009) and Rousseaux et al. (2009) show, to a good extent, the advantages and disadvantages of sap flow measurements to improve water management in olive and other fruit trees. The same can be said for trunk diameter variations on the work by Pérez-López et al. (2008), Moriana et al. (2010) and Cuevas et al. (2010). And the potential of using leaf turgor related measurements with that purpose was evaluated by Ben-Gal et al. (2010), Rüger et al. (2010) and Zimmermann et al. (2010, 2013). Knowledge from these and other Testing the scheduling RDI from leaf turgor pressure 30 publications was collected by Fernández (2014b) in a review on the applicability of those methods to schedule irrigation in commercial orchards. He concluded that the ZIM system was one of the most promising systems for commercial olive orchards. The system is easier to install and use than those of sap flow and trunk diameter variation, and as robust as those two when working under field conditions for the long irrigation seasons common in most olive growing areas. On the outputs, Fernández (2014b) showed that both sap flow and trunk diameter related measurements require high training both to process the collected data and to understand their physiological meaning. For the leaf turgor related measurements, he stated that, although we are still far from fully understanding the physiological meaning of the ZIM probe readings, there was a potential for scheduling irrigation based just on the visual analysis of the Pp daily curves. This is crucial for the acceptance of any method to schedule irrigation by farmers and orchardists without specific training, as previously stated by Naor (2006 and Fernández and Cuevas (2010), among others. Our results suggest that we cannot expect a robust correlation between the tree water status and Pp values for the whole irrigation season (Fig. 4). This is not surprising, since aging induces structural and mechanical changes in the olive leaf that could easily affect the outputs of the ZIM probes. Thus, the water stress history of the leaf can affect the palisade parenchyma (Chartzoulakis et al., 1999; Bacelar et al., 2004), as well as the density and thickness of the leaf (Centritto, 2002). The seasonal course of the Ψstem vs. Pp relationship can also be affected by changes in the elastic modulus (ε) of the leaf cells. It has been observed that, in olive, ε tends to increase with leaf age (Bongi and Palliotti, 1994) and drought (Dichio et al., 2003). All these changes, together with others on leaf response to environmental stimuli (Marchi et al., 2008) can affect the Ψstem vs. Pp relationship along the season. The fact that, at least for olive, the Ψstem vs. Pp relationship changes with time must be taken into account when Pp values are used to derive water stress indices requiring normalization, as for the case described by Bramley et al. (2013). Figure 4, in fact, suggests that establishing an effective normalization procedure of the Pp records is not straightforward. Our irrigation approach (Section 2.1), however, does not rely on absolute Pp values, but on changes between States, which it does not require normalization. Our hypothesis was that findings by Fernández et al. (2011) and Ehrenberger et al. (2012) reported for young trees, also holds for mature, fully productive trees. This is, in fact, supported by Fig. 5, which shows that the relation between the State of the Pp curve and the ranges of tree water status defined by Fernández et al. (2011) and Ehrenberger et al. (2012) holds reasonably well for olive trees of different age growing under a wide range of environmental variables. This supports the potential of the change in the State of the daily Pp curve as a user-friendly, visual indicator for irrigation scheduling. Such potential was confirmed in 2014, when we used our irrigation approach to schedule irrigation of treatment 45RDITP. In period 2 the IA values derived from our approach were lower than those calculated from the crop coefficient approach, i.e. those of the 45RDICC treatment (Figs. 7A,B). For that period, the reduced irrigation in 45RDITP as compared to that in 45RDICC did not lead to differences in plant water status (Fig. 3), despite of the greater Chapter 3 37 evaluation relying on a reduced number of variables monitored for a single irrigation season. The aim of this work was to apply the irrigations scheduling approach proposed by Padilla-Diaz et al. (2016) for two consecutive years at the Sanabria orchard, and to study a wider range of variables with the aim of further assessing the advantages and limitations of the irrigations scheduling approach based on leaf turgor related measurements as compared to the crop coefficient approach. We applied the same 45RDITP and 45RDIcc treatments they used, and stem water potential, stomatal conductance and CO2 net assimilation, as well as the number of internodes in shoots of the current year, leaf area, fruit yield, and seasonal oil accumulation in the fruit and total oil yield, were monitored on representative trees of each treatment, both in 2014 and 2015. We also had a fully irrigated treatment as a control (FI treatment), in which the trees were daily irrigated for the whole irrigation season to replace 100% of the irrigation needs. 2. MATERIAL AND METHODS 2.1. Orchard characteristics and irrigation management The orchard, located at 25 km to the east of Seville (37° 15’ N, −5° 48’ W), was planted in 2007 with one-year-old ‘Arbequina’ trees. The trees were at the top of 0.4 m high ridges and at 4 m × 1.5 m (1667 trees ha−1). Measurements described below were made during the irrigation seasons of 2014 and 2015, when the trees were 8 and 9 years old, respectively. The hedgerows were kept ca. 2.10 m wide and ca. 2.5 m high by pruning in January. We applied the same irrigation treatments described by Padilla-Díaz et al. (2016), i.e. FI, 45RDIcc and 45RDITP, and with the same layout: four 12 m × 6 m plots per treatment, in a randomized block design. Basically, daily irrigation was applied to the full irrigation (FI) trees during the whole irrigation season, aimed to replace 100% of the irrigation needs. Every Monday of the irrigation seasons the irrigation needs (IN) were calculated as IN = ETc−Pe, being ETc the crop evapotranspiration estimated from the crop coefficient approach, and Pe the effective precipitation, estimated as 75% of precipitation recorded by the weather station in the orchard (see Padilla et al., 2016, for details on the Kc and potential evapotranspiration values). For the 45RDI trees, we applied a regulated deficit irrigation strategy consisting on daily irrigation to replace 100% of the irrigation needs on the periods when the crop is most sensitive to water stress (periods 1, 2 and 3 shown in Fig. 1 of Padilla et al., 2016), while reducing to one or two irrigation events per week in between those periods. For both 45RDI treatments, the irrigation supply for the whole irrigation season aimed to 45% of the irrigation needs. The difference between 45RDICC and 45RDITP was that in the 45RDICC treatment irrigation was scheduled from the crop coefficient approach, while in the 45RDITP treatment irrigation was scheduled from the State of the daily Pp curves recorded with ZIM sensors in representative 45RDITP trees, according to the irrigation scheduling approach Confirming the scheduling RDI from leaf turgor pressure proposed by Padilla et al. (2016). Basically, we clamped a ZIM probe in three 45RDITP trees, one per plot in three plots out of the four 45RDITP plots. In 2014 we did not irrigate on period 1, because rainfall was enough, at that time of the year, to replace the crop water needs. In 2015, however, we had to irrigate on that period. Still, the irrigations amounts in on period 1 in 2015 were calculated with the crop coefficient approach for both the 45RDICC and 45RDITP treatments, because at that time we were still installing and checking the ZIM probes and related system. At the beginning of both period 2 and period 3, all the 45RDITP trees showed moderate to severe levels of water stress because of the lack of rain and the reduced irrigation applied on the weeks before each period. Consequently, most ZIM probes were in State 3. We then supplied daily irrigation amounts of 120% of the irrigation needs, on the first three days of both period 2 and 3. This was enough for the daily Pp curves recorded in those trees to change from State 2 or 3 to State 1. For the rest of the period, we adjusted irrigation to the 45RDITP trees according to the State of the Pp curves and the 3-day weather forecast given through Internet. Thus, when one out of the three Pp curves changed from State 1 to State 2 and the weather forecast announced increasing atmospheric demand, irrigation was increased by 15%. If atmospheric demand was expected to decrease, or the most sensitive instrumented 45RDITP tree did not show a change from State 2 to State 3, the irrigation dose was not modified. In case of a change from State 2 to State 3, or a second tree changing from State 1 to State 2, irrigation was increased by 15%. When the State shown by the Pp curves indicated a recovery of the tree water status, irrigation was decreased, also by 15%. Outside of periods 2 and 3, when irrigation was applied just once or twice per week, irrigation amounts were applied according to Fig. 1 in Padilla et al., (2016). On those days, i.e. on the weeks before, in between, and after periods 2 and 3, we used the irrigation needs calculated for the FI treatments with the crop coefficient approach, to derive the required irrigation amounts for the 45RDITP trees. In 2014 the irrigation season started on April 26th, day of year (DOY) 116, and ended on November 20th (DOY 324). In 2015 the irrigation season lasted from April 8th (DOY 98) to October 25th (DOY 298). The calculated irrigation doses were input in an irrigation controller (Agronic 2000, Sistemes Electrònics PRO-GRÉS, S.A., Lleida, Spain) and water was applied through a single pipe per tree row with three 2 L h−1 drippers per tree, 0.5 m apart. Fertilizers were injected once per week into the irrigation system to match the tree requirements (Fernández et al., 2013). All treatments received the same amounts of fertilizers. Climate in the area is typically Mediterranean, with mild, wet winters and little precipitation during the irrigation season (May to October). Average values of precipitation (P) and potential evapotranspiration (ETo) in the area are 522.7 mm and 1531.5 mm, respectively (period 2002-2015). The soil had a 0.4 m deep sandy layer over a clayey layer with low conductance. Most roots were in the top sandy layer (Details in Fernández et al., 2013). Chapter 3 39 2.2. Soil, plant and weather measurements All measurements described in this section and in Section 2.3 were made on both 2014 and 2015. From the 24 trees of each plot, measurements were made in the central 8 trees to avoid any border effect. Soil water status was assessed as described by Fernández et al. (2013). Basically, we used a Profile probe (Delta-T Devices Ltd., Cambridge,UK) to record volumetric soil water content (θv) values in the root zones of one tree per plot in three plots per treatment. The recorded values were used to calculate the seasonal course of the relative extractable water (REW), for all treatments. One central tree per plot, in three out of the four plots of the all treatments, was instrumented with ZIM probes (also called leaf patch clamp pressure probe and now commercially named Yara Water-Sensors, YARA ZIM Plant Technology, Hennigsdorf, Germany). In each instrumented tree, a ZIM probe was clamped on a leaf of the east side of the canopy, at ca. 1.5 m above ground. Once every 5 min the output of the probe was sent via radio to a datalogger with a GPRS modem for data transfer to a server own by YARA ZIM Plant Technology GmbH to which we accessed via Internet. Details on both the ZIM probes and their performance when used in olive are given elsewhere (see review by Fernández 2017). Basically, the probes were able to measure continuously and automatically for the whole irrigation season, since they are not affected by weather conditions, dust, spraying, etc. (Fernández et al., 2011). Precaution should be taken, however, with traffic in the orchard, since they are magnetic and can be easily removed by tractors and machinery. When that happened, or when the sampled leave was detached from the tree by any reason, reclampling was made. That did not affect the application irrigation scheduling approach, since the State of the daily Pp curve collected before and after reclamplig was the same (reclamping was made on a new leaf of the same tree). Measurements of Ψstem were made every two weeks, for the entire irrigation seasons, in one leaf per tree sampled from two representative trees per plot and three plots per treatment. The selected leaves were close to a main branch and wrapped in aluminum foil ca. 2 h before measurements. These leaves were sampled from 11.30 GMT to 12.30 GMT and Ψstem measured with a Scholander-type pressure chamber (see Padilla-Diaz et al., 2016, for details). Leaf gas exchange measurements were made on the same days and trees than Ψstem. Measurements were made at ambient light and CO2 conditions. We sampled young but fully developed leaves from the east part of the canopy, at ca. 1.5 m above ground. Measurements were made with a Licor LI-6400 portable photosynthesis system (Li-cor, Lincoln NE, USA), with a 2 cm x 3 cm standard chamber, at 08.00‒09.00 GMT, the time for maximum daily stomatal conductance (gs,max) and CO2 net assimilation (Amax) in olive (Fernández et al., 1997). Weather measurements were made as described by Fernández et al. (2013). Basically, main weather variables in the orchard were monitored by a Campbell weather station (Campbell Scientific Ltd., Shepshed, UK) located at the center of the area covered by the experimental plots. For the calculation of potential evapotranspiration values (ETo) required to schedule irrigation with the crop coefficient approach data were collected from Confirming the scheduling RDI from leaf turgor pressure a nearby standard weather station belonging to the Agroclimatic Information Network of the local government (https://www.juntadeandalucia.es/). 2.3. Growth and production measurements Shoot growth was assessed by measuring the number of internodes of four current-year shoots per tree, in two trees per plot, each randomly selected from a cardinal point of the canopy and at ca. 1.5 m above ground. Measurements were made once per month from June to November, both in 2014 and 2015. Leaf area (LA) measurements in each plot were made after pruning in January with a LAI-2000 Plant Canopy Analyzer (LI-COR, Lincoln, NE, USA) (see Cuevas et al., 2013, for details) and also during the whole irrigation seasons, one every other week. Fruit and virgin olive oil (VOO) yields were obtained after manual harvesting of three out of the eight trees in the central area of each plot, from the four plots per treatment. Harvesting was made on November 21st 2014 (DOY 325) and on November 6th 2015 (DOY 310). Total fruits per tree were weighted separately. From each plot we took 2 kg of fruit for the oil physical extraction with the Abencor method (Comercial Abengoa S.A., Seville, Spain), as detailed by Martinez et al. (1975). Fruit water productivity (WP) and oil WP were calculated as kilograms of fruit or oil per cubic meter of applied water. Although the standard definition for WP (Molden et al., 2010; Perry, 2011) considers the amount of water actually consumed by the crop and not the water applied, our soil water measurements suggested that all the water applied was consumed by the crop. Oil accumulated in the fruits was determined from June to November, both in 2014 and 2015. We collected manually around 100 fruits from ca. 1.5 m high of five out of the eight central trees of each plot (the other three trees were used to determine yield, as described above). After grinding the fruits we obtained ca. 20 g of fresh paste, which was dehydrated at 105 °C. The oil of the dehydrated paste was chemically extracted by using hexane as a solvent (Soxhlet’s method, García et al., 2013). Results were expressed as percentage of oil per dry weight. 2.4. Statistical analysis We used linear mixed models (LMM) with Tukey´s post-hoc comparisons to analyze the effects of the irrigation treatment (fixed factor) on REW, Ψstem, gs,max, A,max, number of internodes, leaf area and fruit and VOO yield, as dependent variables at α < 0.05. We used leaf identity within plot as the random factor structure in the Ψstem, gs,max, and A,max analyses to describe appropriately our experimental design and deal with the non–independent nature of the spatial experimental design. In the rest of comparisons the random factor was not necessary as we only have one value per plot. When no normal and heterocedastic residuals were obtained, appropriate transformation of the variable was used. Data are shown as mean ± standard error. We used the R software (R Core Team, 2012) to perform Chapter 3 41 the statistical analysis, with help of ‘nlme’ packages (Pinheiro et al., 2011) for LMM and for pair comparisons ‘multcomp’ (Hothorn et al., 2008). 3. RESULTS 3.1. Water supplies, atmospheric demand and soil water content Results on the irrigation amounts (IA) supplied to each treatment are shown in Table 1. The good agreement between the aimed and the actual doses shows a reasonably good irrigation management. Still, in 2015 total water supplies for the 45RDI treatments were closer to 50% than to 45% of the irrigation needs. That was a year of high atmospheric demand (1581.37 mm) and low precipitation (237.60 mm), so the calculated irrigation needs (5855 m3 ha-1) were unusually high (Table 1), as compared to previous years in which this amount was usually below 5000 m3 ha-1 (Fernández et al., 2017). The total IA applied in 2015 to both 45RDI treatments was quite similar, although with marked differences on periods 2 and 3 (Table 1). The seasonal courses of IA for the three treatments are displayed in Fig. 1A, together with the collected precipitation. Values of relative extractable water (REW) calculated from the soil water measurements made in 2015 are shown in Fig. 1B. For the FI treatment they were usually around 0.8, suggesting non-limiting soil water conditions for olive. The low REW value registered in the FI treatment on DOY 225 (Fig. 1B) was probably due to the peak on atmospheric demand registered on the previous days (Fig. 2). For the two 45RDI treatments REW values showed similar trends, with no differences between treatments. Before periods 2 and 3, the soil in the two 45RDI treatments was very dry but recovered quickly after the increase on irrigation of both periods, and hence, no differences in REW were observed among the FI and the 45RDI treatments (Fig. 1B). In between period 2 and period 3, REW values showed severe soil water depletion in the 45RDI treatments, as expected. In the autumn, the rainfall and IA supplied by irrigation avoided the soil becoming so dry. Main features of the REW seasonal courses in 2015 were similar to those in 2014 (see Padilla et al., 2016), with no differences among treatments on periods 2 and 3, for any of the two years. Confirming the scheduling RDI from leaf turgor pressure Table 1. Water supplies (IA = irrigation amounts; P = precipitation) and potential evapotranspiration (ETo) in the experimental orchard for the two irrigation seasons (2014 and 2015). All values are in millimeters. Values of IA are also expressed as percentages of irrigation needs (% IN). DOY = day of year. Values with an asterisk were published by Padilla-Díaz et al. 2016 and with two asterisks by Hernandez-Santana et al. 2017. Whole year Irrigation period Period 2 Period 3 2014 DOY 116–324 DOY 154–185 DOY 238–259 P 549.4* 317.6* 8.2 5.3 ETo 1437.6* 1110.1* 191.3 143.5 IA in FI 462.9 (87.1% IN)* 93.1 (104.3 % INP2)* 69.7 (103.6% INP3)* IA in 45RDICC 238.2 (44.8% IN)* 72.2 (81.0 % INP2)* 66.8 (99.0 % INP3)* IA in 45RDITP 235.9 (44.4% IN)* 61.5 (68.0% INP2)* 67.1 (100.0 % INP3)* 2015 DOY 98-298 DOY 146-173 DOY 237-257 P 237.60** 76.22** 0.00 0.00 ETo 1581.33** 1223.65** 193.48 123.00 IA in FI 585.55 (101.5% IN)** 94.62 (101.7% INP2) 362.16 (105.8% INP3) IA in 45RDICC 286.98 (49.8% IN)** 73.16 (78.6% INP2) 61.21 (107.3% INP3) IA in 45RDITP 290.33 (50.3% IN) 83.79 (90.0% INP2) 50.93 (89.3% INP3) Chapter 3 43 Fig. 1. Seasonal courses of both the irrigation amounts (IA) supplied to each treatment and the precipitation (P) collected in the orchard (A), and the values (avg ± SE) of relative extractable water (REW) derived from the soil water contents measured in the plots of each treatment (B). Measurements were made on the irrigation season of 2015. A similar figure but with data of 2014 was published by Padilla et al. (2016). P1, P2 and P3 correspond to the periods 1, 2 and 3 in which olive is most sensitive to water stress (Fernández et al., 2013). Different letters indicate significant differences between treatments at p < 0.05. Letters are not shown when no differences were found. DOY = day of year. Part of the data of REW was shown in Hernandez-Santana et al. (2017). 100 120 140 160 180 200 220 240 260 280 300 REW 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 5 10 15 20 25 P (mm) IA (mm tree-1 day-1) 0 1 2 3 4 5 FI 45RDICC 45RDITP a b b a b b a b b a b b a b b P2 P3P1 A B DOY 2015 (DOY 160 = June 9) Sep Apr May Jun Jul Aug Oct Confirming the scheduling RDI from leaf turgor pressure Fig. 2. Seasonal courses, for the irrigation seasons of 2014 (A) and 2015 (B), of both the potential evapotranspiration (ETo) in the area and the crop evapotranspiration (ETc) estimated with the crop coefficient approach for the olive orchard at Sanabria. P1, P2 and P3 correspond to the periods 1, 2 and 3 in which olive is most sensitive to water stress (Fernández et al., 2013). DOY = day of year. One of the concerns of Padilla-Diaz et al. (2016) was the suitability of the 15% value for changing IA. Although percentages ranging from 10 to 20% are common in this type of studies (Goldhamer and Fereres, 2004; Conejero et al., 2007; Velez et al., 2007; Fernández et al., 2008a), they had no evidence on 15% being a proper value. To assess whether that value was appropriate, we determined the coefficient of variation (CV, %) of the daily values of the irrigation needs (IN) calculated at Sanabria with the crop coefficient approach, on periods 2 and 3 of the two experimental years (Table 2). The coefficient of variation was calculated as the ratio of the standard deviation of the daily IN values to the average IN for each period, expressed in percentage. With those values we calculated the DOY (DOY 160 = June 9) 100 120 140 160 180 200 220 240 260 280 300 320 0 2 4 6 8 10 0 2 4 6 8 0 2 4 6 8 10 12 0 2 4 6 8 10 P2 ETc (mm) ETo (mm) P1 Sep Apr May Jun Jul Aug Oct 2014 2015 P2 P3 P3 A B Nov P1 Chapter 3 45 mean of the CV and the standard error for daily changes in IN. As shown in Table 2, the average CV for IN was 15.14%, a value similar to the 15% change in the irrigation amount adopted in our irrigation scheduling approach. Table 2. Daily variation of the irrigation needs for each Period under our irrigation scheduling approach based on the ZIM system. CV IN (%) values are the coefficient of daily variation of irrigation needs. DOY = day of year. Year Period CV IN (%) 2014 P2 (DOY 154-185) 16.77 P3 (DOY 238-259) 15.14 2015 P2 (DOY 146-173) 14.58 P3 (DOY 237-257) 16.78 In addition to the accumulated ETo values for the two irrigation seasons, Table 1 shows the ETo values for the whole 2014 and 2015, and for the periods 2 and 3 of each year, and Fig. 2 shows the seasonal ETo courses for both irrigation seasons. Also shown in that figure are the seasonal courses of the crop evapotranspiration (ETc) values calculated with the crop coefficient approach (Section 2.1). 3.2. Plant water status versus leaf turgor related measurements Figure 3 shows the results from applying the Irrigation scheduling approach developed by Padilla-Díaz et al. (2016) to schedule irrigation in the 45RDITP treatment, on periods 2 (Fig. 3A,B) and 3 (Fig. 3C,D) of 2015, as compared to the crop irrigation approach applied. Figures 3A and 3C display the collected Pp curves from the sampled tree that first showed the onset of water stress through the shift from State 1 to State 2. Arrows indicate the days in which the irrigation amount was increased (arrow up) or decreased (arrow down) by 15%, as a response to changes in the State of the daily Pp curve. The resulting irrigation amounts are displayed in Fig. 3B (period 2) and 3D (period 3). On period 2, the total amounts of water supplied to the 45RDICC and 45RDIs treatment amounted to 79% and 90% of the irrigation needs calculated for the FI treatment, respectively (Fig. 3B). On period 3, these values were of 107% and 89%, respectively (Fig. 3D). See Padilla et al., (2016) for a similar figure with data of 2014. Confirming the scheduling RDI from leaf turgor pressure Fig. 3. Time courses of the Pp values recorded on a 45RDITP representative tree in period 2 (A) and period 3 (C) of the 2015 irrigation season (see Fernández et al., 2013, to identify those periods in which olive is most sensitive to water stress). Also shown are the irrigation amounts (IA) supplied to each treatment in each period, expressed as a fraction of the calculated irrigation needs (IN) for the period (B,D). IA was increased (arrows up) or decreased (arrows down) according to changes in the State of the daily Pp curves and the weather forecast of three days (see the irrigation scheduling approach described in the Introduction section). DOY = Day of year. 0 20 40 60 80 100 120 140 FI 45RDICC 45RDITP DOY 2015 (DOY 238 = August 26) 237 240 243 246 249 252 255 258 0 10 20 30 A B Pp (kPa) C D IA (L tree-1 day-1) 20 40 60 80 100 120 Pp (kPa) DOY 2015 (DOY 154 = June 3) 147 150 153 156 159 162 165 168 171 174 IA (L tree-1 day-1) 0 5 10 15 20 FI 45RDICC 45RDITP In 45RDITP, IA = 0.90 IN In 45RDICC, IA = 0.79 IN In 45RDITP, IA = 0.89 IN In 45RDICC, IA = 1.07 IN Period 2 Period 3 May Jun Aug Sep Chapter 3 53 Fig. 8. Seasonal courses of oil accumulation recorded in 2014 and 2015. Values are avg. ± SE and different letters indicate significant differences between treatments at p < 0.05. Letters are not shown when no differences were found. P2 and P3 correspond to periods 2 and 3 represented in Fig. 3. DOY = Day of year. DOY (DOY 160 = June 9) 120 140 160 180 200 220 240 260 280 300 320 0 10 20 30 40 50 Oil content (% dry weight) 0 10 20 30 40 50 60 FI 45RDICC 45RDITP P2 P3 P2 P3 a ab b a ab b a b b a b b A B 2015 2014 Jun JulMay Aug Sep Oct Nov Confirming the scheduling RDI from leaf turgor pressure Table 3. Fruit and oil yield, water productivity of fruit and oil for each treatment and experimental years (2014 and 2015). Values for each treatment are in average ± SE, n = 4. Different letters indicate significant differences between treatments at p < 0.05. Values with an asterisk were published by Padilla-Díaz et al. 2016 and with two asterisks by Hernandez-Santana et al. 2017. Year Treatment Fruit yield (kg ha-1) Oil yield (kg ha-1) WP Fruit (kg ha-1 mm-1) WP Oil (kg ha-1 mm-1) 2014 FI 19283.0 ± 2708.5 a* 1875.8 ± 288.6 a 41.6 ± 5.9 a* 4.1 ± 0.6 a 45RDICC 13443.0 ± 2847.9 a* 1195.0 ± 258.0 a 53.4 ± 12.0 a* 5.0 ± 1.1 a 45RDITP 17025.6 ± 2077.0 a* 1641.5 ± 132.4 a 72.2 ± 8.8 a* 7.0 ± 0.6 a 2015 FI 19724.8 ± 1608.28 a** 1957.4 ± 183.7 a** 33.7 ± 2.7 a 3.3 ± 0.3 a 45RDICC 15073.8 ± 386.04 a** 1387.4 ± 86.1 a** 52.5 ± 1.4 b 4.8 ± 0.3 b 45RDITP 13865.6 ± 757.15 a 1534.5 ± 130.6 a 47.8 ± 2.6 b 5.3 ± 0.5 b Chapter 3 55 4. DISSCUSION 4.1. Impact of the scheduling irrigation approach on crop physiology and crop performance Similarly to 2014 (see Padilla-Díaz et al., 2016, for details), in 2015 no significant differences either on the total irrigation amount applied or on the seasonal course of REW were found between 45RDICC and 45RDITP (Fig. 1). For both period 2 and 3, the irrigation amounts supplied to the 45RDITP trees was ca. 90% of IN (Table 1, Fig. 3), which means that our irrigation scheduling approach based on the use of leaf turgor related measurements was suitable for ensuring non-limiting soil water conditions on these periods of high crop sensitivity to water stress. The water status of the 45RDITP trees showed, in fact, no differences on periods 2 and 3 with the FI trees. Values of Ψstem values were nearly always above ‒1.2 MPa, a threshold value for safe hydraulic functioning in olive (Torres- Ruiz et al., 2013; Hernandez-Santana et al., 2016). This agrees with trees of the 45RDI treatments being in State 1 or 2 most of the days on periods 2 and 3. In fact, the number of days in State 2 shown by the 45RDI trees on both periods was similar to that of the FI trees. This suggests that the occurrence of State 2 on those periods was due more to the high atmospheric demand recorded on those days (Fig. 2) than to a lack of water in the root zone. Values of both gs,max and Amax were also similar for all treatments, on periods 2 and 3 (Fig. 5). This is with the exception of the first days of both periods, on which gas exchange was lower in the 45RDI trees than in the FI trees. This was expected, since it is known than in olive, Ψstem recovers soon after water stress than gas exchange. This, the full recovery of gs,max and Amax may take a few days depending on the severity of the water stress suffered before the recovery irrigation (Fereres et al., 1996; Fernández et al., 1997). The mechanisms behind that behaviour has not been clearly stablished yet. Nevertheless, Torres-Ruiz et al. (2013) suggested that this stomatal behaviour during stress and recovery might be due to the ABA synthetized during the water stress period, which controlled stomata opening at the beginning of the recovery period. The 45RDI strategy applied in the orchard implies a period of water stress before periods 2 and 3, in which the soil can become markedly depleted (Fig. 1B). As described above, in our irrigation scheduling approach we irrigated with 120% of the irrigation needs on the first days of both periods, for a quick recovery of the soil water status, aiming to minimise the described delay in the recovery of both gs,max and Amax. Plant growth was reduced by the 45RDI treatments, as expected, but no differences were observed between 45RDICC and 45RDITP either for the number of internodes in the current-year shoots (Fig. 6) or leaf area (Fig. 7). It is well known that growth in olive is very sensitive to water stress, as occurs in other fruit tree species (Fernández, 2014b; Rosecrance et al. 2015; Cajias et al. 2016), so the lower number of internodes and reduced leaf area in the 45RDI trees was expected. This effect of the 45RDI treatments on growth agrees with the lower levels of Amax found in these trees (Fig. 5), as compared to the FI trees. The reduction both of gs,max and Amax was not necessarily negative. On the one hand, Confirming the scheduling RDI from leaf turgor pressure reducing gs can improve water use efficiency, since the A vs. gs relationship in olive shows A becoming asymptotic for gs values above ca. 0.2 mol m-2 s-1 (Fernández et al., 2008b). On the other hand, a reduction in A contributes to avoid excessive growth (Hernandez- Santana et al., 2017), which may be an advantage in SHD olive orchards both to facilitate mechanical harvesting with the vineyard type straddle-harvesters normally used in these orchards (León et al., 2007) and to avoid competency for light among trees (Gomez-del- Campo et al., 2009; Connor et al., 2012). Both factors ensure a long productive life of the orchard (Fernández et al., 2017). In between periods 2 and 3, when REW values ranged most of the time from 0.2 to 0.3 only for the 45RDI treatments (Fig. 1B), and maximum values of atmospheric demand were recorded (Fig. 2), the 45RDI trees became severely stressed. Thus, from mid-July to the end of August the 45RDI trees showed midday Ψstem values in between ca. ‒3.0 and ‒4.7 MPa (Fig. 4), gs,max values of ca. 0.05 mol H2O m-2 s-1 and Amax values around 4 mol CO2 μm-2 s-1. It is known, however, that the olive tree is quite tolerant to water stress at this time of the year (Moriana et al., 2003; Perez-Martin et al., 2009), and that it is able to recover from severe water stress with a reduced impact on crop performance, providing enough water during both period 3 and ripening is available (Lavee et al., 1990; Lavee and Wodner, 1991). This agrees with the dynamics of oil accumulation observed in our orchard, which showed a recovery of the oil content in the fruits after the beginning of period 3 (Fig. 8). Concerning production, oil accumulation in the fruits during pit hardening was affected in the 45RDI trees, as compared to the FI trees (Fig. 8). During ripening, however, similar values were observed in all treatments. For fruit and oil yield, our statistical analysis showed no differences between treatments (Table 3). Olive presents a linear increase of production with water supply, until a certain level from which little increase on production, if any, is observed when increasing the water supply (Moriana et al., 2003). Other authors have reported little impact on fruit and oil yields despite significant reductions on water supply (Grattan et al., 2006; Rosecrance et al., 2015). In our case, and as indicated in Section 3.3, the number of replicas we had was too low to properly account for the high tree-to-tree variability on both fruit and oil yield existing in the orchard. It is clear, however, that similar results on fruit and oil yield were obtained independently of the approach used to schedule the 45RDI treatment, which means that, for both production variables, our irrigation scheduling approach based on the use of the ZIM system was as effective as that based on the crop coefficient approach. The same can be said for the water productivity (WP) values (Table 3). In fact, at least in 2015 differences on fruit and oil WP were significant in between the FI and the 45RDI treatments, with no differences between 45RDICC and 45RDITP. Chapter 3 57 4.2. Suitability of the scheduling irrigation approach for commercial orchards As detailed above, our results suggest similar crop performance for both 45RDICC and 45RDITP. Previous results obtained by our group at the Sanabria orchard, with similar irrigation strategies as the 45RDI but with different levels of water stress, all scheduled with the crop coefficient approach, suggested that 45RDICC is a suitable option for SHD olive orchards similar to our experimental orchard (Fernández et al., 2013, 2017; García et al., 2017). However, and as Padilla-Diaz et al. (2016) already pointed out, the required values of both the crop coefficient (Kc) and the potential evapotranspiration (ETo) in the area are not available for many olive orchards. On the top of that, the crop coefficient approach implies a delay in the scheduling of irrigation, since weather data must be collected prior to the calculation of ETo. The irrigation scheduling approach proposed by Padilla et al. (2016) and further tested in this work, relies on continuous and automatic leaf turgor related measurements. This real-time data information is suitable for irrigation scheduling in olive (Fernández 2017). Fernández (2014b) made a comparative study on three different methods to schedule irrigation in SHD olive orchards, based, respectively, on sap flow, trunk diameter variations and leaf turgor related measurements. They reported that the ZIM system used for leaf turgor related measurements was the easiest to install and use, and robust enough to withstand field conditions for long irrigation seasons. The potential and limitations of these three plant-based variables to derive suitable plant water stress indicators and to schedule irrigation in commercial orchards was analysed not only by Fernández (2014b), but also by Fernández (2017) and Fernández et al. (2017), and they all agreed on the high potential of leaf turgor related measurements for precision irrigation. Egea et al. (2017) made a financial assessment of the three mentioned methods to schedule regulated deficit irrigation in super high density olive orchards, and concluded that scheduling irrigation with the ZIM method, i.e. from leaf turgor related measurements, was the most profitable one. Our results confirm the advantages of the irrigation scheduling approach proposed by Padilla et al., 2016. Among the main ones is that, contrarily to sap flow and trunk diameter variations records, no data processing is required, because the visual analysis of the Pp daily curves is enough to identify the State, i.e. the level of water stress, of the monitored trees. A procedure for the automatic identification of the State shown by daily Pp curves, based on the use of random forest models, has been recently published by Fernandes et al. (2017), which will facilitate the application of the irrigation scheduling approach to commercial orchards. In addition, the method can be combined with airborne imagery for the zoning or zonification of the orchard. This allows for differential irrigation doses and frequencies depending on the variation in water requirements identified within the orchard. Finally, the combination of leaf turgor related measurements with the 3-day weather forecast allows for an early response of irrigation management to environmental conditions. These features confer our irrigation scheduling approach a high potential for precision irrigation, which may significantly contribute to the rational water use in super high density olive orchards (Fernández, 2017; Fernández et al., 2017). Confirming the scheduling RDI from leaf turgor pressure The irrigation scheduling approach has, however, several limitations. First, caution must be taken when assuming a particular water stress level to the State shown by the Pp curve. Thus, Marino et al. (2016) found lower threshold values of Ψstem for the olive Sicilian genotypes ‘Nocellara del Belice’ and ‘Olivo di Mandanici’ than those reported by Fernández et al. (2011) and Ehrenberger et al. (2012) for ‘Arbequina’ trees. Further studies are then needed to establish the extent at which the ranges of water stress for States 1 to 3 depend on the cultivar and orchard conditions. Second, neither the 120% increase in the irrigation amount on the first days of periods 1 to 3 nor the 15% change in the irrigation amount made afterwards, have been contrasted against the crop physiology or crop performance. The fact that the average coefficient of variation value of the irrigation needs agrees with the 15% percentage of change on the irrigation amount (Table 2) suggests that this percentage is reasonable. This supports that the use of this percentage recommended for our irrigation scheduling approach is suitable for proper irrigation scheduling in orchards similar to our experimental orchard. In any case, the empirical character of the two percentages adopted in the irrigation scheduling approach may curtail their reliability when applied to other orchards with different conditions. Third, the ZIM probe does not work properly beyond a certain level of water stress which, in our case, was reported to be that occurring at ca. Ψstem < ‒1.7 MPa. This is not limiting for scheduling irrigation on periods 1, 2 and 3, because Ψstem is usually greater than ‒1.7 MPa thanks to the high irrigation amounts applied on those days. Still, State 3 was often found in between periods, when the irrigation amounts are low (Fig. 4) and, as explained by Ehrenberger et al. (2012), the information provided by the ZIM probes on those days is not useful to assess the actual water stress of the trees. Irrigation scheduling in between periods must relay, therefore, on a different method. Padilla et al. (2016) suggested that, on those days, the farmer could use whatever knowledge they have on the orchard water needs. This is a rough solution that may lead to imprecise determinations of the irrigation amounts, but the impact on the overall irrigation management will be low because of the reduced number of irrigation events on those days in between periods (see Fig. 1 in Padilla et al., 2016). In any case, we are already working on new water stress indicators derived from the Pp values recorded at State 3. Any advances on this task, together with the approach derived by Fernandes et al. (2017) for the automatic assessment of the daily Pp curve State, could highly contribute to the use of the proposed irrigation scheduling approach in a context of precision irrigation. 5. CONCLUSIONS Our findings, on a greater number of variables than those monitored by Padilla-Diaz et al. (2016), confirm that the proposed irrigation scheduling approach is suitable to schedule regulated deficit irrigation in hedgerow olive orchards with similar conditions than those used in the present experiments. Both calculating the irrigation amounts at the beginning of the periods of high sensitivity to water stress (periods 1 to 3) as 120% of the irrigation Chapter 3 59 needs and changing by 15% the irrigation amounts applied for the rest of each period seem to be adequate for keeping non-limiting soil water conditions on those periods. Nonetheless, the fact that the values recorded by the ZIM probes at State 3 are not informative on the actual leaf turgor, curtails the suitability of the proposed irrigations scheduling approach to calculate irrigation amounts in between periods. It is expected that both recent advances for the automatic identification of the daily Pp curve States and further studies on the information provided by the Pp records at State 3 will increase the suitability of the proposed irrigation scheduling approach for precision irrigation of hedgerow olive orchards. Confirming the scheduling RDI from leaf turgor pressure Chapter 4 Effect of fruit development on turgor related measurements in olive leaves Chapter 4 69 Inc., Osaka, Japan). The Chroma Meter was calibrated by a standard white and CIE coordinates (L, a and b). The fruit colour index was determined by the equation Fruit Colour Index = 𝐿(𝑏−𝑎) 100 (4) where L is the lightness (white = 100, black = 0), a is the range red-green (a > 0 represents the increasing redness whereas a < 0 represents the increasing greenness) and b is the range yellow-blue (b > 0 represents the increasing yellowness whereas b < 0 the increasing blueness). Higher values of colour index mean that the colour of the fruits was shifting from green to yellowish. Lower values correspond to later phases of veraison, when the fruits became purple and black. 2.7. Statistical analysis Data shown of the all physiological variables including REW are mean ± standard error. We used t-test comparisons of mean to analyse the differences between treatments (here considered P ≤ 0.05) for all measured physiological variables and for the Ppmin values for each treatment. Before each t-test comparison, we tested the normality and homoscedasticity of each variable using a Shapiro-Wilk and Levene’s test, respectively. Those variables that were neither adjusted to normality nor homoscedasticity of the variance were transformed prior to apply t-test comparisons. Data from concomitant measurements (i.e. REW, fruit dry weight and fruit colour index) were interpolated assuming linear relationships using the “zoo” package (Zeileis and Grothendieck, 2005). Principal component analyses (PCA) were carried out to assess the effect of the main environmental variables on Ppmin, for each treatment. We developed three PCA tests for the main periods of fruit development with the main daily environmental variables (i.e. maximum VPD, ETo, maximum solar net radiation, P, maximum wind speed, interpolated REW and irrigation amounts applied). This analysis was carried out for each of the three intervals related to olive fruit development (Sanz-Cortés et al., 2002). These intervals are referred to in the literature as period 2 (P2), period 3 (P3) and the period in between (Fernández et al., 2013; Fernández, 2014a; Padilla-Díaz et al., 2016). Both P2 and P3 are considered moments of high sensitivity to water stress (Fernández et al., 2013; Fernández 2014a). Due to the percentage of explanatory variability, two main components of the PCA were taken to perform linear models, using Ppmin as response variable to all three periods and treatment. Distribution of normality and homoscedasticity of the residuals of the models were tested prior to transformation of the variable. Pearson’s correlation tests were performed to understand the percentage of association between fruit dry weight as well as fruit colour index with Ppmin. These analyses were performed using R software (R Core Team, 2012) v. 3.4.2. The fruit effect on leaf turgor 70 3. RESULTS 3.1. Maximum turgor related measurements and the influence of fruit presence We found that the time courses of normalized maximum turgor related measurements (Ppmin) were not affected by the proximity of fruits (Fig. 1A), as shown by theto nonstatistical differences between the records taken in leaves of branches with fruits (F+) and branches without fruits (F‒). Still, Ppmin values were higher in F+ than in F‒ at certain moments along the recorded period: at the end of period 2, at the beginning of the period in between P2 and P3, three days after the sudden pick of VPD on DOY 249, and on the central days of period 3. Both in the P2 and P3 periods, relative Ppmin values increased from 20%, or less, at the beginning of both periods, to over 60% in P2 and around 80% in P3. In between both periods relative Ppmin values decreased to below 20%, except on days in which, either the interruption of irrigation (DOY 208) or peak VPD values (around DOY 249), cause sudden increases on Ppmin (Fig. 1). Moreover, there were no differences neither among trees nor between treatments (i.e. among the 24 ZIM sensors) in the shape of the daily turgor related measurements (daily Pp curves) along the whole experimental season (Fig.S1, supplementary material). 3.2. Relationship between maximum turgor and environmental variables The seasonal courses of Ppmin, for both F+ and F‒ treatments, was plotted against main environmental variables (Fig.1). Irrigation was enough to keep REW values close to 1 for most of the irrigation season, except after DOY 300, when irrigation supplies were reduced and even withheld because of the autumn rains (Fig. 1C). The PCA analysis helped us to evaluate the plant-water relations and, basically, the effect of the environmental variables on Ppmin. With this approach we assessed the effect of the environmental variables shown in Fig. 1, for periods P2 and P3, as well as for in between P2 and P3. Results are shown in Table 1. The main variability was further explained by first principal component (PC1) of the PCA than the second one (PC2) for the three periods studied. Those PC1 and PC2 were also renamed depending on which variable had a greater impact on Ppmin. Thus, in all of cases PC1 was renamed as maximum atmospheric demand and PC2 as water availability. Although the maximum wind-speed (umax) belonged to the PC2 in period 2 and in between periods, we considered renaming PC2 as water availability because of the high effect of the irrigation amounts and REW on the variance. On period 3, both precipitation (P) and umax belonged to PC1 instead of PC2, because of their effect on the atmospheric demand. To test the influence of both the maximum atmospheric demand and water availability on Ppmin for both treatments (F+ and F‒), we performed linear models with Ppmin and those components of the PCA for each treatment and interval period (Table 2). For both treatments, the most important factor that explained the Ppmin daily fluctuations in period 2 (DOY 169‒201), was the maximum atmospheric demand, followed by the water availability. In between periods (DOY 202‒286), the adjustment between PC1 and PC2 with Ppmin was greater than in period 2. Nevertheless, in period 3 (DOY 287‒313) no Chapter 4 71 influence of water availability was found, although the irrigation water supply was withheld due to rainy events. In this case, the maximum atmospheric demand (including umax and P) was the main explanatory variable and the adjustment was lower than in between periods, as occurred in period 2. Fig. 1 Seasonal courses of maximum vapour pressure deficit (VPDmax), calculated from weather records at the experimental orchard, and of the relative maximum daily turgor (Ppmin; avg ± SE) recorded on leaves both close to a cluster of fruits (F+ treatment) and far from them (F‒ treatment) (see text for details on the treatments). Differences on Ppmin between treatments were never significant (P ≤ 0.05) (A). Seasonal courses of potential evapotranspiration (ETo) in the area, maximum net radiation (Rnmax) and maximum wind speed (umax) (B). The seasonal course of relative extractable water (REW; avg ± SE) and values of both precipitation (P) collected in the orchard and irrigation amounts (IA) supplied to the experimental trees (C). P2 and P3 accounts for periods 2 and 3 in which olive is most sensitive to water stress (Fernández et al., 2013; Fernández et al., 2014a). DOY = day of year. 0 2 4 6 8 10 VPDmax (kPa) 5 15 0 10 umax (m s-1) 0 300 600 Rn max (W m-2) 1 3 0 2 4 IA (mm) 0.50 0.75 1.00 REW 0 20 40 60 80 100 F+ F 0 2 4 6 8 10 ETo (mm) Ppm in (%) A B C DOY 2016 (200 = July 18) 160 180 200 220 240 260 280 300 5 15 25 0 10 20 30 P (mm) P3P2 The fruit effect on leaf turgor 72 Period 2 In between periods Period 3 DOY 169 ‒ 201 DOY 202 ‒ 286 DOY 287 ‒ 313 PC1 PC2 PC1 PC2 PC1 PC2 Environmental variables Atmospheric demand Water availability Atmospheric demand Water availability Atmospheric demand Water availability VPDmax (kPa) 0.558 0.06 -0.467 0.164 0.473 -0.002 P (mm) 0.154 -0.465 -0.448 0.380 umax (m s-1) -0.200 0.428 -0.240 -0.551 -0.371 0.105 Rnmax (W m-2) 0.443 0.347 -0.494 0.085 0.455 -0.141 ETo (mm) 0.528 0.369 -0.550 -0.030 0.433 0.213 IA (mm) -0.362 0.518 -0.296 -0.509 0.151 0.632 REW -0.206 0.534 0.254 -0.431 0.142 0.615 Proportion of variance 0.409 0.220 0.426 0.282 0.489 0.233 Cumulative Proportion 0.630 0.611 0.722 Table 1. Results from the principal component analysis (PCA) based on the main environmental variables that could have affected plant performance in every interval of fruit development along the experimental season. PC1 and PC2 are the results of the first and second component of PCA which explain most of the variability in Ppmin. PC1 and PC2 were renamed according to the main daily environmental variables of each group, which appear in bold. Precipitation was removed from the analysis of period 2, since there was no rain in that period. Proportion of the variance is the interval adjustment of the variance which is explained by each component of the PCA, whereas the total variance of the analysis is explained by the cumulative proportion. VPDmax = daily maximum vapour pressure deficit; P = precipitation; umax = maximum daily wind speed; ETo = reference evapotranspiration; IA = irrigation amount; REW = relative extractable water; DOY = day of year. All variables except ETo were taken from the orchard meteorological station. Chapter 4 73 Table 2. Results, for each treatment (F+ and F‒) and interval of fruit development along the whole experimental year of the linear models fitted to the principal component analyses (PCA) for the main environmental daily variables that affects Ppmin, the response variable. Each interval was selected according to the main periods of water sensitivity in olive trees (Fernández et al. 2013, Fernández 2014a). PC1 and PC2 are the results of the first and second component of PCA which explain most of the variability in the response variable. PC1 and PC2 were renamed according to the group of environmental daily variables with more weight on Ppmin. F+ is when the sampled leaf was close to a cluster of fruits and F‒ is when the sample leaf was far from a cluster of fruits (see text for details). DOY = day of year. 3.3. Relationship between of maximum turgor and fruit development Our results show that the time courses of fruit growth, measured as the rate of fruit dry weight, exhibit a double-sigmoid pattern (Fig. 2). The first phase of rapid growth occurred from ca. DOY 168 till DOY 208, i.e. on the P2 period, the maximum rate of pit hardening. The second phase of fast growth was from ca. DOY 294 to 313, i.e. on period P3, when most changes in fruit colour index occurred, the fruits going from light green and yellowish to purple. Those two phases of maximum rate of pit hardening and veraison agreed with an increment on Ppmin in both treatments (Fig. 2A). Both for F+ and F‒, the increment in Ppmin was linearly correlated, in P2, with fruit dry weight (R2 = 0.77, P < 0.0001; Fig. 3A). In P3, however, we found no correlation between both variables (Fig. 3B). This is not surprising because the number of sampling days for both fruit dry weight and fruit colour index was very low in comparison with the high variability of Ppmin on that period (Fig. 2). Period 2 In between periods Period 3 DOY 169 ‒ 201 DOY 202 ‒ 286 DOY 287 ‒ 313 PC1 PC2 PC1 PC2 PC1 PC2 Atmospheric demand Water availability Water availability Atmospheric demand Atmospheric demand Water availability R2 P value P value R2 P value P value R2 P value P value F+ 0.24 0.04 0.03 0.37 < 0.01 < 0.01 0.19 0.03 0.73 F‒ 0.25 0.06 0.02 0.37 < 0.01 < 0.01 0.28 < 0.01 0.53 The fruit effect on leaf turgor 74 Fig. 2 Seasonal courses of maximum vapour pressure deficit (VPDmax), calculated from weather records at the experimental orchard, and of the relative maximum daily turgor (Ppmin; avg ± SE) recorded on leaves both close to a cluster of fruits (F+ treatment) and far from them (F‒ treatment) (see text for details on the treatments). Differences on Ppmin between treatments were never significant (P ≤ 0.05) (A). Seasonal courses of fruit dry weight (n = 6) and fruit colour index (n = 25) measured along the whole experimental season. The shown values are avg ± SE (B). P2 and P3 accounts for periods 2 and 3 in which olive is most sensitive to water stress (Fernández et al., 2013; Fernández et al., 2014a). In 2016, the experimental year, P3 agreed with veraison. DOY = day of year. 0 2 4 6 8 10 VPDmax (kPa) 160 180 200 220 240 260 280 300 0.0 0.2 0.4 0.6 0.8 Fruit dry weight (g) DOY 2016 (200 = July 18) 15 20 25 30 35 40 Fruit Colour Index 0 20 40 60 80 100 F+ F Ppm in (%) P2 Maximum rate of pit hardening P3 Veraison A B Chapter 4 75 Fig. 3 Relationship between fruit dry weight (avg ‒ SE) and relative maximum daily turgor (Ppmin; avg ‒ SE) for periods P2 (A) and P3 (B). Relationship between fruit colour index (avg ‒ SE) and the relative maximum daily turgor (Ppmin; avg ‒ SE) (C). Ppmin values were recorded on leaves both close to a cluster of fruits (F+ treatment) and far from them (F‒ treatment) (see text for details on the treatments). All correlations were made on the average of both treatments and were significant at least as P ≤ 0.05. P2 and P3 correspond to the periods 2 and 3 in which olive is most sensitive to water stress (Fernández et al., 2013; Fernández et al., 2014a). DOY = day of year. P2 (DOY 168 - 200) Fruit dry weight (g) 0.1 0.2 0.3 Ppmin (%) 0 20 40 60 80 100 R2 = 0.77 A P3 (DOY 294 - 313) 0.5 0.6 0.7 0.8 F+ F B P3 (DOY 294 - 313) Fruit Colour Index 20 25 30 35 Ppmin (%) 0 20 40 60 80 100 C F+ F The fruit effect on leaf turgor 76 3.4. Leaf-water relations and the influence of fruit presence Significant differences (P ≤ 0.05) between the F+ and the F‒ treatments were found for maximum daily net CO2 assimilation (AN,max) around the beginning of period 2 (DOY 166 and 169), .No differences were found, hoewver, for maximum daily stomatal conductance (gs,max), predawn leaf water potential (Ψpd) and predawn leaf relative water content (RWCpd) (Fig. 4B‒D). For the rest of the experimental season there were no statistical differences between treatments, for any of the variables. The time courses of AN,max and gs,max (Fig. 4B,C) were affected by events of high VPDmax (Fig. 4A), such as the peak VPD values recorded around DOY 249. At the end pf P3, both gas exchange variables decreased, likely due to those days in November being cold and dark (Fig.1B). The Ψpd values suggest a lack of overnight recovery of the trees water status in mid-summer (in between periods 2 and 3), since values lower than -0.5 MPa, the threshold for satisfactory recovery (Fernández et al., 2014a) were often recorded (Fig. 4D). Likely the high values of VPD on those days (Fig. 4A) were responsible for that. On period 3, however, when VPD values were lower, the trees were able to rehydrate properly during the night. These findings agree with the time course of RWCpd (Fig.4E). Around DOY 253, when a marked peak of VPD occurred, gs,max, AN,max and RWCpd decreased, as corresponded. The results of the calculations derived from the pressure-volume (P-V) curves performed for each sampled day are given in Fig. 5. The seasonal courses showed a nonstationary trend for all the calculated variables. Less negative values were attained for F+ treatment compared with F‒, for both πo and the leaf water potential at turgor loss point ( Y tlp) (Fig. 5B and C), although the high variability of the values led to non-significant differences. A marked decrease on the osmotic potential at full turgor (πo) there was observed on DOY 211 as the response to the interruption of irrigation on DOY 208. However, in period 3 there were significant differences (P values ≤ 0.05) between both F+ and F‒ treatments. In the case of the absolute capacitance at full turgor (CFT) there were statistical differences only on DOY 202 being higher for F+ than F‒. Chapter 4 77 Fig. 4 Seasonal courses of maximum vapour pressure deficit (VPDmax), calculated from weather records at the experimental orchard, and of the relative maximum daily turgor (Ppmin; avg ± SE) recorded on leaves both close to a cluster of fruits (F+ treatment) and far from them (F‒ treatment) (see text for details on the treatments). Differences on Ppmin between treatments were never significant (P < 0.05) (A). Seasonal courses of maximum daily stomatal conductance (gsmax) (B), CO2 net assimilation (Amax) (C), predawn water potential (Ψpd) (D) and relative water content (RWCpd) (E). All values, except VPD, are represented as avg ± SE. Asterisks indicate significant differences between treatments, at P ≤ 0.05. Asterisks are not shown when no differences were found. P2 and P3 accounts for periods 2 and 3 in which olive is most sensitive to water stress (Fernández et al., 2013; Fernández et al., 2014a). DOY = day of year. 0 2 4 6 8 10 VPDmax (kPa) Pp min (%) gs,max (mol H2O m-2s-1) 0.05 0.10 0.15 0.20 0.25 pd (MPa) -1.8 -1.6 -1.4 -1.2 -1.0 -0.8 -0.6 -0.4 DOY 2016 (200 = July 18) 160 180 200 220 240 260 280 300 RWCpd (%) 88 90 92 94 96 98 100 0 20 40 60 80 100 F+ F AN,max ( mol CO2 m-2s-1) 0 5 10 15 20 A B C D E P3 P2 The fruit effect on leaf turgor 78 Fig. 5 Seasonal courses of maximum vapour pressure deficit (VPDmax), calculated from weather records at the experimental orchard, and of the relative maximum daily turgor (Ppmin; avg ± SE) recorded on leaves both close to a cluster of fruits (F+ treatment) and far from them (F‒ treatment) (see text for details on the treatments). Differences on Ppmin between treatments were never significant (P ≤ 0.05) (A). Seasonal courses of osmotic potential at full turgor (πo; avg ± SE) (B), water potential at turgor loss point (ΨTLP; avg ± SE) (C), elastic modulus (; avg ± SE) (D), and absolute capacitance at full turgor (CFT; avg ± SE) (E). The parameters were derived from six pressure-volume curves per treatment and day. Red dots indicated parameters recorded on leaves that were close to a cluster of fruits (F+) treatment and blue dots relatively far from them (F‒) treatment (see text for details). Asterisks indicate significant differences between treatments, at P < 0.05. Asterisks are not shown when no differences were found. P2 and P3 accounts for periods 2 and 3 in which olive is most sensitive to water stress (Fernández et al., 2013; Fernández et al., 2014a). DOY = day of year. 0 2 4 6 8 10 VPDmax (kPa) Ppmin (%) MPa -2.4 -2.2 -2.0 -1.8 -1.6 -1.4 TLP (MPa) -3.2 -3.0 -2.8 -2.6 -2.4 -2.2 -2.0 -1.8 (MPa) 8 10 12 14 16 18 DOY 2016 (200 = July 18) 160 180 200 220 240 260 280 300 CFT (mol m-2 MPa-1) 0.4 0.6 0.8 1.0 1.2 0 20 40 60 80 100 F+ F P2 P3 A B C D E Chapter 4 85 The fruit effect on leaf turgor 86 Chapter 5 Leaf turgor and leaf thickness under drought stress in olive Chapter 5 89 1. INTRODUCTION In the last decades, the knowledge and awareness for a precise use of water in areas of scarcity has allowed the development of new sensors for monitoring plant water status. Among them, those relying on plant-based measurements have the advantage of informing on the plant response to the soil and atmospheric water status (Jones, 2004). The data collected with those sensors, however, are not always easy to interpret, due to the complexity of the soil-water-plant-atmosphere relationships (Jones, 2007; Fernández et al., 2014, 2018) and the adaptive responses to drought stress. Therefore, a greater understanding of the actual meaning of plant-based measurements for assessing water stress is needed for an efficient implementation of the sensors and related systems in commercial orchards. For an irrigation user, the best plant-based sensors are those which allow monitoring the plant water status from non-destructive, automatic and continuous measurements, easy to install and maintain, and implemented with transmission systems for easy access to the collected data through the internet (Fernández 2014b). Among the wide variety of available plant-based methods (Jones, 2004; Fernández 2014b; Fernández 2017), a few rely on the measurements in the leaf. This is the case of the LeafSen sensor (hereafter LS sensor; Sharon and Bravdo, 2001) which measures the diel variations of leaf thickness, and the leaf patch clamp pressure probe or ZIM probe (Zimmerman et al., 2008), that measures a variable inversely related to the cell turgor pressure (Pp; kPa) (Ehrenberger et al., 2012). When the trees have enough water, the daily Pp curves show maximum values around midday, when the daily leaf turgor is minimum due to water loses by transpiration. At night, however, Pp values are closer to zero, in correspondence with greater leaf turgor pressure. When the sampled trees are under water stress conditions, the shape of the daily Pp curve changes. For moderate water stress, half-inversed Pp curves are collected. If the water stress increases, completely inversed Pp curves appears (more details in Fernández et al., 2011, and Ehrenberguer et al., 2012). Fernández et al. (2011) and Ehrenberger et al. (2012) stablished the relationship those three shapes or “States” derived from the daily Pp curves with the plant water status in olive trees. They reported that the State 1 was usually corresponded with a negligible water stress (Ψstem ˃ ‒1.2 MPa), the State 2 with a moderate stress (within the interval between ‒1.2 MPa < Ψstem < ‒1.7 MPa) and the State 3 with severity levels of water stress (Ψstem < ‒1.7 MPa). The two mentioned leaf sensors have been proposed for irrigation scheduling (Sharon et al., 2001; Pardossi and Incrocci, 2011; Cirillo et al., 2013; Fernández et al., 2011a; Bramley et al., 2013, Fernández, 2014b; Padilla-Díaz et al., 2016; Marino et al., 2016; Martínez-Gimeno et al., 2017). The LS was tested as indicator for irrigation scheduling, for the first time, by Seelig et al. (2012), but under greenhouse conditions. The ZIM probe, however, has proved to have a potential to schedule irrigation in commercial orchards, i.e. under field conditions. Padilla-Díaz et al. (2016), for instance, were able to schedule irrigation in commercial hedgerow olive orchard from the shape of daily Pp curves derived Leaf turgor and leaf thickness under drought stress 90 from ZIM-probe readings. Both sensors, in fact, proved useful to schedule deficit irrigation strategies with which 45% and 55% of water savings, by LS and ZIM probes, respectively, were achieved (Seelig et al., 2012; Padilla-Díaz et al. 2016, 2018). Although the close correspondence between daily Pp curves and the water potential (Ψ) in the stem (Fernández et al., 2011; Ehrenberger et al., 2012), there are still gaps in our understanding on the actual turgor fluctuations and the turgor measured by ZIM probes. In the case of the leaf thickness measurements, it is known that daily fluctuations of this variable are related to transpiration (Tyree and Cameron, 1977; Seelig et al. 2012). Also, decreasing Ψ values promote leaf shrinkage due to dehydration (Sylversten and Levy, 1982; Burquez, 1987; Sancho-Knapik et al, 2010, 2011), as well as a reduction in the relative water content of the leaf (Afzal et al., 2017). Furthermore, the relationship between leaf thickness and turgor pressure have been estimated (Scoffoni et al.; 2013;Seelig et al.; 2015) but never measured with a sensor on a leaf. Hence, understanding how both leaf thickness and turgor processes are coupled is complex and far from being completely understood. It is known that a decrease in leaf turgor is a main driver of stomatal closure (Brodribb & Holbrook, 2003; Rodriguez-Dominguez et. al., 2016), causing an inhibition of photosynthesis (Lawlor & Cornic 2002; Brodribb and Holbrook 2003; Bartlett et al. 2012) and plant growth (Frensch and Hsiao, 1994). But the specific role of leaf turgor as a possible driver of other processes is still poorly understood. Cell turgor is recognized as a main driver of cell growth, being responsible for changes in cell wall properties (Lockhart et al., 1965; Cosgrove 1987, 2005, 2016; Kroeger et al. 2011). It is also involved in other processes such as water and carbon transport (Fatichi et al., 2014). A major scientific breakthrough would be the comprehension of diel and seasonal patterns in leaf turgor, which ultimately translates into growth processes (Steppe et al., 2015). The aim of this study was to determine, from the combined use of ZIM probes and LS sensors, the threshold level of leaf turgor that is driving leaf thickness growth and changes in cell wall properties using pressure-volume curves. We also aimed at determining the threshold levels of water-environmental drivers of leaf turgor, and leaf turgor as a driver of leaf growth. We hypothesized that soil water limitation constrains cell growth in leaves. To this purpose, we carried out a greenhouse experiment using the olive tree as a plant model. Several studies have been recently made on this species, involving turgor measurements (Aissaoui et al. 2016; Marino et al. 2016; Padilla-Díaz et al., 2016, 2018, Rodriguez-Dominguez et al; 2016; Fernandes et al.; 2017). We had well-irrigated trees and trees under drought-stress and recovery. For insight into the mechanisms that drive leaf turgor and thickness, we also studied the relationships between plant water potential (leaf water potential at predawn, Y pd; midday leaf water potential, Y leaf, and stem water potential, Y stem), relative leaf maximum water content (RWCleaf) at predawn and maximum leaf osmotic potential (πmax)at midday. Chapter 5 91 2. MATERIALS AND METHODS 2.1. Plant material and experimental setup The experiment was carried out with 2-year old olive seedlings (Olea europaea L. cv. Arbequina). Plants were grown at Seville, south-west Spain, and in February 2016 were sent to the greenhouse facilities at the Institute for Agricultural and Fisheries Research (ILVO) in Melle, Belgium (50° 59' 35.8" N, 3° 47' 03.1" E). Each plant was transplanted to 35-L pots, with a substratum for Mediterranean plants (AGROFINO, Peltracom, Overpelt, Belgium) which consisted in a soil mixture of peat litter (10% organic material), clay, lava and calcium-magnesium based slow-released fertilizers (NPK 14-14-18, 1.2 kg m-3). The irrigation doses were estimated from the daily cumulated plant-water consumption, assessed with Dynagage sap flow sensors (stem heat balance method, Dynamax, 2005) connected to a CR-1000 datalogger. They were estimated once every two weeks and input in an irrigation controller. Water was applied daily by a drip irrigation system providing one 2 L h-1 dripper per plant. At the beginning of the experiment, six trees of similar height and stem diameter (1.24 ± 0.03 m; 8.32 ± 0.17 mm, avg. ± sd, respectively) were selected. Three of them were maintained under well-watered soil conditions (control treatment, each tree was named as C1, C2, C3) during the whole experiment, and the other three trees were under water stress conditions(stress treatment, with trees named as S4, S5, S6). The experiment started on April 18th, when all trees presented similar values for midday stem water potential ( Y stem = -1.40 ± 0.16 MPa, avg ± sd). The stress period in the stress treatment lasted from April 22sd to May 17th, a period in which the irrigation was withheld. From May 17th to 29th, we had daily irrigation with the same amount of water as the control treatment, from the recovery of the stress. 2.2. Soil and meteorological measurements The volumetric water content (VWC; %) was monitored along the whole experimental period with one EC5 soil moisture sensor (EC5, Decagon, Pullman, WA-US) per tree. The EC5 sensors were calibrated following the manufacturer's instructions. At the same time that the calibration was carried out under soil rehydration, soil water potential (Ψsoil) was measured with two tensiometers (TensioTrans, model TT 1531, Bambach GbR Tensio- Technik, Geisenheim, Germany). A water retention curve was fitted to the values recorded during calibration. Meteorological conditions in the greenhouse simulating Mediterranean climate were fixed, and controlled by a MT/MTV sensor unit (HortiMaX). Moreover, a meteorological station was placed in between trees for measuring continuously air temperature (Ta; ºC), relative humidity (RH; %) and photosynthetically active radiation (PAR; μmol m−2s−1). Ta in the greenhouse was measured by a copper-constantan thermocouple (Omega, Amstelveen, The Netherlands) and ranged between 20 − 36 ºC. RH was measured with a capacitive RH sensor (Type HIH-3610, Honeywell, Morristown, NJ, Leaf turgor and leaf thickness under drought stress 92 USA), and ranged from 15% to 85.5%. Vapour Pressure Deficit (VPD; kPa) was calculated from Ta and RH records, with values going from 0.6 to 4.93 kPa. PAR was measured with a quantum sensor (Li-190S, LiCOR, Lincoln, NE, USA) and ranged between 0 and1570 μmol m−2 s−1. Both soil and meteorological measurements were recorded by a datalogger (CR-1000, Campbell Scientific Inc., Logan, Utah, USA) and uploaded to the PhytoSense cloud service (Phyto-IT BVBA, Mariakerke, Belgium) for its storage and processing. Sensor signals were logged every three seconds and averaged every five minutes for its storage. 2.3. Turgor and leaf thickness measurements Healthy and fully developed current-year leaves were selected to monitor turgor (from Pp values; kPa) and leaf thickness variations (µm). Sampled leaves were the 5th or 6th leaf from the apex and from the middle of the canopy. One ZIM probe (Yara Water-Sensors, YARA ZIM Plant Technology GmbH, Hennigsdorf, Germany) per tree was clamped to monitor Pp values (n = 3) . In the opposite leaf, a leaf clip sensor (LS, LeafSen, Netafim, Tel Aviv, Israel) was installed to monitor the leaf thickness (n = 3). In all cases, the central nerve of the leaf was avoided during sensor installation. The sensors were installed on March 31, 2017, early in the morning (4.00 – 5.00 GMT) when both leaf turgor and thickness were around their maximum. The recorded Pp values were sent every 5 min from the Yara ZIM-lab transmitter by USB wire to a local computer for their storage. Leaf thickness values were recorded and stored following the same methodology as detailed for the soil and meteorological measurements (Section 2.2). All sensors worked properly and none of them were reclamped. Still, we had Pp data missing from May 8 to May 15 due to a problem with the data storage in the greenhouse computer. The Pp values were normalized for comparison between treatments, according to the equation: 𝑃p (%)=𝑃p′− 𝑃pmin 𝑃pmax − 𝑃pmin 𝑥 100 , (1) where Pp′ is the current Pp value, Ppmin and Ppmax are the minimum and maximum Pp values, respectively, recorded during the experimental period. The correlation between leaf thickness variations along and air temperature, was explored after Lauriks et al. (2017). Basically, both before and after the experiment, the LS sensors were installed in an aluminium plate of 0.5 mm thickness, in the greenhouse facilities, during 4 days to be under temperature conditions similar as those during the experimental period. The aluminium plate was assumed to be thickness constant and the variations registered by the LS sensors were used to correct for their temperature dependency. The first day of the experimental period, the leaf thickness outputs were adjusted to zero in all trees, for recording relative thickness variations (m). Chapter 5 93 2.4. Plant-water relations and concomitant measurements Measurements of leaf water potential at predawn ( Y pd; MPa; 2.30 – 4.00 GMT) and at midday ( Y leaf; MPa; 12.00 – 16.00 GMT), and of stem water potential at midday ( Y stem; MPa; 12.00 – 16.00 GMT), were made every week, with a Scholander-type pressure chamber (PMS Instrument Company, Albany, Oregon, USA). One leaf per tree, of the same type as described in Section 2.3, was sampled each time. For midday stem water potential we choose leaves close to the main stem, from the inner part of the canopy, and wrapped them at least 2 hours prior to the measurements in plastic bags covered with aluminium foil. Leaf relative water content (RWCleaf; %) was calculated for one leaf per tree, in the 6 trees sampled trees and on the same days as Y pd. Leaves were taken just after the Y pd measurements (03.00 – 04.00 GMT). Again, the leaves were sampled as described in section 2.3. The leaves were immediately introduced in opaque plastic tubes, stored in darkness and taken to the laboratory for determining their fresh weight (FW; g) with a digital balance (DK 6200 with 10-3 g accuracy, Henk Maas, Veen, The Netherlands). Later, the leaves were rehydrated by immersing the petiole in distilled water at 4 ºC during 24 h, in darkness. After rehydration, the excess of water was removed by a soft cloth, and the leaves were weighted for the well-watered weight (WW; g). Then, the leaves were oven dried until dehydration (48 h at 70 ºC) and weighted for the dry weight (DW; g). RWCleaf was calculated as: RWC𝑙𝑒𝑎𝑓 (%)=FW− DW WW−DW 𝑥 100 . (2) The time courses of the maximum adjustment of leaf osmotic potential (πmax;MPa) were derived from two leaves per tree, sampled at midday (12.00 – 14.00 GMT)from the same days as described in Section 2.3. The leaves were cleaned with a damp paper towel, packed in aluminium foil and immediately submerged in liquid nitrogen. Then they were stored at -20 ºC until analysis with a thermocouple psychrometer with standard C-52 chambers (Wescor inc., Logan, UT, USA) connected to a datalogger (PSYPRO, Wescor Inc.). Two discs were punched from each leaf, avoiding the central nerve, and measured under constant temperature for at least 45 min, for sample equilibrium inside the chamber. The data recorded were averaged firstly by leaf and secondly by tree. Both during the stress period and at the end of the recovery period, the maximum stomatal resistance was measured with a porometer (AP4, Delta-T, Cambridge, UK) in the two treatments. One leaf per tree was sampled at 08.00 – 09.00 GMT, with the same criteria as described in Section 2.3. That is when maximum leaf gas exchange occurs in olive (Fernández et al., 1997). Moreover, at the end of the experiment the leaf area of each tree was measured with a LICOR 3100 leaf area meter (LI-COR, Lincoln, NE, USA). Leaf turgor and leaf thickness under drought stress 94 2.5. Pressure-volume curves Again, a leaf with the same characteristics as reported in Section 2.3 was sampled at predawn, on the same days as the other physiological variables mentioned above. Just after cutting, the leaves were rehydrated as mentioned above. Each rehydrated leaf was the used to make pressure-volume (P-V) curves, i.e. to calculate parameters derived from the relationship between the loss of leaf water potential (Ψleaf) with the loss of water volume (g; later transformed into RWCleaf) along the leaf dehydration. For this purpose, Y leaf (MPa) was measured periodically with a Scholander-type pressure chamber till ca. ‒4 MPa, at the same time that the loss of weight was determined with a digital balance. Later, the leaf area (S; m2) was calculated using the ImageJ software (Schneider et al.; 2012), from images scanned with a Canon CanoScan Lide 120 scanner, Finally, the leaves were oven-dried during at least 48 h at 70 ºC, to obtain their DW (g). For calculations, we followed the protocol published by Sack et al. (2011). The first interception point of the linear relationship ‒1/Yleaf (Y axis) vs the RWCleaf (%) (X axis) curve determined the leaf water potential at turgor loss point ( Y tlp; MPa). As a consequence, the osmotic potential at full turgor (πo; MPa) was given by the intercept of that linear relationship with the Y axis. In addition, the elastic modulus (; MPa) corresponded with the slope of the relationship ‒ 1/Yleaf vs RWC from full turgor till turgor loss point. The absolute capacitance at full turgor (CFT; mol m-2 MPa-1) was estimated till Ytlp with the following equation (Blackman and Brodribb, 2011): C 𝐹𝑇 (𝑚𝑜𝑙 𝑚−2 𝑀𝑃𝑎−1)=∂RWC𝑙𝑒𝑎𝑓 ∂𝛹𝑙𝑒𝑎𝑓 (𝐷𝑊 𝑆 )(𝑊𝑊 𝐷𝑊)1 𝑀 . (3) The parameters derived from the P-V curves were averaged by treatment. 2.6. Statistical analysis We fitted generalized additive models (GAM) using environmental variables and internal leaf water relations as explanatory variables to predict leaf thickness and Pp. For a better understanding of the relationship between RWCleaf and Pp values, Y pd and πmax, we also performed GAM. All the cases integrated smoothness estimation and were performed using the “mgcv” package (Wood, 2006) under a Gaussian family distribution with P ≤ 0.05 as significant level. Smooth terms were represented using cubic splines with smoothing parameters selected by generalised cross-validation (GCV). We used the coefficient of determination (R2) to explain the goodness of fit of the observed values vs. predicted values of the models. To analyse the differences between treatments and dates in the parameters of the internal water-relations derived from the P-V curves, we tested with a Shapiro-Wilk the normality and the homoscedasticity of the variance with a Levene’s test prior to Student's t-Test taken a P ≤ 0.05 as significant level. All the statistical analyses were made using R software (RStudio v.1.1.383; RStudio Inc., Boston, MA, USA). Chapter 5 101 Fig. 4 Relationships between soil volumetric water content (VWC; %) and air vapour pressure deficit (VPD; kPa) with predawn water potential (Ψpd; MPa) (a, d), stem water potential (Ψstem; MPa) (b, e) and leaf water potential at midday (Ψleaf; MPa) (c, f) along the whole experimental period. Group of blue dots correspond with control treatment and group of red dots with the stress treatment. Each different dot colour belongs to one tree monitored in the experiment, where the trees C1, C2 and C3 were control trees and S4, S5, S6 were stress trees. Filled dots represent the values at Ψstem and environmental parameters at predawn (a, d), dots with cross represent environmental parameters at midday compared with Ψstem (b, e), and opened dots represent environmental parameters at midday compared with Ψleaf (c, f). The Ψ threshold values fit with the lowest values found for the control treatment, being Ψpd = – Ψstem = – 1.7 MPa and Ψleaf = – 2.7 MPa, all of them are shown in the graphs as horizontal dashed lines. The threshold of VWC represents the soil field capacity, showed as vertical black dashed lines (a, b, c). Leaf turgor and leaf thickness under drought stress 102 The VWC and VPD relationships with plant water status are shown in Fig. 4. The threshold values stablished for water potentials followed the same criterion as for Fig. 3, were the lowest values for the control treatment were chosen as the limit for well-watered trees. Figs. 4a to 4c suggest that the values of Ψpd = ‒1 MPa, Ψstem = ‒1.7 MPa, Ψleaf = ‒2.7 MPa are thresholds for field capacity conditions. As expected, there was no relationship between VPD and plant water status, either for predawn or midday measurements, for both treatments (Figs. 4d, e, f). 3.3. Plant water relations as drivers of turgor and leaf thickness Leaf thickness showed a significant, non-linear relationship between water potentials at predawn and at midday (Fig. 5a, b, c) in contrast to Pp values (Fig. 5d, e, f). In Fig. 6 is represented an approach of the classical Höfler diagram, being in this case constructed from concomitant measurements in both treatments along the whole experimental period. This diagram shows changes in leaf relative water content (RWCleaf; %) together with changes in Pp values at predawn, Ψpd and maximum values of osmotic potential (πmax; MPa). These concomitant measurements were fitted using generalized additive models with integrated smoothness estimation, and the results showed significant relationships at P < 0.001. As expected, when the water stress appeared all variables declined, expected Pp values that increased. The relationship between RWCleaf and πmax was always linear. For the rest of variables, it become linear from 70-80% of RWCleaf, suggesting the point of turgor loss point could be around this interval. By contrast, towards full turgor (i.e. over the threshold of 20% for Pp and over the 94% of RWCleaf taken as threshold, as it is the lowest value found for control treatment) there were high variability including in the control treatment for all the variables measured. Chapter 5 103 Fig. 5 Relationships between leaf thickness (μm) and Pp values and predawn water potential (Ψpd; MPa) (a, d), stem water potential (Ψstem; MPa) (b, e) and leaf water potential at midday (Ψleaf; MPa) (c, f). Values of Pp and leaf thickness were taken at the same time as Ψ measurements, see Fig. 2 to identify those dial values along the whole experimental period. Group of blue dots correspond with control treatment and group of red dots with the stress treatment. Each different dot colour belongs to one tree monitored in the experiment, where the trees C1, C2 and C3 were control trees and S4, S5, S6 were stress trees. Filled dots represent the values at Ψstem and environmental parameters at predawn (a, d), dots with cross represent environmental parameters at midday compared with Ψstem (b, e), and opened dots represent environmental parameters at midday compared with Ψleaf (c, f). The Ψ threshold values fit with the lowest values found for the control treatment, being Ψpd = – 1 MPa; Ψstem = – 1.7 MPa and Ψleaf = – 2.7 MPa, all of them are shown in the graphs as vertical dashed lines. Leaf thickness threshold (a, b, c) was the zero line whilst Pp threshold (a, b) value that was selected as the lowest predawn values for control (d, e, f) both of cases those thresholds are represented as horizontal dashed lines. Grey strait lines represent the fitted generalized additive models to all dataset and the goodness of fit is represented by the coefficient of determination (R2), *** P < 0.001. No grey lines mean non-significant relationship. Leaf turgor and leaf thickness under drought stress 104 Fig. 6 A Höfler diagram approach showing the relationship between the leaf water potential at predawn (Ψpd; MPa), inversely related leaf turgor pressure (Pp; %) derived from the ZIM probes readings, the maximum osmotic potential taken at midday (πmax; MPa) for the variation of the leaf relative water content (RWCleaf) taken at predawn. The concomitant measurements were taken from different leaves and treatments along the whole experimental period. Group of blue symbols correspond with control treatment and group of red symbols with the stress treatment. Each different group symbol colour belongs to one tree monitored in the experiment, where the trees C1, C2 and C3 were control trees and S4, S5, S6 were stress trees. Crosses corresponded with Pp values, dots with values of Ψpd, whereas triangles were πmax. Dashed lines corresponded with the threshold values selected as the las value for the control treatment in all variables measured except πmax, being Ψpd = – Pp = 20 % and RWCleaf = 94%. Grey strait lines represent the fitted generalized additive models to each variable measured and the goodness of fit is represented by the coefficient of determination (R2), *** P < 0.001. Chapter 5 105 Fig. 7 Relation between leaf thickness (μm) and Pp (%) for predawn (filled dots) and midday (opened dots). Group of blue dots correspond with control treatment and group of red dots with the stress treatment. Each different dot colour belongs to one tree monitored in the experiment, where the trees C1, C2 and C3 were control trees and S4, S5, S6 were stress trees. Horizontal dashed line represents the threshold value for leaf thickness growth whereas vertical dashed line was the minimum threshold of Pp values. Leaf turgor and leaf thickness under drought stress 106 The Figure 7 shows the relationship between midday and predawn values for Pp and leaf thickness. This graph suggests that leaf growth in thickness occurred mainly at predawn and, as explained in Section 3.2, under the 20% as Pp threshold value leaf growth only occurred for two out of the tree control trees. When the stress became more severe, there were linear relationships between predawn and midday values for the stress treatment, relationships that not were found for the control treatment. In other to understand if growth were due to changes in the cell walls of the leaves, we constructed pressure-volume (P-V) curves at the beginning and at the end of the experiment. These results appear in Table 1. All the variables derived from the P-V curves are concordant with the values showed in the Höfler diagram approach (Fig. 6). At the beginning of the experiment, and for both treatments, there were not significant differences in any of the variables derived from the P-V curves (Table 1). By contrast, at the end of the experiment there were significant differences between treatments. In the stress treatment occurred an increase in the relative water content at turgor loss point (RWCtlp; %) and in the elastic modulus (; MPa), conversely the absolute capacitance at full turgor (CFT; mol m-2 MPa-1) that decreased. There were non-significant differences for both treatments for the osmotic pressure at full turgor (πo; MPa) and water potential at turgor loss point (Ψtlp), along the whole experiment. Fig. 8 Time courses of the daily Pp (kPa) values and daily leaf thickness (μm) that occurred under severe drought in the drought stress period. Pp shows the typical inversion of the daily curve that normally occurs under severe drought stress whereas the leaf thickness shows the typical shrinkage pattern. It is shown only one sensor (S6) due to a problem in the computer of the greenhouse. Chapter 5 107 The dynamics of the Pp and leaf thickness values recorded on some days in the drought stress period are shown in Fig. 8. Data for the S6 tree only are shown, because we had a problem with the computer and some values recorded on the stress period were lost. The figure shows that the daily variations of the Pp values were similar to those of leaf thickness. 4. DISCUSSION This is the first time that threshold levels of turgor, as a driver of leaf thickness in olive, were derived from the combined used of ZIM probes and LS sensors, together with the analysis of changes in cell wall properties derived from pressure-volume curves. We also determined the threshold levels of water-environmental drivers of turgor and their influence on growth processes in combination with additional measurements of plant water status, both under well-watered and drought conditions. 4.1. Environmental drivers of leaf growth and turgor Our results showed that, under well-watered conditions, the behaviour of both turgor and leaf thickness along the day was mainly influenced by VPD (Fig. 3b and d). However, when water in the soil decreased below field capacity, the main driving variable was soil water availability (Fig. 3a and c). Thus, below field capacity soil water availability started to play a major role on plant performance, being the plant response ensuring its survival by maintaining hydraulic conductivity mainly by closing stomata among other processes. These results agree with recent findings by Rodriguez-Dominguez et al. (2016), who found that the observed decline in stomatal conductance was 87% driven by turgor signals. It is well known that one of the first responses to water scarcity is stomata closing, which limits transpiration and water loss, being also a major limitation to photosynthesis (Flexas and Medrano, 2002; Flexas et al. 2004). However, olive has the ability to conduct photosynthesis under low values of water potential (Ψ) (paper Alfonso, paper Moriana, fotosíntesis con estrés; Padilla-Díaz et al., 2018). Also, the 50% of loss conductivity (P50) in olive is organ dependent (Rodriguez-Dominguez et al. 2018), presenting the roots the largest resistance for taking up water from drying soils. Also, stem P50 values is ca. Ψstem = ‒6 MPa for olive (Torres-Ruiz et al. 2017; Rodriguez-Dominguez et al. 2018; Padilla-Díaz et al., in preparation), which is in agreement with its marked resistance to drought. This low P50 value is one of the features for maximizing photosynthesis under drought stress. Leaf growth, however, did not occurred below field capacity (Fig. 3c, Fig. 7). Although photosynthesis could occur under drying soils conditions, either the rate of CO2 net assimilation was not enough to promote growth or the assimilates are used for other processes, such as fruit growth or oil accumulation, but not for leaf growth. Other authors Leaf turgor and leaf thickness under drought stress 108 have also observed that growth is affected earlier and more markedly than photosynthesis (Fatichi et al. 2013; Boyer, 1970; Hsiao, 1973; Muller et al., 2011; Tardieu et al., 2011). Our results, on Ψstem = ‒ 1.7 MPa being a threshold for water stress, supports previous findings by Fernández et al. (2011) and Ehrenberguer et al. (2012), on Ψstem = ‒ 1.7 MPa value being a limit for the change from state 2 to state 3 (Fernández et al. 2011, Fernández 2014b). Notwithstanding, Marino et al. (2016), for other olive varieties, found lower threshold values for Ψstem and its relationship with the indicators derived from the shape of the daily Pp curves. They attributed these differences to some variability between cultivars and meteorological conditions of the orchards. Thus, they suggested minimum values of Pp, which normally occur at predawn, as an alternative indicator for irrigation scheduling. Our findings show that the relationship between leaf growth in thickness and maximum turgor processes (minimum values of Pp) took place at predawn (Fig.2, 7). For growing processes, Pp values necessary have to be in the interval below 20% of maximum daily values, and Ψpd values must be over ‒1 MPa. Therefore, the threshold of Ψpd = ‒1 MPa and the limit of 20% of turgor loss at predawn could be used as indicators for irrigation scheduling purposes when maximum plant performance is pursued. Below the field capacity threshold, where no leaf thickness growth processes will take place (Fig. 3c), there is still a safety margin for irrigation scheduling of olive orchards (Fig.4a, b), because olive can keep significant plant performance under harsh conditions of water stress (Fernández et al. 2014b). We further address leaf thickness growth, turgor and plant-water relations in Section 4.3. 4.2. Internal plant water relations as drivers of turgor The classical Höfler diagram shows the extent at which changes in cell volume (measured as relative water content, RWC) are related with the water potential (Ψ), turgor pressure (Pc) and osmotic pressure (π) components in cellular water relations. diagram is showed how water potential (Ψ) and turgor pressure (P) decrease as osmotic pressure (π) increases when the relative water content (RWC) drops till incipient plasmolysis. Then, in this point, the turgor is lost and equal to zero (Ψtlp = π), moment in which the decreases in Ψ are equal to π as RWC decreases due to the symplastic water in the cell. In this study, the daily variations in leaf turgor considered as Pp values from the ZIM probes readings in different moments along the day and degree of plant-water status (measured as Ψ) were not apparently related (Fig. 5d, e, f). Although the water relations including P in cells are well known by Höfler diagrams, measurements of leaf turgor taken with ZIM probes seems that there were a not obvious relationship between Ψ and Pp (Fig. 5d, e, f), especially when the inversion of the curve appears (State 3). To understand this behaviour, we could attribute by some possible explanations not mutually exclusive, which are explained in the subsequent paragraphs. Chapter 5 109 On one hand, the accumulation of the air in the leaf tissues during dehydration (Ehrenberger et al., 2012), avoid any possible relation between stem Ψ values lower than ‒ 1.7 MPa and Pp readings. When this phenomenon of air accumulation occurs, the inversion of the daily Pp curve (State 3) appears. And even at lower stress values, i.e. during state 2, there are moments during the day at which there is no correlation between Pp and the actual water status of the plant. Taken into account other leaf Ψ such as at predawn and midday stablished as threshold values (Fig. 5d, e, f), the inversion of the curve also could occur, but only in the stress treatment. Our results from the P-V curves (Table 1) showed that Ψtlp fell into the same range for both treatments as we found as threshold value for midday leaf Ψ (Fig. 5f; Ψleaf = ‒ 2.7 MPa) in the same moment as the Ψstem (Fig. 5e; = ‒ 1.7 MPa). These results suggest that, for our olive trees cv. Arbequina, the response of the inversion of the curve could be related with a loss of leaf turgor. However, in control trees never appeared an inversion of the Pp curve although the Ψleaf taken as threshold at midday (Fig. 5) was similar as Ψtlp from the results of the P-V curves (ca. Ψtlp = ‒2.5 MPa). If we take into account the Höfler diagram, during increasing water stress there were changes in the leaf water relations (Table 1, Fig. 6), to maintain the leaf water and the Ψ before the Ψtlp. Our results showed that there was osmotic adjustment in both control and stressed plants (Fig. 6), as well as a significant increment of the elastic modulus () only in the stress treatment (Table 1). Thus, the response to water scarcity in the soil by the leaves in the stress treatment was to rise , which is related with a stiffness of the leaves. The results of Barlett et al. (2012) showed that the leaves stiffness through sclerophylly and elastic adjustments gives to the leaves the ability of retaining more water, to better tolerate severe drought stress conditions. This tolerance was explained by the combination of changes in  and the increment in the leaf relative water content at turgor loss point (RWCtlp), equal as we found for olive trees under drought stress (Table 1). Although we obtained an increment RWCtlp, we also found a decrease in the capacitance at full turgor (CFT), which means that the capacity to release water to other tissues would be reduced. This could be an effective mechanism to avoid the loss of turgor to maintain the stomatal open and to maximize photosynthesis under extreme drought conditions (Moriana et al. X; Perez-Martin et al X). Moreover, if we bear in mind that leaves are composed by multiple and different cells coordinated, our results about the inversion of the curve could not be attributed only to loss of turgor nor changes in . The different types of tissues that conform the leaves as well as other structures (e.g. trichomes) are also implicated in maintaining the leaf water status before the turgor loss point (Nguyen et al.; 2017). Such that different cell structures, changes in stiffness of the cell walls, an osmotic adjustment, and perhaps the accumulation of air in the mesophyll cells, are responses that olive leaves have to avoid water losses. Their interactions could be implicated in the response of the inversion of the Pp curve. Moreover, other biological processes occurring in the leaves, that are poorly understood, could have an influence on the Pp response. For example, the influence of the Leaf turgor and leaf thickness under drought stress 110 phloem on turgor processes in the leaves seems to be important, since xylem and phloem share their water potentials (Diaz-Espejo and Hernandez-Santana, 2017). Also, there are possible influences of the fruits, since water and photoassimilates are exported from the leaves to the sinks (fruits) (Padilla-Díaz et al, unpublished). All of these complex interactions, together with those explained in the previous paragraphs, illustrates the difficulty of explaining the inversion of the Pp curve, phenomenon that occurs only under drought stress. Going a step forward in this work, during the inversion of the Pp curve the daily dynamic of the curve was practically similar to the daily behaviour of the relative leaf thickness (Fig. 8). However, due to a problem with the computer in the greenhouse, unfortunately we had lost a great fraction of that behaviour for most of inversions of Pp curves. Still, the data set that we could conserve (Fig. 8) shows similar daily dynamics of leaf turgor and leaf thickness, although for Pp values there was more variability during the day than for leaf thickness. At night, however, both sets of readings were similar. This could be explained by changes in plant transpiration occurring during the day, that were registered by the ZIM probes but did not affect the LS sensor readings. This behaviour is also observed in Fig. 7, where leaf thickness and Pp values were related at night but not during the day, when leaf turgor pressure was low and the leaves were shrinking. Nevertheless, Fig. 7 shows a high tree-to-tree variability. Scoffoni et al. (2013) pointed out that, along leaf dehydration, the accumulation of intracellular air space was a consequence of the leaf shrinkage. This was in detriment to mesophyll hydraulic conductance in order to prevent xylem cavitation during daily transpiration. They reported a partial reversibility of the leaf shrinkage. In our case, our results on leaf thickness and the inversion of the Pp curves from living leaves showed a complete recovery (Fig. 1 and 2). Hence, we should conduct a more in-depth drought-stress study, in combination with leaf anatomy and hydraulic conductivity, to unravel the mechanistic processes that are underling the inversion of the Pp curve and its relationship with the leaf thickness. 4.3. Meaning of turgor in the plant water relations as drivers of growth Organ growth depends on cell division and cell enlargement (Cosgrove et al., 2005). Cell differentiation could also be considered as a process associated with growth, and it is subjected to complex genetic regulations. Plant growth is widely studied in pollen, being this cell type used as an organism model given its easy manipulation for experimentation and for mathematical modelling (Kroeger et al, 2017; + more references). When growth is considered at the organ level (e.g. leaf thickness growth), it has to be taken into account that the cells from the tissues in an organ are tightly regulated. That means that different kinds of tissues have differentiated cell structures, each contributing differently to leaf water relations (Nguyen et al.; 2017) and growth. Most studies on leaf thickness refer to shrinkage under drought stress, the importance novel aspect of our experimental study is that we also the leaf thickness growth. Chapter 6 117 The main purpose of this PhD was to validate of the use of the indicators for irrigation scheduling derived from the turgor related measurements in a commercial olive orchard as well as to get deeper in the fundamental understanding of the interaction between the environmental and plant mechanisms that drive these turgor measurements. In the Chapters 2 and 3, the user-friendly of water stress indicator for irrigation scheduling was firstly tested and validated in a regulated deficit irrigation strategy compared with a traditionally scheduled using the crop coefficient approach. The use of these indicators derived just on the visual analysis from the shape of the leaf turgor related measurements from the ZIM probe readings were concluded as the most promising plantbased method for automatic and continuous monitoring of water stress and its use for irrigation scheduling, in periods when the olive tree is most sensitive to water stress, in commercial olive orchards. The impact on crop performance was similar along two years under the same irrigation scheduling approach. Not only was similar in between years, but also was between the crop coefficient approach and to the 100% treatment in which were replaced a 100% of the irrigation needs. The plant water status under the use of the methodology proposed here along the whole experimental years was similar to the crop coefficient approach and similar to the control treatment in periods of high crop sensitivity to water stress. This pointed out the capacity of this methodology to supply the crop water needs in this moments of high sensitivity. Moreover, the vegetative growth was reduced, facilitating the mechanical harvesting as well as a better light interception by the canopy, ensuring a longer productive life of the orchard. In these two chapters is remarkable the 55% of water savings that we obtained following the methodology proposed in this PhD compared with the replacement of the 100% of the irrigation needs. Additionally, this method could be combined with remote sensing which allow a differential irrigation doses and frequencies depending on the water requirements identified in the orchard, contributing to the potential rational use of water for precision irrigation in super high density olive orchards. However, our irrigation scheduling approach has several limitations. On one hand, the 120% of the increment of the irrigation amounts at the beginning of the periods, during 1 to 3 days, in which the olive tree is most sensitive to water stress have never been contrasted with the crop physiology nor crop performance. Moreover, although could be reasonable the changes in the 15% of the irrigation amounts and its relation with the coefficient of variation of our orchard, this 15% might be tested in another different orchard conditions, even if our results showed that those irrigation amounts seems to be adequate for keeping non-limiting soil water conditions. Secondly, Marino et al. (2016) found for Sicilian olives different intervals of plant water status and its correspondence with the shape of the visual daily curves identified by Fernández et al. (2011). These authors attributed those intervals fundamentally to the different genotypes and the different environmental characteristics of the orchards and they proposed a new indicator for irrigation scheduling in commercial olive orchards based on the minimum values of turgor related measurements. This new indicator was explored in the Chapter 4 and 5. And thirdly, ZIM probe does not work properly beyond a certain level of water stress which, in General discussion 118 our case, was reported to be that occurring at ca. Ψstem < −1.7 MPa and the shape of the daily curve suffer an inversion. This behaviour limits the use of the ZIM probes for irrigation scheduling in periods in which the crop can reach high levels of water stress. Nevertheless, the physiological mechanisms that underlie such behaviour with turgor readings were explored in the Chapter 5. The study of other indicators derived from the ZIM probes readings such as the maximum daily turgor would be fundamental to a better comprehension about the responses under the variations of the environmental changes on the plant water status. In the Chapter 4, the environmental variables were explored as well as the proximity of fruits and changes in the elastic modulus on the maximum turgor readings. In this PhD was concluded that, under non-water limiting soil conditions, there were a daily pattern on the maximum turgor resulting from the influence of the atmospheric conditions (mainly driven by the vapour pressure deficit) and, on second place, by the water availability in the soil. The daily behaviour of maximum turgor measurements was complemented with a seasonal pattern that was explained as a result of compensating the carbon allocation from the leaves to the fruits under an equilibrium between water potential in the soil and leaf water potential. This result suggest that, after understanding the mechanisms below, this water stress indicator can become more useful to schedule irrigation on those periods when the plant water status may have a marked effect on fruit production. Also, the increase on the elastic modulus with leaf aging did not have an influence on the maximum turgor as well as the location of the leaf being close or far to the fruits. From these results, we concluded that the ZIM probes could be clamped in any leaf, regardless of being close or far from a cluster of fruits. As was mentioned above, ZIM probe does not work properly when certain level of water stress is reached, being Ψstem < −1.7 MPa in our case, where the shape of the daily curve suffer an inversion. To unravel the mechanisms that underline that behaviour, as well as the maximum turgor under drought stress and its environmental drivers, in the Chapter 5 was described an experiment under greenhouse conditions with a control treatment, in which the plant water needs was replaced, and a stress treatment with a period in which the irrigation was withhold and a later recovery. Also, the turgor related measurements were complemented with the exploration of the continuous leaf thickness measurements. In this PhD was found that, under non-limiting soil water conditions, there was an irreversible growth in leaf thickness mediated by turgor that occurred only exceeded by threshold values of the maximum turgor at Pp > 20%, Ψ = ‒1 MPa and RWCleaf = 94%. To unravel if those mechanisms attributed to growth were determined by the stiffness of the cell walls of the leaves, pressure-volume curves were performed and showed that only in the stress treatment the leaves changed its elasticity in order to tolerate lower values of water stress. These thresholds mentioned offer new opportunities for both precision irrigation and modelling about the fundamental understanding growth processes. Chapter 6 119 In the Chapter 5 also was found that the soil water availability was the main driver of the hydraulic processes that evolve leaf turgor, thickness and plant water status along the continuum soil-plant-atmosphere, as was stablished in the Chapter 4. However the novelty here was that water at field capacity as a threshold of soil water availability was the main turning point as determinant of such processes in the leaves. Under drought stress, exceeded this threshold, the inversion of the daily curve appeared and as a result of the combination of turgor loss responses with other processes such as an osmotic adjustment, an increase in the stiffness of the cell walls and other interactions that could promote the accumulation of air in the mesophyll cells to avoid water losses. Nevertheless, during the inversion of the curve the daily dynamic of the curve was similar to the daily behaviour of the relative leaf thickness which suggests that ZIM probe measure a variable related with leaf thickness during the inversion of the Pp curve. This advance opens a new door for irrigation scheduling at lower values of water potentials and, in the case of regulated deficit irrigation, in periods when the olive tree can tolerate severe conditions of water stress. General discussion 120 121 Chapter 7 Conclusions 122 123 CONCLUSIONS - The irrigation scheduling approach proposed in this doctoral thesis, based on the visual analysis from the Pp curves of the ZIM system and the three days weather forecast, allowed for an effective application of regulated deficit irrigation in a hedgerow olive orchard with high plant density. - Our irrigation scheduling approach can be used by farmers without specific training, since it is based on the State shown by the outputs from the ZIM sensors. - Two years under our irrigation scheduling approach showed a crop performance as good as the crop coefficient approach, where also no differences were found for both approaches on growth, plant water status and gas exchange. - Using our irrigation scheduling approach ca. 50% of water savings could be reached and an increment of the water productivity respect to a full irrigated treatment. - The 120% of the irrigation needs and changing by 15% the irrigation amounts applied for the rest of each period seem to be adequate for keeping non-limiting soil water conditions on those periods when we used our irrigation scheduling approach. - Under non-limiting soil water conditions, relative maximum daily turgor (Ppmin) was mainly driven by the previous daily values of atmospheric demand, followed by water availability. - There were no influence of the proximity of fruits on Ppmin records, suggesting that Ppmin readings can be made in any leaf regardless of being close or far from a cluster of fruits. - The increase on the elastic modulus with leaf ageing did not have an influence on Ppmin. - The pattern of Ppmin changed depending on the fruit developmental stage. Two stages were found to have a clearer influence: maximum rate of pit hardening and rapid fruit growth after the midsummer period. - Ppmin could be a sensitive and reliable water stress indicator on those periods when the plant water status may have a marked effect on fruit production. - Leaf turgor, thickness and plant water status along the continuum soil-plant- atmosphere were dependent on field capacity, as a threshold of soil water availability and, to a lesser extent, on the vapour pressure deficit of the air. - Under severe drought stress, the properties of the cell walls changes and the similar dynamics of the inversion of the daily Pp curve as the leaf thickness measurements suggests that the ZIM probe measures a variable related to 124 leaf thickness during the inversion of the daily Pp curve. This finding opens new possibilities for improving irrigation scheduling in periods when the olive tree is less sensitive to drought stress. 125 CONCLUSIONES - La estrategia de programación del riego propuesta en esta tesis doctoral, basada en el análisis visual de las curvas Pp de los sistemas ZIM y con las predicciones meteorológicas a tres días vista, permitió una aplicación efectiva del riego deficitario controlado en el olivar en seto de alta densidad. - La estrategia de programación del riego puede ser usada por agricultores sin que sea necesario ningún tipo de entrenamiento específico previo, únicamente basándose en el cambio de estado mostrado por las lecturas de las sondas ZIM. - Dos años bajo la programación del riego con la estrategia propuesta en esta tesis doctoral mostraron un buen funcionamiento de la planta tan bueno como el del coeficiente del cultivo, donde tampoco se encontraron diferencias para ambas estrategias en el crecimiento, estado hídrico de la planta ni en el intercambio de gases. - Con la estrategia de programación del riego se consiguieron ahorros de agua más de un 50% y un incremento de la productividad del agua respecto al tratamiento bien regado. - La aplicación de un 120% de las necesidades de riego y los cambios en el 15% en las cantidades aplicadas para el resto de cada periodo parece ser adecuado para mantener al suelo en unas condiciones no limitantes de agua. Estas condiciones son usadas para el riego en la estrategia de programación de esta tesis doctoral. - Bajo condiciones no limitantes de agua en suelo, la turgencia máxima relativa (Ppmin) fue principalmente determinada por la demanda atmosférica seguida por la disponibilidad hídrica del suelo. - No hubo diferencias en la proximidad de los frutos en los registros Ppmin , lo que sugiere que estas lecturas pueden ser usadas en cualquier hoja independientemente de si se encuentra cerca o lejos de del fruto. - El incremento del módulo de elasticidad debido al envejecimiento de la hoja no afecto al Ppmin. - El patron de Ppmin cambió dependiendo del estadío de desarrollo del fruto. Se encontraron dos fases que tuvieron una clara influencia: la tasa máxima de endurecimiento del hueso y el de rápido crecimiento del fruto que se produce a fínales de la segunda mitad del verano. - Nuestros datos apoyan al Ppmin como un indicador sensible y fiable del estrés hídrico en aquellos periodos en los que el estado hídrico de la planta puede tener una marcada influencia en la producción del fruto. 126 - Los procesos hidráulicos que envuelven la combinación de la turgencia de la hoja, el grosor de la misma y el estado hídrico de la planta a lo largo del continuo suelo-planta-atmósfera dependieron de la capacidad de campo, siendo esta última un umbral de la disponibilidad hídrica del suelo y, en menor medida, el déficit de presión de vapor del aire. - Bajo estrés hídrico severo, las propiedades de las paredes celulares de la hoja cambiaron y tuvo lugar la inversión de la curva Pp diaria como las medidas registradas por el grosor de la hoja, sugiriendo que la sonda ZIM durante la inversión de la curva de Pp mide una variable relacionada con el grosor de la hoja. Estos resultados, por tanto, abren nuevas posibilidades para la mejora en la programación del riego en los periodos en los que el olivo es menos sensible al estrés hídrico.