Sostenibilidad del olivar superintensivo Manzanilla de Sevilla

SOSTENIBILIDAD DEL OLIVAR SUPERINTENSIVO MANZANILLA DE SEVILLA Marta Sánchez Piñero TESIS DOCTORAL Presentada al Programa de doctorado Ingeniería Agraria, Alimentaria, Forestal y del Desarrollo Rural. Escuela Técnica Superior de Ingeniería Agronómica UNIVERSIDAD DE SEVILLA Sevilla, noviembre de 2024 Sevilla, noviembre de 2024 TESIS DOCTORAL SOSTENIBILIDAD DEL OLIVAR SUPERINTENSIVO MANZANILLA DE SEVILLA Marta Sánchez Piñero Dirigido por: Mireia Corell González y Mª José Martín-Palomo García Agradecimientos ¡Benditas vueltas y oportunidades que da la vida! Cuando terminé mi trabajo fin de máster y escribí los agradecimientos pensé que serían los últimos, al menos en cuanto a estudios. Sin embargo, con el devenir de la vida, unos años más tarde, me encuentro de nuevo en la misma tesitura. Tengo mucho que agradecer por muchísimas cosas. Tantas que, difícilmente seré capaz de reflejar en lo que serán unas breves palabras de reconocimiento y admiración hacia una serie de personas que de una manera u otra han hecho posible que yo, en mis frecuentes estados de inseguridad, introversión y falta de empatía, pudiese realizar este trabajo. Por tanto, es justo reconocerlo en estas líneas. En primer lugar, quiero dar las gracias a mis directoras de tesis, Mireia Corell y M.ª José Martín-Palomo. Gracias por acogerme desde el primer momento, por creer y confiar en mí, por protegerme y por saber guiarme. A Mireia, por su orden dentro del desorden, por su capacidad de estar en todo y con todos, incluso en las adversidades. A M.ª José, por esa calma y tranquilidad que le caracteriza, consiguiendo el equilibrio necesario en el grupo. A Alfonso Moriana por su inagotable energía, entusiasmo y decisión. Alfonso no deja para mañana lo que puede hacer hoy, de hecho, algunas veces intenta hacer hoy lo que debería dejar para mañana. A Noe, por nuestras charlitas y cotilleos mañaneros, nuestras primeras aceitunas aliñadas y sobre todo por tu ayuda. Gracias ‘burbujas’ por brindarme la posibilidad de realizar este trabajo. Ha sido un placer y un orgullo aprender de personas tan inteligentes, resueltas, diligentes y comprometidas. En ocasiones, he llegado a pensar que tenía más de un padre y una madre. Gracias al CSIC, concretamente a todo el personal que trabaja en la Finca experimental ‘La Hampa’. Ignacio, Oscar y todo el equipo de recolección. Gracias por vuestra ayuda desinteresada y por permitirme realizar este trabajo en vuestras instalaciones. A mis amigos, por su apoyo y amistad, por creer e impulsarme siempre a seguir, por acompañarme en cada paso y celebrar siempre mis triunfos. El mundo, los problemas y las preocupaciones son posible arreglarlos con cervezas y vosotros. A casa Pirelli, sus individuas y el desorden que le caracteriza. Por nuestras conversaciones eternas e innumerables consejos. A mis padres y a mis hermanos, gracias por vuestra paciencia, por enseñarme el valor de la responsabilidad y por poder tener siempre la tranquilidad de que cuento con vuestro respaldo. A todos, gracias. No puedo estar mejor rodeada. Marta. Sevilla, noviembre de 2024 TESIS DOCTORAL SOSTENIBILIDAD DEL OLIVAR SUPERINTENSIVO MANZANILLA DE SEVILLA Marta Sánchez Piñero SANCHEZ PIÑERO MARTA - Firmado digitalmente por SANCHEZ PIÑERO MARTA – Fecha: 2024.10.31 14:49:08 +01'00' Mireia Corell González Mª José Martín-Palomo García CORELL GONZALE Z MIREIA - Firmado digitalmente por CORELL GONZALEZ MIREIA - Fecha: 2024.10.31 14:11:08 +01'00' MARTINPALOMO GARCIA MARIA JOSE - Firmado digitalmente por MARTIN-PALOMO GARCIA MARIA JOSE - Fecha: 2024.10.31 14:43:21 +01'00' Publicaciones La presente Tesis Doctoral se presenta como compendio de publicaciones, las que se citan a continuación: 1. Sánchez-Piñero, M., Martín-Palomo, M.J., Moriana, A, Corell, M., Pérez-López, D., 2022. Endocarp development study in full irrigated olive orchards and impact on fruit features at harvest. Plants 11, 3541. https://doi.org/10.3390/plants11243541 Journal Impact Factor (JIF): 4.50 Plants Sciences: 43/239 Cuartil Q1, Tercil T1, Decil D2 2. Sánchez-Piñero, M., Corell, M., de Sosa, L.L., Moriana, A., Medina Zurita, N., Madejón, E., Girón, I., Castro-Valdecantos, P., Martín-Palomo, M.J., 2024. Assessment of water stress impact on olive trees using an accurate determination of the endocarp development. Irrigation Science, 42 (2). https://doi.org/10.1007/s00271-024-00914-w Journal Impact Factor (JIF): 3,10 Agronomy: 22/125, Cuartil Q1, Tercil T1, Decil D2 FWCI: 1,50 3. Sánchez-Piñero, Marta; Corell, Mireia; Moriana, Alfonso; Pérez-López, David; de Sosa, Laura L; Medina-Zurita, Noemí; Castro-Valdecantos, Pedro; Martín-Palomo, María José. 2024. Response of young super-high density table olive orchard (Manzanilla de Sevilla) to different water stress levels considering an accurate determination of endocarp development. Agricultural Water Management, 303, 109044. http://doi.org/10.1016/j.agwat.2024.109044 Journal Impact Factor (JIF): 5.9 Agronomy: 6/125, Cuartil Q1, Tercil T1, Decil D1 4. Sánchez-Piñero, M., Corell, M., Moriana, A., Castro-Valdecantos, P., Martín-Palomo, M.J., 2023. Managing water stress in olive (Olea europaea L.) orchards using reference equations for midday stem water potential. Horticulturae, 9, 563. https://doi.org/10.3390/horticulturae9050563 Journal Impact Factor (JIF) :3.10 Horticulturae: 7/38 Cuartil Q1, Tercil T1, Decil D2 FWCI: 1,10 5. Sánchez-Piñero, M., Martín-Palomo, M.J., Andreu, L., Moriana, A., Corell, M., 2022. Evaluation of a simplified methodology to estimate the CSWI in olive orchards. Agricultural Water Management, 269, 107729, 1 p.9. https://doi.org/10.1016/j.agwat.2022.107729 Journal Impact Factor (JIF): 6.70 Agronomy: 3/88 Cuartil Q1, Tercil T1, Decil D1. FWCI: 1,02 Capítulo 5 Figura 1 Climatic conditions throughout the experiment during the 2020 season. Pattern of potential evapotranspiration (ETo) and rain ............................................................. 145 Figura 2 Climatic conditions throughout the experiment during the 2021 season. Pattern of potential evapotranspiration (ETo) and rain ............................................................. 146 Figura 3 Relationship between the crop water stress index estimated with Seasonal fit (several equations) vs Daily fit (one equation) ............................................................. 151 Figura 4 Pattern in the 2020 season CWSI estimated with Station data, midday stem water potential, CWSI estimated with Thermohygro data ..................................................... 154 Figura 5 Pattern in the 2021 season CWSI estimated with Station data, midday stem water potential, CWSI estimated with Thermohygro data ..................................................... 156 Figura 6 Midday leaf conductance during the 2020 and the 2021 seasons .................. 157 Figura 7 Relationship between CWSI and midday stem water potential data of Figs. 4 and 5 .................................................................................................................................... 158 Índice de tablas Introducción Tabla 1 Distribución de la superficie (hectáreas) y producción de aceitunas (toneladas) en España según el destino de la producción ...................................................................... 22 Tabla 2 Distribución de la superficie y producción de aceitunas por Comunidades Autónomas ...................................................................................................................... 23 Tabla 3 Resumen del efecto del déficit hídrico en las Fases I-Fase IV del olivar .......... 32 Capítulo 1 Tabla 1 Values of pit-breaking pressure (PBP) parameters and means and variability for the different varieties of measurement, according to the equation PBP ......................... 42 Tabla 2 Characterization of Period I of endocarp development in different cultivars and locations: beginning (full bloom date), length (days), and thermal time (º days) according to three different models ................................................................................................. 43 Tabla 3 Correlation coefficient and number of data in the relationships between relative weight and volume at the end of Period I of endocarp development with thermal time models ............................................................................................................................. 44 Tabla 4 Characterization of Period II of pit-hardening period in different cultivars and locations: length (days), and thermal time (º days) according to three different models (lineal, exponential and cosine) ...................................................................................... 47 Tabla 5 Best fit of the relationships between dry weight and volume at Period 1 and harvest ............................................................................................................................. 50 Tabla 6 Best fits of the relationships between dry weight and volume at Period 1 vs. at harvest in Manzanilla experiments ................................................................................. 50 Tabla 7 Comparison of the two locations where data of the experiments were obtained. Data are the average of monthly data in the period 2006 to 2021 .................................. 56 Capítulo 2 Tabla 1 Parameters (average ± standard error) of the fit adjusted in the individual pitbreaking pressure curves ............................................................................................ 72 Tabla 2 Characterization of Period I and II of endocarp development in the different treatments ....................................................................................................................... 73 Tabla 3 Water status indicators during periods pre-flowering, I, II and III of endocarp development in 2021 and 2022 seasons ......................................................................... 74 Tabla 4 Best fits in the relationship between oil accumulation rate at Period II (R2) or Period III (R3) vs. water status indicators ...................................................................... 80 Capítulo 3 Tabla 1 Parameters of the fit adjusted in the individual pit-breaking pressure curves for the 2023 season. Sigmoid equations for modelling the pit-breaking pressure PBP 100 Tabla 2 Minimum midday stem water potential and Stress Integral in pre-flowering and the three phenological phases according to endocarp development throughout the 2023 season ........................................................................................................................... 102 Tabla 3 Best fits in the relationship between inflorescence set and several water status indicators and average maximum temperature in the period from full bloom until pit hardening ...................................................................................................................... 110 Tabla 4 Quantity and quality of yield, applied water, and percentage of survival trees in the four seasons of the experiment ............................................................................... 113 Tabla 5 Several fruit features related with yield quality ............................................... 114 Capítulo 4 Tabla 1 Best fit of the relationship between midday stem water potential (SWP) vs. maximum daily temperature or vapor pressure deficit ................................................. 134 Capítulo 5 Tabla 1 Water applied (mm) during the 2020 and 2021 seasons ................................. 148 Tabla 2 Best fit of the relationship between the CWSI estimated using a non-water stress baseline calculated with one equation (Daily, D) or considering different parts of the season (Season, S) ........................................................................................................ 151 Tabla 3 Best fit of the relationship between the CWSI estimated with a daily approach using the climatic data from a nearby station or field Thermohygro ........................... 152 INTRODUCCIÓN INTRODUCCIÓN 20 24.60% % 23.60% 1. Introducción 1.1. Importancia del cultivo del olivo El cultivo del olivo (Olea europea L.) está presente en más de 40 países distribuidos a lo largo de todo el mundo, mientras que sus productos derivados se consumen en un total de 179 países. Esto nos muestra la importancia del sector oleícola en el contexto internacional. Aunque la producción está repartida en los cinco continentes, Europa lidera la producción mundial de aceituna con un 47,40 % frente al resto de continentes (FAOSTAT, 2024) (Fig. 1). 0.50% 47.40% 3.90 Europa Asia America Africa Oceania Figura 1: Porcentaje de producción de aceitunas en el mundo en 2022. Fuente: FAOSTAT 2024 La producción se concentra principalmente en la cuenca Mediterránea, debido a que las condiciones climáticas y del terreno lo favorecen. Además, es un cultivo tradicional en esta zona donde es considerado una especie de gran interés socioeconómico. España es líder mundial en producción y exportación de aceituna, con 3.940.0000 toneladas (MAPA, 2024), le sigue Turquía con 2.976.600 toneladas INTRODUCCIÓN 21 producidas e Italia con 2.160.400 toneladas (Figura 2). Siendo Arabia Saudí el único país de los diez principales productores que no forma parte de la cuenca Mediterránea. Figura 2: Principales diez países productores de aceituna. Fuente: FAOSTAT 2022. El sector oleícola mundial ha mantenido un crecimiento continuado durante las últimas décadas, tanto en producción como en superficie (Figura 3). La superficie de olivar cultivado ha experimentado un incremento muy importante, aunque a diferentes tasas. En 1961, la superficie mundial de olivar era de 2.608.804 hectáreas, en 2022 la superficie de olivar llegó a 10.948.521 hectáreas, por lo tanto, en un periodo de 61 años, la superficie mundial ha crecido un 319,68%. Este incremento ha sido especialmente importante a partir del último cuarto del siglo XX (Figura 3). Lo mismo ocurre con la producción, en este caso ha pasado de 8.205.586 toneladas en 1961 a 21.449.867 toneladas en 2022, lo que supone un aumento del 161,41%. En este caso la mayor tasa de incremento de producción tiene lugar a partir de finales de los años 90 del siglo pasado. En este caso, es habitual una evolución en dientes de sierra por la vecería del cultivo, pero incluso con este patrón la producción mundial se ha ido incrementando, manteniéndose todavía en aumento. INTRODUCCIÓN 22 Figura 3: Evolución de la producción (toneladas) y área cosechada (hectáreas) de aceitunas en el mundo. Fuente: FAOSTAT, 2022. A nivel nacional, el olivar abarca 2.635.276 hectáreas, de las cuales 154.643 hectáreas corresponden a olivar de mesa y 2.480.633 hectáreas a olivar de almazara (Tabla 1, MAPA 2024). El olivar de mesa representa un 6 % del total de superficie y se estima que el 81% corresponde a olivar en secano. El 94% de la superficie de olivar en España se destina a aceituna de almazara, considerándose un pilar fundamental del sector agroalimentario español (MAPA, 2024). En el caso de este tipo de aceitunas, la importancia del riego es mayor en valor absoluto (más de 500.00 ha) aunque en valor relativo está alrededor del 22%, un porcentaje similar a la aceituna de mesa. Estos datos suponen que el olivar es uno de los frutales en regadío más importante de España a nivel de superficie cultivada. Tabla 1: Distribución de la superficie (hectáreas) y producción de aceitunas (toneladas) en España según el destino de la producción en 2022. Fuente: MAPA, 2024. Destino Superficie en producción (ha) Producción (t) Secano Regadío Total Olivar de aceituna de mesa 125.867 28.776 154.643 262.276 Olivar aceituna almazara 1.917.653 562.980 2.480.633 3.677.791 Total, España 2.043.520 591.756 2.635.276 3.940.067 INTRODUCCIÓN 23 A nivel regional, la mayor producción se localiza en Andalucía con el 62 % del total, seguida de Castilla la Mancha con el 15% (Tabla 2). Dentro de Andalucía, Jaén, como principal provincia productora, concentra aproximadamente el 22% de la producción total nacional (MAPA, 2024). Tabla 2: Distribución de la superficie (hectáreas) y producción de aceitunas (toneladas) por Comunidades Autónomas (CCAA) 2022. Fuente: MAPA, 2024. CCAA Superficie (ha) Producción (t) Andalucía 1.645.659 3.034.653 Castilla la Mancha 387.234 361.704 Extremadura 271.404 290.240 Cataluña 106.194 73.788 Valencia 93.151 44.699 Este crecimiento potencial que hemos visto previamente, tanto en superficie como en producción en el olivar puede justificarse con los avances en las diferentes técnicas de cultivo, reflejando una modernización de las prácticas agrícolas. El incremento de las densidades de plantación y la recolección mecanizada simbolizaron un proceso de cambio hacia una olivicultura mucho más tecnificada. Las primeras plantaciones superintensivas, o en seto, se realizaron en España en 1995 y provocaron una verdadera revolución en un cultivo que había permanecido prácticamente inalterable en los últimos siglos (Rius- García et al., 2015). La difusión de estas plantaciones superintensivas ha ido más allá de los países tradicionalmente productores de aceite de la cuenca Mediterránea (Italia, Grecia, Portugal, Marruecos o Túnez) y hoy en día es el sistema de cultivo habitual en países como Estados Unidos, Chile o Australia. Sin embargo, España lidera la olivicultura superintensiva a nivel mundial con alrededor de 45.000 hectáreas (Rius-García et al., 2015) de las cuales, el 99% se encuentran en regadío. En los últimos años han proliferado las plantaciones superintensivas de secano que compiten en rentabilidad con las producciones obtenidas en estas condiciones por otros cultivos como el cereal o el viñedo (Roca, 2017). En zonas donde el agua es un recurso limitado, el olivar superintensivo en secano podría representar una opción viable para la producción de aceite. Sin embargo, son grandes los desafíos a los que se enfrenta este tipo de plantaciones superintensivas en INTRODUCCIÓN 24 secano para la producción de aceituna de mesa, dada la limitación de la disponibilidad de agua y la importancia de la calidad de la aceituna. 1.2. Riego Deficitario Controlado en Olivar. El olivo ha sido una especie cultivada tradicionalmente en condiciones de secano dada su capacidad de resistencia a periodos de sequía intensa. Aunque en muchos países productores este sistema sigue siendo el mayoritario, desde los años 90 del siglo pasado, la superficie de este cultivo en regadío ha ido aumentando en todo el mundo. No obstante, en la mayor parte de los países, y muy especialmente en los de olivicultura más tradicional, la cantidad de agua disponible para el cultivo ha sido limitada y habitualmente inferior a las necesidades potenciales. Esta restricción en la disponibilidad de agua supone el manejo de periodos más o menos largos de estrés hídrico en el árbol. El riego deficitario controlado (RDC) se comenzó a desarrollar a mediados del siglo pasado en melocotón y consiste en una programación de riego en el que los periodos de estrés hídrico se seleccionan en función de la fenología del árbol, para minimizar el efecto en la cosecha (Chalmers y Wilson, 1978). El RDC ha sido ampliamente estudiado en muchas especies frutales (Steduto et al., 2012) y su respuesta está relacionada con el nivel, el momento y la duración del estrés hídrico (Hsiao, 1990). La respuesta fisiológica del olivar permitiría un ahorro potencial de agua muy elevado al ser capaz de obtener agua y de sobrevivir en condiciones extremas. Esta especie es extremadamente resistente a la desecación, pudiendo reducir su potencial hídrico foliar a valores muy negativos (Orgaz y Fereres, 2008). Además, cuenta con un sistema radicular profundo y extenso con el que puede explorar horizontes profundos del suelo (Orgaz y Fereres, 2008) y a la vez, presenta hojas coriáceas, con pelos aparasolados en el envés y un fuerte control estomático que limita la traspiración (Rapoport, 2008). Sin embargo, la respuesta productiva del olivar al estrés hídrico es muy diferente en función del momento en el que se realice, niveles que podrían considerarse moderados en floración resultaron ser extremos en cuanto a reducción de cosecha (Moriana et al., 2003). Esta respuesta ha determinado las funciones de producción obtenidas en olivar. La sugerida por Steduto et al. (2012), que tiene como referencia los valores de Moriana et al. (2003), resultó ser muy restrictiva, con una gran respuesta en cosecha al incrementar la Evapotranspiración del cultivo (ETc), mientras que cuando se evitaban estas condiciones de estrés hídrico en primavera, el incremento era menor (Ben- Gal et al. 2021). INTRODUCCIÓN 25 1.3. Importancia de cada fase fenológica. La determinación y el estudio de cada una de las fases fenológicas para poder aplicar un estrés hídrico suele ser poco precisa en los ensayos de riego en campo. Habitualmente sólo se considera un periodo de tiempo fijo, sin determinaciones en el árbol (por ejemplo, Goldhamer, 1999). Sin embargo, una determinación precisa puede conseguir un ahorro sustancial de agua cuando se elige y aplica correctamente una estrategia de RDC, sin mermar el rendimiento y, en ocasiones, mejorando la calidad (Moriana et al., 2003; Tognetti et al., 2006; 2008). Por el contrario, una estrategia RDC mal gestionada puede causar un grave déficit hídrico en las fases en las que el cultivo es más sensible, reduciendo tanto el rendimiento del año en curso como del año próximo (Moriana et al., 2003; Fereres y Evans, 2006). Las distintas fases fenológicas del olivar son determinantes para establecer la respuesta vegetativa y productiva de la planta. Estos periodos se basan especialmente en el desarrollo del fruto, de manera similar a como se describe en otros frutales, al ser la parte del árbol de mayor interés en cosecha (Figura 4). De manera precisa, las fases fenológicas del olivo incluirían periodos de tiempo en los que no está presente el fruto. Aunque estos periodos suelen coincidir con momentos en los que el riego no es necesario o no está disponible, su influencia en cosecha podría ser de interés. INTRODUCCIÓN 32 de la aridez del clima de la cuenca Mediterránea da lugar a que la Fase I de prefloración, y la Fase IV de acumulación de aceite, puedan tener un papel fundamental en la determinación de la respuesta en cosecha y en la gestión de la disponibilidad del agua de riego. En cualquier caso, la determinación del desarrollo del endocarpo es muy importante para establecer de forma clara los efectos del estrés hídrico sobre la cosecha. En la Tabla 3 se muestra un resumen del efecto del déficit hídrico en cada una de las fases mencionadas. Tabla 3: Resumen del efecto del déficit hídrico en las Fases I-Fase IV del olivar FASE RESUMEN FASE UMBRALES DE POTENCIAL HÍDRICO EFECTO EN COSECHA Fase I (Prefloración) Cambios de yema axilar a inflorescencia, y formación de las flores. Importante crecimiento vegetativo. ≤-2 MPa puede reducir el número de flores por inflorescencia y % de flores perfectas. Aumento del aborto pistilar, causando caída de frutos. Fase II (Floración- Cuajado) Desarrollo del fruto y endocarpo. Importante división celular en el mesocarpo. <-2 MPa al amanecer <-3 MPa al mediodía, provocan caídas de frutos y reducen el tamaño final del endocarpo. Disminución del cuajado de frutos y tamaño del endocarpo, afectando al tamaño final del fruto y cosecha Fase III (Endurecimiento masivo del endocarpo) Periodo más resistente al estrés hídrico. Se detiene el crecimiento vegetativo mientras el fruto continúa creciendo en volumen <-4 MPa pueden reducir la cosecha hasta un 30%, -2 < -4 MPa disminuyen la cosecha en 10-20%. Crecimiento más lento del fruto. Las pérdidas se asocian tanto al tamaño del fruto como a la caída de estos. Fase IV (Acumulación aceite) Acumulación de aceite ≥-2 MPa, la acumulación de aceite no se ve afectada. El estrés hídrico puede aumentar el rendimiento de aceite INTRODUCCIÓN 33 1.4. Medidas estado hídrico de la planta Una parte fundamental en la programación del riego deficitario controlado es la determinación del nivel y de la duración del estrés hídrico (Hsiao, 1990). Sólo recientemente han ido apareciendo publicaciones en las que esta parte de la programación se está estudiando, más allá del uso de un porcentaje de la evapotranspiración del cultivo (ETc), por ejemplo, en el trabajo de Moriana et al. (2012). El uso del estado hídrico en las decisiones de riego permite establecer un ajuste de las necesidades hídricas en función de las condiciones edafoclimáticas en las que se desarrolla el cultivo. Se están desarrollando una gran cantidad de metodologías de medida en continuo en planta, que están incrementando cada vez más el número de indicadores a emplear, por ejemplo, los sensores Zim (Zimmermann et al., 2008)). Esto, en ocasiones, hace más difícil aún la evaluación del riego, puesto que son indicadores que aún no están lo suficientemente probados a nivel de campo, pero por otro lado, están mejorando el uso del estado hídrico como herramienta de programación del riego. Muchos de los trabajos de riego en olivar no realizan la programación de RDC basada en estas metodologías, sino que las usan como una forma de describir la estrategia real de riego que es el uso de un porcentaje de la ETc (por ejemplo, Fernández et al., 2013; Gómez del campo, 2013). Habitualmente, las medidas empleadas no son en continuo y la frecuencia de medida no es elevada, por ejemplo una medida al mes, lo que provoca que sólo se consideren umbrales de medidas puntuales, como hemos descrito en el apartado anterior. El efecto de la duración de un cierto nivel de estrés hídrico necesitaría, además, una referencia que pudiese servir de comparación, más allá del uso de un tratamiento Control que podría no estar en condiciones óptimas. Todo esto supone que, además de las limitaciones que tenemos por usar todos los dispositivos que hay en la actualidad y los que se crearan con seguridad en un futuro, sea necesario que podamos definir con un indicador el efecto o no de la acumulación del estrés hídrico. Recientemente, varios trabajos sugieren el efecto de un estrés acumulado sobre la cosecha en olivar, empleando para ello la integral de estrés (IE) del potencial hídrico (Corell et al., 2020; Corell et al., 2022). La IE en estos trabajos es definida como una estimación de la superficie entre un umbral y la curva de evolución del potencial hídrico. Originalmente, la IE tomaba como umbral de referencia el valor más positivo de los medidos (Myers, 1988). Sin embargo, esto dificulta la comparación entre estaciones ya que ese valor podría no ser una referencia adecuada en algunos de ellos. Los trabajos de Corell (Corell et al., 2020; Corell et al., 2022) emplean como referencia un valor de potencial hídrico en hoja cubierta a INTRODUCCIÓN 34 mediodía descrito en trabajos previos como no limitante (-1,2/-1,4 MPa según Moriana et al., 2012). Una de las limitaciones de la medida del estado hídrico para programar el riego es su relación con el ambiente. Esto supone que, a igualdad de estado hídrico, la demanda evaporativa de lugar a valores diferentes. Estas variaciones dan lugar a problemas de interpretación y suponen la necesidad de tener una línea de referencia. En los últimos años se han sugerido varias ecuaciones de referencia en olivar para el potencial hídrico (Corell et al., 2016; Shackel et al., 2021). El uso combinado de estas ecuaciones y la ecuación Myers (1988) podría dar una mayor precisión al estudio de la duración y el nivel de estrés hídrico en la respuesta a cosecha. El uso de estas ecuaciones de referencia también podría permitir un cambio en la descripción de los niveles de estrés hídrico en RDC, empleando, por ejemplo, en lugar de los valores absolutos de potencial hídrico, la distancia de estos a las ecuaciones de referencia. El uso de ecuaciones de referencia o líneas base es relativamente común en alguna de las medidas del estado hídrico sugeridas como herramientas de programación (Temperatura de cubierta, Idso et al., 1981; dendrometría, Goldhamer y Fereres, 2001). Estas referencias permiten corregir el efecto de la demanda ambiental en el indicador que se está considerando. Sin embargo, la misma ecuación de referencia no siempre es válida en ambientes y/o variedades diferentes. Ortuño et al. (2010), en una revisión sobre dendrometría, presenta diferentes líneas base dentro de la misma especie o incluso en el ciclo del cultivo, por ejemplo, en almendro. Esto pone en evidencia un problema de gran importancia para poder emplear estas herramientas en programación de riego deficitario. Un ejemplo podría ser la temperatura de cubierta como indicador que ha tenido un gran uso en olivar en los últimos años por la aparición de drones y de imágenes satélite (Berni et al., 2009; Egea et al., 2017). Su respuesta está basada en el intercambio gaseoso de los árboles, lo que podría suponer un indicador tardío en la identificación del estrés hídrico (Moriana y Fereres, 2002). Egea et al. (2017), concluyeron que en olivar la línea de referencia cambiaría a lo largo del día y también a lo largo del año. Estas variaciones harían aún más difícil su interpretación, añadiendo más limitaciones a las del coste, frecuencia de medida y precisión, en la determinación de la superficie a considerar. El uso de medidas del estado hídrico de la planta podría mejorar la precisión a la hora de decidir los niveles y la duración del estrés hídrico impuesto en una programación en RDC. Sin embrago, aunque hay cada vez más publicaciones empleando diferentes INTRODUCCIÓN 35 indicadores, no está claro que las líneas de referencia que sería preciso emplear sean extrapolables. 36 OBJETIVOS 37 OBJETIVOS 2. Objetivos El objetivo general de esta Tesis Doctoral fue el de estudiar el efecto de la duración y el nivel de estrés hídrico sobre la cosecha de olivar determinando de forma precisa el desarrollo del endocarpo. Este objetivo general lo hemos dividido en varios objetivos parciales: 1. Estudiar la evolución de la dureza del endocarpo en condiciones no limitantes de agua y en varios cultivares para establecer la duración de las diferentes fases de desarrollo. Este objetivo se aborda en el Capítulo 1 de esta Tesis Doctoral. 2. Describir el efecto del estrés hídrico en el desarrollo del fruto (crecimiento y acumulación de aceite) en un rango amplio de estados hídricos en el cv Manzanilla de Sevilla. Este objetivo se aborda en el Capítulo 2 de esta Tesis Doctoral. 3. Caracterizar el efecto del estrés hídrico en la cosecha y en sus componentes en el cv Manzanilla de Sevilla. Este objetivo se presenta en el Capítulo 3 de esta Tesis Doctoral. 4. Estudiar la variación del potencial hídrico en hoja cubierta medido al mediodía respecto a las condiciones ambientales en diferentes niveles de estrés hídrico para establecer los posibles cambios debidos a las condiciones en las que se encuentra el cultivo y validar la respuesta con las líneas base publicadas. Este objetivo se presenta en el Capítulo 4 de esta Tesis Doctoral. 5. Evaluar el uso de la temperatura de cubierta en olivar empleando un sistema de medida de bajo coste y las ecuaciones sugeridas previamente en otras publicaciones. Este objetivo se presenta en el Capítulo 5 de esta Tesis Doctoral. 38 CAPÍTULO I: Endocarp Development Study in Full Irrigated Olive Orchards and Impact on Fruit Features at Harvest CAPÍTULO I 39 3. Capítulo I: Endocarp Development Study in Full Irrigated Olive Orchards and Impact on Fruit Features at Harvest Publicado en Plants 11, 3541. https://doi.org/10.3390/plants11243541 El desarrollo del endocarpo en olivos comprende tres períodos: crecimiento (Período I), esclerificación (Período II) y endurecimiento máximo (Período III). Los dos primeros están muy relacionados con el rendimiento y la gestión del riego. El Período I coincidió con la división celular del mesocarpo y, por lo tanto, con el tamaño final de los frutos. El Período II se considera la etapa fenológica más resistente a la sequía. Sin embargo, se sabe poco sobre la duración de estos períodos y su capacidad para predecir el tamaño final de los frutos en cosecha. El objetivo de este trabajo fue evaluar la duración de ambos períodos en diferentes variedades y diferentes localizaciones en plantaciones de olivar totalmente regadas. También se ha intentado estudiar el impacto de las características de los frutos en la cosecha al final del Período I. Se utilizaron datos de olivares totalmente regados de cv Cornicabra, Arbequina y Manzanilla en dos localidades diferentes (Ciudad Real y Sevilla). El patrón de presión de rotura del endocarpo se midió a lo largo de la temporada con muestras de frutas de varios años (2006 a 2022). Estos datos y los datos climáticos se utilizaron para comparar diferentes métodos de estimación para la duración del Período I y II del desarrollo endocarpo. El volumen de fruta y el peso seco al final del Período I se utilizaron para estimar las características de la fruta en la cosecha. Los resultados sugieren que la duración del Período I fue menos dependiente de la temperatura y la variedad de lo esperado. La duración de este período fue casi constante, alrededor de 49 días después de la plena floración. El tiempo térmico se correlacionó negativamente con el tamaño de la fruta al final del Período I. Por el contrario, un modelo térmico lineal presentó la menor variabilidad al estimar la duración del Período II, que también se vio afectada por la variedad. El mejor ajuste entre peso seco y volumen de fruta en el Período I vs. cosecha fue único para las variedades de aceite (Cornicabra y Arbequina), mientras que cv Manzanilla presentó una relación diferente. Un aumento de la temperatura en el futuro no afectaría a la duración del período I, pero reduciría el tamaño de los frutos al final de este período y en la cosecha. CAPÍTULO I 40 1. Introduction Olive (Olea europaea L.) is a common rainfed fruit tree in the Mediterranean basin. However, from the 1990s, the increase in growing surface in this zone and around the world has been commonly achieved using irrigated orchards. Although this species is considered one of the most drought-resistant, water needs could still be very high (Gucci et al., 2012). In these new orchards, deficit irrigation managements were common because farmers’ available water was lower than the water needs. Regulated deficit irrigation works suggested that endocarp development was a critical period for water management. Endocarp hardening was reported as the most drought-resistant phenological period (Goldhamer et al., 1999). On the other hand, fruit set, coincident with endocarp growth (Rapoport et al., 2013), was determined to be very sensitive and related with reduction in fruit size of the current season and yield in the next season (Fernandez-Escobar et al., 1992; Gomez del Campo et al., 2014; Gucci et al., 2019). However, endocarp development was not commonly measured under field conditions. The determination of the endocarp-hardening period during irrigation scheduling under field conditions and, also, in scientific works was not accurate. Commonly, fixed dates around July (North)/December (South hemisphere) were considered for different climatic conditions (e.g., (Goldhamer et al., 1999; Lavee et al., 2007; Ahumada-Orellana et al., 2017). Fruit longitudinal size could be used to estimate the beginning of the endocarp-hardening period (Rapoport et al., 2013). This indicator has been used under field conditions for regulated deficit irrigation scheduling works with successful results in table (Moriana et al., 2013; Girón et al., 2015; Corell et al., 2020) and oil olive orchards (Corell et al., 2022). However, endocarp development was described as a dynamic process of three different periods (Rapoport et al., 2013), of which endocarp hardening is the second one. Ref. (Hammani et al., 2013) reported that the sclerified area in the endocarp increased from full bloom, first gradually, then abruptly at the end of the endocarp expansion. These periods were also reported with the pit-breaking pressure pattern (Rapoport et al., 2013), with a slow increase in pressure (Period I) and fast increment (Period II). At the end of Period II, the endocarp reached maximum pitbreaking pressure, and the sclerification process was almost null (Rapoport et al., 2013). CAPÍTULO I 41 Endocarp growth occurred only during the first period prior to the fast increase in pitbreaking pressure (Rapoport et al., 2013) at the beginning of Period II (pit-hardening period in irrigation literature). Period I of endocarp development was reported as coincident with the main part of mesocarp cell division activity (Hammami et al., 2011). This period was strongly related with yield, because mesocarp cell division established the final number of cells, and this was reported as the most important process on the determination of the final fruit size (Hammami et al., 2011; Rosati et al., 2020). Then, in absence of water stress, fruit size at the end of Period I could be related with fruit size at harvest. Fruit size is a very important feature in table olive, but it also affects the decision of final use in cultivars either as oil or table (e.g., cv Hojiblanca). This fruit growth pattern is well fitted to Mediterranean weather, but this could vary under changing climatic conditions. As far as we know, there were no scientific works which analyzed the influence of temperature in endocarp development. Endocarp hardening (Period II), the period of fruit growth less sensitive to water stress, occurred during the period of the year with higher water demand in the Mediterranean weather. This synchronization was reported to be closely related to temperature (Pérez-López et al., 2008; Trentacoste et al., 2012). In the current context, global warming could change the best zone for olive production, which would be impacted through the effect on fruit and endocarp development. The aim of this work was to study the length of the first two periods of endocarp development to model them and then use fruit weight or volume at the end of Period I for estimating harvest fruit size. 2. Results 2.1. Endocarp Hardening Model A sigmoid equation was very well fitted for the pit-hardening pattern of all cultivars and locations considered (Table 1). The determination coefficients (R2) were higher than 0.7 in all cases, except in two of them, where variations occurred in the third period. Cv Manzanilla had significantly the hardest pit, with maximum values (a + d) of 167 MPa. Cv Cornicabra and cv Arbequina did not show significant differences for maximum pit hardening (130 MPa cv Cornicabra and 108 MPa cv Arbequina). The slope of pit hardening did not present significant differences between cultivars, with an average value of 0.14. This means an angle of 82 in the inflection point over the horizontal. The CAPÍTULO I 48 so this result can be due to the use of several cultivars with different sizes that introduce a high scatter of data that is bigger than the effect of fruit load. More data per cultivar are necessary to quantify the fruit load effect. Maximum pit pressure at harvest (Table 1) was significantly linked to fruit dry weight at the end of Period I and at harvest (Figure 4). The best fit, in Figure 4, included the three cultivars considered in a linear regression. Each cultivar was grouped according to their weight, with the lowest values in cv Arbequina and the highest in cv Manzanilla. The amount of data for cv Manzanilla was very small, just two seasons (Tables 1–3 and Figure 3). In order to evaluate the relationship between fruit volume at the end of Period I vs. harvest, published data were used (Corell et al., 2020; Martín-Palomo et al., 2020). Fruit volume at Period I was related to the volume at harvest (Figure 5a) and to the fruit size, which was estimated as the number of fruits per kilogram (Figure 5b). The accuracy of the regressions was similar to the ones obtained with Arbequina and Cornicabra (Table 6 vs. Table 5) but with a significantly greater slope. When a multivariable estimation was calculated using the fruit load, the improvements of the regression with volume data was low (Table 6). However, fruit load increased significantly the regression accuracy with fruit size (from 0.65 to 0.8, Table 6). The volume equation (Figure 5a), which did not consider fruit load (first equation in Table 6), included all measured data in cv Manzanilla (Figure 3b) in the prediction interval (95%). CAPÍTULO I 49 Figure 3: Relationship between fruit dry weight (a) and volume (b) at the end of Phase I of endocarp development vs. at harvest. Green circle, cv Cornicabra; Blue triangle, cv Arbequina; Red square, cv Manzanilla. Dashed lines represent the best fit with all data of cvs Cornicabra and Arbequina. Short dash lines represent the best fit obtained for cv Manzanilla with published data (see below). Each point is individual plot data of each season and cultivar. CAPÍTULO I 50 Table 5: Best fit of the relationships between dry weight and volume at Period 1 and harvest (Figure 4). The regression of dry weight included all data of the three cultivars. In the regression of volume, only data of cv Arbequina and Cornicabra were considered. Notes: WH, weight at harvest; W1, weight at Phase I; VH, volume at harvest; V1, volume at Phase I; R2, coefficient of determination; Error, standard deviation; N, number of data; MSE, residual mean square error. *** indicates significate correlation at 0.001. Table 6: Best fits of the relationships between dry weight and volume at Period 1 vs. at harvest in Manzanilla experiments (Figure 5). Notes: VH, volume at harvest (cm3); V1, volume at Phase I (cm3); FL, fruit load (kg m- 3); FS, fruit size (fruits kg-1); R2, coefficient of determination; Error, standard deviation; N, number of data; MSE, residual mean square error. *** indicates significate correlation at 0.001. Figure 4: Relationship between fruit dry weight at harvest (a) and at the end of period I (b) vs. maximum pit-breaking pressure at harvest. Green circle, cv Cornicabra; Blue triangle, cv Arbequina; Red square, cv Manzanilla. Dashed lines represented the best fit with all data. (a) Pressure = 57.40 W + 67.43; N = 11; R2 = 0.78, Error = 12.7 MPa; MSE= 161.3. (b) Pressure = 178.53 W1 + 70.89; N = 11; R2 = 0.68, Error = 15.4 MPa; MSE= 236.6. Data are the average for each cultivar and season. CAPÍTULO I 51 Figure 5: Relationship between fruit volume at Period I of endocarp development and fruit volume at harvest (a) and fruit size at harvest (b) of published Manzanilla experiments. Lines represent the best fit in both figures (equations at Table 6) and the prediction interval (95% confident) only (a). Black symbols at (a) are the Manzanilla data of Figure 3. CAPÍTULO I 52 3. Discussion 3.1. Endocarp Development The length of the Period I of endocarp development was little affected by environmental conditions. The end of this period could be predicted accurately considering 49 days of length from full bloom and the average of all cultivars and seasons taken into account (Table 2). Environmental conditions, mainly temperature, would likely affect the length of this period, but the current approach suggested that the influence was low. Even the cultivar effect was very low, with variations shorter than 2 days when Arbequina and Cornicabra were compared. Moreover, the length of Period II (Table 4) was related to the DAFB of the inflection point (parameter c, Table 1; R2 = 0.6988 **, data not shown). This would support that the length of Period I was almost constant. Several authors with less seasons suggested a similar duration of Period I and did not consider the influence of temperature ((Hammami et al., 2011) around 60 days; (Rapoport et al., 2013) around 50 days). The determination of the end of Period I of endocarp development is very important for the pattern of fruit growth. During Period I, the most important process in the endocarp and mesocarp was cell division (Rapoport et al., 2004). Then, when this period finished, in mesocarp, the cell division decreased while the cell expansion continues, coinciding with Period II of endocarp development (Rapoport et al., 2013). The main differences between cultivars in fruit size at harvest were related to the number of cells in the mesocarp at the end of Period I (Hammami et al., 2011; Rosati et al., 2009). Endocarp was also reported to be the basis for the mesocarp growth (Rapoport et al., 2004), so a smaller endocarp, that grew during Period I, produced a smaller mesocarp and then a smaller fruit (Gómez del Campo et al., 2014; Rapoport et al., 2004). On the other hand, there was a significant influence of thermal time in the fruit growth during Period I (Figure 1, Table 3). Thermal time was negatively correlated with fruit volume and dry weight at the end of this period in all the models considered. This result suggests that also endocarp size would be affected because both are related (Rapoport et al., 2004). Considering that current data correspond to full irrigated orchards, this implies that Period I occurred nowadays in a sub-optimal range of temperatures that could limit fruit development. Although olive trees are a fruit species adapted to warm climates, high temperatures would affect their growth capacity. These results mean that in the context of climatic changes, the increase in temperature would strongly reduce fruit size and, therefore, fruit yield, because Period I, with a constant length, would occur in a CAPÍTULO I 53 worse range of temperature for fruit growth. Olive field experiments simulating climatic changes with a 4 ºC increase in temperature reported that the vegetative growth was reduced when the temperature reached values above 37 ºC (Benlloch-Gonzalez et al., 2016), as was fruit size (Benlloch-Gonzalez et al., 2019). However, (García-Inza et al., 2016) reported that the increase in temperature during a period similar to Period I of the current work did not affect the mesocarp dry weight, but (García-Inza et al., 2014) found a decrease in fruit dry weight for mean temperatures above 25 ºC during the last period of fruit growth (about 110 days). These results could be related with the sort of experiment, where (García-Inza et al., 2014 García-Inza et al., 2016) only modify a branch while the rest of the tree retained the atmospheric temperature. Another possibility could be that cell expansion, the only cell process that occurs in the period in (García- Inza et al., 2014), is more sensitive to temperature than cell division. In both studies (García-Inza et al., 2014 García-Inza et al., 2016), oil concentration was negatively affected by mean growth temperatures above 20 ºC. The length of Period II of endocarp development was clearly affected by the temperature (Table 5). This period coincided with the most important part of sclerification (Table 1) which was different between cultivars (Table 1; Figure 4). This process had been related to a greater demand of reserves, which stops vegetative growth in mature trees (Rallo et al., 1991). The length of Period II varied from around 332 degree-days in Arbequina to 405 in Manzanilla (Table 4, Linear model) and was very closely related to maximum pit hardening (parameter a + d, Table 2; R2 = 0.7442 **, data not shown). Differences in maximum pit-breaking force between cultivars (Table 1) could be associated to a higher degree of sclerification or a greater endocarp with more sclerification cells. The relationship between maximum pit-breaking force (a + d parameters) and the fruit dry weight (Figure 4) suggested the second hypothesis. The increase in temperature in a context of climatic change would also affect Period II, but the final effect on yield would not be clear. This effect could be different between cultivars and would be related to the effect of temperature in the assimilation rate. Ref. (Benlloch-Gonzalez et al., 2019) suggested that an increase in temperature would reduce the pulp–stone ratio and oil accumulation, but such an effect could be related to the decrease in fruit size produced in Period I (Gomez del Campo et al., 2014; Rapoport et al., 2004). The increase in air temperature during Period II in cv Araujo decreased the content of oil but not the dry weight in mesocarp (García-Inza et al., 2016). CAPÍTULO I 54 3.2. Endocarp Development Fruit volume and dry weight at the end of Period I was an adequate predictor of fruit size at harvest (Figures 3 and 5, Tables 5 and 6). The estimation of size at harvest from data obtained early in the season is widely described in the literature for different fruit trees, included olive (among others, such as peach (Scorzal et al., 1991); saskatoon (McGarry et al., 2001); olive (Rosati et al 2009; Rosati et al 2010); apple (Legave et al., 2013). In olive trees, the fruit size is strongly related to the cultivar and the number of cells in the fruit (Rosati et al., 2010). As most of the cell division occurred during Period I (Hammami et al., 2011), the size at the end of this period would be linked to size at harvest. Water stress after this Period I would not affect fruit size when an adequate rehydration was performed (Corell et al., 2020; Girón et al., 2015). This irrigation scheduling had been suggested for table olives, in which size was a very important feature, but oil olives could delay or even miss a complete rehydration (Ben-Gal et al 2021). Moreover, (Corell et al., 2020) reported that deficit management for table olives could be limited when a great fruit load was presented, because fruit size would be too small even in full irrigated conditions. Therefore, the estimation of fruit size at harvest could support the decision of water management during the season and even the final destination of yield if the fruit size expected was too small. Fruit volume estimations suggested that table and oil cultivars could present different patterns of fruit development. In the current work, two oil cultivars (Arbequina and Cornicabra) had a common estimation, while cv Manzanilla, a table cultivar, was different (Figures 3 and 5; Tables 5 and 6). Cultivar Manzanilla presented greater volume at harvest and greater growth velocity throughout the season (Figures 3 and 5). Fruit size is very important in table cultivars but so is the mesocarp vs. endocarp relationship (Gucci et al., 2009). This latter feature is greater in cv Manzanilla than in cv Cornicabra and Arbequina (Barranco et al.,1997), and it could be related to the final fruit size (Rapoport et al 2004). CAPÍTULO I 55 4. Materials and Methods 4.1. Sites Description Data were obtained from three different cultivars: Cornicabra (oil), Arbequina (oil) and Manzanilla (table), throughout several seasons (from 2006 to 2022). Trees were grown in different orchards located in Ciudad Real (central Spain) and Seville (south Spain). Both locations are around 350 km away and have different conditions for olive development (Table 7). Seville is in the main table olive-producing area in Spain, and climatic conditions are the optimum for olive growing. In this location, maximum temperatures were above 34 ºC from May to September as an average for the period 2006–2021. On the contrary, winter was warm with minimum temperature slightly below 0 ºC in January but higher than 6 ºC from April to October. The rain pattern was the typical of a Mediterranean climate with a drought period in summer (from June to August). The average seasonal rain was 497 mm. On the contrary, climatic conditions in Ciudad Real could be limiting for olive development. Maximum temperatures were similar to Seville, but the period when they were above 34 ºC was narrower (from June to September). Winter presented lower minimum temperatures for longer than Seville with values below -3 ºC from November to March. The pattern of rains was similar in both locations, but Ciudad Real presented a lower seasonal rain than Seville with a total amount of 421 mm. • Eleven different seasons were used in the current work (Table 2): • Five seasons for cultivar Cornicabra at “Entresierra” experimental farm near Ciudad Real (39º N, 3’ 560 W; altitude 640 m). • Three seasons for cv Arbequina at the same location. • One season for cv Arbequina at “El Morillo” farm near Seville (37.5º N, 5.7’ W; 102 m altitude). • Two seasons for cv Manzanilla de Sevilla at “La Hampa” experimental farm near Seville (37º 17 ‘N, 6º 3’ W, 30 m altitude). CAPÍTULO I 56 Table 7: Comparison of the two locations where data of the experiments were obtained. Data are the average of monthly data in the period 2006 to 2021. Source of data: Climatic station of “La Rinconada” and “Ciudad Real” of the Spanish Network of agri-climatic stations for irrigation (SIAR, https://eportal.mapa.gob.es/websiar/SeleccionParametrosMap.aspx?dst=1, accessed on 15 October 2022). Note: T Max, average maximum temperature (ºC); T Min, average minimum temperature (ºC); Rain, average rain (mm). Orchards of cvs Cornicabra and Arbequina were mature (more than 10 years at the beginning of the experiment), and only the Manzanilla orchard was young (3 years, first yield in 2021 season). In the orchards located in Ciudad Real, trees were 7 m x 4.75 m apart, around 300 trees ha-1, while the two located in Seville were 4 m x 1.5 m apart, around 1667 trees ha-1. All data presented correspond to full irrigated treatments (100% crop evapotranspiration (ETc) throughout the irrigation period. Details about these orchards were published by (Pérez-López et al., 2007) (Cornicabra), (Corell et al., 2022) (Arbequina, Seville) and (Sánchez-Piñero et al 2022) (Manzanilla). 4.2. Endocarp Development The pattern of endocarp development was studied in these 11 seasons (Table 1). Full bloom was dated in all the seasons. Samples of 20 to 30 fruits were periodically collected during each season in 8 plots at Cornicabra orchard, 10 plots at Arbequina orchard in Ciudad Real, 4 plots at Arbequina orchard in Seville and 3 plots at Manzanilla orchard in Seville. Dry weight and fruit volume were measured weekly. Data from the end of period I and from harvesting were used for this work. The pit-breaking pressure in all fruits was measured according to (Rapoport et al 2013) using a device similar to the one described in this latter work. In short, pressure applied by hand to a lever was CAPÍTULO I 57 transformed into the vertical movement of a probe terminating in a 2 mm diameter tip. Several capacity load cells of different pressure ranges were used. These load cells attached above the tip transformed the force applied to break the pit into an electrical signal. Pressure (MPa) is calculated as the force (N) divided by the area of the tip (m2). The electrical signal was sent to a data acquisition module (Model KUSB 318, Keithley Instruments Inc., Cleveland, OH, USA) connected by USB 2.0 to a computer for data processing. We also developed a specific application based on Visual Basic 6.0 to register the acquired data and to indicate on the screen the máximum pressure reached. Figure 6 shows the expected pattern in full irrigated conditions for fruit dry weight and pit-breaking pressure according to (Rapoport et al 2013). The pattern of pit-breaking pressure was adjusted in each season using the sigmoid Equation (Rapoport et al 2013): PBP = a 1+ℯ −𝑏𝑏(𝐷𝐷𝐴𝐴𝐹𝐹𝐹𝐹−𝑐𝑐) + d (1) where: PBP was the pit-breaking pressure; a is the range of PBP from the minimum to the maximum PBP value; b is the slope coefficient at the inflection point; c is the date at the inflection point; d is the minimum PBP; DAFB days after full bloom. CAPÍTULO II 64 1. Introduction Water is becoming an increasingly limited resource due to reduced meteorological precipitation and rising temperatures, which results in greater demand and a surge in nonagricultural uses (Lorite et al. 2022). For these reasons, farmers are forced to use deficit irrigation scheduling. The aim of regulated deficit irrigation (RDI) is to impose water stress at times where the impact on the yield is very small or inconsequential (Chalmers and Wilson 1978). Irrigation scheduling may cause a yield reduction when implemented without taking into account the phenological stage (e.g., peach, Girona et al. 2012) or the water stress level imposed (e.g., olive, Goldhamer 1999). Olive (Olea europaea L.) trees have a high resistance to water stress (Díaz-Espejo et al. 2018), and their endocarp hardening period is considered the time where such stress has the lowest impact on the final yield (Goldhamer 1999). Nevertheless, the final result of water stress on the yield depends on the point in time in which it is applied, as well as on its intensity and duration (Hsiao 1990). For olive trees, this is even more complex because the production may be intended for different uses, such as table olives, where the most important parameter is fruit size, or oil production. This involves diverse irrigation strategies, since the sensitivity of the affected processes is also different. The fruit growth process is the most sensitive (Martin-Palomo et al. 2020), and excessively harsh water stress conditions (Goldhamer 1999) or insufficient rehydrations (Corell et al. 2020) may result in a yield loss. Conversely, oil accumulation is a more resistant process, and it has even been reported as being minimally affected by water stress (Navas-Lopez et al. 2019; López-Bernal et al. 2021). However, some studies indicated that there was indeed some sensitivity in oil producing crops (Fernández et al. 2013; Ahumada-Orellana et al. 2017; Corell et al. 2022). Therefore, RDI studies with olive trees tend to use endocarp/ pit hardening as a reference, although few actually made measurements to estimate this period, considering fixed dates to establish the water stress application instead (e.g., Goldhamer 1999; Moriana et al. 2003). On occasion, this phenological stage has been described as the time when it is not possible to “cut the fruit with a knife” (e.g., Garcia- Inza et al. 2014). But this type of measurement indicates the end of the hardening stage, not its beginning, and this is a serious simplification. Rapoport et al. (2004) described the endocarp hardening CAPÍTULO II 65 period as the gradual appearance of sclereids that start at the time of pollination. Rapoport et al. (2013) used three different periods to describe the evolution of the endocarp hardening. The sclerification process is initially very slow (Period I), but the rate increases once the endocarp reaches its maximum size (Period II), until reaching its maximum hardness (Period III) (Rapoport et al. 2013). Consequently, Period II of fast hardness growth would be the stage with the greatest resistance to water stress. Recently, Sánchez-Piñero et al. (2022) monitored hardness in different campaigns of different cultivars, in different locations and in the absence of water stress and reached the conclusion that the beginning of Period II takes place 49 days after full bloom. This duration was similar to that suggested by Rapoport et al. (2013) for the Cornicabra cultivar and it could be considered a constant for the species. However, the duration of Period II was variable and dependent on the temperature and the cultivar considered (Sánchez-Piñero et al. 2022). This period was the most common to suggest irrigation restrictions (i.e., Goldhamer 1999). Therefore, if the length changed, water stress could be applied in the most sensitive times of fruit development (such as oil accumulation). In addition, the changes in the pattern of Period II could also be affected by water stress. If the duration of these periods was modified, the real impact of RDI strategies would be uncertain. Hammani et al. (2013) suggested that the beginning of Period II would be moved forward under water stress conditions, while Rapoport et al. (2004) indicate the opposite. The impact of water stress on yield has multi-year components (shoot growth) as well as others within the same year (fruit size, oil amount). These latter processes would be affected at the time when water stress is applied. Because endocarp development was not commonly described, the real effect of water stress on these yield components is unknown. The aim of this work was to determine the impact of water stress on the different processes taking place in the fruit (endocarp formation, fruit growth, and oil accumulation), and accurately measured the pit hardening changes to establish three development periods. In addition, the water stress impact was assessed considering point thresholds (maximum stress values) or cumulative thresholds (water stress duration and intensity). The description of these responses will permit the identification of possible limitations in the use of pit hardening as a reference in deficit irrigations and the response of this yield component to water stress. CAPÍTULO II 66 2. Materials and methods 2.1. Site description and experiment design The experiment was conducted during the years 2021 and 2022 in an experimental plot at the farm “La Hampa”, in Coria del Río (Seville, Spain, 37º 17’ N, 6. 3’W, 30 m). The cultivar used was “Manzanilla de Sevilla,” and the trees were 3 years old at the beginning of the experiment, planted at a distance of 4 x 1.5 m. The first harvest took place during the 2020 campaign and the yield was 800 kg ha− 1. The soil depth was variable but always deeper than 1 m and had a sandy-loam composition. The irrigation water came from a well located on the farm, and it was pumped to a dam before it entered the irrigation system. The mineral composition of the water was adequate for olive irrigation (pH 8.3; 2.37 dS m-1; 73 mg l-1 Ca; 101 mg l-1 Mg; 225 mg l-1 Na). Climate data were collected from the weather station at “La Puebla del Rio” (SIAR 2023). This station belongs to the Andalusian network and is located 6 km away from the site. Figure 1 shows the reference evapotranspiration (ETo) and rainfall during the experiment. The seasonal patterns for both variables are typical for a Mediterranean climate in which rains fall mainly in winter and spring and summer is very dry (Figure 1) Figure 1: Seasonal pattern of reference evapotranspiration (ETo) and rain during the two seasons of the experiment. Source: “La Puebla del Rio” station (SIAR 2023). Vertical lines limited the experimental period. Arrows marked the full bloom date. CAPÍTULO II 67 ETo values reached their peak from the end of spring to the beginning of autumn, standing above 5 mm-1 day on some days. During the period of the experiment, the average measured ETo was above this value. The amount of seasonal rainfall during both years was below the average for the area in the last decade (534 mm (AEMET 2023)), and it was approximately 355 mm, 65% of the long-term average. There was an almost complete absence of rainfall during the experiment period in both years, which is frequent in the area. The below-average rainfall meant that the period immediately before flowering was drier than usual, particularly in 2021 (Figure 1). The experiment design consisted of randomized plots with three repetitions per treatment. Each experimental plot included 5 rows of 13 trees each, and the 3 central rows were used for the measurements. The irrigation system consisted of a simple line of drips of 2.2 l h-1 at a distance of 0.5 m. Irrigation treatments started in 2020 and were based on the combination of the water stress application time point and intensity. The irrigated season was divided in three different phenological stages in which pit hardening was the main reference. Before and after pit hardening periods were considered as sensitive to water stress in relation to yield respond. The former included the main period of vegetative growth, full bloom, and fruit set. The latter occurred in a few weeks before harvest in this table cultivar green style preparation and included the rehydration period in deficit irrigation management. The start of pit hardening was estimated, according to Rapoport et al. (2013), when the rate of olive longitudinal growth started decreasing. The end of this period, in both years, was a fixed date on the third week of August. The irrigation treatments were: • Control. Irrigation scheduling based on the midday stem water potential (SWP) measurements and the estimated evapotranspiration of the crop (ETc). ETc was calculated using the FAO methodology, with a crop coefficient (Kc) of 0.55 and a reduction coefficient (Kr) of 0.7 in 2021 and 1 in 2022. In 2021, the intention was to optimize vegetative growth to achieve maximum yield in the shortest time possible. To that end, the applied water was 100% ETc, except when SWP measurements were more negative than −1.0 MPa, when they were increased to 175%. In 2021, the yield obtained from irrigated treatments was similar to that from an adult plot (approximately 10,000 kg. ha-1), so in 2022, the irrigation scheduling was modified to prevent an excess of vigor. During the second year, the increase in applied water during the pit hardening period was only made when SWP was 0.5 MPa more negative than CAPÍTULO II 68 the baseline established by Corell et al. (2016). This strategy meant that the total amount of water applied was 772 mm in 2021 and 727 mm in 2022. • Regulated deficit irrigation (RDI). Irrigation scheduling was also based on SWP measurements, but different thresholds were taken into account in both years according to the phenological stage. In 2021, the SWP threshold values were those estimated using the baseline from Corell et al. (2016), except during pit hardening, when the threshold was fixed at −2.0 MPa for the entire period. In 2022, the SWP threshold was obtained from the baseline by Corell et al. (2016) minus 0.5 MPa, except during the hardening period, when the differential was increased up to 1.5 MPa. The applied water varied based on the difference between the SWP measured and the threshold at any given time. When the SWP was more negative than the threshold, irrigation started with 1 mm day-1, then increased to 2, 3, and 4 mm day1when the differential was higher than 10, 20, or 30%, respectively. This scheduling meant that the total amount of water applied was 334 mm in 2021 and 200 mm in 2022. • Rainfed. The treatment was not irrigated in June 2020 and received 45 mm of complementary irrigation in September of that year. In 2021 and 2022, no irrigation was provided at all. • Traditional Rainfed. In 2022, a Traditional Rainfed plot with the same cultivar in a 7 x 5 m plantation and 24 years of age at the beginning of the trial was assessed. The water stress in this treatment was shorter but similar in water stress level than the previous one. This permitted a comparison in the effect of the duration. The fruits yield in both years showed a very marked alternate bearing pattern. In 2021, the irrigation treatments produced an average yield close to 10,000 kg ha-1, while in 2022, the yield was between 500 and 1000 kg ha-1 in all treatments, except for Traditional Rainfed, which was approximately 2500 kg ha-1. 2.2. Measurement description The endocarp development was monitored using weekly measurements of pit hardness. Two weeks after full bloom, random samples were taken from ten fruits per repetition of each treatment. The breaking pressure was determined using a device similar to that described in Rapoport et al. (2013). In short, this equipment allows using manual CAPÍTULO II 69 pressure to convert it into the vertical movement of a probe with a 2 mm diameter tip. This tip is connected to a load cell that allows for measuring the force applied. The load cell electrical signal was sent to a data receiver (Model KUSB 318, Keithley Instruments Inc., Cleveland, OH, USA) connected to a computer where data were stored. Data collected during treatment were adjusted to a sigmoid equation: PBP = a 1+ℯ −𝑏𝑏(𝐷𝐷𝐴𝐴𝐹𝐹𝐹𝐹−𝑐𝑐) + d (1) where PBP is the pit-breaking pressure, a is the range from the minimum to the maximum PBD, b is the slope of the inflection point, c is the date when the inflection point occurs, d is the minimum PBD and DAFB are the days after full bloom. Adjusting this equation allowed for the definition of the three development periods described by Rapoport et al. (2013): Period I, slow hardness increase; Period II, fast growth; and Period III, maximum hardness values. To determine the beginning of these periods for each of the samples obtained, Eq. 1 was adjusted for each repetition, so the starting and end dates of Period II were established when the equation estimated that 10% and 95% of hardness were reached, respectively. Determining the amount of oil in the fruit was done by randomly sampling ten olives per repetition on the same date of endocarp development sampling. Fresh and dry weights were determined after reaching constant weight in an oven at 65 ºC. The determination of the oil content was done using a sample from ten dried fruits and a nuclear magnetic resonance spectroscopy (NMR, Minispec, Bruker, Massachusetts, USA). The amount of oil per fruit was determined by multiplying the dry weight by the percentage of oil obtained. The olive size was determined weekly using ten fruits per treatment, selected randomly. These fruits were measured while still on the trees by taking the length and width in the middle area to calculate the volume. The water status of the trees was estimated through a SWP measurement per repetition. The SWP was measured on fully expanded leaves that were covered at least 2 h before taking the measurement using a pressure chamber (Model 1000, PMS, USA). SWP data were analyzed to obtain three indicators in different phenological stages: minimum SWP, average SWP, and stress integral (SI). The SI was estimated using the modified Myers formula (1988), where the maximum value was different on each date using the one estimated with the baseline from Corell et al. (2016): CAPÍTULO II 70 SI = Σ(SWP − (BL)) ∗ n (2) where SI is the stress integral, SWP is the midday stem water potential, BL is the value estimated on the baseline, and n is the number of days. The periods where these indicators were calculated included pre-flowering, Period I, II, and III of endocarp development, according to the estimations based on the hardness measurements. 2.3. Statiscal analysis The statistical analysis was conducted with ANOVA and a separation of means (Tukey test) using the Statistic (SX, v 8.0) program. Significant differences with a p level <0.05 were considered in both tests. The assessment of the wáter stress effect on some parameters was done by obtaining the linear regression between them and the indicators calculated for each period. Multivariate analyses were conducted, and different indicators and/or periods were included; however, they did not improve simple fits. Sigmoid parameters were estimated with nonlinear regression tools with the package “nls2” in R environment (R Core team 2013). 3. Results 3.1. Endocarp development and water status Pit hardening followed a sigmoid pattern in all treatments, both in the 2021 season (Figure 2a) and in the 2022 season (Figure 2b), due to a variation of the pit hardening rate, which remained constant during an initial period (Period I), then started a fast increase (Period II), and ended when the máximum value was reached (Period III). CAPÍTULO II 71 Figure 2: Seasonal pattern of pit-breaking pressure curve during the 2021 (a) and 2022 (b) seasons. Blue circle: Control; green triangle: RDI; red square: Rainfed; yellow diamond: Traditional Rainfed (colour figure online). Afterward, it remained constant for the rest of the fruit’s growth and ripening. The pattern change was the same for all treatments, except the Rainfed one in 2022 (Figure 2b). This treatment showed maximum seasonal values that were almost constant from day CAPÍTULO II 72 of year (DOY) 190 and clearly below the others. Table 1 shows the parameters for the individual fits of the sigmoid equation for these pit hardening curves. Table 1: Parameters (average ± standard error) of the fit adjusted in the individual pitbreaking pressure curves. Sigmoid equations for modeling the pit-breaking pressure (PBP) according to the equation PBP. Each data is the average of three plots. Different letters in the same column indicate significant differences (Tukey test, p < 0.05) * a, maximum pit breaking; b, rate of increase of pit breaking; c, days after full bloom for 50% hardening; d, minimum pitbreaking pressure In 2021, there were no significant differences in any of these parameters for the different treatments. The number of days until half the maximum hardness is achieved (Table 1, parameter c) was very similar, approximately 71 days from full bloom. The hardening rate during Period II (Table 1, parameter b) was also similar at approximately 0.13 MPa day-1, with maximum hardness (Table 1, a+ d) at around 165 MPa. Conversely, the Rainfed treatment displayed a maximum hardness, which was significantly lower (108.4 MPa) than the rest in 2022. Even if the rest of the parameters did not show any important variations, the pit hardening rate (b) tended to have higher values than the rest, and the date when 50% of the hardening value was reached (c) tended to have lower values in the Rainfed treatment. The RDI treatment in 2022 also tended to have values suggesting a faster hardening curve (steeper slope (b), fewer days (c)) and lower hardness (a + d) than the rest, although the differences were much smaller than those seen in the Rainfed treatment (Table 1). The curve fits in Figure 2 and Table 1, together with the bloom date, allowed for estimating the duration of endocarp development Periods I and II (Table 2). In 2021, full bloom occurred on DOY 107 in all treatments, whereas in 2022, this took place on DOY 111 in the Rainfed and Traditional Rainfed treatments, and on DOY CAPÍTULO II 73 118 in Control and RDI treatments. The duration of Periods I and II in the 2021 season was similar for all treatments, and the differences were not significant, as they were never longer than 2 days. Conversely, the duration of Period I in 2022 for the Rainfed treatment was considerably shorter than the rest, and the difference was more than 10 days. However, in the Traditional Rainfed treatment, this Period was of a similar duration than in RDI and Control and ended significantly earlier than Rainfed. Even though there were no differences in the duration of Period II this year, RDI tended to have lower values than the rest, with 6–8 days of difference vs. Control and the two Rainfed treatments. Table 2: Characterization of Period I and II of endocarp development in the different treatments (average ± standard error, n = 3) Different letters in the same column mean significant differences (Turkey test; p < 0.05) * Flower, DOY of full bloom; start P2, DOY of the beginning of Period II of endocarp development; end P2, DOY of the end of Period II. The data in Figure 2 and Table 2 allowed for the establishment of four different periods according to the endocarp development. The water stress indicators for these periods changed throughout the season (Table 3). Measurements during the 2021 preflowering period were scarce, and only one was completed due to rainfall (Fig. 1). During Period I, meaningful differences were found between Rainfed and Control only in the SI, although the most negative values for the other indicators were also found in this treatment. On the other hand, the minimum potential (Min-SWP) in Period II was considerably lower in the Rainfed treatment than in the rest. Despite the lack of important CAPÍTULO II 80 Oil accumulation during Period II (R2) showed a significant relationship with the Min-SWP for the same period, though more so in 2021 than in 2022 (Table 4). Table 4: Best fits in the relationship between oil accumulation rate at Period II (R2) or Period III (R3) vs. water status indicators. * P2 or P3, indicates the period when water status indicator was measured; Min-SWP, minimum midday stem water potential; Av-SWP, average stem water potential; SI, stress integral; N, number of data; R2, signification coefficient; SD, standard deviation; MSE, mean square error. There was a slight variation of R2 data from 2022, despite the wide range of changes in the water status (Figure 5a), resulting in worse fits (Table 4). The oil accumulation rate value during Period III (R3) showed similar values in 2021 and 2022 (Figure 5b) with important relationships with the water status indicators (Table 4). The best fits were obtained with the Min- SWP from Period II (Table 4; Figure 5). Although the best fit was linear (Table 4), the data suggest a constant value up to approximately −3.0 MPa for Min- SWP (Figure 5b) or −2.0 MPa for Av-SWP (Figure 5c). The R3 rate was also significantly related to the SI and the Av-SWP during Period III (Table 4). There were no fits that could link in one equation the indicators for Periods II and III with R3 (data not shown). CAPÍTULO II 81 4. Discussion 4.1. Endocarp development The endocarp development took place simultaneously with fruit volume increase and oil accumulation, but this did not match the growth rate when it was at its peak for the three processes considered (Figures. 2, 3, 4). The seasonal pattern of endocarp development was hardly affected by the fruit load conditions and water status of the trees (Figures 2; Tables 1, 2, 3). The duration of Period I in this work was similar to that described by several authors for varieties and locations when there is no water stress (Sánchez-Piñero et al. (2022), 49 days, and Rapoport et al. (2013), 50 days). Only under very severe water stress conditions, i.e., Rainfed treatment in 2022, a change in endocarp hardening was noticed, with lower maximum values than the rest and a significantly shorter Period I duration (Figure 2; Tables 1, 2). Data suggest that this could be related to water stress in pre-flowering stage (Table 3). Although there were no significant differences in this period for the years studied, the Min-SWP was more negative in the Rainfed treatment in 2022 vs. the rest of the treatments and years. Only one of the three Rainfed plots showed values much lower than the rest (−2.50 MPa Min-SWP), and the differential with the next more negative value (a RDI plot) was 1.14 MPa, indicating the values that could have caused the endocarp development alterations. Hammani et al. (2013), working with cv Picual, presented similar results, with a shortening of Period I and a smaller sclerified area in Rainfed trees under severe water stress (higher than −3.0 MPa during pre-flowering) but not under moderate stress conditions (the water status of this data is shown in Moriana et al. 2003). On the other hand, Rapoport et al. (2004) describe that, for cv Leccino, a severe water stress (more negative than −3.0 MPa of potential at dawn) at this stage caused the extension of Period I. However, the trees studied by Rapoport et al. (2004) received a rehydration period immediately after this, which was not the case in the current experiment, nor in the one by Hammani et al. (2013). This suggests that, under severe stress conditions, the endocarp retains its ability to grow, even at the beginning of the fast hardening period (Rapoport et al. 2004), and a lower pit hardness would be required to indicate incomplete sclerification (Hammani et al. 2013; this work). CAPÍTULO II 82 4.2. Fruit growth Period I of endocarp development has a definite impact on the final fruit size (Rapoport et al. 2004; Sánchez-Piñero et al. 2022) because water stress at this phenological stage may affect the number of cells (Costagli et al. 2003) and their size (Rapoport et al. 2004) in a permanent way. The response of fruit growth to water stress during Period I changed according to the fruit load level. During Period I, in 2021, which was a high yield year, there were significant differences in volume and a trend toward worse indicator values in the Rainfed treatment. However, there were no fruit volume differences in 2022, despite the higher water stress levels in RDI and Rainfed vs. Control during Period I (Figure 3; Table 3). This suggests that fruit growth was more sensitive to water stress than SWP, as previously indicated in other works (Costagli et al. 2003; Gucci et al. 2019; Martin- Palomo et al. 2020). The olive tree physiology changed according to the fruit load level, with less negative water potential values and lower leaf conductance in years with avery low fruit load (Martin-Vertedor et al. 2011; Corell et al. 2016). Nevertheless, such differences in water relations were identified in Period II, when massive hardening began (Martin-Vertedor et al. 2011), but not in Period I. This absence of difference would be related to a lower sensitivity of water relations at leaf level vs. fruit development because it would then be possible to predict size changes during a loading or unloading year with the data from this Period I (Sánchez-Piñero et al. 2022). The sclerification process means a high sink of water and nutrients (Rallo and Suarez 1989; Girón et al. 2015; Trentacoste et al. 2022). So, although at a lower rate, it starts with fertilization (Rapoport et al. 2004). The presence of a lower number of fruits in the low fruit load season could ensure adequate fruit development under severe water stress at leaf level since the fruit would be a preferred sink. Trentacoste et al. (2022) also concluded that, even if the initial fruit growth is a weak sink, severe shadowing during Period I had a bigger impact on the high fruit load than on the low fruit load season and increased abortion due to competence. The water stress levels under high fruit load conditions that could impact volume in Period I would be between RDI and Rainfed values for 2021 (Table 3). Av- SWP data were similar to those suggested by Hueso et al. (2021), at −1.5 MPa. However, for Av and Min-SWP, there was a very narrow range only 0.1 MPa difference between RDI and Rainfed values. This narrow difference in both indicators could suggest that this was actually a cumulative effect of the water stress, represented by a higher SI with a threshold between 6.8 and 16.3 MPa day. Such thresholds have to be checked in future works and with other cultivars. The fruit growth CAPÍTULO II 83 process stopped suddenly with the beginning of the massive hardening (Period II) in the Rainfed treatment, but not for RDI (Figures 2, 3; Tables 1, 2). RDI trees did not even reduce their growth rate at the beginning of this Period II, although the water potential values during Period I were very similar for both treatments and clearly worse than those for Control (Figure 3; Table 3). These results suggest a cumulative effect of water stress, together with an increase in the sclerification process sink effect. The average values when the growth stopped in Rainfed and the rate changed in RDI were around 30 MPa day, with higher SI values for fruit volume decreased in RDI (data not shown). The different responses between Rainfed (stopped) and RDI (growth rate reduction) suggest a differential effect of water stress based on the time point when it occurred—for the first case, just at the beginning of Period II, and for RDI, during the second half of this period. Several studies suggested that SI values higher than 30 MPa day resulted in a fruit growth stop (Girón et al. 2015; Martín-Palomo et al. 2019). Other works also suggested a cumulative impact of water stress on fruit size when such conditions are applied at the end or during Period I and Period II (Gomez del Campo et al. 2014). Although the cell division process in the fruit has been described as not exclusive to Period I (Hammami et al. 2011; Gomez del Campo et al. 2014), the differential response between Rainfed and RDI could be considered related to the main impact of this process instead of cell expansion. The volume reduction process was completely reversible in the RDI treatment, as different authors described using stress time points similar to those in this work from Period II (Moriana et al. 2003; Girón et al. 2015; Corell et al. 2020). However, under conditions of highly severe water stress in Period I or at the beginning of Period II, the size was clearly reduced despite rehydration (Moriana et al. 2003), or there were no significant differences but a clear trend toward being smaller (Gomez del Campo et al. 2014; Gucci et al. 2019). Such responses would support the idea that different processes in fruit development would be affected at different times, and this would affect the possible recovery of fruit size. 4.3. Oil accumulation The oil accumulation process was more resilient to water stress than fruit growth. Accumulation started simultaneously in all treatments, with a seasonal continuous growth pattern in all data collected (Figure 4), although at different rates (Figure 5). This increase took place despite the complete stop of fruit growth in the Rainfed or RDI treatments CAPÍTULO II 84 (Figure 3), and it happened when the water stress levels were more severe, i.e., Periods II and III (Table 3). This response, together with the previous ones, suggests a strong pattern of assimilate distribution, acting as the main sink in the pit hardening process, followed by the oil accumulation, which starts a faster accumulation period when the previous one ends, and, lastly, the fruit growth (Figures 2, 3, 4). Different works suggest the relevance of the hardening process vs. fruit growth (Rapoport et al. 2004) or vegetative growth (Rallo and Suarez 1989; Hernandez-Santana et al. 2018). The absence of a relationship between the rates of oil accumulation (R2 and R3) and indicators of water status in Period I suggests that the cell division could have an influence on the total oil accumulated, which is smaller than previous works suggested (Gomez del Campo et al. 2014). It also seems to indicate that a smaller final fruit size in Period I, and potentially at harvest, could be offset by a more efficient oil production during Periods II and III. In fact, the fruit volume reduction seen between DOY 220 and DOY 240 in both years (Figure 3) did not have an impact on oil accumulation (Figure 4). Different studies suggest that moderate water stress would probably improve the oil percentage in the fruit (Moriana et al. 2003; Grattan et al. 2006; Gucci et al. 2019; Lavee et al. 2007; Fernández et al. 2013; Ben-Gal et al. 2021; Corell et al. 2022). Water stress during Period II of endocarp development apparently had a closer relationship with the oil accumulation rate, both in this period (R2) and in the following one (R3) (Table 4). Garcia- Inza et al. (2014) carried out an experiment on the susceptibility of olive trees to high temperatures and suggested that the fruit development period for the cv Araujo during January– February (Southern hemisphere) had the biggest impact on oil accumulation. Although endocarp development was not accurately established (cutting with a knife), the period of that work (Garcia-Inza et al. 2014) could approximately match Period II in the current work. The linear fit between R3, the steepest oil accumulation rate, and the water status indicators was good (Figure 5; Table 4). But data suggested that this sensitivity would actually be linked to a threshold defined by water stress conditions more negative than −3.0 MPa of Min-SWP. The fits in Table 4 provided good results for SI and Av-SWP too, which could suggest that a lack of rehydration in the Rainfed treatments was the most influential factor on the Min-SWP indicator in this work. Oil accumulation was characterized as a sigmoid, where the period with the greatest oil accumulation rate could be simplified as a linear relationship (López-Bernal et al. 2021). López-Bernal concluded that the rate of this linear equation (R3 in the current work) would be less sensitive to water stress. Current data suggest that the reduction would be important at severe levels, CAPÍTULO II 85 proving the need to determine exactly the water stress to establish a possible oil yield reduction. Studies reporting differences in the oil harvest are related to water stress levels that are very severe and last a considerable amount of time (Moriana et al. 2003; Fernández et al. 2013; Ahumada-Orellana et al. 2017; Corell et al. 2022, among others). The yield reduction could be linked to a loss of fruit due to water stress conditions during Period I (e.g., Moriana et al. 2003). However, works using water stress only during Periods II and III also indicated that yield reduction would be mainly associated to an impact on oil accumulation, seeing that the fruit drop seems less sensitive to water stress in these periods (Corell et al. 2022). Irrigation works with oil production trees that are usually rehydrated due to rainfall or irrigation reported significant yield loss only under water stress conditions with water potentials below −3.0 MPa before rehydration (Ahumada- Orellana et al. 2017) or values close to that, but with a long duration (Fernández et al. 2013). Water stress conditions with minimum potentials and less durations seemed to have a lesser impact with hardly any significance (Moriana et al. 2003; Ben-Gal et al. 2021; Corell et al. 2022). This suggests that rehydration plays a role during Period III that would allow recovering part of the oil production, although not completely. Hueso et al. (2019) suggested that water stress would not affect oil production until values close to −2.21 MPa of SWP were reached in Period III. This threshold was greatly exceeded in this study (Table 3), and this would support the hypothesis that the current study suggests a cumulative rather than an ad hoc effect, in line with the other authors (Moriana et al. 2003; Fernández et al. 2013; Ahumada-Orellana et al. 2017; Ben- Gal et al. 2021). However, because, in the current work, the cultivar was for table use, these conclusions could be limited and not completely useful for oil cultivars. Under no circumstances, did the difference in fruit load affect the oil accumulation period or the amounts produced in the fruit (Figure 4); however, it did affect the relationship with the water stress indicators (Table 4). During the low fruit load year, Period II had a lower sensitivity to water stress than Period III, which, conversely, resulted in some similar relationships between both seasons (Figure 5; Table 4). The massive pit hardening in Period II turned the fruit into a preferred sink for water over the leaves, resulting in the absence of response seen in the low fruit load year. On the other hand, during Period III, once Period II is over, the strength of the fruit as a sink could be reduced, thus making it more sensitive, even in low fruit load seasons. CAPÍTULO II 86 5. Conclusion The absence of changes in endocarp hardening patterns due to water stress conditions ensures an accurate schedule of RDI using this process. However, because experiments similar to the current one were scarce, this has to be checked in further studies with different cultivars. The optimization of deficit irrigation would also need continuous tree water status monitoring because the main yield components (fruit size and oil amount) were related to an accumulation of water trees. The fruit growth stage was the most sensitive process to water stress, being affected when the latter was around 30 MPa day. However, the fruit volume reduction could be reversed to levels equal to Control when it happened at Period II but not earlier. The oil accumulation started during Period II, and the rate was low (R2) in all treatments. This process was moderately resistant to water stress, but during the period of higher accumulation rate (R3), it was linked to the water status indicators of Periods II and III. Only in the year with a very low fruit load, was the fruit growth in Period I and the oil accumulation in Period II not related to the water stress indicators. This suggests that the sclerification process acted as a large water sink that isolated the fruit from the leaves under conditions of a very low load. All these conclusions show the need to accurately establish the process of hardening in the fruit. Although changes between cultivars, mainly because of their final use, could be expected, the value of cumulative stress would be demonstrated. Further work is needed to consider oil cultivars in similar studies and, in general, different levels and durations of water stress that permit the fitting threshold values accurately. 87 CAPÍTULO III: Response of young super-hight density table olive orchard (Manzanilla de Sevilla) to different water stress levels considering an accurate determination of endocarp development. 88 CAPÍTULO III 5. Capitulo III: Response of young super-hight density table olive orchard (Manzanilla de Sevilla) to different water stress levels considering an accurate determination of endocarp development. Publicado en Agricultural Water Management, 303, 109044. http://doi.org/10.1016/j.agwat.2024.109044 Los olivos son especies extremadamente resistentes a la sequía. Esta particularidad podría utilizarse para reducir las necesidades de agua en la programación deficitaria del riego. Sin embargo, la respuesta del rendimiento al déficit hídrico está relacionada con el momento y el nivel de estrés hídrico impuesto. El endurecimiento del endocarpo se ha utilizado comúnmente como etapa para disminuir el riego, pero no se mide en condiciones de campo. El objetivo de este trabajo es describir el impacto de las condiciones de estrés hídrico en la respuesta del rendimiento para los olivos de mesa, considerando una determinación precisa del desarrollo endocarpo. El experimento se realizó durante las temporadas 2020-2023 en un olivar de mesa superintensivo de dos años (cv Manzanilla de Sevilla). Los tres tratamientos de riego (seis repeticiones por tratamiento) fueron: Control, sin estrés hídrico durante 2020 y 2021, y estrés hídrico moderado para reducir el vigor en 2022 y 2023; Riego deficitario controlado (RDC), basado en un estrés hídrico de moderado a grave durante el endurecimiento del endocarpo, utilizando mediciones del potencial hídrico del tallo al mediodía; y Secano. El desarrollo de endocarpo se estableció utilizando la presión de rotura durante las temporadas 2021, 2022 y 2023. Se midieron las relaciones hídricas, el crecimiento vegetativo, el número de inflorescencias y frutos, el rendimiento y la calidad del fruto. El volumen del seto fue limitado en el tratamiento Secano, pero no en el RDC, comparado con el Control. El número de inflorescencias y conjunto de inflorescencias no se vio afectado en RDC y en el Secano, solo se redujo cuando el estrés hídrico fue severo en 2023. El RDC presentó una respuesta al rendimiento similar al Control, con una reducción del 50 % en el agua aplicada y un potencial hídrico del tallo a mediodía significativamente más negativo. Los frutos del tratamiento Secano no eran aptos para la producción de mesa ya que la rehidratación no recupó el tamaño de fruto lo suficiente para ser comercial. El endocarpo mantuvo la capacidad de crecimiento hasta el final de la temporada. CAPÍTULO III 89 1. Introduction The scarcity of water resources limits agricultural activity around the world (Steduto et al., 2012). Olive (Olea europaea L.) was typically a rainfed fruit tree, but nowadays it is frequently irrigated to increase productivity. According to Steduto et al. (2012) the relative increase in crop evapotranspiration of irrigated olive orchard from 50 % to 90 % would enhance relative fruit yield from 40 % to 100 %. Therefore, irrigated olive orchards started spreading widely from the 90’s of last century, when tree density increased greatly. This change in the cultivation system was strongly based on a reduction in harvest costs and the increase of the surface dedicated to super-high density olive orchards. Super-high density orchards would reduce harvest costs around 35 % in comparison with high density orchards (Connor et al., 2014). Therefore, this fruit tree competed for water resources, and, in some areas, it was one of the most important petitioner of irrigation, e.g. in Spain, with more than 400,000 ha of irrigated olive orchards (MAPA, 2024). However, olive trees are considered a drought-resistant species in which an irrigation scheduling based on plant water status could improve yield response to water stress (Ben-Gal et al., 2021). Regulated deficit irrigation can be used to optimize water resources, adjusting water inputs to the phenological periods in which a water deficit does not affect yield and quality of the crop (i.e. Moriana et al., 2003). The duration and timing of the water stress cycle determines the plant vegetative and productive response (Hsiao, 1990). It is, therefore, necessary to establish the different phenological stages of the olive trees to understand the impact of water deficit on the yield response (Fernández, 2014). Phenological stages of olive trees for deficit irrigation scheduling were commonly determined based on endocarp (pit) development. Pit hardening was considered the most resistant period to water stress (Goldhamer, 1999; Moriana et al., 2003; Fernández, 2014, among others). However, it has not been easy to determine this period under field conditions, although it usually occurs in the middle of the summer season. Recently, Sánchez-Piñero et al. (2022) reported endocarp development of different cultivars using the description of the pitbreaking pressure pattern throughout the season, concluding that the beginning of this pit hardening period, Period II, could be estimated as 49 days after full bloom. However, the length of this Period II was not constant and changed according to cultivar and thermal time (Sánchez-Piñero et al., 2022). These results imply that deficit irrigation scheduling based on a constant period of water stress around mid-Summer could include different CAPÍTULO III 96 and maximum number of inflorescences were estimated considering only the date when the maximum number was measured. The number of fruits per inflorescence in these shoots was counted from the beginning of pit hardening until harvest. These latter data only included those shoots in which fruits were observed at the beginning of pit hardening. Shoot expansion, inflorescences and fruit number were not measured during the 2020 season. Each plot was harvested for green style preparation in early September, except Rainfed treatments in 2021 and 2022. The three lines of each plot inside the edge lines (13 trees each) were hand harvested and weighted. In each plot, a sample of 2 kg was collected for quality determination. These samples were processed in the laboratory. Around a subsample of 0.5 kg of fruits was counted to obtain an estimation of the fruit size (number of fruits per kilogram, USDA, 2019). One hundred fruits were selected according to a colour maturity index (IM, Hermoso et al., 1997). In addition, three subsamples of 10 fruits per plot were randomly collected to estimate the pulp to pit ratio in fresh and dry weight (60ºC until constant weight). Fruit texture was determined for 10 fruit per plot by measuring the maximum peak force of the first compression (Szychowski et al., 2015) using a force gauge (FM 200, PCE Instruments, Spain). Applied water was measured in the main pipe at the beginning of each plot with a water-meter (Hydrojet, Hidroconta, Spain). At the end of each season, the number of trees in each plot was counted to obtain the percentage of tree survival. 2.3. Statistical analysis In the endocarp development, the Sigmoid parameters were estimated with nonlinear regression tools with the package “nls2” in R environment (R Core team, 2013). The statistical analysis in most of the measurements was conducted with ANOVA and a separation of means (Tukey Test) using the Statistic (SX, v 8.0) program. Significant differences with a p-level <0.05 were considered in both tests. The assessment of the water stress effect on inflorescence set was done with multivariate and simple analyses. Outliner data in this relationship was estimated with the relationship between error and expected values. Data which error were higher than 4-fold the average of each season was consider outliner. Comparison between fits considered R2, standard error and mean square error to discuss the best relationship. CAPÍTULO III 97 3. Results 3.1. Climate Reference evapotranspiration (ETo) and rainfall presented the typical pattern of a Mediterranean climate, although there were some extreme seasons (Fig. 1). The climatic conditions showed a warm winter with an extremely hot and dry summer. Seasonal ETo was 1273 mm only in 2020, while it was 1373 mm in 2022 and 1365 mm in 2023. Maximum daily ETo was greater than 7.5 mm in all seasons and greater than 8 mm in 2022 and 2023. The period in which daily ETo was higher than 5 mm was usually from the beginning of May until mid-September, although in 2022, it started in mid-April and, in 2023, on the 3rd of April (Figure 1). Figure 1: Reference evapotranspiration (ETO), rainfall and maximum temperature throughout Maximum temperatures showed a similar pattern to ETo. Absolute seasonal maximum temperature varied between 39.9 ºC in 2020 and 44.2ºC in 2022, with values higher than 43ºC in the other two seasons (44.2ºC in 2021 and 43.7ºC in 2023). All of them were greater than the historical seasonal average of maximum temperatures (36.0 ºC, AEMET, 2024). In the period when fruit set occurred, from April to the end of June, maximum daily temperature oscillated between 36.6ºC in 2021 and 42.9º C in 2023. The average maximum temperatures in this period showed the lowest values in 2020 (26.7ºC) CAPÍTULO III 98 while 2023 was the hottest (30.8ºC), with intermediate values in the other two years, around 28ºC. The distribution and amount of rain also presented extreme values throughout the 4 seasons of the experiment (Fig. 1). All years experienced a lower amount of seasonal rain than the 30-year average (539 mm., AEMET, 2024). Only in 2020 this reduction was lower than 10 % (493 mm), but in the rest of the seasons, the decrease was greater than 30 % (357 mm, 2021; 362 mm, 2022; 289 mm, 2023). The distribution was also different in comparison to the average season (Fig. 1). Summer was the driest period every year, but the interval from January to April was received more rain than the average year. In this period, more than 50 % of the seasonal rain occurred and the volume was greater than the historical average (186 mm, average year vs 268 mm, 2020; 210 mm, 2021, 194 mm, 2022), except in 2023, when only 51 mm were measured (27 % of the seasonal average). The beginning of the rainy period, typically from September onwards, started on those dates in all seasons, except in 2022, when it was delayed until mid- October. However, the amount of rain in September, when harvest occurred, was lower than the average year for this month (27 mm, AEMET, 2024) in 2020 (16 mm) and 2022 (4 mm). Only in 2023 season the amount of rains were greater (54 mm) with a unique event of 47 mm at the beginning of the month. 3.2. Endocarp development The pit-breaking pressure (PBP) of the endocarp showed the three different phases in which water relations data will be discussed (Figure 2). Figure 2 depicts data from the 2023 seasons. Control and RDI presented a similar pattern throughout the 2023 season. PBP was almost constant in the three treatments until around 50 days after full bloom (DAFB), but the Rainfed curve changed from that date on, marking the beginning of Phase II. While irrigated trees showed a fast increase of PBP in this period, Rainfed ones displayed a significantly lower hardness at 56 and 70 DAFB, with a recovery of PBP from 70 DAFB. At the beginning of Phase III, when PBP was at its maximum level, Rainfed trees presented lower values than irrigated trees, significantly different by 112 DAFB. Only on the last date of measurement, after a heavy rain, hardness was almost the same in all treatments. CAPÍTULO III 99 Figure 2: Pattern of endocarp pit-breaking pressure during the 2023 season. Each symbol is the average of 3 data. Vertical bars represent the standard error. Vertical lines define the different phases. Control, blue circle; RDI, green diamond; Rainfed, red square. Asterisks indicate significant differences between treatments on that date. The adjustment of these data to a sigmoid equation is presented by the parameters of Table 1. Only maximum hardness (parameters a and a+d) was significantly different between Rainfed and the irrigated trees. In Phase III, fruit from Rainfed presented PBP values around 10 % lower than irrigated trees. In addition, the determination coefficient (R2) was also lower, on average, in Rainfed trees. This decrease in R2 was likely related to the recovery pattern during Phase II. The rate of increase during Phase II (parameter b) was equal in all treatments. However, the date when 50 % of PBP occurred (parameter c) took place with approximately 1 day later in RDI and Rainfed versus Control trees, although such differences were not significant. CAPÍTULO III 10 Table 1: Parameters (average±standard error) of the fit adjusted in the individual pitbreaking pressure curves for the 2023 season. Sigmoid equations for modelling the pitbreaking pressure (PBP) according to the equation 𝐹𝐹𝑃𝑃𝐹𝐹 = 𝑎𝑎 1+𝑒𝑒 −𝑏𝑏(𝐷𝐷𝐴𝐴𝐹𝐹𝑃𝑃−𝑐𝑐) + 𝑑𝑑. Each data is the average of 3 plots. Different letters in the same column indicate significant differences (Tukey Test, p<0.05). *a, Maximum pit-breaking pressure (MPa); b: rate of pit-breaking pressure increase (MPa day-1); c. days after full bloom for 50% hardening; d, minimum pit-breaking pressure (MPa). 3.3. Water relations The SWP pattern throughout 2023 showed strong changes in the most stressed trees due to the rain period (Fig. 3 and Fig. 1 around DOY 150). Significant differences were measured on most dates between Rainfed and Control, while RDI was often near Control. Control SWP was near to the baseline equation in Phase I and during preflowering period. From Phase II, this treatment presented a small decrease respect to this equation. The decrease in SWP was sharp in Rainfed and RDI, although in the former it was disrupted by several rain events that allowed an almost complete recovery (on DOY 144, 34 DAFB; DOY 166, 56 DAFB; DOY 249, 139 DAFB). Minimum values in Rainfed SWP were around -6 MPa from DOY 186 until harvest when a previous rain almost recover this treatment. On the other hand, RDI was irrigated to be able to reach a maximum water stress level of approximately -3 MPa at Phase II. Differences between RDI and Control were significant mainly in the period from DOY 151 to 193. In addition, during Phase I, an irrigation problem increased RDI water stress more than expected in that period with significant more negative values than Control and around -3 MPa though irrigation permitted a completely recover with near value to the baseline equation. CAPÍTULO III 101 Figure 3: Pattern of Stem water potential (SWP) throughout the 2023 season. Each point is the average of six data. Vertical bars represent the standard error. Solid lines indicate full bloom (left) and harvest (right). Dash lines limit the Phase II of fruit development. Control, blue circle; RDI, green diamond; Rainfed, red square. Solid line represented Corell et al. (2016) baseline. Asterisks indicate significant differences between treatments on that date (p<0.05, Tukey Test). Table 2 showed data of minimum SWP and SI in four different phases of the 2023 season. These phases were delimited by the PBP curves (Figure 2 and Material and Methods). Only in the pre-flowering phase there were no important differences between treatments. This period presented the lowest water stress level in all trees, although in Control, SI data were near to the ones estimated in Phase II or Phase III. Greater differences than previously were found in the rest of phases in both indicators (Min SWP and SI). Minimum SWP values on Control trees were always considerably more positive than RDI, and this, more so than in Rainfed. These values decreased in all treatments throughout the season, and the most negative SWP were measured in the Phase III of endocarp development. Similarly, SI values of Control were significantly lower than RDI and Rainfed in all phases. The maximum SI was estimated in Phase II in all treatments using the greatest value in Rainfed (more than 200 MPa*day, only in this period). CAPÍTULO III 10 Table 2: Minimum midday stem water potential (SWP, MPa) and Stress Integral (MPa*day) in pre-flowering and the three phenological phases according to endocarp development throughout the 2023 season. Each value is the average of 6 data. Different letters in the same row indicate significant differences (p<0.05, Tukey Test). Gas exchange was considerably reduced in Rainfed trees during summer of the 2023 season (Fig. 4). These measurements started from the beginning of June, just after a period of rain (Fig. 1). Under those conditions, trees presented a similar midday net photosynthesis (Fig. 4). From this date, all treatments trended to reduce their values. Photosynthesis in Control and RDI treatments had a similar pattern with values between 4 to 6 µmol m-2s-1. In that period, from DOY 172 to 242, maximum temperature was between 30 to 40 ºC with ETo values higher than 6 mm day-1 (Fig. 1). The last data of these treatments (DOY 258), after harvest, reached their maximum photosynthesis values, between 7 to 8 µmol m-2s-1 in cooler conditions with a maximum temperature of 29.4 ºC and ETo around 4.5 mm day-1 (Fig. 1). In the hottest period, DOY 172 to 242, Rainfed net photosynthesis sharply decreased and reached values near zero. These minimum values were measured in conditions of very high temperature (40ºC, DOY 200). The last date of measurements was performed after harvest and a great rain even on DOY 246 (46.7 mm, Fig.1). Such conditions permitted the recovery of Rainfed photosynthesis which were near to irrigated trees and around the maximum values of the dataset. CAPÍTULO III 103 Figure 4: Pattern of Midday Net Photosynthesis during the 2023 season. Each symbol is the average of 6 data. Vertical lines represent the standard error. Control, blue circle; RDI, green diamond; Rainfed, red square. Asterisks indicate significant differences between treatments on that date (p<0.05, Tukey Test). 3.3 Vegetative growth. The crown volume patter was similar between seasons, with the greatest differences at the end of the growth period (Figure 5). In the first season (2020), significant differences were found between all treatments from September onwards, just after harvest. The decrease of this measurement (the two last data point) in this season was related to pruning just after harvest. After this season, more frequent and earlier measurements were possible. Only in 2021, the three treatments were significantly different from the beginning and throughout the season. On the first date, RDI presented around 20% and Rainfed around 50% less than Control. The seasonal pattern was also different between treatments, Control presented a greater increase of crown volume than RDI and Rainfed. The latter displayed a very low growth throughout the season, only around 5% increase from the beginning of the season. The two irrigated treatments increased their differences at the end of 2021 season up to 27%. CAPÍTULO III 10 Figure 5: Pattern of crown volume and rainfall throughout the four seasons of the experiment. Each point is the average of six measurements. Vertical lines indicate the standard error. Dash lines limit each season. Control, blue circle; RDI, green triangle; Rainfed, red square. Asterisks show the date when statistical differences were found (p<0.05, Tukey Test). The last two seasons, 2022 and 2023, Control and RDI showed the common pattern of mature orchards. Crown volume at the beginning was clearly lower than at the end of the previous season because of the winter pruning. Such decrease caused, in 2022 and 2023, both irrigated treatments to be very similar at the beginning, approximately 10% lower in RDI, with no significant differences. On the contrary, Rainfed did not reduce crown volume between seasons because pruning was almost null. Control and RDI presented a continuous growth throughout 2022 and 2023 but at a lower rate in RDI than in Control. This pattern caused important differences between treatments at the end of the season, with a decrease of around 25% in RDI versus Control. Conversely, Rainfed presented a concentrated growth period during the spring rainfalls, greater in 2022 than in 2023, and with significantly lower crown volume values than both irrigated treatments. The pattern of shoot expansion changed greatly between seasons (Fig. 6). Each year, the beginning of data collection varied between DOY 80-120. In all the tree years, the seasonal pattern was similar with an important increase of expansion on the first dates and a strong reduction from the beginning of pit hardening. The year 2021 presented the CAPÍTULO III 105 lowest rate of increase, showing no significant differences between treatments. In this season, Control started with a lower shoot expansion than Rainfed and RDI, but, in the last measurements, Control had a higher shoot growth rate than the rest, although without significant differences. Spring rainfall in this season was around 98.6 mm, greater than the 10 years-average (95.4 mm, AEMET 2023). In the year 2022, the three treatments showed a very fast expansion before pit hardening, but there was also a sharp change around this date. Rainfed treatment had a significantly greater expansion than the other treatments on the first two dates and remained higher during all the measurements. Control and RDI treatments had a similar expansion and did not stop from pit hardening onwards, as Rainfed did. Figure 6: Pattern of shoot expansion and rainfall throughout the three seasons of the experiment. Each point is the average of six measurements. Vertical lines indicate the standard error. Control, blue circle; RDI, green triangle; Rainfed, red square. Vertical dash lines limit the period of pit hardening. Asterisks show the date when statistical differences were found (p<0.05, Tukey Test) CAPÍTULO III 11 In 2021 and 2023, there were no marked differences between treatments on any of the dates considered. Only in 2022, RDI presented significantly higher values than Rainfed on both sample dates, and Control was an intermediate treatment. In this latter season, RDI presented the maximum values of the three years, while Control and Rainfed were very similar. There were no differences between dates within the same season in any of the three treatments. 3.5 Quantity and quality of harvest. Table 4 presents the data related to yield (quality and quantity), seasonal applied water and tree survival throughout the four seasons of the experiment. The pattern of the yield during the 4 years showed the changes from a young to a mature orchard. The first commercial yield was harvested in the 2020 season and was the lowest of all seasons in the three treatments. From 2021, an alternate bearing pattern was observed in irrigated trees, with the highest yield in 2021 and the lowest in 2022. Rainfed treatment presented a significant reduction in yield in comparison to Control in all seasons, except 2022. These reductions were important and varied from around 55% in the lowest yield seasons (2020 and 2022) to 93% in the two greatest yield ones (2021 and 2023). Differences between RDI and Rainfed were lower than the ones with Control and only significant in 2021 and 2023. Differences in yield between RDI and Control were important only in 2021, with a reduction of around 27%. In the rest of the seasons, RDI also tended to lower values with reductions between 12% in 2020 and 29% in 2022. Fruit size is one the most important features of yield quality (Table 4). Rainfed trees presented significantly higher values of fruits per kilogram, therefore lower fruit size, than the other two irrigated treatments. These differences were obtained even though these trees were harvested always after a short period of rehydration with irrigation (2020 season) or rain (rest of seasons). The values obtained in Rainfed trees were higher than 350 fruits per kilogram in all seasons (Petite class, USDA 2019), and the fruits were considered unsuitable for the table industry. RDI and Control values presented no meaningful differences in fruit size. Minimum values in Control were measured in 2022 (199 fruits kg-1, extra-large class, USDA 2019), while in RDI were obtained in 2023 (196 fruits per kg-1, extra-large class, USDA 2019). The worst fruit sizes in both treatments were harvested in 2021, the highest yield season (Control, 233, and RDI, 251 fruits kg-1, Medium class, USDA 2019). In all seasons, RDI tended to smaller fruit size than Control, CAPÍTULO III 113 with a reduction between 8 and 15%, except in 2023 when RDI produced the largest fruits. Table 4: Quantity and quality of yield, applied water, and percentage of survival trees in the four seasons of the experiment (average [6 data] and standard error). Different letters at the same measure and season indicate significant differences (p<0.05, Tukey test). * CON, Control treatment; RDC, Regulated Deficit Irrigation treatment; RAF, Rainfed treatment; Yield, kg ha-1; Size, fruit size, number of fruits per kg; Class, Commercial Class (USDA); AW, seasonal applied water, mm; Survival, Percentage of survival trees Irrigation management produced meaningful greater differences in applied water (Table 4). Rainfed trees were not irrigated from the end of 2020. During this first season, this treatment was irrigated before the beginning of the experiment and the week before harvest. RDI reduced the water applied in comparison to Control by 52% (2020), 57% (2021), 72% (2022) and 43% (2023). Despite these large differences in water applied, the percentage of tree survival in each treatment was not significant differences from 2020 to 2022. However, Control presented the highest percentage of trees (around 95%) and RDI the lowest (88%) in the irrigated treatments. Only from 2022, there was a decreasing trend in Rainfed, which presented significantly lower percentage of tree survival in 2023 (84 vs 88 and 94%). Data related with several fruit features are presented at Table 5. Results of fruit texture and maturity index (MI) were usually related. When fruit ripening progressed, fruit texture decreased and MI increased (Table 5). Control and RDI data were not markedly different in any of the seasons. In 2020, when harvest was delayed, fruit texture CAPÍTULO III 11 was considerably lower in Control and RDI than Rainfed because the MI was slightly, but significantly, greater in the former treatment. In the following two seasons, 2021 and 2022, Control and RDI were harvested for green style preparation, but Rainfed trees were harvested later, after the first rains. Rainfed presented a significantly lower values of fruit texture because they were in a more advanced MI (around 3, fruit veraison, vs around 1, green fruit). In 2023, all treatments were harvested on the same date because rain occurred earlier than in previous seasons. This year, fruit texture was almost equal in all treatments and similar to the values obtained in the green fruit stage of previous seasons (around 50 N), but MI was significantly different. Rainfed presented values greater than 3, while irrigated trees were around 1, the optimum for green style preparation. Pulp/stone ratio (PS) is another important feature in yield quality. No differences were found in PS in fresh or dry weight between Control and RDI (Table 5). Although PS in fresh weight in RDI tended to be lower than in Control. On the other hand, PS was significantly reduced in Rainfed in comparison to irrigated trees. Only in the 2021 season and in dry weight, those differences were not significant. Such reductions were more severe in the 2022 and 2023 seasons, when Rainfed was around half the PS than irrigated trees. Table 5: Several fruit features related with yield quality (average [6 data] and standard error). Different letters at the same measure and season indicate significant differences (p<0.05, Tukey test). * CON, Control treatment; RDC, Regulated Deficit Irrigation treatment; RAF, Rainfed treatment; Texture, Fruit texture, N; PSF, pulp stone ratio in fresh weight; PSD, pulp stone in dry weight CAPÍTULO III 115 4. Discussion Yield response in fruit trees is related to several factors, which change their relevance at different stages in the orchard’s lifecycle. The current work presented data from the transition from a young to a mature orchard, when these changes occurred. Crown volume was the main factor in the yield respond in 2021, when irrigated trees reached yield levels close to the maximum (Figure 5, Table 4). Tree size was probably the main limitation that strongly reduced Rainfed yield throughout the experiment but did not delay the moment in which the first yield was harvested. Similar results in relation to the beginning of flowering were reported in irrigated hedgerow orchards (Gómez-del- Campo, 2010) and would suggest a strong drought-resistance of the flower induction process. In fact, shoot and crown volume growth have been reported as the main limitations of yield response in olive trees (Connor et al. 2014; Pierantozzi et al. 2020). However, because the vegetative growth was very sensitive to water stress (Hsiao, 1990), the amount of irrigation to achieve maximum crown volume as soon as possible would be considerable (as in current work, Table 4) and not always clearly effective (Pérez- López et al., 2007; Gómez-del-Campo, 2010). A large amount of irrigation in Control of the current work to maximize growth secured a significant yield increase, mainly in 2021 (Table 4, Figure 5), but it also increased pruning cost and shading problems for future yield (Pastor et al., 2007; Trestaconte et al., 2022). When maximum crown volume was reached, according to each orchard conditions, shoot growth could be less limiting and the current season yield components would be more important. Rainfed treatments presented an adequate shoot growth in all seasons and some of them were not different to the irrigated trees (Figure 6), even under very low rainfall conditions (Figure 1). Moreover, both irrigated treatments were almost equal in shoot growth (Figure 6) but not in crown volume patterns (Figure 5). Therefore, under conditions of maximum crown volume, the yield decrease would present a greater resistance to water stress, because vegetative growth would be not the limiting factor. Then, the increase in crown volume could be minimized during the season, and just a short period of shoot growth would be enough. These would explain the lack of yield differences in some irrigation publications with water stress conditions during the vegetative growth period (Lavee et al, 2007; Gómez-del-Campo et al., 2010; Pierantozzi et al., 2020). The number of fruits and inflorescences were two of the most important components affecting yield response in the current work. Water stress during two CAPÍTULO III 11 different periods probably reduced the maximum number of inflorescences. Preflowering water stress was the least common, even in Rainfed trees, and only related to the reduction in all treatments during the 2023 season (Table 2, current work and Table 3 in Sánchez-Piñero et al., 2024). Rapoport et a (2012) reported, in a pot experiment, a great decrease in the number of inflorescences with a dry cycle during inflorescence formation, greater than during flower formation. Values of SWP suggested a moderate water stress in this period (between -1.3 to -1.8 MPa; Table 2), but SI values in 2023 increased until more than 20 MPa*day (Table 2), and they were lower than 15 MPa*day in the rest of seasons (Sánchez-Piñero et al., 2024). Several authors suggested threshold values of SWP during this period in relation to inflorescence and fruit number. Beyá-Marshall et al. (2018) reported that values more negative than -2 MPa would lower the number of flowers per inflorescence. Hueso et al. (2021) reported more positive values as thresholds for different processes with general recommendations of values more positive than -1.5 MPa during the pre-flowering period. Minimum SWP values of the current work during 2023 would be near Hueso et al. (2021) values, although SWP Control was slightly more positive (Table 2). These results would suggest that the duration could also be important, and the SI could be an additional indicator for managing this uncommon period of stress. On the other hand, the pattern of maximum number of inflorescences in Rainfed trees was a continuous slow decrease of the maximum number of inflorescences during the three seasons, different to the alternate bearing pattern of irrigated trees in 2021 and 2022 (Figure 7). Inflorescence set and fruit number per inflorescence was less sensitive to water stress than the maximum number of inflorescences (Figures 7, 8, 9, 10 and Table 3). The fruit number was not affected by the water stress during pit hardening, even under Rainfed conditions (Figure 10). Several authors reported no significant fruit drop under moderate water stress levels (Gomez-del Campo, 2013 a and b; Corell et al., 2022) or even if there was a relationship between them, the impact on yield was small (Corell et al., 2020). Conversely, inflorescence set was strongly related to the SI (Figure 9 and Table 3). In this period, temperature was typically reported as one of the most limiting factors of fruit set in olive trees (Fernandez-Escobar et al., 1983, Cuevas and Rapoport, 1994; Benlloch- González et al., 2018). However, in a short period of time and with a narrow range of variation in temperature, the water stress becomes the most important factor. Several authors reported a decrease in fruit set under water stress conditions during this period, with water stress levels ranging from -2 MPa of pre-dawn water potential (Caruso et al., CAPÍTULO III 117 2013) to -4 MPa of midday water potential (Gómez-del-Campo, 2013 a and b). However, in these research publications, the reduction in fruit set was not clearly linked to these threshold values of water potential. In the current work, Rainfed and RDI presented more positive values of SWP in this period (Table 2, current work, and Table 3 in Sanchez- Piñero et al. 2024) than in the literature (Caruso et al. 2103; Gómez-del-Campo 2013 a and b), but to a longer duration of the stress conditions. An SI of around 20 MPa*day could be a threshold value for deficit irrigation management in Phase I (Figure 9). On the other hand, the reduction in the number of inflorescences in Rainfed trees would likely relate with the affection of flower induction. This process is complex but part of it occurred during the early fruit development before the finish of pit hardening (Stutte and Martin, 1986). Severe water stress conditions of Rainfed trees during this Phase II likely reduced the amount of assimilate which could affect to the partitioning toward the potential flower buds of the next season. Data of plant water status from Rainfed suggested that such reduction was likely related to the extremely severe water stress during pit hardening in the previous season. In the scientific literature, SWP data around -3MPa during pit hardening did not reduce the maximum number of inflorescences in the next season in “Manzanilla de Sevilla” cultivars (Girón et al., 2015; Corell et al., 2020) and oil cultivar (cv Morisca, Moriana et al., 2012). However, in this latter research publication, the season with a SI around 300 MPa*day presented a decrease in the number of inflorescence next season (Moriana et al., 2012). Although this latter data was seasonal, most of the SI values in RDI scheduling are reached in Phase II (i.e. Table 2). The values reached in Rainfed trees were 219 MPa*day in 2023 and 104 and 242 MPa*day in 2021 and 2022 (Sánchez-Piñero et al., 2024). In this way, Trestaconte et al. (2022) reported a strong reduction of number of inflorescences in severe shading after post-fruit set. Although these authors related such response with a decrease in shoot growth, this mainly occur before fruit set and, in fact, could be related with a strong restriction in the assimilate. An accurate management of water stress could optimize the amount of irrigation. RDI management obtained a yield quality and quantity close to Control with around 50% of applied water in most of the seasons, with just one season with significant difference, 2021 (Tables 4 and 5). The trend to lower yield and fruit size in RDI was likely related to a slightly restricted rehydration (Tables 2 and 4, Figure 3) as described in the literature (Girón et al. 2015; Corell et al. 2020). On the other hand, Rainfed was unsuitable for table olive production because fruit size was strongly affected, even after a rehydration by CAPÍTULO III 11 irrigation or rain (Table 4). This latter treatment could be considered for oil production because trees were able to survive extremely severe water stress (Table 4). Fruit size in Rainfed trees was affected by the early water stress in Period I and at the beginning of Period II of endocarp development and a too short rehydration before harvest (Table 4, Fig. 3, current work, and data at Sanchez-Piñero et al. 2023 and 2024). The most important reduction in pulp/pit ratio in fresh and dry weight in Rainfed trees occurred in 2022 and 2023 (Table 5), when endocarp development was significantly affected (Fig. 2 and Sanchez-Piñero et al., 2024). These changes occurred because of the water stress during the early period of endocarp development (Period I and beginning of II) and would void the rehydration period just before harvest. In Period I, cell division and endocarp growth were the most important processes. Water stress in this period reduced the final fruit size, even in experiments with a rehydration period longer than the one of the current work (Rapoport et al., 2004; Gucci et al. 2007; Caruso et al. 2013; Gomez del Campo, 2013a; Gomez del Campo et al. 2014; Gucci et al. 2019). Smaller fruit sizes were related to the decrease of endocarp size and not linked to the number of mesocarp cells (Rapoport et al. 2004; Gómez del Campo et al., 2014). The recovery of the endocarp hardness at the end of 2023 (Figure 2) showed that endocarp development was a preferential part of the plant in the rehydration process. This recovery capacity could worsen fruit quality with the greatest reduction in pulp/pit ratio (Table 5) because would compete with the mesocarp growth (Rapoport et al., 2004). This did not occur in RDI with the same pattern of endocarp development than Control (Figure 2, current work and data at Sánchez-Piñero et al. 2024). Rapoport et al. (2004) reported an endocarp recovery after a water stress before 8 weeks after full bloom but fruits at harvest presented a significant smaller size. This latter research suggested that water stress changed the length of Period I of endocarp development (Rapoport et al. 2004). However, endocarp would actually retain the capacity of increasing its hardness throughout the season (Figure 2). These pattern could be also affect to the production of phytohormones which could be consider in further works. Current data suggest that irrigation needs at super-high density table olive orchards could be reduced with an accurate water management as other authors suggested for oil production (Ben-Gal et al., 2021, among others). In mature orchards, periods of Phase I and Phase II of fruit development could reduce irrigation and delay the beginning of the irrigation season. Fruit set, one of the most sensitive yield component, could be managed with midday water potential if SI was not greater than 20 MPa*day. Probably the most CAPÍTULO III 119 dangerous part of the season would be the end of Phase I and beginning of Phase II when cell division is still important. According to the literature, this process, cell division, would be more drought sensitive than fruit set because values between 6 to 16 MPa*day has been suggested (Sánchez-Piñero et al., 2024). All these latter thresholds would have to adjust accurately in future works. Along Phase II, massive pit hardening, according to current data, water stress level could increase greatly until -3 MPa. However, rehydration before harvest should be long enough to reach the maximum potential fruit size as in RDI. This latter part of the season would be the most limiting in table production in comparison to oil orchards. CAPÍTULO III 12 5. Conclusions Crown volume was the most important yield constraint, but other effects would be more limiting when trees reached maximum crown volume values in the orchard’s lifecycle. Then, the maximum number of inflorescence and fruit set appear as the most important constraint in water stress management. However, the decrease of these two yield components in comparison to Control during Phase I of endocarp development occurred only under severe water stress conditions. Moderate water stress, such as in RDI trees, would secure maximum fruit set and adequate cell division phase in fruit development. Therefore, when a moderate water stress was imposed, even under Rainfed conditions, rehydration was the real limiting factor to obtain an adequate yield response in terms of quantity and quality. Extremely severe water stress levels in Rainfed trees during Phase II, endocarp massive hardening, also reduced the number of inflorescences in the following season. Moderate water stress conditions would consider the water stress level and duration, then SI indicator could be adequate to an accurate deficit irrigation scheduling. In this way, all continuous measurement (such as IoT systems) would have a great important in the future RDI scheduling. These moderate water stress conditions, as in RDI, produced a normal endocarp development, which would allow an efficient rehydration. Otherwise, endocarp would increase its hardness during the rehydration and reduce the increase in fruit size. Those restraints mean that a Rainfed treatment in superhigh density orchards will be unsuitable for table production. However, the survival capacity of trees and the level of production of Rainfed trees under the current experiment conditions would suggest that the irrigation needs could be strongly lowered under conditions of severe water scarcity. This latter conclusion, which is linked to an irrigation scheduling based on plant water status, could provide new approaches to water resources administrators. The most severe water stress level would be interesting even in irrigated trees during drought periods when water availability would be strongly limited. However, the long term effect of such conditions would have to consider in further works, mainly the response of leaf conductance after the recovery of the trees. 121 CAPÍTULO IV: Managing Water Stress in Olive (Olea europea L.) Orchards Using Reference Equations for Midday Stem Water potential. CAPÍTULO IV 128 approximately 6 km away from the experimental station. Slopes of these regressions were compared to published baselines, for temperature (Corell et al., 2016) and for VPD (Shackel et al., 2021) using a T-test (p < 0.05). 3. Results Water relations were significantly affected in the three seasons. The midday stem water potential (SWP) showed extremely low values in the Rainfed treatment (Figure 2). Although the decrease in Rainfed trees was delayed in 2020 due to the COVID-19 pandemic (Figure 2a), the seasonal pattern was very similar with a fast decrease down to values between −5 and −6 MPa. During the 2020 season, Rainfed trees completely recovered in two weeks by the end of the experiment (Figure 2a), while in 2021 and 2022, the absence of rains kept the SWP values at very severe water stress levels (Figure 2 b,c). Figure 2. Midday stem water potential pattern throughout the (a) 2020, (b) 2021, and (c) 2022 seasons. Each point is the average of six measurements. The green solid line represents Corell’s baseline. Vertical bars are the standard error. Asterisks show the date when statistical differences were found (p < 0.05, Tukey Test). Significant differences between Rainfed and Control were found early in the season, although at the beginning they were very similar. The Control trees presented a CAPÍTULO IV 129 similar pattern in the three seasons. The SWP values in the Control were close to Corell’s baseline, reaching minimum values in mid-summer, at around DOY 200, in all seasons. Conversely, the RDI treatment trees showed a response that could be placed between Rainfed and Control. Although the water status of the RDI was more similar to the Control, major differences between the RDI and Rainfed trees were delayed until severe water stress conditions of Rainfed occurred at around −3 MPa. The RDI and Control trees presented significant differences but only on a few dates in the 2021 and 2022 seasons. Although these differences were not always important, Figure 2 shows clear periods of water stress in the RDI. The length of these periods and the level of water stress also changed for the different seasons. During 2020 (Figure 2a), the RDI presented the shortest period, around three weeks, while in the 2021 season it was the longest with more than two months (Figure 2b). The pattern of the SWP in the RDI was similar but not parallel to the Control, mainly during the 2022 season (Figure 2c). Midday leaf conductance also presented significant differences throughout the season (Figure 3). CAPÍTULO IV 130 Figure 3: Midday leaf conductance pattern throughout (a) 2020, (b) 2021, and (c) 2022 seasons. Each point is the average of six measurements. Vertical bars are the standard error. Asterisks show the date when statistical differences were found (p < 0.05, Tukey Test). Leaf conductance in the Rainfed trees was considerably lower than the Control from mid-season but several weeks later than the SWP. In 2020, the rehydration period increased the rainfed leaf conductance to the same level compared to the Control and at a similar time compared to the SWP (Figure 2a), but no recovery was measured in the 2021 and 2022 seasons in rainfed trees. Differences between the Control and Rainfed treatments by the end of both seasons were smaller but still significant and likely related to a decrease in the Control values (Figure 3b,c). Control leaf conductance was very variable between seasons and days. Most Control measurements were around 400 mmol m-2 s-1 and tended to decrease by the end of all seasons. The RDI was an intermediate treatment between Rainfed and Control in all seasons, but with only a few significant differences (one or two per year) with Control commonly around DOY 200. The RDI CAPÍTULO IV 131 trees tended to greater leaf conductance values than the Rainfed ones in all seasons, but they were important only from DOY 200 in 2021 (Figure 3b). The pattern of soil water extraction changed along the season with the irrigation management (Figure 4). Figure 4. Soil moisture at different soil depths at three different dates of the 2021 season. (a) At the beginning of the season (DOY 109); (b) At mid-water stress period (DOY 203); (c) Just before harvest (DOY 245). Each symbol was the average of six data. Horizontal bars represent standard error. Asterisks indicated significant differences at the depth where they are located (p < 0.05, Tukey Test). Figure 4 presented an example of the 2021 season (the high fruit load year) at three different dates: DOY 109 (around flowering), DOY 203 (mid-water stress period), and DOY 245 (just before harvest). At the beginning of the season, when small differences were found in the SWP or leaf conductance, the pattern of soil water uptake was different in the first 0.4 m. The Rainfed and RDI trees uptaked more water than the Control trees, but this water uptake was not observed in the deeper part of the profile. Such differences in water uptake were similar at DOY 203 (Figure 4b) when differences in the SWP were very clear. The RDI and Rainfed trees almost depleted the shallow part of the soil (until 0.4 m). Below that depth, there were no significant differences, but there was a trend of drier conditions in the Rainfed than in the RDI, which was near to the Control values. CAPÍTULO IV 132 Finally, at DOY 245, at the end of the rehydration period, when the Rainfed presented significant, very severe SWP values but the RDI were recovered, the Rainfed profile was almost depleted. Although no significant differences were found because of the great variability, there was a clear increase in soil moisture in the profile at the RDI treatment that was very similar to the Control. Figure 5 showed the pattern of the percentage of soil cover along the three seasons of the experiment. Soil cover data were obtained at the beginning of each season, after pruning, and at the end, after harvest. No significant differences were found in 2020 because irrigation treatments started after the main growth period. However, the Control trended to higher values than the Rainfed and RDI, and this fact better supports the water status conditions described with the rest of data. Figure 5: Percentage of soil cover along the three experimental seasons. First data (03/2020) were the average of the orchard. Each symbol is the average of six data. Vertical bars represent standard error. Asterisks indicated the date when significant differences were found (p < 0.05, Tukey Test). The decrease in the Rainfed from 2021 season was produced for a more severe pruning to reduce the water stress compared to the others. No pruning was carried out in this treatment from this date. The percentage of soil cover in the Rainfed showed a very different pattern between seasons. In 2021, the highest fruit load season, no increase was measured while growth was significant during 2022 season. However, significant differences were found between the Rainfed and the rest of treatments in all dates from 2021. The decrease in the Control and RDI at the beginning of each season was related to pruning. On those dates, there were not significant differences between both, though the RDI was around five points lower than the Control. At the end of the 2021 and 2022 CAPÍTULO IV 133 seasons, such differences in the percentage of soil cover were significant between both treatments. As in Rainfed, during the 2022 season, the one with very low fruit load, RDI and Control showed maximum growth. Figure 6 presents the relationship between the SWP and maximum daily temperature. Each season data were grouped according to their percentage of leaf conductance decrease. Table 1 shows the best fit of all data for the same interval of leaf conductance reduction. Data presented different levels of scattering in the three groups of conductance reduction, but this was smaller in the ones between 90% and 100% of maximum leaf conductance (Figure 6). The best fit of the 90%–100% was linear with a low R2 but displayed a significant relationship with Tmax. (Table 1). The slope of this relationship was not significantly different from Corell’s baseline, just slightly lower. Therefore, if they were considered as parallel, the equation in Table 1 would be around −0.3 MPa lower than Corell’s baseline (y-interception was 0 vs. 0.294 in Corell’s equation (Corell et al., 2016). Data from 90%–100% varied from values less negative than −1 to near −2 MPa (Figure 6). Figure 6: Relationship between midday stem water potential and maximum daily temperature. Each symbol is the average of several data (different each date) in the same group of leaf conductance reduction. The black solid line represents the equation ISO (Table 1). The blue line represents the equation of (Corell et al., 2016) (solid) and plus 0.5 MPa (dash). CAPÍTULO IV 134 Table 1: Best fit of the relationship between midday stem water potential (SWP) vs. maximum daily temperature (Tmax, Figure 6) or vapor pressure deficit (VPD, Figure 7) with data from the groups 90%–100%, 70%–80%, 45%–55%; and the two first groups more positive than −1.5 MPa (ISO) from Figure 6. 1 Data, amount of data in each equation; R2, coefficient of determination; SD, standard error; MSE, mean square error; Sig, slope significantly different from published baseline ((Corell et al., 2016) in Tmax and [13] in VPD); ***, p<0.001: **, p<0.01. The relationship of 70%–80% leaf conductance reduction with Tmax was also significant but weaker than the ones from 90%–100% (Table 1). These data presented a greater scattering than the previous group and varied from values less negative than −1 to slightly more negative than −3 MPa (Figure 6). Most data of 70%–80% were less negative than −2 MPa (48 of 56 data). The slope of the regression, as in the previous group, was not significantly different from Corell’s baseline (Table 1). Finally, the group 45%–55% of leaf conductance reduction had the greatest scattering in the SWP vs. Tmax relationship. The relationship between the SWP and Tmax were significant but very weak (Table 1). The slope of the equation was not significantly different to Corell’s baseline but tended clearly to more negative values. Data from this group, 45–55%, changed from less negative than −1 to around −6 MPa. The average of all data in this group was more negative than −2 MPa (20 of 48 data). According to (Díaz-Espejo et al., 2018), a SWP less negative than −1.5 MPa would be related to an isohydric respond. Then, the decrease in leaf conductance would be closely related to the evaporative demand instead of the water stress conditions. The relationship of the SWP vs. Tmax of 90%–100% and 70%– 80% data more positive than −1.5 MPa improved the fit of the previous regression (ISO Tmax equation, Standard error and MSE in Table 1). The slope of this latter regression was significantly much lower than Corell’s one (Table 1, −0.023 vs. −0.046 (Corell et al., 2016)). Then, according to this equation, the SWP would be almost constant with the increase in Tmax (Figure 6). The relationship of the SWP and maximum vapor pressure deficit (VPD) is shown in Figure 7, also with the same group of leaf conductance reduction than in Figure 6. The pattern of scattering in the three groups considered was the same as in the Tmax CAPÍTULO IV 135 relationship. The group 90%–100% presented the lowest variability and significantly lower slope than the (Shackel et al 2021) baseline. Conversely, data from the 70%–80% group had a greater variability also in the VPD relationship. The relationship of this latter group was linear too, but with a low R2 (Table 1). The worse relationship was for the 45%–55% group. The slope of this latter equation was not significantly different from the Shackel baseline, although it tended clearly to more negative values. Values from 90%– 100% and 70%–80% more positive than −2 MPa presented a significant linear relationship (ISO VPD, Table 1). The slope of this latter equation was significantly less negative than the Schackel baseline slope. As in the Tmax relationship, the SWP variation with this latter equation was very small. The (Corell et al., 2016) baseline based on the VPD data is also presented in Figure 7. This equation is almost parallel to the Shackel estimation but approximately 0.3 MPa more negative. Figure 7: Relationship between midday stem water potential and maximum daily VPD. Each symbol is the average of data in the same group of leaf conductance reduction. The black solid line represents the equation ISO (Table 1). The blue line represents the equation of (Shackel et al 2021) (solid) and plus 0.5 MPa (dash). The gray line represents the equation of [12]. CAPÍTULO IV 136 4. Discussion The pattern of the SWP and leaf conductance showed that the former was more sensitive to water stress under field conditions than the latter. Most of the seasons, the SWP decreased under conditions of water stress and recovered several weeks earlier than leaf conductance (Figures 2 and 3). The differences between treatments were also greater in the SWP than in leaf conductance. These responses were reported in different irrigation works in young (Moriana et al., 2002; Gómez del Campo, 2010) and mature olive orchards (Girón et al., 2015; Moriana et al 2007). However, this was not the case in other orchards, where significant lower leaf conductance was found at the beginning of the season, when the SWP was very positive and not different (Aguero-Alcaras et al., 2016; Corell et al., 2020; Moriana et al., 2003). These changes in the sensitivity of leaf conductance would be related to two factors: percentage of roots in dry soil and isohydric response. Some authors (Pérez-López et al 2008) concluded that the delay in the recovery of leaf conductance from the water stress conditions was related to the volume of moist soil. Thus, partial root drying experiments showed that leaf conductance was affected by the presence of stress signals from roots, even with the same SWP (Morales-Sillero et al., 2013). But leaf stomata closure at the beginning of the season was not likely related to this. In this part of the season, when the SWP was still very positive (Figure 3) and likely vegetative growth was occurring in all treatments (Figure 5), this parameter would be linked to an isohydric response. Although traditionally plants were classified according to their isohydric and non-isohydric response (Fereres.,1984), recent works suggested that some species could combine both. Some authors (Díaz-Espejo et al., 2018) suggested that both responses were possible in olive trees. At the beginning of the season, a great amount of soil water would promote the isohydric response, with a sharp decrease in leaf conductance before a small variation of the SWP (Díaz-Espejo et al., 2018). Several authors reported this pattern in the leaf conductance vs. water potential relationship (Marino et al., 2018; Ahumada-Orellana et al., 2019; Fernández et al., 2014). In the present work, these conditions could occur in the Control but were not clear in the RDI or Rainfed according to the soil moisture profile and differences in soil cover (Figures 4 and 5). Differences in soil cover suggested a decrease in vegetative growth. These commonly occurred in olive trees before the SWP and leaf conductance affection (Pérez- López et al., 2007). The lack of response from leaf conductance to water deficit in some works, such as this one, would be likely related to data variability. According to several CAPÍTULO IV 137 authors, transition from isohydric to non-isohydric would be between −1.5 and −2.3 MPa (Marino et al., 2018; Ahumada-Orellana et al., 2019 Díaz-Espejo et al., 2018). This change in the physiological response would affect water relations in the plant and the response to the environment. Reference equations of the SWP changed when data considered the leaf conductance reduction (Figures 6 and 7; Table 1). The scattering of the data increased with the decrease in leaf conductance, but relationships were usually weak (Table 1). Best fits in the 90%– 100% and 70%–80% groups were parallel, and most of them were also parallel to the published reference equations (Table 1). All the above suggested that the same leaf conductance reduction would include a different position of the leaf conductance vs. water potential relationship, which increased the variability. Several authors reported that the SWP vs. leaf conductance relationship presented a great scattering, with values more positive than −2 MPa (Marino et al., 2018; Ahumada-Orellana et al., 2019). Only in the ISO data (Table 1, SWP more positive than −1.5 MPa and small leaf conductance reduction), the variability was small, and the reference equations presented a very slight slope, not parallel to the published baselines (Table 1). However, even in these ISO data, the SWP presented a meaningful relationship with evaporative demand (Temperature or VPD). Therefore, according to the current work, the selection of a constant value as a threshold for full irrigated conditions would not be advisable. Several authors reported a threshold of the SWP around −2 MPa or even more positive (−1.2 MPa, (Moriana et al., 2012)) as indicator of full irrigated conditions (Marino et al., 2018; Ahumada-Orellana et al., 2019). However, these works assumed that the SWP measurements would be under isohydric conditions. This response in commercial orchards is likely to occur at the beginning of the season, but it would not be a common occurrence by mid-summer or dry spring, when part of the root zone could be under dry conditions (Torres-Ruiz et al., 2013; Morales-Sillero et al., 2013). Thus, reference equations that do not consider leaf conductance, such as (Corell et al., 2016) or (Shackel et al 2021), could be more appropriate for defining baseline equations than the ISO equation of this work or constant values. Management of deficit irrigation based on the SWP baseline equations is not common. Some authors (Shackel et al 2021) suggested that baseline equations were defined as under “wet soil conditions”, and most of the time they would not be adequate for defining the target water status in a commercial orchard. Only under conditions that consider the tree growth as the main target of irrigation scheduling, the baseline could be the best CAPÍTULO V 144 2020) that could be affected by the variability of temperature in the upper or lower part of the canopy (Ramírez- Cuesta et al., 2022). Noguera et al. (2020), using a low-cost device for olive orchards, concluded that, although the system worked, further works would be needed to test different environmental and water status conditions. This mainly refers to the estimation of NWSB, which would affect to CWSI determinations (Egea et al., 2017; Ramírez-Cuesta et al., 2022). The aim of this work was to evaluate a simplified methodology to estimate NWSB in a young super-intensive density olive orchard using a low-cost device for a wide range of water statuses. This simplified approach would verify the NWSB estimations by Egea et al. (2017) and the use of different sources of climate data in the estimation of the CWSI. Two consecutive seasons with three different irrigation systems (Control, RDI and Rainfed) would allow evaluating a wide range of water statuses. The low-cost device was a commercial thermal camera attached to tablet, as in Garcia-Tejero et al. (2018), and the data will be compared with water potential measurements. 2. Materials and Methods 2.1. Sites and treatment description The experiment was carried out in an olive orchard in La Hampa, an experimental farm of the Instituto de Recursos Naturales y Agrobología (IRNAS-CSIC) in Coria del Rio, near Seville (Spain) (37º 17´N, 6º 3´W, 30 m altitude) during the 2020 and 2021 seasons. This orchard was 2 years-old at the beginning of the experiment and the first yield occurred in 2020 and was around 800 kg ha-1. The trees (Olea europaea L. cv Manzanillo) were planted in a super-intensive system, 4 x 1.5 m apart (1667 tree ha-1) and were trained in a central leader system. The soil was a sandy loam with more than 1 m depth. Climatic data were obtained from the Andalusian weather network, station “La Puebla” (SIAR, 2022), about 7 km away from the experimental plot. The climatic conditions of the location are typically Mediterranean. The potential evapotranspiration (ETo) and rainfall in both years presented the typical seasonal pattern (Figures 1 and 2). CAPÍTULO V 145 Figure 1: Climatic conditions along the experiment during 2020 season. (a) Pattern of potential evapotranspiration (ETo) and rain. Vertical lines limited the period of measurements (b) Pattern of daily radiation in the measured dates. (Source: Climatic station of “La Puebla”). CAPÍTULO V 146 Figure 2: Climatic conditions along the experiment during 2021 season. (a) Pattern of potential evapotranspiration (ETo) and rain. Vertical lines limited the period of measurements (b) Pattern of daily radiation in the measured dates. (Source: Climatic station of “La Puebla”). CAPÍTULO V 147 ETo was greater than 5 mm day-1 during most of the experimental period in both seasons and more common in 2021 than in 2020. Rainfall occurred from mid-autumn until early-spring, with a dry summer. In both seasons, the volume of rain was low, particularly during 2021 (492 mm in 2020 and 357 mm in 2021) compared to the average of the zone (539 mm, AEMET, 2022). Rains during the experimental period in both years were almost null. Most of the days when measurements were carried out, the weather was clear without radiation constrains (Figs. 1b and 2b). Only on day of the year (DOY) 274 in 2021 (Fig. 2b) there was a clear reduction of radiation at midday. The irrigation system consisted of a single line of emitters (2.2 L h-1) which were spaced 0.5 m. The water applied was measured with a water-meter in each replication and monitored weekly during the irrigation season. Three irrigation treatments were applied using a completed randomized design, with 6 repetitions/plot from 2nd of June in 2020 due to the coronavirus pandemic. The irrigation season started before that date, and all trees were irrigated with the same amount of water (56 mm until the beginning of June). Each plot had 5 lines of 13 trees each, and the 3 central lines were used as measurement points. Irrigation scheduling considered four phenological phases: a) before pit hardening; b) pit hardening (until the last week of August); c) rehydration period (until harvest) and d) post-harvest. The beginning of pit hardening was determined by the variation in longitudinal growth of the fruits (Rapoport et al., 2013). The irrigation treatments were: • Control: The objective of this treatment was to maximize vegetative growth. Water application was scheduled based on midday stem water potential (SWP) measurements. Irrigation was the traditional Control. The objective of this treatment was to maximize vegetative growth. Water application was scheduled based on midday stem water potential (SWP) measurements. Irrigation was the traditional estimation of FAO 66 (crop coefficients (Kc) 0.55, reduction coefficient (Kr) 0.7), but the water applied was increased when SWP values were lower than -1 MPa until 175% crop evapotranspiration (ETc). The seasonal water applied reached 553 and 772 mm in 2020 and 2021 respectively (Table 1). • Regulated deficit irrigation (RDI). The objective of this treatment was to achieve a mild or moderate water stress throughout the season. Trees were irrigated with 1 mm per day in all the irrigation season. This water applied was increased by 1, 2 or 3 mm according to the difference between the CAPÍTULO V 148 measured SWP and threshold values. The threshold value was as expected according to the baseline of Corell et al. (2016) for most of the season. Only from the beginning of pit hardening until the second week of August, the threshold SWP was reduced down to -2 MPa. Irrigation scheduling applied 267 and 334 mm, respectively in 2020 and 2021 (Table 1). During the post-harvest period, the threshold value was constant and equal to -1.2 MPa. • Rainfed. During the 2020 season, these trees were not irrigated from June until the first week of September, just before harvest, when they received around 45 mm. No irrigation was applied during 2021. Table 1: Water applied (mm) during the 2020 and 2021 seasons. Each data is the average of 6 plots. 2.2. Measurements The water relations of the trees were described using the SWP and midday leaf conductance. SWP was determined in one leaf per plot. Leaves near the main trunk were covered with aluminum foil two hours before the measurement, which was made using the pressure chamber technique at midday (PMS 1000, USA). The abaxial leaf conductance was measured with a porometer (SC-1, Decagon devices, USA) in a fully expanded sunlit leaf per plot. Both the SWP and leaf conductance were measured in the same tree every 7–10 days. Canopy temperature (Tc) was measured by thermal imaging per plot at the same time that the SWP and leaf conductance. These measurements were obtained using a lowcost thermal camera (Flir One, Flir System, USA), which was attached to a tablet, and the image included three trees. The SWP and leaf conductance were measured in the central one. This camera used a thermal sensor (8–14 μm, 80 × 60 pixels) and a digital sensor (1440 ×1080 pixels), and had the capability to overlap two images (a false-colored one with a digital image). The error margin was 5% of the difference between ambient and scene temperature (http://www.flir.es/flirone/ios-android/). According to Costa et al. CAPÍTULO V 149 (2013), matt emissivity was used (ε= 0.95) and each pixel corresponded to an actual temperature reading. The imaging was obtained at a distance of approximately 2 m from the trees and in the sunlit side at around 11:00 a.m. GMT. Images were analyzed with the software “Flir Tools” _by Android. Each image included 3–4 sunlit areas with only leaves, and the Tc was the average temperature of these zones. The crop water stress index (CWSI) was calculated empirically according to the expression of Idso et al. (1981): Where: 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = ∆𝑇𝑇 − ∆𝑇𝑇𝑙𝑙𝑙𝑙 ∆𝑇𝑇𝑢𝑢𝑙𝑙 − ∆𝑇𝑇𝑙𝑙𝑙𝑙 ∆𝑇𝑇 is the difference between Tc and the air temperature (Ta). ∆𝑇𝑇𝑙𝑙𝑙𝑙 is the difference between the canopy temperature and the air temperature in the lower limit (canopy transpiring at potential rate) for a given vapor pressure deficit (VPD) and was estimated with a non-water stress baseline (NWSB). ∆𝑇𝑇𝑢𝑢𝑙𝑙 is the maximum expected difference between the canopy temperature and the air temperature, which correspond to a non-transpiring canopy. This was assumed constant and equal to 5ºC (Ben-Gal et al., 2009). The CWSI was estimated using different approaches to evaluate the influence of climatic data measurements and NWSB. Regarding climatic data, two different sources of data were used for Ta and relative humidity. On one hand, these data were obtained from daily records of the weather station “La Puebla” of the Andalusian weather network (Maximum daily air temperature and Minimum daily relative humidity, SIAR, 2022). This data source (hereinafter, the “Station”) simplifies the estimation process in commercial orchards. On the other hand, data were obtained using a thermo-hygrometer (TFA 31.1028 Dewpoint Pro, TFA, Germany) (hereinafter, the “Thermohygro”), located at the orchard when the measurement was made to obtein an instant measurement of these variables. Regarding the estimation of the NWSB, two different approaches were used, both according to Egea et al. (2017). The NWBS is a linear function of VPD that changes in olive trees depending on the time when measurements were made and seasonal time point (Egea et al., 2017). The first approach considered just one time point per day, and the general equation estimated by Egea et al. (2017) for 11:00 GMT was used for the whole season (hereinafter, “Daily”). The second approach considered the time, but also the seasonal time point (hereinafter, “Seasonal”). Then, three different equations were used CAPÍTULO V 150 (i) until the end of July, (ii) during August, (iii) from September, all of them for 11:00 GMT (Egea et al., 2017). 2.3. Statiscal análisis The data were analysed by means of the one way-ANOVA and the Statistix program (SX 8.0, Analytical Sofware, USA). Data normality was tested using a Shapiro- Wilk test. Differences in ANOVA were considered significant when the p-level < 0.05 in both tests. The number of samples measured is indicated in the text and figures. A linear regression analysis was carried out to explore the relationships between the estimation of CWSI using the different approaches. The accuracy of the regressions was described with the residual mean square, standard deviation and mean absolute error of each fit. Differences between regressions and 1:1 line were determined using a T-test of the slope and y-intercept. 3. Results Different comparisons of the CWSI estimation were obtained using the equations for the NWSB and climatic data when calculating the VPD (Fig. 3). In all relationships considered, there were significant fits between the seasonal and daily approach (Fig. 3a, Table 2). These fits changed significantly according to the season or climatic data source used (Station or Thermohygro). CAPÍTULO V 151 Figure 3: Relationship between crop water stress index (CWSI) (a) estimated with seasonal vs daily adjustment. Solid line represents the 1:1 line. Dash line the best fit using the pool data. Red symbol, Station in 2021; Black symbol Thermohygro in 2021; Blue, Station in 2020; Green, Thermohygro 2020. (b) estimated with climatic station data vs thermohygro. Solid line represents the 1:1 line. Dash line the best fit using all data. Red symbol 2020 data; black symbol 2021 data. In both figures, each symbol is the average of 6 data and vertical and horizontal bars represent standard error. Table 2: Best fit of the relationship between the CWSI estimated using a non-water stress baseline calculated with one equation (Daily, D) or considering different parts of the season (Season, S). For each season, the climatic data were obtained from two sources: climatic station (Station) and field thermo-hygrometer (Thermohygro). * Notes: CWSI, crop water stress index; R2, coefficient of determination; Error, Standard deviation; N, number of data; MSE, Residual mean square error; MAE, mean absolute error. CAPÍTULO V 152 However, all of them were very similar, particularly when the same season is considered (Table 2, Fig. 3a). When data pool for both seasons were used, the total equation suggested that the differences between both approaches (Seasonal and Daily) were small (Table 2, Fig. 3a). The CWSI estimated with a Seasonal approach was around 6% greater than those values estimated with the Daily approach (Table 2). On the other hand, the CWSI varied more when using the same approach and comparing climatic data sources (Figure 3b, Table 2). For both seasons, the relationship between the CWSI estimated with different climatic data sources were close to 1:1 (Figure 3b), but th values were significantly different. The best fit of the 2020 season tended to be slightly higher than the 1:1 line (Figure 3b and Table 3). This means that, for that season, the CWSI values estimated using climatic station data were greater than those with based on Thermohygro data. In 2021, the fit was weaker than in 2020 (Table 3) and the CWSI estimated with Thermohygro data tended to be greater than those based on Station data (Figure 3b, Table 3). Table 3: Best fit of the relationship between the CWSI estimated with a daily approach using the climatic data from a nearby station (Station) or field Thermohygro (Thermo). * Notes: CWSI, crop water stress index; R2, coefficient of determination; Error, Standard deviation; N, number of data; MSE, Residual mean square error; MAE, mean absolute error. The pattern of CWSI and SWP throughout the 2020 season presented significant differences between Control and Rainfed treatments on most dates (Fig. 4). The CWSI was estimated only with the Daily approach, using climatic data from the Station (Fig. 4a) and Thermohygro (Fig. 4c). Most values of the CWSI were in the expected range (between 0 and 1) for both climatic data sources. Only on the last date of measurements, for both, and on DOY 246, only in Thermohygro, the CWSI was lower than -0.5. The pattern for both CWSI estimations were very similar, with a cycle of peaks and falls on the same dates and with the same significant differences between treatments (Figure 4a CAPÍTULO V 153 and c). On five dates, values in both CWSIs were significantly greater in Rainfed than in Control, but only on the last two it was significantly greater than in RDI. The first period of increase and decrease of the CWSI was between DOY 177 and 211. In this cycle, all treatments presented a similar pattern, although RDI and Control tended to a lower CWSI than Rainfed, and this was more significant on DOY 183, 191 and 211 (Figure 4a and c). The pattern of SWP during the same period was very different from the CWSI (Figure 4b). The Rainfed SWP decreased sharply under severe water stress conditions (around - 4 MPa), while Control and RDI were greater than -2 MPa (Fig. 4b). In this period, Rainfed presented the lowest SWP values, which were coincident with the maximum CWSI. However, differences in SWP between RDI and Rainfed were greater than in CWSI, which were almost equal. From DOY 211, Rainfed presented again a cycle of increase and decrease in CWSI, but this was not observed in Control and RDI (Figure 4 and c). This cycle was not identified in the SWP data in Rainfed, as they were almost constant at around -4 MPa until recovery on DOY 246 (Figure 4b). On this latter date, CWSI was equal to RDI and, from here on, all treatments showed similar great increases and decreases in CWSI which, again, were not measured in the SWP data.