Tetraploid Citrumelo 4475 (Citrus paradisi L. Macf. x Poncirus trifoliata L. Raf.) Rootstocks Improve Common Clementine Tolerance to Long-term Nutrient Deficiency (Citrus clementina Hort. ex Tan)

Nutrient de�ciency alters growth and the production of high-quality nutritious food. In Citrus crops, rootstock technologies have become a key tool for enhancing tolerance to abiotic stress. The use of doubled diploid rootstocks can improve adaptation to lower nutrient inputs. This study investigated leaf structure and ultrastructure and physiological and biochemical parameters of common clementine scions (C) grafted on diploid (2x) and doubled diploid (4x) Carrizo citrange (C/CC2x and C/CC4x) and Citrumelo 4475 (C/CM2x and C/CM4x) rootstocks under optimal fertigation and after seven months of nutrient de�ciency. Rootstock ploidy level had no impact on structure but induced changes in the number and/or size of cells and some cell components of common clementine leaves under optimal nutrition. Rootstock ploidy level did not modify gas exchanges in Carrizo citrange but induced a reduction in the leaf net photosynthetic rate in Citrumelo 4475. By assessing foliar damage, changes in photosynthetic processes and malondialdehyde accumulation, we found that C/CM4x were less affected by nutrient de�ciency than the other scion/rootstock combinations. Their greater tolerance to nutrient de�ciency was probably due to the better performance of the enzyme-based antioxidant system. Nutrient de�ciency had similar impacts on C/CC2x and C/CC4x. Tolerance to nutrient de�ciency can therefore be improved by rootstock polyploidy but remains dependent on the rootstock genotype.


Introduction
Fruit crops, especially citrus fruits, require large amounts of fertilizers to ensure good production and fruit quality. Today, the challenge for sustainable agriculture, and particularly organic agriculture, is to reduce the use of inputs in crops. Reducing inputs optimizes the economic outcome while limiting the environmental impact.
Water and minerals absorbed by roots in the soil are essential for plant development, growth and reproduction. Fourteen minerals are considered essential and these can be divided into two groups: the macroelements (N, K, P, Ca, Mg and S) which are constituents of organic matter (proteins, nucleic acids) or play a strong osmotic role, and microelements (Zn, Cu, Fe, Mn, B, Mo, Cl and Ni), which are only involved as speci c cofactors or constituents of certain enzymes 1-3 . Mineral-de cient plants present various visual symptoms, such as necrosis, chlorosis, dark green foliage, or stunted growth 4 . Essential mineral de ciency alters plant primary metabolism and this disrupts the physiological and biochemical processes leading to changes in leaf structure and ultrastructure 5,6 . Moreover, cells must cope with an overproduction of reactive oxygen species (ROS) such as singlet oxygen (O * ), hydroxyl radicals (OH • ), superoxide anion (O 2 •− ) and hydrogen peroxide (H 2 O 2 ) which cause membrane leakage due to lipid peroxidation and damage to proteins and nucleic acids [7][8][9] . As a result, ROS defence mechanisms are activated by a set of antioxidant compounds (metabolites such as ascorbate and proline) and antioxidant enzymes (superoxide dismutase (SOD), catalase (CAT), ascorbate peroxidase (APX) and dehydroascorbate reductase (DHAR)). Adapted genotypes are therefore sought in order to propose cultural itineraries more suitable to low input conditions.

Page 3/16
In citrus crops, improving the performance of varieties is based on the scion/rootstock combination forming the aerial parts and roots of the plant, respectively. The impact of the rootstock on the scion lies in its in uence on owering, fruit quality, canopy size, and resistance, among other parameters 10 . Good relations between rootstock and scion are essential for the e cient translocation of water and mineral nutrients and to promote biomass production and tolerance to biotic and abiotic factors, such as nutrient de ciency 11,12 . Improved root system vigour in citrus rootstocks results in increased soil nutrient and water uptake 13,14 18 . The use of rootstocks better adapted to environmental constraints seems to be a promising eco-friendly strategy.
Many Citrus genotypes are used as rootstock for citrus cultivation. Genotypes belong either to the Citrus genus such as Volkamer lemon or are obtained by hybridization between Citrus and Poncirus genus progenitors such as Citrumelo 4475 and Carrizo citrange. Volkamer lemon which is used as rootstock for lemon, is adapted to dry, calcareous and saline soils and presents tolerance to Tristeza, cachexia and exocortis. Citrumelo 4475 imparts cold tolerance to the scion. Citrange Carrizo is frequently used in acidic and neutral soils but not in dry areas because of its limited performance under drought conditions. These two genotypes inherited Tristeza tolerance from their Trifoliate orange progenitor and give clementine varieties that produce a high yield and fruit quality 19 .
In a previous study, we compared leaf structure and ultrastructure under nutrient de ciency of two genotypes used worldwide as rootstock for citrus cultivation, the Citrumelo 4475 (Citrus paradisi L. Macf. × Poncirus trifoliata L. Raf.) and Volkamer lemon (Citrus limonia Osb.) with both 2x and 4x genotypes 20 .
Results showed an increase in tolerance to nutrient de ciency in 4x genotypes. Doubled diploid genotypes presented less degradation of ultrastructural components such as chloroplasts, thylakoids, mitochondria and starch grains associated with a smaller decrease in leaf net photosynthetic rate (P net ), stomatal conductance (g s ) and chlorophyll uorescence (F v /F m ) compared to 2x genotypes.
The higher tolerance to nutrient de ciency was linked to the reduced accumulation of malondialdehyde (MDA) and H 2 O 2 in Citrumelo 4475 4x than in the 2x, implying a more e cient antioxidant system in the 4x genotype. However, few differences in the antioxidant system and oxidative status were observed between 2x and 4x Volkamer lemons.
The aim of this study was therefore to determine the impact of rootstock ploidy level on the 2x common clementine scion during nutrient de ciency. We selected common clementine grafted on two rootstocks used worldwide for clementine cultivation, i.e. Citrumelo 4475 (C/CM2x) and Carrizo citrange (C/CC2x) in both 2x and 4x types (C/CM4x and C/CC4x, respectively). The effect of rootstock ploidy level on common clementine scions under nutrient de ciency was investigated by studying common clementine leaf structure and ultrastructure and physiological and biochemical parameters.

Results And Discussion
Effect of rootstock ploidy level on anatomical properties of common clementines under control conditions Whatever the rootstock genotype and ploidy level, microscopic examination of leaf surface imprints con rmed the presence of stomata only on their abaxial surface that were surrounded by ordinary epidermal cells 29 (anomocytic organization) (Fig. 1). Rootstock tetraploidy did not induce any changes in the location of stomata or the epidermal cell structure of common clementine leaves. This agrees with this anomocytic organization observed on leaves of tetraploid seedlings 20 .
Stomata size was unchanged in C/CM4x and C/CC4x combinations. Conversely, stomata density decreased and ostiole sizes increased in C/CC4x compared to C/CC2x combinations (Table 1). Studies have shown a positive correlation between stomata and ostiole size and a negative correlation between stomata density and ploidy level in both 4x ungrafted genotypes and 3x clementine in comparison to their 2x counterparts 20,29,30 . Our study showed that 4x rootstocks had no effect on stomata sizes in 2x common clementine scions and a non-systematic impact on ostiole size and stomatal density, putatively through changes in hydraulics or mineral inputs. The decrease in stomatal density caused by rootstock tetraploidy in the C/CC4x combination was not associated with any changes in gas exchanges compared to its C/CC2x counterpart (Table 1; Fig. 1 and 5). The increase in ostiole size in the C/CC4x combination compared to its C/CC2x counterpart suggests an adjustment of the stomata opening to compensate for the reduced stomatal density and maintain stomatal conductance 31 . This adjustment could be due to the considerable degradation of starch grains in the guard cells in the rst hour of light contributing to a rapid increase in the opening of the stomata in parallel with the activation of membrane ion transport 32,33 .
Despite identical physical stomatal attributes between C/CM2x and C/CM4x combinations, analysis of the gas exchange parameters revealed a decrease of P net in the C/CM4x combination (Table 1; Fig. 1 and 5). Other factors than leaf structure therefore appear to be involved in the regulation of photosynthesis in the C/CM4x combination.
At ultrastructural level, rootstock tetraploidy initiated changes in the number and size of the main cell components of the 2x common clementine leaves (Tables 2 and 3). Cell size and thickness in both palisade and spongy mesophylls were reduced in C/CC4x compared to C/CC2x, whereas they increased in C/CM4x compared to C/CM2x (Tables 2 and 3; Fig. 2 and 3). These ultrastructural modi cations were the only identical changes in C/CM4x and the ungrafted rootstock counterpart 20 . In contrast to its ungrafted counterpart, C/CM4x showed a lower P net and similar chloroplast numbers to C/CM2x 20 (Tables 2 and 3; Fig. 2, 3 and 5A). Chloroplast enlargement in palisade mesophyll brought about by rootstock tetraploidy in common clementine leaves was not associated with an increase in chloroplast numbers. This phenomenon appears to reduce photosynthetic capacity 34 . Overall, under optimal conditions and depending on the rootstock genotypes, the structural and/or ultrastructural modi cations of common clementine leaves induced by rootstock tetraploidy could either compensate each other with no change in gas exchanges or induce a modi cation of gas exchanges. The number and size of transitory starch grains and mitochondria were either reduced or similar between 2x and 4x scion/rootstock combinations in palisade and spongy mesophylls (Tables 2 and 3; Fig. 2 and 3). The number of plastoglobuli increased in C/CM4x and C/CC4x compared to C/CM2x and C/CC2x in the chloroplasts of the palisade and spongy mesophylls. Only the C/CM4x plastoglobuli were larger than those of its 2x counterpart (Tables 2 and 3; Fig. 2 and 3). Thus, rootstock tetraploidy involved a potential increase in lipid (plastoquinone-9 (PQ-9), plastoquinol-9 (PQ-9H2) and a-tocopherol (a-T)) reserves that could not be deposited in the thylakoids of chloroplasts.
Rootstock tetraploidy can induced modi cations in the number and/or size of the investigated ultrastructural components.
Differences in photosynthetic properties and redox status of scion/rootstock combinations under nutrient stress could be related to their leaf and cell anatomy Complete starvation resulted in a signi cant decrease in N but similar or higher levels of P, K, Mg, Ca and Na than controls in all scion/rootstock combinations (Table 4). A concentration effect induced by the transfer of N to other tree areas could explain the increase in P, K and Mg contents in all scion/rootstock combinations 35 . Leaf chlorosis occurs when plants do not have the nutrients needed for chlorophyll synthesis, which in turn affects the photosynthetic e ciency 36 . However, the decrease in N, which was an important factor in chlorosis, was similar in all scion/rootstock combinations (Table 4). C/CM4x showed less chlorosis as indicated by its light green colour compared with the yellow colour of other scion/rootstock combinations (Fig. 4). These results suggest a better integrity of chlorophyll content probably related to a better protection against ROS [37][38][39] . The improvements in the redox status of scions by tetraploid rootstocks is probably linked to an improvement in photosynthesis.
Chlorosis was associated with structural and ultrastructural foliar changes and a disruption of photosynthetic properties in all scion/rootstock combinations 6,40 (Tables 1, 2 and 3; Fig. 5). At structural level, stomata and ostioles of all scion/rootstock combinations showed a narrowing which was associated with an increase in density (except in C/CC2x) response to nutrient de ciency (Table 1; Fig. 1).
The rootstock apparently has a different effect on the structural components of common clementine leaves depending on genotype and/or ploidy level. P net and g s decreased concomitantly in all scion/rootstock combinations under nutrient de ciency (Fig. 5A-B). These results suggest a critical role of stomatal structure in the process of CO 2 availability 41 . However, the smaller decrease in P net , g s and F v /F m in C/CM4x than in other scion/rootstock combinations (Fig. 5) implies that other factors than stomata are needed to sustain photosynthesis [42][43][44] .
At ultrastructural level, nutrient de ciency resulted in cell size enlargement (except in C/CM2x spongy mesophyll), thylakoid with grana degradation and a decrease in chloroplast size in leaf mesophylls of all scion/rootstock combinations (Tables 2 and 3; Fig. 2 (Tables 2 and 3; Fig. 2 and 3).
The overproduction of reactive oxygen species (ROS) leads to alterations in the integrity of the membrane structures of cells, chloroplasts or thylakoids by a process called lipid peroxidation, one of the indicators of which is malondialdehyde (MDA). The lower MDA content in C/CM4x compared to other scion/rootstock combinations was consistent with the reduced damage to ultrastructure (Tables 2 and 3; Fig. 2, 3, and 6D). Thylakoid size and structure is dependent on the formation of PSII-LHCII supercomplexes 49 . In C/CM4x, the lower decrease in F v /F m suggests a more limited degradation of the PSII-LHCII supercomplexes reducing the disruption of electron transport for the photosynthetic reaction and thus the production of ROS 50 (Fig. 5C and 6D). Maintaining the redox balance in C/CM4x decelerates the degradation of thylakoids resulting in a more e cient photosynthetic capacity (P net and F v /F m ) compared to C/CM2x, C/CC2x and C/CC4x after nutrient de ciency (Tables 2 and 3; Fig. 2, 3, 5A, 5C and 6D). Plastoglobuli are bound to thylakoids via their membranes. Studies have shown that plastoglobuli are involved in the formation and degradation of thylakoids during plant growth, development and senescence but also when plants are exposed to drought, high-light, N starvation or fungal infection 46,[51][52][53] . The reduced damage of the thylakoid membrane in C/CM4x would explain the maintenance of the plastoglobuli numbers and the small increase in their size due to the accumulation of catabolites formed by thylakoid degradation in their hydrophobic core compared to other scion/rootstock combinations 54,55 (Tables 2 and 3; Fig. 2 and 3).
Rootstock tetraploidy improves the tolerance of 2x common clementine scions by maintaining the redox status and delaying ultrastructural changes and damage with a consequent improvement in photosynthetic capacity. However, as suggested by the similar impact of nutrient de ciency on C/CC2x and C/CC4x, rootstock tetraploidy does not automatically result in better tolerance of the photosynthetic properties of the scion.
Differences in tolerance to nutrient stress between scion/rootstock combinations may be explained by a better antioxidant system Nutrient deprivation led to a modi cation of oxidative metabolism 35,39,56 . Tolerance differences between scion/rootstock combinations may be related to differences in ROS production and/or antioxidant system e ciency. The enzymatic antioxidant system response differs according to plant species and the de cient minerals [56][57][58][59][60] . Enzymatic antioxidant systems in common clementine leaves differed depending on the rootstock genotypes (Fig. 6).
Nutrient de ciency impaired SOD activity in C/CC4x and C/CM4x (Fig. 6A). It was interesting to note that APX activity either increased or remained similar to the control in all scion/rootstock combinations (Fig.  6C) whereas CAT activity only increased in C/CM4x under nutrient de ciency (Fig. 6B) Conversely, in C/CM4x, the synergistic activity of CAT and APX maintains the MDA content in C/CM4x (Fig. 6B, C and D).As in their ungrafted counterparts, the increased enzymatic activity may explain the reduced ultrastructural damage and decrease in photosynthetic activity in C/CM4x in comparison with the other scion/rootstock combinations. The improvement in performance of the enzyme-based antioxidant system in the scion is due to the tetraploidization of the rootstock and the compatibility between rootstock and scion.
In conclusion, rootstock ploidy level had no effect on the structure of the common clementine scion leaves (except stoma density in Carrizo citrange) whereas it induced modi cations in the ultrastructural components. The impact of prolonged nutrient de ciency on the structure, ultrastructure, physiology and biochemistry of the common clementine scion differed according to the variety and ploidy level of the rootstock. Among the four scion/rootstock combinations, common clementine grafted with 4x citrumelo 4475 rootstock (C/CM4x) was the most tolerant to nutrient de ciency as indicated by the limited changes in leaf cell structures and photosynthetic activity. The improved tolerance of common clementine grafted with 4x citrumelo 4475 rootstock may be related to a better antioxidant system. Tolerance to nutrient de ciency can therefore be improved by rootstock polyploidy but remains dependent on the rootstock genotype. The next step of this study will be to test the impact of rootstock ploidy level on the quality and yield of clementine fruit under nutrient de ciency.

Plant material and experimental design
The experiment was carried out on the AREFLEC experimental station located in San Giuliano, Corsica

Scanning electron microscopy (SEM)
Scanning electron microscopy measurements were carried out on three leaf pieces per scion/rootstock combination and fertigation level (typically 1 cm²) (n = 3) cut with a razor blade from mid-laminar areas at between 10:00 and 11:00 am. As described in Oustric et al. 20  Specimens were mounted on aluminum stubs with carbon double-sided adhesive disks, coated with gold/palladium in a SC7640 sputter coater (Quorum Technologies Ltd, Newhaven, U.K.) and examined under a S-3400N scanning electron microscope (Hitachi High-Technologies Corporation, Tokyo, Japan) at an accelerating voltage of 5 kV.

Transmission electron microscopy (TEM)
Transmission electron microscopy measurements were carried out on ve leaf pieces per scion/rootstock combination and fertigation level (typically 1 mm²) (n = 5) cut with a razor blade from mid-laminar areas at between 10:00 and 11:00 am. As described in Oustric et al. 20 , leaves were immediately xed in cold (4°C ) 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer at pH 7.2, rinsed in a 0.1 M cacodylate buffer at pH 7.2, post-xed in cold (4 °C) 1% osmium tetroxide in the same buffer for 1 h, dehydrated through a graded ethanol series (70% and 100%) and propylene oxide, embedded in Spurr, and polymerized at 60 °C for 24 h. Ultra-thin sections (60-90 nm) were cut using a Power tome PC ultramicrotome (RMC Boeckeler, Tuscon, U.S.A.). Sections were placed on 200-and 300-mesh copper grids and stained with UranyLess (Delta Micoscopies, France) and lead citrate. They were then examined using a Hitachi H-7650 (Hitachi High-Technologies Corporation, Tokyo, Japan) at an accelerating voltage of 80 kV.
Mineral content was measured on a pool of eight fully expanded leaves for the three plants per combination and fertigation level (n = 3) between 10:00 and 11:00 am. Fresh leaves were placed in a forced air oven at 65 ± 10° C overnight and then transferred into a desiccator for cooling. The dehydrated leaves were then sent to the CIRAD "Analyses des eaux, sols et végétaux service unit" at Montpellier (France) for analysis of macro-and micro-nutrients.
Leaf P, K, Ca, Mg and Na contents were measured using an Agilent 720 simultaneous ICP-OES after double calcination with silica removal by adding hydro uoric acid.
The leaf total N content was evaluated after combustion using a Leco TruMac N determinator.

Measurements of gas exchange and chlorophyll a uorescence
All measurements were made on three fully developed leaves for each of the three plants per combination and fertigation level (n = 9). A portable photosynthesis system (LI600) was used to measure the leaf net photosynthetic rate (P net ), stomatal conductance (g s ) and transpiration rate (E) at between 7:00 and 11:00 am. The carbon dioxide concentration (CO 2 ), air ow rate, light intensity and temperature of the leaf chamber were maintained at 380 μmol.mol -1 , 500 μmol.s -1 , 1400 μmol.m -2 .s -1 and 25 °C, respectively.
Chlorophyll a uorescence parameters were measured using an OS1p (Hansatech, Instruments Ltd) at between 9:00 and 11:00 am. Leaves were dark-acclimated for 30 min using special leaf clips. Chlorophyll a uorescence was recorded after illumination with red actinic light (650nm, 3000 μmol photon.m -2 .s -1 ) for 1 s and this was used to calculate the maximum uorescence [F v /F m = (F m -F o )/F m ] 23 . Leaves were exposed to an actinic light to evaluate the current uorescence yield (F s ) and the actual light-adapted uorescence (F m ′). Formulas were applied to this data in order to determine the effective quantum yield of PSII Y(II) = (F m '−F s )/F m '], the Y(NO) = F s /F m ], the non-photochemical quenching coe cient [Y(NPQ) = (F s /F m ')-Y(NO)] 24,25 , and the electron transport rate through PSII [ETR(II) = Y(II) x PAR x 0.5 x 0.84] 26 . The ETR/P net ratio was calculated to estimate the use of electrons in other processes unrelated to the photosynthetic CO 2 assimilation rate.

Determination of oxidative stress and antioxidant levels
Biochemical analyses were performed on three samples for each scion/rootstock combination, i.e. one per tree, obtained by pooling eight fully-expanded leaves (n = 3) collected between 10:00 and 11:00 am and immediately immersed in liquid nitrogen and stored at -80 °C. Immediately prior to biochemical analysis, each leaf and root sample was ground to a ne powder in liquid nitrogen.
Proline content was measured as described by Oustric et al. 28 A V-630 spectrophotometer was used for all measurements (Jasco Inc., Tokyo, Japan).

Statistical analyses
All statistical measurements were performed with R statistical software (v.2.12.1) (http://www.Rproject.org) and the Rcmdr package. The qualitative factors studied were sampling date (D0 and D210 after nutrient de ciency), the comment clementine scion grafted onto rootstocks subjected to nutrient stress (C/CC and C/CM) and the ploidy level of nutrient stressed rootstocks (C/CC2x, C/CC4x, C/CM2x and C/CM4x). The in uence of these three factors was analyzed using a two-way ANOVA followed by LSD test at p < 0.05.
The microscopic data obtained on leaves of common clementine scion grafted onto the various rootstocks at D0 (control) and D210 of nutrient de ciency were analyzed by heatmaps generated by Heatmap.2 function of the gplot package 3.0.1 for Rstudio (v.1.3.1093) (https://rstudio.com).