Boron Uptake by Navel Orange Seedlings as Influenced by Irrigation Water, Rootstock and Soil Texture

Abstract

Boron (B) is an essential but narrow-range micronutrient for citrus, with toxicity risks heightened in dry regions due to potentially high-B irrigation water and limited soil leaching. ‘Forner-Alcaide 5’ (FA5) is a promising rootstock for enhancing B-tolerance of sweet orange, but it had not been sufficiently tested before this study, specifically considering soil texture. Therefore, this greenhouse study investigated the effects on B absorption and biomass buildup of irrigating navel orange seedlings (cv. Navelina) grafted onto ‘Carrizo’ citrange (CC) and FA5 rootstocks, with 0.11, 2, or 5 mg B L⁻¹ waters and grown in clay loam or sandy loam soils. The results of this complete three-factor trial revealed that leaves are the primary sink for B (24–1300 mg kg⁻¹), indicating passive, transpiration-driven uptake and limited phloem redistribution. The presumed absence of sugar alcohols, and the weak binding affinity of B to the abundant sucrose, may account for the restricted phloem mobility of B in citrus, consistent with the mechanistic interpretation proposed in this study. FA5 rootstock showed greater B tolerance, sustaining 28% higher biomass than CC at 2 mg L⁻¹ B. Plant B uptake was found to be more related to soil soluble B than adsorbed B. Interestingly, the relationship followed a diminishing-returns pattern, thereby suggesting a balancing feedback mechanism, potentially based on B-induced stomatal closure. This analytical link between irrigation B and plant accumulation offers a framework for managing B toxicity, pending field validation.

1. Introduction

Boron (B) is considered an essential micronutrient for crop development, but the margin between its nutritional and toxic concentrations is narrow, making its agricultural management challenging [1,2,3,4]. Crop B tolerance is assessed based on the concentration of this element in the irrigation water, with tolerance thresholds that range between 0.05 mg L⁻¹ for very sensitive crops and 15 mg L⁻¹ for very tolerant ones [5]. In particular, citrus is very sensitive, with a tolerance threshold ranging between 0.50 and 0.75 mg L⁻¹ [6,7].

Interestingly, the thresholds used to define crop B tolerance were established in trials performed in sandy soils [7] to minimize interference from soil properties. However, plant B availability strongly depends on soil characteristics. Therefore, the same irrigation water B concentration can have markedly different effects on crops depending on soil properties [4,8,9].

Soil pH, texture, and soil organic matter (SOM) are key factors controlling the distribution of B in soils between soluble and adsorbed forms, with the soluble fraction representing the pool available for plant uptake [10]. Soluble B consists primarily of boric acid (B(OH)₃) and borate (B(OH)₄⁻), which exist in a pH-dependent equilibrium that favors B(OH)₃ below pH 9.2 [11]. Because B(OH)₃ is a small, uncharged molecule, biological membranes are highly permeable to it, allowing passive diffusion into root cells, subsequent loading into the xylem, upward transport to the shoot [12], and concentration in the leaves because phloem redistribution is limited, particularly in citrus [13]. In contrast, soil adsorbed B is bound to clay-sized minerals and SOM, mainly through inner-sphere complexes with hydroxyl groups [14]. Like the soluble forms, adsorbed B is in dynamic, pH-dependent equilibrium with soluble B [15,16].

B essentiality is traditionally considered multifaceted because it influences diverse plant physiological processes, in which its unique capacity to bond to compounds with cis-hydroxyl groups, i.e., cis-diols, makes it irreplaceable [17,18]. Primarily, B plays a key role in cell wall maintenance through the formation of ester B bridges between polysaccharides within the cell wall and between the cell wall and the plasma membrane [19,20,21]. However, precisely because of its strong affinity for cis-diols, B also readily binds to ribose, both as a free sugar and as part of RNA, ATP, and NAD(P)H, thereby interfering with the functions of these important metabolites [22].

The disruption of plant metabolism caused by excess B leads to elevated production of reactive oxygen species (ROS), particularly in organs where B accumulates the most, such as the leaves, thereby impairing the photosynthetic function [23,24,25]. Ultimately, B excess manifests as leaf injuries, growth inhibition, and yield reduction [26,27,28,29,30]. Crop-damaging levels of B are most often observed in drylands, where soil B is usually high, because, if added through the irrigation water, the limited drainage prevents its leaching [31,32]. Precisely, in these water-scarce areas, non-conventional irrigation with desalinated seawater (DSW) and reclaimed water (RW), which are usually high in B (>1 mg L⁻¹), is increasingly introduced [33,34]. Therefore, the safe use of DSW and RW requires an understanding of how crop B uptake depends on irrigation water B and soil characteristics, specifically in citrus, due to their low B tolerance.

Beyond irrigation water and soil characteristics, the rootstock is central to crop B uptake, since it provides the entry point for B into the plant and mediates its transport to the scion. This key role has been made particularly evident for grafted citrus in which successful adaptation to excess B has been linked to, specifically, root and xylem morphology and reciprocal signaling between rootstock and scion [35].

Among citrus rootstocks ‘Citrange’ Carrizo (CC) has recently become a benchmark for B tolerance testing [23,25,36,37,38,39]. This hybrid of Citrus sinensis (L.) Osbeck and Poncirus trifoliata originated from a mutant seed of ‘Troyer’ citrange in ‘Carrizo’ Springs, Texas (USA). CC is widely used as a rootstock in major citrus-growing regions worldwide because of its vigor and broad cultivar compatibility [40]. However, CC has also notable limitations, including sensitivity to soil salinity, lime and B [36,38]. As a result, research efforts have focused on developing alternative rootstocks for citrus cultivation in regions where such unfavorable soil conditions prevail.

One promising alternative to CC is ‘Forner-Alcaide 5’ (FA5), which is a hybrid of Citrus reshni (‘Cleopatra’ mandarin) and Poncirus trifoliata. FA5 is valued for its tolerance to soil salinity, lime and waterlogging, as well as resistance to citrus tristeza virus (CTV), nematodes, and Phytophthora spp. [41]. In terms of B tolerance, ungrafted FA5 has shown strong root resilience and reduced B accumulation in all plant organs compared with CC, though its leaves exhibit equal or greater physiological vulnerability [23]. These findings underscore the need to advance to the next stage of germplasm testing, in which FA5 should be evaluated as a grafted rootstock for a representative citrus cultivar. Sweet orange is a particularly suitable candidate, as it represents nearly 50% of global citrus production [42]. To our knowledge, such an evaluation of FA5, especially under contrasting soil conditions, has not yet been carried out.

Therefore, this study aims to evaluate the B uptake capacity of navel orange grafted onto FA5 in comparison to the CC benchmark, across soils with contrasting textures and to develop, consequently, a quantitative model that relates B uptake by citrus seedlings to soil B.

2. Materials and Methods

2.1. Experimental Layout

A total of 108 Citrus sinensis (cv. Navelina) seedlings grafted onto two rootstocks: CC (n = 54) and FA5 (n = 54) were cultivated in pots (14.5 cm in diameter × 17.5 cm in height) filled with two soils of contrasting textures and irrigated with water containing three different B concentrations. A complete factorial design of 12 treatments (2 rootstocks × 2 soil textures × 3 B concentrations) was set up, with nine seedlings, i.e., replicates, randomly assigned per treatment. The seedlings were randomly arranged on a bench in a temperature-controlled (18–26 °C) greenhouse and the trial lasted 336 days, from planting on 18 October 2022 to 19 September 2023.

2.2. Plant Material

The rootstock seedlings grafted with Navel orange were provided by the plant nursery Viveros Beniplant S.L. (Peñíscola, Castellón, Spain).

2.3. Soil and Irrigation

A silty clay loam soil, very low in organic matter and strongly calcareous, was obtained from the peat and soil supplier Abonos Conde (Paterna, Valencia, Spain). The soil was sieved to 2 mm and divided into two portions. One portion was mixed with 50% silica sand (0.6–0.8 mm), while the other was left unmixed. This procedure yielded two soil types that could be commonly found under real-world conditions: a clay loam and a sandy loam (

Table 1. Main characteristics of the soils in the trial: SOM, soil organic matter; CCE, calcium carbonate equivalent; B sol, soluble boron; B ads, adsorbed boron.

Soil	Sand	Clay	SOM	CCE	pH	B sol	B ads
	%				−	mg L−1	mg kg−1
Clay loam	20	30	0.3	28	7.9	0.12	0.02
Sandy loam	57	16	0.4	17	8.0	0.08	0.01

). Both presented comparable pH and low SOM, which was intended to avoid their interference in this study (Table 1). Each pot was filled with 3.6 kg of soil, achieving an average bulk density of 1.24 g cm⁻³, representative of non-compacted agricultural soils.

Irrigation was supplied via a drip system with one 2 L h⁻¹ emitter per pot, using tap water of almost constant composition (

Table 2. Average characteristics of the tap water that was used to make the irrigation water: Alk, alkalinity.

Na+	Ca2+	Mg2+	Cl−	NO3−	SO42−	Alk.	EC25	pH
mg L−1						meq L−1	µS cm−1	−
17	25	9	26	40	31	52	277	7.25

), in which analytical grade boric acid was dissolved to obtain three B concentrations (0.11, 2, and 5 mg L⁻¹). A total of 37.8 L per pot was applied to keep the soil water content near field capacity, thereby avoiding plant water stress and minimizing drainage.

2.4. Samplings and Laboratory Analyses

Over the 336-day experiment (18 October 2022–19 September 2023), three seedlings, together with their corresponding soils, were randomly and independently selected from each treatment at three dates (27 March, 22 June, and 19 September 2023) and taken to the laboratory for analysis.

Soils were left to dry at room conditions. Then, the air-dried soils were disaggregated to pass a 2 mm mesh sieve. The soil hygroscopic water content was determined by weight difference before and after drying a subsample at 105 °C for 24 h in a DIGIHEAT laboratory oven (J.P. Selecta, Barcelona, Spain). Next, two operationally defined soil B fractions were obtained: soluble B and adsorbed B.

The soluble B was extracted by equilibrating 10 g air-dried soil subsamples with 25 mL of CaCl₂ 0.025 M for 2 h in a reciprocal shaker, the suspensions centrifuged, and the supernatants decanted, filtered and saved at 4 °C. The adsorbed B was extracted by resuspending the sedimented soil materials and equilibrating with 15 mL of CaCl₂ 0.025 M for 30 min on a reciprocal shaker. This procedure was repeated five times to ensure the extraction of all readily desorbable adsorbed B, thereby simulating the effects of intensive leaching with water of low–moderate salinity. The five filtered supernatants were joined together and saved at 4 °C. The B concentration in both extracts was determined by inductively coupled plasma atomic emission spectrophotometry (ICP-AES) following the ISO 11855:1996 water analysis standard [43] using an iCAP 6500 Duo spectrometer (Thermo Fisher Scientific Inc., Waltham, MA, USA).

Collected plants were split into three different organs, i.e., leaves, stems, and roots. Then, every plant material was weighed, briefly rinsed with tap water, followed by type 1 water to remove all external dirt, but without extracting internal B. Then, the plant material was dried in an air-forced Dry Big laboratory oven (J.P. Selecta, Barcelona, Spain) at 65 °C for 5 days. The oven-dried plant samples were weighed to determine the water content they had when fresh, and subsamples were taken and ground to pass a 200 µm mesh sieve. The B concentration in the plant samples was determined by ICP-AES in concentrated HNO₃ and 33% H₂O₂ microwave digestion extracts [44], following the same standard and using the same spectrometer as stated for soil extracts.

2.5. Data Analyses

Before the statistical analyses, the normality of empirical distributions for B concentrations in plant organs (leaves, roots, and stems), soil fractions (soluble and adsorbed), and plant biomass, was assessed by calculating skewness and kurtosis coefficients. If these coefficients were outside the ±1 bounds, the distributions were considered non-normal [45]. Remarkably, non-normality in dependent variable distributions undermines the assumption of homogeneity of variances required to obtain reliable analysis of variance (ANOVA) results [46]. Consequently, Box–Cox transformations, such as logarithmic, square root, etc., of the original variables were tried, and the skewness and kurtosis of the transformed data distributions were evaluated again to check if the transformations had worked. The SPSS statistical package v19 (Chicago, IL, USA) was used to perform all the data analyses.

Four or three-way ANOVA was performed to determine how the B concentration in the different soil–plant system compartments, namely, soil solution and solids, plant leaves, stems, and roots, depended on the experimental factors: the B concentration factor at three levels (0.11, 2 and 5 mg L⁻¹), the soil texture factor at two levels (sandy loam and clay loam), and the rootstock factor at two levels (CC and FA5). The sampling date at three levels (27 March, 22 June, and 19 September 2023) was treated as another fixed effects factor. Interactions were tested until the second-order and if the experimental factors presented a significant effect (p < 0.05) on any of the B concentrations, the treatment means were separated using Duncan’s multiple range test because of its power to detect significant effects while controlling for Type I errors [47].

Following the rationale outlined in the discussion section below, Mitscherlich functions of the form given by Equation (1) were fitted to relate plant B concentration (y) to soil B concentration (x), with a, b and x₀ as fitting parameters, with the aim of developing a model for plant B uptake.

$$\mathrm{y}=a (1-e^{-b \left(x+x_0\right)})$$

(1)

Model parameters (a, b and x₀) were estimated by minimizing the sum of squared differences between predicted and observed B concentrations in plant organs, using the generalized reduced gradient (GRG) nonlinear optimization algorithm [48]. Goodness of fit was evaluated using the coefficient of determination (R²). Confidence intervals were estimated using bootstrap resampling.

3. Results

3.1. Boron Concentration in Plants and Soils

Foliar B concentration ranged from marginal (25 mg kg⁻¹) to toxic (up to 1300 mg kg⁻¹), surpassing on average (270 mg kg⁻¹) the established toxicity threshold for citrus (260 mg kg⁻¹ [49]. Next, root B followed with concentrations within a lower range (7–690 mg kg⁻¹) and finally, stem B with between 12 and 55 mg kg⁻¹ (

Table 3. Descriptive statistics for both measured and Box–Cox-transformed boron concentrations in the different compartments of the soil–plant system.

	Variable	Unit	N	Mean	Std. Dev.	Kurtosis	Skewness	Min.	Max.
Plant	Leaf	(mg kg−1)	123	268.72	307.77	2.13	1.68	24.96	1312.86
	Root		130	66.78	66.25	60.30	6.55	6.97	688.85
	Stem		48	26.09	9.87	0.52	0.89	12.39	54.91
	Log Leaf	Log (mg kg−1)	123	2.14	0.49	−1.22	0.21	1.40	3.08
	Log Root		130	1.71	0.31	0.13	0.06	0.84	2.84
	Log Stem		48	1.39	0.16	−0.62	0.11	1.09	1.74
Soil	Solution	(mg L−1)	129	1.74	2.34	1.80	1.59	0.01	10.15
	Adsorbed	(mg kg−1)	129	4.67	3.67	0.75	0.96	0.08	17.16
	SQRT solution	Sqrt (mg L−1)	129	1.01	0.85	−0.58	0.75	0.10	3.19
	SQRT adsorbed	Sqrt (mg kg−1)	129	1.95	0.90	−0.56	−0.05	0.27	4.14
	Log Solution	Log (mg L−1)	129	−0.40	0.92	−1.09	−0.29	−2.00	1.01
	Log Adsorbed	Log (mg kg−1)	129	0.46	0.55	1.41	−1.34	−1.12	1.23

). Accordingly, leaves appear to be the primary plant site for B accumulation, presenting significantly more B concentration than roots or stems (Table 3). Interestingly, root and foliar B exhibited highly positively skewed and leptokurtic distributions, whereas stem B more closely approximated normality (Table 3). For root and foliar B, normality was achieved after logarithmic transformation. Although not needed for stem B, this was also subjected to logarithmic transformation to ensure consistency in the ensuing data analyses.

Soil B concentrations covered an even wider range of order-of-magnitude than plant B concentrations, with adsorbed B (0.08–17 mg kg⁻¹) consistently exceeding soluble B (0.01–10 mg L⁻¹), indicating dominant adsorption (Table 3). Soluble B showed a moderately positively skewed and leptokurtic distribution, whereas adsorbed B fulfilled the criteria for normality (Table 3). For both, the logarithmic transformation did not sufficiently improve normality. However, the square root did, as the skewness and kurtosis values closer to zero show (Table 3).

3.2. Plant Biomass

Total plant dry biomass at trial end ranged between 60 and 225 g with a mean of 120 g (

Table 4. Descriptive statistics for both the measured and Box–Cox transformed dry biomass of each plant organ at trial end.

Variable	N	Unit	Mean	SD	Kurtosis	Skewness	Minimum	Maximum
Leaf biomass	45		32.45	9.05	−0.25	0.445	15.7	53
Roots biomass	48		44.22	18.75	3.13	1.565	18.8	113.5
Stem biomass	48	(g plant−1)	43.83	11.73	1.21	0.884	20.6	79.1
Total biomass (TDB)	45		120.22	34.86	1.41	1.088	60.2	224.6
Log leaf	45		1.494	0.123	−0.31	−0.201	1.2	1.72
Log root	48	Log (g plant−1)	1.613	0.166	0.30	0.357	1.27	2.05
Log stem	48		1.627	0.113	0.43	0.023	1.31	1.9
Log TDB	45		2.063	0.119	0.33	0.233	1.78	2.35

). Root and stem biomass were quite similar, whereas leaf biomass seemed to be slightly lower. In absolute terms, biomass at trial end was 15–20% lower than that reported for younger trees in comparable studies [23,50]. However, in relative terms it remained consistent with those studies [50]. Biomass partitioning showed roots as the dominant sink (37.2% of total), followed by stems (35.7%) and leaves (27.1%), reflecting typical resource allocation in young grafted citrus seedlings.

In general, plant biomass presented a positively skewed and leptokurtic distribution, with root biomass as the farthest away from normality, leaf biomass as the nearest and total biomass in between both, but strictly not fulfilling the requirements for normality (Table 4). Application of the logarithmic transformation successfully normalized the biomass distribution in all organs by reducing skewness and kurtosis to under 0.5 in absolute value (Table 4), thereby making them meet parametric test assumptions for the subsequent analyses.

3.3. Boron in Soils and Plants Throughout Treatments

The results of the four-way ANOVAs indicated that the irrigation water B concentration determined to a large extent the soil B concentration, both soluble and adsorbed, and also the B concentration in leaves and roots, but not in stems (

Table 5. p-values of the four-way analyses of variance of the B concentration in the different compartments of the soil–plant system and of the dry plant biomass as a function of the factors: boron concentration in the irrigation water (Boron), soil texture (Soil), rootstock and date. Italics indicates higher than 95% significance.

Factor	Logarithm to Base Ten of Plant B Concentration in…			Square Root of Soil B Concentration as…		Logarithm to Base Ten of Dry Biomass in…
	Leaves	Roots	Stems	Soluble	Adsorbed	Leaves	Roots	Stems	Total
Boron	<0.001	<0.001	0.291	<0.001	<0.001	0.006	0.446	0.599	0.161
Soil	<0.001	0.193	0.869	0.675	<0.001	0.622	0.840	0.919	0.740
Rootstock	0.376	0.942	0.157	0.784	0.050	0.366	0.002	0.021	0.007
Date	<0.001	<0.001		<0.001	<0.001
Boron × soil	0.853	0.657	0.045	0.382	0.139	0.856	0.896	0.490	0.751
Boron × rootstock	0.268	0.726	0.015	0.219	0.253	0.099	0.639	0.210	0.240
Boron × date	<0.001	<0.001		<0.001	<0.001
Soil × rootstock	0.499	0.699	0.615	0.269	0.898	0.743	0.226	0.265	0.321
Soil × date	0.006	0.122		0.043	0.012
Rootstock × date	<0.001	0.271		0.294	0.447

). Furthermore, a significant interaction between irrigation water B concentration and sampling date revealed (Table 5) that its effect on both soil B fractions, as well as on B in leaves and roots, changed in magnitude or direction over time, thereby pointing towards an accumulation effect on both soils and plants.

Interestingly, the rootstock factor (CC versus FA5) had no significant effect on the B concentration in any compartment of the soil–plant system. In addition, the effects of the irrigation B concentration did not differ between CC and FA5 rootstocks. However, the significant rootstock-by-date interaction for leaf B (Table 5) suggests a change in the speed of B plant accumulation depending on the rootstock. Specifically, at the beginning of the trial (up to 160 days), CC accumulated B in leaves more rapidly than FA5. However, by the end of the trial, the FA5 rootstock closed the gap of B concentration with CC.

Soil texture exhibited an effect on soil adsorbed B and on leaf B. Specifically, the clay loam accumulated more adsorbed B than the sandy loam. However, soil texture did not change the effect of the irrigation water B on B concentration in any compartment of the soil–plant system. On the contrary, for soil adsorbed and leaf B, a significant soil-by-date interaction emerged, with the sandy loam accumulating B more rapidly at the beginning of the trial (up to 160 days), which suggests delayed equilibrium (Table 5).

3.3.1. Boron Build-Up in Soils

Increasing irrigation water B concentration raised both soluble and adsorbed B in the soil (Figure 1). In contrast, higher clay content did not affect soluble B but did increase adsorbed B. Specifically, the clay loam contained significantly more adsorbed B (maximum of 14.2 mg kg⁻¹) than the sandy loam (4.37 mg kg⁻¹; Figure 1).

At the end of the experiment, the equilibrium between soluble and adsorbed soil B featured an adsorbed-to-soluble B ratio of 2.75 (Figure 2).

3.3.2. Boron Uptake by Plants

Leaf and root B increased progressively with irrigation water B. In contrast, stem B at the end of the experiment remained lower and independent of irrigation water B (Figure 3).

3.4. Plant Biomass Throughout Treatments

The results of the three-way ANOVAs at trial end indicate that the irrigation water B significantly determined the biomass of leaves, but not of roots or stems or in total (Table 5). Interestingly, the absence of significant interactions between irrigation water B and the other factors in the experimental layout (Table 5) suggests that the effects of water B on plant biomass were consistent across soil textures and, notably, did not differ between the CC and FA5 rootstocks. Soil had no effect on plant biomass, nor did it influence its response to water B. In contrast, the rootstock factor significantly affected stem and root biomass, and consequently total biomass, but had no effect on leaf biomass (Table 5).

Specifically, under low-boron conditions, no significant differences were observed between the two rootstocks. In contrast, under boron application, the FA5 rootstock produced 39% more total biomass than CC (

Table 6. Comparison of total dry biomass per plant at trial end between the two rootstocks at the three water boron concentrations.

Rootstock	Water Boron (mg L−1)	Total Dry Biomass (g plant−1)	Std. Dev.	Log (Total Dry Biomass)	Std. Dev.	p-Value
Carrizo	0.11	111.18	37.17	2.03	0.12	0.82
FA5	0.11	112.10	19.39	2.04	0.08
Carrizo	2	110.75	17.21	2.04	0.06	0.012
FA5	2	154.14	39.54	2.17	0.10
Carrizo	5	96.10	23.09	1.97	0.10	0.026
FA5	5	133.73	39.41	2.11	0.11

). In particular, FA5 biomass increased from 0.11 to 2 mg L⁻¹, peaking at this level, and then declined between 2 and 5 mg L⁻¹ (Table 6). Remarkably, however, biomass at 5 mg L⁻¹ remained higher than at 0.11 mg L⁻¹ (Table 6). These results indicate that FA5 exhibits greater tolerance to B than CC, with a toxicity threshold likely around 2 mg L⁻¹ in irrigation water, which corresponds to approximately 220 mg kg⁻¹ in leaves (Figure 4).

3.5. Modeling Plant Boron Uptake

The analysis of B concentrations in plant organs and soils showed a broad concentration range, supporting the suitability of the dataset for developing a model that simulates B uptake by citrus. To support model development, we examined the plots of B concentration in leaves and roots in particular, which serve as primary B storage sites to plant-available soil B.

The plotted data suggest the existence of curvilinear relationships between the decimal logarithms of both leaf and root B concentrations and the square roots of the three soil B fractions (soluble, adsorbed and total). In all cases, leaf and root B showed a gradually diminishing increase that approached a saturation limit, which was markedly higher in leaves than in roots (Figure 4). The shapes of these curves resembled a Mitscherlich-type response, and this function was therefore fitted to the experimental data in each case (Figure 4). In contrast, stem B was almost constant throughout the range of variation in the soil B fractions.

The determination coefficients (R²) of the fitted Mitscherlich function that relate leaf and root B to all three plant-available soil B fractions were significantly over 0.5 (

Table 7. Parameters of the Mitscherlich functions (Equation (1)) fitted to the relationships between the decimal logarithm of leaf and root B concentrations (y) and the square root of soil B fractions (x), along with the total sum of squares (TSS), the model sum of squares (SSM), and the coefficient of determination (R² = SSM/TSS). All parameters are expressed as the 95% confidence interval for the mean.

Plant Organ	Soil B Fraction	a	b	x0	R2
Leaf	Soluble	3.1 ± 0.4	0.6 ± 0.2	0.9 ± 0.3	0.81 ± 0.04
Root	Soluble	2.2 ± 0.1	0.7 ± 0.2	1.0 ± 0.3	0.79 ± 0.06
Leaf	Adsorbed		0.06 ± 0.11	2.7 ± 1.1	0.60 ± 0.08
Root	Adsorbed		0.20 ± 0.18	2.5 ± 1.5	0.62 ± 0.09
Leaf	Total	5 ± 4	0.14 ± 0.10	2.7 ± 1.2	0.72 ± 0.07
Root	Total	2.6 ± 0.6	0.22 ± 0.12	2.7 ± 1.1	0.73 ± 0.08

). However, the functions featuring soil soluble B presented highly significantly higher R² than the functions featuring soil total B (p < 0.003) and soil adsorbed B (p < 3 × 10⁻¹⁰) in this order. There were also highly significant differences between adsorbed and total (p < 2 × 10⁻⁴; Table 7). Conversely, the functions featuring foliar and root B did not present significantly different R² (p > 0.1 and Table 7). These results indicate that foliar and root B are closely linked to soil B and, especially, to the soluble fraction.

The diminishing-returns pattern of the relationship between leaf and root B and soil B fractions, i.e., approximately linear at low concentrations but gradually leveling off towards a saturation ceiling at higher concentrations, is consistent with a physiological uptake limit. Conversely, stem B content showed no significant dependence on soil B (Figure 4), supporting the role of stems as B transport pathways rather than storage sites.

4. Discussion

The pronounced foliar accumulation of B, together with the markedly low concentrations in roots and stems observed in this study, supports the notion that leaves act as the primary sink for B in citrus (Table 3), which is consistent with previous findings [51,52]. This pattern aligns with the passive uptake of B, its upward transport via the transpiration stream, and the limited redistribution from leaves [53].

Leaf and root B increased progressively with irrigation water B, supporting passive uptake via transpiration-driven xylem transport. In contrast, stem B at the end of the experiment remained lower and independent of irrigation water B (Figure 3), corroborating limited phloem mobility [51,54]. In plants, boric acid enters the roots primarily by passive diffusion due to the small size and neutral charge of the B(OH)₃ molecule [12]. Boron is subsequently loaded into the xylem and transported to the shoots via the transpiration stream, ultimately reaching the leaves [13]. Within leaf cells, B binds to complex cell wall polysaccharides, particularly rhamnogalacturonan II, thereby becoming key to the structural integrity of the cell wall [19]. In addition to its retention in the cell wall, boron, especially under conditions of excess, can bind to anthocyanins, facilitating its compartmentalization within vacuoles [18]. However, especially in citrus species, the substantial accumulation of B in leaves may exceed its vacuolar compartmentalization capacity, which may partially explain the pronounced B sensitivity observed in this genus [55].

A common feature of plant compounds to which B binds is the presence of hydroxyl groups (–OH) on adjacent carbon atoms. These neighbor diols enable the formation of highly stable dioxoborolane rings when B binds as a monoester, or spiro-bis (dioxoborolane) structures when it binds as a diester (Figure 5). From leaves, some B redistribution to other plant tissues via the phloem may occur [56,57], although the efficiency of this process seems to be species-dependent [52,58,59]. In fact, while root permeability and xylem transport for B is relatively widespread throughout plants, phloem translocation appears to be restricted to certain species that produce glycitols (sugar alcohols), which are polyols such as sorbitol, mannitol, galactitol, dulcitol, etc. [58,60,61,62].

Research has shown that glycitol-producing species, such as peach, apple, pear or olive, form esters between boric acid and hydroxyl groups in adjacent carbon atoms along these polyols that facilitate phloem B transport [58,60]. In contrast, non-glycitol-producing species like citrus present lower phloem B transport capacity, even though boric acid can also bind to sucrose, which is the most abundant disaccharide in citrus phloem [63]. However, the lower B-bonding capacity of sucrose may be due to the formation of two halfesters instead of one monoester, even though sucrose also presents hydroxyl groups in adjacent carbon atoms [56].

The high B-binding capacity of glycitols is attributable to the predominantly non-cyclic structures they adopt, which allow their hydroxyl substituents full rotational freedom around the C–C backbone of the molecules. As a result, adjacent hydroxyl groups can assume conformations in which the –O–C–C–O– atoms lie in a common plane, thereby fulfilling the requirements for B binding through the formation of dioxoborolane cycles (Figure 5). If in these five-membered heterocyclic rings the C–C–O angles are taken to be 107.8° as in ethanol [64], in addition to assuming average internuclear distances of 1.52, 1.42 and 1.36 Å for the C–C, C–O and B–O bonds [65,66], then the O–B–O angle is estimated to be 122.8° (Appendix A.2). The close match between this value and the trigonal angle of 120° characteristic of the sp² hybridization of B valence atomic orbitals in boron–oxygen compounds suggests effective atomic-orbital overlap between B and O and, consequently, strong covalent bonding of B in the dioxoborolane cycle formed with glycitols.

In contrast, sucrose predominantly adopts closed-ring conformations in its glucose and fructose units, forming glucopyranose and fructofuranose, respectively (Figure 5). These cyclic structures exhibit trans orientations on all pairs of adjacent hydroxyl groups with lack of rotational freedom around the C–C bonds (Figure 5). As a consequence, the –O–C–C–O– atoms cannot achieve a coplanar arrangement. Indeed, if values of 64.3, 62.8 and 78.4° are assumed for the three relevant O–C–C–O torsion angles in sucrose [67], then the oxygen atoms are positioned at estimated distances of 2.77, 2.79 and 2.94 Å (Appendix A.1), which are 0.05, 0.07 and 0.22 Å longer than double the average B–O distance of 1.36 Å in boron–oxygen compounds, i.e., 2.72 Å [66], and therefore incompatible with effective overlap of the sp² atomic orbitals from B with, simultaneously, the two sp³ atomic orbitals from the O atoms. Therefore, in sucrose, the stereoelectronic requirements for the formation of dioxoborolane rings are not met. As a result, sucrose cannot bind B in compounds of comparable stability to those formed by B with, e.g., rhamnogalacturonan II in cell walls, anthocyanins in vacuoles and glycitols in phloem. Consequently, sucrose cannot take B from rhamnogalacturonan II or anthocyanins and cannot function as a B transporter as effectively as glycitols.

In plants, glycitols are synthesized in leaves by reduction in the corresponding monosaccharides, then are translocated through the phloem, and frequently reoxidized again to the monosaccharides in sink organs, a reduction–oxidation cycle that requires ATP, NADH, and NADPH [68,69]. Therefore, the glycitol metabolism is energetically costly, and its evolutionary development suggests that glycitols must confer significant adaptive advantages to the plant. Among these, enhanced tolerance to abiotic stresses, such as drought, salinity, and extreme temperatures, seems key [68,69].

Notably, glycitols stand out as transport carbohydrates in trees of the Rosaceae family, particularly within the genera Malus (apple), Pyrus (pear), and Prunus (peach, plum, apricot, etc.) [70]. The Rosaceae family is widely distributed across the warm temperate and subtropical regions of the Northern Hemisphere, with a major center of diversification in central China [71,72]. The pronounced seasonal contrasts in temperature and rainfall that characterize these regions, now and presumably during most of the Neogene–Quaternary, may have favored the evolution and maintenance of the glycitols’ biosyntheses in the Rosaceae family as adaptive metabolites to the Northern mid-latitude climate features.

By contrast, the primary radiation center of citrus lies in Southeast Asia [73], a region characterized now and presumably during most of the Neogene–Quaternary by warmer, wetter and more climatically stable conditions with lower seasonal variability in temperature and rainfall. Therefore, under such low-latitude climate conditions, glycitols do not seem able to confer selective advantages, potentially explaining their reduced prominence in citrus.

Therefore, because of the scarcity of glycitols in citrus phloem, B remains quite immobile in their foliage, resulting in preferential accumulation in leaves, with comparatively smaller proportions retained in roots and stems [52,59,74]. The findings in this work are consistent with this characteristic B distribution pattern in citrus, as evidenced by: (1) significantly higher B concentrations in leaves than in other organs, and (2) stem B levels that do not exhibit any dose–response relationship (Figure 3).

Rootstocks may modulate citrus tolerance to B toxicity through two main mechanisms: (i) variations in the cell wall and plasma membrane permeability to B; and (ii) differences in the secretion of root exudates featuring hydroxyl groups in adjacent carbon atoms and, therefore, as capable as the aforementioned glycitols to bind B, thereby keeping the element within large molecules, likely charged similarly to those with spiro-bis(dioxoborolane) ring systems (Figure 5), to which plasma membranes are quite impermeable [62]. Differences in active B transport capacity by root cells may also influence B uptake. However, the variations in active B transport capacity are expected to play a role only if B is deficient, but not if it is excessive.

In this work, two rootstocks with contrasting a priori tolerance to B were compared: CC (B-sensitive) and FA5 (B-tolerant). These genotypes have previously been shown to differ in their capacity to accumulate B with lower concentrations in FA5 compared to CC under transpiration conditions similar to those in the present study [38]. Remarkably, however, both rootstocks displayed comparable B accumulation in the present study. Despite this similarity in B accumulation in plant material, the FA5 rootstock showed higher biomass production than CC under the B stress imposed by irrigating with 2–5 mg B L⁻¹, underscoring its superior vegetative development under elevated B conditions. The decline in biomass of CC in response to increasing B in irrigation water is consistent with previous findings [23,38], which reported reduced biomass production at higher water B concentrations for this rootstock. Additionally, the higher biomass accumulation exhibited by FA5 aligns with previous observations by Simon-Grao et al. [23], who also documented better vegetative growth of FA5 at high B concentrations. Accordingly, the higher tolerance of FA5 with regard to CC can be attributed to higher tissue tolerance, not adsorption.

Evaluating the relation between soluble and adsorbed B found in the soil, the particularly high adsorbed/soluble ratios we observed may reflect the enhanced B binding capacity of carbonate-rich soils, as shown by Goldberg et al. [15]. This behavior arises because B adsorption involves reactions between surface functional groups such as X–OH and X–O⁻ where X = Si, Al, Fe, etc., which act as Lewis bases, and the soluble species B(OH)₃, which behaves as a Lewis acid. The concentrations of these reactive species are strongly pH-dependent, as predicted by neglecting activity effects (Figure 6). As a result, the pH that maximizes the combined concentration products of B(OH)₃ with X–OH and X–O⁻ is expected to correspond to maximum B adsorption (Figure 6). This optimal pH for B adsorption is close to that characteristic of calcareous soils.

The nonlinear plant B uptake response observed in this study, characterized by a gradual diminishing increase, which flattens to a plateau at high soluble B concentrations in soil (>3 mg L⁻¹), suggests the operation of regulatory mechanisms that limit excessive B uptake and accumulation. One plausible contributing mechanism would be as simple as the abscission of leaves with the highest B loads. Indeed, the treatments receiving the highest B rates exhibited markedly greater levels of defoliation, supporting this interpretation. However, defoliation is the last effect of B phytotoxicity in a sequence that starts with other effects that diminish plant transpiration, such as stomatal closing [75], as represented in Figure 7. In this hypothesized causal-loop diagram, leaf B accumulation reduces transpiration by inducing stomatal closure; water uptake correspondingly decreases, leading to reduced B uptake which, in turn, limits further leaf B accumulation (Figure 7). This mechanism constitutes a negative feedback loop within the soil–plant system, manifesting as the dampened response the diminishing-returns law so well captures.

The practical implications of the findings in this work are especially pertinent to Mediterranean citrus production, where the increasing reliance on desalinated seawater [33] and treated wastewater [75] has heightened concerns regarding B toxicity. The results of this study confirm that FA5 may be a suitable rootstock under such conditions, whereas CC continues to be appropriate for low-boron environments, consistent with previous recommendations [23,76]. Furthermore, the demonstrated rootstock-dependent differences in B response patterns, together with the soil-specific patterns of B availability, underscore the need for management strategies that integrate both plant genetic traits and local edaphic conditions.

5. Conclusions

The leaf was identified as the main organ for B accumulation in citrus, in contrast to the stem, which acts primarily as a conductive pathway, and the roots, which serve as the main site for B absorption. Both CC and FA5 displayed similar patterns of B uptake, transport and distribution to the scion. In particular, no evidence of B exclusion was detected because absorption likely occurs through passive transport in both rootstocks, and phloem redistribution may be very limited because of lack of sugar alcohols and low B-transport capacity of sucrose as mechanistically interpreted. Nevertheless, FA5 exhibited greater B tolerance than CC, as evidenced by its higher biomass at elevated B levels (2–5 mg L⁻¹), which may support its suitability over CC for citrus propagation in regions with B-rich irrigation waters.

Robust relationships suggesting the law of diminishing returns were observed between the logarithm of B concentration in leaves and roots and the square root of soil soluble B. These relationships can be interpreted as reflecting a balancing feedback loop within the soil–plant system involving B, whereby B-induced stomatal closure likely reduces transpiration and, in turn, limits further B uptake. Accordingly, Mitscherlich-type functions were successfully fitted to predict leaf and root B buildup from soil soluble B.

The model presented here that connects citrus B uptake to soil B availability represents a novel contribution to B toxicity management in citrus production. It offers a final analytical link connecting irrigation water B concentration to plant B concentrations in which the other link would connect irrigation water B with soil soluble B concentration by simulating soil water flow and solute transport and soil B partitioning among soluble and adsorbed fractions.

This approach can be particularly valuable for the sustainable management of orchards irrigated with non-conventional and quality-degraded irrigation waters. However, before this seedling-based model can be implemented, it must be validated under open-field conditions with young and adult trees, and across a range of contrasting soil types, irrigation regimes and rootstocks. Under these conditions, leaching will play a greater role, and carry-over effects will emerge, enabling refinement of the model prior to its incorporation into decision-support tools for sustaining citrus production under non-conventional and deteriorating water quality conditions. Therefore, future research should assess the long-term impacts of sustained irrigation with B-rich waters under real open-field conditions, also studying the effects on citrus fruit quality and overall productivity.

Finally, the theoretical framework and mechanistic interpretations presented in this work advance understanding of B dynamics in citrus systems while highlighting key knowledge gaps. In this regard, future research should prioritize experimental testing of the mechanisms presented here to elucidate how plant B tolerance develops, with particular emphasis on FA5’s superior performance as a citrus rootstock.