Linking soil physicochemical properties to leaf nutrient composition in olive orchards on semi-arid calcareous soils

Abstract

This study examines whether bulk soil nutrient pools reliably reflect leaf nutrient status in olive orchards under alkaline, calcareous conditions, where carbonate buffering is known to constrain nutrient availability. We evaluated soil–leaf nutrient relationships in 25 rainfed olive orchards (‘Kilis Yaglik’) in southeastern Türkiye over two consecutive years (2017–2018). Soils remained strongly alkaline and highly calcareous, while organic matter declined by approximately 35% between years. Leaf nutrient composition did not consistently track soil nutrient dynamics. While some soil nutrients exhibited interannual variation, corresponding leaf responses were element-specific and not uniformly aligned with soil changes. Correlation and principal component analyses revealed variable and year-dependent associations between soil and leaf nutrients, with no reproducible, strong soil–leaf relationships across years. These findings indicate that, under carbonate-buffered conditions, bulk soil nutrient pools are not reliable predictors of canopy nutrient status. Instead, leaf nutrient composition appears to reflect integrated and element-specific availability patterns rather than direct soil–plant coupling.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-54157-0.

Introduction

The olive tree (Olea europaea L. ssp. europaea var. sativa) is among the oldest cultivated woody species in the Mediterranean region and has long held cultural and symbolic significance across ancient societies. Olive production is strongly concentrated in the Mediterranean region, which accounts for the vast majority of global output, with Türkiye, Greece, Italy, Spain, and several North African countries forming the core production zone¹. In Türkiye, olive cultivation occupies a notable share of tree crop cultivation and is economically important; unlike the global tendency toward table olive consumption, oil extraction remains the primary use in the country².

A defining constraint across many Mediterranean agroecosystems is the prevalence of alkaline, calcareous soils characterized by high calcium carbonate (CaCO₃) content and low organic matter. These conditions reduce the solubility and plant availability of key micronutrients—particularly Fe, Zn, Cu, and Mn—through precipitation and adsorption processes, and they promote nutrient antagonisms under strong pH buffering³. Consequently, nutrient deficiencies in olive leaves may occur despite seemingly adequate total soil nutrient concentrations⁴.

In perennial crops such as olive, leaf nutrient composition provides an integrated indicator of nutritional status over time, but it does not necessarily track soil supply in a simple, linear manner. Remobilization and storage processes can decouple soil availability from foliar concentrations across seasons⁵. As a result, interpreting individual nutrients in isolation may overlook relevant inter-element interactions, whereas multivariate approaches can better capture coordinated nutrient behavior. Correlation analysis and principal component analysis (PCA) are therefore widely applied to summarize nutrient interaction patterns and to relate them to underlying environmental gradients^6–8.

This study evaluates whether soil physicochemical properties (pH, CaCO₃, and organic matter) are associated with leaf nutrient composition in rainfed olive orchards under semi-arid, calcareous conditions. Based on current understanding of carbonate-buffered soils, we formulated the following testable hypotheses: (i) Leaf micronutrient concentrations are negatively associated with high soil pH and CaCO₃ levels. (ii) Higher soil organic matter is associated with relatively improved leaf nutrient status⁹. (iii) Soil–leaf nutrient associations are not consistent across years, reflecting temporal variability in soil–plant coupling8. To test these hypotheses, we characterized soil physicochemical properties and nutrient contents, determined leaf nutrient composition, evaluated soil–leaf correlations, and applied principal component analysis (PCA) to summarize multivariate patterns across years¹⁰.

Results

Soil physicochemical characteristics

The soils of the studied orchards were moderately to strongly alkaline, with pH values ranging from 7.60 to 8.04, and exhibited high calcium carbonate content (28.0–36.2%) together with low organic matter levels. Between 2017 and 2018, mean soil pH decreased significantly from 7.86 to 7.80 (p < 0.001), while calcium carbonate content remained statistically unchanged (p = 0.327). Organic matter declined from approximately 1.82% to 1.12%, corresponding to a reduction of about 35% (p < 0.001) (Fig. 1).

Fig. 1.

Interannual variation in soil pH, CaCO₃, and organic matter across 25 rainfed olive orchards during the 2017 and 2018 growing seasons. Values represent mean Interannual differences were tested using repeated-measures ANOVA or non-parametric alternatives when assumptions were not met. Exact p-values are reported in the Results section.

Soil nutrient composition

The macronutrient and micronutrient dynamics of the orchard soils showed temporal variation between 2017 and 2018. Nitrogen, potassium, calcium, magnesium, copper, and iron all exhibited statistically significant changes (p < 0.001; see Table S3 and S4 for mean differences). Phosphorus, manganese, and zinc remained stable across the two years (p > 0.05) (Figs. 2 and 3).

Fig. 2.

Interannual variation in leaf and soil macronutrient concentrations across the sampled olive orchards in 2017 and 2018. Data points represent individual orchards, and markers indicate the mean. Leaf and soil nutrient concentrations are presented separately because analytical fractions and units differ between plant and soil matrices.

Fig. 3.

Interannual variation in leaf and soil micronutrient concentrations across the sampled olive orchards in 2017 and 2018. Data points represent individual orchards, and markers indicate the mean. Soil micronutrients represent extractable concentrations, whereas leaf micronutrients represent tissue concentrations.

Leaf nutrient composition

Leaf macronutrient concentrations showed interannual variation. Nitrogen showed no significant change across years (p = 0.071. Phosphorus exhibited a significant difference (p < 0.001), changing from 0.041 to 0,051. Potassium, calcium, and magnesium also varied significantly between years (p < 0.001), with mean concentrations shifting from 0.58% to 0.21% for potassium, from 2.97% to 0.838% for calcium, and from 0.126% to 0.254% for magnesium. Among micronutrients, copper and iron decreased (p < 0.001), zinc increased (p < 0.001), and manganese remained statistically unchanged (p = 0.313) (Figs. 2 and 3).

Multivariate structure of leaf nutrient profiles

Principal Component Analysis (PCA) evaluated the interrelationships among leaf nutrients and their seasonal variation (Fig. 4). In 2017, two principal components explained 59.11% of total variance (PC1 = 31.97%, PC2 = 27.15%), while in 2018 the explained variance increased to 63.16% (PC1 = 40.40%, PC2 = 22.76%). Macronutrients (N, P, K, Ca) predominantly loaded on PC1, whereas micronutrients (Fe, Mn, Zn, Cu) clustered on PC2. The combined-year PCA explained 38.59% of total variance and separated macro- and micronutrient variables.

Fig. 4.

Principal Component Analysis (PCA) of leaf nutrient composition for 2017 and 2018. The biplots illustrate the patterns of covariation and covariation patterns between macronutrient and micronutrient domains across seasons. PCA was used as a descriptive ordination method; the positions of variables indicate covariation patterns and should not be interpreted as causal relationships.

Multivariate structure of soil nutrient profiles

PCA of soil data (Fig. 5), two components accounted for 56.65% of total variance in 2017 (PC1 = 43.12%, PC2 = 13.53%), and 57.75% in 2018 (PC1 = 44.02%, PC2 = 13.73%). PC1 was primarily associated with K, Mg, and Cu. PC2 grouped Ca and Fe. The combined two-year model explained 46.38% of total variance. Organic matter and lime were positioned near orthogonal axes in the PCA space.

Fig. 5.

Principal Component Analysis (PCA) of soil physicochemical characteristics and nutrient profiles for 2017 and 2018. The plots display the clustering of Ca–Fe and K–Mg–Cu variables within the strongly alkaline, calcareous soil matrix. PCA was used as a descriptive ordination method; the positions of variables indicate covariation patterns and should not be interpreted as causal relationships.

Leaf nutrient contents across orchards

Leaf nutrient concentrations for all sampling sites and both years are presented in Supplementary Table S1 and Table S2. Across the 25 olive orchards, macronutrient levels displayed moderate spatial variability and remained within a relatively bounded range across orchards. Nitrogen concentrations ranged between 0.94 and 1.03% in 2017 and 0.91–1.17% in 2018, whereas phosphorus values were relatively consistent (0.04–0.06%).

Potassium, calcium, and magnesium showed stronger site-dependent variability, with 2018 values generally lower than in 2017 (g kg⁻¹ for K, Ca, and Mg). Micronutrients exhibited pronounced spatial variability across orchards, together with interannual variation between sampling years. Copper ranged from 2.4 to 7.9 mg kg⁻¹ in 2018, while iron ranged from 65 to 196 mg kg⁻¹. Manganese (12–33 mg kg⁻¹) and zinc (9–44 mg kg⁻¹) values indicated considerable heterogeneity among orchards. Statistical evaluation confirmed significant inter-site differences for K, Ca, Mg, Cu, Fe, and Zn (p < 0.001), while N, P, and Mn remained largely uniform.

Soil characteristics and nutrient contents across orchards

Soil physicochemical properties and nutrient element concentrations for the 25 olive orchards are summarized in Supplementary Table S3 and Table S4. The soils were moderately to strongly alkaline (pH 7.61–8.05) with high calcium carbonate contents (26–47%) and low to moderate organic matter (0.5–2.6%). Despite the slight interannual decline in organic matter, variability across orchards remained substantial. Macronutrient concentrations showed strong site dependency. Nitrogen ranged from 0.03 to 0.52%, while potassium varied widely from 70 to 410 mg kg⁻¹, confirming considerable heterogeneity among orchards.

Phosphorus levels were generally low (4.3–9.9 mg kg⁻¹) with non-significant interannual differences (p > 0.05). Calcium and magnesium were dominant cations, with calcium reaching up to 2257 mg kg⁻¹ and magnesium 480 mg kg⁻¹ in 2017, both showing significant decreases in 2018 (p < 0.001). Micronutrient variability was also pronounced: copper and iron ranged from 0.3 to 0.9 and 2.3–9.9 mg kg⁻¹, respectively, while manganese and zinc varied between 3.4 and 25.8 and 0.5–1.9 mg kg⁻¹. All micronutrients displayed highly significant inter-site variation (p < 0.001).

Leaf nutrient correlations (2017)

Nitrogen showed positive associations with phosphorus (r = 0.214) and iron (r = 0.414) (Table 1). Calcium exhibited negative correlations with magnesium (r = −0.459) and manganese (r = −0.653). Copper correlated positively with calcium (r = 0.433) but negatively with phosphorus (r = −0.466). Iron and manganese showed a weak relationship (r = −0.056).

Table 1.

Pearson correlation matrix of leaf macronutrient and micronutrient concentrations across the 25 olive orchards in 2017.

Variables (Leaf)	L-N	L-P	L-K	L-Ca	L-Mg	L-Cu	L-Fe	L-Mn	L-Zn
L-N	1	0.214	−0.582	0.054	−0.483	−0.244	0.414	−0.300	−0.131
L-P	0.214	1	−0.235	0.140	−0.346	−0.466	0.205	0.040	0.327
L-K	−0.582	−0.235	1	−0.191	0.349	0.185	−0.389	0.371	0.069
L-Ca	0.054	0.140	−0.191	1	−0.459	0.433	−0.155	−0.653	−0.120
L-Mg	−0.483	−0.346	0.349	−0.459	1	0.254	−0.145	0.443	0.153
L-Cu	−0.244	−0.466	0.185	0.433	0.254	1	−0.445	−0.340	−0.321
L-Fe	0.414	0.205	−0.389	−0.155	−0.145	−0.445	1	−0.056	0.070
L-Mn	−0.300	0.040	0.371	−0.653	0.443	−0.340	−0.056	1	0.425
L-Zn	−0.131	0.327	0.069	−0.120	0.153	−0.321	0.070	0.425	1

Significant values are in bold.

Leaf nutrient correlations (2018)

In 2018 (Table 2), calcium exhibited positive correlations with magnesium (r = 0.896), manganese (r = 0.790), iron (r = 0.619), and zinc (r = 0.508). Nitrogen and phosphorus were also positively correlated (r = 0.654). The Fe–Mn relationship was positive in 2018 (r = 0.550).

Table 2.

Pearson correlation matrix of leaf macronutrient and micronutrient concentrations across the 25 olive orchards in 2018.

Variables (Leaf)	L-N	L-P	L-K	L-Ca	L-Mg	L-Cu	L-Fe	L-Mn	L-Zn
L-N	1	0.654	0.218	−0.007	0.140	0.255	0.079	0.163	−0.129
L-P	0.654	1	0.223	−0.117	0.061	0.323	0.132	0.192	0.022
L-K	0.218	0.223	1	0.018	−0.053	0.191	0.202	0.116	−0.024
L-Ca	−0.007	−0.117	0.018	1	0.896	−0.321	0.619	0.790	0.508
L-Mg	0.140	0.061	−0.053	0.896	1	−0.284	0.603	0.827	0.565
L-Cu	0.255	0.323	0.191	−0.321	−0.284	1	−0.139	−0.199	−0.271
L-Fe	0.079	0.132	0.202	0.619	0.603	−0.139	1	0.550	0.409
L-Mn	0.163	0.192	0.116	0.790	0.827	−0.199	0.550	1	0.399
L-Zn	−0.129	0.022	−0.024	0.508	0.565	−0.271	0.409	0.399	1

Significant values are in bold.

Soil characteristics and nutrient correlations (2017)

The 2017 soil correlation matrix (Table 3) showed that soil pH was negatively correlated with calcium carbonate content (r = −0.544) and positively with potassium (r = 0.590) and magnesium (r = 0.641). Calcium carbonate exhibited negative correlations with K (r = −0.860) and Cu (r = −0.720).

Table 3.

Pearson correlation coefficients among soil physicochemical properties (pH, CaCO₃, organic matter) and extractable soil nutrients in 2017.

Variables (Soil)	pH	CaCO₃	S-N	S-P	S-K	S-Ca	S-Mg	S-Cu	S-Fe	S-Mn	S-Zn
pH	1	−0.544	−0.295	−0.339	0.590	−0.017	0.641	0.412	0.388	−0.046	−0.136
CaCO₃	−0.544	1	0.422	0.569	−0.860	0.345	−0.314	-0.720	−0.423	−0.318	0.017
S-N	−0.295	0.422	1	0.384	−0.255	−0.137	−0.215	−0.185	−0.227	0.187	0.459
S-P	−0.339	0.569	0.384	1	−0.462	−0.102	−0.320	−0.247	−0.468	−0.097	−0.042
S-K	0.590	−0.860	−0.255	−0.462	1	−0.164	0.421	0.712	0.521	0.360	0.183
S-Ca	−0.017	0.345	−0.137	−0.102	−0.164	1	0.045	−0.275	0.072	−0.449	−0.136
S-Mg	0.641	−0.314	−0.215	−0.320	0.421	0.045	1	0.457	0.366	0.171	0.110
S-Cu	0.412	−0.720	−0.185	−0.247	0.712	−0.275	0.457	1	0.211	0.278	0.200
S-Fe	0.388	−0.423	−0.227	−0.468	0.521	0.072	0.366	0.211	1	0.233	−0.182
S-Mn	−0.046	−0.318	0.187	−0.097	0.360	−0.449	0.171	0.278	0.233	1	0.469
S-Zn	−0.136	0.017	0.459	−0.042	0.183	−0.136	0.110	0.200	−0.182	0.469	1

Significant values are in bold.

Soil characteristics and nutrient correlations (2018)

In 2018 (Table 4), potassium correlated positively with copper (r = 0.729) and zinc (r = 0.834). Calcium showed a positive relationship with iron (r = 0.517).

Table 4.

Pearson correlation coefficients among soil physicochemical properties (pH, CaCO₃, organic matter) and extractable soil nutrients in 2018.

Variables (Soil)	pH	CaCO₃	S-N	S-P	S-K	S-Ca	S-Mg	S-Cu	S-Fe	S-Mn	S-Zn
pH	1	0.166	0.276	−0.297	−0.286	−0.147	−0.041	−0.298	−0.027	−0.138	−0.270
CaCO₃	0.166	1	−0.112	0.296	0.025	−0.030	−0.050	−0.028	0.001	−0.082	0.052
S-N	0.276	−0.112	1	0.145	0.169	−0.202	−0.325	−0.075	−0.319	−0.008	0.056
S-P	−0.297	0.296	0.145	1	−0.139	−0.142	−0.220	−0.221	−0.462	−0.049	0.017
S-K	−0.286	0.025	0.169	−0.139	1	−0.062	0.035	0.729	0.165	0.306	0.834
S-Ca	−0.147	−0.030	−0.202	−0.142	−0.062	1	0.196	−0.030	0.517	−0.282	−0.240
S-Mg	−0.041	−0.050	−0.325	−0.220	0.035	0.196	1	0.111	0.347	0.070	−0.022
S-Cu	−0.298	−0.028	−0.075	−0.221	0.729	−0.030	0.111	1	0.183	0.240	0.695
S-Fe	−0.027	0.001	−0.319	−0.462	0.165	0.517	0.347	0.183	1	0.310	0.029
S-Mn	−0.138	−0.082	−0.008	−0.049	0.306	−0.282	0.070	0.240	0.310	1	0.379
S-Zn	−0.270	0.052	0.056	0.017	0.834	−0.240	−0.022	0.695	0.029	0.379	1

Significant values are in bold.

Diagnostic nutrient ratios

Diagnostic nutrient ratios were calculated exclusively from leaf nutrient concentrations. The K: (Ca + Mg) ratio was calculated on a mass basis, whereas Fe: Mn was expressed on both a mass and molar basis; the K: Ca: Mg ratio was calculated on a molar basis with K normalized to 1. The K: (Ca + Mg) ratio across orchards showed limited interannual variability, clustering around 0.19–0.20 (2017: 0.191 ± 0.081; 2018: 0.197 ± 0.037) with narrow year-to-year spread (Table 5).

Table 5.

Diagnostic leaf nutrient ratios (K: (Ca + Mg), Fe: Mn, and K: Ca: Mg) calculated on mass and molar bases for 2017 and 2018. Averages for alkaline-calcareous soil subsets (pH > 7; CaCO₃ ≥ 15%) are provided to highlight context-specific nutrient balances.

Year	Subgroup condition	K: (Ca + Mg) Mean ± SD	K: (Ca + Mg) Min–Max	Fe: Mn(mass) Mean ± SD	Fe: Mn(mass) Min–Max	Fe: Mn(molar) Mean ± SD	K: Ca: Mg (molar, K = 1)
2017	All orchards (n = 20)	0.191 ± 0.081	0.014–0.374	5.35 ± 1.67	5.27 ± 1.65	5.27 ± 1.65	1: 8.21: 0.56
2018	All orchards (n = 20)	0.197 ± 0.037	0.144–0.280	7.05 ± 2.82	6.93 ± 2.78	6.93 ± 2.78	1: 3.91: 1.96
2017	pH > 7; CaCO₃ ≥ 15% (subset defined by soil data) (n = 14)	0.162	0.014–0.260	5.36	2.30–9.45	5.27	1:11.03:0.74
2018	pH > 7; CaCO₃ ≥ 15% (subset defined by soil data) (n = 14)	0.196	0.144–0.276	7.34	2.99–16.10	7.23	1: 3.96: 1.98
Combined	Contextual average	0.179	–	6.35	–	6.25	1: 7.5: 1.4

The Fe: Mn ratio exhibited moderate variation (2017: 5.27 ± 1.65; 2018: 6.93 ± 2.78 molar), suggesting a broadly comparable Fe–Mn balance despite interannual spread. Under the alkaline–calcareous subset conditions (pH > 7; CaCO₃ ≥ 15%), the ratios showed context-specific average values of approximately 0.18 for K: (Ca + Mg), 6 for Fe: Mn on a molar basis, and 1:7.5:1.4 for K: Ca: Mg on a molar basis. These context-specific means were derived from subset averages for 2017 (0.162, 5.27, 1:11.03:0.74) and 2018 (0.196, 7.23, 1:3.96:1.98).

Soil–leaf nutrient correlations

Correlations between soil and corresponding leaf nutrient concentrations varied among elements and between years (Tables 6 and 7). In 2017, CaCO₃ was positively associated with leaf Ca (r = 0.574), while organic matter showed a negative relationship with leaf K (r = −0.652). Soil Cu was negatively correlated with leaf Ca (r = −0.560), and CaCO₃ exhibited a negative association with leaf Mn (r = −0.523). In 2018, the strongest associations were observed between soil Mg and leaf Cu (r = 0.494), and between soil pH and leaf Cu (r = −0.470). Across both years, no single nutrient exhibited consistently strong soil–leaf coupling.

Table 6.

Pearson correlation coefficients (r) indicating the degree of coupling between soil physicochemical properties/nutrients and corresponding leaf nutrient concentrations across the sampled orchards in 2017.

	L-N	L-P	L-K	L-Ca	L-Mg	L-Cu	L-Fe	L-Mn	L-Zn
pH	−0.459	−0.273	0.459	−0.240	0.380	0.219	−0.429	0.352	0.167
CaCO₃	0.381	0.210	−0.504	0.574	−0.369	0.205	0.170	−0.523	−0.188
OM	0.506	0.182	−0.652	0.476	−0.334	0.148	0.357	−0.573	−0.268
S-N	0.322	0.299	−0.487	0.362	−0.196	0.086	0.328	−0.341	−0.132
S-P	0.528	0.180	−0.485	0.225	−0.496	0.078	0.243	−0.250	−0.157
S-K	−0.372	−0.254	0.484	−0.387	0.281	0.001	−0.227	0.316	−0.064
S-Ca	−0.249	−0.197	0.288	0.187	0.262	0.457	−0.537	−0.186	−0.117
S-Mg	−0.258	−0.360	0.316	−0.151	0.409	0.265	−0.361	0.136	−0.084
S-Cu	−0.261	−0.262	0.466	−0.560	0.371	−0.142	−0.055	0.416	−0.054
S-Fe	−0.444	−0.035	0.218	−0.329	0.416	0.138	−0.316	0.364	−0.014
S-Mn	0.436	0.046	−0.270	−0.288	0.020	−0.324	0.369	−0.013	−0.147
S-Zn	0.289	0.189	−0.166	0.095	−0.217	−0.215	0.438	−0.455	−0.076

Significant values are in bold.

Table 7.

Pearson correlation coefficients (r) indicating the degree of coupling between soil physicochemical properties/nutrients and corresponding leaf nutrient concentrations across the sampled orchards in 2018.

	L-N	L-P	L-K	L-Ca	L-Mg	L-Cu	L-Fe	L-Mn	L-Zn
pH	−0.245	−0.370	0.137	0.207	0.023	−0.470	0.104	−0.054	−0.033
CaCO₃	−0.057	0.236	0.106	−0.079	0.045	−0.075	−0.055	−0.056	0.302
OM	0.263	0.173	0.215	0.197	0.297	−0.295	0.278	0.178	0.076
S-N	0.002	0.019	0.149	−0.182	−0.136	−0.143	0.270	−0.226	−0.241
S-P	0.227	0.259	−0.055	−0.267	−0.044	0.051	0.065	−0.196	−0.169
S-K	0.021	0.269	0.049	−0.197	−0.071	0.206	0.084	0.056	0.054
S-Ca	0.117	0.044	0.132	−0.145	−0.174	−0.089	−0.068	−0.080	−0.207
S-Mg	0.147	0.154	0.181	−0.163	−0.128	0.494	0.033	0.111	−0.013
S-Cu	0.072	0.305	0.229	−0.033	−0.052	0.191	0.105	0.173	0.143
S-Fe	0.026	0.106	−0.056	−0.109	−0.058	0.381	0.086	0.050	−0.008
S-Mn	−0.091	0.032	−0.181	−0.053	0.003	0.118	0.051	0.077	−0.014
S-Zn	0.125	0.348	0.064	−0.228	−0.160	0.344	0.150	−0.033	−0.039

Significant values are in bold.

Discussion

The present results should first be interpreted as descriptive evidence of weak and year-dependent soil–leaf associations in rainfed ‘Kilis Yaglik’ olive orchards grown on alkaline, calcareous soils. Because this study measured bulk soil properties, extractable soil nutrients, leaf nutrient concentrations, and their correlations across two years, it can identify association patterns but cannot directly resolve the physiological or rhizosphere mechanisms behind them. Therefore, apparent soil–leaf decoupling is discussed here as an observed correlation pattern, not as a directly demonstrated mechanism.

Mediterranean olive groves in semi-arid settings are commonly characterized by calcareous soils and low organic matter¹¹. In our dataset, soils were consistently alkaline and CaCO₃-rich, organic matter was low and declined between years, and carbonate levels remained broadly stable. The decline in soil organic matter (from 1.82% to 1.12%) represents a substantial interannual change in the measured topsoil pool. This change may be compatible with processes such as management-related disturbance, rainfall variability, erosion, or enhanced mineralization under semi-arid conditions; however, these processes were not directly measured in the present study. Accordingly, the observed decline should be interpreted as a measured temporal shift rather than as evidence for a specific underlying mechanism. Regardless of the specific mechanism driving this interannual decline, these boundary conditions are agronomically consequential because high pH and carbonate-rich chemistry can restrict micronutrient solubility and effective availability via precipitation and sorption, while low organic matter limits nutrient retention and metal complexation^12,13. Under these conditions, soil nutrient status should be interpreted against the specific analytical fraction (e.g., DTPA-extractable rather than total content) reported (e.g., total vs. extractable), because critical thresholds are method-dependent. Recent syntheses further emphasize that plant nutrition in calcareous soils is constrained by CaCO₃-buffered alkalinity and carbonate/bicarbonate-driven chemistry that governs ion speciation and phytoavailability, and that these constraints can be sensitive to environmental variability¹⁴. The interannual leaf and soil shifts should be interpreted in the context of a persistently calcareous system experiencing declining organic matter and a modest pH change, not as a fundamental shift in soil constraint intensity. Although the field data reflect the 2017–2018 seasons, the robust buffering capacity and highly calcareous nature of these semi-arid soils represent persistent, long-term physicochemical constraints. Therefore, the observed soil–leaf association patterns and the proposed descriptive diagnostic approach may remain relevant for interpreting nutrient status in comparable Mediterranean olive orchards.

A second overarching feature is that interannual variation is superimposed on substantial spatial heterogeneity across orchards. Wide ranges in soil macro- and micronutrient pools and pronounced inter-site differences in several leaf nutrients indicate that spatial variability is agronomically relevant rather than incidental. Orchard-to-orchard differences in carbonate status and associated soil properties may contribute to variation in effective nutrient availability and soil–plant association patterns, such that temporal patterns are best interpreted within a site-specific context¹⁵. This framing aligns with evidence that Mediterranean olive-grove soils can exhibit marked within-region variability in fertility indicators relevant to olive production¹⁶ and with broader observations that heterogeneous landscapes express temporal trends within a wide spatial range rather than as uniform, system-wide shifts¹⁷. Here, “year effects” are best treated as site-conditioned signals expressed on top of high baseline between-orchard variability.

Under these persistent edaphic constraints, soil nutrients separated into temporally dynamic versus comparatively stable elements, indicating that interannual change was selective rather than system-wide. In this dataset, soil N, K, Ca, Mg, Cu, and Fe shifted across years, whereas P, Mn, and Zn remained comparatively stable. Such selectivity is consistent with expectations for calcareous soils, where CaCO₃ buffering, dissolution–precipitation dynamics, and seasonal environmental variability can differentially affect nutrient speciation and the measured pools, making some indicators more responsive to temporal perturbations than others¹⁴. The soil results therefore support an element-specific interpretation of temporal change, and they establish a necessary baseline for discussing canopy dynamics: leaf responses should not be assumed to track a uniform soil shift when only a subset of soil nutrients exhibit temporal movement. In short, the soil record supports an element-specific (not “overall fertility”) reading of interannual change.

Phosphorus illustrates how “temporal stability” can coexist with agronomic constraint. Although soil P was comparatively stable between years, P levels were generally low across orchards, suggesting that stability may reflect a persistent baseline limitation rather than a non-limiting status. In calcareous–alkaline soils, carbonate minerals and Ca-rich conditions commonly suppress effective P availability through sorption and Ca–P phase formation, so a low-but-stable soil P signal can remain agronomically meaningful even without a year effect¹⁸. Therefore, the low and temporally stable soil P values are consistent with a persistently low soil-test P condition; however, the present data do not directly establish the physiological severity of P limitation. This highlights the importance of interpreting P against method-specific critical levels rather than as a method-free “adequacy” classification^19–21. Soil-test ‘stability’ therefore reflects the measured pool and does not guarantee nutritional sufficiency. Given that soil-test pools can remain low-but-stable while effective availability is constrained, bulk soil data alone may not accurately reflect canopy nutrient status; accordingly, we next evaluate whether leaf macronutrients track soil temporal behaviour or diverge under these carbonate-buffered conditions.

Leaf macronutrient behaviours did not mirror soil dynamics in a one-to-one manner, and the N–P pattern is particularly informative in that respect. Leaf N showed no significant interannual shift, whereas leaf P differed between years, even though soil P remained comparatively stable over time. This divergence indicates that leaf macronutrient concentrations did not directly track bulk soil nutrient pools. A possible interpretation, consistent with perennial crop physiology, is that leaf nutrient composition integrates soil availability with internal plant processes; however, these processes were not directly measured in the present study. Consistent with this view, soil–leaf decoupling has been observed in olive under field conditions, including cases where soil P is higher in control soils while plant P is higher under contrasting management, demonstrating that leaf P can depart from bulk/soil-test patterns²². Under calcareous boundary conditions, such divergence is plausible because leaf composition integrates uptake constraints, whole-plant physiological integration, and phenological timing, and it can therefore register meaningful change even when corresponding soil indicators show limited temporal responsiveness. This interpretation is consistent with documented seasonal shifts in olive leaf nutrient concentrations that highlight strong sensitivity to plant phenology and sampling timing^23–25. Practically, this argues for treating leaf P as an integrated canopy indicator rather than assuming soil-test P stability implies a stable canopy P status.

Micronutrients showed a clearer temporal signature in the canopy than macronutrients: Cu and Fe declined, Zn increased, and Mn remained comparatively stable. Notably, leaf Zn increased while soil Zn remained comparatively stable, reinforcing that canopy Zn dynamics were not a simple reflection of bulk soil Zn change. In parallel, Cu and Fe shifted in the soil across years, yet the canopy moved toward lower Cu and Fe, consistent with a constrained effective supply environment in which bulk soil change alone does not capture plant-available Cu and Fe under carbonate constraints. In alkaline–calcareous settings, Fe and Cu availability is sensitive to carbonate chemistry and the soil chemical/binding environment, whereas Zn availability is also pH-dependent and micronutrient behaviour can be element-specific due to differences in speciation and retention processes that govern micronutrient mobility and plant availability¹⁴. The divergent directions of Cu–Fe versus Zn–Mn in the leaf therefore argue against treating “micronutrients” as a homogeneous domain; instead, the canopy pattern points to element-specific constraints operating within the same carbonate-buffered system. Although pH declined significantly, the system remained strongly calcareous; carbonate-related constraints likely remained relevant, and the canopy micronutrient shifts are consistent with element-specific availability constraints within a persistently calcareous framework, although the underlying mechanisms were not directly tested. Because several leaf nutrients shifted in different directions across years, PCA was used to summarize multivariate covariation patterns among nutrients rather than to infer underlying mechanisms. The observed separation of nutrient groups indicates structured co-variation, but does not identify the physiological processes responsible for these patterns²⁶. The separation of macro- and micronutrients along PCA axes is best interpreted as a descriptive pattern of covariation rather than evidence of distinct regulatory processes. While such separation is consistent with known differences in nutrient demand and mobility in perennial plants, these processes were not directly examined in this study. Accordingly, the PCA structure is treated here as a summary of coordinated variation among nutrients without implying specific physiological control mechanisms.

The soil multivariate structure indicates that the system is not organized along a single “fertility axis,” but instead reflects coupled variability among several cations (including K, Mg, and Cu), with lime and organic matter behaving more independently—a pattern consistent with calcareous-soil chemistry where CaCO₃ buffering and Ca–ion interactions govern ion speciation and phytoavailability while organic matter contributes distinct sorption and biogeochemical functions²⁷. This organization is consistent with the limited ability of individual soil variables to account for leaf nutrient composition in this dataset.

Correlation patterns in both leaf and soil reorganized between years, indicating temporal plasticity rather than fixed nutrient coupling. This year-to-year reconfiguration is consistent with evidence that year-to-year climatic or management variation can weaken or reconfigure nutrient coupling in soils²⁷ and with the view that effective micronutrient availability in calcareous systems reflects a moving balance among carbonate chemistry and interacting binding/ligand processes rather than carbonate percentage alone²⁷. Accordingly, correlation structures are best treated as year-contingent descriptors of a prevailing “nutrient balance state,” not as stable physiological constants.

The year-dependent and element-specific soil–leaf correlation patterns observed in this study mirror broader findings in olive and other perennial systems grown on alkaline soils, where pH and carbonate content strongly influence nutrient availability. Specifically, soil pH has been shown to be a dominant factor controlling nutrient solubility and uptake in olive orchards, with alkaline conditions reducing availability of several macro- and micronutrients and reshaping soil–plant nutrient relationships across seasons and treatments²⁸. In such carbonate-buffered environments, weak or transient soil–leaf correlations are expected because bulk soil pools do not consistently translate into plant-available forms, and these observations align with previous studies suggesting that nutrient availability in calcareous systems may not be adequately captured by bulk soil measurements alone^14,29,30. Recent work highlights that under calcareous constraints, canonical soil–leaf transfer functions may not hold, especially for micronutrients like iron and copper, whose phytoavailability is strongly modulated by soil chemical context rather than total content alone^14,30,31.

The absence of reproducible strong soil–leaf correlations across both sampling years suggests that canopy nutrient status cannot be explained by bulk soil nutrient pools alone. This perspective aligns with field evidence that nutrient uptake and leaf nutrient responses can vary with rhizosphere pH and carbonate effects, and that leaf nutrient composition often reflects dynamic physiological adjustments rather than static soil signals^32,33. The conditional nature of soil–leaf associations underscores the importance of integrated diagnosis combining soil chemical constraints and plant tissue analysis, particularly in calcareous Mediterranean systems where carbonate buffering and pH interact to shape nutrient availability and plant responses^34–36.

The diagnostic ratios derived from leaf data provide an integrative way to summarize these interactions without over-interpreting individual elements³⁷. A relatively constrained K: (Ca + Mg) ratio across years, together with Ca-dominant K: Ca: Mg molar patterns, is consistent with a base-cation environment in which K–Mg balance may be sensitive to competitive interactions under calcareous conditions³⁸, even when element-wise concentrations fall within conventional sufficiency interpretations³⁷. At the same time, shifts in the internal partitioning between Ca and Mg within the K: Ca: Mg pattern suggest rebalancing within the base-cation pool rather than a simple “K stability” narrative. This ratio-based framing aligns with mechanistic work discussing K–Mg antagonism and ratio-oriented mitigation approaches³⁹, although antagonistic effects and mitigation responses were not tested in this study. Meanwhile, Fe: Mn ratios that remained within a broadly comparable range (despite interannual spread) support the utility of Fe–Mn balance as a descriptive indicator; in calcareous soils, the leaf Fe/Mn molar ratio has been used as a practical index to diagnose Fe–Mn balance and recovery from Fe chlorosis⁴⁰. Overall, ratio-based diagnostics are best interpreted as concise indicators of nutrient balance unless directly tested mechanistically. Ratios are most defensible here as compact summaries of interaction patterns rather than as mechanistic proof.

From an applied perspective, the combined evidence supports a management framing that is responsive to element-specific dynamics rather than broad, undifferentiated fertility programs. Given the observed temporal plasticity and soil–leaf decoupling, interventions aimed at stabilizing effective micronutrient supply (e.g., organic inputs, targeted foliar programs, or practices supporting the soil chemical environment governing availability) can be considered alongside conventional soil-based approaches, rather than relying on carbonate-focused soil modification as a primary lever¹⁰. Strong spatial heterogeneity further implies that orchard-level diagnostics and site-specific adjustment may outperform uniform regional prescriptions, consistent with evidence that spatial variability in soil and leaf nutrient status affects fertilizer requirements and supports variable-rate management in olive orchards⁴¹. From a practical perspective, the results highlight the value of combining soil and leaf analyses when evaluating nutrient status, rather than relying on soil indicators alone in calcareous systems.

Interpretive boundaries should be kept explicit, as the present study is based on a two-year, observational design and the analyses are inherently correlative rather than mechanistic. The dataset cannot distinguish the degree to which year effects reflect climate variability versus management differences among orchards, and it does not directly measure rhizosphere processes needed for mechanistic attribution. Accordingly, mechanistic explanations remain interpretive rather than demonstrated. Furthermore, it is important to acknowledge that soil water availability acts as a major driver and constraint for plant nutrient uptake, particularly in rainfed semi-arid Mediterranean orchards. Therefore, integrating soil moisture dynamics with nutrient mobility should definitely be a key factor addressed in future research. Future work coupling repeated soil–leaf sampling with rhizosphere availability metrics and clearly documented management histories would strengthen causal inference and allow direct testing of whether targeted interventions (e.g., foliar programs, organic amendments, or bio/biochemical fertilization) can improve canopy micronutrient status and related performance outcomes⁴². In sum, the present results delineate a coherent, calcareous-soil–consistent pattern of temporal nutrient rebalancing in olive and identify the key measurements required to translate these patterns into actionable management strategies.

In conclusion, this two-year observational dataset revealed weak, element-specific, and year-dependent associations between soil nutrient pools and leaf nutrient composition in rainfed ‘Kilis Yaglik’ olive orchards under alkaline, calcareous conditions. These findings indicate that bulk soil measurements alone have limited capacity to describe canopy nutrient status in this system. Because the underlying processes were not directly examined, the mechanisms driving these patterns remain unresolved. Overall, the results support the use of combined soil and leaf assessments when interpreting nutrient status in comparable rainfed olive orchards on alkaline, calcareous soils.

Several limitations should be considered when interpreting these results. This study is based on a two-year observational dataset from a single region and cultivar, which restricts the generalizability of the findings. Orchard-level management practices (e.g., fertilization, soil disturbance) were not controlled and may have contributed to the observed variability. In addition, the analyses relied on bulk soil and leaf measurements and did not include rhizosphere-scale processes or direct assessments of nutrient uptake and internal redistribution; therefore, the results describe association patterns rather than underlying mechanisms.

The statistical framework is also subject to important limitations. Pearson correlation analysis captures only linear bivariate relationships and may not detect non-linear responses, threshold effects, or interactions among variables. Moreover, the large number of pairwise correlations increases the risk of false-positive associations; accordingly, correlation results should be interpreted as exploratory, particularly in the absence of multiple-testing correction. Finally, PCA was used solely as a descriptive tool to summarize multivariate structure and should not be interpreted as evidence of causal relationships or nutrient regulation mechanisms.

Materials and methods

Study area and plant material

The study was carried out during 2017–2018 in olive orchards located in Kilis province in south-eastern Türkiye. Sampling sites included Doğançay, Tahtalı, Duruca, Hıcıpoğlu, Süngütepe, Karaçavuş, Başmağara, Tamburalı, Çukuroaba, Demirciler, Meşetlik, and Akçabağlar villages. The experimental material consisted of 25 soil and 25 leaf samples collected from Olea europaea L. ssp. europaea var. sativa cv. ‘Kilis Yaglik’ trees grown under rainfed, semi-arid Mediterranean conditions. The study was conducted in olive orchards of the ‘Kilis Yaglik’ cultivar, which is one of the oldest and most widespread varieties in the region. The orchards consisted of old plantations with tree ages exceeding 65 years and a planting density of 8 × 8 m. The trees were grown under natural rainfed conditions and managed using traditional low-input practices, including basic soil tillage and pruning, without systematic fertilization. No fertilizer applications were carried out during the two-year study period, representing traditional, low-input and rainfed olive-growing systems of the region. Figure 6 illustrates the spatial distribution of the sampled olive orchards across the Kilis region. A total of 25 orchards were geographically mapped, covering representative microclimatic and edaphic zones across the study area. The sites are distributed between approximately 36.68°–36.82° N latitude and 36.95°–37.15° E longitude, reflecting the regional variability in soil characteristics and elevation. This distribution ensured comprehensive coverage of the major pistachio–olive production belt of the province and served as the spatial basis for subsequent soil and leaf sampling analyses.

Fig. 6.

Spatial distribution of the 25 sampled olive orchards across Kilis Province, Türkiye. The selected sites represent the microclimatic and edaphic variability of the semi-arid regional olive production belt. (Map data: Google Earth, Image © 2026 Airbus.).

The olive leaf and soil samples were collected from privately owned orchards, and permission was obtained from the orchard owners prior to sampling⁴³. Since the study involved a widely cultivated agricultural species (Olea europaea L.) and did not involve endangered or protected species, no specific national or international permits were required. To characterize the interannual climate variability, which directly impacts rainfed orchard dynamics, monthly precipitation (mm) and mean air temperature (°C) data for the 2017 and 2018 growing seasons were obtained from the Turkish State Meteorological Service (Kilis Station). Reflecting the climatic conditions driving the respective sampling periods, the total annual precipitation for the 2016–2017 production season (preceding the 2017 harvest) was 254.8 mm, while the 2017–2018 season (preceding the 2018 harvest) recorded 327.2 mm. The climate of Kilis is characteristic of the Mediterranean type, with hot, dry summers and mild, wet winters^44,45. The detailed interannual meteorological data are summarized in Supplementary Tables S5 and S6. Composite sampling was conducted by pooling subsamples from multiple trees and the number of replicates (n) per site is specified. The year (2017 or 2018) was tested as a factor in the statistical analyses.

Soil sampling and preparation

Composite soil samples were collected prior to any fertilization or irrigation events during the sampling period from the midpoint between two adjacent tree canopies at a depth of 0–30 cm, following a zig-zag pattern to represent each orchard. Each composite sample (~ 1 kg) was placed in labelled polyethylene bags, air-dried, sieved through a 2 mm mesh, and stored in sealed containers until analysis⁴⁶.

Leaf sampling and preparation

Leaf samples were collected each year from 8 to 10 trees per orchard to ensure representation of the entire site. Sampling was conducted in late summer, after the olive trees had completed fruit set, following standardized procedures for foliar nutrient analysis. To ensure representativeness, leaves were picked from all four quadrants of the tree canopy at about human height (approximately 1.5–2 m above ground), specifically from the middle portion of current-season (one-year-old) shoots⁴⁷. In the laboratory, the leaves were rinsed with distilled water, oven-dried at 40 °C, and ground to a fine powder prior to analysis.

Soil analyses

Soil pH was determined potentiometrically in a 1:2.5 soil-to-water suspension using a calibrated pH meter⁴⁸.

Calcium carbonate (CaCO₃) was measured using a Scheibler calcimeter and expressed as a percentage⁴⁹.

Organic matter (OM) was quantified by the modified Walkley–Black method⁵⁰ and expressed as percentage⁴⁸, following the classification of Thun et al.⁵¹.

Available phosphorus (P) was extracted with 0.5 M NaHCO₃ (pH 8.5) and determined calorimetrically using the ascorbic acid method^52,53.

Soil exchangeable potassium (K) content was determined using the ammonium acetate extraction method. Air-dried soil samples were sieved through a 2 mm mesh, and exchangeable K was extracted with 1 N NH₄OAc (pH 7.0). Potassium concentrations in the extracts were measured using a flame photometer (Jenway, Optima 7000 DV). Calibration was performed using standard potassium solutions, and the results were expressed as mg kg⁻¹ dry soil according to the method described by Thomas (1982)⁵⁴.

Available Fe, Mn, Cu, and Zn were extracted using the DTPA–CaCl₂–TEA method (pH 7.3) and quantified by inductively coupled plasma–optical emission spectrometry (ICP-OES)⁵⁵.

Leaf nutrient analyses

Total nitrogen (N) was determined by the macro-Kjeldahl digestion–distillation method⁵⁶. Phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg), iron (Fe), copper (Cu), zinc (Zn), and manganese (Mn) concentrations were determined after microwave-assisted digestion (HNO₃–H₂O₂–HClO₄, closed-vessel system) of leaf samples with concentrated HNO₃ + H₂O₂ + HClO₄, followed by ICP-OES measurement.

Calculation and statistical treatment of diagnostic nutrient ratios

Given the observational design (25 orchards across two consecutive years), all statistical analyses were conducted to describe associations and multivariate patterns rather than to infer causality or underlying regulatory mechanisms. Mass-based ratios were utilized to align with conventional agronomic reporting, whereas molar ratios were calculated to better represent the stoichiometric and physiological balance among competing ions (e.g., K, Ca, and Mg).

Leaf macronutrients (K, Ca, Mg) were expressed on a mass basis (%) and micronutrients (Fe, Mn) on a concentration basis (mg kg⁻¹). Diagnostic ratios were computed exclusively from leaf data as follows: K: (Ca + Mg) (mass-based), K: Ca: Mg (molar, normalized to K = 1), and Fe: Mn (both mass and molar).

Molar ratios were calculated using the atomic weights: K = 39.0983, Ca = 40.078, Mg = 24.305, Fe = 55.845 and Mn = 54.938. All ratios are unitless.

Soil data were used only to define the contextual subset (“context filter”) and were not included in the ratio calculations. The alkaline–calcareous subset was defined as pH > 7 and CaCO₃ ≥ 15%, following WRB⁵⁷ and NRCS⁵⁸ criteria.

In our dataset, soil pH values ranged from 7.2 to 8.1, with a mean of ≈ 7.8, reflecting the typical alkaline condition of calcareous soils described by WRB⁵⁷ and NRCS⁵⁸. Under such high-pH conditions, the solubility and plant availability of several micronutrients—particularly Fe, Zn, and Mn—are known to decrease markedly⁵⁹. Therefore, the diagnostic ratios derived in this study represent nutrient relationships characteristic of alkaline–calcareous environments.

For each ratio type, descriptive statistics (mean ± standard deviation and minimum–maximum range) were calculated separately for 2017 and 2018, as well as for the alkaline–calcareous subset. The pooled mean (“context average”) was obtained by averaging the 2017 and 2018 subset means and was used only as a descriptive summary of nutrient balance under the sampled alkaline–calcareous conditions.

All calculations were performed in Microsoft Excel and cross-checked in IBM SPSS Statistics v.24 (IBM Corp., Armonk, NY, USA).

Data processing and statistical analysis

All statistical analyses were carried out to evaluate the variability and interrelations of nutrient concentrations in soil and leaf matrices. Data were examined for normality using the Shapiro–Wilk test and for homogeneity of variance using Levene’s test. Because the same 25 orchards were sampled over two consecutive years, interannual differences were evaluated using a repeated measures analysis of variance (repeated measures ANOVA), where ‘year’ (2017 and 2018) was included as a within-subject fixed effect to properly account for the non-independence of the temporal data. When assumptions of normality or homoscedasticity were not satisfied, non-parametric tests were used. The factor year (2017 or 2018) was included as a fixed effect to assess interannual variability.

Pearson correlation coefficients (r) were calculated to identify bivariate associations among nutrients within and between soil and leaf datasets. All variables were standardized to zero mean and unit variance prior to multivariate analysis. Principal component analysis (PCA) was performed separately for soil and leaf datasets as an exploratory descriptive tool to reduce dimensionality and summarize covariance patterns among variables. PCA was not used to infer causal mechanisms or physiological regulation. Components with eigenvalues greater than 1.0 were retained following the Kaiser criterion. Loadings with absolute correlation coefficients (|r|) ≥ 0.45 were considered significant contributors to component formation. The proportion and cumulative percentage of variance explained by each component were reported. Complete loading matrices and eigenvalue statistics are provided in the supplementary material. Statistical analyses were conducted using IBM SPSS Statistics v.24 (IBM Corp., Armonk, NY, USA).

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

AK and HC conceptualized the study. HC developed the methodology. AK conducted the formal analysis and curated the data. ST and HG prepared the visualizations. AK wrote the original draft. HC, ST, and HG reviewed and edited the manuscript. HC supervised the study.

Funding

The authors did not receive support from any organization for the submitted work.

Data availability

All data generated or analysed during this study are included in this published article and its supplementary information files.

Declarations

Competing interests

The authors declare no competing interests.