Environmental, Genetic and Structural Interactions Affecting Phytophthora spp. in Citrus: Insights from Mixed Modelling and Mediation Analysis to Support Agroecological Practices

Abstract

This study investigates the complex interactions between environmental, genetic, and structural factors that influence two key parameters: the density of Phytophthora spp. propagules per gram of dry soil (NPSS) and the number of colonies (NC). Using advanced statistical approaches, we examined the combined effects of variables such as soil moisture, dry weight, temporal fluctuations, and rootstocks. The results show a significant linear relationship between NPSS and soil moisture, as well as a strong positive correlation between NPSS and NC. Genetic analyses reveal a predominant contribution of environmental factors to trait variability, with high phenotypic coefficient of variation (PCV) and low broad-sense heritability. Mixed models highlight the synergistic impact of soil moisture, NC, and dry soil weight on NPSS, as well as significant temporal effects. Mediation analysis confirms that soil moisture influences NPSS primarily through an indirect effect transmitted by NC, with a mediated proportion exceeding 94%. Finally, multivariate analysis reveals significant differences between rootstocks, with Citrus Volkameriana B2 28613 (R4) and Mandarin Sunki x P.T. B2 38581 (R7) standing out as the most performant. These results highlight the importance of an integrated management of environmental variables and rootstocks to optimize soil productivity and agronomic quality. The implications of this study provide a solid foundation for guiding genetic improvement and soil management strategies, balancing environmental constraints and the opportunities offered by targeted genetic selection.

1. Introduction

Citrus is a species of the Rutaceae family and is one of the most cultivated and economically important fruit trees in the world [1]. They occupy a central place in agriculture, not only because of their high production levels, but also because of their role in international trade and their contribution to food security [2,3]. Citrus fruits are rich in vitamins, minerals and bioactive compounds. They are highly valued for their nutritional value and their many uses in the food, pharmaceutical and cosmetic industries [4,5]. The cultivation of Citrus fruits in Morocco covers some 128,000 hectares, and production totalled 2.4 million tons. These include 766,500 tonnes destined for export, generating added value of over 5.7 billion dirhams Moroccan equivalent to approximately 546,561,215 euros [6].

Citrus production relies on effective management of interactions between environmental factors, plant genetic characteristics, and agronomic practices [7,8]. Indeed, grafting is frequently used to improve yields [9], in this context, rootstocks play a crucial role in optimizing yields as they directly influence key parameters such as resistance to abiotic stress, soil health, and nutrient uptake because rootstocks impact various citrus characteristics [10,11]. The use of adapted rootstocks is particularly important in agricultural systems subject to environmental constraints, where factors such as soil moisture, microbial colony density, and soil physical quality significantly influence crop productivity [12,13,14,15,16].

Phytophthora spp. causes severe diseases in citrus, including root rot, seedling damping-off, and gummosis, impacting tree health and productivity [17]. Among them, P. nicotianae is the most widespread in subtropical regions [18]. Rootstocks, being in direct contact with the soil, play a crucial role in tree resilience against Phytophthora infections [19]. In Morocco, studies have evaluated the inoculum density of Phytophthora spp. in citrus orchards, but no damage threshold has been defined [16], unlike in Florida where treatments are recommended from 5 to 15 propagules/cm³ [20,21]. This study is part of the ongoing research on the presence of Phytophthora citrophthora and P. parasitica in the Gharb region, further exploring the environmental factors that promote their development and distribution.

The number of propagules per gram of dry soil (NPPS) and the number of colonies (NC) are two fundamental indicators for assessing soil health and quality [15,16,21,22]. NPPS reflects the dynamics of fungal and microbial populations, while NC is a key indicator of soil biological activity, directly influencing nutrient availability for crops [23,24]. However, the variability of these traits is influenced by genetic and environmental factors, requiring a deep understanding of the underlying mechanisms to guide agronomic practices and genetic selection strategies [25].

Previous studies have shown that rootstocks can significantly influence these traits by modulating the interaction between plants and their environment [26,27]. However, the exact mechanisms by which these interactions occur, particularly through mediating factors such as colony density, remain poorly understood [28]. Moreover, the impact of seasonal environmental variations and long-term dynamics on these traits remains poorly understood, despite their importance for optimizing soil management [29].

In this context, the use of advanced statistical models, such as linear mixed models and mediation analyses, provides powerful tools to disentangle the combined effects of environmental, genetic, and structural factors on the studied traits [30]. These approaches allow quantifying the relative contributions of direct and indirect effects, while integrating temporal and spatial variations, to provide evidence-based recommendations for improving agricultural systems [31].

This study aims to explore the complex interactions between environmental, genetic, and structural factors influencing the number of propagules per gram of dry soil (NPSS) and the number of colonies (NC) of Phytophthora spp. in the Gharb region of Morocco. The specific objectives are as follows:

• Describe the fundamental relationships between environmental variables (humidity, soil dry weight, colony density) and NPSS, as well as their temporal fluctuations.
• Quantify the combined effects of temporal, environmental, and structural factors using linear mixed models, while examining key interactions and random effects associated with rootstocks.
• Identify the mediation mechanisms between humidity, colony number, and NPSS to better understand indirect relationships and their relative importance.
• Evaluate the genetic and phenotypic variability of the studied traits to determine the relative contributions of environmental and genetic factors.
• Compare the performance of rootstocks in terms of NPSS and NC to identify the most suitable rootstocks for optimal soil management and sustainable agronomic improvement.

2. Materials and Methods

2.1. Plant Material and Experimental Conditions

In a plot of Valencia late grafted onto 15 rootstocks aged 17 years (

Table 1. List of citrus rootstocks with their main characteristics (origin, tolerance, and compatibility).

Code	Rootstocks	Code ICVN
3	Poncirus trifoliata. B6 CZ 24	ICVN 0110139
6	Mandarin Sunki x P.T. B2 38581	ICVN 0110204
5	P.T B 6 C Z 13	ICVN 0110107
7	Citrange Carrizo 28608	ICVN 0110181
11	Citrumelo 4475 B2 G3	ICVN 110145
16	Mandarin sunki x P.T. 30591	ICVN 0110211
17	Mandarin sunki x P.T. 30588	ICVN 0110208
18	Mandarin cleopatra x P.T. 30584	ICVN 0110155
23	Gou-Tou SRA 506	-
24	Citrus macrophylla	ICVN 0110058
25	Citrus volkameriana B2 28613	ICVN 0110025
30	Mandarin Cleopatra X C.C. 30577	ICVN 0110223
34	Bigaradier P6 R26 A16	-
39	Mandarin sunki x P.T. 330590	ICVN 0110210
41	Citrumelo 1452 B6 C	ICVN 0110282

), three trees were randomly selected for each rootstock. Four soil sbsamples were collected following the cardinal directions. The sampling was performed using a corer one meter from the trunk and at a depth of 5 to 20 cm [32,33].

Soil samples were sieved separately through a 2 mm mesh sieve and stored at an ambient temperature of 21–24 °C [32,34,35]. To estimate the inoculum density of Phytophthora spp. in the soil samples, we used the dilution technique. Each subsample, representing a given geographical orientation, 10 g of soil was diluted in 90 mL of 0.25% water agar. After agitation for 20 min, one ml was spread onto a Petri dish containing BARPHY 72 culture medium [36,37]. This is a selective culture medium for Phytophthora spp. Petri dishes were incubated in the dark at 28 °C for 48 h [32,35]. To count Phytophthora spp. propagules, the plates were washed with sterile distilled water to remove soil particles, and then the colonies were counted. These were then transferred separately into test tubes containing cornmeal agar (CMA) medium. Knowing the amount of soil inoculated in each plate, an approximate value of the number of propagules per gram of dry soil of Phytophthora was calculated.

Number of Colonies (NC): This measurement is obtained directly by counting the visible colonies that develop on the culture medium after inoculation using the plate dilution method [38,39].

Number of Propagules per Gram of Dry Soil (NPSS): We have clarified that this estimate is based on the NC, taking into account the dilution factor and the dry weight of the soil. Indeed, only propagules capable of forming viable colonies on the culture medium are considered.

Identification was based on the taxonomic criteria of [39,40] and on the morphological characteristics of the colonies, mycelial features, culture medium, presence or absence of chlamydospores, morphology of mycelial filaments, and sporangia dimensions, referring to various identification keys, such as that of [41]. The colonies that developed from the samples collected in this study all appear to be characteristic of P. parasitica and P. citrophthora, with a clear dominance of P. citrophthora populations over those of P. parasitica. No other Phytophthora species were observed in the analyzed samples [42].

Soil humidity was determined gravimetrically for each sample by taking 10 g of soil and placing it in an oven for 48 h at 75 °C [43]. Once dried, the dry weight (Pm) of the tested sample was measured, and the moisture content (expressed as a percentage) was calculated using the following equation:

H (%) = [(10 − Pm)/10] × 100

2.2. Statistical and Genetic Analysis

All statistical analyses were performed using Stata 18 software (StataCorp, College Station, TX, USA), which enabled regression modeling, multivariate analysis (MANOVA), mediation analysis, and linear mixed models. Graphics were generated using Python 3.10, providing additional flexibility to visualize the relationships between the studied variables with scatter plots, trend lines, and appropriate confidence intervals. These tools allowed for rigorous analysis and clear visualization of the results, in accordance with scientific standards.

2.2.1. Genetic Analysis

The variability of traits was evaluated using genotypic variances (GV) and coefficients of variation, according to the method described by [40]. The genetic coefficient of variation (GCV) was calculated using the formula:

$$GCV(\%)={\displaystyle \frac{\sqrt{GV}}{\overline{X}}}\times 100$$

where represents the overall mean of the trait. Similarly, the phenotypic coefficient of variation (PCV) was estimated according to:

$$PCV(\%)={\displaystyle \frac{\sqrt{PV}}{\overline{X}}}\times 100$$

where VP corresponds to the total phenotypic variance. Broad-sense heritability (H²) was determined as the ratio of genotypic variance to phenotypic variance:

$$H^2={\displaystyle \frac{GV}{PV}}$$

As proposed by [40]. The average genetic gain (AGG) expressed as a percentage of the mean was calculated following the method described by [44]:

$$AGG(\%)={\displaystyle \frac{K\times \sqrt{GV}\times H^2}{\overline{X}}}\times 100$$

where K is the selection differential (K = 2.06 for a selection intensity of 5%). The genotypic correlations (Rg) and phenotypic correlations (Rp) between traits were determined using the following formulas:

$$Rg={\displaystyle \frac{Cov(G_1,G_2)}{\sqrt{V{G}_1.V{G}_2}}}\mathrm{and}Rp={\displaystyle \frac{Cov(P_1,P_2)}{\sqrt{V{P}_1.V{P}_2}}}$$

where Cov(G1, G2) and Cov(P1, P2) represent the genotypic and phenotypic covariances between two traits, respectively, and VG1, VG2, and VP1, VP2 their associated variances. Traits with high genotypic correlations were identified as priorities for improvement, as they reflect intrinsic relationships that are less influenced by environmental variability. Finally, the classification of heritability followed the criteria proposed by [45], with thresholds defined as follows: low (H² < 40%), moderate (40% ≤ H² < 60%), high (60% ≤ H² < 80%) and very high (H² ≥ 80%).

2.2.2. Linear Mixed Model

In this study, a linear mixed model was used to analyse the factors influencing the number of propagules per gram of dry soil (NPSS). The choice of this approach is justified by the hierarchical nature of the data, where observations are grouped by rootstocks (PG). The model allows to account for the inter-group variability via a random effect associated with PGs, while evaluating the contributions of fixed factors such as year, month, humidity, number of colonies and dry soil weight (PSS). The linear mixed model can be represented by the following equation:

$$Y_{ij}={\beta }_0+{\beta }_1{X}_{1ij}+{\beta }_2{X}_{2ij}+\dots +{\beta }_p{X}_{pij}+b_{0i}+{ϵ}_{ij}$$

where Y_ij represents the NPSS for the i-th unit (rootstock) at the j-th measurement, Xpij denotes the explanatory variables associated with the fixed effects, β_p their estimated coefficients, b0i is the random effect specific to the i-th rootstock, assumed to be distributed according to $b_{0i}\sim N(0,{\sigma }_b^2)$ and ${ϵ}_{ij}\sim N(0,{\sigma }^2)$ represents the residual error.

In this context, fixed effects include year and month to capture temporal effects, as well as a three-way interaction between humidity, number of colonies, and PSS, to model the complex relationships between these environmental and agronomic variables. The random effects of PG allow to model the variations specific to the groups and to capture the unobserved characteristics specific to each rootstock. Model parameters were estimated by restricted maximum likelihood (REML), and the significance of fixed effects was assessed by p-values associated with z statistics. The validity of the model was confirmed by checking the assumptions of normality and independence of residuals, as well as by comparing it with a simple linear model using a likelihood ratio test (chibar2).

2.2.3. Causal Mediation Analysis

A mediation analysis was conducted to explore the effect of humidity (%H) on the number of propagules per gram of dry soil (NPSS) through the mediator, the number of colonies. The model employed is based on a decomposition of effects according to the following relationships (Figure 1): humidity directly influences the mediator (number of colonies) as represented by, where M is the mediator, X is humidity, a is the coefficient describing the effect of humidity on the number of colonies, and εM is the error term. The outcome (NPSS) is modeled by, where Y represents the NPSS, c′ is the direct effect of humidity on NPSS, b is the effect of the mediator (number of colonies) on NPSS, and εY is the error term. This approach allows for the decomposition of the total effect (TE) of humidity on NPSS into a natural indirect effect, representing the impact transmitted through the mediator, and a natural direct effect, measuring the direct influence independent of the mediator. Comparisons were made between five humidity classes defined as percentages (4.4% < H ≤ 8.4%, 8.4% < H ≤ 12.3%, 12.3% < H ≤ 16.2%, 16.2% < H ≤ 20.2%, and 20.2% < H ≤ 24.1%) relative to a reference class (0.5% ≤ H ≤ 4.4%). Effect estimates were obtained using a linear model fitted with robust standard errors, and statistical significance was assessed at a p < 0.05 level. For each humidity category, the proportion of mediation was calculated to quantify the portion of the total effect attributable to the indirect effect. This proportion was obtained using the following formula:

$$MediatedProportion\left(\%\right)={\displaystyle \frac{IndirectEffect(a\times b)}{TotalEffect\left(c\right)}}\times 100$$

2.2.4. Multivariate Analysis of Variance (MANOVA) of Rootstocks

To analyse the overall effect of different rootstocks on the dependent variables (number of propagules per gram of dry soil and number of colonies), a multivariate analysis of variance (MANOVA) was conducted. Multivariate tests were performed using Wilks’ lambda, Pillai’s trace, Lawley-Hotelling trace, and Roy’s largest root statistics to assess the overall significance of differences between rootstocks. Post-hoc analyses included adjusted predictions (margins) to compare the specific performance of rootstocks with 95% confidence intervals and p-values to test statistical significance.

Adjusted predictions for the number of propagules per gram of dry soil (NPSS) and the number of colonies (NC) were obtained using marginal linear predictions following a multivariate analysis of variance (MANOVA). The model accounted for the fixed effects of rootstocks. Post-hoc pairwise comparisons were conducted using Tukey’s test to assess significant differences between the adjusted predictions of each rootstock. The adjusted predictions were reported along with 95% confidence intervals and p-values for each comparison. Results were presented in tables and figures to illustrate the variation in NPSS and NC among different rootstocks.

3. Results

The Figure 2 shows a linear increase in the number of propagules per gram of dry soil (NPSS) as a function of humidity (%) with a clear positive relationship, while NPSS decreases linearly with increasing soil dry weight. A strongly positive linear relationship is observed between NPSS and the number of colonies, indicating a progressive increase in NPSS with increasing colony density. Regarding temporal evolution, NPSS exhibits marked fluctuations with peaks around April 2013 and March 2014 and troughs around September 2013, revealing a variable dynamic depending on the date.

The Figure 3 presents a distribution of means and standard deviations for the number of colonies and the number of propagules per gram of dry soil according to the different rootstocks. For Citrus volkameriana B2 28613, the mean number of colonies is high with a large variability, a trend that is also reflected in the number of propagules per gram of dry soil. In contrast, Bigaradier P6 R26 A16 presents a low mean and a reduced variability for both colonies and propagules, reflecting significant stability. Other rootstocks display varied profiles, with different means and standard deviations for each case, revealing specific behaviors depending on the rootstock. This parallel reading highlights the similarities and differences in the performance of rootstocks in terms of colonies and propagules (Figure 3).

3.1. Genetic and Phenotypic Variability

The Figure 4 presents the genetic and phenotypic metrics calculated for the number of colonies and the number of propagules per gram of dry soil, traits influenced by rootstocks. The genetic coefficient of variation (GCV) and phenotypic coefficient of variation (PCV) show a major contribution of environmental factors to the variability of both traits, with high PCV values (87.7% for colonies and 91.6% for propagules) significantly exceeding GCV (43.2% and 45.3%, respectively). Broad-sense heritability (H²) is low for both traits (24.3% for colonies and 24.4% for propagules), indicating a predominant environmental influence on their expression. The average genetic gains (AGG) are moderate (21.6% and 22.8%), suggesting that genetic improvement through selection is possible, although limited by environmental constraints. Moreover, the extremely high genotypic (Rg = 0.9994) and phenotypic (Rp = 0.9981) correlations between the number of colonies and the number of propagules indicate a strong intrinsic relationship, mainly determined by genetic factors, with a low genotype-environment interaction. This relationship suggests that targeted improvement of one trait will lead to a concomitant improvement of the other, thus simplifying genetic improvement strategies. These results highlight the importance of combining moderate genetic selection to exploit the strong genetic relationship between traits, while optimizing environmental conditions to reduce constraints related to phenotypic expression.

3.2. Mixed Linear Model

A mixed linear model was used to analyze the number of propagules per gram of dry soil (NPSS) as a function of year, month, the interaction between humidity, number of colonies, and soil dry weight (SDW), with rootstocks (RS) included as a random effect (

Table 2. Fixed Effects on the Number of Propagules per Gram of Dry Soil (NPSS).

Number of Propagules per Gram of Dry Soil	Coefficient	Std. Err.	z	p-Value	95% CI
Month	4.10	1.43	2.86	0.004	[1.29; 6.92]
Year	38.86	7.66	5.07	0.000	[23.83; 5388]
Hum × Nc × PSS	0.70	0.007	91.80	0.000	[0.69; 0.72]
Constante	78.21622	15.43885	−5.07	0.000	[−108.4758; −47.95663]

and

Table 3. Random Effects Parameters and Overall Model Fit Statistics.

Random-Effects Parameters	Estimate	Std. Err.	95% CI
Variance of rootstocks	63.94	52.98	[12.61; 324.34]
Residual variance	1418.67	125.69	[1192.52; 1687.70]
Parameter
Log-vraisemblance	1367.33	-	-
Wald chi2 (3)	9393.57	-	p < 0.0001
LR test (chibar2(01))	2.93	-	p = 0.0433

).

The analysis using a mixed model highlighted several key relationships influencing the number of propagules per gram of dry soil (NPSS). The temporal effect of year was significant (p < 0.001), with an average increase of 38.86 NPSS units per year, likely reflecting improvements in agronomic practices or long-term environmental conditions. Moreover, the seasonal effect of month, although less important in previous models, is now significant (p = 0.004), indicating variability linked to monthly fluctuations (e.g., temperature, precipitation).

The three-way interaction between humidity, number of colonies, and soil dry weight (SDW) is a key factor in this model (p < 0.001). This interaction reveals that the combined effect of these three variables is synergistic, where an increase in humidity and the number of colonies amplifies the impact of soil dry weight on NPSS.

These results highlight the need for integrated management of environmental and agronomic variables to optimize propagule production.

The random effects associated with rootstocks (var(cons) = 63.94) indicate a moderate contribution to the total variability of NPSS. The comparison test between the mixed model and the simple linear model (chi-squared = 2.93, p = 0.0433) shows that adding random effects significantly improves the model. This underscores the importance of including rootstocks in the modeling, although their influence is less pronounced than that of environmental and agronomic interactions (Figure 5).

Below is a visualization of the three-way interaction between humidity, the number of colonies, and soil dry weight (SDW), illustrating their combined effect on NPSS. The surface plot shows how humidity and the number of colonies interact to amplify the effect of soil dry weight on propagule production.

3.3. Causal Mediation Analysis

The

Table 4. Direct, indirect, and total effects of humidity on the number of propagules per gram of dry soil via the number of colonies (causal mediation model).

Number of Propagules per Gram of Dry Soil		Robust
	Coefficient	Std. Err.	Z	p-Value	95% CI
Natural indirect effect
Humidity (H) %
4.4 < H ≤ 8.4 vs. 0.5 ≤ H ≤4.4	27.17	11.44	2.37	p < 0.05	4.72	49.61
8.4 < H ≤ 12.3 vs. 0.5 ≤ H ≤4.4	169.85	21.04	8.07	p < 0.001	128.60	211.11
12.3 < H ≤ 16.2 vs. 0.5 ≤ H ≤4.4	261.77	21.68	12.07	p < 0.001	219.27	304.27
16.2 < H ≤ 20.2 vs. 0.5 ≤ H ≤4.4	420.91	42.65	9.87	p < 0.001	337.31	504.52
20.2 < H ≤ 24.1 vs. 0.5 ≤ H ≤4.4	453.39	128.10	3.54	p < 0.001	202.30	704.48
Natural direct effect
Humidity (%H)
4.4 < H ≤ 8.4 vs. 0.5 ≤ H ≤4.4	1.57	0.32	4.80	p < 0.001	0.93	2.22
8.4 < H ≤ 12.3 vs. 0.5 ≤ H ≤4.4	2.65	0.72	3.68	p < 0.001	1.24	4.07
12.3 < H ≤ 16.2 vs. 0.5 ≤ H ≤4.4	4.54	1.09	4.15	p < 0.001	2.39	6.70
16.2 < H ≤ 20.2 vs. 0.5 ≤ H ≤4.4	1.69	2.08	0.81	0.417	−2.39	5.78
20.2 < H ≤ 24.1 vs. 0.5 ≤ H ≤4.4	5.21	3.47	1.50	0.133	−1.59	12.02
Total effect
Humidity (%H)
4.4 < H ≤8.4 vs. 0.5 ≤ H ≤4.4	28.74	11.26	2.55	p < 0.05	6.66	50.83
8.4 < H ≤12.3 vs. 0.5 ≤ H ≤4.4	172.51	20.70	8.33	p < 0.001	131.92	213.10
12.3 < H ≤ 16.2 vs. 0.5 ≤ H ≤4.4	266.32	21.24	12.54	p < 0.001	224.69	307.95
16.2 < H ≤ 20.2 vs. 0.5 ≤ H ≤4.4	422.61	41.33	10.22	p < 0.001	341.60	503.62
20.2 < H ≤ 24.1 vs. 0.5 ≤ H ≤4.4	458.61	126.73	3.62	p < 0.001	210.21	707.00
Proportion d’humidity (%)	Proportion mediated (%)
4.4 < H ≤8.4 vs. 0.5 ≤ H ≤4.4	94.5%
8.4 < H ≤12.3 vs. 0.5 ≤ H ≤4.4	98.5%
12.3 < H ≤ 16.2 vs. 0.5 ≤ H ≤4.4	98.3%
16.2 < H ≤ 20.2 vs. 0.5 ≤ H ≤4.4	99.6%
20.2 < H ≤ 24.1 vs. 0.5 ≤ H ≤4.4	98.9%

H: Humidity(%).

illustrates the direct, indirect, and total effects of humidity (%H) on the number of propagules per gram of dry soil (NPSS), with comparisons between different humidity classes expressed as a percentage relative to the reference class 0.5% ≤ H ≤ 4.4%. For the indirect effect (NIE), which captures the impact of humidity on NPSS via the mediator number of colonies, the coefficients increase significantly as humidity classes increase. For example, for 4.4% < H ≤ 8.4% vs. 0.5% ≤ H ≤ 4.4%, the indirect effect is estimated at 27.17 units (p < 0.05), while it reaches 453.39 units (p < 0.001) for 20.2% < H ≤ 24.1% vs. 0.5% ≤ H ≤ 4.4%, reflecting a strong positive relationship between humidity and the increase in NPSS via the mediator number of colonies. Regarding the direct effect (NDE), which measures the influence of humidity on NPSS without going through the mediator, significant coefficients are observed for the first three humidity classes: 1.57 units for 4.4% < H ≤ 8.4% vs. 0.5% ≤ H ≤ 4.4%, 2.65 units for 8.4% < H ≤ 12.3% vs. 0.5% ≤ H ≤ 4.4%, and 4.54 units for 12.3% < H ≤ 16.2% vs. 0.5% ≤ H ≤ 4.4% (p < 0.001).

However, for the higher classes (16.2% < H ≤ 20.2% and 20.2% < H ≤ 24.1%), the direct effect becomes non-significant (p > 0.05), suggesting that the overall impact of humidity on NPSS in these classes is primarily transmitted through the mediator number of colonies. Regarding the total effect (TE), which combines direct and indirect effects, a marked increase is observed in the higher humidity classes: 28.74 units for 4.4% < H ≤ 8.4% vs. 0.5% ≤ H ≤ 4.4%, and 458.61 units for 20.2% < H ≤ 24.1% vs. 0.5% ≤ H ≤ 4.4% (p < 0.001), highlighting a strong overall relationship between humidity and NPSS, dominated by indirect effects via the mediator number of colonies in the higher humidity classes. These results confirm the importance of the number of colonies as a key intermediate factor in the relationship between humidity (%H) and the number of propagules per gram of dry soil (NPSS).

For all humidity classes, the proportion mediated remains remarkably high, ranging from 94.5% to 99.6%, indicating that nearly the entire total effect of humidity on NPSS is transmitted through the number of colonies. This highlights the essential role of colonies as an intermediary in this relationship. The slight variation in the proportion mediated between humidity classes suggests that as humidity increases, the indirect pathway becomes even more predominant.

3.4. Multivariate Analysis of Rootstocks (MANOVA)

To assess the overall effect of different rootstocks on the dependent variables, a multivariate analysis of variance (MANOVA) was performed. Multivariate tests revealed a significant effect of rootstocks on the combined variables (number of propagules per gram of dry soil and number of colonies). Results indicate that the overall effect of rootstocks is highly significant, with the following statistics:Wilks’ lambda: λ = 0.7201, F(14, 28) = 508.0, p < 0.0001 (exact); Pillai’s trace: V = 0.2951, F(28, 510) = 3.15, p < 0.0001 (approximatif); Lawley-Hotelling trace: T = 0.3678, F(28, 506) = 3.32, p < 0.0001 (approximatif); Roy’s largest root: Θ = 0.2970, F(14, 255) = 5.41, p < 0.0001 (upper bound). These results indicate that the different rootstocks have a statistically significant effect on the two combined dependent variables, suggesting that the choice of rootstock significantly influences the studied characteristics (

Table 5. Multivariate Analysis of Variance (MANOVA) Test Statistics for the Effect of Rootstock on Number of Propagules per Gram of Dry Soil (NPDS) and Number of Colonies (NC).

Source	Statistic	df1	Df2	F	p-Value
Wilks’ Lambda	0.7201	14	28	508	p < 0.0001
Pillai’s Trace	0.2951	28	510	3.15	p < 0.0001
Lawley-Hotelling	0.3678	28	506	3.12	p < 0.0001
Roy’s Largest Root	0.2970	14	255	5.41	p < 0.0001

). A post hoc analysis was then conducted to examine the effect of individual rootstocks on the number of propagules per gram of dry soil (NPSS) and the number of colonies (NC). The results show significant variations between rootstocks. Regarding NPSS, rootstocks exhibit marked differences. Rootstock R4 (“Citrus volkameriana B2 28613”) stands out with the highest NPSS value (601.53 ± 50.66, p < 0.001), followed by R7 (“Mandarin Sunki x P.T. B2 38581”) (408.08 ± 50.66, p < 0.001). These results reflect a significant performance for these rootstocks in optimizing NPSS. Conversely, rootstock R10 (“Bigaradier P6 R26 A16”) shows the lowest value (100.44 ± 50.66, p = 0.048). For the number of colonies (NC), a significant variability is also observed between rootstocks. R4 (“Citrus volkameriana B2 28613”) again shows a superior performance with a maximum NC value (55.22 ± 4.58, p < 0.001), followed by R7 (“Mandarin Sunki x P.T. B2 38581”) (37.61 ± 4.58, p < 0.001) (Figure 6). Rootstocks R10 (“Bigaradier P6 R26 A16”) and R13 (“Citrus macrophylla”) showed the lowest values for NC, with 9.72 ± 4.58 (p = 0.035) and 14.22 ± 4.58 (p = 0.002), respectively. Overall, the results confirm that R4 (“Citrus volkameriana B2 28613”) is the best performing rootstock for both parameters, followed by R7 (“Mandarin Sunki x P.T. B2 38581”), while R10 (“Bigaradier P6 R26 A16”) is the least performing.(

Table 6. Adjusted Predictions for Number of Propagules per Gram of Dry Soil (NPSS) and Number of Colonies (NC) by Rootstocks.

Rootstock	Adjusted Prediction	T	p-Value	95% CI
Bigaradier P6 R26 A16	100.44	1.98	p < 0.05	0.67	200.20
Number of propagules per gram of dry soil (NPDS)
Citrange Carrizo 28608	276.42	5.46	p < 0.001	176.66	376.19
Citrumelo 1452 B6 C	238.69	4.71	p < 0.001	138.93	338.46
Citrumelo 4475 B2 G3	172.34	3.40	p < 0.01	72.57	272.10
Citrus macrophylla	151.64	2.99	p < 0.01	51.88	251.41
Citrus volkameriana B2 28613	601.52	11.87	p < 0.001	501.76	701.29
Gou-Tou SRA 506	232.23	4.58	p < 0.001	132.47	332.00
Mandarin Cleopatre X C.C. 30577	216.68	4.28	p < 0.001	116.92	316.45
Mandarin Cleopatre x P.T. 30584	291.85	5.76	p < 0.001	192.08	391.61
Mandarin Sunki x P.T. 30588	218.69	4.32	p < 0.001	118.92	318.45
Mandarin Sunki x P.T. 30591	220.31	4.35	p < 0.001	120.54	320.08
1#Mandarin Sunki x P.T. 330590	272.34	5.38	p < 0.001	172.57	372.11
Mandarin Sunki x P.T. B2 38581	408.07	8.06	p < 0.001	308.31	507.84
P.T B 6 C Z 13	287.88	5.68	p < 0.001	188.11	387.64
Poncirus trifoliata. B6 CZ 24	214.20	4.23	p < 0.001	114.43	313.96
Number of colonies (NC)
Bigaradier P6 R26 A16	9.72	2.12	p < 0.05	0.70	18.73
Citrange Carrizo 28608	26.16	5.71	p < 0.001	17.14	35.18
Citrumelo 1452 B6 C	22.38	4.89	p < 0.001	13.37	31.40
Citrumelo 4475 B2 G3	16.61	3.63	p < 0.001	7.59	25.62
Citrus macrophylla	14.22	3.11	p < 0.01	5.20	23.23
Citrus volkameriana B2 28613	55.22	12.06	p < 0.001	46.20	64.23
Gou-Tou SRA 506	22.05	4.82	p < 0.001	13.03	31.07
Mandarin Cleopatre X C.C. 30577	21.27	4.65	p < 0.001	12.26	30.29
Mandarin Cleopatre x P.T. 30584	27.22	5.95	p < 0.001	18.20	36.23
Mandarin Sunki x P.T. 30588	20.88	4.56	p < 0.001	11.87	29.90
Mandarin Sunki x P.T. 30591	21.44	4.68	p < 0.001	12.42	30.46
Mandarin Sunki x P.T. 330590	25.38	5.54	p < 0.001	16.37	34.40
Mandarin Sunki x P.T. B2 38581	37.61	8.21	p < 0.001	28.59	46.62
P.T B 6 C Z 13	26.77	5.85	p < 0.001	17.76	35.79
Poncirus trifoliata. B6 CZ 24	20.94	4.57	p < 0.001	11.92	29.96

) These differential performances highlight the importance of rootstock selection based on specific agronomic objectives. These results emphasize the relevance of a thorough analysis of rootstocks to optimize soil productivity and quality and confirm the significant impact of rootstocks on the studied parameters (Figure 6, Figure 7 and Figure 8).

4. Discussion

The results obtained reveal the importance of environmental, genetic, and structural factors in regulating the number of Phytophthora spp. propagules per gram of dry soil (NPDS) and the number of colonies (NC), thus providing a comprehensive understanding of the complex interactions between these variables. Initial descriptive analyses reveal significant linear relationships between NPDS and environmental variables such as humidity, dry soil weight (DSW), and number of colonies, as well as a marked temporal dynamic [15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46]. These trends underline the critical influence of environmental conditions on the expression of these traits, consistent with observations reported by [16], who documented similar variations linked to seasonal fluctuations and weather conditions.

The analysis of genetic and phenotypic metrics revealed a predominant contribution of environmental factors to the variability of the traits studied, as evidenced by high phenotypic coefficient of variation (PCV) values compared to genetic coefficient of variation (GCV) values. These results agree with the work of [47], who reported that traits influenced by genotype-environment interactions often have low broad-sense heritability (H²). However, the high genotypic and phenotypic correlations between NPSS and NC suggest a strong intrinsic relationship, indicating that breeding strategies targeting one trait could lead to concomitant improvements in the other [48]. These results support the hypothesis that a moderate genetic selection, combined with optimal environmental practices, could maximize the productivity of the studied systems.

A linear mixed model analysis clarified the combined impact of temporal, environmental, and structural variables on NPSS. The significant temporal effect reflects potential improvements in agricultural practices or long-term environmental changes, corroborating the results of [49], who identified a similar trend in comparable environments. Moreover, the synergistic effect between humidity, number of colonies, and PSS highlights the need for an integrated management of agronomic variables to minimize propagule production [50]. These results are reinforced by the random effects of rootstocks, although their contribution is moderated compared to environmental variables, as observed in similar studies [51,52].

Mediation analysis revealed that humidity mainly influences NPSS through an indirect effect via the number of colonies, with high proportions mediated (>94%), suggesting that the mediator plays a crucial role in this relationship. These results confirm the observations of [53,54], who demonstrated the importance of interactions mediated by biological factors in complex agricultural systems. This predominance of indirect effects highlights the impact of the number of colonies as a key factor in optimizing NPSS, validating the approach of targeting this mediator in agricultural management practices. Various factors can influence the density and spread of Phytophthora inoculum in the soil. Flood irrigation creates more favorable conditions for infection than localized irrigation. Since our orchard employs this irrigation method, the rootstocks may be at a higher risk of severe root infections, potentially leading to an increased density of Phytophthora in the soil [55].

Finally, the results of the multivariate analysis confirm that rootstocks significantly influence NPSS and NC, with important variations between rootstocks. Citrus volkameriana B2 28613 (R4) and Mandarin Sunki x P.T. B2 38581 (R7) stood out as the best performers, corroborating the conclusions of [13,16,56,57], who highlighted the superior performance of certain rootstocks in similar environmental conditions. In contrast, the weaker performance of Bigaradier P6 R26 A16 (R10) underlines the need for a careful choice of rootstocks based on specific agronomic objectives [36,58]. These results reinforce the idea that rootstocks play a strategic role in soil management and maximizing agricultural productivity [59].

The selection of rootstock plays a crucial role in determining the presence and proliferation of Phytophthora in the soil. Resistant rootstocks, such as Troyer and Macrophylla, suppress fungal development by producing elevated levels of phenolic compounds, which possess antifungal properties [60,61,62]. In contrast, Citrus volkameriana, despite its ability to synthesize these compounds, facilitates the expansion of Phytophthora due to its rapid root regeneration. This study underscores the essential role of rootstock selection in managing Citrus gummosis and highlights the intricate mechanisms underlying plant resistance to pathogens [63,64,65].

This study provides a comprehensive analysis of factors influencing the number of Phytophthora spp. propagules per gram of dry soil (NPDS) and the number of colonies (NC) by integrating descriptive, multivariate approaches, and advanced models such as mixed models and mediation analysis. The results highlight the complementarity between environmental, genetic, and structural factors, while identifying strong relationships between the studied traits, reinforced by precise genetic and phenotypic metrics. The use of robust statistical techniques such as MANOVA ensures the reliability of the conclusions, and visualizations provide essential clarity to complex data. However, the results are specific to a given environment, limiting their generalization to other agro-ecological contexts. The low heritability of the studied traits, coupled with the predominance of environmental factors, complicates genetic improvement efforts in various contexts. Moreover, although the models used are powerful, they do not capture non-linear relationships or higher-order interactions. Finally, the impact of unmeasured factors, such as microscopic variations in soil structure, has not been integrated, which could influence the results.

5. Conclusions

This study highlights the complex interactions between environmental, genetic, and structural factors influencing NPSS and NC. The results demonstrate the importance of integrated management of rootstocks and environmental conditions to optimize productivity and soil quality. Although contextual, the findings provide a solid foundation for targeted agronomic strategies and future work to extend these conclusions to different environments. These results also open up perspectives for refining genetic selection approaches and agricultural practices, with a focus on key factors such as humidity and colony density, to maximize agronomic benefits in complex systems.