Climate variability governs mulch performance and stability in date palm (Phoenix dactylifera L.) in arid oases

Plastic mulch recommendations for date palm cropping systems in arid oases rest on demonstration trials conducted in favorable seasons. Yet the yield value of mulch-induced phenological advancement may depend on whether the season’s thermal regime places reproductive development within the effective pollination window, a contingency single-season trials cannot expose. We tested this hypothesis across three seasons (2023 baseline, 2024, 2025) using four mulching treatments: no mulch (NM), shredded date palm fronds (SDPF), polyethylene film over SDPF (PP+SDPF), and polypropylene woven ground cover over SDPF (PWGC+SDPF), on 64 mature Phoenix dactylifera L. cv. Najda trees at Errachidia, Morocco. Canopy development, spathe production, and marketable yield were recorded on individually tagged trees, and inter-seasonal stability was assessed using the Alternate Bearing Index and a Risk Index. PP+SDPF maintained the largest canopy in both 2024 and 2025, whereas treatment effects on spathe production and yield differed between seasons. In 2024, PP+SDPF recorded a marketable yield of 29.92 ± 3.56 kg tree −1, 4.2× the unmulched control; in 2025, it fell to the lowest marketable yield recorded (11.84 ± 2.22 kg tree −1), 7.93 kg tree −1 below the unmulched control (treatment × season interaction: p < 0.001). The association between frond number and spathe count was significant in 2024 (r s = +0.455, p < 0.001) but not in 2025 (r s = +0.206, p = 0.122), indicating that relationships among canopy development, spathe production, and yield differed between seasons. Across three seasons, PWGC+SDPF achieved the lowest Alternate Bearing Index (0.419 ± 0.065) and Risk Index (8.3%). These results are consistent with the phenological-window hypothesis and indicate that mulch permeability was associated with lower inter-seasonal yield volatility under the conditions of this trial. They suggest a possible trade-off between productivity and stability: PWGC+SDPF was associated with lower observed inter-seasonal variability, whereas PP+SDPF achieved a higher three-season mean yield but greater variability. As the trial was conducted at a single site with a single cultivar over three seasons, these relationships require verification across other locations, cultivars, and climates before broader recommendation.

Agronomists recommend plastic mulch to farmers for a reason grounded in physics: polyethylene film suppresses evaporation, conserves root-zone moisture, and buffers thermal extremes. Each mechanism improves crop performance so reliably that the recommendation has become the agronomic default in arid production systems (Kasirajan and Ngouajio, 2012; Gao et al., 2019; Demo and Asefa Bogale, 2024). The thermal effect, however, differs by mulch class: polyethylene film displaces the soil thermal regime in a direction set by its optical properties (Lamont, 2005), whereas organic surface mulches redistribute heat vertically, warming the surface while cooling the subsurface (Kader et al., 2017; Alharbi and Ghonimy, 2026a). The evidence base is broad and largely consistent. Plastic mulching has raised yield and water-use efficiency across a wide range of annual crops and climatic zones, with reported water-use efficiency gains of 20–60% (Kader et al., 2017) and yield increases spanning 7–47% depending on crop and material (Iqbal et al., 2020). A recent second-order meta-analysis reported a mean yield increase of 28.7% with 98% of trials recording significant improvement, and found these gains to hold across mulch types, thicknesses, and coverage levels, and across gradients of elevation, temperature, and precipitation (Wang et al., 2026). Benefits are also well documented in arid perennial systems: in a Saudi date palm orchard, palm-leaf basin mulching reduced daytime soil evaporation by 44–58% and lowered root-zone temperature by 3.6–6.8 °C between 2 and 10 cm depth, while raising surface temperature by roughly 3 °C (Alharbi and Ghonimy, 2026a), and companion column work under the same arid conditions showed organic mulch suppressing cumulative evaporation by 70–79% (Alharbi and Ghonimy, 2026b). These syntheses and trials characterize response across sites, climatic zones, and management configurations rather than across successive seasons at a single site. Interannual stability under a fixed management regime is a different quantity, and it remains largely unquantified.

The gap is structural, not accidental: demonstration trials typically run for one or two seasons under favorable conditions, and a technology that raises yield by 24% in a warm, dry year looks validated (Gao et al., 2019). But climate variability alone explains roughly a third of global crop yield variability, and more than 60% in some of the world’s major production regions (Ray et al., 2015), which means that a mulch evaluated in one season has been tested against a single draw from a distribution whose tails determine whether the recommendation helps or harms.

That single draw is insufficient because the most comprehensive meta-analysis in the field, pooling 1,700 trial-years of plastic mulch use across dryland crop systems, has not made a distinction that turns out to be decisive (Gao et al., 2019). Mulching consistently advances spathe emergence by 4–7 days in contrasting seasons (Uwase et al., 2026), but what determines yield is not when flowers arrive but whether they arrive at a thermal regime that sustains female receptivity. Effective pollination in date palm peaks within a narrow range of 20–35 °C, and female receptivity contracts sharply below it (Zahiri et al., 2023); in a cool season, flowers that emerge earlier under plastic mulch develop under exactly the thermal constraints that limit receptivity, producing no reproductive advantage despite the mechanism operating perfectly, a decoupling of consistent mechanism from contingent outcome. That decoupling means the same phenological advancement that drives 2024-season yield to 4.2× the unmulched control generates no gain at all when the season shifts, and it is precisely what single-season trials cannot expose. The pattern is not unique to date palm: across 837 observations from 50 published studies, the yield benefit of plastic mulching in maize was greatest under optimal conditions and diminished systematically as growing-season temperature rose, reaching zero above a mean of 24 °C (Yu et al., 2018), confirming that plastic mulch outcomes are governed not by the mechanism alone but by whether the ambient thermal regime supports the crop’s critical developmental phase.

Date palm, however, poses a synchrony challenge that monoecious crops do not: dioecism demands that male and female inflorescences open simultaneously within a window that climate sets annually and management cannot extend (Monselise and Goldschmidt, 1982; Hachef et al., 2023). Even when the thermal regime is adequate, pollination success therefore depends on a second climate-regulated alignment, compounding the contingency.

However, these reviews largely synthesize responses across crops, sites, mulch materials, and climatic zones, providing limited information on interannual performance under fixed site and management conditions (Kader et al., 2017; Iqbal et al., 2020; Wang et al., 2026). Similarly, the cited date palm studies characterize hydrothermal or phenological responses but do not quantify yield stability across successive contrasting seasons under the same management regime (Alharbi and Ghonimy, 2026a; Uwase et al., 2026). That cost falls directly on the growers now adopting plastic mulch across the Drâa-Tafilalet, growers who have no evidence-based way to anticipate what the next season will do to their investment. Development agencies in the region are scaling polyethylene film adoption based on demonstration trials conducted under warm-season conditions, precisely the conditions under which the contingency stays invisible (Bouzelmate et al., 2025). A grower who invests after seeing 2024-season yield reach 4.2× the unmulched control cannot anticipate that the trees leading production that season will fall to the bottom of the distribution the following season, or that the highest-yielding tree in one season will produce 91% less fruit in the next. What these growers need is an answer to a different question: not whether mulching works, but under what conditions each mulch type works, and at what cost in inter-seasonal volatility. Answering requires a design that holds the field, the trees, and the management constant while letting climate vary: three seasons at a single site, with individual-tree tracking across the full chain from canopy development through reproductive initiation to marketable fruit. We conducted a three-season field experiment applying four mulching treatments to 64 mature Phoenix dactylifera L. cv. Najda trees across three climatically contrasting seasons at a single site in Errachidia, Morocco. We propose the phenological window hypothesis: the yield value of mulch-induced phenological advancement is set not by the advancement itself but by whether the ambient thermal regime provides a receptivity window into which advanced reproductive development can land.

Experimental site, climate, and baseline conditions

The study was conducted at the Regional Centre for Agronomic Research, Errachidia, Morocco (31°55′N, 4°26′W; 1,036 m a.s.l.; Figure 1) during 2023 to 2025. Sixty-four mature Phoenix dactylifera L. (cv. Najda) trees, established in 2006 at a planting density of 156 trees ha −1, served as the experimental material. The experiment was arranged as a randomized complete block design with eight blocks. Blocks were defined by field position to account for spatial heterogeneity, and within each block the four treatments were randomly assigned to experimental units of two adjacent trees, giving one replicate of each treatment per block and 16 trees per treatment. Individual trees served as experimental units for statistical analysis.

Initial soil samples were collected in March 2023 from three depth intervals (0–30, 30–60, and 60–90 cm) prior to treatment establishment and analyzed at an accredited laboratory. Soil pH and electrical conductivity (EC) were determined in a 1:5 aqueous extract (NM ISO 10390:2013; NM ISO 11265:1995); calcium carbonate by calcimetry (NM ISO 10693); organic matter by wet oxidation (MOS11-PR09 V1, Skalar Analytical B.V., Breda, Netherlands); total nitrogen by the Kjeldahl method (NF ISO 11261); and cation exchange capacity by ammonium acetate extraction (NF X 31-108:2024). Soil texture was determined by the Robinson pipette sedimentation method (NF EN ISO 23470). Results are presented in Table 1. NM and NF designate Moroccan and French national standards, respectively, both aligned with ISO equivalents.

The study spanned three growing seasons with marked inter-annual variation in rainfall timing and total precipitation (Figure 2). Total annual rainfall was 136.0 mm (2023), 146.5 mm (2024), and 241.0 mm (2025). In 2023, rainfall was concentrated in February (69.0 mm), with minimal spring and summer precipitation (March to August: 21.0 mm) and moderate rain during fruit ripening (September to October: 46.0 mm). In 2024, the spring and summer period was extremely dry (February to April: 14.5 mm; March to August: 32.5 mm), with rainfall concentrated heavily in the late season (September to October: 111.0 mm). In 2025, rainfall showed a bimodal distribution, with substantial spring precipitation (February to April: 115.0 mm) and a second peak in December (76.0 mm), while summer and autumn remained comparatively dry (17.0 mm and 20.0 mm, respectively). Mean monthly maximum and minimum temperatures followed typical semi-arid Mediterranean patterns across all three years, with spring temperatures (March to May) ranging from 15.2 to 16.7 °C (Tmin) and 22.4 to 24.0 °C (Tmax), summer peaks (July to August) reaching 36.2 to 37.1 °C (Tmax), and autumn declines (September to October) to 25.5 to 28.4 °C (Tmax) (Figure 2). This inter-annual contrast, a modest 1.4–1.5 °C difference in spring mean temperature (2024: Tmin 16.6 °C, Tmax 24.0 °C; 2025: Tmin 15.2 °C, Tmax 22.4 °C) alongside an eight-fold difference in spring rainfall (14.5 mm in 2024 vs. 115.0 mm in 2025), provided the natural variation needed to test whether mulch-induced phenological advancement translates into a pollination advantage under contrasting spring climatic regimes.

Mulching treatments and experimental management

Four treatments were applied within a 2 m radius circular basin centred on each tree (Figure 1): (1) NM, no mulch (control); (2) SDPF, shredded date palm fronds applied at 100 kg per tree to a depth of 3 cm; (3) PP+SDPF, black polyethylene film (50 µm thickness) applied over the SDPF layer; and (4) PWGC+SDPF, polypropylene woven ground cover (water-permeable) applied over the SDPF layer. Synthetic mulches were inspected periodically for physical damage and repaired or replaced as needed to maintain treatment integrity. To maintain consistent organic mulch specifications across years, fresh SDPF (30 kg per tree) was applied each March, restoring the mulch layer to approximately 3 cm depth and 100 kg per tree total mass, consistent with standard annual pruning cycles in date palm oases.

Irrigation was delivered via a drip system (four emitters per tree; 400 L h −1 combined flow rate), with frequency adjusted seasonally from twice weekly during the cool season (November to February) to daily during the hot season (July to September), as described in full by (Uwase et al., 2026). Mineral fertilization was applied uniformly to all trees by fertigation during the critical growing period (May to July), supplying N 1.318 kg tree −1, P 2 O 5 0.214 kg tree −1, and K 2 O 0.855 kg tree −1 per year. Both irrigation and fertilization were kept identical across all four treatments throughout the three study seasons to isolate mulch effects from potential water-input or nutritional confounds.

Soil moisture dynamics under the same four treatments, in the same experimental field and during the same study period, were characterized as part of the broader experiment reported by Uwase et al. (2026). Both synthetic mulch treatments increased soil moisture at 0–40 cm depth, whereas PP+SDPF exhibited significantly lower moisture at 60–100 cm than PWGC+SDPF and NM. These companion measurements are used here only to support the interpretation of the observed treatment responses.

Canopy development monitoring

Canopy status was assessed at two fixed points within each annual cycle. In February of each year, during post-dormancy vegetative reactivation and prior to spathe emergence, two variables were recorded per tree: (i) total frond number, comprising all live fronds present at the time of assessment; and (ii) new-leaf number, defined as fronds that had physically emerged at the apical growing point since the previous assessment. To ensure objective identification across years, newly emerged fronds at the heart of the crown were marked at each February assessment using coloured spray paint, allowing unambiguous discrimination of new growth from pre-existing fronds in subsequent seasons. In September of each year, following harvest and after peak summer heat stress, dried-leaf number was recorded as the count of fronds exhibiting complete desiccation of the entire rachis and leaflets, identified in the field by characteristic downward bending. All canopy assessments were conducted by the same trained field team over a one-week observation window per assessment point, with team members cross-checking counts on each tree to minimise observer error.

Phenological monitoring and pollination protocol

Spathe emergence was monitored from March through May using daily visual observations conducted in the morning (0800 to 1200 h), when spathe activity is most readily observable. Each newly emerged spathe was recorded with the date of first observation and assigned a sequential identity number. First emergence date per tree was extracted as day of year (DOY). Total spathe count per tree was taken as the highest sequential identity number assigned within the season, equivalent to the total number of spathes that emerged. Within-season spathe count analyses were restricted to producing trees (those recording at least one spathe per season); non-producing trees were excluded from spathe count and emergence date comparisons. Pollination was performed by inserting fresh male spathes directly into open female spathes during peak receptivity, identified by the flower color transition from green to creamy-white to pale yellow, and securing them with thread to maintain contact while allowing natural wind-assisted pollen dispersal within the enclosed inflorescence. Pollination was conducted from 21 March to 25 April 2024 and from 14 April to 14 May 2025, with timing determined individually according to female spathe receptivity. Male spathes were sourced from dedicated male date palms in an adjacent plot and used within 3 to 5 days of collection to ensure pollen viability.

Yield assessment and stability metrics

Fruits were assessed for commercial maturity based on: (i) external color transition to dark mahogany-red across 80% or more of the fruit surface; (ii) slight skin wrinkling indicating moisture loss; and (iii) flesh yielding to gentle pressure without collapse. Harvest was conducted on 23 October 2024 and 25 October 2025. Total harvest weight (THW, kg tree −1) was recorded per tree using a calibrated field scale. Marketable yield (Mkt, kg tree −1) was derived as (Equation 1):

where WSF is the weight of spoiled fruit (kg tree −1). Fruits were classified as spoiled if exhibiting any of the following: (1) extreme desiccation; (2) underdevelopment due to pollination failure; (3) bird damage affecting 25% or more of the surface; (4) insect damage affecting 25% or more of the surface; (5) fungal rot or fermentation; or (6) advanced tissue collapse. Non-producing trees were retained in the analysis with a yield of zero to reflect farm-level productivity. The 2023 dataset provided the pre-treatment baseline for equivalence verification; because spoilage was not partitioned in 2023, THW 2023 equals the recorded 2023 yield.

Three-season yield stability was quantified from THW using four metrics. (i) Three-season mean yield (MY) per tree per treatment (Equation 2):

(ii) Coefficient of variation (CV) (Equation 3):

where σ is the standard deviation of THW across three seasons.

(iii) Alternate Bearing Index (ABI) (Equation 4):

where Y t is THW in season t; ABI ranges from 0 (perfectly stable) to 1 (complete alternate bearing) and was averaged per tree, then per treatment.

(iv) Risk Index (RI) (Equation 5), reported descriptively per treatment as the proportion of zero-yield tree-season observations across three seasons:

where n (zero) is the number of tree-season observations with zero THW and n (total) is the total number of tree-season observations (48 per treatment across three seasons). Individual tree identity was tracked across all seasons using Line_ID × Tree_ID, enabling within-tree cross-season correlation analyses.

Statistical analysis

All analyses were performed in R (version 4.5.1; R Core Team, 2025). Within-season treatment effects on continuous response variables were tested using one-way ANOVA with the individual tree as the experimental unit (n = 16 per treatment; error df = 60 for canopy and yield variables). For spathe count and first emergence date, analyses were restricted to producing trees, reducing error degrees of freedom to between 44 and 54 depending on season. Pairwise comparisons were conducted using Tukey HSD. Season-dependent treatment effects were assessed using two-way ANOVA (Treatment × Season) applied to canopy variables, spathe count, first emergence date, and yield components across 2024 and 2025. Spearman rank correlation coefficients (r s) were computed between total frond number, spathe count, and total harvest weight for all trees pooled across treatments within each season (n = 56 to 58 per season); ordinary least squares regression lines were fitted for graphical purposes only, and statistical inference was based exclusively on the Spearman coefficient. Yield stability metrics were analyzed as follows: (i) three-season mean yield and CV % were reported descriptively, with CV % additionally evaluated using one-way ANOVA; (ii) ABI was tested using one-way ANOVA with Tukey HSD pairwise comparisons; (iii) Risk Index was reported descriptively as the proportion of zero-yield tree-season observations per treatment. All tests used α = 0.05.

Canopy development and leaf senescence under mulching treatments

Total frond number and dried-leaf count were recorded for all four treatments in 2024 and 2025 to characterize the canopy response to mulching across contrasting seasons (Figures 3a, b). The pre-treatment baseline (2023) confirmed equivalence among treatments (p = 0.878). In 2024, PP+SDPF recorded the highest total frond number (66.4 ± 1.3 fronds tree −1), significantly exceeding all other treatments, which ranged from 56.6 to 61.6 fronds tree −1 (p < 0.001); this ranking was maintained in 2025, with PP+SDPF again highest at 69.3 ± 0.9 fronds tree −1 against 59.6 to 65.4 fronds tree −1 for the remaining treatments (p = 0.001), and no treatment × season interaction was detected (p = 0.866). For dried-leaf count, PP+SDPF carried the highest mean in 2024 (13.4 ± 0.9 fronds tree −1) and the lowest in 2025 (4.5 ± 0.6 fronds tree −1), a directional reversal that produced a significant treatment × season interaction (p = 0.0004); no other treatment showed a comparable shift between seasons (Figure 3b). New-leaf production increased across all treatments between seasons and differed significantly among treatments in 2024 only (p = 0.037), with PP+SDPF recording the highest mean (3.31 ± 0.15 fronds tree −1); no treatment differences were detected in 2025 (p = 0.190; Supplementary Figure 1).

Spathe development and production

Spathe count per producing tree and first spathe emergence date were recorded for all four treatments in 2024 and 2025 to characterize the reproductive response to mulching across contrasting seasons (Figures 4a, b). In 2024, spathe count differed significantly among treatments (p < 0.001), with PP+SDPF producing 13.6 ± 1.0 spathes tree −1, more than 1.5 times the count of any other treatment, which ranged from 7.4 to 8.6 spathes tree −1 (Tukey HSD, all p < 0.005 vs PP+SDPF). In 2025, no significant treatment effect was detected (p = 0.599), with all treatment means converging between 10.1 and 11.8 spathes tree −1, producing a significant treatment × season interaction (p < 0.001; Figure 4a). First spathe emergence differed significantly among treatments in 2024 (p = 0.007), with PP+SDPF emerging earlier than NM and PWGC+SDPF (Tukey HSD, p = 0.017); no treatment differences were detected in 2025 (p = 0.275), and the treatment × season interaction was not significant (p = 0.116; Figure 4b). The divergence between these interaction patterns suggests that season-dependent treatment responses were more pronounced for spathe production than for first-emergence timing, although the climatic mechanisms underlying this difference were not directly tested.

Marketable yield, total harvest weight, and fruit spoilage

Marketable yield, total harvest weight, spoiled fruit weight, and reproductive initiation across contrasting seasons are presented in Table 2. Treatments were equivalent at baseline in 2023 (marketable yield: p = 0.878), with reproductive initiation already lower under NM (12/16 trees, 75.0%) than under the plastic-based mulches, though this difference was not yet statistically tested at baseline. In 2024, reproductive initiation differed significantly among treatments (p = 0.016): PP+SDPF and PWGC+SDPF achieved complete initiation (16/16 trees, 100%), while SDPF reached 13/16 (81.2%) and NM recorded the lowest initiation at 11/16 (68.8%). This upstream difference in the number of reproducing trees compounded treatment effects on yield: PP+SDPF recorded the highest marketable yield (29.92 ± 3.56 kg tree −1) and total harvest weight (33.60 ± 3.36 kg tree −1), significantly exceeding all other treatments (Tukey HSD, p < 0.001), while NM recorded the lowest marketable yield (7.12 ± 1.98 kg tree −1); the gap between the two treatments was 22.8 kg tree −1. Spoiled fruit weight also differed significantly in 2024 (p = 0.014), with PWGC+SDPF recording the highest spoilage (5.29 ± 0.70 kg tree −1), significantly above NM (2.52 ± 0.65 kg tree −1), suggesting that a proportion of PWGC+SDPF’s total harvest did not meet marketable grade. In 2025, reproductive initiation converged across all treatments (p = 0.623), with NM and PWGC+SDPF each recording 15/16 producing trees (93.8%), PP+SDPF 14/16 (87.5%), and SDPF 13/16 (81.2%). Consistent with this convergence, treatment rank order for yield reversed: PP+SDPF fell to the lowest marketable yield (11.84 ± 2.22 kg tree −1), 7.93 kg tree −1 below NM (19.77 ± 3.82 kg tree −1), while no significant treatment differences were detected for marketable yield (p = 0.281), total harvest weight (p = 0.303), or spoiled fruit weight (p = 0.897). The absence of treatment differences in spoilage in 2025 confirms that PP+SDPF’s reduced marketable yield in the 2025 season reflected reduced fruit set rather than elevated post-harvest quality loss. The treatment × season interaction was highly significant (p < 0.001), confirming that treatment rank order reversed between seasons. The absence of treatment differences in spoilage in 2025 indicates that PP+SDPF’s lower marketable yield was not explained by greater fruit spoilage.

Relationships among canopy development, spathe production, and yield

Spearman rank correlations among total frond number, spathe count, and total harvest weight were computed for all trees in 2024 and 2025 (Figure 5). In 2024, total frond number was significantly correlated with spathe count (r s = +0.455, p < 0.001), spathe count with total harvest weight (r s = +0.606, p < 0.001), and total frond number with total harvest weight directly (r s = +0.557, p < 0.001). In 2025, the frond–spathe relationship weakened substantially and was no longer significant (r s = +0.206, p = 0.122); the frond–yield relationship showed the same pattern (r s = +0.225, p = 0.090). By contrast, the spathe–yield relationship not only held but strengthened (r s = +0.717, p < 0.001), remaining the only robust predictor of harvest weight across both seasons while the frond–spathe and frond–yield links held only under 2024.

Yield stability across seasons

Yield stability metrics computed from total harvest weight across the three seasons are presented in Table 3. PP+SDPF recorded the highest three-season mean yield (22.48 ± 1.21 kg tree −1), followed by PWGC+SDPF (20.18 ± 1.72 kg tree −1), SDPF (18.25 ± 1.82 kg tree −1), and NM (16.24 ± 1.92 kg tree −1). Despite its yield advantage, PP+SDPF did not achieve the lowest inter-seasonal oscillation: PWGC+SDPF recorded the lowest Alternate Bearing Index (ABI = 0.419 ± 0.065) and Risk Index (8.3%), marginally below PP+SDPF (ABI = 0.448 ± 0.053; Risk Index = 10.4%). NM recorded the highest ABI (0.674 ± 0.077) and Risk Index (22.9%), reflecting its large yield swings between seasons. ABI differed significantly among treatments (p = 0.039), with Tukey HSD confirming that PWGC+SDPF’s ABI was significantly lower than NM (p = 0.045); no other pairwise comparisons were significant. The coefficient of variation did not differ significantly among treatments (p = 0.153). For individual trees, the Risk Index shows that under no-mulch conditions nearly one in four tree-season observations recorded zero harvestable yield, a frequency of total loss that any mulch treatment reduced, and that PWGC+SDPF reduced to fewer than one in twelve.

Across the two treatment seasons, the experiment revealed clear season-dependent differences in mulch performance. PP+SDPF maintained the largest canopy in both seasons and produced the highest marketable yield in 2024, whereas it recorded the lowest numerical marketable yield in 2025, although treatment differences were not significant that year. Spathe production and the relationships among frond number, spathe count, and yield also differed between seasons. These findings, together with the significant treatment × season interaction, constitute the direct evidence from this study. However, the experiment did not isolate the climatic factors responsible for these patterns or directly measure root-zone and spathe temperatures, pollination success, or carbohydrate reserves. Temperature-driven phenological-window misalignment and carryover carbohydrate depletion are therefore presented as possible, non-exclusive explanations requiring direct testing.

Canopy development: stable mechanism, seasonal modulation

PP+SDPF sustained the largest canopy in both treatment seasons, significantly exceeding all other treatments in total frond number with no treatment × season interaction (Section 3.1; Figure 3a). The stability of this advantage across climatically contrasting seasons is consistent with the primary mechanism of impermeable plastic mulch: suppression of soil evaporation throughout the vegetative period, conserving root-zone moisture and reducing leaf water-deficit stress (Kasirajan and Ngouajio, 2012; Demo and Asefa Bogale, 2024). Regional irrigation studies in the Tafilalet confirm that soil water availability is the primary driver of frond production in mature date palm under semi-arid conditions (Sabri et al., 2017). The frond-count advantage of PP+SDPF across both years is therefore best interpreted as a persistent reduction in effective soil-water deficit, regardless of ambient temperature, establishing a larger photosynthetic source capacity on which reproductive output can subsequently draw.

The dried-leaf count told a markedly different story. PP+SDPF carried the highest post-harvest senescence load in September 2024 but the lowest in September 2025, a directional reversal, unique among the treatments, that produced a significant treatment × season interaction (Section 3.1; Figure 3b). Two non-exclusive interpretations are consistent with this pattern. First, in 2024 the larger, more productive canopy under PP+SDPF may have sustained faster frond-cohort turnover: trees that initiated more new fronds also senesced older fronds more rapidly, a pattern documented in high-productivity perennial canopies where source-sink activity accelerates leaf cycling (Monselise and Goldschmidt, 1982; Goldschmidt and Sadka, 2005). Second, wet 2025, with 115 mm of spring rainfall, elevated background soil moisture may have extended the hydraulic viability of leaf tissue across all treatments; PP+SDPF, maintaining the highest moisture through both mulch conservation and enhanced infiltration, would have shown the most pronounced senescence delay. This reversal in senescence dynamics prefigures the yield reversal documented in Section 4.3 and represents the first detectable signal of seasonal context modulating the expression of plastic mulch effects.

New-leaf production reinforced the picture of a 2024-specific canopy response: treatment effects were significant in 2024 only, with PP+SDPF highest, and absent in 2025 (Section 3.1; Supplementary Figure 1). This suggests that plastic mulching amplifies vegetative meristematic activity specifically when soil thermal enhancement adds marginal benefit to moisture conservation, conditions that prevailed in the dry 2024 spring (February–April rainfall: 14.5 mm) but not in wet 2025 spring (115 mm). These results align with the finding of (Uwase et al., 2026) that mulching acts as a coupled hydro-thermal regulator: its effects are strongest when ambient conditions leave a large thermal-moisture gap for mulch to fill and attenuated when the season independently satisfies water and temperature requirements for vegetative development.

Spathe production: the climate-regulated node

In 2024, the larger canopy built under PP+SDPF translated directly into substantially greater reproductive output: it produced more spathes than any other treatment and advanced first spathe emergence by 7.2 days relative to unmulched controls, with reproductive initiation reaching 100% against 68.8% under NM. This response is mechanistically consistent with plastic mulching operating through two reinforcing pathways at the canopy-to-spathe transition: sustained soil moisture conservation enabling uninterrupted meristematic activity through winter, enlarging the source capacity available for inflorescence initiation (Goldschmidt and Golomb, 1982); and surface thermal enhancement reducing growing degree day requirements for Phase A, which Uwase et al. (2026) found PP+SDPF lowered by 37% in the 2024. Together, these pathways explain why the frond-to-spathe link was robust in 2024.

In 2025, this link broke. Treatment means converged and no significant treatment effect was detected, the 7.2-day advancement of 2024 contracting to an undetectable 0.8 days, while the treatment × season interaction for spathe count remained highly significant (Figure 4). This is consistent with the hierarchical climate-mulch control established by Uwase et al. (2026), who found that year explained 45% of the variance in phase duration while mulching explained only 12%: climate sets the phenological framework and management modulates expression within those bounds. Independent records reinforce the primary thermal driver. Faci and Salah Eddine (2021) found that warmer late winters advanced spathe emergence by nearly three weeks in cv. Deglet Nour, and Hachef et al. (2023) showed that male flowering onset varied by nearly one month across years as a function of minimum air temperature. Within the bounds these thermal limits set, mulch acts only as a secondary modulator.

The treatment-specific response within 2025 adds a further dimension. SDPF achieved the narrowest emergence window of any treatment and a 5.1-day advance over NM, outperforming PP+SDPF in phenological advancement in 2025. This reversal of treatment ranking suggests that organic permeable mulch maintained effective moisture conservation without the deep-soil hydraulic restriction documented for plastic mulching by Uwase et al. (2026), who found PP+SDPF exhibited statistically lower moisture at 60–100 cm than PWGC+SDPF and NM. In wet years where deep infiltration is not limiting, permeable organic mulch may provide a more effective microclimate for Phase A advancement by avoiding the root-zone drying that accompanies impermeable film during sustained rainfall. Spathe production thus appears as the climate-regulated node in the production chain: the point at which seasonal context decides whether upstream canopy investment converts to reproductive output.

Yield reversal: mechanism-outcome decoupling across seasons

PP+SDPF’s marketable yield advantage over NM in 2024, at 4.2× the control, far exceeds the 24.3% mean yield advantage reported for plastic mulching across 266 annual-crop studies in dryland China (Gao et al., 2019), consistent with the compounding effect of mulching in a perennial crop where reproductive organs accumulate across multiple spathe-emergence events within a single season. Within the 2024 pattern, higher frond and spathe counts under PP+SDPF were associated with higher yield. Earlier spathe emergence may also have increased overlap with temperatures favourable for date palm pollination (20–35 °C) (Zahiri et al., 2023), although pollination and fertilization success were not directly measured. In 2025, PP+SDPF recorded the lowest numerical marketable yield, 7.93 kg tree −1 below NM, although no pairwise treatment difference was significant because of high inter-tree variability (Table 2). This contrasting seasonal pattern demonstrates season-dependent treatment performance but does not establish the underlying mechanism.

Two non-exclusive mechanisms, both consistent with the data but neither directly tested here, are proposed. First, phenological window misalignment: in 2025, the same plastic mulch that advanced flowering by 7.2 days in 2024 advanced it by only 0.8 days, a consequence of the reduced marginal thermal and moisture benefit of plastic mulch against a background of substantially wetter spring conditions (115 mm vs. 14.5 mm in 2024) and modestly cooler ambient temperatures (spring Tmin 15.2 °C vs. 16.6 °C, Tmax 22.4 °C vs. 24.0 °C in 2024). Flowers that did emerge did so in a slightly cooler thermal environment, closer to the lower bound of the 20–35 °C receptivity window, compounded by the dioecious nature of date palm, in which male and female inflorescences must open simultaneously, and by documented disruptions to flowering synchronicity in the Drâa-Tafilalet, where 30–40% of farmers in Errachidia already observe multiple asynchronous flowering cycles per year (Bouzelmate et al., 2025). Second, carryover source-sink depletion: trees that produced at high intensity under PP+SDPF in 2024 may have depleted meristematic and carbohydrate reserves, suppressing inflorescence initiation in 2025. Goldschmidt and Golomb (1982) demonstrated that carbohydrate reserves are the primary nutritional constraint on inflorescence initiation in the subsequent season across alternate-bearing perennials (Monselise and Goldschmidt, 1982); date palm presents analogous source-sink dynamics, and a high-productivity year under plastic mulching may amplify the carbohydrate depletion that suppresses reproductive initiation in the following season.

The spoilage data in 2024 reveal a mechanistically informative subsidiary pattern. PWGC+SDPF recorded the highest spoiled fruit weight, significantly above NM, while PP+SDPF was intermediate (Table 2). The 2024 season was characterized by extreme late-season rainfall concentration (111 mm in September–October), coinciding with the harvest window. Bouzelmate et al. (2025) note that heavy rainfall at the ripening stage causes Khamaj, a fungal rot and fruit collapse prevalent in the Drâa-Tafilalet oases. The water-permeable woven cover may have created localized soil moisture conditions during this rainfall event that facilitated such disorders, whereas the impermeable PP film would have intercepted the rain pulse, reducing the moisture reaching fruit bunches and explaining its lower spoilage in this specific event. In 2025, when late-season rainfall was minimal (~20 mm), no spoilage differences were detected, consistent with this interpretation.

Seasonal variation in relationships among canopy, spathe production, and yield

The Spearman rank correlations showed that the relationships among total frond number, spathe count, and total harvest weight differed between seasons, providing evidence of seasonal variation in the associations among vegetative development, reproductive production, and yield.

In 2024, all three links were simultaneously significant (fronds–spathes r s = +0.455; spathes–yield r s = +0.606; fronds–yield r s = +0.557; all p < 0.001), meaning the canopy advantage built by plastic mulching propagated through the biological system to yield. In 2025, the chain structure changed in a theoretically precise way: the frond–spathe link weakened to non-significance (r s = +0.206), and the frond–yield link likewise (r s = +0.225), while the spathe–yield link not only held but strengthened (r s = +0.717, p < 0.001). This pattern indicates that, in 2025, higher frond number was not significantly associated with greater spathe production, whereas spathe count remained strongly associated with yield. This is consistent with a seasonal constraint acting between canopy development and reproductive output, but the correlation analysis cannot identify the nature or cause of that constraint. Uwase et al. (2026) established that Phase A, the dormancy-to-emergence interval of the reproductive cycle, has a climate-contingent heat requirement measured in growing degree days (GDD): mulch shortened this interval by 37% in 2024 but only marginally in 2025. The correlation analysis provides a tree-level pattern consistent with an attenuation of the relationship between canopy development and spathe production when the mulch-related reduction in GDD was smaller.

This structural finding has a direct implication for farm monitoring. A grower assessing canopy condition in February of a dry-spring year can use frond count as a meaningful predictor of that season’s yield; the same assessment, wet year provides little predictive information at the tree level. Monitoring frameworks built around canopy assessment, common in commercial date palm management in Morocco (Sabri et al., 2017; Bouzelmate et al., 2025), therefore require seasonal calibration against background climate to retain predictive validity across variable years.

Mulch permeability as the stability determinant

Considered across all three seasons, the yield stability metrics reveal a trade-off between productivity and inter-seasonal reliability that maps directly onto mulch permeability (Table 3). PP+SDPF achieved the highest three-season mean yield, but PWGC+SDPF delivered the most stable performance: the lowest Alternate Bearing Index (0.419, significantly below NM), the lowest Risk Index (8.3%), and the lowest coefficient of variation (62.0%). NM recorded the highest ABI (0.674) and Risk Index (22.9%), meaning nearly one in four tree-season observations produced zero production, reflecting a comparatively high frequency of zero-yield observations under NM.

The most direct mechanistic explanation lies in differential soil moisture dynamics across the profile. Uwase et al. (2026) documented this in the same field: PP+SDPF and PWGC+SDPF increased surface moisture (0–40 cm) similarly in both years, but PP+SDPF exhibited statistically lower moisture at depth (60–100 cm), where date palm roots concentrate, whereas PWGC+SDPF did not differ from NM there. The permeable woven cover allows rainfall to infiltrate through the mulch layer, maintaining moisture at multiple depths across contrasting seasons and avoiding the boom-bust root-zone hydraulic cycle that impermeable film may create during alternating dry and wet years (Demo and Asefa Bogale, 2024).

All mulch treatments recorded lower descriptive Risk Index values than NM, ranging from 8.3% to 18.8% compared with 22.9% under NM. This pattern suggests a potential reduction in the frequency of zero-yield tree-season observations, although the Risk Index was not subjected to inferential testing. PP+SDPF recorded the highest three-season mean yield, whereas PWGC+SDPF showed the lowest observed interseasonal variability. The economic implications of these differences cannot be determined without considering material and labor costs, market prices, and financial returns.

A proposed phenological-window framework

Together, the results show that mulch responses varied between seasons. The canopy advantage under PP+SDPF was maintained, whereas spathe production and yield responses differed between 2024 and 2025. This observed pattern is consistent with the phenological-window hypothesis but does not demonstrate its underlying thermal or pollination mechanisms. The framework is therefore proposed as a testable explanation for the observed season-dependent responses.

We propose the phenological window hypothesis: mulch-induced phenological advancement is agronomically beneficial only when ambient thermal conditions place the resulting reproductive development within the thermally effective window for pollination. When the season is thermally constrained, the same advancement either fails to manifest (0.8 days in 2025 versus 7.2 days in 2024, because the thermal ceiling blocks Phase A GDD reduction) or occurs without the thermal environment needed to sustain receptivity and fertilization success. In 2024, the spring thermal regime placed flowers in the 20–35 °C range optimal for date palm pollination (Zahiri et al., 2023), and mulch-advanced flowers entered that window ahead of the general emergence period, compounding the advantage. In 2025, the thermal regime set a lower ceiling for Phase A GDD reduction (Uwase et al., 2026), advancing flowers only marginally and into the same suboptimal conditions faced by control trees, producing the convergence in spathe counts and yield documented in Sections 4.2 and 4.3. The dioecious nature of date palm amplifies this effect relative to monoecious crops: male and female inflorescences develop on separate trees and can exhibit distinct thermal requirements (Chao and Krueger, 2007; Hachef et al., 2023), so climate-induced misalignment between pollen availability and female receptivity compounds the effect of suboptimal temperatures at the individual-flower level.

The phenological window hypothesis is testable; its confirmation would require: (a) spathe-level temperature measurement during the receptivity period in mulched and unmulched trees across contrasting seasons simultaneously (Uwase et al., 2026); (b) pollen viability assays at peak spathe receptivity under contrasting mulch × season conditions (Zahiri et al., 2023); and (c) carbohydrate reserve measurements across seasons to quantify the carryover depletion mechanism independently of the window misalignment mechanism. Beyond date palm, the hypothesis may extend to other dioecious perennials managed through microclimate modification, such as pistachio (Pistacia vera) and kiwifruit (Actinidia deliciosa), in which mulch-induced phenological advancement under cool conditions could similarly produce misalignment with the effective pollination window.

Scope, limitations, and future research priorities

This study’s three-season, individual-tree design provides the minimum evidential base for detecting and characterising the contingency pattern. Five scope boundaries constrain generalization and define research priorities.

Location and cultivar. The experiment used a single site and a single Bayoud-resistant cultivar (cv. Najda). Interactions between mulch type, irrigation regime, and cultivar thermal requirements vary substantially across the Moroccan and international date palm range; multi-location trials across the Drâa-Tafilalet, Ziz, and Draa valleys, and internationally, are required before regional recommendations can be scaled.

Climate coverage. Two treatment years captured one dry-spring and one wet-spring contrast, sufficient to demonstrate the contingency but insufficient to characterize its probability distribution. Five to ten contrasting seasonal cycles would be needed to assess pattern stability as baseline temperatures continue rising (Bouzelmate et al., 2025).

Thermal measurement resolution. Root-zone and spathe temperatures were not directly measured; therefore, the proposed thermal mechanism remains inferential. Future studies should monitor temperatures at both levels across contrasting seasons to test the phenological-window hypothesis.

Reproductive mechanism. Pollen viability, fertilization success, fruit-set percentage, fruit number, and bunch number were not quantified as response variables. Consequently, the stage at which reproductive development was constrained cannot be determined from yield alone and requires direct investigation.

Carbohydrate reserves. Carbohydrate reserves were not measured across seasons; therefore, carryover source–sink depletion remains a proposed explanation requiring direct testing.

Economic and environmental analysis. Agronomic performance and sustainability are distinct questions, and the present study evaluates only agronomic performance and yield stability. It does not assess economic returns, plastic-waste accumulation, potential microplastic formation, material recovery, labor costs, or carbon footprint.

Mulch material characterization. Manufacturer information and the optical, thermal, and quantitative permeability properties of the synthetic mulches were not retained in the procurement records, limiting exact material replication and detailed mechanistic interpretation.

Despite these constraints, the results consistently show that treatment performance varied between seasons and that PWGC+SDPF had the lowest observed interseasonal yield variability under the conditions of this trial. These patterns are consistent with the proposed climate-contingent framework but do not establish the causal roles of climate or mulch permeability. They provide a foundation for multi-location and multi-year investigations aimed at translating these findings into adaptive management. As interannual climate variability intensifies across arid oasis agroecosystems, mulch recommendations should consider cross-season stability alongside mean yield.

This three-season study showed that mulch performance varied between seasons in mature date palm. PP+SDPF maintained the largest canopy in both treatment seasons and produced the highest marketable yield in 2024, but recorded the lowest numerical yield in 2025, with a significant treatment × season interaction. These findings demonstrate seasonal variation in treatment performance but do not establish the climatic or physiological mechanisms responsible. The proposed phenological-window misalignment and carryover carbohydrate depletion mechanisms therefore require direct testing. Across the three seasons, PP+SDPF achieved the highest mean yield, whereas PWGC+SDPF recorded the lowest Alternate Bearing Index and Risk Index, suggesting a potential trade-off between productivity and stability under the conditions of this trial. The results highlight the importance of evaluating mulch treatments across multiple seasons rather than relying only on single season mean yield. Because the study was conducted at one site with one cultivar, further validation across locations, cultivars, and longer climatic sequences is required before broader recommendations are made.