Carbon and Water Balances in a Watermelon Crop Mulched with Biodegradable Films in Mediterranean Conditions at Extended Growth Season Scale

Abstract

The carbon source/sink nature and the water balance of a drip-irrigated and mulched watermelon cultivated under a semi-arid climate were investigated. Biodegradable films, plants and some fruits were left on the soil as green manure. The study spanned from watermelon planting to the subsequent crop (June–November 2023). The eddy covariance technique was employed to monitor water vapor (H₂O) and carbon dioxide (CO₂) fluxes, which were partitioned into transpiration, evaporation, photosynthesis and respiration, respectively, using the flux variance similarity method.This method utilizesthe Monin–Obukhov similarity theory to separate stomatal (photosynthesis and transpiration) from non-stomatal (respiration and evaporation) processes. The results indicate that mulching films contribute to carbon sequestration in the soil (+19.3 g C m⁻²). However, the mulched watermelon crop presented in this study functions as a net carbon source, with a net biome exchange, representing the net rate of C accumulation in or loss from ecosystems, equal to +230 g C m⁻². This is primarily due to the substantial amount of carbon exported through marketable fruits. Fixed water scheduling led to water waste through deep percolation (approximately 1/6 of the water supplied), which also contributed to the loss of organic carbon via leaching (−4.3 g C m⁻²). These findings recommend further research to enhance the sustainability of this crop in terms of both water and carbon balances.

1. Introduction

Watermelon [Citrullus lanatus (Thun.) Matsum & Nakai] is a crop of considerable economic importance worldwide. Currently, Spain leads watermelon production in Europe, followed by Italy, whose production has been continuously increasing both in yield and surface terms, reaching approximately 650,000 Mg in 2023 [1]. Watermelon crop, when managed in irrigated conditions, requires substantial quantities of water to improve yield due to its high growth rate, short growing period, and water content of 90–92% in the fruits [2]. Since irrigated watermelon, like other cucurbit crops, benefits from mulching in open field [3], the adoption of soil partial mulching is a cropping management practice that is continuously expanding in the Mediterranean region, where water is a limiting factor for stable and high-quality productions [4,5].

The commercial value of watermelon fruits is strongly linked to marketable quality standards; therefore, non-saleable fruits (due to non-standard sizes, anomalous shapes or imperfect skin) are usually not harvested [6] and are left on the soil surface, until subsequent chopping. This final crop management step, following the scaled harvesting of watermelon fruits, has never been studied in terms of carbon dioxide (CO₂) and water (H₂O) balances, despite it being part of the cropping season.

Mulching efficiently reduces ineffective water consumption by reducing soil evaporation [7]; therefore, it is often recommended as an agrotechnical practice to enhance water use efficiency (WUE) [8] in high-value crops, both herbaceous and arboreal [9].

On the other hand, mulching with films was found to significantly affect greenhouse gas (GHG) emissions during the crop growth cycle, although the actual impacts on the GHG balance are still debated. Gao et al. [10] found a general consistent reduction in GHG emissions in mulched rice cropping; Li et al. [11] noted that a film-mulched cotton crop acts as a stronger carbon sink compared to a non-mulched canopy; Gong et al. [12] observed that partial mulching in maize stimulates assimilation more than respiration throughout the growing season, resulting in more efficient carbon storage; Qian et al. [13] found that film mulching combined with straw incorporation in a maize crop increases CO₂ and nitrous oxide (N₂O) emissions and, at season level, increases soil C storage with respect to other treatments. Conversely, other studies (see the review by [14]) have reported that mulching films increase N₂O emissions, an effect largely dependent on the agricultural system adopted. Furthermore, Ming et al. [15] found that plastic mulching combined with drip irrigation in rice enhanced both carbon assimilation and respiration, with the latter increasing more than the former. Similarly, Xiong et al. [16] reported that in wheat, plastic mulching plus irrigation decreased the C footprint at high nitrogen (N) fertilization rates, whereas the C footprint increased under low N rates compared to cultivation without mulch.

The processes involved when a film uniformly covers part of the cultivated surface include alterations in the soil’s hydrological and thermal conditions, which, subsequently, influence the soil’s biogeochemical processes. These alterations stimulate (i) the decomposition of soil organic carbon (SOC) and (ii) the fermentation and nitrification/denitrification processes, producing methane (CH₄) and N₂O, respectively. On the other hand, the increase in the soil temperature and moisture content due to film mulching contributes to the rapid loss of SOC. These two contrasted processes can result in enhancing, stabilizing, or decreasing SOC conservation [17,18,19,20] depending on environmental factors, agricultural systems, crop types and field management practices [11,13]. However, in general, as reported by 21. Birkhofer et al. [21], short-term studies cannot provide a clear picture of changes in agroecosystem support services due to agricultural systems since soil carbon requires at least 8–10 years to detect meaningful changes.

The extensive use of plastic mulching films in agriculture has strong environmental impacts [22,23] which can potentially be mitigated by using biodegradable materials that degrade in the environment by between 85% and 95% depending on their components [24]. However, aside from the general improvement in WUE when biodegradable mulching films are used [25], information on carbon and water balances in such scenarios is scarce, especially considering that the film is typically not collected and removed but left on the soil surface post-harvest until it decomposes.

The eddy covariance (EC) technique for measuring scalar fluxes in the soil–crop–atmosphere continuum (see, e.g., [26]) is a suitable method for accurately establishing H₂O and CO₂ balances at the ecosystem level [27,28]. This capability of the EC technique can be notably enhanced to understand the separate active roles of the crop and the soil on H₂O and CO₂ exchanges when these EC fluxes are partitioned into transpiration (T) and evaporation (E) and photosynthesis (P) and respiration (R), respectively. The well-established flux variance similarity (FVS) method can be utilized for this purpose; this technique was first introduced by Scanlon et al. [29,30]. FVS can distinguish between stomatal (T and P) and non-stomatal (E and R) fluxes when active plants are present on the soil surface [31].

The aim of this study was to assess the dynamics of canopy CO₂ assimilation and respiration, H₂O crop transpiration and soil evaporation for a watermelon crop subjected to a Mediterranean semi-arid climate. Furthermore, water use efficiency has been determined over six months from planting to the next cultivation.

2. Materials and Methods

2.1. The Site, the Crop and the Mulching

The field site was at the CREA-AA Research Unit experimental farm located in southern Italy (Rutigliano–Bari, 41° 01′ N, 17° 01′ E, altitude 147 m a.s.l.). The site is characterized by a Mediterranean semi-arid climate, categorized as Csa according to the Köppen–Geiger classification. Over the past thirty years, the average annual rainfall has been 535 mm, primarily occurring between autumn and late winter. The year 2023 was hot and very dry compared to the preceding twenty years. The soil, classified as “Lithic Rhodoxeralf”, features a clay texture, stable structure, shallow profile (0.6–1.1 m) and rapid drainage due to an underlying cracked limestone subsoil. The SOC content averages around 12.0 g kg⁻¹. The field capacity (FC) and the permanent wilting point (WP) volumetric water contents are 0.36 and 0.21 m³m⁻³, respectively. Consequently, with a bulk density of 1.15 Mg m⁻³, the available soil water ranges from 80 to 140 mm.

Measurements were conducted on a watermelon crop (seedless var. Lion king), following a broccoli cabbage crop harvested in April and partially incorporated (0.81 kg m⁻² of fresh biomass in a soil layer depth of 0.30 m, corresponding to 0.69 kgH₂O m⁻²) as green manure on 25 May 2023. Main tillage at medium depth ploughing (0.30 m) and seedbed preparation were performed between 25 and 30 May 2023; the biodegradable film mulch (model PC 100 d8, BASF, Italy, 1 m width) was applied on 1 June 2023. On the same day, driplines (2.1 Lh⁻¹ emitters, 0.60 m apart) and the main organic fertilization (Orga-Kem 6.11.8 + 11CaO, 300 kg ha⁻¹) were also applied. The watermelon plants were transplanted on 9 June at a spacing of 2.70 m between rows and 1 m between plants, covering an area of about 4.0 ha, with a density of approximately 3200 plants ha⁻¹. Every 6 rows, the inter-row distance was 5 m to facilitate machinery passage. The first irrigation was performed the day before planting. Crop management adhered to the usual treatments in the area including mechanical weed removal every 4 weeks, irrigation around three times per week to maintain optimal soil water conditions and monthly fertigation (ammonium sulphate 50 kg ha⁻¹, magnesium nitrate 30 kg ha⁻¹, calcium nitrate 60 kg ha⁻¹, mycorrhizae 20 kg ha⁻¹). The scalar harvest of marketable fruits occurred between 28 and 31 August 2023. After harvesting, on 25 September 2023, the fresh plant residues (0.6 kg m⁻² of fresh biomass, corresponding to 0.49 kgH₂O m⁻²), unharvested fruits (4.0 kg m⁻² of fresh material, corresponding to 3.7 kgH₂O m⁻²) and the mulching film were chopped by a tractor shredder and ploughed in two steps, on 2 and 13 October 2023, to a soil depth of 0.30 m. Measurements concluded at the end of November 2023, when tillage for the new winter crop commenced.

For the development during the growing cycle, plants were harvested weekly or bi-weekly with randomized samples (3 plants). Each was analyzed for fresh biomass (FB) weight, dry matter (DM), leaf area index (LAI) using a Licor 3100 (Li-COR Inc., Lincoln, NE, USA) and plant height.

The field coverage ratio on sampled dates was obtained by NDVI. The instrumentation for multispectral remote sensing was the Phantom 4 RTK unmanned aerial vehicle (UAV) with DJI GS PRO software (v2.0.15) for planning the survey area. The flight altitude was 33 m with 80% front and side overlap and a flight speed of 2.3 m s⁻¹. The Phantom 4 is equipped with six 1/29” CMOS sensors in the Blue (450 nm ± 16 nm), Green (560 nm ± 16 nm), Red (650 nm ± 16 nm), Red edge (730 nm ± 16 nm), Near-infrared (840 nm ± 26nm) and RGB bands for visible light. During image acquisition, the Phantom 4 RTK UAV was coupled to a DJI D-RTK 2 Mobile Station differential positioning system (https://www.dji.com/it/d-rtk-2/info; last accessed on 12 June 2024). PIX4D mapper v.4 software (https://www.pix4d.com/product/pix4dmapper-photogrammetry-software/; last accessed on 12 June 2024) was used to mosaic and orthorectify the images, achieving a resolution of 0.02 m. The UAV’s solar sensor was used for reflectance calibration (https://dl.djicdn.com/downloads/p4-multispectral/20200717/P4_Multispectral_Image_Processing_Guide_EN.pdf; last accessed on 12 June 2024). After importing the images produced by PIX4D into the QGIS environment (https://qgis.org/en/site/; last accessed on 12 June 2024), for each acquisition date, the NDVI vegetation index was calculated to determine the area covered by vegetation in the study area.

2.2. Continuous Monitoring of Agro- and Micro-Meteorological Variables

The EC flux tower was established in the center of the field a few days prior to the installation of the mulching film, specifically on 25 May 2023. The equipment comprised a three-dimensional sonic anemometer (uSonic 3 Scientific, Metek GmbH, 25337 Elmshorn, Germany) and a fast response open-path infrared gas analyzer (LI-7500, Li-COR Inc., Lincoln, NE, USA). The three wind components, sonic temperature and atmospheric concentrations of CO₂ and H₂O were continuously measured at 1.5 m above the crop canopy, with the sensor height adjusted to follow crop growth, reaching a maximum of 1.75 m. The gas analyzer was positioned 0.32 m laterally from the anemometer to avoid airflow distortions, endeavoring measurement of the same air parcels. Micrometeorological data were recorded at a frequency of 10 Hz on a dedicated computer using the MeteoFlux software (Servizi Territorio, S.n.c., Cinisello Balsamo, Italy) and were stored on an hourly scale. Post-processing and computation of hourly fluxes of H₂O (Fq, mmol m⁻² s⁻¹) and CO₂ (Fc, μmol m⁻² s⁻¹) were conducted using EddyPro^® software, v7.0.9 (http://www.licor.com/eddypro, last accessed on 12 June 2024), applying 60 min block averaging, double coordinate rotation, the statistical test [32], the maximum cross-covariance method, and the WPL density correction [33].

Continuous hourly meteorological data, including air temperature (Ta, °C), relative humidity (RH, %), precipitation (P, mm), global radiation (Rg, W m⁻²) and photosynthetically active radiation (PAR, mmol m⁻²) were collected from an automatic weather station equipped with standard sensors, situated in proximity to the experimental field.

Soil water content in volume (θ, m³ m⁻³) was measured by capacitive probes (10HS, Decagon Devices Inc., Pullman, WA, USA). Three points were monitored: two along the row below the mulched surface (θ_r) and one between the rows (θ_ir). At each point, two capacitive probes were installed horizontally into the soil profile and transversely to the row, at depths of −0.15 m and −0.30 m. These sensors were connected to dataloggers (Tecno.el srl, Roncade, Italy) with data stored hourly and averaged daily. Integrated soil water content for the profile was calculated by integrating values measured at each depth, as described in [34]. The θ measurements from the three points were pooled to obtain a single mean value as [35,36]:

$$\theta =0.7\overline{{\theta }_r}+0.3{\theta }_{ir},$$

(1)

where $\overline{{\theta }_r}$ is the mean of the two θ measurements along the rows. Soil water availability was determined by the relative extractable water (REW, unitless, [37]) calculated as:

$$REW={\displaystyle \frac{\theta -{\theta }_{min}}{{\theta }_{max}-{\theta }_{min}}},$$

(2)

where θ_min and θ_max represent the minimum and maximum soil water content, respectively, observed during the experiment.

2.3. The Net Biome Exchange, the Carbon and Water Balance

The hourly Fq and Fc fluxes were cumulated over the course of the day during the extended growth season to establish the water and carbon balance in this cropping system.

The net flux of CO₂, which represents the net ecosystem exchange (NEE), also referred to as ecosystem productivity (NEP), is given by:

$$NEE=GPP+R_{eco},$$

(3)

where all terms are in g C m⁻²; GPP denotes the gross primary production (carbon assimilated via photosynthesis) and R_eco represents the total ecosystem respiration. Both parameters were estimated by the EC measurements and FVS partitioning. In accordance with atmospheric conventions, fluxes moving downward from the atmosphere to the ecosystem (GPP) are negative, while fluxes moving upward are positive (R_eco).

To determine if the watermelon crop under the present conditions can be classified as a carbon sink (C-sink, indicating carbon uptake from the atmosphere, negative) or a carbon source (C-source, indicating carbon loss from the ecosystem to the atmosphere, positive), the net biome exchange (NBE, [38,39]) is calculated by:

$$NBE=NEE+{FC}_{harvest},$$

(4)

where FC_harvest denotes the organic carbon exported through watermelon fruit harvesting.

The carbon budget of the watermelon field can be assessed by considering all C inputs and outputs. Rigorously, the CH₄ flux should also be included in the C balance [40]. However, for this study, this term was not considered, and a simplified version was used for the variation in soil organic carbon stock (ΔSOC) over time (Δt):

$$∆SOC/∆t=NEE+{FC}_{in}+{FC}_{out},$$

(5)

where FC_in and FC_out are the organic carbon imported (positive) and exported (negative), respectively. Negative values of NEE by Equation (3) indicate C import to the field; hence, for this balance, the sign must be inverted.

The total organic carbon input is given by:

$${FC}_{in}={FC}_{green.manure}+{FC}_{org.fert}+{FC}_{mulching.film},$$

(6)

where FC_green.manure is the organic carbon imported via green manure from broccoli, watermelon plants and fruits, FC_org.fert is the organic carbon imported by fertilization and FC_{mulching.film} is the organic carbon imported via the incorporation of the biodegradable mulching film into the soil.

The total organic carbon out is given by:

$${FC}_{out}={FC}_{harvest}+{FC}_{leach}$$

(7)

where FC_leach represents the organic carbon lost by leaching.

ΔSOC was measured in a 0–0.30 m soil layer, where all green manure operations were conducted and more than 80% of the total root length was found [41].

To determine the C content (%) in the fresh biomass and the biodegradable mulching film, a CHN Elemental Analyzer (Flash EA 1112, Thermo Scientific, Waltham, MA, USA) was employed. Broccoli biomass, watermelon plants and harvested fruit biomass were collected in triplicate samples of 1 m², randomly collected in the area around the EC tower.

The ΔSOC of Equation (5) was computed by sampling soil before seeding after the green manure of the broccoli crop and at the end of the monitoring period (24 May 2023 and 30 November, respectively). On air-dried and sieved soil samples (0.5 mm particle size), SOC was measured by dry combustion [42] with a Vario Select analyzer (Elementar, Hanau, Germany).

C lost via leaching was estimated according to [43], by measuring the soil water-extractable organic carbon (WEOC) content at the beginning and end of the trial. WEOC was measured according to a protocol reported by [44].

Finally, the uncertainties on the presented terms of the C balance were evaluated using the error propagation law [45], where the final uncertainty is the sum of the absolute errors calculated through the relative error on the sampled values.

In arid and semi-arid areas with minimal slopes, surface runoff can be neglected, and capillary rise is nearly non-existent in shallow soils [4]. Thus, a simplified water balance was established on a daily scale as:

$$∆\theta =P+I+E+T+DP,$$

(8)

where I is the water supplied via irrigation, E and T are the actual evaporation and transpiration, respectively; DP is the water lost via deep percolation, and $∆$θ is the soil water content variation in the time interval. All terms in Equation (8) are in kg H₂O m⁻².

The FVS partitioning method was employed to determine the terms GPP, Reco, E, and T, in Equations (3) and (8), respectively. The FVS approach applies the flux variance similarity, derived from the Monin–Obukhov similarity theory, separately to stomatal (GPP and T) and non-stomatal (Reco and E) processes. The FVS method provides analytical expressions to partition CO₂ and H₂O fluxes simultaneously; it requires inputs of (i) the variables commonly derived from high-frequency EC measurements and (ii) the leaf-level water use efficiency (WUEl) for the analyzed crop. Here, the FVS calculation proposed by [46] was utilized; the H₂O and CO₂ EC fluxes were partitioned using an adaptation of the code in Phyton provided by [47] and downloaded from https://github.com/usda-ars-ussl/fluxpart, v0.2.10, last accessed on 26 May 2024. WUEl values were directly estimated by the cited code. Further details about the partitioning method used can be found in the study by [31] conducted at the same experimental site with the same equipment.

2.4. The Water Use Efficiencies

To provide practical agricultural and environmental indicators, several WUE values [39,48,49,50] are calculated on a seasonal growth scale.

The agronomic water use efficiency (WUE_agro, kg m⁻³), often referred to as “water productivity” by several authors (see [49] for a review), is determined as the ratio between the final harvested crop production (yield of fresh fruits biomass, in kg m⁻²) and the cumulative evapotranspiration (ET, in m³ m⁻²) during the period from planting to harvest:

$${WUE}_{agro}={\displaystyle \frac{yield}{ET}}.$$

(9)

The irrigation water use efficiency (WUE_irr, kg m⁻³) is calculated as the ratio between yield (kg m⁻²) and the sum of irrigations (I, m³ m⁻²) and precipitations (P, m³ m⁻²) accumulated during the period from planting to harvest:

$${WUE}_{irr}={\displaystyle \frac{yield}{P+I}}.$$

(10)

The physiological water use efficiency (WUE_ph, g C kg H₂O⁻¹; [48,51]) over the extended growing period is given by:

$${WUE}_{ph}={\displaystyle \frac{GPP}{T}}.$$

(11)

More commonly, this indicator is defined as [39,52,53]:

$${WUE\prime }_{ph}={\displaystyle \frac{GPP}{ET}}.$$

(12)

The environmental water use efficiency (WUE_env, g C kg H₂O⁻¹; [54]) at the ecosystem level over the extended growing period is:

$${WUE}_{env}={\displaystyle \frac{NEE}{ET}}$$

(13)

3. Results and Discussion

3.1. The Weather, the Soil Water Content and Crop Development

The seasonal daily time courses of the main agrometeorological variables—air temperature (Ta), photosynthetically active radiation (PAR), vapor pressure deficit (VPD, kPa) and precipitation (P)—during the experimental period (1 June–30 November) are shown in Figure 1a,b. Ta was 24.6 ± 3.4 °C, with a very hot period in the second half of July when daytime hourly temperatures exceeded 41 °C. During late spring and early summer, 48.8 mm of precipitation was recorded. The radiation pattern expressed by PAR follows the usual trend for the area, while VPD reached high values corresponding to high Ta in the second part of July.

The patterns of soil water content, expressed by θ, are illustrated in Figure 2, along with irrigation and rainfall events for clarity. In the figure, mean-field θ values are reported, distinguishing between mulched rows and non-mulched rows, along with the wilting point and field capacity. At the beginning and end of the experimental period, missing values were due to equipment malfunction. After irrigation began, θ values evolved differently for θ_r and θ_ir; while soil water content below the mulched surface remained almost constant and around field capacity (FC), the water content in soil not covered by the films decreased rapidly as evaporation proceeded, even dropping below the WP in the middle part of the growth cycle and more severely after the crop harvesting. In the latter case, the sudden increase in θ_ir corresponds solely to precipitation events. Mean θ values at the field level remained above WP. These high mean θ values were due to substantial water supply via irrigation (583 mm), resulting in high relative extractable water (REW) values in the first half of the growth cycle, well above the water stress threshold for food herbaceous crops, which ranges between 0.4 and 0.5 (i.a., [55,56]).

The development of the crop plants is shown in Figure 3, in terms of LAI and fresh green biomass. Both variables exhibit similar trends during the growth cycle, showing a slow increase in June, a rapid increase in July with a maximum in early August, followed by a decrease until the harvest of the watermelon fruits. The final yield of freshly harvested fruits for the market was 10.00 kg m⁻².

3.2. The Fluxes of CO₂ and H₂O

The trends of hourly EC fluxes Fc and Fq for the watermelon crop are reported in Figure 4 and Figure 5, respectively; the patterns are illustrated at a decadal scale to capture the dynamics during the entire experimental period.

During the first month after planting (June), Fc was always positive due to bare soil respiration, with values decreasing from the first to the third decade as vegetation developed, increasing LAI and FB and enhancing CO₂ assimilation (Figure 3). During July and August, when the crop was fully developed, the system acted as a complete carbon sink during daytime, reaching maximum values from −15 up to −20 μmol m⁻² s⁻¹ in the third decade of July. In the third decade of August, during the scalar harvest of the fruits, the system became a weak carbon sink, gradually turning into a carbon source from September until the end of the experiment. In early November, a small CO₂ assimilation likely resulted from weeds grown in the field before tillage for the successive crop. During nighttime, the system was always a very weak source of carbon.

The patterns of Fq (Figure 5), always at a decadal scale, followed the standard actual crop evapotranspiration trend throughout the experimental period. The only deviation from the standard ET trend can be seen in the third decade of September, when heavy rainfall (Figure 1) supplied water to the soil, which was available for evaporation. This feature can also be partially attributed to the inclusion of fresh chopped biomass rich in fully available water to the soil [57]. The maximum Fq values (around 8 mmol m⁻² s⁻¹) were reached in the third decade of July through the first decade of August, corresponding with the highest Fc values (see Figure 4). Furthermore, the Fq trend was very similar in the first and second decades of September, when a decrease in evaporation would have been expected in the absence of rainfall and irrigation (see Figure 1).

The partitioned fluxes of CO₂ and H₂O are presented at a daily scale for the entire experimental period in Figure 6 and Figure 7 for the stomatal and non-stomatal components, respectively; surface coverage is also shown in both figures.

The trends of plant transpiration and photosynthesis were naturally opposite (Figure 6), aligning perfectly with canopy development as indicated by coverage, from the beginning of the growth season until residual chopping in early October. Afterward, both transpiration and photosynthesis ceased until the end of the experiment.

The soil evaporation and respiration (Figure 7) followed the typical trend of a well-irrigated crop; however, in this case, the CO₂ emissions from the soil presented quite high values a few days after harvesting of watermelon fruits and, further, after the residual chopping. This pattern was also reported for evaporation, even if with less evidence.

This increase in CO₂ emissions post-harvest was likely due to biodegradation of the fresh biomass (see following C balance) from non-marketable watermelon fruits left on the soil, the incorporation of fresh plant biomass [57], and the initiation of mulching film degradation after chopping and incorporation into the soil in October, contributing a total of 2.96 g C m⁻² to the soil

Furthermore, after incorporating watermelon plants at the end of September, H₂O fluxes showed a clear decreasing trend (Figure 5e,f), as transpiration ceased and only evaporation from the soil-fresh biomass system was detected (Figure 7).

The overall trends of CO₂ and H₂O fluxes during the watermelon growing season were similar to those reported by [31,58] for fava beans cropped in the same environment under a similar fresh biomass incorporation process.

The patterns of the cumulative fluxes of assimilated and emitted CO₂, expressed in Figure 8 as photosynthesis and respiration, clearly illustrated that GPP consistently increased (negatively) from planting until watermelon fruit harvest, and then plateaued in correspondence with the green residual chopping, followed by a short gentle decrease due to reduced photosynthesis activity in non-irrigated plants. Conversely, total emitted CO₂ (R_eco) continuously increased as the soil maintained high carbon dioxide emissions post-harvest and after incorporating the green fresh residuals and biodegradable mulching films (Figure 8). The CO₂ emissions from the crop residuals left on the soil after harvesting until the end of measurements accounted for approximately 14% of the total.

3.3. The Carbon and Water Balances and the Water Use Efficiencies

Over the entire extended watermelon cultivation period of six months, from June to November, photosynthetic CO₂ uptake exceeded CO₂ losses from respiration, resulting in a net C input of 49.0 g C m⁻². By subtracting the C removed through harvesting (278.9 ± 70.1 g C m⁻²), the NBE yields a value of +229.9 g C m⁻² over the experimental period, indicating that this biodegradable mulched watermelon crop, managed with organic fertilization and residual green manure, is a net carbon source.

In terms of carbon balance in the field, all terms of Equation (5) are reported in Figure 9. The largest amount of imported organic C is due to the green manure of non-marketable watermelon fruits left on the field, followed by the green manure of watermelon plants and broccoli, and the C input from NEE. The biodegradable mulching film contributes 12% to the carbon input, while organic fertilization has the lowest contribution. The substantial amount of C removed by harvest reduced the C stored in the soil to 19.3 g C m⁻² over the investigated period. This finding is in accordance with results by [59,60] who reported increased carbon sustainability through conservative practices applied to horticultural water-consuming crops under Mediterranean conditions.

However, the directly measured ΔSOC from soil analysis of samples in the 0–0.30 m layer (352.2 ± 52.2 g C m⁻²) was much higher than the estimated C temporarily stored in the soil over the six-month experimental period (19.3 g C m⁻²). Conversely, this latter value is much lower than the uncertainties of the SOC measurement, being the 7% of the standard deviation of the carbon stock measured in the soil samples at the beginning of the trial (289.8 g C m⁻²). This result indicates the following: (i) it is extremely difficult to unambiguously establish the net C stored in the soil, as many input/output terms must be evaluated for a complete carbon balance [40,61,62]; (ii) C losses can be determined on a crop growth seasonal scale, but several years of continuous cultivations and green manure incorporation are necessary to detect a significant difference in soil carbon storage [40]; (iii) measurements of SOC are subject to great uncertainties, as indicated by large value spreads around the mean [63], likely linked to the spatial variability [64,65] and short-term time processes at the atmosphere-pedosphere interface [66], often making single point measurements ambiguous.

The water balance on an extended seasonal scale showed that the cumulative inputs were: precipitation, 139 mm; irrigation, 583 mm; and 5 mm of water supplied by the green manure of fresh materials (broccoli, watermelon plants and non-harvested fruits). Conversely, the outputs of evaporation and transpiration were 178 and 186 mm, respectively. These amounts, added to the water stored in the soil (272 mm), resulted in a significant amount of water lost through deep percolation (91 mm). This quantity represents 16% of the water supplied by irrigation. Similar values of deep percolation water losses were found by [67] in watermelon cultivation under a Mediterranean climate, whereas [68] found higher values of deep percolation in semi-arid continental conditions.

From the above-reported values of total gross primary production, net ecosystem exchange, and water terms, along with the marketable yield, it is possible to calculate all defined water use efficiencies. In the present case, we found WUE_ph = −1.38 g C kg H₂O⁻¹, WUE^’_ph = −0.70 g C kgH₂O⁻¹, WUE_env = −0.14 g C kg H₂O⁻¹. Any comparison with published values of eco-physiological water use efficiencies is challenging because they have not been previously published for the watermelon crop. However, the watermelon WUE^’_ph, in absolute value, is lower than the values reported by [39] for biomass food and non-food crops.

Regarding agronomic water productivity indicators, a comparison of our results with other values published in peer-reviewed papers (

Table 1. Mean agronomic and irrigation water use efficiencies (WUE_agro and WUE_irr) for watermelon cropped using plastic mulched and drip optimal irrigation.

Climate	WUEagro [kg m−3]	WUEirr [kg m−3]	References
Mediterranean		21.28	[5]
Mediterranean		26.76	[69]
Semi-arid	27.60	60.00	[70]
Warm-temperate-humid		33.40	[71]
Arid oasis desert	39.00		[72]
Mediterranean	38.38	15.82	Present work

) shows that our biodegradable mulched watermelon has WUE_agr values comparable to those found in plastic-mulched cultivations under semi-arid and arid climates. Concerning the WUE_irr, our value is lower than those found for watermelon cropped under similar Mediterranean climates.

4. Conclusions

In terms of C balance, despite a NEE of −49 g C m⁻², the watermelon cropping system became a net C source primarily due to substantial C removal through harvested fruits resulting in an NBE of 229.9 g C m⁻². The considerable amount of fresh biomass left in the field (broccoli, watermelon plants, and non-harvested fruits) contributed 85% of the organic C input. This, in conjunction with the biodegradable mulching film, allows the addition of C in the soil, although only a small percentage of the directly measured C stock could be explained by the C balance terms.

The results indicated that the mulching technique maintained soil and plants in optimal water conditions. However, the fixed irrigation schedule only partially matched the actual evapotranspiration needs, leading to water inefficiencies. Consequently, the water lost through deep percolation represents about one-sixth of the supplied water, which also has implications for carbon leaching. The water applied through the green manure of fresh biomass left in the field accounted for less than 1% of the total water supply.

Finally, these results show that the biodegradable mulched watermelon crop, when fully irrigated under Mediterranean conditions, exhibits limited sustainability from both carbon and water perspectives. Further research is needed to enhance the ecosystem impacts and sustainability of this practice. Despite the noted general improvement in WUE whit the use of biodegradable mulching films, information on carbon and water balances in such scenarios is scarce, especially in terms of the effects of film left on the soil surface and their time of decomposition.