Potential productivity of Arundo donax in two contrasting environments from medium-term experiments

ABSTRACT Marginal land exploitation devoted to versatile crops, such as giant reed, is related to the degree of land marginality. Two experiments were conducted in two sites of Sardinia (Italy) to investigate the effects of management on dry matter yield and water productivity (WP). Experiment 1 compared four crop evapotranspiration restoring treatments (100%, 75%, 50%, 25%, and a rainfed control) over 5 years. Experiment 2 lasted 7 years and investigated zero-input supply on crop response. Experiment 1, notwithstanding well-watered conditions (Irr100), showed a higher value of biomass production during the last year monitored, but yielding only 13.5 Mg DM ha−1 and with no significant difference from Irr75. Moreover, being equal the average dry matter yield, in the Irr100, we would save 3630 m3 of irrigation water by achieving the same WP recorded in the Irr25 (25% ETc restoring). Experiment 2, being equal rainfall amount, raised WP values that picked 8.2 kg m−3 during 2018. Results indicate that in less favourable soils as the Experiment 1, well-watered conditions do not ensure better crop growth and productivity, whereas in deep soils (Experiment 2) promising productive results, as well as ecosystem services, could be obtained without any input supply.


Introduction
Arundo donax L. (giant reed) is a perennial herbaceous plant worldwide known for its suitability to be used for different and several purposes (Zhang et al. 2021). Interest in cultivating giant reed is related to its use as a bioenergy crop for biogas (Corno et al. 2016;Ceotto et al. 2021) and second-generation bioethanol production (Zanetti et al. 2019;De Bari et al. 2020). Moreover, giant reed has gained increasing interest for the potential to immobilize a broad range of metals from the environment Danelli et al. 2021;Rai 2021), to decontaminate soils contaminated by red mud (Alshaal et al. 2013), and to reduce soil erosion (Fagnano et al. 2015). Furthermore, giant reed finds valuable uses as a non-wood plant fiber resource for paper and plant production (Shatalov and Pereira 2006;Abrantes et al. 2007), as unconventional material for the building sector (Vitrone et al. 2022), in sewage treatment (Calheiros et al. 2014;Toscano et al. 2015), and alternative herbal medicine (Tuttolomondo et al. 2014;Mir et al. 2018). From the environmental sustainability perspective, giant reed was demonstrated to have a positive influence on marginal lands by delivering an ecosystem service through soil organic carbon sequestration (Cattaneo et al. 2014;Allesina et al. 2018;Martani et al. 2021). Several authors have already indicated that perennial energy crops might contribute to soil carbon sequestration through cyclic addition of aboveground and belowground residues and reduced soil mechanical disturbance (Rowe et al. 2013;Agostini et al. 2015;Chimento and Amaducci 2015). For the above-mentioned reasons, the giant reed can be considered as one of the non-food crops potentially able to adapt in variable environments including marginal lands avoiding competition with lands devoted to food or feed production (Amaducci et al. 2017;Sacristán et al. 2021). Scientific literature reported that dry biomass production could range from 20 to 37 Mg ha −1 year −1 shifting from marginal soil to a fertile one, respectively (Nassi O Di Nasso et al. 2013a, 2013b. In Mediterranean areas, land marginality is related to a regime of precipitation and evaporative demand  interacting and combined at various levels with soil characteristics such as granulometry, bulk density, stoniness, and depth (Van Orshoven et al. 2013;Scordia and Cosentino 2019).
Therefore, it is important to increase the information so far available regarding giant reed performance in diversified marginal contexts.
Moreover, in the context of the Sardinia Region, there is a need to support new bio-based productions. In fact, in 2006 the only sugar factory operating was reconverted into a biomass plant power station following a major reform led to a greater market orientation of the EU's sugar policy (Regulations (EU) 2006/318 and 2006/320) and recently, relevant investments were promoted to reconvert the old petrochemical site to a bio-based chemical complex (Yazan et al. 2017).
Therefore, following an approach aimed at exploiting marginal land and pursue a sustainable use of natural (e.g. nitrogen and water) and available 2014 resources (e.g. crop residues; Cocco et al. 2014), resilient perennial bioenergy crops could represent in Mediterranean environments and in Sardinia a valuable source not only for bioenergy but also for producing multiple services, contributing at the development of a bio-economy, as it has been emphasized at EU's New Green Deal level (European Commission 2019).
In our study, we set up two independent experiments to provide complementary information to giant reed cropping and production at a relative different degree (thereinafter as severe vs less severe) of land marginality (Table 1). Experiment 1 was established in a location characterized by shallow, stony soils with low plant available water capacity, and mainly devoted to winter cereal or grassland cultivation; the site of Experiment 2 had less severe limitations with deeper soils, rich in organic matter. Our research questions aimed to fill the gap in knowledge regarding what happens if (i) nitrogen and water are supplied in a 'severe' marginal situation; (ii) water and N fertilizer are not supplied in a 'medium' or less severe marginal situation, in terms of productive traits and possible influence on carbon sequestration.  located in a flat terrain area of the Campidano Plain with a Mediterranean-type climate, with warm and dry summer and mild winter. Rainfall events are concentrated during the period from autumn to early spring. Based on a long-term series , total mean annual precipitation of the area is 434 mm, whereas mean temperature values range from 4.8°C in January to 33°C in August. The soil of Ussana site is a sandy-clay-loam, with a sand percentage greater than 50% (Table S1). The drainage is moderate and the stone percentage is about 10%. USDA classification is Typic Xerothent (Soil Survey Staff 1999). Experiment 2 (Exp. 2) was conducted at the experimental farm of National Research Council, Leccari-Sassari (40°45 ′N, 8°29′ E, 27 m a.s.l.) from 2012 to 2019. The experimental field is located in a flat area of the Nurra Plain, with a Mediterraneantype climate characterized by autumn-winter rains (553 mm, 50-year long-term series) and summer droughts. The annual mean temperature trends are in the range 7.2-25.9°C in February and July, respectively. The soil, classified as a Typic, Vertic, Aquic and Mollic Xerofluvents (Soil Survey Staff 1999), is sandy-clay-loam, alkaline with an adequate content of organic matter (Table S1). At this site, the soil depth exceeds 150 cm, without limitations for root development.

Experiment 1
Experiment 1 was laid out according to a randomized complete block design with three replicates. The experimental site was 1440 m 2 with a plot size of 96 m 2 . The treatments comprised five levels of irrigation based on crop evapotranspiration (Etc): 100% (Irr100, full Etc restoring), 75% (Irr75), 50% (Irr50), 25% (Irr25) of Etc restoring and a rainfed Control. Planting occurred in March 2006 by placing rhizomes of a local ecotype at a density of 0.90 × 1.1 m and at 6-cm depth. In the establishment year P 2 O 5 was distributed at a rate of 150 kg ha −1 (as diammonium phosphate) to all plots as a pre-plant fertilizer, whereas no P 2 O 5 was applied in the following years. The N fertilization was applied in the establishment year at a rate of 60 kg N ha −1 (diammonioum phosphate). In the following years, N was applied at the start of growth (within April) at a rate of 90 kg N ha −1 year −1 . Irrigation was provided by a drip irrigation system equipped with self-regulating drop emitters. The amount of water (V) to supply was calculated according to the maximum available soil water content within 60-cm soil depth, where most of the root is expected to grow and according to the formula V = 0.66 · (FC -WP) · φ · D · 103; where V is the water amount (mm); 0.66 is the easily available water; FC is the field capacity; WP is the wilting point, φ is the bulk density; and D is soil depth where is expected to find the most of the roots. Field capacity and wilting point were derived with pedotransfer functions based on soil parameters (Toth et al. 2015). Water was supplied when Etc equaled to V. For the calculation of daily Etc (Doorenbos and Pruitt 1977), Kc from Miscanthus × giganteus was used according to Cosentino et al. (2007). Weed control was performed by hoeing only during the establishment year. No crop pests or diseases were detected during the study.

Experiment 2
Rhizomes of a local accession of giant reed were extracted from the soil and used as propagation material in the experimental field. Tillage was carried out in two stages: first in November 2011 and next in March 2012, consisting of medium depth (30 cm) ploughing and harrowing with a field cultivator, respectively. After the second tillage, rhizomes were planted in three different sections of the same field in east-west rows (plot size: 100 m 2 ), each 100-cm wide, at a density of 2 rhizomes m -2 . Weed control was performed manually only during the establishment year. During the study, nor any plant protection event neither weeding was performed. Just before transplanting and therefore only in the establishment year, fertilization was applied with 40 and 90 kg ha −1 of N and P 2 O 5 , respectively.

Experiment 1
The meteorological data was measured at a weather station located 600 m away from the experimental site. The following biometric and productive traits have been registered during each harvest time, in a small sampling area (2 m 2 ): stalk height (cm), stalk density (stalk m −2 ) and aboveground dry biomass yield (Mg ha −1 ). Harvest occurred manually once a year in wintertime (February). Aboveground biomass production was recorded in the center of each plot, in a sub-plot harvestable area of 10 m 2 . Fresh sub-samples of biomass were kept in an oven at 65°C until a constant weight was reached to determine dry matter content of biomass. The irrigation water productivity (IWP) was calculated as the ratio of total dry matter production to the total volume of irrigation water consumed by the system (Pereira et al. 2012).

Experiment 2
The meteorological data for the 2012 and 2019 giant reed growing seasons were obtained from an in-situ weather station. Plant growth traits and yield crop parameters were determined in the central part of each plot at the end of the growing cycle. For each plot (1 m 2 of sub-sampled area) the number of stalks per unit area (stalks m −2 ) was determined. On 10 main stalks chosen randomly in each experimental area, stalk height (as the distance from the ground to the ligule of the upper expanded leaf of the main plant stalk), basal, medial and distal diameter of stalks (Vernier measurement), were measured and number of leaves and nodes counted. At the end of each growing cycle (February), aboveground biomass production was determined by cutting the stalks 5 cm above the ground level and weighing them to have the fresh matter yield per unit area. Plant samples were then partitioned into stalk and leaves (lamina), to determine the proportion of the different components on the aboveground biomass. Dry matter content was obtained by oven drying at 65°C until constant weight. The water productivity (WP) was calculated as the transformation efficiency of rainfall water through the cultivation system into dry matter yield (Pereira et al. 2012).
Soil was sampled in 2012 to assess initial soil conditions (Table S1) and at the end of the last crop cycle in 2019 to evaluate the potential trend of variation of soil organic matter, total nitrogen and C/ N. In each plot, three composite soil samples were collected at 0-30, 30-60 and 60-90-cm depths, respectively, and analysis was carried out in accordance with the standard methods of the Italian Soil Science Society (2000).

Statistical analysis
For Exp. 1, the Irrigation and the Year effects on biometric and yield-related parameters were tested by a two-way analysis of variance using R software (R Core Team 2012), given the normality of distributions (Shapiro and Wilks test) and the homogeneity of variances (Levene's mean-based test).
Year was considered as fixed factor in the analysis of variance model because of its higher potential effect as stand age in the case of perennial crops (Piepho et al. 2003) on irrigation treatments across the perennial crop cycle. Blocks and interaction with blocks were treated as random factors. In the case of significance of the Year and Irrigation effect, means were separated by the Tukey's post hoc test and the means were considered significant at P ≤ 0.05. For the Exp. 2, results were expressed as the mean ± standard errors of the mean. One-way analysis of variance was conducted using R package to identify the effect of Year on biomass yield and biometric traits, and the effect of soil depth over the difference between initial and final soil condition. After a significant F ratio was found, a Tukey's post hoc test was used for means separation. The differences between the means were considered significant at P ≤ 0.05.

Weather conditions
In Exp. 1, mean air temperatures increased from March to August, with minimum values above 6°C and maximum values above 15°C starting from March onward ( Figure S1). The rainfall distribution was typically concentrated during the autumn-winter season, then, usually, a long drought period occurred. The mean cumulative precipitation observed from 2006 to 2011 (474 mm) was about 30 mm below the long-term average. The lowest annual precipitation values were observed in 2007 (305 mm), the highest during 2010 (613 mm; Figure S1). In Exp. 2, mean air temperatures increased from March to August, with minimum values above 7°C and maximum values above 10°C starting from March onward ( Figure S2). The rainfall distribution typically showed a peak between autumn and early spring, then, usually, a long drought period occurred. The mean cumulative precipitation observed from 2012 to 2019 (483.6 mm) was about 70 mm below the long-term average. Summer 2017 was a particular dry season with only 1 mm from July to August ( Figure S2).

Experiment 1
Averaging over the 5-year study, 600 mm year −1 of irrigation water was distributed under Irr100 (full restoring of ETc) from early summer until September. Analysis of variance showed that the effect of irrigation treatment was statistically significant for stalks density, plant height, dry biomass production, relative humidity, and irrigation water productivity (Table S2). Moreover, for each variable, a significant two-way interaction Year × Irrigation was found. Stalks density resulted significantly affected by year and irrigation treatments (Table S2, Figure 1) and it was always significantly higher in Irr100 ranging from 16 (1 st year, 2007) to 9 (3 rd year, 2009) with an average value of 13 stalks m −2 . The lowest value was found in Control during 2009 harvesting (3 rd year) and it was equal to 4 stalks m −2 ( Figure 1). Each year, stalks under rainfed conditions (Control) were significantly shorter than stalks under Irr100 and Irr75 (Figure 2). Within each year, Irr100 and Irr75 did not significantly differed in terms of stalk height. Comparison among treatments within each year confirmed the rainfed Control to be the less productive option, mean values in the range 1.2 (3 rd year) -5.6 (5 th year) Mg DM ha −1 were observed (Figure 3). The highest value was recorded in Irr100 during the harvest of the last year monitored (5 th year) yielding 13.5 Mg DM ha −1 . In 2010, no significant difference emerged among watered treatments, that significantly differed only by Control one. The water treatments had significant effects on IWP (Table S2 and Figure 4), and the responses showed similar variation across the 5-year study (Figure 4).

Experiment 2
The number of stalks per area was significantly affected by years (P < 0.00233, Table S3). A marked and significantly different reduction was recorded in 2015 (9 stalks m −2 ) that resulted the lowest value observed across the whole lasting of the experiment (Figure 5). At the end of the first year, the stalk height reached 222.5 cm and it was also the shortest value, significantly different (P < 0.032, Table S3) from subsequent years, when stalk height ranged between 333.8 cm (2015) to 414.2 cm (2019). Stalks measured at basal level were thicker and larger than stalks measured in the middle and distal part (Figure 6), this trend was rather stable and not significantly correlated with stalk number per unit area. Statistically differences were found only at distal diameter level among years (Table S3; Figure 6). For number of leaves per stalk a statistically difference (P < 0.000269) was observed (Table  S3). In particular, a significant reduction was observed in 2015 and 2019 that recorded the lowest number of leaves per stalk yielding 10 and 6 number of leaves per stalk, respectively ( Figure S5). The plant dry matter content at harvest showed that the partitioning to stalk resulted in low but significant differences (P < 0.0221) among years (67% during 2013 significantly lower than other years) (Table S3, Figure S5). Two peaks of production were recorded, the first one during the 2014 growing season and the second during 2018 sampling (27.2 and 32.2 Mg ha −1 , respectively), with no significant difference between them (Figure 7). Stalk density influenced dry matter production mainly during the 2014, 2017 and 2018 growing seasons. Water productivity (WP) ranged between 1.4 kg m −3 in 2013 and 8.2 kg m −3 in 2018 (Figure 8). When cumulated growing season rainfall (March to February) was slightly lower than long-term rainfall (2014-15, 2015-16, and 2016-17 growing seasons) dry matter yield averaged 16.6 Mg ha −1 with WP values in the range 2.5-5.0 kg m −3 . During 2018-19 growing season, a substantial reduction in stalk density, presumably due to crop age was enough to drop dry matter yield and, as consequence, WP to 2.7 kg m −3 .
Soil organic matter content at the end of the experiment at 30 cm was 38% significantly higher than that of the initial soil conditions, and it was the only depth that successfully accrued soil organic matter over time (Figure 9). At 60 and at 90 cm we observed that soil organic matter was more likely to be maintained stable over time (Figure 9). At 90 cm depth, significant differences in soil total N were found compared with initial soil conditions. At 60 and at 90 cm the C/N ratio between the initial soil conditions and that in 2019 was not significantly different, whereas significant difference was found between the two periods at 30 cm, with the same pattern previously demonstrated for soil organic matter (Figure 9).

Discussions
It is acknowledged that Arundo donax is suitable to be grown in several pedo-climatic conditions across the sub-tropical and Mediterranean areas. However, a mosaic of distinctive pedoclimatic conditions features Mediterranean areas (Scordia and Cosentino 2019), even within the same region (Cervelli et al. 2020). Additionally, research findings that clearly show how crop responds when grown in a less severe marginal environment without external input vs in a severe situation but in a well-watered conditions and regularly supplied with fertilizers are still scarce or lacking. The results of our experiments lead to an advancement in knowledge on the giant reed dry matter yield throughout a mid-term period of growth (5 years in Exp. 1 and 7 years in Exp. 2) as influenced by growing conditions subjected to different degree of land marginality and provide indications on which is the best sustainable way to agronomical manage this crop to attain stable production over time. Besides, we demonstrated that fertilization and well-watering condition do not necessarily provide a higher response in dry matter yield, concluding that probably, what really accounts for dry matter yield is, more than irrigation and fertilization, soil depth, water retention capacity, annual rainfall amounts quite similar to long-term series even with variability for seasonal distribution of rainfall among years.

Experiment 1
In our study, the interaction between Year × Irrigation treatments was nearly always statistically significant, which suggested different performance of giant reed crop stand age to the water regime change. Despite well-watered treatments provided restoring 75% and 100% of maximum evapotranspiration, the highest stalk height recorded (371 cm during the 4 th year harvesting) was closer to the 50% of maximum evapotranspiration treatment of Cosentino et al. (2014) being equal the stand age. Indeed, the latter authors, who worked in Southern Italy (Sicily), found 410 cm stalk height for the 100% ETm treatment and 353.2 cm for the 50% ETm treatment. The dry matter production recorded over the 5-year study did not differ between the two treatments which returned 75% and 100% of the maximum evapotranspiration. Thus, strongly different values, if compared to those reported in the literature being equal the treatment (restoring 100% of ETm) and crop age (5 th year), were found in our study. Mean dry biomass production observed by Mantineo et al. (2009) for a giant reed stand watered at 75% ETm in Southern Italy was 42 Mg DM ha −1 (2 nd and 3 rd stand age, 100 kg N ha −1 fertilization). Cosentino et al. (2014) recorded 28.8 Mg DM ha −1 (3 rd and 4 th stand age, 120 kg N ha −1 fertilization) for another irrigated at 100% ETm restoration experiment carried out in Southern Italy, confirming the importance of other crop production factors, as well as the effect of particular environment conditions, in determining the final crop dry matter production. Furthermore, maximum dry matter production values observed in Exp. 1 were highly far from those recorded in Exp. 2, the crop age being equal, but in that case (Exp. 2) in rainfed condition. This latter finding validates   similar results recorded on cardoon perennial bioenergy crop tested in another experiment (Deligios et al. 2017), but at the same site and under low input conditions. It is worth noting, however, that the extreme high temperatures conditions recorded in 2008 at Exp. 1 (3 rd year of the stand; ARPA Sardegna, 2009;http://www.sar.sardegna.it/pubblicazioni/periodiche/analisi_10.2007_09.2008 influenced the final dry matter production of the crop (harvest of 2009), and this was a common factor to all the treatments, even when the full restore of maximum evapotranspiration occurred. Apparently, the 3 rd and the 4 th year had a similar thermometric trend, indeed, the period between January and September showed mean air temperature of +1.9°C in 2008 (around +3°C in January, March, April and July) and of +1.6°C in 2009 with respect to the long-term series. In contrast, for the same period, the pluviometric trend was 3-fold higher in 2009 (460.5 mm in 2009 vs 169.6 mm in 2008). Even if this latter data might appear as a remarkable one only for the Control treatment, however, probably the pluviometric trend recorded during 2009 promoted cooler microclimatic conditions at canopy level (despite the anomalous thermometric trend) allowing regular photosynthetic process and dry biomass production. Indeed, it is well documented that in C3 plants, as giant reed is, the increase in leaf temperature negatively affected Rubisco activase activity and quantum yield reducing consequently net assimilation rate and biomass production (Spreitzer et al. 2003;Taiz and Salsbury 2012). Even if Jampeetong et al. (2020) study showed that giant reed had a high photosynthesis capacity at high temperature regimes (36/32° day-night temperatures), their monitoring period lasted only 4 weeks and the experiment was carried out under constant and controlled environment, but it is acknowledged heat stress is a complex function of intensity, duration, and the rate of the increase in air temperature (Irmak 2016). The lack of significance in WP means that the   same amount of dry biomass yield fully depends on the amount of irrigation water used (Pereira et al. 2012). The irrigation water productivity was higher in Irr25 treatment because dry matter yield being equal supported a significant contraction in irrigation water used. Overall, being equal the water productivity of Irr25, on average 3630 m 3 (60%) of water could be saved in the Irr100 to achieve the same yields. This means that, also in environmental limiting conditions, producers could opt for a significant saving in water supply to the crop to maximize and stabilize over seasons the crop yield, to obtain satisfactory yields and ensure, at the same time, a better environmental-friendly management of water. Besides the improvement of yield increasing effect, the economic benefit of irrigation is another factor need to be considered. Consistently, Scordia et al. (2019) stated that the possibility of using very unproductive lands remains questionable if even yield-increasing measures (irrigation, fertilizer application, etc.) failed to realize actual yields. Accordingly, neither the exploitation of a local ecotype (surely adapted to the site-specific land marginality level) nor irrigation water supply resulted a winning strategy in our study.

Experiment 2
Stalks density and dry matter yield values resulted positively correlated to the number of stalks per unit area. However, with deep soils, as was the one considered in Exp. 2, starting by an initial number of stalks per hectare of 123.000, which corresponded 8.8 Mg DM ha −1 , we observed an increase rate in stalks per unit area over 3.300 year −1 , until the final stalks number of 147.000 stalks ha −1 that we recorded during the last sampling (18.7 Mg DM ha −1 ). As consequence and consistently with other studies focused to find the optimal plant density for giant reed and bearing in mind the rate of stalks density increase that we observed in Exp. 2 and the typical plasticity of the species in producing tillers, it is possible to state that initial establishment might also be reduced, as tested by Alexopoulou et al. (2015) in Northern Italy, with positive economic advantages for the farmers in lowering initial establishment costs per hectare (Testa et al. 2016). Consistently with scientific literature (Nassi O Di Nasso et al. 2010;Bonfante et al. 2017), our results showed that, excluding the first year of planting and establishment during which an expected lower biomass yield was recorded, and the third year, during which an occasional pest attack by Ostrinia sp. caused a yield drop by affecting stalks density and number of leaves per stalk, dry matter yield was quite stable (19.5 Mg DM ha −1 averaging over 7-year study). Even if it might seem that average dry matter yield was low when compared to the broad range reported in literature (Zhang et al. 2021), we should underscore that dry matter yield was attained under rainfed condition and unfertilized management. The pattern of dry matter production across the 7-year study resulted almost similar to findings of Fagnano et al. (2015), which however showed a lower potential of giant reed (14.6 Mg DM ha −1 ) across the first seven years of stand in comparison with our results. It is worth noting that also in this latter study giant reed was grown without irrigation and in a Mediterranean-type climate, but comparing two nitrogen doses treatments (50 vs 100 kg ha −1 year −1 ). Moreover, the peak of dry matter yield (32.2 Mg DM ha −1 ) did not differ greatly from the peak of dry matter yield reported by Angelini et al. (37.5 Mg DM ha −1 as average of fertilized and unfertilized treatments) in their longterm experiment (2005). Presumably, soil properties at Exp. 2 might have had the main weight in supporting the productive performance at the given cumulative precipitation of the site and without N fertilizer supply. What lead us to state this, is the consideration that even if cumulative annual rainfall resulted well under 550 mm in four out of seven growing seasons, it did not cause any significant negative effects on the biomass yield as well as on the final amount of soil organic matter and total N in the 0-30 cm and 60-90 cm soil layer, respectively. Thus, stable yields might be reached in zero-input management systems. However, the conversion toward a mixed agro-energetic system raises new questions on the sustainability of bioenergy species cultivation. Because of this, rather than leading to a generalized increase in using external inputs (e.g. water, fertilizers, and pesticides) with related issues of pollution and water scarcity (Scarlat et al. 2015), it might be preferable to drive ecological transition by recovering abandoned land devoting the most suitable one to resilient perennial bioenergy crops such as giant reed is. Taken together, our results suggest to carefully identify the most suitable land where giant reed might be grown at minimum input levels and to avoid generalization indicating an overall proven adaptation of this species to the Mediterranean environments. Further, a key role is played by the marginal profits achieved by farmers who grow bioenergy crops even on marginal land. Mantziaris et al. (2017), considering the option of farm gate delivery, observed that giant reed generated the highest positive gross profits with respect to poplar and miscanthus. Moreover, they assessed that giant reed can partially replace durum wheat in low input lands. Soldatos (2015) analyzing the economic aspects of bioenergy production from perennial grasses in marginal lands of Southern Europe concluded that giant reed appears to be the most profitable and promising crop with respect to switchgrass and miscanthus. Specifically, this is due to the biomass sales prices needed to reach the break-even point in the future market. Profitability by energy crops in marginal land should be carefully verified on the basis of some factors such as the site location (in particular the type of soil and the availability of water) affects production costs and agronomic management as highlighted by Giannoulis et al. (2014) on switchgrass in Greece. Moreover, according to Pereira et al. (2012) improving crop yields is markedly related to the adaptation of the crop ecotype to the cropping environment. However, a crop variety with a higher WUE than another variety has the potential for using less water than the second when achieving the same yield. Therefore, discussing how improving WP could lead to water saving in irrigation requires the consideration of several factors, namely: (a) the contribution of rainfall to satisfy crop water requirements, (b) the management and technologies of irrigation, (c) the agronomic practices, (d) the adaptability of the crop variety to the environment, and (e) the water use efficiency of the crop and variety under consideration.
Despite the relevant amount of aboveground biomass (on average 19.4 Mg DM ha −1 as in Figure 5) and nitrogen (on average 94.5 kg N ha −1 , data not shown) removed each year as a result of harvesting, soil organic matter content and total N remained rather stable over time and no soil depletion was detected. It appears therefore that giant reed cropping may contribute to C sequestration in the mid-term as already argued by different authors (Nassi O Di Nasso et al. 2013b;Cattaneo et al. 2014;Bosco et al. 2016). This valuable increase (+38% of soil organic matter, Figure 5) is consistent with crop management; unlike annual crops, perennial grasses only need tillage in the year of establishment. The ecological advantages, in terms of ecosystem services, of the long period without tillage are reduced risk of soil erosion and soil C matter conservation (Martani et al. 2021). Interestingly, it has been demonstrated that giant reed is colonized by N 2 -fixing bacteria and therefore might episodically assimilate nitrogen fixed by non-nodulating bacteria living in the rhizosphere (Hu et al. 2018). Presumably, this additional N source might have contributed to the above-mentioned quite stable soil total N content over time at Exp. 2 ( Figure 5). Taken together, the results from both medium-term experiments supply complementary and novel information at different Mediterranean pedoclimatic conditions and giant reed management. Additionally, our findings support additional insights on giant reed based on wide ranges of biometric traits, potential biomass yields, WP values as well on its proven ability to maintain over time both stable biomass yields and soil organic matter content under zero-input management systems.

Conclusions
This study, from two experiments under Mediterranean marginal land conditions, highlighted effects of different management intensities on yield and sustainability of giant reed crop. Results from Exp. 1 indicated that well-watered conditions in marginal land did not ensure satisfactory crop growth and production if compared with results attained in Exp. 2 where no external input was supplied during the seven-year experiment. Indeed, findings from Exp. 2, evidenced that in deep soils, promising results in terms of both biomass yields and ecosystem services could be obtained without water and N input. Our results could also be useful for allocating the right crop in the right place within the range of distinctive pedoclimatic conditions of Mediterranean environment. LCA-specific analyses will be needed for a more precise evaluation of the environmental sustainability of the studied options.