Introduction
The presence of weeds in maize (Zea mays L.) fields is a major concern for maize growers because their presence diminishes yield and their removal is time-consuming and requires considerable resources. Specifically, the estimated loss in maize production due to weeds is 32%, and this loss is greater than that caused by pests (18%) and pathogens (15%) (Oerke and Dehne Reference Oerke and Dehne2004). Weeds can also harbor crop pests and diseases that need to be controlled, and consequently increase production costs. Moreover, the presence of weeds makes harvesting more difficult and devalues the crop by reducing its quality. In Spain, maize was cultivated in 2017 in Mediterranean semiarid conditions under flood or sprinkler irrigation on about 330,000 ha, most of which are located in Castilla y León (26%), Aragón (25%), Extremadura (14%), and Cataluña (11%). The annual maize production in Spain is about 4 × 109 kg with an average yield of 10,000 kg ha−1 (MAPA 2018). Chemical control is the most widely used method for controlling weeds in maize production. In Europe, herbicides are used to control weeds in greater than 90% of maize cultivation areas (Meissle et al. Reference Meissle, Mouron, Musa, Bigler, Pons, Vasileiadis, Otto, Antichi, Kiss, Pálinkás, Dorner, van der Weide, Groten, Czembor and Adamczyk2010). In general, the currently used herbicides are highly effective, very reliable, and provide broad-spectrum control of weeds without damaging the crop.
Nevertheless, resistance to commonly used herbicides is an emerging problem. Currently, there are fields in the maize cropping areas in Spain in which weed populations of the dicotyledonous species of Amaranthus, Chenopodium, and Solanum genera present problems with control when conventionally used photosystem II–inhibiting herbicides such as terbuthylazine are used. On the other hand, the monocotyledonous weed species of Echinochloa, Sorghum, Setaria, and Digitaria genera are becoming resistant to acetolactate synthase (ALS) inhibitors (CPRH 2018). Resistance to ALS- and acetyl CoA carboxylase–inhibiting herbicides, which are widely used for controlling weeds in other annual crops, is also increasing, as evidenced by the resistance of blackgrass (Alopecurus myosuroides Huds.) and Italian ryegrass [Lolium perenne L. ssp. multiflorum (Lam.) Husnot.] in the United Kingdom (Hicks et al. Reference Hicks, Comont, Coutts, Crook, Hull, Norris, Neve, Childs and Freckleton2018; Hull et al. Reference Hull, Tatnell, Cook and Moss2014) and ripgut brome (Bromus diandrus Roth) (Escorial et al. Reference Escorial, Loureiro, Rodriguez-Garcia and Chueca2011) or rigid ryegrass (Lolium rigidum Gaudin) in Spain (Loureiro et al. Reference Loureiro, Escorial, Hernández-Plaza, González-Andújar and Chueca2017). Accordingly, farmers are now compelled to use other strategies for weed control before sowing the crop, such as false seedbeds and delayed sowing to promote the early emergence of weed (van der Weide and Bleeker Reference Van der Weide and Bleeker1998), commonly followed by glyphosate application. However, modeling studies to compare the rates of evolution of glyphosate resistance under crop rotation and annual use of glyphosate pre-sowing have identified increased glyphosate use on stale seedbeds, often in systems with reduced or no-tillage, as a major driver for evolution of glyphosate resistance in Australian populations of L. rigidum (Neve et al. Reference Neve, Diggle, Smith and Powles2003).
Of all the herbicides, glyphosate is the most widely used globally, because it has high efficacy against a broad spectrum of weeds (Duke and Powles Reference Duke and Powles2008). Glyphosate use enables the application of new crop production systems, such as conservation agriculture and no-till practices, and new weed management approaches that rely on the cultivation of glyphosate-resistant (GR) crops. However, the cultivation of GR crops increases glyphosate use, which can result in less use of other herbicides, an increased number of weed species that cannot be controlled by glyphosate, weed shifts, and weed resistance to glyphosate (Bonny Reference Bonny2016; García-Ruiz et al. Reference García-Ruiz, Loureiro, Farinós, Gómez, Gutiérrez, Sánchez, Escorial, Ortego, Chueca and Castanera2018; Johnson et al. Reference Johnson, Davis, Kruger and Weller2009). Although glyphosate is viewed as a low-risk herbicide with regard to the evolution of resistance, the emergence of glyphosate-resistant weed populations, especially in monocultures with limited rotation or minimal tillage, could threaten the utility of both glyphosate and GR crops. The results of several surveys among American scientists and farmers revealed that 80% of respondents attributed shifts in the weed species to the use of GR crops (Culpepper Reference Culpepper2006; Gibson et al. Reference Gibson, Johnson and Hillger2006; Johnson and Gibson Reference Johnson and Gibson2006). It has also been reported that the extensive and continuous use of glyphosate can promote glyphosate resistance in weeds (Heap and Duke Reference Heap and Duke2017). More recently, Heap (Reference Heap2019) reported that 43 different weed species had developed resistance to glyphosate, although the use of a GR crop did not always account for this development. The infestation of cultivated crops with glyphosate-resistant Amaranthus species, especially Palmer amaranth (Amaranthus palmeri S. Watson), has become one of the biggest weed problems in U.S. agriculture (WSSA 2016).
The reduced herbicide rates to control weeds have been applied in more than 50% of the areas cultivated in maize in the Netherlands, and more than 80% of the maize cultivation areas in Denmark, Germany, and France (Meissle et al. Reference Meissle, Mouron, Musa, Bigler, Pons, Vasileiadis, Otto, Antichi, Kiss, Pálinkás, Dorner, van der Weide, Groten, Czembor and Adamczyk2010). However, the use of low herbicide doses can result in the rapid evolution of herbicide resistance because of the development of non–target site resistance (Manalil et al. Reference Manalil, Renton and Powles2011; Norsworthy et al. Reference Norsworthy, Ward, Shaw, Llewellyn, Nichols, Webster, Bradley, Frisvold, Powles, Burgos, Witt and Barrett2012). Neve and Powles (Reference Neve and Powles2005) claimed that low application rates of a herbicide could accelerate the evolution of herbicide resistance in a weed population with broad genetic diversity, such as L. rigidum in Australia. Collavo and Sattin (Reference Collavo and Sattin2014) also reported the development of glyphosate resistance in Lolium spp. due to the continuous low-dose application of glyphosate to cereals in Italy, which could be attributed to both target-site and non–target site mechanisms.
Determining the sensitivity of target pests to an active substance is advantageous, because it gives baseline information about the level of resistance to a particular plant protection product in a pest population (EPPO 2015). Sensitivity data also enable comparisons to be made between the same and different populations at various times to detect any sensitivity shifts and resistance development (Moss Reference Moss2001). These data are especially important for detecting non–target site resistance when less sensitive weed populations may be selected and resistance slowly evolves in each subsequent generation (Gressel Reference Gressel2011). Differential sensitivity to glyphosate has been identified in several dicotyledonous weed species, such as common lambsquarters (Chenopodium album L.) (Westhoven et al. Reference Westhoven, Kruger, Gerber, Stachler, Loux and Johnson2008), Amaranthus spp. (Norsworthy et al. Reference Norsworthy, Griffith, Scott, Smith and Oliver2008; Patzoldt et al. Reference Patzoldt, Tranel and Hager2002; Smith and Hallett Reference Smith and Hallett2006; Volenberg et al. Reference Volenberg, Tranel, Hager and Patzoldt2007), Erigeron spp. (González-Torralva et al. Reference González-Torralva, Cruz-Hipólito, Bastida, Muelleder, Smeda and De Prado2010), and kochia [Bassia scoparia (L.) A. J. Scott] (Waite et al. Reference Waite, Thompson, Peterson, Currie, Olson and Stahlman2013). Differential glyphosate sensitivity has also been identified in monocotyledonous weed species, such as quackgrass [Elymus repens (L.) Gould] (Espeby et al. Reference Espeby, Fogelfors, Sjödal and Milberg2014), A. myosuroides (Davies and Neve Reference Davies and Neve2017), L. rigidum, and B. diandrus (Barroso et al. Reference Barroso, Loureiro, Escorial and Chueca2010).
Against this background, we undertook an investigation whose aims were (1) to assess the response to glyphosate of the most commonly found weeds in the maize fields of Castilla y León and (2) to generate sensitivity data to establish the basis for monitoring the response to glyphosate across a range of weed populations before this herbicide is used extensively in the maize fields of this region.
Materials and Methods
Plant Material
The investigation comprised glyphosate dose–response assays that used seeds from 85 different populations of eight dicotyledonous or broadleaf weed species and five monocotyledonous or grass weed species. The eight dicotyledonous weed species were velvetleaf (Abutilon theophrasti Medik.), five populations; redroot pigweed (Amaranthus retroflexus L.), nine populations; C. album, 10 populations; jimsonweed (Datura stramonium L.), eight populations; common purslane (Portulaca oleracea L.), 10 populations; black nightshade (Solanum nigrum L.), five populations; two species of Xanthium, namely spiny cocklebur (Xanthium spinosum L.), four populations, and common cocklebur (Xanthium strumarium L.), six populations. The five monocotyledonous weed species were large crabgrass [Digitaria sanguinalis (L.) Scop.], 10 populations; barnyardgrass [Echinochloa crus-galli (L.) P. Beauv.], five populations; and three Setaria spp., namely adherent bristlegrass [Setaria adhaerens (Forssk.) Chiov.], five populations, bristly foxtail [Setaria verticillata (L.) P. Beauv.], three populations, and green foxtail [Setaria viridis (L.) P. Beauv.], five populations.
Seed Sampling
Seeds were collected in field surveys conducted randomly during 2013 and 2014 in maize fields of provinces with the largest areas of maize cultivation in Castilla y León, namely León (64,547 ha), Zamora (18,507 ha), Salamanca (18,230 ha), and Valladolid (9,082 ha) (Figure 1). Sampling was done by extensive driving through the region and stopping at 10-km intervals to sample the nearest maize area. The sample sites were georeferenced using a global positioning system. A total of 59 field sites were visited. At each site, mature seeds from 25 to 50 plants that were randomly selected from different patches in the maize field were collected for each weed species. The seed from each weed species in the same maize field was bulked to form a population. Each seed sample contained at least 10,000 mature seeds. The seeds from each weed sample were placed in paper bags, dried at room temperature in the laboratory, manually cleaned and threshed, and stored at room temperature until use. One hundred and seventy-six seed samples from 13 weed species were collected. Only 87 weed samples with good germination (> 70%) with a minimum of five and a maximum of 10 populations per species were used in the study. We are unable to guarantee that the sampled weed populations were not previously exposed to glyphosate, because this herbicide is one of the most widely used herbicides in Spain. However, we assumed that the sampled populations were not exposed to glyphosate, because it is not commonly used in conventional maize farming.
Figure 1. Percentage of the maize cropping area in 2016 in Castilla León, Spain. Data were obtained from the Agricultural Statistics, Studies and Planning Service of the Department of Agriculture and Livestock of the Regional Government of Castilla y Léon, 2016.
Glyphosate Dose–response Assays
The glyphosate dose–response assays were conducted in a growth chamber under a 16-h photoperiod and 300 μE m−2 s−1 photosynthetically active radiation and 8 h of darkness at 30 ± 2 C/16 ± 1 C (day/night).
The seeds from each population were first pre-germinated in trays, and the germinated seedlings were then transplanted at an early seedling stage to 200-ml plastic pots filled with a 75% soil:mulch:sand (1:1:1) and 25% vermiculite mixture, at a rate of 3 uniform seedlings per pot. For Xanthium species, the seeds were sown directly in the plastic pots at a rate of 1 plant per pot. When the plantlets of the monocotyledonous weeds were at the 2- to 3-leaf stage (BBCH 12-13) or the plantlets of the dicotyledonous weeds were at the 2- to 4-leaf stage (BBCH 12-14), glyphosate (Roundup®, 360 g ai L−1, Monsanto Agricultura, Madrid, España) was applied at doses of 0, 16.8, 33.6, 67.5, 135, 270, and 1,080 g ai ha−1. Three replicates of five pots were made for each population and dose, and each dose–response assay was repeated twice. The glyphosate treatments were applied using an automatic sprayer (Devries Manufacturing, Hollandale, MN, USA) equipped with a TeeJet® 8002-E flat-fan nozzle (TeeJet Technologies, Orléans, France) that was calibrated to spray 175 L ha−1 at 130 kPa.
Once treated, the plants were returned to the growth chamber and watered as required throughout the experiment. At 15 d after treatment (DAT) for all weed species, the aboveground plant parts were first cut down and weighed, and then dried in an oven at 80 C for 48 h, and weighed again. For the development of herbicide dose–response curves, doses should cover the whole range of plant responses, from almost no apparent effects to complete kill of the plants, so the aboveground fresh weight for the Xanthium spp. was measured at 21 DAT to ensure that the full effects of the herbicide were visible.
Data Analysis
For determining the dose–response curve of each weed species’ population for the different glyphosate doses, the dry-weight parameter was first transformed to a percentage of the untreated control and a log-logistic model (Seefeldt et al. Reference Seefeldt, Jensen and Fuerst1995) was then fit to estimate I50 values (the effective dose for 50% growth reduction = GR50) according to Equation 1.
$$y = C + {\rm{ }}\left[ {\left( {D - C} \right)/\left( {1{\rm{ }} + {\rm{ }}{{\left( {x/{\rm{ }}{I_{50}}} \right)}^b}} \right)} \right]$$
where C is the lower limit and corresponds to the mean response at highest glyphosate dose, D is the upper limit and corresponds to the response of the control, and b is the slope of the curve around the GR50. These parameters were estimated by curve fitting with an iterative adjustment approach using the Table Curve® 2D program v. 5.01 (Systat Software, San José, CA, USA). A dose–response curve for each weed species was then generated using the data from all populations, and the mean GR50 value (baseline) for each weed species was estimated. The 95% confidence intervals (CI95%) for GR50 were calculated. The ratio between the GR50 values of the least and most sensitive populations was used to determine the SI50 (sensitivity index at 50% efficacy) for each weed species.
The percentage values of dry weight after herbicide treatment were arcsine square-root transformed before a two-way ANOVA using the general linear model procedure. The population effect was considered as the random effect, and glyphosate dose was considered to be the fixed effect. When the F-test was significant at P = 0.05, the mean dry weight of the populations and glyphosate doses were compared using the Newman-Keuls test. All statistical analyses were done using computerized statistical software (Statgraphics Centurion XVI.II, StatPoint, Herndon, VA, USA).
Results and Discussion
Establishing the baseline sensitivity of a weed population to herbicides is critical for monitoring the development of herbicide resistance and managing this resistance in weed populations (Moss Reference Moss2001; Paterson et al. Reference Paterson, Shenton and Straszewski2002; Ulber et al. Reference Ulber, Nordmeyer and Zwerger2013). Sensitivity data for a particular herbicide may be considered as a baseline when they are obtained from a weed population that has not been previously exposed to that herbicide or to herbicides with the same mode of action (EPPO 2015). Although GR crops have not been authorized for cultivation in the European Union, glyphosate is the most commonly used herbicide (Benbrook Reference Benbrook2016), and also in Spain (MAPA 2013). The use of glyphosate is mainly associated with reduced-tillage or no-till farming systems (Wiese et al. Reference Wiese, Schulte, Theuvsen and Steinmann2018).
The weed species we selected for our surveys are among the most prevalent weed species in maize-growing regions in Spain (San Martín et al. Reference San Martín, Andújar, Fernández-Quintanilla and Dorado2015) and elsewhere in Europe (Dewar Reference Dewar2009; Jensen et al. Reference Jensen, Bibard, Czembor, Dumitru, Foucart, Froud-Williams, Jensen, Saavedra, Sattin, Soukup, Palou, Thibord, Voegler and Kudsk2011). Analyses of glyphosate dose–response curves were performed separately for each weed population of the different species of weeds, and the GR50 values were determined. The GR50 values were used as a measure of sensitivity. The applied doses were appropriate for describing the dose–response curves for all weed species. Table 1 displays the GR50 values for each population of the dicotyledonous weed species, and Table 2 displays the GR50 values for each population of the monocotyledonous weed species. Our results reveal that all populations of the weed species selected to determine glyphosate sensitivity are susceptible to this herbicide. The GR50 values ranged from 4 g ai ha−1 (one population of D. stramonium) to 83 g ai ha−1 (one population of A. theophrasti), and both these values are much lower than the recommended glyphosate dose of 540 g ai ha−1. The GR50 values for the different populations of the same species were relatively similar. Table 3 displays the SI50, which is a measure of the variability of the response among the weed populations. For dicotyledonous weed species, the SI50 values showed a 1.4- to 3.3-fold difference in sensitivity, and there was a 1.4- to 2.4-fold difference in sensitivity for monocotyledonous weed species (Table 3).
Table 1. GR50 values for each population of the selected dicotyledonous weed species.
a The identification numbers of populations correspond with their numbers in the collection of the Weed Control Group of the Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria (INIA). The GR50 values are displayed as mean ± standard error (SE) and were estimated by the log-logistic equation used to calculate the glyphosate dose that caused a 50% growth reduction in the bioassays conducted under controlled growth chamber conditions.
Table 2. GR50 values for each population of the selected monocotyledonous weed species.
a The identification number of populations corresponds with their numbers in the collection of the Weed Control Group of the Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria (INIA). The GR50 values are displayed as mean ± standard error (SE) and were estimated by the log-logistic equation used to calculate the glyphosate dose that caused a 50% growth reduction in the bioassays conducted under controlled growth chamber conditions.
Table 3. Indices of glyphosate sensitivity of selected weed species. a
a The SI50 (sensitivity index at 50% growth reduction) values for each weed species are displayed as the ratio between the GR50 values of the least and most sensitive populations. The SI50b values for each weed species were calculated as the ratio between the GR50 value of the least sensitive population and the baseline GR50 value estimated for a range of susceptible populations of the same species.
For monocotyledonous weed species, the ANOVA results revealed no significant differences in glyphosate response among populations (P > 0.05). For the dicotyledonous weed species A. theophrasti, D. stramonium, and S. nigrum and the two Xanthium species, the ANOVA did not show any significant differences (P > 0.05) in the glyphosate response among populations. We found significant differences in sensitivity among the studied populations of A. retroflexus (F (9,124) = 3.69, P = 0.0004) and C. album (F (9,124) = 3.33, P = 0.001). However, the SI50 values for these species were 2.1 and 1.6 for A. retroflexus and C. album, respectively (Table 3). These SI50 values are consistent for susceptible weed populations, as small resistance indices (< 3) can occur among susceptible populations due to natural variation in intrapopulation sensitivity to herbicides (EHRAC 2017; Espeby et al. Reference Espeby, Fogelfors and Milberg2011; Patzoldt et al. Reference Patzoldt, Tranel and Hager2002; Schulz et al. Reference Schulz, Mathiassen and de Mol2014). However, this variation can also be due to differences in the selection pressure exerted through the recurrent use of the same herbicide or mode of action (Claerhout et al. Reference Claerhout, Reheul and De Cauwer2015; Kniss et al. Reference Kniss, Miller, Westra and Wilson2007). It could be possible that those weed populations with the highest GR50 values may be in a more advanced stage of selection for glyphosate resistance. It has been reported that the rate at which herbicide resistance evolves in weed populations is influenced by biological and genetic factors, which are inherent to each weed species, and by selection pressure, which can be manipulated by the cropping system and the weed management strategy (Harker Reference Harker2013).
The baseline GR50 values (GR50b) of glyphosate sensitivity for the main dicotyledonous and monocotyledonous weed species are displayed in Figures 2 and 3, respectively. The GR50b values for the dicotyledonous weed species ranged from 10.25 g ai ha−1 (CI 95% = 9.08 to 11.42) for D. stramonium, the most glyphosate-sensitive species, to 53.23 g ai ha−1 (CI 95% = 41.96 to 64.51) for A. theophrasti, the least sensitive species (Figure 2). We must consider that the evaluation of the herbicide treatment for the Xanthium spp. was extended by 1 wk, so despite being a similar response, it might not be comparable with responses of the rest of the weed species. For monocotyledonous weed species, the GR50b values ranged from 16.05 g ai ha−1 (CI 95% = 13.22 to 18.88) for S. verticillata, the most sensitive species, to 66.34 g ai ha−1 (CI 95% = 59.86 to 72.83) for E. crus-galli, the least sensitive species (Figure 3).
Figure 2. Effect of glyphosate on the growth of eight dicotyledonous weed species. The black line represents the dose–response curve fit to the mean aboveground dry biomass values in response to increasing glyphosate doses from all the populations assessed in each species. Dotted gray lines represent the 95% confidence interval (CI) for the dose.
Figure 3. Effect of glyphosate on the growth of five monocotyledonous weed species. The black line represents the dose–response curve fit to the mean aboveground dry biomass values in response to increasing glyphosate doses from all the populations assessed in each species. Dotted gray lines represent the 95% confidence interval (CI) for the dose.
When the baseline GR50b values for these species (instead of the GR50 value of the most sensitive population) were used to calculate the sensitivity index among populations, the differences in glyphosate sensitivity diminished to a maximum of 1.6-fold instead of 3.3-fold for dicotyledonous weed species (Table 3). When this calculation was done for monocotyledonous weed species, the differences in glyphosate sensitivity diminished to a maximum of 1.2-fold, instead of 2.4-fold (Table 3). Therefore, baseline data should take into account the natural variation of the sensitivity of weed populations and may be a more useful parameter than the value of the most sensitive population for establishing sensitivity indices.
The GR50 values for the response to glyphosate estimated in this study are consistent with those reported in other studies. Tharp et al. (Reference Tharp, Schabenberger and Kells1999) studied the response of several annual weed species to glyphosate and found GR50 values of 96 g ai ha−1 for giant foxtail (Setaria faberi Herrm.), 120 g ai ha−1 for C. album and D. sanguinalis, and 160 g ai ha−1 for E. crus-galli. Boutin et al. (Reference Boutin, Elmegaard and Kjær2004) reported GR50 values for the response to glyphosate at 29.2 and 18.5 g ai ha−1 for cornflower (Centaurea cyanus L.) and corn poppy (Papaver rhoeas L.), respectively, while White and Boutin (Reference White and Boutin2007) reported a GR50 of 77 g ai ha−1 for S. nigrum. Our GR50 values also agree with those estimated in a Danish study on the sensitivity to glyphosate of six non-target plant species and 10 crop species (Strandberg et al. Reference Strandberg, Mathiassen, Bruus, Kjær, Damgaard, Andersen, Bossi, Løfstrøm, Larsen, Bak and Kudsk2012). Those authors reported GR50 values that ranged from 29 to 130.9 g ai ha−1 for the non-target weeds common yarrow (Achillea millefolium L.) and herb-robert (Geranium robertianum L.), respectively. For crop species, they reported GR50 values that ranged from 1.6 g ai ha−1 for common sunflower (Helianthus annuus L.) to 84.6 g ai ha−1 for onion (Allium cepa L.). However, it should be noted that GR50 values depend to a large extent on the experimental conditions. Tharp et al. (Reference Tharp, Schabenberger and Kells1999) showed that the GR50 values for the response to glyphosate of A. theophrasti varied from 28 to 120 g ai ha−1 depending on the growth stage of the plants. Ou et al. (Reference Ou, Stahlman and Jugulam2018) reported GR50 values for the response to glyphosate of two B. scoparia populations which varied from 42 to 67 g ha−1 to 171 to 187 g ha−1 when the assays were conducted at 25/15 C (day/night temperature) and 32.5/22.5 C, respectively. The GR50 values should be only compared when the experiments have been carried out under similar conditions of plant growth (temperature, relative humidity, and light intensity), phenological stage of the plants, herbicide products used, or the timing of evaluation of the experiment.
Herbicide resistance is increasing worldwide, with an increasing number of cases of cross- and multiple resistance (Hicks et al. Reference Hicks, Comont, Coutts, Crook, Hull, Norris, Neve, Childs and Freckleton2018; Loureiro et al. Reference Loureiro, Escorial, Hernández-Plaza, González-Andújar and Chueca2017; Peterson et al. Reference Peterson, Collavo, Ovejero, Shivraind and Walsh2018; Powles Reference Powles2014), thereby limiting the number of available herbicide options for weed control for farmers. Glyphosate resistance is not an exception. Although surveying and characterizing herbicide sensitivity variation across a large number of populations with dose–response experiments can be expensive and time-consuming, such proactive monitoring studies are essential for identifying shifts in herbicide sensitivity and response before widespread development of herbicide resistance. The sensitivity data generated in this study provide an important reference for determining any time-dependent changes in glyphosate sensitivity of the commonly found weed species in the maize fields. Therefore, subsequent monitoring of the glyphosate sensitivity of these weeds will be needed to ensure the continued use of glyphosate and to minimize and delay the development of resistance.
Acknowledgments
This work was supported by the Spanish Ministry of Agriculture, Food and Environment (MAGRAMA), under grant EG13-75. No conflicts of interest have been declared.
