**Phytotoxic E**ff**ects of Three Natural Compounds: Pelargonic Acid, Carvacrol, and Cinnamic Aldehyde, against Problematic Weeds in Mediterranean Crops**

### **Marta Muñoz 1,2, Natalia Torres-Pagán 1, Rosa Peiró 3, Rubén Guijarro 2, Adela M. Sánchez-Moreiras 4,5 and Mercedes Verdeguer 1,\***


Received: 28 April 2020; Accepted: 27 May 2020; Published: 2 June 2020

**Abstract:** Weeds and herbicides are important stress factors for crops. Weeds are responsible for great losses in crop yields, more than 50% in some crops if left uncontrolled. Herbicides have been used as the main method for weed control since their development after the Second World War. It is necessary to find alternatives to synthetic herbicides that can be incorporated in an Integrated Weed Management Program, to produce crops subjected to less stress in a more sustainable way. In this work, three natural products: pelargonic acid (PA), carvacrol (CV), and cinnamic aldehyde (CA) were evaluated, under greenhouse conditions in postemergence assays, against problematic weeds in Mediterranean crops *Amaranthus retroflexus*, *Avena fatua*, *Portulaca oleracea,* and *Erigeron bonariensis*, to determine their phytotoxic potential. The three products showed a potent herbicidal activity, reaching high efficacy (plant death) and damage level in all species, being PA the most effective at all doses applied, followed by CA and CV. These products could be good candidates for bioherbicides formulations.

**Keywords:** weeds; abiotic stress; natural herbicides; secondarymetabolites; postemergence; phytotoxicity

#### **1. Introduction**

One of the main challenges for the agriculture in this 21st century is to be capable to feed the increasing world population in a sustainable way, because natural resources are becoming even more scarce [1]. Crop protection measures can prevent yield losses due to pests [2]. Herbicides have been the most used method to control weeds since their development, at the end of the Second World War because they are effective and economical [3,4].

Herbicides cause stressin crops and canmake themmore susceptible to other pests [5]. Other problems derived from the overuse of herbicides are environmental pollution, toxicity for nontarget organisms, and the development of herbicide-resistant weed biotypes [6]. In the latest 10 years, integrated weed management (IWM) strategies have been promoted worldwide [7,8] to control weeds. They consist of a combination of methods: cultural, mechanical, physical, biological, biotechnological, and chemical. In Europe, IWM has been promoted through the European Union Directive 2009/128/EC [8].

The society is demanding new solutions for weed control and "greener" weed management products. The use of natural products as bioherbicides could be one alternative to reduce the stress that synthetic herbicides promote in crops and all their negative impacts aforementioned. Bioherbicides could be incorporated in IPM programs as an innovative weed control method. They are less persistent than synthetic herbicides and are potentially more environmentally friendly and safe [9] and also, they have different modes of action, which can prevent the development of herbicide-resistant weed biotypes [10].

Bailey [11] defined bioherbicides as products of natural origin for weed control. The EPA (USA Environmental Protection Agency), considers three categories of biopesticides: (1) biochemical pesticides, which include naturally occurring substances that control pests; (2) microbial pesticides or biocontrol agents, which are microorganisms that control pests; and (3) plant-incorporated protectants, or PIPs, which are pesticide substances produced by plants that contain added genetic material) [10]. In recent years, the search for natural substances that can act as bioherbicides has been very extensive.

The weeds selected for this study were *Amaranthus retroflexus* L., *Avena fatua* L., *Portulaca oleracea* L., and *Erigeron bonariensis* L. because of their importance in many crops worldwide and their difficult management. *A. fatua* is a very important weed mainly in cereals and also in other crops around the world [12], and this weed is on the fourth position in resistance to herbicides worldwide, having developed resistance to nine different modes of action [13]. *A. retroflexus* is a serious and aggressive weed in summer crops, with cosmopolite distribution [14]. It has developed resistance to five modes of action and is on the eight position worldwide in resistance to herbicides [13]. *E. bonariensis*, which can be found both in summer or winter crops, especially with no-tillage practices [15], is on the ninth position in resistance to herbicides worldwide, with resistance to four modes of action. *P. oleracea*, which is a summer weed difficult to control in Mediterranean crops [16], has developed resistance only to two modes of action [13]. *A. fatua* and *E. bonariensis* have developed resistance to glyphosate, which is the herbicide most commonly used around the world [13,17].

There are several examples of natural products that have been tested as potential bioherbicides to control *A. fatua*, *A. retroflexus*, *E. bonariensis,* and *P. oleracea*, mainly essential oils (EOs) [14,18–26], or extracts from plants with different solvents [27–29], or their isolated compounds [30,31]. Most studies have been carried out only in in vitro conditions. Of the weeds considered, *A. retroflexus* has been the most tested. In vitro studies with EOs from *Artemisia vulgaris*, *Mentha spicata*, *Ocimum basilicum*, *Salvia o*ffi*cinalis*, and *Thymbra spicata* from Turkey demonstrated high phytotoxic effects on seed germination and seedling growth of *A. retroflexus*, with stronger effects with higher doses [18]. EOs from *Tanacetum* species growing in Turkey, rich in oxygenated monoterpenes, inhibited completely *A. retroflexus* germination in in vitro assays [19]. In addition, EOs from *Nepeta meyeri*, with high content in oxygenated monoterpenes controlled completely *A. retroflexus* germination [20]. The phytotoxic potential of 12 EOs was studied in vitro against *A. retroflexus* and *A. fatua,* and the most phytotoxic EOs were those constituted mainly by oxygenated monoterpenes [21]. Other EOs which showed strong herbicidal potential against *A. retroflexus* seed germination and seedling growth were *Rosmarinus o*ffi*cinalis*, *Satureja hortensis,* and *Laurus nobilis*[14], and a nanoemulsion of *S. hortensis* EO was tested against *A. retroflexus*in greenhouse conditions killing the weed at 4000 μL/mL dose [22]. *P. oleracea* germination was completely inhibited by *Eucalyptus camaldulensis* EO in in vitro conditions [23]. The application of leaf extracts (obtained using water, methanol, and ethanol as solvents) of cultivated *Cynara cardunculus*in in vitro bioassays inhibited seed germination and germination time in *A. retroflexus* and *P. oleracea* [27].

Different natural compounds have demonstrated herbicidal potential against the germination and seedling growth of *A. fatua*, such as EOs from *Artemisia herba-alba* [24] and *Eucalyptus citriodora* EOs [25] and extracts from *Sapindus mukorossi*, which inhibited *A. fatua* and *A. retroflexus* growth in vitro and in pots [28] or from *Iris sibirica* rhizomes [29].

EOs from *Thymbra capitata*, *Mentha piperita*, *Eucalyptus camaldulensis*, and *Santolina chamaecyparissus* were tested in vivo against *E. bonariensis*. *T. capitata* EO, with high content in carvacrol, was the most effective to control *E. bonariensis*, showing an excellent potential to develop bioherbicide formulations [26].

Some studies carried out in recent years relate the herbicidal activity of plant extracts or EOs to their composition in monoterpenes, and these substances are postulated as the future of natural herbicide components [32–35]. For example, eugenol, a monoterpene that can be found in many EOs as the major compound, like in *Syzygium aromaticum* EO, has shown strong phytotoxic potential against *A. retroflexus* [30] and *A. fatua* [31]. In *A. fatua*, eugenol inhibited its seedling growth, affecting more the roots than the coleoptiles. In addition, sesquiterpenes, secondary metabolites in plants, present in some EOs, have demonstrated strong herbicidal activity [36,37].

The natural products studied on this work for their potential as bioherbicides were pelargonic acid, trans-cinnamaldehyde and carvacrol. Pelargonic acid (PA) (CH3(CH2)7CO2H, n-nonanoic acid), which is present as esters in the EO of *Pelargonium* spp., is a saturated fatty acid with nine carbons in its structure [28–40]. PA and its salts are used like active ingredients in bioherbicide formulations for garden and professional uses worldwide. They are applied as burndown herbicides, which in a short time, attack cell membranes, causing cell leakage, followed by breakdown of membrane acyl lipids [41], and finally causing visible effects of desiccation of green areas of the weeds [38]. All the symptoms caused by PA on weeds involve extreme phytotoxicity for the plants and their cells, which rapidly begin to oxidize, causing necrotic lesions on aerial parts of plants [42,43].

Herbicidal fatty acids have been used for a long time in weed management, and some of them are used as natural herbicides. Still, the high dosage and the high cost are some of the drawbacks of its practical application in the current agriculture. In 2015, the bioherbicide Beloukha® was authorized as plant protection product to be marketed in Europe [44]. It is derived from oleic acid from different origin. Actually, it is authorized also for markets in USA and Canada. This work aims to find an optimal formulation of PA capable to be effective at reduced doses compared to the existing products in the market.

*Trans*-cinnamaldehyde (CA) (C9H8O) is one of the major components of two different cinnamon species (*Cinnamomum zeylanicum* and *Cinnamomum cassia*) and their EOs [45–48]. This compound has shown strong antioxidant properties and is responsible for various observed biological activities of cinnamon like bactericidal, fungicidal, or acaricidal [49–52]. The antimicrobial activity of CA is well known, however, its potential as bioherbicide has been less studied. Despite that, recent research demonstrated the herbicidal activity of CA against *Echinochloa crus-galli* by reducing the fresh weight and growth of this important weed [53]. To our knowledge, the mode of action of CA on weeds has not been elucidated.

The third natural compound evaluated was carvacrol (CV), a phenolic monoterpene frequently present on EOs obtained from many species belonging to Lamiaceae family like *Thymus* spp., *Thymbra* spp., and *Origanum* spp. [34]. CV presents antimicrobial properties that make it helpful for controlling diseases in crop protection [54–58]. In relation to its mode of action, CV exhibited membrane-disrupting activity that was dependent on long exposure at high concentration [33]. Postemergence exposure of plants to high concentrations of CV causes severe phytotoxicity. One of the effects associated with the mode of action of CV is the reduction of weed growth [22,41,54].

This work is a collaboration between the Universitat Politècnica de València (UPV) and the company Seipasa S.A., which develops and commercializes biopesticides, with the purpose to manage agricultural ecosystems in a more sustainable way. The objective of the present study was to evaluate the herbicidal potential of the natural compounds pelargonic acid, trans-cinnamaldehyde, and carvacrol against important cosmopolite weeds (*Amaranthus retroflexus* L., *Portulaca oleracea* L., *Erigeron bonariensis* L., and *Avena fatua* L.) as an alternative to synthetic herbicides to reduce the abiotic stress that they cause on crops. Effective compounds were formulated as emulsifiable concentrates (ECs) by Seipasa S.A., and evaluated for their postemergence herbicidal activity in greenhouse conditions in the UPV (Spain).

#### **2. Materials and Methods**

*2.1. Postemergence Herbicidal Assays against Targeted Weed Species*

#### 2.1.1. Weeds

Seeds of *Amaranthus retroflexus* L., *Portulaca oleracea* L., and *Avena fatua* L. purchased from Herbiseed (Reading, UK) (year of collection 2017), which have been previously tested in a plant growth chamber EGCHS series from Equitec (Madrid, Spain) (30 ± 0.1 ◦C, 16 h light and 20 ± 0.1 ◦C, 8 h dark for *A. retroflexus* and *P. oleracea*; 23.0 ± 0.1 ◦C, 8 h light and 18.0 ± 0.1 ◦C 16 h dark for *A. fatua*) to assure their germination viability, were sown in pots (8 × 8 × 7 cm) filled with 2 cm of perlite and 5 cm of soil collected from a citrus orchard nontreated with herbicides. In Figure 1, the location (39◦37'24.8" N, 0◦17'25.6" W Puzol, Valencia province, Spain) and a view of the citrus orchard (0.4 ha) from which the soil was collected is reported. Table 1 shows the main physical characteristics of the soil used for the experiments.

**Figure 1.** Location (**A**) and view (**B**) of the citrus orchard where the soil for the herbicidal tests was collected.

**Table 1.** Physical properties of the soil used for the experiments [59].


*Erigeron bonariensis* L. seeds were collected from an ecological weed management persimmon orchard located in Carlet (Valencia province, Spain) in July 2018. They were previously tested in the plant growth chamber described before (30 ± 0.1 ◦C, 16 h light and 20 ± 0.1 ◦C 8 h dark) to assure their germination capability and after that, sown in plastic pots filled with a mix of three-fourth peat and one-fourth perlite instead of soil because it was very difficult to germinate the seeds on the soil, as *E. bonariensis* germinates better in lighter soils [60] and, therefore, the properties of the soil collected from the citrus orchard (Table 1) did not fit the needs for their germination.

All weeds were irrigated by capillarity from trays (43 cm × 28 cm × 65 cm) placed under the pots and filled with water, until the plants were ready for the herbicidal experiments.

#### 2.1.2. Treatments

Ten pots were prepared for each treatment, described in Table 2. The treatments were applied when plants reached the phenological stage of 2-3-true leaves, corresponding to stage 12-13 BBCH (Biologische Bundesanstalt, Bundessortenamt und Chemische Industrie ) scale for the monocotyledonous *A. fatua*, and 3-4-true leaves, corresponding to stage 13-14 BBCH scale for the dicotyledonous *A. retroflexus* and *P. oleracea* and in rosette stage for *E. bonariensis*, stage 14-15 BBCH scale (Figure 2). Pelargonic acid, cinnamic aldehyde and carvacrol were provided formulated as emulsifiable concentrates (ECs) by the company Seipasa S.A. (L'Alcudia, Valencia province, Spain). Beloukha® was purchased from Ferlasa (Museros, Valencia province, Spain) and Roundup® Ultra Plus was purchased from Cooperativa Agrícola Nuestra Señora del Oreto (CANSO, L'Alcudia, Valencia province, Spain).


#### **Table 2.** Treatments tested.

**Figure 2.** Pots ready for the postemergence treatments. (**A**) *A. fatua*, (**B**) *A. retroflexus*, (**C**) *P. oleracea*, and (**D**) *E. bonariensis.*

In Table 3, the dates of the herbicidal tests and the greenhouse conditions during the experimental periods are reported. Data were registered using a HOBO U23 Pro v2 data logger (Onset Computer Corporation, Bourne, MA, USA).


**Table 3.** Greenhouse conditions during the herbicidal tests.

#### *2.2. Evaluation of the Herbicidal Activity of Each Natural Product*

During the experiments, images from the plants were taken 24 h and 3, 7, 15, and 30 days after the treatments application to be processed with Digimizer v.4.6.1 software (MedCalc Software, Ostend, Belgium, 2005–2016).

To evaluate the herbicidal activity, two variables were measured for each plant: the efficacy, which was scored 0 if the plant was alive and 100 if the plant was dead, and the damage level, which was

assessed between 0 and 4 as reported in Table 4 and Figure 3. The efficacy and damage level for each treatment were calculated as the mean of the 10 treated plants.



**Figure 3.** Damage scale for each species: (**A**) *A. fatua*, (**B**) *P. oleracea*, (**C**) *A. retroflexus*, and (**D**) *E. bonariensis*.

#### *2.3. Statistical Analyses*

Data were processed using Statgraphics® Centurion XVII (StatPoint Technologies Inc., Warrenton, VA, USA) software. A multifactor analysis of variance (ANOVA) was performed on efficacy and damage level including species, treatments, time after treatments application, and their double significant interactions as effects, followed by Fisher's multiple comparison test (LSD intervals, least significant difference, at *p* ≤ 0.05) for the separation of the means.

#### **3. Results and Discussion**

#### *3.1. E*ffi*cacy of Pelargonic Acid, Cinnamic Aldehyde, and Carvacrol against Target Weeds*

*A. retroflexus* was the weed species most susceptible to the treatments tested, with 73.50 efficacy (Table 5). No significant differences were observed between the other species, which showed around 55 efficacies. The fact that all species tested were susceptible to all treatments with natural products assayed confirm that they could be a more sustainable alternative to synthetic herbicides, and they also offer new modes of action to control weeds that have developed resistant biotypes to many herbicides.


**Table 5.** Efficacy according to the species, time, and treatment.

Values are efficacy ± standard error. Means followed by different letters in the same column differ significantly (*p* ≤ 0.05).

Efficacy increased with time after treatments application, with values close to 90 between 7 and 15 days (Table 5). This happened because PA, at all doses applied, and the higher doses of CA and CV acted very quickly in the treated species, causing the death of all plants between 24 h and 3 days after application of treatment (Figures 4–7, Tables S1–S4). The same happened for the bioherbicide reference BE (as PA was also the active compound on it), while GL acted more slowly, depending on the species against which it was applied; it killed *A. retroflexus* plants after 3 days, *A. fatua* and *P. oleracea* after 15 days, and *E. bonariensis* after 30 days (Figures 4–7, Tables S1-S4). It has been reported that weed damage caused by PA can be observed visually few hours after application [61]. Thymol, *trans*-cinnamaldehyde, eugenol, farnesol, and nerolidol were tested in postemergence in *E. crus-galli* applied at two-leaf stage, and significantly reduced the shoot growth and the fresh and dry weight 2 days after the foliar treatments with 0.5%, 1.0%, and 2.0% concentrations. All treatments except

thymol controlled the weed completely when applied at 1.0% and 2.0% [52]. The concentrations of CA used in this work were higher, and this could explain the quicker toxic effect observed on weeds. It is also remarkable that weed species displayed different sensitivity to low doses of CA; *E. bonariensis* and *P. oleracea* showed more resistance to this compound than the other weeds tested (Figures 4–7, Tables S1-S4), as the lowest concentration (6%) used took more time (15 days) to kill all the plants in *E. bonariensis* than in *A. retroflexus* (24 h) or *A. fatua* (3 days), whereas in *P. oleracea,* this dose reached 50 efficacy, i.e., only 50% of plants were dead at the end of the experiment (30 days). Previous studies also confirmed the rapid activity of carvacrol in plants; in a greenhouse experiment, a nanoemulsion (NE) of *Satureja hortensis* L. EO, rich in carvacrol (55.6%), was applied against *A. retroflexus* and *C. album,* and after 30 min, the weeds were exhibiting injury symptoms, reaching the maximum lethality within 24 h of treatment application. The lethality percentage was dependent on the doses applied and the species against which NE was applied [21]. As observed with CA, also weed species showed different sensibility to CV application, especially at the lower dose, which took more time to control the weeds (Figures 4–7, Tables S1–S4): *A. retroflexus* was the more sensitive species, being controlled by all doses 24 h after application of treatment (Figure 4, Table S1), whereas in *A. fatua* and *E. bonariensis*, the lowest dose took 7 and 15 days, respectively, to reach 100 efficacy (Figures 5 and 6, Tables S2 and S3), being again *P. oleracea* the most resistant weed species, 7 days after treatment application, all plants were killed in all CV treatments, although then some regrew 15 and 30 days after treatments application (Figure 7, Table S4).

All the treatments managed to control the weed species tested, and the results of the treatments were statistically significant compared to CW (Table 5). The most effective treatment was the PA formulation at 10%, achieving 74.50 efficacy. This treatment did not show significant differences compared to the results obtained by the commercial product used as biological reference, also containing PA as active ingredient, which obtained an efficacy of 78.50. Moreover, there were no significant statistical differences in the efficacy between the three doses of the PA-based formulations (5%, 8%, and 10%). The next most effective treatment was the CA-based formulation, which exhibited the same efficacy values for the two higher doses applied (12% and 24%), while the lowest dose (6%) had significant less efficacy. This can be explained by the different sensitivity of the weed species to low doses of CA, as commented above. Finally, the treatments with carvacrol did not show significant differences in efficacy between doses, but with the control, and were also very effective, reaching an efficacy between 60.50 and 65.00 (Table 5).

All treatments tested with natural products showed higher efficacy for the control of weeds than GL, which showed efficacy values of 36. This was because of its slower activity. Mechanism of action of GL is by affecting the enzyme 5-enolpyruvlyshikimate-3-phosphate synthase (EPSPS), and it is the only herbicide with this mode of action. The inhibition of EPSPS reduces levels of amino acids needed for the synthesis of proteins, cell walls, and secondary plant products. In addition, the inhibition of EPSPS causes deregulation of the shikimic acid pathway, promoting the disruption of plant carbon metabolism [62]. GL is translocated in plants and differential responses of weed species may be caused by differences in herbicide translocation, i.e., weeds capable to translocate GL more efficiently are more severely damaged [63]. In field experiments conducted for 2 years, it was verified that GL controlled more effectively *A. retroflexus* than other species [64], which supports our results. Decreased herbicide translocation to the meristem causes reduced glyphosate efficacy [65]. The necessity of being translocated explains the slow effect of GL compared with the natural compounds, as 14C translocation throughout the plant demonstrated that glyphosate took 3 days to reach and accumulate in the meristematic tips of the roots and shoots [66]."

**Figure 4.** Evolution of efficacy of the tested treatments (**A**) pelargonic acid, (**B**) cinnamic aldehyde and (**C**) carvacrol in *A. retroflexus* during 30 days after application.

**Figure 5.** Evolution of efficacy of the tested treatments (**A**) pelargonic acid, (**B**) cinnamic aldehyde, and (**C**) carvacrol in *A. fatua* during 30 days after their application.

**Figure 6.** Evolution of efficacy of the tested treatments (**A**) pelargonic acid, (**B**) cinnamic aldehyde, and (**C**) carvacrol in *E. bonariensis* during 30 days after their application.

**Figure 7.** Evolution of efficacy of the tested treatments (**A**) pelargonic acid, (**B**) cinnamic aldehyde, and (**C**) carvacrol in *P. oleracea* during 30 days after their application.

3.1.1. Efficacy of Pelargonic Acid, Cinnamic Aldehyde, and Carvacrol on *A. retroflexus*

In the species *A. retroflexus* (Figure 4, Table S1) all the treatments tested obtained 100 efficacy (all treated plants were dead) one day after the application of the treatment, except for the chemical reference. The treatment with GL managed to control the species on the third day after its application. In this trial, there was a relevant percentage of mortality in the CW, especially at the end of the trial.

#### 3.1.2. Efficacy of Pelargonic Acid, Cinnamic Aldehyde, and Carvacrol on *A. fatua*

All the tested treatments managed to control completely the species *A. fatua* from the third day after application (Figure 5, Table S2), except CV6, which achieved 100 efficacy after 7 days, and GL, which reached 100 efficacy 15 days after application. The treatments that showed phytotoxic effects more quickly were, starting from the first day after application, the bioherbicide reference (BE), AC12, and PA10.

#### 3.1.3. Efficacy of Pelargonic Acid, Cinnamic Aldehyde, and Carvacrol on *E. bonariensis*

All treatments were able to control *E. bonariensis* (Figure 6, Table S3). The higher doses of the treatments performed with CA- and CV-based formulations achieved a total control of this species faster than their lower doses. It should be noted that despite this, all of them managed to control it completely 15 days after the application. The bioherbicide reference (BE) reached 100 efficacy 24 h after its application, instead GL took 30 days to reach 100 efficacy (death of all treated plants).

#### 3.1.4. Efficacy of Pelargonic Acid, Cinnamic Aldehyde, and Carvacrol on *P. oleracea*

The most effective treatments to control *P. oleracea* were the three treatments carried out with the PA-based formulation (PA5, PA8, and PA10) (Figure 7, Table S4). A dose effect was observed in this species for the tested natural products, being higher doses more effective and showing phytotoxic effects faster than lower ones. The treatment AC6 reached 50 efficacy at the end of the experiment (30 days after application), while the higher doses of this compound (AC12 and AC24) killed all plants after 3 days of application. The treatments CV8, CV16, and CV32 decreased their efficacy from day 7, when some of the evaluated plants regrew. It should be noted that the treatment with the chemical reference, GL, exhibited a slower action than the rest of the treatments with natural products, showing phytotoxic effects on this species between 7 and 15 days after application.

When analyzing the effect of the interaction between species and time after treatments with respect to efficacy, the species that showed the highest sensitivity most rapidly was *A. retroflexus*. On the other hand, the species that took longer to show phytotoxic effects was *A. fatua*. However, at the end of the trials, all species showed high mortality rates, which were slightly higher in *A. retroflexus* and *A. fatua* than in *P. oleracea* and *E. bonariensis* (Figure 8).

**Figure 8.** Effect of the interaction between treatment and days after treatment application in the efficacy per species.

#### *3.2. Damage Level of Pelargonic Acid, Cinnamic Aldehyde, and Carvacrol against Target Weeds*

*A. retroflexus* was the species which presented higher damage level, followed by *P. oleracea* and *A. fatua* (without significant differences between them), and finally *E. bonariensis* (Table 6). All species exhibited damage level near 2 or higher, which means severe damage (Table 4). It is important to consider the damage level caused by the treatments on the weed species in addition to their efficacy because it represents the state of the plants that were not killed. If the plants remaining alive were more damaged, it would mean that in field conditions, they would be less competitive with crops, causing less stress to them.


**Table 6.** Damage level depending on the species, time after application, and treatment.

Values are mean of damage level ± standard error (ten replicates). Different letters in the same column indicate significant differences (*p* ≤ 0.05).

Throughout time, more severe levels of damage were reached as more days after treatment applications passed, with significant differences in the damage level assessment between different days after the applications (Table 6). All the treatments tested successfully controlled the weed species inducing a high level of damage compared with CW. The treatments that showed the strongest phytotoxicity on weeds were PA10 and BE, with no significant differences between them. PA10 showed no significant differences with the other two doses of PA-based formulations tested (PA5 and PA8), neither with the two highest doses of CA based formulations tested (CA12 and CA24) nor with the highest doses of CV tested (CV32) (Table 6).

The damage level increased in all species with time after treatments (Figure 9). *A. retroflexus* was confirmed as the most susceptible species to the treatments, as it showed a higher level of damage than the other species 24 h after the treatments were administrated. No differences between species were observed 15 days after treatment, as all showed similar levels of damage.

**Figure 9.** Effect of treatment and time after treatment interaction on damage level.

The effects induced by the different treatments on *E. bonariensis* 24 h after their administration are presented in Figure 10. This species is shown because of its intermediate response to all treatments as compared with *A. retroflexus* that was more sensitive or *P. oleracea*, which was more resistant and because phytotoxic effects can be better visualized in it than in *A. fatua*. The intermediate concentration tested for PA, CV, and CA is shown to be representative of the effects of the other concentrations tested. All the natural compounds tested caused more severe plant damage than the synthetic herbicide GL 1 day after treatment. The effects of 8% PA were very similar to those induced by the positive bioherbicide control Beloukha (also containing PA as active compound). Probably due to the effect of PA, the cuticles exhibited alteration on membrane permeability and peroxidation of thylakoid membranes [67] and leaves appeared desiccated, with reduced photosynthetic pigments but without punctual damages on the leaves, which resulted in a stoppage of growth and development of the whole plant. In contrast, CV-treated leaves showed signs of dehydration, resulting in curling and punctual damages on the leaves with increased necrotic spots related to application spots, which could be due to the disruption of cell membranes [68]. Finally, CA treatment resulted in growth reduction and loss of photosynthetic pigments, which could be related to oxidative damage induced by this compound. This oxidative damage has to be further investigated as no mode of action of CA has been reported in the literature up to now.

Bioherbicides are new products on the international markets and consequently, the processes for obtaining natural raw materials are not yet very efficient or the final cost of its extraction is elevated compared to synthetics. This fact affects the final cost of these formulated products, making them more expensive in some cases than conventional herbicides for farmers. Nevertheless, it is important to evaluate the cost–benefit factor of bioherbicides, including sustainability, reduction of soil and water contamination, or the absence of residues on crops. In line with legal framework, policies, and global sustainability objectives, the higher price of bioherbicides justifies the benefits that can be achieved with their implementation [69]. On the other hand, the rapid action, broad spectrum, and eco-friendly profile make bioherbicides molecules more attractive to the pesticide market, which is increasingly concerned with the sustainability of treatments applied in agriculture. Herbicide market is expected to reach a value of \$37.99 billion by 2025 [38]. Improving the efficiency of raw material extraction, decreasing the applied doses per hectare using improved formulations, as well as combining active substances in search of synergies may be the future of new sustainable herbicides.

The natural products tested, PA, CV, and CA, performed strong herbicidal activity in all the treated weeds, causing high lethality and damage levels; hence, they demonstrated that they could be good candidates for bioherbicides formulations. Further investigations should focus on determining the dose–response of different weed species to these compounds in order to find the optimal doses, which is very important in the context of integrated weed management and sustainable agriculture. Another key point is to find out the optimum phenological stage in which the products should be applied to weeds and crops, to achieve the maximum phytotoxic effect on weeds minimizing their phytotoxic effects and consequent stress on crops. A better understanding of their mode of action could lead to a more efficient administration. Finally, different combinations between these natural products could be a powerful tool for weed management. Their synergies and antagonisms must be also considered and studied.

**Figure 10.** Images of *Erigeron bonariensis* plants 24 h after treatment applications.

#### **4. Conclusions**

The natural products PA, CV, and CA showed great herbicidal activity against the weeds *A. retroflexus*, *A. fatua*, *E. bonariensis,* and *P. oleracea* and could be good candidates for bioherbicides formulations. *A. retroflexus* was the most sensitive weed to all the applied treatments. For CV and CA, the higher doses applied exhibited greater and quicker phytotoxic effects than the lowest, with different responses in the weed species, while there were no significant differences in the herbicidal activity between the tested doses of PA. This study demonstrates that natural products could be sustainable as well as effective alternatives to synthetic herbicides, and they contribute to integrated weed management.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4395/10/6/791/s1, Table S1. Efficacy of the tested treatments on *A. retroflexus* after 1, 3, 7, 15 and 30 days of application. Table S2. Efficacy of the tested treatments on *A. fatua* after 1, 3, 7, 15 and 30 days of application. Table S3. Efficacy of the tested treatments on *E. bonariensis* after 1, 3, 7, 15 and 30 days of application. Table S4. Efficacy of the tested treatments on *P. oleracea* after 1, 3, 7, 15 and 30 days of application.

**Author Contributions:** Conceptualization, M.V., M.M., and A.M.S.-M.; methodology M.V., M.M., N.T.-P., and R.G.; formal analysis, M.V., M.M., and N.T.-P.; investigation, M.V., M.M., A.M.S.-M., R.P., N.T.-P., and R.G.; resources, M.V., M.M., and R.G.; data curation, N.T., M.M., R.G. and R.P.; writing—original draft preparation, M.V., M.M., and N.T.-P.; writing—review and editing, M.V., A.M.S.-M., and R.P.; visualization, M.V., M.M., N.T.-P., A.M.S.-M., R.P., and R.G.; supervision, M.V., M.M., and A.M.S.-M.; project administration, M.V.; and funding acquisition, M.V. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by SEIPASA.

**Acknowledgments:** Thanks to Vicente Estornell Campos and the Library staff from Polytechnic University of Valencia that assisted us to get some helpful references.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Resilience Capacity Assessment of the Traditional Lima Bean (***Phaseolus lunatus* **L.) Landraces Facing Climate Change**

**María Isabel Martínez-Nieto 1, Elena Estrelles 1, Josefa Prieto-Mossi 1, Josep Roselló <sup>2</sup> and Pilar Soriano 1,\***


Received: 30 March 2020; Accepted: 18 May 2020; Published: 26 May 2020

**Abstract:** Agriculture is highly exposed to climate warming, and promoting traditional cultivars constitutes an adaptive farming mechanism from climate change impacts. This study compared seed traits and adaptability in the germinative process, through temperature and drought response, between a commercial cultivar and Mediterranean *Phaseolus lunatus* L. landraces. Genetic and phylogenetic analyses were conducted to characterize local cultivars. Optimal germination temperature, and water stress tolerance, with increasing polyethylene glycol (PEG) concentrations, were initially evaluated. Base temperature, thermal time, base potential and hydrotime were calculated to compare the thermal and hydric responses and competitiveness among cultivars. Eight molecular markers were analyzed to calculate polymorphism and divergence parameters, of which three, together with South American species accessions, were used to construct a Bayesian phylogeny. No major differences were found in seed traits, rather different bicolored patterns. A preference for high temperatures and fast germination were observed. The 'Pintat' landrace showed marked competitiveness compared to the commercial cultivar when faced with temperature and drought tolerance. No genetic differences were found among the Valencian landraces and the phylogeny confirmed their Andean origin. Promoting landraces for their greater resilience is a tool to help overcome the worldwide challenge deriving from climate change and loss of agrobiodiversity.

**Keywords:** *Phaseolus*; landrace; seed; germination; drought tolerance; genetic approach; sustainable agriculture; climate change

#### **1. Introduction**

Agriculture is highly exposed to environmental changes, such as climate warming and aridification, as farming activities depend directly on climate conditions. Indeed, the role of agriculture is fundamentally improving natural resources management, rural development, food production and preserving environmental heritage by the conservation of seminatural habitats, landscape and biodiversity [1,2].

Accordingly, the cultivation and conservation of traditional landraces and crop diversification can be effective adaptation strategies to respond to these changing conditions [3], mainly given the increase in aridity and rainfall unpredictability that derive from these changes in environmental conditions.

Loss of crop diversity is a worldwide challenge. Modern cultivars have replaced local landraces, which are now threatened in food production systems, including cultural heritage, local knowledge and traditional farmer skills. This decline, supported by worldwide globalization, leads to reduced

agrobiodiversity on a massive scale, and mainly in developed countries where the industrial food system moves towards genetic uniformity. With the disappearance of traditional species and cultivars, wide ranges of unharvested species also disappear. Promoting local cultivars, which are theoretically more competitive, is one of the major adaptive mechanisms of agriculture to climate change impacts [4–6].

Legumes, specifically *Phaseolus lunatus*, are considered one of the most valuable sources of nutrients in developing countries [7,8]. *Phaseolus lunatus* (Fabaceae), commonly known as "lima bean", and locally termed "garrofó", is the second genus *Phaseolus* species to follow *P. vulgaris* in terms of its economic interest. Attention is paid mainly to its food use worldwide [9], even though other relevant aspects are under study, such as the role on plant protection of the cyanogenic glycosides present in the seeds of this species [10,11].

Cultivated varieties have a South American origin and initially concentrated in northern Peru, where an in-depth selection was developed by the Inca civilization for a long time [12]. According to current germplasm and herbarium records, the conspecific wild ancestor of lima bean is widely distributed from Mexico to Argentina [13]. These landraces are classified into two major groups, Mesoamerican and Andean, according to their geographic origin and seed characteristics [14].

Although it originally comes from Mesoamerica and the Andes, it is currently cultivated throughout Latin America, the southern United States, Canada, and many other world regions including Mediterranean countries, where it is associated with local gastronomy.

On the coasts of the Mediterranean Basin, it is cultivated in warm sunny places in deep well-drained soil. Its strong roots allow plants to thrive on lands where other legumes cannot. It is a highly demanding crop with special requirements. These plants have a type IV climbing growth habit [15] with considerable vegetative development, which means they need a structure that supports, ventilates and illuminates their branches. Nowadays, this cultivation is maintained only for the value of its tender pods and dried grains, and for its special link with traditional cuisine. Currently, the traditional cultivars of this species are being replaced with commercial varieties and represent a testimonial crop in small areas on the European continent.

Hence the present research intends to compare and assess the resistance and adaptability of local cultivars and a commercial variety to face the environmental alterations deriving from climate change. The commercial cultivar is imported from Peru and can be purchased in most retail stores. Primitive landraces, known as 'Pintat', 'Ull de Perdiu' and 'Cella Negra', are traditionally used in the western Mediterranean Basin and are especially cultivated in east Spain (Valencian Community). The use of this species in the eastern Iberian Peninsula in that traditional cuisine is very ancient. Today, we only have references to using these four cultivars in the last 100 years in this region. Our main aim in this work was to recover forgotten crops for the future. In fact, some of the studied cultivars, in particular 'Cella Negra', have practically disappeared today and it has been very difficult to find seeds of this plant.

Furthermore, barcoding is a method to identify taxonomic units using short DNA sequences that allow the determination of the genetic polymorphism and divergences between them. The aim is to identify a region or a combination of regions capable of discriminating taxonomic units, such as species, subspecies, cultivars, or even gene lineages within species [16] and references therein]. Although chloroplast DNA barcoding is utilized mainly to identify plant species, its application can be extended to the food industry, evolution studies and forensics [17]. Various regions of the plastid genome have been proposed to serve as DNA barcodes in plants, such as those put forward by Shaw et al. [18] or Taberlet et al. [19], internal transcribed spacers (ITS) [20] or other specific genes like FRO1 and Phs7 used in legumes phylogenetics by Diniz et al. [21]. This method has been useful in Leguminosae phylogenetics and wild gene pool identifications in *Phaseolus lunatus* [16,21–23]. Thus, it might be a useful tool for typifying local landraces.

This study focuses on seed characterization and providing new information about seed response to temperature and water stress tolerance, estimated during the germinative process, in the *Phaseolus lunatus* traditional cultivars from Mediterranean Europe in line with the future global warming and water deficit scenario.

It also aims to characterize molecularly cultivars—by determining the genetic polymorphism and divergences among local, traditional and commercial, as well as American accessions—of *P. lunatus* in an attempt to genetically delimitate landraces, and to find the potential correlation of these genetic characteristics and germination responses. Moreover, the phylogenetic origin of the Valencian cultivars is studied as part of its molecular characterization.

#### **2. Materials and Methods**

Four lima bean (*Phaseolus lunatus*) cultivars were tested, three of which were from local Valencian traditional crops ('Pintat', 'Ull de Perdiu' and 'Cella Negra'), mainly provided by the Estación Experimental Agraria de Carcaixent (EEA-Carcaixent) (province of Valencia, Spain). A fourth commercial cultivar imported from Peru to Spain (hereinafter referred to as 'Peru') was bought for the study. The seeds provided by the EEA-Carcaixent were collected during the previous season, nearly one year before starting the germination tests. We did not collect data on the seeds of commercial origin.

#### *2.1. Seed Features*

Seed dimensions were measured on a digital image using the ImageJ software [24]. Seed weight was determined by an Orion Cahn C-33 microbalance. All the data were obtained from *n* = 50 seeds from each cultivar.

In order to detect differences in variance levels and to identify homogeneous groups, a one-way ANOVA and Tukey's test (*p* < 0.05) were applied, respectively, for each parameter among the different cultivars.

#### *2.2. Germination Assays*

Seed germination assays were performed with the 'Pintat' and the commercial cultivar, 'Peru', for the low seed availability of the rarest landraces, 'Ull de Perdiu' and 'Cella Negra'. Sporadic tests were conducted with them to provide the initial data for future studies. Data were included as Supplementary data.

Tests were carried out using four replications of 10–15 seeds (depending on seed availability) per treatment for each cultivar. Tests were conducted on 14-cm diameter Petri dishes with paper filters kept in climate-controlled cabinets. Illumination was provided by daylight fluorescent tubes with a 12-h photoperiod and a mean irradiance of 100 <sup>μ</sup>mol·m−2·s<sup>−</sup>1. The germination process was evaluated for 15 days. Germinated seeds were counted daily.

Firstly, the optimum germination conditions for successive experiments were set. Temperature screening, using six constant temperatures (15 ◦C, 20 ◦C, 25 ◦C, 30 ◦C, 35 ◦C, 40 ◦C), was applied to determine the optimal germination temperature.

The water stress effect was evaluated by the controlled osmotic potential levels generated using polyethylene glycol (PEG 6000) solution at 30 ◦C according to Villela et al. [25] to obtain 0 (control), −1, −2, −3, −4 and −5 bar. In order to minimize the evaporation and concentration of the effect of solutions and to maintain the known osmotic potential stable, seeds were moistened every 24 h with fresh PEG solutions and plates were kept in double plastic zip lock bags. After 15 days, non-germinated seeds were transferred to distilled water. Thereby, germination capacity recovery was tested to check the potential influence of PEG exposure on seed germination behavior of *Phaseolus lunatus* cultivars.

Germination Percentage and Mean Germination Time (MGT) were considered to compare seed responses. The base temperature (Tb), by back extrapolation [26], and the thermal time requirement [27] were also calculated to compare thermal responses. Then, the base potential (Ψb) and hydrotime (θ) for each cultivar were calculated [28,29].

Variance levels and homogeneous groups were determined by the one-way ANOVA and Tukey's test (*p* < 0.05), respectively, for each parameter among cultivars.

#### *2.3. Genetic Assays*

#### 2.3.1. Plant Material and DNA Extraction

The plant material used in the molecular analysis was obtained from the seeds germinated in Germplasm Bank (UV) or Estación Experimental Agraria (EEA-Carcaixent). Eight individuals of each cultivar were analyzed, except for 'Cella Negra', where only five individuals were available at the time of when the genetic assays were done. For the 'Ull de Perdiu' and 'Pintat', we studied two different samples; one seed accession was obtained from the EEA-Carcaixent, while the other was bought from a traditional market. All the 'Cella Negra' seeds came from EEA-Carcaixent and all the 'Peru' ones were obtained from a market as local farmers do not traditionally cultivate them. All the accessions were identified according to seeds' distinctive morphological features. DNA was extracted from young leaves using Doyle and Doyle [30] protocol, modified by Soltis laboratory (2002; https://www. floridamuseum.ufl.edu/wp-content/uploads/sites/95/2014/02/CTAB-DNA-Extraction.pdf). In order to phylogenetically locate the Valencian cultivars, all the accessions provided by Serrano-Serrano et al. [22] in NCBI were used.

#### 2.3.2. Molecular Analyses

A pool of five chloroplast and three nuclear markers (see Table 1) was analyzed to characterize the Valencian landraces. These markers were variable in other studies related specifically to *Phaseolus lunatus* or *Phaseolus* spp. [21–23]. A standard PCR protocol following GoTaq® Polymerase (Promega, Madison, WI, USA) instructions was used for all the markers, except for *Phs*7 and *FRO3*, which were amplified following Diniz et al. [21]. The PCR products were purified using the Real Clean PCR Kit (Durviz, Valencia, Spain) and sequenced in an ABI 3100 Genetic Analyzer with the ABI BigDye Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems, Foster City, CA, USA).


**Table 1.** The markers analyzed for the *P. lunatus* Valencian cultivars, primer names, Tm (primer melting temperature) and original references in which they were described.

#### 2.3.3. Phylogenetic Analyses

As Serrano-Serrano et al. [22] used ITS, *Atpb-rbcL* and *trnL-trnF* fragments to construct a wide *P. lunatus* phylogeny, these fragments were employed to locate the origin of the Valencian cultivars. Two individuals of each Valencian landrace and all the accessions provided by Serrano-Serrano et al.[22] in NCBI were analyzed. MAFFT v. 7.402 [33,34] was utilized to generate a multiple sequence alignment. The preconfigured MAFFT strategy, which favors accuracy with the FFT-NS-I algorithm (an iterative refinement method that performs 1000 iterations), and default parameters were selected. The ambiguously aligned regions were automatically dealt with using GBlocks v. 091b [35] by implementing the least stringent parameters, but allowing for gaps in 50% of sequences. (NCBI accession numbers: MT072230–MT072258, ITS; MT080626–MT080654, *atpB-rbcL*; MT090972–MT091000, *trnL-trnF*; MT110491–MT110519, *rpoB-trnC*; MT124955–MT124983, *psbA-trnH*; MT154089– MT154117, phs7; MT154118–MT154146, FRO3).

A Bayesian phylogenetic MCMC analysis was run with MrBayes v. 3.2.2 [36]. Indels were coded with SeqState v. 1.4.1 [37] according to modified complex coding. The coded indels were considered to be a partition of standard data (states = 0, 1, 2, 3, ?), with the gamma rate and hyperprior fixed at 1.0 to allow different stationary state frequency proportions to be explored by the MCMC procedure. The optimal substitution models for the nucleotide section were inferred with PartitionFinder2 [38] by considering a model with linked branch lengths for the codificant and non-codificant regions of nrITS and chloroplast fragments, respectively, and using the Bayesian information criterion (BIC). Finally, three partitions were considered: two within the ITS (ITS1 + ITS2 and 5.8S), as well as *Atpb-rbcL* + *trnL-trnF*. This analysis favored the HKY + G model for the ITS1 + ITS2 partition, K80 + I for 5.8S, and also GTR + I for the chloroplast region. Then, a MrBayes analysis was conducted with two parallel and simultaneous four-chain runs, executed over 5 <sup>×</sup> 106 generations, starting with a random tree, and sampling after every 500th step. The first 25% of the data was discarded as burn-in. The 50% majority-rule consensus tree and the corresponding posterior probabilities were calculated from the remaining trees. Chain convergence was assessed by ensuring that the average standard deviation or split frequencies (ASDSF) values were below 0.01, and the potential scale reduction factor (PSRF) values approached 1.00. iTOL v. 4.4.2 [39,40] was used to construct the 50% majority rule consensus tree. The programs MAFFT, MrBayes and PartitionFinder2 were hosted at the CIPRES Science Gateway [41].

#### 2.3.4. DNA Polymorphism and Divergence

The MAFFT original alignment without outgroups was employed to evaluate DNA polymorphism and divergence by taking in account the studied Valencian cultivars and all the accessions, including those used by Serrano-Serrano et al. [22], respectively. All the analyzed markers were utilized to study the Valencian landraces, as well as ITS, *Atpb-rbcL* and *trnL-trnF*, for the whole analysis. Five parameters were calculated by DnaSP v. 6 [42]: segregating sites (s), nucleotide diversity (π), number of haplotypes (h), haplotype diversity (Hd), and the nucleotide genetic differentiation estimate Kst.

#### **3. Results**

#### *3.1. Seed Features*

The seed dimensions of these four *Phaseolus lunatus* cultivars were similar (Table 2). It is noteworthy that the 'Pintat' seeds obtained higher values for the length and width parameters, and had a more rounded contour. The thickness analysis indicated significant differences among cultivars, with the lowest values for the traditional landraces. The 'Peru' and 'Pintat' seeds were the heaviest, while the 'Cella Negra' seeds were lightweight.

Seed coat color is an important consumer trait. In this group, it is a relevant distinctive character for these traditional cultivars (Figure 1; Table 2). The studied commercial cultivar, identified herein as 'Peru', has a completely white seed coat showing no type of pigmentation. The traditional 'Pintat' depicts an irregular spotted pigmentation over the whole external cover, from dark maroon to brown, depending on the maturation stage. The 'Ull de Perdiu' cultivar has a characteristic black eye surrounding the hilum seed zone. Finally, the cultivar known as locally 'Cella Negra' is identified by having a dark brown to black seed tip close to the embryo radicle lobe.

**Table 2.** Seed morphological features for the different studied cultivars. Length (L), width (W) and their relation (L/W), thickness, as well as weight and color trait of seed coat, are indicated. The same letters indicate homogeneous groups among temperatures (*p* < 0.05) for each cultivar.


**Figure 1.** Seed morphological traits and pigmentation for the studied *Phaseolus lunatus* cultivars; 'Peru'; 'Pintat'; 'Ull de Perdiu'; 'Cella Negra'.

#### *3.2. Germination Assays*

#### 3.2.1. Germination Response to Temperature

High germination percentages were achieved at almost all the tested temperatures. The lowest values were for 35 ◦C in the two studied cultivars, while no germination was observed in any of them above this temperature.

After taking into account the values obtained for the germination percentage and mean germination time, the optimal germination temperature for the studied group of *Phaseolus lunatus* cultivars was set at 30 ◦C (Figure 2; Table 3). Good results for germination percentages were also obtained at 15 ◦C and 25 ◦C, mainly for the 'Pintat' cultivar, but germination was slower in both cases. The values with the same letters did not significantly differ at the 5% level. No significant differences were found when comparing germination velocities among the cultivars at each specific temperature.

The regression lines, indicating the response of germination velocity to increasing temperature (Figure 3), showed a steeper slope for the local 'Pintat' cultivar than for the commercial one, labelled as 'Peru', given the shorter mean germination time; i.e., faster germination. This effect became evident at the temperatures exceeding 19 ◦C. When the thermal time, S and Tb parameters were calculated from the regression line equations, the 'Pintat' seeds gave values of 131.6 ◦C·day−<sup>1</sup> and 5.2 ◦C respectively, with 185 ◦C·day−<sup>1</sup> and <sup>−</sup>15.0 ◦C for 'Peru'.

**Figure 2.** The germination percentage values obtained at different temperatures for the studied *Phaseolus lunatus* cultivars. The same letters indicate homogeneous groups (*p* < 0.05).

**Table 3.** Mean germination time (days) at different temperatures (◦C) for 'Pintat' and 'Peru' cultivars. The same letters indicate homogeneous groups among temperatures (*p* < 0.05) for each cultivar.

**Figure 3.** The linear regression of the germination rates (MGT) related to the tested temperatures for two cultivars: 'Pintat' and 'Peru'.

#### 3.2.2. Germination Response to Drought Stress

Characteristically, germination was affected by rising PEG concentrations. In both cases, a drastic reduction in germination was recorded from −4 bars, and no germination took place at −5 bar. However, Figure 4 and Table 4 show better tolerance to induced water stress for the 'Pintat' cultivar, which obtained higher germination percentages and velocity under all the tested conditions. At −2 bar, no significant differences appeared in relation to the control for 'Pintat' landrace, while germination lowered by 28.8% for the cultivar 'Peru'.

**Figure 4.** The accumulative germination percentages of the studied cultivars at increasing osmotic pressures obtained with PEG from 0, the control, to a maximum of −5 bar.

**Table 4.** Mean germination time (MGT), expressed as days, for the *Phaseolus lunatus* cultivars analyzed at increasing PEG 6000 concentrations. The same letters indicate homogeneous groups among cultivars and the tested concentrations (*p* < 0.05).


The germination test conducted at increasing water stress pointed out differences in the seeds of the studied cultivars for their physiological potential to face water deficit. A drastic drop in germination was recorded at −4 and −5 bars. The cultivar 'Pintat' demonstrated better tolerance to water stress, which obtained values above 50% for the germination percentage for all the tested osmotic potentials up to −4 bar.

The 'Pintat' cultivar displayed a faster response to germination velocity under all the conditions, and only showed a clear decrease from −2 bar (Table 4; Figure 5).

**Figure 5.** The relation between osmotic potential (bar) and germination rate (1/MGT) for the studied cultivars at 30 ◦C.

The hydrotime calculated from the linear regression slope was 37.5 and 44.1 bar·day for the cultivars 'Peru' and 'Pintat', respectively. The theoretical values calculated for the minimum osmotic potential (Ψb) at which radicle emergence was prevented were respectively −9.8 and −12.5 bar for these same cultivars. When PEG exposure ended, non-germinated seeds were transferred to the non-stressed medium. After 15 days of incubation in distilled water, no recovery was observed at any tested concentration.

#### *3.3. Genetic Assays*

The dataset herein considered comprised new 29 sequences, including three nuclear and five chloroplastic concatenated fragments that belong to the four more common *P. lunatus* cultivars in Spain. The phylogenetic analyses included the ITS, *Atpb-rbcL* and *trnL-trnF* fragments, two individuals of each Valencian landrace and all the accessions provided by Serrano-Serrano et al. [22] in NCBI. Seventy-eight individuals were analyzed. The MAFFT algorithm produced an alignment of 1828 bp with outgroups and 1410 without them. After the automatic removal of ambiguously aligned positions in GBlocks v. 0.91b, 97% (1781 nucleotides) of the original length, 13 selected blocks were kept after taking the outgroups into account. This final alignment included 110 variable positions, of which 73 were parsimony informative and 37 were singletons. The MrBayes analysis reached an average standard deviation of split frequencies of 0.01 after 156 generations. The resulting topology is presented in Figure 6, where the Valencian cultivars were clustered in the AI gene pool, together with the Andean Cordillera accessions from Ecuador and Peru with high clade support (BI ≥ 0.9). These landraces also formed a high supported clade inside the AI gene pool (BI = 0.98). The main groups also displayed good clade support (BI ≥ 0.9), except for the MII gene pool (BI = 0.61), which was clustered in a wider and well-supported Mesoamerican group, split inside.

**Figure 6.** Phylogram depicting the phylogenetic relations among the *P. lunatus* accessions from Spain and South America obtained with MrBayes and based on nrITS and cpDNA data. Support values are given for the main nodes (BI). Colors correspond to the gene pools for wild *P. lunatus*: black branches belong to outgroups, purple branches to AI (Andean I) and the purple triangle inside represents collapsed clades of Valencian landraces (local and 'Peru') as they were almost genetically identical, the orange triangle represents the MI (Mesoamerican I) and MII (Mesoamerican II) collapsed clades as they did not provide any relevant information for our purposes. The whole tree is shown in the Supplementary Material (Figure S2). COL = Colombia, ECU = Ecuador, PER = Peru, ESP = Spain.

Polymorphism and divergence analyses were conducted throughout two groups: only the Valencian cultivars and Valencian and South American cultivars from Serrano-Serrano et al. [22], excluding outgroups. All the analyzed fragments were used in the 29 sequences of the Valencian group with very low genetic diversity estimates. There were no gaps and 4080 sites, of which only four were variable and none showed any pattern of change. This group presented nine haplotypes, 3.8 <sup>×</sup> 10−<sup>4</sup> of nucleotide diversity and a non-significant genetic differentiation estimate Kst of 0.022. The group including the South American varieties comprised 67 sequences of concatenated ITS, *Atpb-rbcL* and *trnL-trnF* fragments, 1800 sites and 1413 sites excluding gaps, 44 of which were variable. The group showed 37 haplotypes, a nucleotide diversity of 3.89 <sup>×</sup> 10−<sup>3</sup> and a significant genetic differentiation estimate Kst of 0.528 (Table 5).

**Table 5.** The polymorphism and divergence data of the two *P. lunatus* cultivars groups. The Valencian landraces included the most frequently used cultivars in Spain ('Peru', 'Pintat', 'Ull de Perdiu', 'Cella Negra'). The Valencian + South American group included the Valencian and South American accessions from Serrano-Serrano et al. [22]. N: number of individuals, n: number of sites, n': number of sites excluding sites with gaps/missing data, S: number of variable sites, h: number of haplotypes, Π (s.d.): nucleotide diversity and standard deviation in brackets, Hd (s.d.): haplotype diversity and standard deviation in brackets, Kst: genetic differentiation estimate and its *p*-value (n.s.: non-significant, \*\*\*: *p* < 0.001).


#### **4. Discussion**

A landrace differs from a variety that has been selectively modified to improve particular characteristics. These traditional landraces, cultivated continuously for years, are severely threatened by genetic extinction because they are replaced with modern varieties, selected mainly for their higher productivity instead of their resistance to climate change consequences [43].

Currently, the commercial white-seed bean ('Peru') is the cheapest and the most widely sold among lima beans in the Valencian Community, and probably the only one known to most people. Seeds of 'Pintat', and rarely of 'Ull de Perdiu' are sold only in a few local markets, while the cultivar 'Cella Negra' has practically disappeared. The EEA-Carcaixent conserves and multiplies a few accessions of the cultivar 'Cella Negra' for its preservation, from the few seeds that it has been able to find from some farmers who still cultivate it for their own use. We focused our research according to the assumption that the commercial predominance of the different cultivars is not a question based on consumer preferences, but on local farmers' low profitability.

Local landraces are associated with one specific geographical location and, therefore, present climatic adaptability. They are generally better adapted to abiotic stress than modern cultivars [44] and supporting the recovery of their cultivation can mean advantages to face the climate change threat, especially if consumer demand increases. Hence, this climatic adaptability reveals the need to conserve the landrace germplasm as a means to provide information about adaptations to drought and heat stress, and because it constitutes a tool to identify stress-tolerant alleles to improve productivity when faced with climate change [45,46].

Baudoin [47], after thoroughly reviewing the diversity of *Phaseolus lunatus*, already indicated this species as an underexploited crop with a very high cultivation potential given its ability to withstand several types of stress, including severe drought. Moreover, the necessity of carrying out preservation programs for germplasm banks of wild forms and landraces was highlighted.

Regarding seed morphology, clear variability that depends on cultivars is described in the literature. Additionally, variations in dimensions, test patterns and color in cultivars from different countries

are known as Potato with small rounded seeds, along with Sieva, with medium-sized reniform seeds, while Andean ones are known as Big Lima and have large, but flat, seeds [14].

In the four studied landraces, no major differences in seed dimensions appeared. Seeds have the morphological characteristics of most of the individuals cultivated for commercial purposes, mainly with big, attractive and nutritional seeds, which indicate their Andean origin.

Regarding seed color and according to bibliographic references, the most frequent seed coat color of the cultivated plants differs depending on the considered geographic area. White seeds are one of the most frequently found among Cuban cultivars [48], while a predominant bicolored pattern is observed in the cultivars grown in Peru [12]. The studied landraces exhibit different bicolored patterns, which also agrees with their geographical provenance.

The seeds of the studied landraces underwent fast germination with no primary dormancy trait, even though dormancy was detected in some colored lima beans [49]. The marked preference for high temperatures in this species stood out, as clearly evidenced by our results with an optimal germination response at 30 ◦C. Indeed, Polock and Toole [50] and Polock [51] indicated that temperatures below 25 ◦C in the imbibition phase can be harmful. Other authors have indicated a good response of lima beans at high temperatures (25 ◦C and 30 ◦C), even when they were exposed to different salt concentration levels [52].

When we compared germination behavior of the commercial cultivar and the landrace 'Pintat' for the different tested temperature regimes, a stronger competitiveness of the local cultivar was observed from 19 ◦C to the optimum temperature, close to 30 ◦C.

Drought is one of the most important problems in agriculture as it leads to reduced yields and loss of crops. Water availability is essential for plants, as they need a good water supply throughout their life cycle. Therefore, water deficit in plants affects all phases of their development, physiological processes, growth and production which, under extreme conditions, can lead to plants dying [43,53,54]. Like the exposed response to temperatures, our results also support the hypothesis of the higher tolerance of those landraces cultivated for years that better adapt to changes in environmental conditions deriving from the Mediterranean climate. In fact, the 'Pintat' landrace was the most tolerant to water stress, simulated by lowering the osmotic potential of PEG solutions. In fact, the thermal time and hydrotime parameters have proven to be good discrimination tools to identify drought- and high temperature-tolerant common bean cultivars [55].

Conversely, DNA barcoding has not found any differences between the local and commercial cultivars used in the Valencian Community, not even when using different sorts: chloroplastic, nuclear, codificant and non-codificant markers. Nevertheless, their origin can be clearly situated. Recent phylogenetic studies have used genome-wide SNP markers polymorphisms [56] to indicate that the wild lima bean is structured into three gene pools, as previously proposed by Serrano-Serrano et al. [22]: the Mesoamerican one (MI); the Mesoamerican two (MII); the Andean one (AI). Their geographic ranges do not generally overlap. In addition, Chacón-Sánchez and Martínez-Castillo [56] also suggest the existence of another Andean gene pool (AII) in central Colombia. Our phylogenetic analyses, based on the data of Serrano-Serrano et al. [22], placed the Valencian cultivars in AI in relation to the 'Big Lima' morphology. These cultivars were phylogenetically grouped with the Andean Cordillera accessions from Ecuador, this being the domestication area of the Andean gene pool located between Ecuador and northern Peru [22,57].

For the Mesoamerican landraces, recent evidence indicates a scenario of a single domestication event in the gene pool MI for all the Mesoamerican landraces, perhaps in central-western Mexico, and the subsequent admixture among landraces and wild populations within the distribution range of gene pool MII, which gave rise to the MII landraces [56]. Therefore, and according to our results, these previous studies have shown that domestication was accompanied by strong founder effects that decreased the genetic diversity of the landraces in the Andes and MI of Mesoamerica. Thus, low polymorphism and divergence statistics have been found in the cultivars used in the Valencian Community (Spain), even between traditional ('Pintat', 'Ull de Perdiu', 'Cella Negra') and commercial ones ('Peru'). However, they all came from the same original gene pool in which a split occurred, as the earliest, when Europeans arrived in America 500 years ago, which is a negligible time in evolutionary terms.

Although other genome-wide barcoding techniques can be used [56,58], the different responses of these genetically close landraces can be explained by epigenetic mechanisms or by a few genes that play a relevant role in crop stress responses [59]. Indeed, rather than DNA barcoding, the search for relevant genes and local landrace alleles related to water stress tolerance could lead to new research works to help preserve these cultivars, by identifying the particular genetic features and their purity. When considering crop tolerance to overcome climate change-related stresses, natural variance among different cultivars can act as genetic reservoir for adaptation capability [60]. This idea, combined with an interest in providing added value to local landraces to defend their use recovery and agro-biodiversity conservation, could supply key future tools that promote local activities to face climate change effects on crops in order to contribute to the auto-sustainability of agronomy activities.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4395/10/6/758/s1.

**Author Contributions:** Conceptualization, E.E., J.R. and P.S.; Investigation, M.I.M.-N., E.E. and P.S.; Methodology, M.I.M.-N., E.E., J.P.-M. and P.S.; Project administration, E.E. and P.S.; Visualization, M.I.M.-N., E.E. and P.S.; Writing—Original draft, M.I.M.-N., E.E. and P.S.; Writing—Review & Editing, M.I.M.-N., E.E., J.P.-M., J.R. and P.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors wish to thank the Estación Experimental Agraria de Carcaixent (Valencia, Spain) for its support, particularly Fernando Amorós Ortega for providing us with some of the plant materials (seeds and leaves) used in the experiments. The authors sincerely acknowledge the valuable comments, corrections and suggestions made by anonymous reviewers that have significantly improve the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
