Glyphosate effects on non-target plants: collateral impacts on elemental composition and growth of yerba mate

ABSTRACT This work investigated collateral effects on non-target yerba mate (Ilex paraguariensis A. St.-Hil) plants when using glyphosate for weed control; growth and leaf elemental composition were examined. Special emphasis was placed on examining the heavy metals Cd and Pb due to regulations defining maximum limits in South American infusion products. The experiment was conducted in pots using glyphosate [applied to Congo grass (Brachiaria ruziziensis), soil surface application, and control], two P rates (with and without P), two clonal yerba mate cultivars, and two different soils (basalt- and rhyodacite-derived). When using glyphosate to control Congo grass, total dry matter and basal diameter of yerba mate plants decreased. This effect was enhanced by lack of P. Collateral effects of glyphosate use resulted in increased K, P, and Cu in clone 1 cultivated in basalt-derived soil; with rhyodacite-derived soil, increased levels of Ca, Mg, Mn, Fe, and Zn were observed in clone 1, while clone 2 displayed increases in B, Cd, and Pb. These observations demonstrate that glyphosate use to control undesirable plants can impact initial development and elemental composition of yerba mate, with variations between cultivars and soil type.


Introduction
For centuries, yerba mate (Ilex paraguariensis A. St.-Hil) leaves have been exploited to produce infusion drinks in southern regions of South America. Historically, there was a predominance of extractivism of this species, which grows naturally in forest understories (Cardozo-Junior and Morand 2016) under very acidic soil conditions (Reissmann et al. 1999;Toppel et al. 2018). However, increased consumption and consequent increase in demand for raw material have resulted in cultivations outside of native forests . Such commercial operations often use herbicides (especially glyphosate) and fertilizers to increase productivity. Due to positive yerba mate response, phosphate fertilization  has also become popular among producers.
Worldwide, glyphosate is the most widely used herbicide (Beckie et al. 2020). This compound inhibits the activity of the enzyme 5-enolpyruvyl-shikimate-3-phosphate-synthase in the shikimate pathway and hinders synthesis of amino acids (tryptophan, phenylalanine, and tyrosine), which leads to plant death (Zablotowicz and Reddy 2004). In addition, C synthesis is disrupted and causes oxidative damage (Zobiole et al. 2011(Zobiole et al. , 2012. When applied to plants, glyphosate penetrates cell interiors, is rapidly translocated due to high mobility in the phloem, and quickly reaches roots and apical meristems (Yamada and Castro 2007). Glyphosate can be exuded from plant roots, which can promote changes in the rhizosphere and negatively impact soil microbiological communities (Lane et al. 2012;Kumar et al. 2017), especially colonizing mycorrhizae (Carvalho et al. 2014). Furthermore, Rodrigues et al. (1982) and Neumann et al. (2006) observed the absorption of glyphosate by nontarget plants from root exudates of target plants (via transfer through contact between root systems).
Glyphosate also reaches the soil via foliar wash and spray residue (Ellis and Griffin 2002). Due to their high sorption capacity, glyphosate molecules can remain in soil as a residue by binding to humic substances, metallic atoms, and oxides (Yamada and Castro 2007). However, there is evidence that simultaneous application with phosphate fertilizer can reduce soil sorption of glyphosate due to the preference of phosphate for binding sites (Munira et al. 2016;Pereira et al. 2019).
Studies evaluating nutrient accumulation in glyphosate-contaminated plants show a decrease in leaf Mn in sunflower (Tesfamariam et al. 2009) and soybean (Duke et al. 2012a), and a reduction in Fe, Mn, Ca, and Mg in soybean seeds ). Changes are attributed to these nutrients forming insoluble complexes with glyphosate that decrease nutrient absorption and translocation. França et al. (2010) reported a decrease in Mn, Fe, and P in coffee leaves (with different behavior among evaluated cultivars) when evaluating the effects of glyphosate on the nutrients N, P, K, Ca, Mg, Fe, Zn, Mn, and Cu. Most other studies with coffee (Barbosa et al. 2020a), citrus (Gravena et al. 2012), Camellia sinensis (L.) (Rana et al. 2020), and Eucalyptus grandis (W. Hill ex Maiden.) (Pereira et al. 2019) only evaluated glyphosate effects on plant growth or physiology.
The use of glyphosate and phosphate fertilizer raises concern regarding possible changes in elemental composition of yerba mate leaves, especially for the heavy metals Cd and Pb. Low concentrations of these potentially toxic elements are naturally present in yerba mate leaves (Valduga et al. 2019;Frigo et al. 2020;Pardinho et al. 2020;Barbosa et al. 2020b) and commercial products (Ulbrich et al. 2022). In 2013, MERCOSUR (Common Market of the South) technical regulations defined maximum limits for Cd and Pb in teas and infusion drinks sold in South American countries. This legislation included yerba mate infusion products and set maximum limits of 0.40 and 0.60 mg kg −1 for Cd and Pb, respectively (ANVISA 2013). Research to date indicates that natural concentrations of Cd and Pb in yerba mate leaves may be very close to or above maximum limits established by this legislation Magri et al. 2021).
Since these elements can pose risks to human health (Boskabady et al. 2018;Hamid et al. 2019), investigating the presence of Cd and Pb in food crops is essential. In addition, no studies have evaluated the impacts and interaction of glyphosate and phosphate fertilizer use on levels of potentially toxic elements in yerba mate leaves. We believe that the use of glyphosate can have collateral impacts on growth and elemental composition of yerba mate, and possible interactions with phosphate fertilization. Thus, our objective was to test how glyphosate and phosphate fertilizer usage alters growth and leaf elemental composition (especially Cd and Pb) of yerba mate cultivars grown on two different soils (basalt and rhyodacite parent material).

Experimental design
An experiment was conducted with four factors: Factor 1 -glyphosate application (applied to Congo grass (Brachiaria ruziziensis), soil surface application, and control); Factor 2 -phosphorus supply (with and without P); Factor 3 -two clonal yerba mate cultivars; Factor 4 -two different soils (basalt and rhyodacite parent materials). There were five replications for each treatment. Thus, the experiment was a 3 × 2 × 2 × 2 factorial. Soils of igneous origin were collected from the state of Rio Grande do Sul in Brazil. Rhyodacitederived and basalt-derived soils were collected from the municipalities of Ilópolis (28°54ʹ50.89" S and 52°7ʹ52.38" W) and Barão de Cotegipe (27°33ʹ50.57" S and 52°24ʹ4.01" W), respectively. Soils from these parent materials were selected since they represent regionally predominant types used in yerba mate production (Magri et al. 2021). The collection sites had no history of using glyphosate, fertilizers, or soil correctives.
Prior to soil collection, weeds and plant debris were removed from the soil surface. Soils were collected from the 0-20 cm depth and roughly screened in the field to remove roots and stones. Sieving (4 mm aperture) and homogenization were performed prior to placing 7 L of soil into 8 L plastic pots.
For phosphorus supply, solutions containing Dibasic Ammonium Phosphate [DAP; (NH 4 ) 2 HPO 4 ] were added to each pot at a rate of 64.93 mg kg −1 , and homogenized before seedlings were planted. Since DAP has NH 4 + in its formulation, urea (CO(NH 2 ) 2 ) was supplied to pots not receiving phosphate fertilization (980 mg of urea per pot); this was equivalent to the N rate supplied by DAP. After this preparation, yerba mate clonal seedlings were planted in each pot; the two clonal cultivars (Clone 1: BR5-BLD Yary and Clone 2: BR5-BLD Aupaba) were propagated by the microcutting technique (Wendling et al. 2017). These seedlings were grown in 120 cm 3 tubes (125 mm high × 34 mm in diameter) and had attained heights of ~15-20 cm at transplanting. In December 2017, the experiment was initiated in an open experimental area located at the Agricultural Sciences Sector of the Federal University of Paraná, Curitiba-PR, Brazil. The experimental site had an altitude of ~910 m and a Cfb climate according to Köppen classification (Alvares et al. 2013).
Three-hundred fifty-four days after planting yerba mate seedlings, 40 Congo grass (Brachiaria ruziziensis Germ. & C.M. Evrard; Syn. Urochloa ruziziensis) seeds were sown to simulate a weed infestation. Commercial glyphosate (Monsanto Roundup Original® DI glyphosate, acid equivalent 370 g L −1 ) was applied 48 days after planting Congo grass. For this procedure, a 15 mL L −1 glyphosate solution was prepared; from this solution, an equivalent of 2.5 L of glyphosate ha −1 was applied to each pot requiring an application using a calibrated CO 2 pressurized sprayer (set at a constant pressure of 250 kPa) equipped with a boom having two fan-type spray tips (50 cm apart). Although glyphosate is not labelled for yerba mate, the amount used was typical of that used in commercial yerba mate productions. For glyphosate application in weed and soil treatments, the base of yerba mate plants was wrapped with plastic to avoid contact with glyphosate. Glyphosate applications were conducted in a windfree environment and were protected from water for 24 h to avoid washing the product from leaves. In the control treatment and in the treatment that received glyphosate to the soil, manual weeding was conducted periodically.
One-hundred sixty-three days after glyphosate application (~1.5 year from study initiation), stem diameters 2 cm above the ground were measured with digital calipers and plants were harvested. Six mature leaves of similar physiological age were also collected from the upper third of each plant canopy. These leaves were washed under running water and rinsed three times with distilled water. Samples were dried in a forced-air oven (60°C) until mass remained constant before determining total dry matter.

Soil and plant analysis
Prior to fertilizer addition, samples were collected from each homogenized soil for chemical and granulometric characterization (Table 1 and Table S1). Samples were air-dried and disaggregated to pass through a 2 mm sieve. Granulometry of these samples was determined by the Bouyoucos hydrometer method using a mixture of sodium hydroxide (4 g L −1 ) and sodium hexametaphosphate (10 g L −1 ) as a dispersing solution (Gee and Or 2002). Soil chemical determinations were made using the methodologies of Marques and Motta (2003): determinations of pH (0.01 mol L −1 CaCl 2 ), potential acidity (H+ Al), exchangeable Al, Ca 2+ , and Mg 2+ (extracted with KCl 1 mol L −1 ; Al quantification by titration; Ca and Mg by atomic absorption spectrophotometer), P and K + (extracted by Mehlich I; P determination by colorimetry and K by flame photometer), organic matter (Walkley Black method), and available levels of Mn, Fe, Zn, Cu, Ni, and Ba (extracted by Mehlich I; quantified by Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES, Varian 720-ES)) .
For pseudo-total determination, soil samples were again disaggregated and passed through a 0.21 mm sieve. This determination followed the USEPA 3051A methodology (USEPA 1995) involving the digestion of 400 mg of soil in 9 mL of 65% nitric acid and 3 mL of 36-38% hydrochloric acid. Digestion was performed in a microwave oven (Mars 6 -CEM Corporation) for 4.5 min at a power setting between 1200 and 1800 W, a heating ramp of 5.5 min, and a maximum temperature of 175°C. After digestion, samples were brought to volume (50 mL) using ultra-pure water (Milli-Q system with resistance of 18.2 MΩ cm -Millipore Milli-Q Academic). Elements were quantified by ICP-OES.
Leaf samples were crushed in a knife mill to pass through a 0.25 mm sieve. According to methodology described by Magri et al. (2021), leaf digestion was conducted in a microwave oven using 200 mg of sample, 4 mL of nitric acid, and 1 mL of hydrogen peroxide. Digestion of certified reference material (CRM -C. sinensis GBW 10052) and blank samples was also performed. Quantification of K, Ca, Mg, P, Mn, Al, Fe, Zn, Ba, B, Cu, and Ni was performed by ICP-OES. Quantification of Cd and Pb was conducted by atomic absorption with a graphite furnace -GFAAS (model AA 6800 -Shimadzu). Equipment parameters used followed those described by Magri et al. (2021). Recovery of quantified elements ranged from 74% to 104% (Table S2).

Statistical analysis
Data were tested for normality of residuals using the Shapiro-Wilk test. Variables not showing normality were transformed using Box-Cox transformation. Growth parameters and plant tissue elements were compared between treatments with glyphosate and with addition of phosphate fertilizer by mean comparison tests. This comparison was carried out individually for each clonal cultivar in each soil, adopting a two-way ANOVA followed by Tukey's test in cases of significant F (p < 0.05). Additionally, from the correlation matrix of normalized and standardized data, principal component analysis and discriminant analysis were performed to verify differences between clonal cultivars and soil parent materials. Statistical analyses were performed using R software version 4.0.3 (R Core Team 2022).   Clay and Silt (hydrometer method); pH (CaCl 2 0.01 mol L −1 ); Ca 2+ , Mg 2+ , Al 3+ (KCl 1 mol L −1 extraction); H+ Al 3+ (calcium acetate 0.5 mol L −1 extraction); organic matter (OM; volumetric method by potassium dichromate); K + , P, micronutrients, and Ba (Mehlich-1 extraction).

Plant growth attributes
Glyphosate application to control Congo grass reduced basal stem diameter and total dry matter of yerba mate clone 1 grown in basalt-derived soil without phosphate fertilization (Figure 1). A similar effect was also noted for stem diameter of clone 2 plants grown in the same soil, regardless of P addition. Although no statistical differences were identified for plants cultivated in rhyodacite derived-soil, yerba mate plants in treatments with glyphosate applied to Congo grass had the lowest stem diameter and total dry matter values. Furthermore, yerba mate plants from both clones and soils showed leaf yellowing after application of glyphosate to Congo grass.

Glyphosate and phosphorus effects on elemental composition
When glyphosate was applied to Congo grass, clone 1 plants cultivated on basalt-derived soil had higher K and P in leaves relative to the control ( Table 2), independent of P fertilization. When P was added, leaf Cu was lower in the control compared to Congo grass and soil glyphosate applications for this same clone and soil (Table 3). On the other hand, clone 1 cultivated under rhyodacite-derived soil had higher Zn and Fe when glyphosate was applied on Congo grass compared to the control and less Ca, Mg, Mn, B, and Ba with soil glyphosate application compared to other treatments, independent of P fertilization. Different values for uppercase letters (glyphosate effect) and lowercase (P effect) indicate significant difference by two-way ANOVA followed by Tukey test (p < 0.05). The presence of an asterisk represents a significant F value. For clone 2 cultivated on basalt-derived soil, treatments using glyphosate on Congo grass had higher leaf P compared to the control. In comparison, this clone cultivated in rhyodacite derived-soil displayed higher values of B, Cd, and Pb (Table 2, Table 3, Table 4) .
For clone 2 grown in rhyodacite-derived soil, leaf Cd concentrations were two times higher when P was not added to soil and when glyphosate was applied on Congo grass. Also, P fertilization resulted in a decrease in leaf K compared to the no P treatments of both clones in both soils evaluated. Fertilization decreased leaf Zn in clone 1 cultivated on rhyodacitederived soil, while leaf Cd only decreased for clone 2 in this soil. Concentrations of Ca, Mg, P, Mn, Fe, Cu, Ni, and Pb were generally higher in plants that received P fertilization. Among these elements, we highlight the small statistical increase of Pb in clone 2 cultivated on basalt- Different values for uppercase letters (glyphosate effect) and lowercase (P effect) indicate significant difference by two-way ANOVA followed by Tukey test (p < 0.05). The presence of an asterisk represents a significant F value. derived soil when P was applied compared to no applied P in both glyphosate treatments. In addition, Ni increased in both clones cultivated on rhyodacite-derived soil (doubling in clone 1 and increasing three to five times in clone 2).

Soil and plant genetic effects on elemental composition
Leaf concentrations of Mg, Zn, and Cd were higher in plants grown on basalt derived-soil, while Al and Ni were higher when cultivated on rhyodacite derived-soil regardless the addition of glyphosate. Calcium and Mn concentrations were also higher for basalt derived-soils, but this effect was restricted to clone 2. Regarding comparison between clones, the main difference was higher P and K in clone 2 (Table 2). Discriminant analysis had an efficiency of 87% when separating samples according to soils and clones used in the experiment. Discrimination took into account elemental levels of K, Ca, Mg, P, Mn, Zn, Al, Ni, Cd, and Pb. As seen in the discriminant analysis matrix (Table 5), soil was more sensitive than plant genetic variation in sample discrimination. Although samples were not efficiently discriminated by clone type, there was a clear difference in leaf elemental composition when considering the two soils used in this study. These differences can also be observed in raw values (Table 2,  Table 3, Table 4) and in the principal components analysis (Figure 2 and Table S3).

Discussion
According to Eker et al. (2006), observed yellowing of yerba mate leaves after applying glyphosate to Congo grass may be associated with Fe and Mn inactivation due to accumulation of glyphosate in leaf tissue. This process may occur due to the high capacity of glyphosate to bond with divalent cations, resulting in immobilization in plant tissue (Bellaloui et al. 2009;Cakmak et al. 2009;Zobiole et al. 2011). In glyphosate-resistant soybean plants, yellowing after use of glyphosate is also common (Zablotowicz and Reddy 2007). Rather than a soybean nutritional imbalance, toxicity from aminomethylphosphonic acid (main metabolite generated in the degradation of glyphosate; Duke et al. 2012b) affects the synthesis of chlorophyll Serra et al. 2013) possibly due to an increase in ethylene (Yamada and Castro 2007).
Uptake of glyphosate by both leaves and roots, as well as mobility in the phloem and xylem, is directly affected by the availability of P (Pereira et al. 2019). When plants are grown under low P availability, increased expression of high-affinity P transporters is induced (Gu et al. 2016). Studies indicate that these transporters are responsible for transporting glyphosate within plants (Fitzgibbon and Braymer 1988;Pereira et al. 2019). Under low P availability in our study, glyphosate was likely absorbed in greater amounts by Congo grass, translocated and exuded in greater quantity by their roots, and absorbed in greater quantity by yerba mate roots. Since the clay content in basalt derivedsoil is much higher than in rhyodacite-derived soil (Table 1), more P was adsorbed and not readily available to plants (Poggere et al. 2020). For this reason, growth effects associated with P use were more pronounced for the basalt derived-soil (Figure 1).
Several studies suggest that plant uptake of glyphosate from soil is practically nil, due to the high adsorption force on soil colloids (Neumann et al. 2006;Tesfamariam et al. 2009;Bott et al. 2011). However, there is no doubt concerning resorption after exudation by plants treated with glyphosate (Rodrigues et al. 1982;Neumann et al. 2006) and contact of yerba mate roots with Congo grass roots may have possibly allowed some transport and absorption of glyphosate. In our case, effects on yerba mate growth ( Figure 1) and elemental composition (Table 2, Table 3, Table 4) were more pronounced when glyphosate was directly applied to Congo grass rather than direct application to soil.
The occurrence of chlorosis without nutritional deficiency was indicative of glyphosate absorption by yerba mate plants. Santos et al. (2008) identified the presence of glyphosate in all eucalyptus plants intercropped with Congo grass that received glyphosate, but not in sufficient quantity to cause injury. Tong et al. (2017) evaluated the uptake and translocation of glyphosate in C. sinensis plants grown in hydroponic solution, and found that glyphosate was absorbed by roots and transferred to leaves. This process occurred from application until the fifth day when glyphosate concentration in leaves began to decrease. This implies that tea consumers can be exposed to glyphosate when applied a few days before harvest. Specific studies are required to determine if glyphosate is present in yerba mate leaves. If true, wait periods between glyphosate application and leaf harvest may need to be determined.
In the present study, glyphosate application to Congo grass did not statistically decrease leaf concentration of any evaluated element (Table 2, Table 3, Table 4). There was only a decrease when glyphosate was applied directly to the soil. However, changes in plant elemental composition caused by glyphosate use are very contradictory since some studies indicated decreased elemental concentrations França et al. 2010), increased concentrations (Zobiole et al. 2010(Zobiole et al. , 2011(Zobiole et al. , 2012, or no concentration change (Bailey et al. 2002;Rosolem et al. 2010). In this regard, soybeans have also displayed both increases and decreases in elemental concentrations (especially metals) that primarily varied by cultivar (Cavalieri et al. 2012). Findings generally indicate small variations between species or even between cultivars. The use of yerba mate cultivars is still incipient, since cultivation are predominantly produced from seeds and plants extracted from native habitats. Thus, large variation in collateral effects of glyphosate use on elemental composition of field-grown yerba mate can be expected.
Considering the potentially toxic elements Cd and Pb, there is a concern about their presence in yerba mate leaves since it is classified as a food and fall under specific legislation. However, even control treatments had values (Table 4) above those established by legislation (Brasil 2013). Additionally, even with observed alterations due to glyphosate use, values were lower than natural accumulation potential (Magri et al. 2021). These authors also indicated that foliar Cd was dependent on Zn, as was also shown in our study (Figure 2). In addition, intake of Cd and Pb by yerba mate consumers is very low (Barbosa et al. 2015;Magri et al. 2021;Ulbrich et al. 2022), indicating that legislation governing elemental levels in yerba mate products is inconsistent with the natural concentration present in yerba mate leaves. The results of this study suggest that soil type and plant genetic factors may have a greater influence on Cd and Pb levels than the use of glyphosate or phosphate fertilization.

Conclusion
The use of glyphosate to control unwanted plants can harm initial development of yerba mate on P-deficient soils, with distinct effects noted between clonal cultivars and type of soil in which plants were grown. Elemental composition of yerba mate leaves can be altered, as reflected by increases in some elements that differed between cultivars and soil type. Regarding the potentially toxic elements Cd and Pb, there were small increases in some cases that exceeded maximum limit values proposed by South American legislation regulations. Nonetheless, these changes were not expressive to point of being a matter of concern to producers and consumers.