Introduction
Weeds are the main biotic factor responsible for yield losses in agricultural crops and a threat to global food security (Délye et al., Reference Délye, Jasieniuk and Le Corre2013). Herbicides are the most practical, efficient and least costly method of controlling these species (Harker and O'Donovan, Reference Harker and O'Donovan2013; Heap Reference Heap2014). However, excessive use leads to the emergence of herbicide-resistant weeds (Bagavathiannan and Davis, Reference Bagavathiannan and Davis2018).
Herbicide resistance (HR) in weeds is a natural response to selection pressure (Vencill et al., Reference Vencill, Nichols, Webster, Soteres, Mallory-Smith, Burgos, Johnson and McClelland2012). It is the inherent and inheritable ability of some individuals in a population to survive and reproduce after exposure to a lethal dose of the product (Norsworthy et al., Reference Norsworthy, Ward, Shaw, Llewellyn, Nichols, Webster, Bradley, Frisvold, Powles, Burgos, Witt and Barret2012; Busi et al., Reference Busi, Vila-Aiub, Beckie, Gaines, Goggin, Kaundun, Lacoste, Neve, Nissen, Norsworthy, Renton, Shaner, Tranel, Wright, Yu and Powles2013; Heap, Reference Heap2014). Herbicides eliminate susceptible plants, allowing resistant individuals to predominate, limiting or precluding chemical management, a major concern in modern agriculture (Burgos et al., Reference Burgos, Tranel, Streibig, Davis, Shaner, Norsworthy and Ritz2013).
In recent years, glyphosate has become the most widely used herbicide in chemical management (Heap and Duke, Reference Heap and Duke2018). It is a non-selective foliar-applied product used to manage annual and perennial weeds (Kleinman et al., Reference Kleinman, Bem-Ami and Rubin2015) and the only herbicide that inhibits the enzyme 5-enolpyruvylshikimate-3-phosphate synthase on the shikimate pathway (Heap and Duke, Reference Heap and Duke2018), causing plant death (Gomes et al., Reference Gomes, Smedbol, Chalifour, Hénault-Ethier, Labrecque, Lepage, Lucotte and Juneau2014). However, intensive use in no-till systems, genetically modified and cover crops (González-Torralva et al., Reference González-Torralva, Gil-Humanes, Barro, Brants and De Prado2012) has increased resistance (Heap and Duke, Reference Heap and Duke2018).
Among the main glyphosate-resistant species worldwide, Italian ryegrass (Lolium multiflorum Lam.) biotypes are particularly prominent, occurring on four continents: the Americas, Europa, Asia and Oceania (Heap, Reference Heap2021). Lolium multiflorum is an annual winter grass considered one of the most important temperate forage species in the world (Wang et al., Reference Wang, Pembleton, Cogan and Forster2016). It is widely used as a cover crop or in integrated crop-livestock systems (ICLS) (Sandini et al., Reference Sandini, Moraes, Pelissari, Neumann, Falbo and Novakowiski2011) that are characterized by the annual rotation of pastures and crops in a no-till system where the main integrated farming is rotation or succession of summer crops with winter annual grazing grasses or successive natural pastures (Moraes et al., Reference Moraes, Carvalho, Anghinoni, Lustosa, Costa and Kunrath2014).
Due to its adaptability to different soil types, water stress tolerance, and easy dispersal and persistence in the soil seed bank, L. multiflorum interferes in winter crops, orchards, vineyards, maize, rice and soybean (Vargas et al., Reference Vargas, Moraes and Berto2007; Galvan et al., Reference Galvan, Rizzardi and Scheffer-Basso2011; Niinomi et al., Reference Niinomi, Ikeda, Yamashita, Ishida, Asai, Shimono, Tominaga and Sawada2013; Nandula, Reference Nandula2014). The presence of ryegrass reduces wheat yield by 62% (Paula et al., Reference Paula, Agostinetto, Schaedler, Vargas and Silva2011), making it a major weed worldwide (Ge et al., Reference Ge, D'Avignon, Ackerman, Collavo, Sattin, Ostrander and Preston2012).
As such, strategies are needed to mitigate the evolution of glyphosate resistance in Italian ryegrass. One of these strategies is preventing seed migration by reducing the dispersal of the resistance gene (Beckie and Harker, Reference Beckie and Harker2017; Bagavathiannan and Davis, Reference Bagavathiannan and Davis2018). This strategy is vital in cases involving glyphosate because the frequency of initial resistance is considered low. In practical terms, the dispersal of resistance is far more important as a source of new infestations than new cases that emerge in situ (Heap and Duke, Reference Heap and Duke2018).
Endozoochory is one of the main sources of seed dispersal in ICLS, with ruminants as dispersal agents, enabling the migration of herbicide-resistant weed seeds (Viero et al., Reference Viero, Schaedler, Azevedo, Santos, Scalcon, David and Rosa2018). Endozoochory, the dispersal of seeds via ingestion by vertebrate animals and their subsequent elimination in faeces (Alvarez et al., Reference Alvarez, Leparmarai, Heller and Becker2016), is an important process for weed species (Chuong et al., Reference Chuong, Huxley, Spotswood, Nichols, Mariotte and Suding2016). However, seeds ingested by ruminants undergo physical and chemical processes as well as microbial activity, affecting their recovery and germination (Lisboa et al., Reference Lisboa, Medeiros, Azevedo, Patino, Carlotto and Garcia2009; Oliveira et al., Reference Oliveira, Neto, Adelson and Valença2013; Milotić and Hoffmann, Reference Milotić and Hoffmann2016; Viero et al., Reference Viero, Schaedler, Azevedo, Santos, Scalcon, David and Rosa2018; Wang et al., Reference Wang, Hu, Zhang and Hou2019).
The aim of this study was to assess the recovery and germination of glyphosate-resistant L. multiflorum seeds that have passed through the digestive system of cattle.
Materials and methods
The research protocol was reviewed and approved by the Animal Ethics Committee of the Department of Agricultural Diagnostics and Research (DADR).
Lolium multiflorum seed collection and experimental procedures
Glyphosate-resistant L. multiflorum seeds (LOLMU-R) were collected from a biotype at the EMBRAPA Trigo company (28°16′S and 52°24′W) in the Passo Fundo city, Rio Grande do Sul state (RS), Brazil. The biotype was previously identified by dose-response curve studies after surviving chemical treatment in apple orchards (Vargas et al., Reference Vargas, Roman, Rizzardi and Silva2004). The recovery and germination experiments were conducted between June and December 2015, in two stages: recovery, carried out at the Center for Forage Research (30°21′S and 54°16′W) in São Gabriel (RS); and germination and resistance assessment at the Itaqui Campus of the Federal University of Pampa (RS) (29°12′S and 56°18′W).
Seed recovery
A completely randomized experiment with six replications was conducted to assess the recovery of LOLMU-R seeds that had passed through the digestive system of cattle. The treatments consisted of the following seed recovery times: 1, 2, 3, 4, 5 and 6 days after ingestion. Six Hereford steers with an average age of 2 years and an average weight of 350 kg were used, with each animal considered one replicate.
The animals were allocated in metabolism cages for 15 days for metabolic assays. Over the first 9 days, the animals were allowed to adapt to the environment and diet. The diet offered was composed of fresh forage from native grassland in sufficient quantity to meet the animal's nutritional requirement. Its botanical composition was mainly constituted by Axonopus affinis, Desmodiun incanum, Paspalum notatum and Paspalum plicatulum; and a chemical composition of 409 g/kg of dry matter, 71 g/kg of ash, 92 g/kg of crude protein and 604 g/kg of neutral detergent fibre. On day 9, LOLMU-R seeds were supplied via a flexible feeding tube manually inserted into the glottis, with 25 g of seeds/animal (12 112 seeds), estimated based on the 1000 seed weight of the LOLMU-R biotype.
The steers were fitted with polyvinyl chloride saddles and bags. The bags were changed three times a day to prevent the accumulation of faeces. The samples were homogenized every 24 h and a 10% subsample removed for analysis. These samples were washed under running water in a 1 mm mesh sieve to separate the LOLMU-R seeds. After recovery, the results were converted to 100% of faecal volume. The seeds were counted and pre-dried, then placed in plastic bags and stored at 6°C (±2°C) under 85% relative humidity (RH) until the germination tests.
Germination of the recovered seeds
A completely randomized design was used; with six replications (the seeds recovered from each animal were considered as different replications). The recovered LOLMU-R seeds were deposited onto Germtest® paper soaked in distilled water (equivalent to 2.5 times the weight of the paper), which was rolled up and placed into sealed plastic bags. The rolls were placed into a Biochemical Oxygen Demand (BOD) incubator at a constant temperature of 20°C, under a 12 h photoperiod and 70% RH.
For germination, only seeds recovered on the 2nd, 3rd and 4th days were considered, given the small number retrieved on the remaining assessment days. A germination test was also performed in the control treatment (no gut passage), using 50 seeds per repetition. At 14 days after sowing, the germination percentage (% in relation to the control) was determined considering normal seedlings with a root and shoot, in accordance with Regulations for Seed Analysis (Brasil, Reference Brasil2009).
Dose-response curve
In order to establish the resistance of the LOLMU-R seeds recovered after animal ingestion, a greenhouse experiment was carried out using a completely randomized design with four replications. After germination assessment, 100 seedlings were transplanted into 0.2 litres plastic pots (one plant per pot) filled with commercial substrate and maintained at field capacity. When the plants reached phenological stage 23 (BBCH, 1997), the maximum recommended dose (1080 g a.e/ha) of glyphosate [Roundup Original®, 360 grams of acid equivalent per litre – (g a.e/l)] was applied.
Twenty days after treatment, the tillers were transplanted to new experimental units (0.2 litres) – two tillers per pot, totalling 28 pots. The treatments were applied when the plants obtained from the tillers reached stage 23 (BBCH, 1997), with the following glyphosate doses: 0, 1080, 2160, 4320, 8640, 17 280 and 34 560 g a.e./ha. Application was performed using a CO2 pressurized sprayer positioned 0.5 m from the target, equipped with a boom containing flat spray tip nozzles (XR 110.015) spaced 0.5 m apart, at a working pressure of 250 kPa. The average temperature, RH and wind speed during application were 24°C, 69% and 6.5 km/h, respectively.
At 28 days after application, population control (%) was visually assessed. The control was assessed using a percentage scale where a score of zero (0%) was established for no control and 100% for plant death (Burril et al., Reference Burril, Cardenas and Locatelli1976). The remaining plants were cut at ground level and dried in a forced-air oven (60°C) until constant weight, when shoot dry weight (SDW) was determined.
Statistical analysis
The data obtained were analysed using R software (R Core Team, 2021). The recovered seeds and dose-response curve data were fitted via the three or four-parameter logistic model (3 or 4PL) (Ritz et al., Reference Ritz, Baty, Streibig and Gerhard2015), using the drc package (Eqns (1) and (2), respectively).
where Y is the resulting response value for dependent variables, either number of recovered seeds, control or SDW; d is the upper limit, defined by the maximum response from the number of recovered seeds or from non-treated plants (control or SDW); c is the lower limit, determined by the response levels from a high dose of herbicide – if c = 0, then the four-parameter model (Eqn (2)) reduces to the three-parameter model (Eqn (1)), with the lower limit being zero; D50 is the time or dose that causes 50% of recovery seed, control or SDW reduction; and b the slope of the curve at D50 (Ritz et al., Reference Ritz, Baty, Streibig and Gerhard2015).
The regression model was submitted to lack of fit testing using the modelFit function and when not significant (P ≥ 0.05) indicates that the data are well described by the selected model. The parameters and their standard errors were estimated using the drm function, with P values (P ≤ 0.05) dictating whether the parameters were significant. D50 was estimated using the ED function, with a 95% confidence interval. Data on SDW and control were converted into a percentage in relation to the control treatment.
Germination of the recovered seeds data was fitted via linear regression (Eqn (3)), using lm function (Kniss and Streibig, Reference Kniss and Streibig2019).
where Y corresponds to germination (%); a is the intercept of regression line, the predicted value when x = 0; b is the slope of the regression line; x = days after ingestion.
The regression model was submitted to lack of fit testing, treating the independent variable as a factor variable and comparing with the fitted linear model using the ANOVA function, when not significant (P ≥ 0.05), regression analysis describes data appropriately. The lm function was used to analyse whether parameters were significant (P ≤ 0.05), to estimate the standard errors and to determine the adjusted r2 (Kniss and Streibig, Reference Kniss and Streibig2019).
Results
Seed recovery
The number of glyphosate-resistant L. multiflorum (LOLMU-R) seeds recovered showed a sigmoidal behaviour from the 2nd day after ingestion by the steers (lack of fit, P ≥ 0.05). The average time to recover 50% of the seeds (D50) was 2.7 days (Table 1). Of the 12 112 seeds supplied per animal, 1109 (9.1%) were recovered during the assessment period, with 1027 retrieved in the first 3 days (92.6% of the total). However, during the first day after ingestion, no seed was recovered. Maximum recovery (728 seeds or 65%) occurred 2 days after ingestion, which declined to 3.33 seeds (or 0.3%) on day 6 (Fig. 1).
Fig. 1. Observed and predicted values for the number of glyphosate-resistant Lolium multiflorum seeds recovered from cattle faeces as a function of time (days after ingestion). Horizontal bar represents the 95% confidence interval to obtain 50% recovery.
Table 1. Estimated parameters (b, d and D50 – with standard error and P value) and lack of fit by the non-linear regression equationa, based on the glyphosate-resistant Lolium multiflorum seeds recovered from cattle faeces as a function of time (days after ingestion)
a Y = d/(1 + exp[b(log(x)–log (D50))].
Germination of the recovered seeds
The LOLMU-R seed germination showed a linear reduction in response to recovery time (days) after ingestion by cattle (lack of fit, P ≥ 0.05). The model indicated a reduction around 12% in germination for each day after passage through the digestive system, compared to the control treatment (no digestive system passage) that showed a germination of 67% (Table 2). However, to the end of assessment period or 4 days after ingestion, the germination potential was 18% (or 27% when compared to the control treatment) (Fig. 2).
Fig. 2. Observed and predicted values for the germination of glyphosate-resistant Lolium multiflorum seeds as a function of recovery time after passage through the digestive system of cattle (compared with the control treatment, no digestive system passage), 14 days after sowing. r2 = 0.85.
Table 2. Estimated parameters (a and b – with standard error and P value) and lack of fit by the linear regression equationa, based on the germination of glyphosate-resistant Lolium multiflorum seeds as a function of recovery time after passage through the digestive system of cattle, 14 days after sowing
a Y = a + b × days after ingestion.
Dose-response curve
The dose-response curve showed a sigmoidal behaviour (lack of fit, P ≥ 0.05), where control increased and SDW decreased in resistant L. multiflorum as a function of larger glyphosate doses. However, the doses that obtained 50% control (C50) and 50% SDW reduction were 7610 and 5057 g a.e./ha, corresponding to 7 and 5x, respectively (Table 3). Thus, LOLMU-R survived the maximum recommended dose (1080 g a.e./ha = 1x) (Fig. 3(a) and (b)).
Fig. 3. Dose-response curve for Lolium multiflorum plants after seed passage through the digestive system of cattle, 28 days after glyphosate application. (a) Observed and predicted population control values (% in relation to the control treatment). Horizontal bar representing the 95% confidence interval to obtain 50% control (C50). (b) Observed and predicted shoot dry weight values (% in relation to the control treatment). Horizontal bar representing the 95% confidence interval to obtain 50% shoot dry weight reduction (GR50).
Table 3. Estimated parameters (b, d, c and D50 – with standard error and P value) and lack of fit by the non-linear regression equationa or b, based on the control and shoot dry weight (both in % relative to the control) of Lolium multiflorum plants after seed passage through the digestive system of cattle, 28 days after glyphosate application
a Y = d/(1 + exp[b(log(x)–log (D50))].
b Y = c + {d – c1 + exp[b(log (x)–log (D50))]}.
Discussion
Endozoochory involves the capture and ingestion of seeds, which are influenced by digestive action as they pass through the digestive system and then eliminated in the faeces (Fazelian et al., Reference Fazelian, Kohyani and Shirmardi2014). Seeds can be destroyed by mastication and rumination (Alvarez et al., Reference Alvarez, Leparmarai, Heller and Becker2016), which may result in more than 80% seed loss (Wang et al., Reference Wang, Lu, Narkes, Ma, Zhang and Wang2017). As such, the recovery rate is directly related to seed characteristics (shape, size and presence or not of an integument), the species that ingests the seeds, diet quality and total retention time in the gastrointestinal tract (Deminicis et al., Reference Deminicis, Vieira, Araújo, Jardim, Pádua and Neto2009; Fazelian et al., Reference Fazelian, Kohyani and Shirmardi2014; Wang et al., Reference Wang, Hu, Zhang and Hou2019).
In general, small round seeds are less likely to be damaged during chewing (Brochet et al., Reference Brochet, Guillemain, Gauthier-Clerc, Fritz and Green2010; Picard et al., Reference Picard, Papaïx, Gosselin, Picot, Bideau and Baltzinger2015). Lolium multiflorum seeds are compact and medium-sized for a forage grass (Fontaneli, Reference Fontaneli2009), justifying the 9.1% recovery rate recorded in the present study and similar to the 12% of Brachiaria decumbens seeds recovered from ruminants (Simão Neto et al., Reference Simão Neto, Jones and Ratcliff1987).
It should be noted that the highest recovery rate was obtained 2 days after ingestion. Viero et al. (Reference Viero, Schaedler, Azevedo, Santos, Scalcon, David and Rosa2018) assessed the recovery of weedy rice (Oryza sativa L.) and barnyardgrass (Echinochloa crus-galli L.) seeds and found that they passed through the gut of the cattle on the first day after ingestion, with zero recovery. However, maximum recovery was obtained 2 days after ingestion, exhibiting sigmoid behaviour and declining to zero (Viero et al., Reference Viero, Schaedler, Azevedo, Santos, Scalcon, David and Rosa2018), similar to the results of the present study.
The processes involved in endozoochory affect the physical and physiological characteristics of seeds, which are exposed to physical processes, microbial activity in the rumen, chemical action in the abomasum, temperature and internal pH (Oliveira et al., Reference Oliveira, Neto, Adelson and Valença2013; Fazelian et al., Reference Fazelian, Kohyani and Shirmardi2014). These processes influence seed viability and, consequently, germination (Blackshaw and Rode, Reference Blackshaw and Rode1991; Gardener et al., Reference Gardener, McIvor and Jansen1993; Wang et al., Reference Wang, Hu, Zhang and Hou2019).
Studies that simulated the mastication, body temperature corporal and digestive fluids of ruminants observed inhibitory effects on the germination of most species assessed (Milotić and Hoffmann, Reference Milotić and Hoffmann2016).
In weedy rice and barnyardgrass, germination declined in seeds that had passed through the gut of cattle (Viero et al., Reference Viero, Schaedler, Azevedo, Santos, Scalcon, David and Rosa2018). This phenomenon may be directly related to the loss of seed viability as a function of retention time in the rumen (Lisboa et al., Reference Lisboa, Medeiros, Azevedo, Patino, Carlotto and Garcia2009). Some seeds tolerate a certain amount of time in the rumen, followed by a rapid decline in viability (Blackshaw and Rode, Reference Blackshaw and Rode1991). These previous results corroborate those found here, whereby the germination of LOLMU-R seed decreased as a function of time after ingestion, that is, recovery time. Ruminants, especially cattle, are considered the main vectors of endozoochory, particularly in the dispersal of species from the family Poaceae (Fazelian et al., Reference Fazelian, Kohyani and Shirmardi2014). In this respect, according to Nakao and Cardoso (Reference Nakao and Cardoso2010), cattle are considered a legitimate dispersal agent of glyphosate-resistant L. multiflorum seeds.
The occurrence of glyphosate-resistant L. multiforum impacts production systems in which the species is used as a winter cover crop (no-till), plant cover (orchards and vineyards) or forage (ICLS) (Peterson et al., Reference Peterson, Collavo, Ovejero, Shivrain and Walsh2018). A previous study confirmed the resistance of this species, demonstrating that glyphosate doses above the recommended maximum were ineffective (Vargas et al., Reference Vargas, Roman, Rizzardi and Silva2004). HR can be confirmed by studying dose-response curves (Burgos et al., Reference Burgos, Tranel, Streibig, Davis, Shaner, Norsworthy and Ritz2013). Thus, the results presented here prove resistance in LOLMU-R, where the control dose is greater than the recommended maximum.
Conclusions
Cattle are a legitimate dispersal agent of glyphosate-resistant L. multiflorum seeds. This may worsen the evolution of resistance, since in these cases, dispersal is far more important as a source of new infestations. Thus, in order to mitigate the evolution of glyphosate resistance, a quarantine period of at least 7 days is recommended before animals move from one infested area to another.
Acknowledgements
The authors thank the National Council for Scientific and Technological Development (CNPq) and Coordination for the Improvement of Higher Education Personnel (CAPES) – Brazil.
Financial support
This research received a grant from the CNPq.
Conflicts of interest
None.
Ethical standards
The research protocol was reviewed and approved by the Animal Ethics Committee of the Department of Agricultural Diagnostics and Research (DDPA).
