How the Management and Environmental Conditions Affect the Weed Vegetation in Canary Grass (Phalaris canariensis L.) Fields

Abstract

Canary grass (Phalaris canariensis L.) is a versatile crop with global significance; it is primarily cultivated for its small elliptical seeds, which are used as bird feed and for human consumption. This crop is adapted to various climates and soils, so it can be grown successfully in Hungary. However, challenges such as weed control, climate change impacts, and soil factors require strategic management for sustained success in canary grass cultivation. Our study investigated the impact of management and environmental (as seasonal and soil) factors on pre-harvest weed vegetation in canary grass fields in Southeast Hungary between 2017 and 2020. In addition to showing the weed vegetation of the canary grass, the aim of our work was to promote more effective weed management of canary grass by revealing correlations between soil, seasonality, and management variables, influencing weed diversity and coverage. Using the analysis of covariance (ANCOVA) and correlation tests, we tested significant variables, providing insights into the complex interactions affecting weed composition. A redundancy analysis (RDA) further unveiled the relationships between explanatory variables and weed species’ composition. The findings offer valuable information for effective weed management strategies in canary grass cultivation. Our comprehensive study on canary grass fields in Southeast Hungary sheds light on significant factors influencing weed composition and abundance. The average weed coverage was 10.8%, with summer annuals and creeping perennials being the most prevalent life forms. Echinochloa crus-galli, Cirsium arvense, Xanthium italicum, and Setaria viridis were among the dominant species. ANCOVAs revealed the impact of soil, management, and seasonal factors on weed cover, species richness, diversity, and yield levels. Soil properties like texture, pH, and nitrogen content showed varying effects on weed parameters. The vintage effect, tillage systems, and farming practices also played crucial roles. The redundancy analysis highlighted the influence of the year, soil sulfur content, and winter preceding crops on weed composition. In conclusion, the herbaceous vegetation in the studied area is dominated by summer germinating and creeping perennial species. Despite slight differences in average coverage and occurrence, a well-defined set of significant species is evident. Multicollinearity among variables suggests limitations to further increase the number of variables that can be included in the analysis. The ANCOVAs showed that the soil, seasonal, and farming variables significantly influence overall weed vegetation and crop yield, with a lesser impact on species richness and diversity. The reduced RDA model highlights the strong influence of the year on species’ composition, emphasizing the inherent factors during canary grass cultivation that are challenging to modify through farming practices.

1. Introduction

The canary grass (Phalaris canariensis L.), a member of the Poaceae family, is a plant similar to millet. However, it is more elongated, 100–120 cm tall, with thin stems and thin long leaves [1], as well as with 2–4 cm long and 1–2 cm wide panicles [2]. Canary grass is grown mainly for its seeds, which are small, elliptical, and similar to flaxseed, with a shiny surface. These seeds are covered with microscopic silicon spines that can cause severe skin irritation and are potentially carcinogenic [3]. It is used in several countries to make pasta, bread, or porridge, but it is mainly grown worldwide as bird feed [4,5,6,7,8,9]. Its most important value-measuring property lies in the chemical composition of its grains, including starch, protein, and amino acids, making it a very promising crop in the right areas of use. It has a higher protein content (22%) than oats (13%) and wheat (16%) [10]. Its proteins contain more lysine and threonine than wheat, and also have a high content of cysteine, tryptophan, and phenylalanine. Overall, the compositions of starch and gluten-like, tryptophan-rich proteins give canary grass unique functional and nutritional properties [3,11]. Due to its protein content, it has been traditionally used to treat hypertension and diabetes [12,13]. It is most widely grown in Canada [2,14,15], as well as in the Czech Republic, Hungary, the Netherlands, the USA, Australia, Thailand, Mexico, and Argentina [16]. It was first recorded as a wild plant in Britain in 1632 [17], from whence it spread across Europe and then the Americas [9,18,19]. The majority of the canary grass produced in Hungary is processed abroad, but due to its favorable nutritional values and its increasing distribution, its domestic use is also increasing [20]. According to the FAO data [21], between 2000 and 2022, the area of canary grass harvested globally varied between 199 and 443 thousand hectares, with production varying between 150 and 377 thousand tons from year to year. Hungary, among other countries, plays an important role in meeting Europe’s canary grass requirements.

The canary grass is a fast-growing, heat-demanding plant with good drought tolerance. It requires temperatures of 6–8 °C at germination and tolerates mild cooler temperatures in late spring, but prolonged colder temperatures below freezing can cause major damage to the already germinated plant. During its short growing season, it requires a total of 160–180 mm of rainfall. It is sensitive to drought during the period of grain formation; otherwise, it tolerates dry periods well [22,23].

Canary grass is a less demanding plant in terms of soil. It can be grown successfully almost anywhere except in sandy areas. It prefers more compacted clay soils, but it is not sensitive to the soil pH. It is potentially suitable in all areas except extreme soils [6,22,24]. It is not sensitive to previous crops, but it is not suggested to grow it after millet or flax, because the control of volunteer plants is not solved. The canary grass is sensitive to herbicides; therefore, the carryover effects of soil residual herbicides used in previous crops should be considered, as well as the presence of common pathogens or pests, e.g., Fusarium spp. or barley leaf beetle (Oulema melanopus) [25].

Generally, the effect of autumn- and spring-sown pre-crops is different on weed composition because different growing seasons allow different weeds to grow and produce seeds. In the case of canary grass, the maize and sunflower as spring-sown row crops are good pre-crops, but the sorghum, with its limited monocotyledonous weed control potential, should be avoided. Similarly, of the fall-sown pre-crops, it is better to choose those in which effective monocotyledonous weed control options are available, e.g., winter canola, winter peas, and vetch [26,27]. However, the most important is the possibility of adequate soil preparation after the harvest of pre-crops [20].

To achieve uniform and rapid emergence, a sufficiently compacted, fine, and smooth seedbed is essential. It is also important to carry out medium–deep cultivation in autumn. Due to its very early sowing time, it is advisable to base soil preparation on autumn work, so that a suitable seedbed can be prepared in spring with minimum disturbance [25]. Sowing times depend largely on the weather in late winter and early spring. It can be sown from the end of February at the earliest, but it is usually sown in early to mid-March. Sowing can start as soon as the soil temperature rises steadily above freezing, and it requires a temperature of 5–6 °C for germination. The recommended sowing depth is 2–3 cm, and the seed rate is 6–8 million sprout ha⁻¹, which means a mass of seed requirement of 70–80 kg ha⁻¹ [28]. Canary grass is a nutrient-demanding crop with a specific nutrient requirement per 1 ton of grain yield: 70–90 kg N, 30–40 kg P, and 50–60 kg K active ingredient. The first year of fertilization with organic fertilizers may not be rewarding but it can be treated with liquid nitrogen fertilizer at temperatures up to 20 °C until the end of tillering [25].

Seed banding by fungicides may be useful in monoculture production or in the case of cereal pre-crops for Fusarium spp., or other common diseases. Pathogens damaging the green parts of the plant (Claviceps purpurea, Spodoptera frugiperda, Schizaphis graminum, Metopolophium dirhodum, Gibberella zeae, and Rhynchosporium secalis) are rare and do not require fungicidal control from an economic point of view [29,30]. Insect pests such as Elateridae, Melolonthidae, Opatrum sabulosum, and Zabrus tenebroides are known to be present in the standing crop, but apart from Oulema melanopus, they rarely cause economic damage. The barley leaf beetle can be controlled in the same way as in cereals [20,31].

The biggest challenge in growing canary grass is keeping the crop weed-free, which can be promoted by proper, judicious site selection and soil preparation. If the germination and initial development of the crop is uniform and rapid, weed control will not be critical due to the closure of crop canopy [20]. The number of registered herbicides in Hungary is quite limited; only herbicides with active ingredients such as MCPA, carfentazone-ethyl, and the combination of 2,4-D + florasulam are authorized for post-emergent chemical control [32]. These active ingredients are only effective against dicotyledonous weeds, so only preventive and non-chemical control methods are available against monocotyledonous weeds [33].

In Hungary, canary grass ripens in mid-July due to its short vegetation. For safe storage, a moisture content of 12–14% is required. Harvesting losses can be caused by fallen stock, overripe grains, or incorrectly adjusted harvesters. Harvesting and threshing are facilitated if the bushes are exposed to rain before harvesting at full maturity [20].

The different physical and chemical properties of the soil have effects on each other [34]. Thus, soil properties influence the emergence, growth, and competition of weeds in a very complex way. One of these important properties is the transmittance of light to the soil, which has a decisive influence on the germination of weed seeds. The transmittance of light depends on the size of the soil particles, the moisture content, the color of the particles, and the presence of organic matter. Various studies have observed that, as the particle size of the soil decreases, the light permeability of the soil decreases, which in turn changes the rate of germination of weed seeds [35,36]. Even a slight change in soil chemistry has a major impact on the emergence and growth of vegetation, including weeds. Generally, plants do not prefer more acidic soils, but some weeds may have an advantage over the crop in such conditions. The competitiveness of the crop can be improved by the addition of lime [35] and by organic and mineral fertilization [36].

Fertilization, as a long-term factor influencing soil properties, plays a crucial role in the development of weed flora and weed population, as well as in competition with crop plants [37]. For example, Sinapis arvensis was a major problem in acid soils, but it has now declined in importance, partly due to liming and fertilizer use [38].

Soil salinization is one of the most critical soil factors affecting yields. Salinization affects 20% of arable farmland, including 33% of irrigated farmland. Agricultural crops show a variety of adverse responses to salt stress, and salinity adversely affects several different soil properties [39]. Salinity also determines morphological, physiological, and biochemical processes, including seed germination, plant growth, water, and nutrient uptake for both crops and weeds. Therefore, salinity plays a crucial role in all aspects of plant development, from germination, through the vegetative growth stages, to reproduction [40].

The vegetation can adapt to the variability of weather factors, but climate change is of paramount importance for both crops and weeds in agriculture [41]. Both the amount and distribution of rainfall [42], as well as the number of hot days [43], contributed to the shift in the climatic zones in Hungary [44]. Sudden changes in weather stress our crops, making them more susceptible and less competitive against weeds [45]. Air temperature plays a crucial role in the geographic distribution of weeds, as it alters weed proliferation and competitive behavior in the plant population [46,47]. In connection to this factor, further northward expansion of several weeds, such as Amaranthus retroflexus, Setaria spp., Digitaria sanguinalis, and Sorghum halepense, has been observed in the last decades [48,49]. Milder and wetter winters tend to increase the overwintering ability of annual weeds, but thermophilic annuals also grow with greater intensity where the summer is longer and warmer. Climate change will lead to the spread of field weeds such as Datura stramonium, which needs high temperatures to grow [45,50,51]. Rainfall patterns and the increasing frequency of droughts associated with climate change are also altering the occurrence and spread of weeds, thereby impacting crop production. Extreme weather conditions and drought-affected agricultural areas are expected to become more prevalent in the near future [52].

In addition to the climate, crop rotation and tillage together influence the emergence of weeds [53]. In general, reduced tillage systems, such as no-tillage, stratify the soil’s weed seed bank closer to the soil surface, whereas intensive tillage provides a uniform distribution of weed seeds down to the depth of cultivation [54]. Weed species respond quite differently to different tillage regimes, with some species increasing under reduced tillage and others decreasing [55]. The no tillage method significantly encouraged the proliferation of certain broadleaf weeds, notably Capsella bursa-pastoris. In contrast, Viola arvensis occurred more frequently in conventional tillage plots. Echinochloa crus-galli was identified in all types of tillage systems, with a notably elevated occurrence observed in the minimum tillage treatment, accounting for 22% of the total weed population [56]. Crop rotation, especially alternating spring row crops and dense cereals, can significantly reduce weed infestation, but this effect may vary from place to place, depending on the environmental conditions [53].

Reducing the number of tillage operations and avoiding rotation will help to reduce the presence of some weeds by reducing large-seed, deep-germinating weed species, e.g., Abutilon theophrasti, Xanthium spp., and Datura stramonium. In contrast, the incidence of small-seeded and broad-leaved weeds such as Chenopodium album, Amaranthus retroflexus, and Ambrosia artemisiifolia, and annual grasses e.g., Echinochloa crus-galli and Setaria species, may increase [57]. These tillage regimes can also promote the establishment and germination of perennial weed species such as Epilobium ciliatum, Poa trivialis, Cirsium arvense, Taraxacum officinale, Equisetum arvense, and Elymus repens [58].

Basically, arable weed species have different nutrient requirements and are adapted to different nutrient supply levels [59]. Thus, some weed species can also serve as indicators of soil nutrient status. Rumex acetosella, Spergula arvensis, and Scleranthus annuus tolerate calcium deficiency and are often referred to as acidophilic weeds, while Chenopodium album, Stellaria media, and Tripleurospermum inodorum prefer calcareous soils and are more able to use fertilizers than cultivated crops. The latter are also known as nitrophilic weeds [60,61,62]. The effects of many of the factors highlighted in the introduction can be generalized to other crops, but the specificities of canary grass cultivation in terms of management factors are on a different scale than for other crops.

2. Materials and Methods

2.1. Data Collection

The aim of our study was to examine the weed composition of the surveyed region and to evaluate the effect of environmental, management, and soil factors on the pre-harvest weed vegetation of canary grass fields. A total of 31 canary grass fields were examined between 2017 and 2020 in Southeast Hungary (in the region of Gyomaendrőd city) when weed vegetation was sampled between late-May and mid-July, depending on the maturity of the canary grass.

The study area is characterized by alluvial and meadow soils with a too-high water-retaining capacity and short timeframe of proper tillage, resulting in high weed populations in general. Because of their relatively high homogeneity, the fields were assessed in eight randomly placed 1 × 1 m quadrats in each field [63]. Each weed species was surveyed based on coverage, i.e., the percentage of ground covered by aboveground plant parts [64]. The analyses were conducted using the field-level average for all species. On all surveyed fields, soil and management variables and factors affecting the environment (termed ‘environmental variables’) were also recorded by soil analyses, other observations, or farmer interviews (

Table 1. Units and ranges of continuous variables and values of categorical variables.

Variable (Unit)	Range/Recorded or Calculated Values
Soil factors	–
Soil texture (KArany) B	53–66
Soil pH (KCl) B	4.15–7.33
Soil properties (m/m %)
Salt A	0.066–0.7
Humus B	1.45–6.64
CaCO3 B	0.04–1.8
Soil properties (mg kg−1)
N B	1.4–29.1
P2O5 B	31–2020
K2O B	228–880
Na B	26–252
Mg	355–1420
S	15–42
Cu	3.1–18.1
Mn	82–475
Zn	0.8–8.53
Seasonality factors
Year	2017–2020
Sampling date (Julian day) A	149–202
Management factors
Farming B	Conventional, organic
Preceding crops a
Wintering crops b	0–0.7
Spring row crops c	0.1–0.7
Spring dense crops A d	0–0.7
Tillage system B	Ploughing, loosening, shallow tillage
Tillage depth (cm) A	15–35
Sowing date (Julian day) B	58–110
Nr. of mechanical weed control applications A	0–1
Nr. of herbicide applications A	0–1
Amount of Nitrogen fertilizer (kg a.i. ha−1) A	0–54

^A Variables not included into the analysis due to multicollinearity. ^B Variables dropped during the backward selection process. ^a Calculated by last three preceding crops: (pre-crop 1 × 0.6) + (pre-crop 2 × 0.3) + (pre-crop 3 × 0.1). ^b Winter wheat, spelt wheat, winter barley, and winter oilseed rape. ^c Maize, sunflower, and oilseed pumpkin. ^d Spring pea and canary grass.

).

Tillage practices were based on either shallow cultivation by disc harrow to a depth of 15 cm, ploughing to a depth of 30 cm, or loosening via a shank ripper to a depth of 35 cm, depending on the field. Crop rotation in a three-year period before data collection was recorded, and preceding crops were classified into three groups, as winter crops (winter wheat and spelt wheat), spring row crops (spring sowing crops with a row spacing of 45 cm or more; maize, sunflower, and oilseed pumpkin) and spring dense crops (canary grass and spring oat). The relative frequency (between 0 and 1) of these groups was calculated/weighted by the years of production: 0.6 for previous crop #1, 0.3 for previous crop #2, and 0.1 for previous crop #3. The analysis included the relative frequency of these groups. Organic fields were certified by Biokontroll Hungária Nonprofit Ltd. (certification number: HU-ÖKO-01). The cultivar, amount of seeds, phosphorus and potassium fertilizer rate, and geographical location (latitude, longitude, and ALS) variables were also recorded but were excluded from the analyses, as the cultivar and amount of seeds were the same for all fields (Kisvárdai 41 and 80 kg ha⁻¹), no application of phosphorus and potassium fertilizer was performed on neither conventional nor organic fields, and the fields were located very close to each other (between N 46.93328° and N 47.01637°, E 20.76288° and E 20.91864°, approximately 100 km square) and were on very similar altitude (78–81 m ASL).

2.2. Statistical Analyses

The data preparation and the analysis can be divided into four parts. In the first step prior to the analysis, cover values of each species were aggregated within the eight plots from each field to calculate the average weed composition of the individual fields. These field averages were then examined through statistical analyses. To demonstrate the overall importance of each species, both the average cover value and frequency (at the field level) were calculated across all examined fields. The intercorrelations between soil, seasonality, and management factors, as explanatory variables, were assessed by calculating generalized variance inflation factors (GVIF). During this process, soil salinity (highly correlated with soil reaction), sampling date (highly correlated with sowing date), ‘tillage depth’ (highly correlated with tillage method), ‘number of mechanical weed control applications’, ‘number of herbicide applications’, nitrogen fertilizer rates (N) (all highly correlated with a farming system where mechanical weed control and the abandonment of fertilizer use characterized organic fields, while fertilizer and herbicide [2,4-D and florasulam a.i.] use dominated in conventional fields), and ‘spring dense preceding crops’ (highly correlated with other preceding crops) were excluded from the analysis (Table 1). The rest of the variables showed only a limited collinearity, where the highest value of GVIF (adjusted by degree of freedom) was 20.51 [65].

In the second step, the Shannon diversity index [66] was calculated based on relative coverage of weed species of each field according to the following equation:

$$H^{\prime }=-\Sigma \sum _{i=1}^R{p}_i\ln p_i$$

where R means number of species, and p_i is proportion of coverage of individuals belonging to ith species in the sample.

In the third step total weed coverage, species richness, calculated diversity indexes, and yields were analyzed together with the soil, seasonality, and management variables—not eliminated in the first step—by analysis of covariance (ANCOVA) [67] and by Pearson correlation (numeric variables) [68] and Tukey test (factor variables) [69] in the case of significant (p < 0.05) variables of ANCOVAs.

In the fourth step, average cover values of each species and each field were subjected to Hellinger transformation [70] and examined in a redundancy analysis (RDA), together with the soil, seasonality, and management variables—not eliminated in the first step—to describe the effect of explanatory variables on the composition of the weeds. The number of explanatory variables was decreased by stepwise backward selection using a p < 0.05 threshold for type I error, which resulted in a minimal adequate model which contained eight independent variables (Table 1). This analysis (RDA) also included the estimation of the gross and net effects of each explanatory variable of the reduced model [71]. Based on the results, a common rank of ‘importance’ was settled among all explanatory variables according to the adjusted R² values of the net effects of the partial RDA models. To show the relations between significant factors and weed species, for each partial RDA model, 10 species with the highest explained variation in the constrained axis were identified. The presented statistical analysis was performed in the R Environment (R Development Core Team, version 4.3.2), using the Vegan add-on package (vegan 2.5-1).

3. Results

3.1. Weed Composition

On the examined fields, the total weed coverage was 10.8% on average. According to Figure 1, the summer annuals (SAs) were the most abundant, with the total cover value of 8.4%, and the presence of creeping perennial (CP) species was also significant, with 2.0% in total. The most abundant species were Echinochloa crus-galli (4.2%, SA), Cirsium arvense (1.6%, CP), Xanthium italicum (1.6%, SA), and Setaria viridis (1.0%, SA). Additionally, species such as Hibiscus trionum (0.8%, SA), Chenopodium album (0.4%, SA), Convolvulus arvensis (0.3%, CP), Ambrosia artemisiifolia (0.2%, SA), Persicaria lapathifolia (0.2%, SA), and Avena fatua (0.1%, SA) had coverage values below 1% but were still significant (Figure 1).

Figure 2 shows that, similar to the cover values, the most frequently occurring weeds belong to the summer annual (SA) and creeping perennial (CP) life-form groups. Six species were present with a frequency exceeding 50%, such as Hibiscus trionum (83.9%, SA), Cirsium arvense (80.6%, CP), Xanthium italicum (77.4%, CP), Echinochloa crus-galli (74.2%, SA), Setaria viridis (71.0%, SA), and Chenopodium album (58.1%, SA). Looking at the cover and frequency data together, we can see that some species (e.g., Datura stramonium and Amaranthus retroflexus) were very frequent in the study areas but showed low cover values. In contrast, Echinochloa crus-galli, Cirsium arvense, and Xanthium italicum, in addition to their high frequency, also showed high cover values. On the other hand, Hibiscus trionum was the most frequently occurring weed, but it covered less (0.8%).

3.2. Effect of Variables on Total Weed Coverage, Species Richness, Diversity, and Canary Grass Yield

The analyses of covariance (ANCOVAs) confirmed that several soil and management factors, as well as management seasons (year), had a significant effect on total weed cover, species richness, weed diversity, and yield levels (

Table 2. Results of ANCOVAs, Pearson correlation (in case of significant numeric variables), and Tukey test (in case of significant factor variables) on total weed coverages, species richness, calculated Shannon diversity indexes, and yields effected by soil, management, and seasonality variables.

Factor	Total Weed Coverage [%]	Species Richness	Shannon Diversity	Yield [kg ha−1]
	p-value of ANCOVAs (Pearson correlations)/ [mean and sign. classes of Tukey tests]
Soil texture	0.007 (−0.11)	<0.001 (−0.49)	0.032 (−0.20)	<0.001 (−0.18)
Soil reaction	0.002 (−0.14)	<0.001 (+0.58)	0.003 (+0.34)	<0.001 (−0.14)
Soil properties
Humus	<0.001 (−0.08)	ns	0.018 (+0.01)	<0.001 (+0.28)
N	<0.001 (+0.27)	0.004 (−0.68)	<0.001 (−0.63)	<0.001 (−0.42)
P2O5	<0.001 (−0.11)	ns	ns	ns
K2O	ns	ns	0.018 (−0.14)	<0.001 (−0.09)
CaCO3	0.009 (−0.16)	ns	Ns	<0.001 (−0.15)
Na	<0.001 (−0.11)	ns	ns	<0.001 (−0.07)
Mg	<0.001 (−0.18)	ns	0.036 (−0.02)	<0.001 (+0.09)
S	0.001 (+0.03)	ns	ns	ns
Cu	<0.001 (+0.06)	ns	0.003 (−0.12)	0.001 (+0.06)
Mn	<0.001 (−0.10)	ns	ns	ns
Zn	ns	0.004 (−0.14)	ns	<0.001 (+0.02)
Year	<0.001 [2017—6.1 a 2018—10.9 b 2019—2.4 a 2020—23.8 b]	ns	ns	<0.001 [2017—1861 b 2018—1348 a 2019—1822 b 2020—1304 a]
Farming     conv (conventional)     org (organic)	<0.001 [conv.—4.53 a org.—17.45 b]	ns	0.031 [conv.—1.68 b org.—1.24 a]	0.007 [conv.—1565 a org.—1620 b]
Preceding crops a
Wintering crops b	0.016 (+0.34)	ns	0.036 (−0.29)	0.003 (−0.35)
Spring row crops c	ns	ns	ns	<0.001 (−0.27)
Tillage system     MT (minimum tillage)     LO (loosening)     PL (ploughing)	0.003 [MT—0.9 a LO—11.7 b PL—10.6 b]	0.017 [MT—8.5 b LO—7.8 a PL—8.0 ab]	0.004 [MT—1.8 b LO—1.4 a PL—1.6 ab]	<0.001 [MT—2037 b LO—1453 a PL—1827 b]
Sowing date	0.022 (+0.21)	ns	ns	ns

a Calculated by last three preceding crops: (pre-crop 1 × 0.6) + (pre-crop 2 × 0.3) + (pre-crop 3 × 0.1). b Winter wheat and spelt wheat. c Maize, sunflower, and oilseed pumpkin. ns not significant.

).

In the case of soil texture, the highest correlation was observed for the species richness (−0.49), but lower values were also observed for cover, diversity, and yield on more compact soils. The study showed that the higher pH value increased the species richness (corr: +0.58), and weed diversity (+0.34) had a negative effect on the yield level (−0.14). Among the macronutrients, the soil nitrogen content had the greatest effect on weediness, with a positive effect on weed cover (+0.27) but negative effects on species richness (−0.68) and weed diversity (−0.63). The vintage effect had the greatest impact on weed coverage. Compared to the very high weed cover values of 10.9% and 23.8% in the rainy years of 2018 and 2020, the total weed cover was 6.1% and 2.4% in the drier years of 2017 and 2019, respectively. The vintage effect is also clearly detectable on the yield level. In weedier years, the average yield was about 39% lower than in the less weedy years. The weed cover values of fields managed under conventional production systems were more than double (14.9%) that of organic areas (6.4%). The difference in yields between fields managed in different ways was also significant, but this difference was less pronounced. Neither the species richness nor the diversity had changed as a result of used farming systems. Of the tillage systems, minimum tillage was the most effective in reducing weed encroachment, but resulted in the highest weed species richness and diversity, as well as the highest yield. The sowing date of the crop had a significant influence on the subsequent weed-infestation values: the earlier the sowing date, the less weed infestation was observed (corr: +0.21).

3.3. Effect of Variables on Weed Composition

According to the redundancy analysis (RDA), the weed composition was mainly affected by the year of data record (21.06%), soil S content (5.96%), and winter preceding crop (5.93%), but in total, eight factors were significant in our model. The total explained variation in order of all included soils, year, and preceding cop were 21.62, 21.06, and 9.26% % (

Table 3. Effects of the soil, seasonality, and management variables on weed composition.

Factors	Df	Gross Effect		Net Effect
		Explained Variation (%)	R2adj	Explained Variation (%)	R2adj	F	p-Value
Soil S content	1	13.43	0.104	5.96	0.060	3.387	0.030
Soil Cu content	1	6.38	0.031	4.86	0.044	2.761	0.002
Soil Mg content	1	4.94	0.017	3.52	0.025	1.996	0.032
Soil Mn content	1	6.98	0.038	3.87	0.030	2.195	0.026
Soil Zn content	1	5.02	0.017	3.41	0.024	1.937	0.045
Year	3	23.70	0.152	21.06	0.206	3.987	0.001
Winter preceding crop	1	6.88	0.037	5.93	0.060	3.370	0.003
Spring row preceding crop	1	4.73	0.014	3.33	0.022	1.893	0.042

).

The sulfur, as a nutrient, affected the growth of weed species differently. The high soil sulfur content increased the spread of Xanthium italicum and Convolvulus arvensis, but it strongly inhibited the growth of Setaria viridis and Echinochloa crus-galli in the experimental area (

Table 4. Names, score values, and fit of species, giving the highest fit along the first constrained axis in the partial redundancy analysis (pRDA) models of Soil variables in open-field canary grass experiment (Hungary, 2017–2020).

Species	Ax 1 Score	Fit	Species	Ax 1 Score	Fit
Soil S content (+ high; − low)			Soil Mn content (+ high; − low)
Xanthium strumarium	0.321	0.275	Cirsium arvense	0.227	0.095
Convolvulus arvensis	0.112	0.057	Hibiscus trionum	0.098	0.033
Ambrosia artemisiifolia	0.104	0.096	Sinapis arvensis	0.086	0.144
Persicaria lapathifolia	0.101	0.092	Abutilon theophrasti	0.076	0.142
Calystegia sepium	0.094	0.084	Datura stramonium	0.050	0.057
Amaranthus retroflexus	0.082	0.069	Avena fatua	0.037	0.025
Sonchus asper	0.059	0.101	Sonchus asper	−0.067	0.130
Sinapis arvensis	−0.050	0.048	Ambrosia artemisiifolia	−0.107	0.103
Setaria viridis	−0.166	0.077	Calystegia sepium	−0.109	0.112
Echinochloa crus-galli	−0.185	0.063	Xanthium strumarium	−0.159	0.067
Soil Cu content (+ high; − low)			Soil Zn content (+ high; − low)
Xanthium strumarium	0.146	0.057	Xanthium strumarium	0.168	0.075
Triticum spelta	0.043	0.134	Convolvulus arvensis	0.147	0.098
Sonchus asper	0.034	0.032	Sonchus asper	0.060	0.103
Avena fatua	−0.028	0.015	Ambrosia artemisiifolia	0.045	0.018
Datura stramonium	−0.045	0.046	Calystegia sepium	0.042	0.017
Calystegia sepium	−0.049	0.023	Triticum spelta	0.029	0.062
Abutilon theophrasti	−0.071	0.125	Sinapis arvensis	−0.045	0.040
Amaranthus retroflexus	−0.082	0.068	Abutilon theophrasti	−0.047	0.056
Hibiscus trionum	−0.241	0.203	Helianthus annuus	−0.118	0.040
Helianthus annuus	−0.264	0.199	Hibiscus trionum	−0.194	0.131
Soil Mg content (+ high; − low)
Hibiscus trionum	0.241	0.202
Helianthus annuus	0.121	0.042
Chenopodium album	0.077	0.019
Avena fatua	0.059	0.065
Setaria viridis	0.055	0.008
Calystegia sepium	0.049	0.023
Sinapis arvensis	0.024	0.011
Triticum spelta	−0.029	0.060
Xanthium strumarium	−0.068	0.012
Echinochloa crus-galli	−0.187	0.065

).

The coverage of Cirsium arvense and Hibiscus trionum could increase to the greatest extent by high manganese content, but other species (e.g., Calystegia sepium, Xanthium italicum) preferred the low manganese content of soils more.

The weed composition was also well differentiated by copper content. Fields with higher copper content were more covered by Xanthium italicum and Triticum spelta, but Hibiscus trionum and Helianthus annuus had a higher coverage in the case of lower copper content.

The higher zinc content caused increased cover of Xanthium italicum and Convolvulus arvensis, but the cover of the Helianthus annuus and the Hibiscus trionum was lower.

In the case of soil magnesium content, there were significant differences in weed coverage as well. The Hibiscus trionum and Helianthus annuus showed higher cover values, in contrast to Xanthium italicum and Echinochloa crus-galli, which had less significant coverage there.

The order of weed dominance varied between years over the survey period. Most important species were Trifolium repens and Calystegia sepium in 2017; Persicaria lapathifolia and Chenopodium album in 2018; Cirsium arvense, Hibiscus trionum, and Helianthus annuus in 2019; and Echinochloa crus-galli and Avena fatua in 2020 (

Table 5. Names, score values and fit of species giving the highest fit along the first constrained axis in the partial redundancy analysis (pRDA) models of Year variable in open-field canary grass experiment (Hungary, 2017–2020).

Species	Ax 1 Score	Fit	Species	Ax 1 Score	Fit
2017 (+ high; − low)			2018 (+ high; − low)
Trifolium repens	0.208	0.365	Persicaria lapathifolia	0.186	0.317
Calystegia sepium	0.136	0.174	Chenopodium album	0.155	0.079
Sinapis arvensis	0.097	0.186	Echinochloa crus-galli	0.147	0.040
Sonchus asper	0.073	0.153	Setaria viridis	0.130	0.047
Datura stramonium	−0.054	0.065	Amaranthus retroflexus	0.053	0.029
Avena fatua	−0.075	0.105	Abutilon theophrasti	0.045	0.051
Persicaria lapathifolia	−0.080	0.058	Datura stramonium	−0.045	0.045
Amaranthus retroflexus	−0.096	0.094	Trifolium repens	−0.080	0.054
Chenopodium album	−0.137	0.062	Helianthus annuus	−0.183	0.095
Echinochloa crus-galli	−0.450	0.376	Cirsium arvense	−0.273	0.138
2019 (+ high; − low)			2020 (+ high; − low)
Cirsium arvense	0.256	0.121	Echinochloa crus-galli	0.438	0.357
Hibiscus trionum	0.235	0.191	Avena fatua	0.149	0.421
Helianthus annuus	0.202	0.117	Datura stramonium	0.063	0.089
Chenopodium album	0.136	0.061	Triticum spelta	−0.029	0.062
Amaranthus retroflexus	0.134	0.182	Trifolium repens	−0.087	0.064
Triticum spelta	0.038	0.106	Amaranthus retroflexus	−0.089	0.080
Persicaria lapathifolia	−0.055	0.027	Hibiscus trionum	−0.121	0.051
Calystegia sepium	−0.058	0.032	Chenopodium album	−0.146	0.070
Echinochloa crus-galli	−0.129	0.031	Helianthus annuus	−0.147	0.062
Setaria viridis	−0.174	0.084	Cirsium arvense	−0.156	0.045

).

Preceding crops resulted in having both differences and similarities in weed dominance. The most important species in both cases (after wintering crops and spring row crops) were Cirsium arvense and Hibiscus trionum. At the same time, winter preceding crops increased the presence of Echinochloa crus-galli to the second highest extent, but spring row preceding crops strongly decreased the importance of this species (

Table 6. Names, score values, and fit of species, giving the highest fit along the first constrained axis in the partial redundancy analysis (pRDA) models of management variables in open-field canary grass experiment (Hungary, 2017–2020).

Species	Ax 1 Score	Fit	Species	Ax 1 Score	Fit
Wintering preceding crop (+ high; − low)			Spring row preceding crop (+ high; − low)
Cirsium arvense	0.265	0.130	Cirsium arvense	0.157	0.045
Echinochloa crus-galli	0.120	0.027	Hibiscus trionum	0.144	0.072
Hibiscus trionum	0.109	0.041	Convolvulus arvensis	0.108	0.053
Sinapis arvensis	0.072	0.101	Xanthium strumarium	0.078	0.016
Triticum spelta	0.023	0.040	Ambrosia artemisiifolia	0.073	0.048
Calystegia sepium	−0.087	0.072	Avena fatua	0.073	0.100
Helianthus annuus	−0.089	0.023	Persicaria lapathifolia	0.029	0.008
Ambrosia artemisiifolia	−0.125	0.138	Triticum spelta	0.023	0.039
Xanthium strumarium	−0.126	0.042	Setaria viridis	−0.070	0.014
Setaria viridis	−0.265	0.197	Echinochloa crus-galli	−0.214	0.085

).

The overall model also confirms relationships between several variables. The occurrence of Xanthium italicum and Persicaria lapathifolia is often associated with high soil sulfur content. Conversely, Setaria viridis occurs with higher zinc concentrations. The crop rotation also resulted in different weed vegetation. The presence of Echinochloa crus-galli is highly correlated with autumn pre-crops, while Cirsium arvense was more problematic after spring row-crops. Both Hibiscus trionum and Helianthus annuus (volunteers) were most affected by the seasonal effect, with the former being the most prevalent weed in 2019 and the latter in 2017 (Figure 3).

4. Discussion

The studied area is characterized by bound soils that are unfavorable for crop production, with a too-high water-retaining capacity and unbalanced nutrient supply. The average rainfall in the region is below the national average, and the distribution of rainfall over the years has not been favorable during the period under study. All of this limited the initial development of the canary grass, which opened the way to easier and more extensive weed growth and resulted in weed infestation above 10%, in spite of the weed-control applied.

Comparing our results with the weed cover of maize fields in the Sixth National Arable Weed Survey in Hungary [72], we see that there are similarities in many weed species and weed cover values. Eight of the top twenty weed species are identical, with Hibiscus trionum and Cirsium arvense showing the least difference in cover value, but Echinochloa crus-galli, Setaria spp., Chenopodium album, Convolvulus arvensis, Ambrosia artemisiifolia, and Persicaria lapathifolia were also found among the weed species with the highest cover values in both studies.

There are also differences between the results of the two studies. Whereas Sorghum halepense, Datura stramonium, Amaranthus retroflexus, Helianthus annuus, Abutilon theophrasti, and Chenopodium hybridum all showed near or above 1% cover in the Sixth National Arable Weed Survey, they were absent or present in low abundance in the present study. In contrast, our results showed that both Xanthium italicum and Avena fatua were common species, and Cirsium arvense gave significantly higher cover values.

The similarities in the two studies can be explained by the fact that both crops are spring sown, and similar initial weed infestations can develop [73]; however, different durations of growing seasons and different weed control options—mechanical and chemical—have had different effects on the weed vegetation [74].

Comparing these results with those of our previous study, which was carried out on pea fields but in the same area as the current one, we can see that there is a high degree of similarity in terms of weed species’ cover and frequency. Most of the top ten weeds with the highest cover in peas and canary grass are identical, and the cover values of Cirsium arvense and Xanthium italicum differ by less than 0.1%. The exceptions are only Chenopodium polyspermum and Tripleurospermum inodorum, which characterized the pea fields, but not the canary grass [75]. This suggests that the species most characteristic of the landscape will be of greater importance irrespective of the type of crop [76].

The effect of crop is reflected in the higher weed abundance values in pea compared to canary grass, but the frequency of individual weed species was more uniformly higher in the canary grass fields, where Hibiscus trionum, Cirsium arvense, Xanthium italicum, Echinochloa crus-galli, and Setaria viridis, also had a frequency above 70%, compared to the results from the pea fields, where only Echinochloa crus-galli was above 70%. In addition to the similar crop density, the differences in weed cover between the two crops are also based on different herbicide use potential [75].

Analyzing the findings in more detail, it can be inferred that the year when the survey was conducted (21.06%) exerted the most significant impact on weed infestation. Regarding the survey periods, the majority of variations can be attributed to differences in rainfall [77]. In the surveyed region, the average rainfall for the period between January and June over the aggregated 20-year span (2001–2020) amounted to 277 mm. This is in contrast to the respective figures of 184, 330, 328, and 293 mm observed between 2017 and 2020 [78]. Notably, the years with the highest and lowest rainfall exhibited the most significant differences.

In the case of conventional farming, the chemical weed management of canary grass is possible by use of herbicides [32], which can be used effectively against most dicotyledonous weed species. In contrast, in organic farms, post-emergent weed control can only be carried out by mechanical methods, such as by use of fixed or flexible tine weed harrow or finger weeder. In the study, the average weed cover of organic areas was more than three times higher (17.43%) than that of conventional fields (4.52%). Higher cover values in organic farming are consistent with the results of Dorner et al. [79] and Nam et al. [80], who found that the weed cover in organic areas was 25.18% compared to 5.42% in conventional areas. Weed infestation values in organic areas appear high but are tolerable with the cultivation of a crop with a high germination rate [81], as our results show, too, that the difference in yield between the two production systems was less than 3%, and the weed flora was not significantly changed by farming systems according to RDA model.

The influence of pre-seeding crops on weed flora was consistent in our study with many previous studies [82,83,84], as 9.26% of the total weed flora variability was accounted for by the two classes of pre-seeding crops included in the RDA model. It is important to point out that, of the management variables, only this factor had a significant effect on weed composition, so it is important to take this variable into account when predicting weediness [85,86].

Similar to the results of Pinke et al. [87], our study also found that many soil properties (magnesium, sulfur, copper, manganese and zinc content; Table 3) had a significant effect on the composition of weed species. On the contrary, the soil texture, pH, nitrogen, and calcium content had a significant effect only in their study with a wider geographical distribution. This suggests that there may not be significant differences in some variables at small geographical scales [71,83], making their impact in such cases difficult or impossible to assess.

Weeds, as indicator species, could highlight the presence of certain nutrients. Xanthium strumarium was more common in soils with higher sulfur, copper, and zinc concentrations than in soils rich in manganese and magnesium; however, the accumulation of these microelements is not outstanding in this species [88]. The occurrence of Cirsium arvense was significant, with high manganese content. This result contradicts the findings of Alda et al. [89], as manganese accumulation was higher in all the crops (maize, sugar beet, and alfalfa) and the weed (Elymus repens) they studied, and this manganese accumulation was higher than that of C. arvense. The presence of Hibiscus trionum was influenced by the presence of several micronutrients, as it favored high concentrations of magnesium—similarly, a high concentration was detected in its foliage by Steyn et al. [90]—and manganese in contrast to copper and zinc content. Additionally, macronutrients also increased the density of this species [91]. Ambrosia artemisiifolia prefers soils enriched with sulfur and zinc, an these elements can highly accumulate in this plant [92], and it is less common in soils with a high manganese content.

Although the effect of soil texture was not significant on weed composition in our study, it is a major determinant of other soil properties, including soil fertility and the development of weed species [93], so it may have had an indirect effect.

Similarly, the soil pH had no significant effect in our RDA model, but the amounts of essential elements often vary depending on the pH of the soil, and there are correlations between the uptakes of different nutrients [94]. In addition, the prevalence of weeds is most influenced by the soil structure and pH, so it is necessary to interpret these parameters and nutrient preferences together when investigating the effects of soil parameters on weed flora [26,95,96].

5. Conclusions

The weed flora of canary grass and its relationship with soil, environmental, and management factors are less researched topics due to canary grass’s small area under cultivation. However, because of the similar growing parameters and conditions to cereals, the results of the present study are comparable to those described for cereals.

Weed composition on agricultural habitats is influenced by a number of factors. Our study investigated the effects of 26 variables on weeds. The simultaneous monitoring of these variables was essential, as they can have positive or negative effects on each other.

During the data processing, several variables were excluded from further analysis due to their collinearity. Herbicides, fertilizers, and mechanical weed control were related to farming systems; the tillage system determined the depth of the tillage; and the frequencies of different preceding crop groups also interacted with each other. These interactions suggest that, despite the most basic need, it is not recommended to increase the number of study variables affecting weed control indefinitely, but rather to select the variables that have the most impact.

There were several study variables that affected total weed cover but had no significant effect on weed composition. Weed control, for example, is a key issue in both conventional and organic farming, because, without effective weed control, profitable production cannot be achieved. Although the weed control abilities of farming systems differ greatly, there are no major differences in the species composition of the weed flora, but significant differences in weed cover values were observed.

The results of the redundancy analysis showed that the year of the study had the largest effect on the weed flora, which could be explained by the different rainfall conditions, as changes in rainfall were consistent with changes in weed abundance. Other important descriptive factors were the nutrient (S, Cu, Mg, Mn, and Zn) content of the soil and the preceding crops, which also had a clear influence on the composition of the weed vegetation.

Despite the clear results of the statistical analyses, it is important to note that the effect of these factors can only be applied to the regions under study or to regions with similar soil, weather, and farming parameters. Regions already affected by different factors may have different weed flora.