Effect of long-term fertilisation on the weed community of a winter wheat field

Effects of fertilisation and other management techniques on a weed community were evaluated during wheat growth in a rice-wheat cropping system. Fertiliser treatments were C0 (C means chemical, C0 means zero chemical fertiliser.), CN (N fertiliser), CNK (N plus K fertiliser), CNPK (N plus P and K fertiliser), CNP (N plus P fertiliser), and CPK (P plus K fertiliser). Weed density, biomass, and bio-diversity were determined. Redundancy analysis (RDA) was used to investigate the relationship between fertiliser management, weed species, and weed density. The overall weed densities in the C0 and CPK treatments were the greatest during wheat seeding and ripening periods and were significantly greater than densities in the other treatments. N, P and organic matter in soil were highly correlated with weed species and density, whereas K in soil was not significantly correlated with weed species and weed density. N fertiliser significantly reduced weed density. Balanced fertilisation maintained weed species richness and resulting in a high yield of wheat. CNPK application reduced weed damage and improved the productivity and stability of the farmland ecosystem.


Results
Differences in soil nutrients and wheat yields. There were six treatments: no fertilisation (C0), N fertiliser only (CN), N fertiliser and P fertiliser (CNP), N fertiliser and K fertiliser (CNK), P fertiliser and K fertiliser (CPK), and all three fertilisers (CNPK). The experiment was initiated in 1980. The initial values for the soil in the experimental farmland were a pH of 6.8 and a bulk density of 1.26 g·cm −3 . The basic physical and chemical indicators of the topsoil at a depth of 0-15 cm were as follows: organic matter 24.20 g·kg −1 , total N 1.43 g·kg −1 , total P 428.00 mg·kg −1 , available P 8.40 mg·kg −1 (Total P includes organic phosphorus and inorganic phosphorus. While Olsen-p is the result of soil available P determination using sodium bicarbonate extraction method (0.5 mol/L NaHCO 3 , i.e. Olsen method), and Olsen-p can be directly absorbed by crops), available K 127.00 mg·kg −1 .
There were significant differences in farmland soil nutrients and wheat yields among the different fertiliser treatments (Table 1). For the CN and CNK treatments without P fertilisation, the total P content in the soil was equal to the initial P content. The total P and Olsen P in the soil significantly increased in all of the treatments with P fertilisation. The available K content was reduced compared to initial values, depending on the treatments: available K increased by 18.11% in CPK treatment, but decreased by 6.81% −49.61% in the rest treatments. Wheat yield was highest in the balanced NPK fertilisation treatment. N or P deficiency in the soil significantly decreased the wheat yield but differences in available K content did not significantly affect wheat yield.
Composition of the weed community on the field surface. Seedling and ripening periods are the key periods of wheat growth. According to long-term observation, the weeds in ripening and seedling periods basically covered main weed species of wheat growing periods. And ripening period was the optimum observation time that can directly and obviously looked into the impacts caused by weeds and wheat competition.
During the wheat seeding period (Table 2), the total weed density in the different treatments was (high to low) C0 > CPK > CN > CNK > CNP > CNPK. In addition to the three dominant weed species (Vicia sativa, Lapsana apogonoides, and Conyza canadensis), the dominant weed community contained other species. These included Ammannia arenara in the C0 treatment, Geranium carolinianum, Alopecurus aequalis Sobol, and Mazus japonicas in the CNP treatment, Hemistepta lyrata and Alopecurus aequalis Sobol in the CNPK treatment, and Mazus japonicas and Alopecurus aequalis Sobol in the CPK treatment. These differences suggest that the fertilisation structure can affect the species composition of the dominant weeds in the early stages of wheat growth. Survey results during the wheat ripening period identified a total of 11 weed species, this was 6 fewer species than the 17 weed species found during the seeding period. The total weed density among the different treatments during the wheat ripening period was (high to low) CPK > C0 > CNK > CN > CNP > CNPK. This order was not consistent with the order observed during the seeding period. The weed species and communities observed during the wheat ripening period differed considerably among the various fertilisation treatments. The dominant weed communities observed during the wheat ripening period significantly differed from those observed during the seeding period. For instance, Scirpus juncoides and Beckmannia syzigachne were components of the dominant community for C0, CNPK, and CPK treatments during the wheat ripening period but were not in the dominant community during the seeding period. The dry biomass of the weeds increased as weed density increased. Under CPK and C0 treatments dry biomass was significantly greater than in treatments containing N and CPK was significantly greater than C0. Dry weed biomass in the CN, CNPK treatments was significantly lower than in the other treatments, so the N fertilisation and balanced NPK fertilisation appear to be most beneficial for weed reduction.
Community diversity characteristics of weeds. The community diversity of weeds under the long-term treatment with different fertilisers is shown in Fig. 1. The differences in the diversity indexes among the fertilisation treatments were smaller during the seeding period than during the ripening period. In the seedling periods, the CNPK treatment had the highest values and the CPK treatment had the lowest values for all the four diversity indexes. In the ripening periods, C0 treatment had the highest and CNP treatment had the lowest values for Shannon, Simpson and Margalef indexes, while CNK treatment had the highest and CPK treatment had the lowest for Equitability index. The data indicate that the balanced fertiliser application and the no fertiliser treatments increased the diversity of weeds in wheat fields.

Relationships between weed distribution, soil nutrient factors and wheat yield.
Redundancy analysis (RDA) ordination (Table 3, Fig. 2) reflects the general relationship between the weed distribution in the wheat fields and the soil nutrient factors and wheat yields. It also reflects the similarity among weed species for growing requirements. During the wheat seeding period, the eigenvalues of the first and second ordination axes were 0.560 and 0.282, respectively. During the wheat ripening period, the eigenvalues of the first and second ordination axes were 0.439 and 0.349, respectively. In each ordination, most weeds were located in the opposite direction, opposite the arrow of the soil fertility factor and had a negative correlation with the soil fertility factor, showing a tolerance for unfertile land. Therefore, the application of N and P fertilisers had a significant impact on the weed distribution.

Discussion
Fertilisation can help to manage weeds by enhancing the competitive advantage of crops 16 . The competition between crops and weeds can be significantly affected by the application of N fertiliser 17 . Many agriculturally important weeds are equally or more responsive than crops to higher soil N levels 18 . Without the use of herbicides, N fertilisation significantly increased weed density and dry matter. The opposite conclusion may be due to the lack of herbicide or short-term N application and other reasons. However, there are fewer articles that are the same as our experimental conditions 19,20 . Our study was based on a 31-year long-term experiment of fertiliser plots. Weeds were planted in conventional high-yielding cultivation of rice and wheat. NPK fertilisers were used in accordance with the growth requirements of wheat, and there will be the use of herbicides during 30-40 days after wheat seedling emergence. Our results indicated that the application of N fertiliser can significantly decrease the weed density in wheat fields. This is likely because N is the primary nutrient factor required to increase crop yield 21,22 . This is likely because, in our study, wheat had higher nitrogen requirements associated to a higher nitrogen uptake rate than the weed species. Fertiliser application can significantly enhance the NPK content in the soil, alter farmland nutrient habitats, and affect the plant species structure of farmland ecosystems 23 . After P fertilisation, the density of Mazus japonicas and Alopecurus aequalis Sobol significantly increased, while the density of Conyza canadensis considerably decreased. In a 47-year long-term experiment, Banks et al. found that Amaranthus weeds grew better in fields receiving only P fertiliser. Mollugo pentaphylla grew relatively well in fields receiving treatment with only P fertiliser or a NP fertiliser combination, whereas broadleaf weeds did not grow well in fields receiving balanced fertiliser treatments 24 . Different weeds have significantly different demands for various nutrients [25][26][27] . Katharine et al. 28 demonstrated that a decrease in soil N can increase the intraspecies competition of weeds, whereas P has a greater effect on the interspecies competition. The differences among nutrient levels in the soil caused by fertilisation can determine weed species presence. Therefore, fertilisation can affect the intra-and interspecies competition of farmland weeds, while potentially increasing the competitive advantage of crop plants.
Weeds are important for farmland diversity and changes in agricultural management have reduced the weed diversity. Loreau et al. 29 suggested that the diversity of plant communities is positively correlated with the primary productive forces and is related to soil fertility. Fertilisation can alter the diversity of the weed community in farmland 30,31 . The results of the present study indicate that for the C0 treatment without fertilisation and the CNPK treatment with balanced fertilisation, the individual diversity indexes attained relatively high levels. The wheat yield was also highest with the CNPK treatment. The CPK treatment without N fertilisation, and the CNP treatment, without K fertilisation had relatively low diversity indexes. The CNPK treatment, with balanced fertilisation maintains relatively high diversity and evenness index values and the richness index value was also higher than the other treatments.   Ecological system functioning is influenced by the characteristics and number of species. Differences in the requirements of species for different nutrients result in increased nutrient retention in the system 32 . Proper nutrient management can improve the competitive relationship between crops and weeds. This can reduce the influence of weeds on the crop yield, while maintaining diversity among the controlled weeds. Our RDA results indicated that the N and P nutrient levels in the soil and the organic matter content are the primary environmental factors that affect the distribution of weeds.
Our findings indicated that significant differences in the composition of weeds can be related to fertilisation treatments. N fertiliser had a greater effect on the density of different weed communities, P fertiliser had a greater effect on the weeds species present, and the impact of K fertiliser was unclear. The formation of the weed community in farmland derives from interactions among multiple environmental factors. The combination of NPK fertilisers can significantly reduce the weed density, increase the community diversity index, and achieve high yield and stability of a wheat crop production system.

Materials and Methods
Location of the experiment. The long-term experimental farmland study area was located at the Suzhou Academy of Agricultural Sciences in the Taihu area of Jiangsu, China (31°27′45″N, 120°25′57″E). This area has a northern subtropical monsoon climate; annual sunshine is 3,039 h, precipitation is 1,128 mm, average temperature is 15.7 °C, and the effective accumulated temperature is 4,947 °C. The soil used for the experiments was the acidic, heavy, loamy, paddy soil derived from loess sediments.
Materials for the experiment. The experiment was repeated in triplicate. The area of each small plot was 20 m 2 (4 m × 5 m), and the plots were permanently separated by ridges using granite slabs and cement, between which the individual small plots were connected with irrigation channels. The crops in this experiment were rice and wheat grown in rotation. Taihu Lake Cultivated Area, encompassing 1.68 million ha, is a high yield rice-wheat region and most fields use a rice-wheat double cropping system. This area is also one of the highest chemical N use areas in China. The total amount of chemical N used in both the rice and the wheat seasons ranges from 500-600 kg ha −1 .
During the wheat season, the N treatment consisted of urea at 150-300 kg ha −1 yr −1 (total N) between different years, with approximately 50% as a basal fertiliser, 20% as a tillering fertiliser, and 30% as a wheat earing fertiliser. The P treatment applied P 2 O 5 at 55.8 kg ha −1 yr −1 , all of which was basal fertiliser. Treatment K applied K 2 O at 137.5 kg ha −1 yr −1 , with 50% as base fertiliser and 50% as wheat earing fertiliser. Chemical weeding was used during both the rice and the wheat seasons. The same herbicides were used in all six treatments. During the rice season, the herbicides were applied 3-5 d after rice transplantation, and consisted of benzyl butyl isoproturon at 750-900 g ha −1 and 30% benzyl ethyl pretilachlor wettable powder at 1.5-1.8 L ha −1 . During the wheat season, 30% benzyl ethyl pretilachlor wettable powder was applied 30-40 d after sowing at 1.5-1.875 L ha −1 . During The arrow represents the environmental factor, the quadrant where the arrow is located represents the positive and negative correlation between the environmental factor and the ordination axis, and the length of the arrow connection represents the degree of correlation between an environmental factor and the distribution of the research objects. The longer the connection, the impact of this environmental factor on the distribution of objects is greater. The angle between the arrow connection and the ordination axis represents the correlation between the certain environmental factor and the ordination axis. The smaller the angle is, the higher the correlation is. herbicide applications in 2011 and 2012, shields were used to block measuring areas to avoid chemical weeding, we inserted shields around each plot with an area of 1 m 2 . Herbicides application was during 30-40 days after wheat seedling. As the wheat grew not high at that time, we used covers to block the test plots when weeding, and after weeding, we removed the covers. Survey method. Weed surveys were performed during the wheat seeding and ripening periods of 2011-2012. An iron frame with dimensions of 0.3 m × 0.3 m was used to randomly collect weeds at five sites in each plot. The weed identification was mainly accomplished by referral to the "China's Farmland Weeds Colored Map", and identification was verified using China's Weed Information System. During ripening periods, all the weeds on the field surface were collected and transported to the laboratory. Soil was removed from the weed and they were washed and then dried to a constant weight to determine the total dry matter.
Soil sampling and nutrient measurement. After the wheat harvest, a soil auger was used to collect soil samples at a 0-15 cm depth from each plot using the diagonal distribution method. The soil collected from the plots was numbered according to the treatment. After being fully mixed, the samples were taken to the laboratory for measurement. The total N in the soil was digested and measured using the Kjeldahl method. The total P was digested with H 2 SO 4 -HC1O 4 and measured using the molybdenum blue colorimetric method. Available P was determined using a soil extraction followed by colorimetric or inductively coupled plasma atomic emission spectroscopic (ICP-AES) analysis. Olsen-P is one of the most common extractants used, i.e. 0.5 mol/L sodium bicarbonate (NaHCO 3 ), which was in our preference due to the soil characteristics and pH of the experimental field. Nascimento et al. has reported that analysis of digested samples can in some cases cause the determination of inaccurate P concentration; however, the methods employed in the current manuscript were only methods available 33 . The alkaline hydrolysis N was measured by the alkaline hydrolysis diffusion method. Available P was measured with the sodium bicarbonate method. Available K was measured with the ammonium acetate extraction method. Organic matter was measured by the potassium dichromate volumetric method.
Data processing and analysis. The species diversity of the weeds was characterized by applying the following indexes.
The Species richness was determined using the Margalef index: R S N 1 ln = − . Simpson index and Shannon index are aggregative indicators to evaluate species diversity and individual distribution evenness. Shannon index brings out the uncertainty of the community, bases on the species number of weed community. The value of this index is larger, the samples have better evenness. Simpson index refers to the probability that two individuals randomly selected from a sample will belong to different species. Ranging from 0 to 1, the higher the index, the higher the diversity.
Margalef index refers to the species richness degree in a community or environment. The number of species per sample is a measure of richness. The more species present in a sample, the 'richer' the sample. Pielou index intends to estimate the evenness rate of weed species distribution. Evenness is a measure of the relative abundance of the different species. Pielou bases on Shannon index (H′) and biggest diversity value (Hmax), ranges from 0 to 1.
The basic data processing and statistical analyses were performed using Excel 2003 and SPSS 16.0, and the redundancy analyses of the soil nutrient data and the weed density data were performed using Canoco 4.5. The figures were drawn using Origin 8.5.