Causes of differences in the distribution of the invasive plants Ambrosia artemisiifolia and Ambrosia trifida in the Yili Valley, China

Abstract Ambrosia artemisiifolia and Ambrosia trifida are two species of very harmful and invasive plants of the same genus. However, it remains unclear why A. artemisiifolia is more widely distributed than A. trifida worldwide. Distribution and abundance of these two species were surveyed and measured from 2010 to 2017 in the Yili Valley, Xinjiang, China. Soil temperature and humidity, main companion species, the biological characteristics in farmland ecotone, residential area, roadside and grassland, and water demand of the two species were determined and studied from 2017 to 2018. The area occupied by A. artemisiifolia in the Yili Valley was more extensive than that of A. trifida, while the abundance of A. artemisiifolia in grassland was less than that of A. trifida at eight years after invasion. The interspecific competitive ability of two species was stronger than those of companion species in farmland ecotone, residential, and roadside. In addition, A. trifida had greater interspecific competitive ability than other plant species in grassland. The seed size and seed weight of A. trifida were five times or eight times those of A. artemisiifolia. When comparing the changes under simulated annual precipitation of 840 mm versus 280 mm, the seed yield per m2 of A. trifida decreased from 50,185 to 19, while that of A. artemisiifolia decreased from 15,579 to 530.

They spread to other continents as early as 1836  or 1829 (Verloove, 2016) without considering its cultivation in botanical gardens. According to the global geographical distribution of the two species (CABI Invasive Species Compendium, https://www.cabi.org/ isc/searc h/index ?q=Ambrosia, accessed February 21, 2020), the distribution of A. artemisiifolia is more extensive than that of A. trifida. In addition, A. artemisiifolia and A. trifida occur in 80 and 40 countries, respectively .Why does A. artemisiifolia invade a larger area than A. trifida worldwide? In order to provide information necessary for the early warning of invasion by the two species, a more in-depth study is necessary.
Differences in distribution between species are normally caused by differences in genetic adaptation to environmental conditions. Ambrosia artemisiifolia and A. trifida often invade roadsides, farmland ecotones, wastelands (Essl et al., 2009;Milakovic et al., 2014;Pinke et al., 2013), residential habitats (Ziska et al., 2003), and other disturbed areas (Bassett & Crompton, 1982;Essl et al., 2015;Fumanal et al., 2008;Milakovic et al., 2014). Ambrosia artemisiifolia is rarely found in grasslands (Bullock et al., 2012); however, A. trifida occurs in grasslands (Regnier et al., 2016). In terms of specific regions, there is partial overlap between the two species' niches, but these two plants tend to invade different types of microhabitats. The main reason for the differences in habitat that they invade is not clear.
Generally, invasive plant species have strong performance-related traits, including those related to physiology, leaf-area allocation, shoot allocation, growth rate, size, and fitness than do noninvasive plant species (van Kleunen et al., 2010(van Kleunen et al., , 2015. Ambrosia artemisiifolia and A. trifida both have relative strong interspecific competitive ability . The effect of interspecific competitive ability on the distribution differences of two species is unclear. Water availability affects plant seed germination, growth, and reproduction, factors that are the basis of species distribution and competition, especially in arid and semiarid areas. Leiblein-Wild and Lösch (2011) found that A. artemisiifolia grew well under moist soil conditions and that it can survive in dry soils. Ambrosia trifida needs more water than A. artemisiifolia (Abul-Fatih & Bazzaz, 1979;Bassett & Crompton, 1982). It is not clear how the water use capacity affects the distribution difference of the two species. Moreover, the link between the differences in distribution and water demand of these two species during seed germination, plant growth, and reproduction period remains unclear.
The Yili Valley, Xinjiang, China, covers an area of 56,400 km 2 and contains a rich variety of habitats, including grasslands, farmlands, mountains, and residential areas (Jia et al., 2011). Our previous study found that A. artemisiifolia and A. trifida simultaneously invaded the same area of the Yili Valley in 2010, and we also found that the dominant habitat distributions of two species were different (Dong et al., 2017). Therefore, the Yili Valley provides a large, relatively closed field experiment site in which to study the beginning of an invasion by the two species along with their subsequent diffusion.
This study can therefore help to explain the distribution differences and causes for successful invasion of A. artemisiifolia and A. trifida, providing insight into the reasons for the resulting distribution of these two species worldwide.
Distribution and abundance of A. artemisiifolia and A. trifida were surveyed and measured from 2010 to 2017 in the Yili Valley, Xinjiang, China. The soil physical and chemical properties, soil temperature and humidity, and the main companion species were determined in farmland ecotone, residential area, roadside, and grassland in 2017. Also, biological characteristics, such as density and coverage, plant height, number of seeds per plant, 100-seed weight, and seed size of these two species and companion species (density and coverage, plant height) in four habitats, were measured in 2017. Moreover, the differences in water demand between the two species were studied through seed germination and garden experiments from October 2017 to October 2018. The following questions were explored: What were the differences in the distribution of these two species in the Yili Valley? What caused the differences in the distribution of these two species?  (Chen, 1993).
Xinyuan County (43°03′-43°40′N, 82°28′-84°56′E) is located in the hinterland of the Gongnaisi grassland in the eastern part of the Yili Valley. This site is the main distribution area of A. artemisiifolia and A. trifida. The average annual temperature and precipitation are 8.1°C and 480 mm, respectively. We studied the interspecific competitive ability, seed size, and water demand differences between the two species in farmland ecotone, residential area, roadside, and grassland in Xinyuan County, because these four habitats were the main distribution areas of the two species (Dong et al., 2017), and there were relatively large differences in water status, temperature status, and companion species between those four habitats.

Distribution area
The distribution areas of A. artemisiifolia and A. trifida in the Yili Valley were surveyed and measured during the growth periods from 2010 to 2017. Every year, through a large number of field censuses, new distributional points of the two species were recorded with GPS to determine the current distribution boundaries of the two species.
In order to ensure the accuracy of the measurements, we included a sufficient number of boundary points, with the distances between two consecutive points limited to 2-3 km. These points were then marked on a Google map to calculate the distribution areas.

Soil physical and chemical properties
The differences in soil physical and chemical properties were compared between farmland ecotones, residential areas, grasslands, and roadsides. The upper 0-20 cm of soil from four habitats (selected at observation points determined in 2010 in Experiment 1 where A. artemisiifolia and A. trifida were present) was divided into two layers. The soil in each 10-cm layer was sampled, and soil properties were determined in July 2017 as follows. Total nitrogen, total phosphorus, and total potassium were determined using the micro-Kjeldahl, sodium hydroxide melting-molybdenum anticolorimetric, and flame photometry methods, respectively. Soil pH was measured using a Mettler-Toledo pH meter (UB-10, USA), and soil conductivity was measured using a conductivity meter (Hach, USA). Soil organic matter content was checked using the K 2 CrO 7 -H 2 SO 4 external heating method. Alkaline hydrolysis nitrogen, available P, and available K were measured using the alkaline hydrolysis diffusion method, Mo-Sb colorimetry, and the ammonium acetate method, respectively. Soil samples from each habitat were taken three times in three individual sites (more than 5 km apart), and a total of 3 (repetition) × 4 (habitat) × 2 (soil layer), which resulted in 24 samples being collected.

Soil temperature and humidity
In order to compare the water demand between the two species analyzed, soil temperature and humidity meters (Watch Dog 1200, USA) were placed in the 10-cm soil layer in the four habitats on 1 September 2017; the meters were removed on 2 October 2018.
Each temperature and humidity meter recorded data every hour.
The data from 1 October 2017 to 30 September 2018 were used to analyze the annual conditions. Three temperature and humidity recorders were placed in three individual sites for each habitat, and a total of 3 (repetition) × 4 (habitat) = 12 recording units were set up.
The data for temperature and humidity were divided into four

| Experiment 3: Observation of biological characteristics
In (2) Total abundance = (number of occurrences in all habitats∕100) × 100%. each plot. If there were fewer than 30 Ambrosia or companion species plants, we measure all of them. Three plots from each habitat were taken in three individual sites, for a total of 3 (repetition) × 4 (habitat) sampling units in 12 plots being set up.
Six A. artemisiifolia and A. trifida plants were randomly selected from each sample plot, and all seeds counted on these plants were removed in September 2017. If some seeds had fallen, we estimated the number based on the locations of the seeds. A total of 100 seeds from each plant were randomly selected, air-dried, and weighed with 0.0001 g precision on an electronic balance (BDS, China). Twenty seeds were randomly selected from each plant, and the lengths and widths of these seeds were measured with Vernier calipers (BDS, China) to calculate the average seed size using Equations (3) and (4), with three repetitions for each:

Seed germination
Seed germination was analyzed in the laboratory. In October 2016, the seeds of A. artemisiifolia and A. trifida were collected from four habitats in the Yili Valley and combined. The amount of seeds between habitats was set to be equal, and the seeds were initially stored in the dark at 0-5°C in a cold storage room with 40% relative humidity (Bae et al., 2016). In June 2017, 50 g heat-dried in situ soil samples were weighed, and each sample was placed in a Petri dish. Next, 2.5, 5, 7.5, 10, and 12.5 g of distilled water was added to each Petri dish, resulting in the soil moisture contents in the various Petri dishes of 5%, 10%, 15%, 20%, and 25%, respectively. For each sample, 20 fully developed undamaged same-sized seeds of A. artemisiifolia or A. trifida were uniformly spread on the soil surface in Petri dishes. Each group of seeds was evenly placed in Petri dishes.
The seeds were treated in a climatic chamber China) for 60 days at 20-10°C, 12-hr/12-hr light/darkness, and 3,000 lx light intensity, after which the germination rate was calculated by counting the number of germinated seeds. Seeds with the seed radicle at least 0.2 mm long were considered to have germinated. Seed germination was checked every day. When no seeds germinated in a single Petri dish for five consecutive days, it was regarded as the end of germination.

| Statistical analysis methods
One-way analysis of variance (ANOVA) and multiple least significant difference comparisons were used to explore the differences in soil physical and chemical properties (Table 1) and soil temperature and humidity (Figure 3) between the four habitats, while 100-seed weight, seed size, number of seeds per plant, and seed yield per m 2 were compared between A. artemisiifolia and A. trifida ( Figure 5). ANOVA was also used to examine differences in densities, coverage, and plant heights of A. artemisiifolia, A. trifida, and companion species between the four habitats ( Figure 4). ANOVA, multiple least significant difference comparisons, and t tests were used to explore the differences in seed germination (Table 3), density and plant height (Figure 6), and seed yield (Table 4) of A. artemisiifolia and A. trifida in different water gradients.IBM SPSS Statistics 20 was used for data analysis, and OriginPro 8.5 was employed for graphics.

| Distribution differences between A. artemisiifolia and A. trifida (Experiment 1)
Ambrosia artemisiifolia and A. trifida invaded the Yili Valley starting in 2010. Since 2014, the areas occupied by these two species have increased rapidly, although A. artemisiifolia is distributed over a larger area than A. trifida. By 2017, these two species had occupied 1,322 and 311 km 2 , respectively; thus, the former occupied 4.25 times the area inhabited by A. trifida (Figure 1).
(3) Seed size = seed length × seed width. The species abundances were measured in the 25 plots in each habitat. From 2010 to 2017, the abundance of A. artemisiifolia was higher than that of A. trifida and increased rapidly in farmland ecotone, residential area, and roadside habitats. By 2017, total abundance of A. artemisiifolia and A. trifida was 57% and 39%, respectively, so that A. artemisiifolia was 1.46 times more abundant than A. trifida.
However, the abundance of A. artemisiifolia in grassland was less than that of A. trifida, where the latter was 3.5 times more abundant than the former (Figure 2).

| Soil physical and chemical properties, soil temperature and humidity, and companion species in four habitats (Experiment 2)
Soil physical and chemical properties showed little difference among different habitats; however, the soil total nitrogen levels in farmland ecotones and roadsides were higher and lower, respectively, than those in other habitats. The contents of available phosphorus and available potassium in grasslands were lower than those in other habitats. Soil organic matter content in grasslands was higher than that in other habitats (Table 1).
In SP, the soil temperature of farmland was significantly higher than in other habitats, and that of grassland was significantly lower than in other habitats; in GP, the soil temperatures of grassland and roadside were significantly lower than those of other habitats; in FFP, the roadside temperature was significantly lower than those of other habitats. The soil moisture in different habitats showed significant differences in different periods, and the values of soil moisture were ranked as follows: grassland > farmland ecotone > residential area > roadside (Figure 3).
The types of companion species in farmland ecotones, residential areas, and roadsides were similar and quite different from those of grassland (Table 2).

| Biological characteristics of A. artemisiifolia, A. trifida and companion species (Experiment 3)
Ambrosia trifida was significantly taller than other plant species in all habitats from LS period to MR period, being 3.45-8.3 times taller than the companion species in the FR period. Ambrosia artemisiifolia was significantly taller than companion species in the farmland ecotone and residential area from ES period to MR period (Figure 4).
The density of A. trifida was significantly higher than that of the other species in all habitats, reaching 1.35-4.4 times that of the companion species in the FR period. The density of A. artemisiifolia was higher than that of the companion species in the farmland ecotone, residential area, and roadside, at 1.39-2.23 times that of the companion species in FR period. However, the density of A. artemisiifolia was lower than that of the companion species in grassland, at only 0.37 times the density of the companion species in FR period (Figure 4).
The coverage of A. trifida was significantly greater than that of the other species in all habitats from LS period to MR period, at 1.31-2.8 times that of the companion species in FR period, respectively.
The coverage of A. artemisiifolia was significantly higher than that of the companion species in the farmland ecotone and residential area from EG period to MR period, at 1.84 and 1.7 times that of the companion species in FR period, respectively. However, the coverage was significantly lower than the companion species in grassland, at 0.53 times that of the companion species in FR period ( Figure 4).

| The water demand for seed germination, plant growth, and reproduction in A. artemisiifolia and A. trifida (Experiment 4)
The seed germination rates of A. artemisiifolia and A. trifida increased with increasing soil moisture. However, no significant difference in seed germination rate was observed when comparing these two species under the same soil moisture content (Table 3).  (Table 4). (mg/kg) 10-20 88.3 ± 10.3a 43.1 ± 5.12b 11.9 ± 1.97c 40.9 ± 8.29b

F I G U R E 3
Soil temperature and humidity in the seedling period (SP), growing season (GP), flowering and fruiting period (FFP), and winter season (WP) of the four habitats analyzed in the present study in the Yili Valley. Different letters indicate significant differences at p < .05 using least significant difference tests for different habitats with relatively poor water availability and weak interspecific competition, A. artemisiifolia was much more abundant than A. trifida. In grassland with relatively rich water availability and strong interspecific competition, A. trifida was much more abundant than A. artemisiifolia ( Figure 2). In the study area, more types and larger areas of suitable habitat are available to A. artemisiifolia than to A. trifida, which is consistent with the distribution of these two species worldwide (Bullock et al., 2012;Chauvel et al., 2006;Follak et al., 2013;Montagnani et al., 2017;Regnier et al., 2016).
Greater population density, higher plant height, and greater coverage are conducive to successful plant invasion (Chapman et al., , 2016. Although the density, height, and coverage of A. trifida were higher than those of A. artemisiifolia and companion species in roadside, farmland ecotone, and residential area in the present study (Figure 4), the distribution points of A. trifida were all located in low-lying and waterlogged areas (Figure 2). Ambrosia artemisiifolia is highly competitive in continuously disturbed habitats such as roadsides and farmland ecotones (Bullock et al., 2012;Gentili et al., 2015Gentili et al., , 2017Kazinczi et al., 2008;Novak et al., 2009) as the disturbances decrease competition. Ambrosia trifida is widely distributed in grassland as the density, height, and coverage of A. trifida are higher than those of A. artemisiifolia and companion species ( Figure 4). The life-history strategy of A. trifida is mostly based on rapid growth that allows the plants to quickly reach a greater height and biomass than other plants (Abul-Fatih & Bazzaz, 1979). Stronger interspecific competitive ability of A. trifida may explain larger distribution of the species in grassland.

The primary means of dispersal of A. artemisiifolia and
A. trifida seeds are barochory (Basset & Crompton, 1975;Montagnaniet al., 2017).The medium-distance and long-distance dispersal of A. artemisiifolia and A. trifida is driven by human activities and obstruction in many ways (Bullock et al., 2012). Seed size is an important factor affecting seed diffusion and species distribution (Washitani & Nishiyama, 1992). Ambrosia artemisiifolia has lighter and smaller seeds ( Figure 5), so A. artemisiifolia seeds are easier to spread in habitats with more human activity such as residential area and roadside (Bullock et al., 2012;Essl et al., 2009;Skálová et al., 2017).
Easier spread of seeds of A. artemisiifolia may explain larger distribution of the species in the Yili Valley. In addition, the long-term seed bank of A. artemisiifolia (Fumanal et al., 2008;Webster et al., 2003) may be mentioned as a factor stabilizing populations, especially in very dry years when seed production is low.
Ambrosia artemisiifolia can grow well and produce more seeds than A. trifida with a limited water supply when the latter produces almost no seeds (Table 4). This shows that A. artemisiifolia has a stronger ability than A. trifida to tolerate drought. The net photosynthetic rate of A. artemisiifolia decreases during periods of reduced soil water content (Bazzaz, 1973), but the plants recover rapidly from short-term droughts (Bazzaz, 1973(Bazzaz, , 1974. In unusually dry years or on dry sites, A. artemisiifolia plants have   Figure 3). Therefore, we believe that the existing distribution pattern of the two species is not mainly affected by temperature in the Yili Valley.
Since the causes of species distribution include factors other than interspecific competition, seed size, and water demand, other issues need to be discussed in future work if researchers wish to better explain the reasons for the differences between these two species. Additional factors to investigate include the following: (a) How temperature and water work collectively to affect the germination, growth, and reproduction of these two TA B L E 3 Seed germination of Ambrosia artemisiifolia (A. a.) and Ambrosia trifida (A. t.) in different soil moisture contents

F I G U R E 6
The density and plant height in the early seedling (ES) period, late seedling (LS) period, early growth (EG) period, late growth (LG) period, flowering (FR) period, and maturity (MR) period in 280 mm, 560 mm and 840 mm of simulated annual precipitation for Ambrosia artemisiifolia (A. a) and Ambrosia trifida (A. t). Different capital letters indicate significant differences at p < .05 using independent t tests for A. artemisiifolia and A. trifida. Different lowercase letters indicate significant differences at p < .05 using least significant difference tests for different water gradients species; and (b) quantitative analysis of the influence of the difference in seed size on the difference in distribution of the two species.

ACK N OWLED G M ENTS
This work was financially supported by the Natural Science Zhigang Li: Software (equal). Note: Different capital letters indicate significant differences at p < .05 using independent t tests for A. artemisiifolia and A. trifida. Different lowercase letters indicate significant differences at p < .05 using least significant difference tests for different water gradients.