The factors affecting a native obligate parasite, Cuscuta australis, in selecting an exotic weed, Humulus scandens, as its host

In weed management, using native parasites to control exotic weeds is considered a better alternative than classical biological control. But the risk must be assessed because of the potential damage caused by these agents. We conducted this project to investigate the mechanism driving the choice of a native obligate parasite, Cuscuta australis, between the exotic, Humulus scandens, and native plants as its host through field and pot experiments. The results showed that C. australis preferred the exotic weed over native (naturalized) hosts and caused a notable reduction in the biomass of H. scandens in the field. In contrast, the results of the pot experimentindicated that C. australis preferred a mix of native (naturalized) hosts over the exotic weed. Both texperiments indicated that the parasitic preference of C. australis was induced more by light irradiance than plant water, carbon (C), nitrogen (N) and phosphorus (P) contents, indicating that the native parasite can only be used to control H. scandens when the exotic weed forms mono-cultures or dominates the community. Accordingly, induction and release of C. australis to control H. scandens should be conducted with great caution.

deliberately parasitize a mixture of host species to either obtain various types and amounts of nutrients or minimize the toxic effects of a single host 17 . Accordingly, it is very possible that parasitic plants would select a combination of plants with higher nutrient contents as their preferred hosts. In native communities, parasitic plants can increase species biodiversity by parasitizing and suppressing competitively superior and dominant host species 18 . Exotic invasive plants are usually the most competitive and dominant species in invaded communities 19 . So parasitic plants may prefer to parasitize and reduce the coverage of these first-order competitors (exotic plants in invaded communities) as keystone species and ecosystem engineers 20,21 . Experimental and observational results have actually showed that some exotic plants were more susceptible to these novel native generalist parasites than native and naturalized plants in their invaded ecosystems [7][8][9] , possibly because these invasive species have not evolved to resist or mitigate the virulence of the parasites 21,22 . However, whether the parasitic preference for exotic invasive plants is caused by the lower red light:far-red light ratio under the dense canopy (and/or higher nutrient contents) of invasive plants or the origin of the host requires more extensive research.
We previously found that a native obligate parasite, Cuscuta australis R. Br., was able to parasitize and suppress an exotic weed, Humulus scandens, in a field survey (Fig. 1), so our goal in this study was to compare the preferences of parasite for native and exotic plants and assess the risk of using this potential biocontrol agent before its induction and release. We hypothesized that 1) C. australis would prefer exotic H. scandens over single native (naturalized) hosts because H. scandens has higher nutrient contents and an exotic origin; 2) C. australis would prefer a mixture of native (naturalized) hosts over the exotic H. scandens due to greater benefits and 3) the host preference of dodders would mainly be induced by light cues resulting from the plant canopy rather than the properties of the host plant.

Results
Field experiment. In the field survey, the dodder biomass in the parasitized H. scandens subcommunity was significantly higher (74.4%) than that in the native subcommunity, while the light irradiance under the canopy of the parasitized H. scandens subcommunity was much lower (49.4%) than that under the canopy of the native subcommunity ( Fig. 2A, ANOVA, p < 0.01). The aboveground biomass of H. scandens in the parasitized subcommunity was much lower (37.8%) than that in the nonparasitized subcommunity (Fig. 2B, ANOVA, p < 0.01).
Pot experiment. The results from the single and mixed native (naturalized) host treatments were similar in the pot experiment (Figs 3 and 4). For the single native host treatment, the percent parasitism was not  significantly different between the native and exotic subcommunities in the LL and HH treatments, but the percent parasitism in the H. scandens subcommunity was distinctly lower in the LH treatment and higher in the HL treatment relative to the native subcommunity ( Fig. 3A, ANOVA, p < 0.01). Light irradiance did not significantly different between the native and exotic subcommunities in the LL and HH modes, but that in the H. scandens subcommunity was higher in the LH mode and lower in the HL mode compared to the native subcommunity ( Fig. 3B, ANOVA, p < 0.01). The water content of H. scandens was higher than that of native plants in three of the planting density modes (no differences in the LL mode), whereas the C content exhibited an opposite pattern (Fig. 3C,D, ANOVA, p < 0.05). The N and P contents were both higher in native plants than in exotic plants in the LH mode, while they did not differ markedly between native and exotic plants in the other three modes (Fig. 3E,F, ANOVA, p < 0.05). The results of two-way ANOVA showed that the differences in percent parasitism were mainly caused by density rather than plant origin or the interaction of the two factors (Table 1, ANOVA, p < 0.01).
In the mixed native host treatment, percent parasitism in the exotic subcommunities was higher in the HL mode, whereas it showed the opposite pattern in the other three modes (Fig. 4A, ANOVA, p < 0.01). Light irradiance, water content (except in the LH mode) and C content were the same as those in the single native host treatment ( Fig. 4B-D, ANOVA, p < 0.01). The N content was higher in native plants than in exotic plants in all modes, and the P content was higher in native plants than in exotic plants in the LL mode but not evidently different in the other three modes (Fig. 4E,F, ANOVA, p < 0.05). The results of two-way ANOVA showed that the differences in percent parasitism were caused by density and plant origin but not an interaction between the two ( Table 1, ANOVA, p < 0.05). Stepwise multiple regression analysis. The Pearson correlations among the variables showed that the percent parasitism by dodder was not significantly related to the plant water ( Table 2, R = 0.014, p = 0.901), or C ( Table 2, R = 0.193, p = 0.086) and P ( Table 2, R = 0.195, p = 0.084) contents, but the relationships between the percent parasitism by dodder and the plant N content ( Table 2, R = 0.314, p = 0.005) and irradiance ( Table 2, R = −0.582, p = 0.000) were both significant. However, the results of stepwise multiple regression analysis showed that only irradiance could be best fitted to the percent parasitism data, and the adjusted R 2 value was 0.330 (Table 3).

Discussion
Firstly, we report that a native parasite, C. australis, can parasitize and suppress the exotic weed, H. scandens, and it can cause substantial biomass loss (37.8%) of the exotic weed in the field, which indicates that C. australis can be used as a potential agent for the biocontrol of H. scandens. However, the parasitic preferences between native plants and exotic weeds were distinctly opposite in the two experiments: C. australis was observed to prefer exotic weeds over native hosts in the field, while it preferred mixed native hosts over exotic hosts in the pot experiment. Therefore, identifying the causes of the differences in parasitic preferences addresses the knowledge gap of the potential use of C. australis to control H. scandens. Because C. australis can parasitize both types of plants to a great extent, we can speculate that its parasitic preferences are not caused by the defenses of native and exotic plants or by its own detoxification mechanism. Generally, parasites are thought to prefer hosts with higher N contents [13][14][15][16] , and although our results also showed that the percent parasitism by C. australis is significantly related to higher plant N content ( Table 2) that is not the main cause. In our pot experiment, C. australis preferred the exotic weed over native hosts in the HL mode, while the N contents of the exotic plants were lower than those of native plants in the mixed native host treatment (Fig. 4A,E). Furthermore, stepwise multiple regression analysis indicated that parasitic preference of C. australis was not primarily caused by plant N content (Table 3). Accordingly, plant N content is not the main determinant of parasitic preferences in C. australis. Similarly, the parasitic preferences are also not induced by plant water, or C or P contents because percent parasitism by C. australis was not significantly correlated to these indexes (Tables 2 and 3). Furthermore, this parasite prefers mixed native hosts, enen though H. scandens had a higher water content and lower carbon (C) content in the mixed native host treatment (Fig. 4A,C,D, Table 1), so it can be concluded that the preferences of C. australis are mainly caused by factors other than plant properties. Luijckx et al. also concluded that parasite infectivity largely depends on factors other than host nutrients at the beginning of the infection process 23 .
Our data indicated no parasitic preference between native and exotic hosts occurred in the single native host treatment, whereas there was a significant parasitic preference for mixed native hosts in the mixed native host treatment (Table 1). Parasitic preferences for mixed hosts indicate that the parasite can improve its nutrient balance and minimize toxic effects, which is consistent with previous findings 20 . In addition, a mixed diet increases the ability of a parasite to encounter high-quality hosts by supporting growth in the gaps between preferred plants 17 , and parasites can obtain more nutrients from mixed hosts because of the increasing nutrient availability and production provided by their mutualism 24 , which is consisted with the higher nutrient contents (especially N) in the mixed native hosts observed in our study (Figs 3 and 4). Even though we did not investigate this factor in this experiment, another advantage of mixed hosts is that parasites can protect themselves to some extent from suffering the effects of environmental stressors, particularly herbivory 25 .
The results show that the percent parasitism by C. australis is significantly negatively related to irradiance under the plant canopy and that the plant density greatly affects the percent parasitism (Tables 1-3), supporting the results from the field experiment ( Fig. 2A), and these findings highlight that the parasitic preferences of C. australis are more affected by light irradiance than by plant properties 10,25 . High plant density and low irradiance indicate a dense plant canopy 9,10 , which is especially prominent among exotic plants in invaded communities 19 . Thus, the probability of C. australis encountering and parasitizing a host plant is much higher if it grows toward such environments 11,12 . In addition, the percent survival of C. australis will increase because the successful search for a suitable host plant is the most important step in its survival due to the limited resources available to its seedling 10 .   Parasites prefer hosts with a dense canopy (low irradiance) in the ecosystem regardless of whether the nutrient contents in the hosts are high or low, as observed in our field and pot experiments, which not only increases the survival rate of parasites but benefits them to a much greater extent. The total nutrient quantity (at least per unit area) in the dominant host is very likely greater than that in all other plants due to its high biomass, so it can support more parasites and sustain them for a longer time 13,26 . In addition, for parasites with low migration capacities, their spread can also be promoted by host density 11,12 . Moreover, competitive and dominant plants (especially exotic plants) usually have higher growth rates, and parasite performance is strongly correlated to the growth rate of the hosts 27 . As the most competitive and dominant plants in an ecosystem, exotic plants are more likely to be encountered by the seeds and seedlings of parasites 16 . Therefore, we can conclude that parasitizing dominant and competitive exotic hosts provides many advantages for a parasite, and the probability of a parasite encountering exotic weeds is much greater in invaded ecosystems. Both of these factors contribute to the parasitic preference of C. australis for the exotic weed H. scandens in our field study 13,20 .
In summary, C. australis prefers the exotic weed H. scandens over native hosts and causes great biomass reduction in the field, but the pot experiment indicates that C. australis prefers mixed native hosts over the exotic weed and that parasitism is caused more by light irradiance than by plant properties. Accordingly, we suggest that the induction and release of C. australis to control H. scandens should only be used when H. scandens forms mono-cultures or solely dominates a community.

Materials and Methods
Field experiment. In late autumn 2014, we selected six H. scandens communities parasitized by C. australis in the city of Changsha, Hunan Province. In each community, three subcommunities were identified located at least 10 m apart: a parasitized H. scandens subcommunity, a nonparasitized H. scandens subcommunity and a native subcommunity (including a naturalized species, Medicago sativa). To assess the suppressing effects of C. australis on the H. scandens community, three 1.0 × 1.0 m 2 plots were established in each of the parasitized and nonparasitized H. scandens subcommunities to measure the aboveground biomass of H. scandens after removing (only for the parasitized subcommunity) all C. australis organs from the H. scandens. Similarly, to compare the parasitism preference between H. scandens and native (naturalized) species, three 1.0 × 1.0 m 2 plots were also established in each of the parasitized and native subcommunities and dodder biomass on the plants and light irradiance under the plant canopy were measured in each plot. The dodder biomass was weighed after collecting all C. australis organs dissected from the stems and leaves of hosts. Light irradiance under the plant canopy in each plot was measured three times, and the mean value was considered to be the irradiance of the plot.

Pot experiment. From April to June 2014, a pot experiment was conducted in a greenhouse at Hunan
Agricultural University, Changsha, China. In April 2014, seedlings of the exotic weed H. scandens and four co-occurring native (naturalized) species were selected from an abandoned field (113°4′25.39″E, 28°11′3.92″N) that included Artemisia rubripes, Geranium carolinianum, Medicago sativa and Polygonum perfoliatum, all of which can be parasitized by C. australis in the field (Fig. 1). To compare parasitic preferences between native and exotic plants, an exotic subcommunity and a native subcommunity were constructed and planted separately (spaced 10 cm apart) on the both sides of a pot. Each pot had a rectangular area of 1500 cm 2 (soil surface), a depth of 20 cm and 25 kg of paddy soil (organic matter: 20.2-26.7 g kg −1 , total N: 6.2-6.67 g kg −1 , total P: 0.52-0.67 g kg −1 and pH: 5.86-6.12, soil layer approximately 15 cm). Four density modes were established: low-density exotic subcommunity vs. low-density native subcommunity (LL), low-density exotic subcommunity vs. high-density native subcommunity (LH), high-density exotic subcommunity vs. low-density native subcommunity (HL) and high-density exotic subcommunity vs. high-density native subcommunity (HH). Four individual plants were planted in the low-density subcommunities, and eight individual plants were planted in the high-density subcommunities. At the same time, two types of native subcommunity were established: only one native (naturalized) species was planted, which was treated as the single native host treatment; and all four native (naturalized) species were planted together, which was treated as the mixed native host treatment. The plants were evenly planted on both sides of the pots, and each native host treatment was replicated six times for the mixed native host treatment and four times for the single native host treatment. We measured light irradiance under the plant canopy of each subcommunity in triplicate twenty days after the hosts were planted, and the mean value was considered the light irradiance under the subcommunity. Simultaneously, ten pregerminated C. australis seedlings (approximately 4 cm with ring-like bases) were carefully and evenly placed in the center of the space between the native and exotic subcommunities in each pot, and the tip of each seedling was gently manipulated to point upward. In this experiment, we used 4 density modes, 2 native host treatments, and six or four replicates, resulting in a total of 40 pots. To prevent plants (especially the vines) from growing toward one another, a bamboo cane (80 cm long) was placed vertically beside each plant to support it. During the experiment, 300-500 ml of tap water was added to each pot every morning (7:00) to maintain soil moisture. Weeds (all unintentionally cultivated plants) were carefully removed from the pots, and other appropriate care was taken to minimize any disturbance. The survival and parasitic success of C. australis were considered to have occurred only when the tip extended more than 5 mm outward from the stem or remained coiled. The percent parasitism by C. australis in each subcommunity was determined one month after the planting, and it was calculated as the total number of surviving C. australis in each subcommunity divided by the total number of surviving individuals in each pot. Then, the shoots of each species were harvested and weighed after removing all C. australis organs from the stems and leaves of the hosts in each pot. The shoots were then oven dried at 65 °C for 72 h after drying at 105 °C for 30 minutes, weighed, and processed into fine powder for analysis. The water content of the plants was measured by dividing the weight of the water (wet weight minus dry weight) by the wet weight. Laboratory analysis. The total N concentrations (% dry mass) of the shoots were determined using an automatic N analyzer (Büchi-339), and the total P (% dry mass) contents were analyzed using the vanadium molybdenum yellow colorimetric method after digestion in H 2 SO 4 and H 2 O 2 and standardization against known reference materials. The total C concentrations (% dry mass) of the shoots were measured using a TOC analyzer (WIN TOC1010, O. I. Corporation, USA) after combustion at 650 °C with glass fiber.
Data analysis. All the raw data were analyzed using the SPSS 17.0 software package (SPSS Inc., Chicago).
Homogeneity of variance was tested by using Levene's test, and differences between means were determined using Duncan's test (P < 0.05). One-way ANOVA was used to analyze differences in dodder biomass and irradiance between the native subcommunity and the parasitized H. scandens subcommunity as well as differences in H. scandens biomass between parasitized and nonparasitized H. scandens subcommunities in the field. Differences in percent parasitism, irradiance, water content, C content, N content and P content between the native and H. scandens subcommunities in the pot experiment were also determined by one-way ANOVA. Two-way ANOVA was used to determine the effects of origin (native vs. exotic) and density (low vs. high) on the percent parasitism by dodder in the different native host treatments. To assess the primary factors inducing the parasitic preferences of the dodder, a correlation matrix was constructed using Pearson's linear correlation to assess the relationships among the variables (percent parasitism, light irradiance and plant water, C, N and P contents). Then stepwise multiple regression analysis was carried out with light irradiance and plant water as well as C, N and P contents as independent variables and percent parasitism as the dependent variable. Multicollinearity of the results was also checked by examining the variance inflation factors (VIF) of all predictor variables.

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.