Host preference of Thrips hawaiiensis for different ornamental plants

Thrips hawaiiensis is a common thrips pest that damages the flowers of various plants. The differing population sizes of T. hawaiiensis among host plants suggest its preference and performance vary among host plants. In this study, the host fitness of T. hawaiiensis for different flowers was assessed through field investigation. The behavioral responses of T. hawaiiensis to the color and volatiles of flowers eliciting different apparent fitness levels and their development and survival on the plants were also studied. Adults and larvae of T. hawaiiensis were found in the flowers of 21 species, which were classified into four fitness levels for this thrips species. T. hawaiiensis showed significantly different visual responses to the color and olfactory responses to the volatiles of four tested flowers (each representing one of the four fitness levels), with the rankings of visual preferences for Dianthus caryophyllus > Tulipa gesneriana > Hydrangea macrophylla > Rosa rugosa, and olfactory preferences for H. macrophylla ≥ T. gesneriana > D. caryophyllus > R. rugosa. Plant species had significant influences on the development and survival of T. hawaiiensis, with developmental times from egg to adult of 9.58 d, 9.92 d, 10.35 d and 10.75 d on H. macrophylla, T. gesneriana, D. caryophyllus and R. rugosa, respectively, and corresponding survival rates of 76.33%, 71.33%, 64.00% and 59.00%. In summary, this study shows that olfactory preferences were consistent with the field performance of T. hawaiiensis on the four flower plants tested. Further, fitness levels of host plant flowers are correlated with development rate and survivorship of T. hawaiiensis. Our study adds to the understanding of the mechanism of host selection by thrips and provides basic information to underpin the management of T. hawaiiensis on horticultural plants.


Introduction
Thrips hawaiiensis (Morgan) (Thysanoptera: Thripidae) is a common flower-dwelling thrips (Mound 2005) that is native to Oriental and Pacific regions. Through increasing international trade, it has expanded its geographical range to Africa, Australia, Europe, and America (Reynaud et al. 2008;Goldarazena 2011;Atakan et al. 2015). T. hawaiiensis has a wide host range among crops, including various fruits, vegetables and ornamental plants (Murai 2001;Aliakbarpour and Salmah 2011;Cao et al. 2018). Thus, T. hawaiiensis is becoming a more important agricultural and horticultural pest worldwide, particularly in many provinces of China. It is a dominant thrips pest on banana and mango and results in large annual economic losses through yield reductions (Wu et al. 2014, Fu et al. 2018. Thrips hawaiiensis prefers to feed and live in flowers, rather than in other plant structures (Murai 2001;Mound 2005;Fu et al. 2020). Different population sizes and associated degrees of damage have been found on different host plant species (Huang et al. 2009;Cao et al. 2018). These differences in populations indicate that T. hawaiiensis exercises host preferences in the field.
For thrips, host orientation and selection is driven by visual and olfactory cues Gillespie 1990, 1995;Teulon et al. 1999;Pearsall 2000;Cao et al. 2019). The rates of, for example, development, survival, and oviposition vary significantly when different plant species are supplied as food sources (Brown et al. 2003;Li et al. 2015;Cao et al. 2018). Thrips also show different physiological responses to different host plants. Frankliniella occidentalis (Pergande) and F. intonsa (Trybom) have significantly different activities of detoxifying enzymes and protective enzymes when feeding on different host species (Liu et al. 2017). Therefore, both behavioral and physiological responses of thrips to the physical and chemical properties of plant species affect host selection and performance of these insects.
In this study, the host suitability of flowering plants for T. hawaiiensis was assessed through field investigation in the Guiyang area, Guizhou Province, China. In conjunction with field observations, further research was conducted on how fitness is related to the host preferences of T. hawaiiensis. Thus, we studied the visual responses to color, olfactory responses to volatiles, and the development and survival of thrips on different host plants with different levels of fitness, as determined in the field investigations. The results of this study help to explain the differences in the extent of damage caused by T. hawaiiensis among different host plants, and provide basic data to explore the host selection mechanisms of this thrips species. Ultimately, this information will be useful for devising targeted strategies to control of this pest on particular horticultural plants.

Host associated fitness of T. hawaiiensis under field conditions
Field sampling was undertaken in Guiyang area, Guizhou Province, China, from April 2015 to December 2016, during the early, middle and late period of each month. No pesticides were used in the sampling areas. In total, 10 main locations were selected for sample collection, with necessary changes according to the richness of plant species in the flowering stage in different sampling periods. All the flowers present in each sampling area were sampled, so all of the flowering plants in Nanming district, Guiyang, China, were included in our study. Each sampling area was approximately 400 m 2 , and a five-point sampling method was used. The center of the diagonals and four points equidistant to this point were fixed as sampling sites. Two plants were selected per host at each location. The number of adults of T. hawaiiensis were counted in six flowers in total per plant, two collected from the lower, middle and upper part of the plant, respectively. These flowers were taken to the laboratory to allow eggs to hatch. The emerged larvae were reared in plastic containers (20 × 14 × 9 cm) with snap-on lids, in climate-controlled rooms at 26 ± 1 °C, 70 ± 5% relative humidity (RH), under a 14:10 h light/dark photoperiod (Cao et al. 2014). The larvae were fed with the flowers of the plant they were collected from until they reached the adult stage. All of the adult thrips were identified under a microscope (Olympus CX41, ZSA300) to evaluate the proportion of T. hawaiiensis (Cao et al. 2018(Cao et al. , 2019. A selectivity index (I) and suitability index (P) were used to evaluate the host fitness for T. hawaiiensis (Yuan et al. 2011;Cao et al. 2019), and were calculated as follows: where N is the number of times T. hawaiiensis was found on each flowering host plant, M is the total number of investigation times for each host plant, and La and LA are the largest numbers of, respectively, larvae and adults (male and female combined) of T. hawaiiensis observed among all replicates of each host plant. For each plant species, the number of thrips per flower was used to assess the host fitness for T. hawaiiensis, but the influence of flower size was not considered among different plants. The host plants were classified into different fitness levels for T. hawaiiensis, as measured using the method described by Bo et al. (1997) and Yuan et al (2011), which are based on the comparison between the P value of each plant flower and X (the mean value of P for all the flower species). The most suitable host plants (****) (P ≥ 2X), suitable host plants (***) (X ≤ P < 2X), relatively suitable host plants (**) (X/2 ≤ P < X), and least-suitable host plants (*) (P < X/2).

Insects and plants
Thrips hawaiiensis adults were collected from various plant species in Guiyang area, Guizhou Province, China, and used to establish laboratory colonies. These thrips were transferred onto green bean pods Phaseolus vulgaris L. (Fabales: Leguminosae) in plastic containers (20 × 14 × 9 cm) with snap-on lids. A hole (5 cm × 4 cm) was cut in the lid and covered with thrip-proof organdy (40-μm mesh) to allow for ventilation (Cao et al. 2014). The thrips were kept in a climate-controlled room at 26 ± 1 °C, 70 ± 5% RH, under a 14 L:10D h photoperiod. Subsequent colonies were reared continuously for more than five generations before being used in bioassays.
The flowering plants used in these experiments were determined to be hosts of T. hawaiiensis with different fitness levels in our field investigation. The species used were  (Cao et al. 2018). Greenhouses were maintained free of insect pests by insect-proof netting, and plants were cultivated without application of insecticides. Plants were sampled at the same flowering stage. Flowers at anthesis with intact petals were collected for olfactory tests and investigation of development and survival of thrips.

Free choice of flower color by thrips
Colored paper was used to create different visual stimulations (Cai et al. 2015) using the RGB color mode of Byers (2006). The paper samples were purple (RGB: 128, 0, 128), yellow (RGB: 255, 255, 0), white (RGB: 255, 255, 255) or pink (RGB: 255,192,203), similar colors to the flowers of H. macrophylla, T. gesneriana, D. caryophyllus and R. rugosa, respectively. The selectivity of T. hawaiiensis among the colors was tested in plastic Petri dishes (diameter 15 cm), using the method of Pang et al. (2004) and Tian et al. (2017) with necessary modifications. A circle (diameter 1 cm) was drawn in the center of the bottom of each Petri dish, and a " + " shape that filled in the whole dish was drawn through the center of the circle to divide the dish into four quadrants. Paper of the different colors was cut into small circles (diameter 3 cm), and one of four different papers was placed in each quadrant. T. hawaiiensis 2-3-day-old adults after emergence were starved for 6 h before bioassays and then introduced in groups of ca. 80 individuals into the Petri dish through a hole (diameter 0.5 cm) in the center of the dish cover. After 30 min the number of thrips on each of the different colored papers was counted; thrips in the center of the dish (a circle with diameter 3 cm) were considered as not having chosen a particular color. The paper locations were rotated (90º) after each test to minimize positional effects, and the color paper randomized within each dish. Bioassays were replicated four times, and all were carried out between 09:00 and 17:00. Each replicate contained three dishes. Females and males T. hawaiiensis were tested separately.

Olfactory responses of thrips
The olfactory responses of T. hawaiiensis to the volatiles of different plant flowers were tested in a Y-tube olfactometer by the method of Cao et al. (2014Cao et al. ( , 2019. In these tests, we made two types of two-way comparisons: (1) each flower (no leaves or stems) vs. clean air (CA), and (2) all possible flower pairings. For each comparison, 60 T. hawaiiensis (2-3-day-old) were tested individually. Compared with male T. hawaiiensis, females are more sensitive in responding to the volatiles from host plants (Cao et al. 2020), thus only female thrips were used in these tests. Thrips were starved for 3 h before testing, and the flower material (15 g) was replaced after every 10 individuals tested. A thrips was considered to have made a choice when it crossed 2/3 of one arm within 5 min. Otherwise, the thrips was considered to be non-responding. Airflow through each arm of the olfactometer was maintained at 300 ml/min (Cao et al. 2014(Cao et al. , 2019. All bioassays were conducted between 08:00 and 18:00.

Development and survival of thrips
Flowers of each host plant were placed in plastic containers (20 × 14 × 9 cm) with snap-on lids. A group of ca. 150 adult thrips (female: male = 1: 1) was placed in each container to allow oviposition for 24 h (Cao et al. 2014). Because eggs are laid inside the flower tissue and are not visible, their mortality could not be determined. Newly emerged larvae were placed on fresh flower disks, as described by Gaum et al. (1994) and Van Rijn et al. (1995) with modifications (Cao et al. 2018). The development from egg to adult of T. hawaiiensis was observed daily, and juvenile survival was assessed every 12 h. Developmental periods of eggs were determined by recording the time until the appearance of larvae (Van Rijn et al. 1995;Zhang et al. 2007). One hundred first instars were counted for each plant species as initial observation number; one larva being placed on each flower disk. This was replicated three times with total of 300 larvae tested per type of flower. These investigations were conducted at 26 ± 1 °C, 70 ± 5% RH, and a 14:10 h (L:D) photoperiod.

Statistical analysis
Data were analyzed using SPSS software (version 18.0; SPSS, Chicago, IL, USA). Data were checked for normality and homoscedasticity before further analyses. Statistically significant differences (P < 0.05) in the color selectivity, developmental time, and survival rate of thrips among different flowers were analyzed by one-way analysis of variance (ANOVA) and Tukey's test. The null hypothesis that T. hawaiiensis adults showed no preference for either Y-tube arm (a response equal to 50:50) was analyzed using a chisquare goodness-of-fit test.

Host associated fitness of T. hawaiiensis under field conditions
In our field investigations, we found that T. hawaiiensis occurred on 21 plant species with different population sizes, indicating differential host preferences by the thrips (Table 1). Host fitness of these flowers for T. hawaiiensis was measured quantitatively using the selectivity index (I) and suitability index (P). The highest value of I was 1, and was recorded for three flower species, namely H. macrophylla, Gardenia jasminoides Ellis, and T. gesneriana. The lowest value of I was 0.27, which was recorded for Paeonia suffruticosa Andr. The highest P value was 0.87 for G. jasminoides, and the lowest was 0.01 for P. suffruticosa. The host plants were classified at four fitness levels for T. hawaiiensis based on the comparison between the P value of each plant flower and X for 0.16 (the mean value of P for all the investigated flower species). The most suitable host plants comprised two species (P ≥ 2X), G. jasminoides (P = 0.87) and H. macrophylla (P = 0.65). Suitable host plants comprised three species (X ≤ P < 2X), T. gesneriana (P = 0.30), Gerbera jamesonii Bolus (P = 0.33), and Telosma cordata (Burm. F.) Merrill (P = 0.28). Relatively suitable Table 1 Host fitness for T. hawaiiensis based on field investigation I is the selectivity index (I), P is the suitability index (P), and L a and L A are the largest numbers of larvae and adults (male and female combined), respectively, of T. hawaiiensis observed among all replicates of each host plant

Number
Host I L a L A P Fitness    Fig. 1A). Male T. hawaiiensis showed a similar preference ranking as females for these four flower colors, except there was no significant difference between purple and yellow (F = 99.79, P < 0.0001, Fig. 1B).

Development of thrips
There were significant differences in the developmental time from egg to adult of T. hawaiiensis among the different host plant flowers (F 8, 812 = 48.99, P < 0.01). Development required the least amount of time on H. macrophylla (Table 2). Development was over 12% faster on H. macrophylla than on R. rugosa. This was caused by the significantly different developmental durations of first instars (F 8, 1011 = 12.96, P = 0.002) and second instars (F 8, 899 = 18.39, P = 0.001). There were no significant In each Petri dish, 80 thrips were released per replicate. Different lowercase letters above bars indicate significant differences (one-way analysis of variance followed by Tukey's HSD test, P < 0.05) differences among host plants for any of the non-feeding stages of immature thrips ( Table 2).

Survival of thrips
Flower species significantly influenced the survival rates of T. hawaiiensis at the first instar (F 8, 1011 = 58.25, P < 0.01), prepupa (F 8, 849 = 4.94, p = 0.032), and from egg hatching to adult stages (F 8, 812 = 106.27, P < 0.01) (Fig. 3) As euryphagous herbivorous insects, thrips show clearly different feeding preferences and population sizes among different host plants (Mound 2005). For T. hawaiiensis, based on the thrips collected from all the flowers in our sampling area, we identified 21 flower plants that were attacked by this pest species, and we classified the host plants at four different fitness levels for T. hawaiiensis based on its performance on these plants in the field. There were 14 herbaceous Fig. 2 Olfactory responses of T. hawaiiensis females to the volatiles of different flowers. CA: clean air. Asterisks indicate highly significant (**P < 0.01) and significant (*P < 0.05) differences in the selectivity of 60 adult T. hawaiiensis females between two odors (χ 2 test). n.s. indicates no significant difference (P > 0.05) in selectivity of T. hawaiiensis between two odors  (Qin and Wang 2001;Schoonhoven et al. 2005). It has been reported that the pupation rates of F. occidentalis in the soil and host plants can fluctuate depending on whether the plants are in the flowering or non-flowering stage. Buitenhuis and Shipp (2008) found that the proportion of F. occidentalis remaining on plants to pupate significantly increased in the presence of flowers, but a large proportion of thrips pupated in the soil. In the present study, we only counted the number of T. hawaiiensis pupa in plant flowers. Therefore, the number of T. hawaiiensis pupating in the soil is unknown. Further studies should consider the proportions of pupa in the soil and in flowers to determine whether this is a significant factor in the relationship between thrips and their hosts. Further studies should also evaluate the preferences of T. hawaiiensis among host plants at the non-flowering stage to better understand the interactions between thrips and hosts. Visual and olfactory cues are important for thrips when searching for and selecting host plants. As reported extensively for F. occidentalis, thrips are attracted by the color and volatiles of host plants (Teulon et al. 1999;Mainali and Lim 2011;Cao et al. 2018Cao et al. , 2019Avellaneda et al. 2019). We observed similar results for T. hawaiiensis in this study. Importantly, the olfactory preference rankings of female thrips for the four flowers tested in this study were in accordance with their fitness levels as hosts for the thrips. These results indicate that volatiles play an important role for T. hawaiiensis in selecting the most suitable host plant species for their offspring. This pattern was also observed in F. occidentalis in our previous studies (Cao et al. 2014(Cao et al. , 2019. Therefore, it may be reasonable to pay more attention to the control of T. hawaiiensis on its preferred host plants because more thrips might aggregate on these plants, causing more serious damage. In addition, the volatiles from the preferred plant might act as aids for mating and oviposition, leading to more pest offspring (Groot et al. 2003;Olsson et al. 2006;Xu and Tarlings 2018).
There might be differences in host-discrimination mechanisms between the olfactory and visual responses of T. hawaiiensis, because its visual preference rankings were not in accordance with the host fitness level for the thrips. However, only simulated flowers using paper of fixed shape and size in different colors were tested in this study. The behavioral responses of T. hawaiiensis to flowers with natural color, size and shapes should be tested to evaluate comprehensively the visual preference of this thrips. For example, F. occidentalis has shown different preferences to different shapes of flowers in laboratory tests (Mainali and Lim 2011). Sticky traps with actual flower shapes enhance their attractiveness to thrips Lim 2008, 2010). Visual cues are only effective over relatively short distances and are not affected by air movement (Prokopy and Owens 1983), whereas olfactory cues can drift over a long distance via the wind (Cardé and Willis, 2008). Therefore, it is necessary to understand how olfactory and visual cues interact in the host selection process by herbivorous insects (Pinero Fig Goyret et al. 2007;Wenninger et al. 2009). Such knowledge will be helpful in aiding the management of pest thrips in the field (Vernon and Gillespie 1995;Teulon et al. 1999;Davidson et al. 2007;Mainali and Lim 2011). Host plant species significantly influences the development and survival of T. hawaiiensis (Murai 2001;Zhang et al. 2014;Cao et al. 2018Cao et al. , 2019. In our study, development rate for T. hawaiiensis was the fastest on H. macrophylla, followed by T. gesneriana, then D. caryophyllus and was the slowest on R. rugosa. The survival rates (from high to low) of immature T. hawaiiensis on these flowers were in the same order. Based on these results, we conclude that T. hawaiiensis has faster development and greater survival rates on more attractive and suitable host plants. This phenomenon is also true for other thrips species, such as F. occidentalis, F. intonsa and Megalurothrips usitatus (Bagnall) (Li et al. 2015;Tang et al. 2015). Host plants with higher fitness levels should have better nutritional composition for the thrips, which would be beneficial for thrips development (Brown et al. 2003). The critical nutrients for the development or survival of T. hawaiiensis need further biochemical identification (Mollema and Cole 1996;Brodbeck et al. 2001).
Oviposition is another important factor in the host preference of insects, but we did not examine oviposition in this study. Characterizing oviposition and fecundity of T. hawaiiensis on different host plants would aid in interpreting the results of choice tests with T. hawaiiensis. In choice tests, Frankliniella fusca Hinds and. F. occidentalis have greater oviposition on more suitable larval hosts (Chaisuekul and Riley 2005). Therefore, the influence of the host plant species on the physiology of thrips should also be studied to explore their physiological adaptation to different plants (Veenstra et al. 1995;Kojima et al. 2010;Liu et al. 2017). By integrating such approaches, we will better understand the mechanisms of host plant selection, and how they result from behavioral and physiological responses to various plants. Then we will be able to take effective measures for improved and more sustainable management of T. hawaiiensis on preferred horticultural plants.
We did not determine the influence of pollen on the population development of thrips in this study. Previous studies have shown that pollen can affect the population development of F. occidentalis (Zhi et al. 2005;Riley et al. 2011), and that F. occidentalis shows olfactory preferences for volatiles from pine pollen (Abdullah et al. 2014). Thus, further research is required to determine whether the preferences of T. hawaiiensis for host plant flowers are related to the pollen. We note that other thrips species, for example, F. occidentalis, F. intonsa, and Thrips tabaci, were also detected in our study, and one or more of them coexisted with T. hawaiiensis in certain plant flowers. Therefore, the population performance of T. hawaiiensis may be influenced by interspecific competition with, or displacement by, other thrip species (Atakan and Uygur 2005, Paini et al. 2008, Northfield et al. 2011. These conditions should be taken into consideration to better assess and understand the host preferences of T. hawaiiensis.

Author contribution statement
YC, YLG and CL conceived and designed the research. YC, WC, LJW, and SYY conducted the experiments. YC and YLG analyzed the data. YC, SRR, GSG and WQZ wrote the manuscript. All of the authors read and approved the manuscript.