Individuals of T. diversifolia used in experiment were germinated from seeds collected during the summer of 2015 (between October and December) from a population located in an urban area of Uberlândia, state of Minas Gerais, Brazil (18°53’10” S, 48°13’56” W). Seeds were germinated in polyethylene pots (12 x 10 cm; 500mL) containing commercial organic substrate (Bioplant^®), and maintained until the end of the study in a greenhouse located in Uberlândia (18°53’10” S, 48°15’36” W). Local illumination was provided by natural light, and the photoperiod was not controlled. Irrigation was performed homogeneously by an automatic system activated for five minutes four times daily. There were no plant species other than T. diversifolia in the greenhouse during the experiment.

Individuals of Aphis gossypii spontaneously started colonization in the greenhouse two weeks after seeds were planted, at which time 168 seedlings had germinated and all had two or more pairs of expanded leaves. Only a small group of seedlings (~ 20 individuals) that were located next to each other had aphids (~ 10 aphids per individual) visible to the naked eye (i.e., third instar or later stages). After this initial attack, we assessed all the leaves, shoots, soil and pots of the other plants with a handheld magnifying glass to isolate individuals that were not attacked by aphids. A total of 53 seedlings had no aphids present, which were reallocated to a stand 6-m away from the attacked plants. Since aphids of early instars could still be present in these apparently non-attacked plants, no effort was made to establish any type of physical isolation between the plants and the environment (e.g., anti-aphid screen) so that natural enemies could settle on these plants.

Two weeks after colonization the number of aphids increased dramatically on the initially-attacked plants (~ 100 individuals per plant). All individuals of this group hosted apterous A. gossypii of many instars and winged individuals. Winged individuals of this species occur in high population densities, which facilitates dispersal to other plants (Ebert & Cartwright, 1997). Ladybug larvae (Coleoptera: Coccinellidae) and aphids parasitized by braconid wasps (i.e., round form and straw yellow coloration) (Kavallieratos & Lykouressis, 2004) were present on the infested seedlings.

Due to the differences of the number of aphids between the two seedlings groups, we categorized T. diversifolia individuals with high densities of A. gossypii and natural enemies as the “infested group”, and plants with no visible aphids as the “non-infested group”. We kept these plant groups spatially separated until the end of experiment and did not perform any method to interfere with the attack of A. gossypii and other arthropods on the plants.

Approximately four weeks after the beginning of the treatments some infested seedlings started to exhibit signs of mortality (i.e., presented only senescent leaves), at which time the experiment was terminated — 22 days after the initial attack of aphids.

Thirty random seedlings of each of the two groups (60 seedlings in total) were used for collecting data on vegetative variables and the aphids, natural enemies and other insects present.

To determine arthropod abundance, including A. gossypii and its natural enemies, we collected the two most apical and expanded leaves of each seedling and placed them in Petri dishes, which were labeled and sealed with adhesive tape to prevent any arthropods from escaping. The Petri dishes were frozen for 48 hours at -15ºC to kill any arthropods present without damaging morphological structures. Quantification and identification of arthropods was done using a stereomicroscope (Opton TIM-2B; optic zoom: 10–160x). Aphids and parasitoids were identified to species, while ladybugs were identified to the level of family.

We measured above and below ground vegetative structures of seedlings to compare how the two levels of aphid infestation affected the performance of T. diversifolia. We measured shoot length (cm) with a metric tape and counted the number of leaves for each seedling of each group. We scanned (600 dpi) the two previously-collected leaves and measured their area (mm²) using ImageJ software (Schneider, Rasband, & Eliceiri, 2012); seedling leaf area was calculated as the mean of the areas of two leaves. We also dried the aerial and radicular portions of each seedling in a stove (50ºC for 48 hours) to determine shoot and root dry weights (g), which were then used to determine the root-to-shoot ratio.

Normality of the data was assessed by the Kolmogorov-Smirnov test (p < 0.05) and, when necessary, residual homoscedasticity was analyzed graphically. Simple linear regression was applied to determine the relationship between aphid abundance (independent variable) and quantity of natural enemies (dependent variable). Since natural enemies of T. diversifolia were absent from the non-infested plants, the relationship between natural enemies and aphids was analyzed using only the infested group. We used the Student t-test for independent samples to compare foliar area, shoot length, shoot weight, root weight and total weight of infested and not infested groups. Due to the data being non-normal, we used the Mann-Whitney non-parametric test to compare the number of leaves and the root-to-shoot ratio between groups. All analyses were executed in SYSTAT 13 (2009)

The high density of A. gossypii caused infested seedlings to present visually-apparent reduced vegetative performance relative to seedlings of the non-infested group (Figure 1a). Damage from the aphids also caused several necroses along apical leaf blades of infested plants, characterizing foliar senescence (Figure 1b). These impacts were also reflected in approximately 50% higher values for leaf number, leaf area, stem length and stem, root and total plant weight in non-infested than infested T. diversifolia (Table 1).

Figure 1.
Vegetative characteristics of Tithonia diversifolia seedlings on different levels of infestation by Aphis gossypii in greenhouse. (a) Habitus of non-infested seedling (left) and infested seedling (right); and (b) leaf of non-infested individual (left) and leaf of infested seedling (right) with black arrows indicating marks of necrosis.

Table 1
Vegetative performance of T. diversifolia seedlings infested and non-infested by Aphis gossypii. Quantitative values are presented as mean ± standard error (c±EP).

The density of A. gossypii achieved in the present study is likely due to the high nutritional quality and ineffective defense of T. diversifolia. Many invasive plants allocate macronutrients to vegetative structures for growth and reproduction (Gioria & Osborne, 2014). Thus, the nutritional quality of the vegetative organs of this weed (e.g., carbohydrates and nitrogen) can increase developmental and reproductive rates of herbivores, thus leading to increased abundance of A. gossypii (Hosseini et al., 2010).

Aphis gossypii negatively affected vegetative performance of infested seedlings. When feeding, these aphids suck sap from the plant and inject saliva and other compounds into the hosts, thereby altering its physiological processes. As a result, aphid feeding is facilitated and the nutritional quality of the host plant is increased (Michaud, Zhang, & Bain, 2017). Aphids can remove large quantities of macronutrients, such as nitrogen, by sucking sap from host plants (Guerrieri & Digilio, 2008). In addition, they cause degradation of chlorophyll and increase the nitrogen content of leaves due to substances injected with saliva, which increases the nutritional quality of the plant for the aphid (reviewed by Michaud et al., 2017). Thereby, displacement and retention of these nutrients may occur in other vegetative parts of host plants (Goggin, 2007). Aphids can also inject substances in plants that reduce photo-assimilates by concentrating free amino acid content at the site of feeding (reviewed by Michaud et al., 2017). The nutritional manipulation of hosts by aphids can result in cell death (Goggin, 2007), which would explain the marks of necrosis observed on the leaves of T. diversifolia in the present study. Alternatively, individuals of this plant species can induce early leaf senescence as an induced defense to limit resource availability for A. gossypii (Goggin, 2007).

In relation to plant defense, Ambrósio et al. (2008) reported that leaves of T. diversifolia have glandular trichomes with secondary anti-herbivory compounds. Although similar structures inhibit aphids (Guerrieri & Digilio, 2008), they might not prevent the attack of the cotton aphid on some plants (Zarpas, Margaritopoulos, Stathi, & Tsitsipis, 2006; Leite, Picanço, Zanuncio, & Gusmão, 2007). Increased mortality of initial instars (Soglia, Bueno, & Sampaio, 2002) and decreased fecundity (Soglia, Bueno, Rodrigues, & Sampaio, 2003) of A. gossypii were observed to be related to increased density of trichomes of another species of Asteraceae, Dendranthema grandiflora Tzvelev. Therefore, foliar trichomes and chemical defenses of T. diversifolia can reduce herbivory by this aphid, but are insufficient to prevent it. Future studies should evaluate which characteristics of the Mexican sunflower reduce A. gossypii infestation, since other environmental variables, such as temperature (Ebert & Cartwright, 1997; Soglia et al., 2003; Soglia et al., 2002; Zamani, Talebi, Fathipour, & Baniameri, 2006), also influence the population growth of this aphid.

We observed that infestation by A. gossypii can drastically limit the energy allocated by T. diversifolia seedlings to vegetative growth, resulting in low performance. It is worth mentioning, however, that vegetative growth of Mexican sunflower seedlings may have been limited in the present study by the size of the pots in which they were growing, thus enhancing the effects of the aphid attack on the growth variables of the analyzed seedlings. Furthermore, the environmental conditions of the greenhouse may have been favorable for population growth by polyphagous insects with high potential of population growth, such as A. gossypii (Brødsgaard & Albajes, 1999), thereby enhancing the effects of A. gossypii on the vegetative growth of T. diversifolia. Nevertheless, our results strongly evidence the effects of this aphid on the plant.

A total of 28,141 individuals of A. gossypii were present on the collected leaves of infested T. diversifolia, with a mean of 938 ± 61 aphids per seedling (c±EP, n = 30 seedlings). No aphids were found on seedlings of the non-infested group, which reinforces the discrepancy in aphid presence between the two groups of plants and justifies considering them as different levels of infestation.

No organisms other than A. gossypii, ladybugs (Coleoptera: Coccinellidae) and A. gossypii parasitized (mummified) by A. platensis (Braconidae: Aphidiinae) were present on the infested plants. Of these, there were 118 ladybug larvae and 440 individuals of A. platensis parasitizing aphids. The abundance of aphids was positively and significantly related to the number of ladybug larvae (p < 0.001) (Figure 2a) and mummified aphids (p = 0.01) (Figure 2b).

Our results demonstrated that populations of A. gossypii can be regulated by ladybugs and A. platensis. The joint action of these two natural enemies results in high consumption of aphids in a short amount of time, thus characterizing them as good agents for biological control of the cotton aphid in plantations (Ebert & Cartwright, 1997; Kavallieratos, Stathas, & Tomanovic, 2004). In our study, however, the cotton aphid achieved a high population density, even in the presence of these natural enemies. Although the attack rate of the parasitoid A. platensis can be estimated by the number of mummies formed, the experiment was short (only 22 days), and consumption of aphids by the predatory ladybugs was not directly observed. These limitations prevented us from observing the actual efficacy of the biological control of the cotton aphid on T. diversifolia.