Introduction

Polyploidy is one of the most important mechanisms of evolution, and might be responsible for wide genotypic variation in plants. Two kinds of polyploidy are recognized: autopolyploidy (possess more than two sets of chromosomes through intraspecific genome duplication) and allopolyploidy (possess more than two sets of chromosomes through inter-specific hybridization and subsequent genome duplication) (Soltis and Soltis 2009). Polyploidy can be advantageous for plants by conferring resistance to attack by herbivores and pathogens (Janz and Thompson 2002; Nuismer and Thompson 2001), and may result in release from natural enemies (Broz et al. 2009; Keane and Crawley 2002) and subsequent proliferation of exotic plants (Hufbauer and Sforza 2008; Pandit et al. 2011; Saad et al. 2011; Thébault et al. 2011). Alternatively, allopolyploidy could be disadvantageous (Janz and Thompson 2002) when the hybrid serves as a bridge between parent species, expanding the host range of herbivores (Pandit et al. 2011; Saad et al. 2011; Schlaepfer et al. 2010; Thébault et al. 2011). Nevertheless, most of the studies on phytophagous insects and plant polyploidy are concentrated on autopolyploidy (Arvanitis et al. 2010; Boalt et al. 2010; Nuismer and Thompson 2001; te Beest et al. 2011) but several of the well-known invasive plants are allopolyploids (e.g. Fallopia spp., Pilosella spp., Alternanthera spp.) (Bzdega et al. 2016; Šingliarová et al. 2011; Sosa et al. 2008).

The invasive amphibious plant, alligator weed, Alternanthera philoxeroides (Martius) Grisebach (Amaranthaceae), is an allopolyploid, native to South America. In its native range in Argentina the plant possess two cytotypes, tetraploids (4X = ca. 68) in cold and temperate regions and hexaploids (6X = ca. 102) in warm regions, where it is suggested that this polyploidy could affect the physiology and invasiveness of the plant (Sosa et al. 2008; Zhang et al. 2019). Alligator weed hexaploids adapt better than tetraploids to different environments and, therefore, are more invasive than tetraploids (Okada 1985), so they would be expected to invest more of their resources in the growth and production of biomass than in the production of resistance structures (Pan et al. 2012). In the exotic range two cytotypes have been recorded so far: pentaploids/aneuploids (USA) and hexaploids (China) (Chen et al. 2015) with high phenotypic plasticity and epigenetic variability (Gao et al. 2010; Geng et al. 2013, 2016).

In the exotic range (USA, China, Australia) alligator weed invades aquatic and terrestrial habitats where different management strategies are utilized, including biological control, although with varying success. The biological control agent that is used most is Agasicles hygrophila Selman and Vogt (Chrysomelidae), from Argentina (Winston et al. 2014). It successfully controls the weed in aquatic habitats with mild winters (Julien et al. 2012; Stewart et al. 1999) whereas, in cooler and terrestrial situations, its success is restricted. Different explanations have been suggested to understand the varying success of biological control, including mineral deficiencies (Maddox and Rhyne 1975), low temperatures (Guo et al. 2012; Julien et al. 1995; Stewart et al. 1999) and restrictions to insect development due to pupation failure on stems with a smaller diameter and variability of host plant populations, including genetic and cytogenetic variation (Lu et al. 2010; Pan et al. 2013; Telesnicki et al. 2011). In relation to the genetic variation, it is well known that the efficacy of biological control agents as indicated by preference and performance, often varies amongst weed genotypes (Goolsby et al. 2013; Manrique et al. 2008; Paterson et al. 2012). Therefore, the genotype of host plant is an important factor that needs to be considered in this system to maximize the likelihood of A. hygrophila controlling alligator weed.

Biological control practitioners need to understand factors that govern host plant selection and resulting agent fitness and efficacy. The preference–performance hypothesis states that females choose a host plant (or plant genotype) to lay eggs on those selected host plants that may confer benefits to their progeny in terms of nutrition, survival, and increased maternal fecundity (Refsnider and Janzen 2010). Another alternative proposed to explain female oviposition choice is maximizing maternal survival (Refsnider and Janzen 2010; Wise et al. 2008). In this context, we studied how A. philoxeroides ploidy level affects the oviposition preference of A. hygrophila and how this influenced female feeding preference, and egg and larval survival. Addressing this would help to improve biological control worldwide through the understanding of host plant interactions and the variation in efficacy of the biological control agent. As alligator weed hexaploids are more successful invaders than tetraploids (Okada 1985), it is expected that most of its resources are spent on the growth and production of biomass rather than on the production of resistance structures (Pan et al. 2012). Consequently, we hypothesized that females of A. hygrophila would show preference for the alligator weed genotype with the weakest deterrent mechanisms (i.e. hexaploids) to ensure the highest egg and larval survival. It is expected that this genotype would also be preferred for feeding.

Materials and methods

Insects and plants

Agasicles hygrophila is a flea beetle that has three larval instars. Larvae and adults feed on alligator weed leaves. Pupation occurs inside the stem, the last stage larva makes a hole in the internode of the plant and regurgitates the ingested stem to cover the entrance of the hole. However, under laboratory conditions, this insect may pupate outside the plant (Telesnicki et al. 2011). Five days after emergence, the female is ready to oviposit a clutch of around 32 eggs, in the shape of elongated ovals (1.25 mm length × 0.38 mm diameter), in parallel rows on the leaves of A. philoxeroides (Maddox 1968). Insects used for this study came from a laboratory colony originating from field collections in sites near Hurlingham city, Buenos Aires Province, Argentina. They were fed with plants of the same locality (hexaploid), and were maintained for four generations in rearing chambers (25 °C, 80–90% RH) before starting the experiments.

Plants came from four geographically distant locations in the native range, with different morphological characteristics that suggest genetic variability (Sosa et al. 2008). Plant cytotypes were determined by counting chromosomes using regular cytogenetic techniques and flow cytometry (Singh 2016). Two hexaploids and two tetraploids populations were chosen: (1) Hexaploids—HB (Hurlingham, Buenos Aires Province 34.59° S, 58.64° W) and RS (Reconquista, Santa Fe Province 29.80° S, 59.82° W); (2) Tetraploids—BB (Ruta Nacional 226.7 km North of Balcarce, Buenos Aires Province 37.80° S, 58.25° W) and MB (Ruta Nacional 2.9 km North of Maipú, Buenos Aires Province 36, 80° S, 57.81° O) (Fig. 1). Plants were grown in a greenhouse for two months, after which 30 pieces of three-node stems were taken from each population and cultured for three months in individual pots (50:50 mixture of peat and sand, and 1 g of slow release fertilizer Osmocote, Argensem SA, Argentina). This procedure was repeated twice to reduce any maternal effects and homogenized variability provided by the original environment (Geng et al. 2016; Xu et al. 2010). Furthermore, we used plants of Alternanthera paronynchioides as a negative control in the experiments.

Fig. 1
figure 1

Populations of alligator weed from Argentina utilized for the experiments. Grey squares indicate tetraploids (4X) and black pentagon hexaploids (6X). BB Balcarce, MB Maipú, HB Hurlingham and RS Reconquista

Full size image

Alligator weed in the exotic range is known to have pentaploid genotypes in USA that are not found in the native range. As we conducted the experiments in the greenhouse outside and not in a quarantine facility, we did not include the exotic pentaploid in our experiment.

Experimental design

Female feeding and oviposition preference

To test female feeding preference we conducted a multiple choice experiment. Newly emerged adult females (N = 15) were randomly selected and placed individually in plastic containers (3 l, 13 cm diameter, 20 cm high). Five plants were placed in each container for seven days: four from the different populations of A. philoxeroides and one A. paronynchioides as a negative control. We replaced the plants daily considering consumption. We photographed leaves before and after the experiment to calculate the leaf surface using ImageJ (Rasband 1997-2005) and estimated feeding consumption per plant as the difference in leaf surface before and after feeding.

After seven days, the motivation level for oviposition was maximized (Singer 2003), so we introduced one male per container. Males were not removed from replicates until the female had died. As A. hygrophila females lay eggs in clutches on alligator weed leaves, we counted the number of egg clutches per plant and number of eggs per clutch. We replaced plants daily if they had at least one clutch of eggs on them.

Egg survival and larval performance

To test if females chose to lay eggs on a host plant (or plant genotype) to confer benefits to their offspring (Refsnider and Janzen 2010), we counted and compared the number of larvae hatched per clutch per plant in the multiple-choice experiment. Furthermore, to determine if larval performance was correlated with female oviposition preference, we followed larval development on A. philoxeroides plants from the different populations corresponding to tetraploids and hexaploids, and plants of A. paronynchioides. We isolated egg clutches from the colony reared in the laboratory. We isolated newly hatched larvae and placed each individual in a plastic container (3 cm in diameter and 5 cm depth) with wet tissue paper, and fed them with leaves of only one population of alligator weed (N = 150, 30 replicates per plant), in a completely randomized design. We recorded and compared daily mortality among treatments.

Statistical analysis

For analysis of the oviposition preference, we used a generalized linear mixed model (GLMM) (Bolker et al. 2009) based on negative binomial distribution with a log link function, considering proportion of egg clutch laid as the response variable, plant population or cytotype as fixed factors and replicate as a random factor. As females laid only two egg clutches in all replicates of A. paronynchioides, this plant was excluded from oviposition preference models. For the analysis of leaf consumption, we used a general linear model (GLM) based on Gaussian distribution with an identity link function. We considered leaf surface consumed (cm2) as the response variable and plant population or cytotype as fixed factors. We modeled variance with the varIdent function per cytotype. We tested the surface of plants as covariate in two models but its contribution was not significant so we excluded it from the analysis. Replicates as random factor had been originally included in a GLMM, but its contribution was not significant so we excluded it from the analysis. A. paronynchioides was not consumed, so it was excluded from feeding models. For egg survival analysis, we used a GLMM based on binomial distribution with a logit link function. We considered proportion of eggs that hatched as the response variable, plant population and cytotype as fixed factors and replicate as a random factor. Only two egg clutches were obtained in all replicates of A. paronynchioides, so this plant was excluded from egg survival models. For survival analysis, we modeled the survival curves depending on the plant population on which larvae were reared (Kaplan and Meier 1958). We compared survival curves with log-rank tests. Analysis were performed with ‘lme4’ (Bates et al. 2015) and ‘survival’ packages (Therneau 2012) of the R 3.6.0 (R Core Team 2019) software with Infostat software interface.

Results

Oviposition preference of A. hygrophila female varied between plant cytotypes (F1,14 = 6.98; p = 0.0193), and amongst plant populations (F3,28 = 7.38; p = 0.0009) (Table 1). The proportion of egg clutches laid by the females was higher on hexaploids than tetraploids. However, plants from Hurlingham (HB) were preferred over plants from Reconquista, whereas there were no clear preferences among tetraploid populations. The plants of A. paronynchioides were not chosen for oviposition (Table 1).

Table 1 Biological parameters measured of Agasicles hygrophila (mean ± SE) to determine adult feeding, egg survival and larval performance by population (Fig. 1) and cytotype
Full size table

The females of A. hygrophila showed different feeding patterns between the cytotypes (F1,7 = 14.63; p = 0.0065) and among different populations of alligator weed (F3,21 = 3.87; p = 0.0238). The hexaploids plants from Hurlingham (HB) and Reconquista (RS) were consumed more than plants from the two tetraploid populations. A. paronynchioides was not consumed.

The proportion of hatched eggs was not affected by the oviposition substrate (cytotypes: F1,30 = 3.701; p = 0.064) or by the different populations (F3,28 = 1.20; p = 0.3291). The proportion of hatched eggs was close to 88.19 ± 2.25% for all cases (Table 1). We recorded about 23.2 ± 0.19 eggs per clutch for all plants.

Larval development of A. hygrophila showed different patterns between cytotypes (\(\chi_{2}^{2}\) = 66.84; p = 0.0001) and among different populations of alligator weed (\(\chi_{4}^{2}\) = 59.98; p < 0.0001) (Fig. 2; Table 1). Larvae fed with plants from Hurlingham (HB) and Reconquista (RS) survived 6.07 ± 0.51 more days on average than those fed with the plants from Balcarce (BB) and Maipú (MB). They reached the adult stage only on hexaploids (21.7%, plants from Hurlingham and Reconquista), whereas on tetraploids (from Maipú and Balcarce) they only reached the third larval instar (33.3%). On A. paronynchioides, all insects died in the first larval instar.

Fig. 2
figure 2

Kaplan–Meier survival curves of Agasicles hygrophila for the different populations of Alternanthera philoxeroides (hexaploids, 6X and tetraploids, 4X) and Alternanthera paronynchioides. Hurlingham population (HB, black dotted line); Reconquista population (RS, black whole line); Balcarce population (BB, dark grey whole line); Maipú population (MB, dark grey dotted line); Alternanthera paronynchioides (AP, light grey dotted line)

Full size image

Discussion

According to our results, A. hygrophila females recognize the different ploidy levels of its host plant, alligator weed plants. Females of A. hygrophila discriminate and preferentially oviposit on hexaploids of alligator weed, which maximises larval performance (measured as larval survival, Fig. 3). This is in accordance with the preference–performance hypothesis or mother knows best hypothesis. This hypothesis is based on the principle that the females will seek to oviposit on host plants that confer high nutritional benefits to their larvae (Pöykkö 2006). The larvae, whose eggs were laid in host plants preferred by the females, are larger (Rausher 1983), have shorter development times (Vacek et al. 1985), have a greater chance of survival (current study) and greater digestive efficiency (Sadeghi and Gilbert 1999). Such decision makes progeny less vulnerable to predation, desiccation or other adversities of the environment. Selective female oviposition can also give the larvae the possibility to sequester chemical defenses (Thompson and Pellmyr 1991) and minimize intraspecific competition on the host plant. Li and Ye (2006) stated that variation in mortality recorded on different genotypes of alligator weed was explained by the variation of internal stem morphology. They stated that high pupal mortality was mainly due to plants with a narrower internal stem, which is characteristic of tetraploids in native range (Sosa et al. 2008). However, in the present study we forced insects to pupate outside the stems, thus the higher larval survival would be related to the leaf characteristics and their nutritional quality rather than their morphological differences. Another possible explanation for the difference in larval survival is that alligator weed populations south of the Buenos Aires Province, that is the tetraploids, could invest most of their resources in the production of resistance structures and defenses (Pan et al. 2012) or produce more constitutive defenses than hexaploid alligator weed populations due to their lower degree of ploidy (Broz et al. 2009).

Fig. 3
figure 3

Relationship between larval survival and oviposition preferences of Agasicles hygrophila for the different populations of Alternanthera philoxeroides and Alternanthera paronynchioides. Hurlingham population (HB, black hexagon); Reconquista population (RS, black hexagon); Balcarce population (BB, grey square); Maipú population (MB, grey square); Alternanthera paronynchioides (AP, empty circle). Linear regression (solid line); 95% confidence bands (dotted line)

Full size image

Insect selection behavior is governed by an array of sensory mechanisms that allow for the discrimination of attractants and repellents, and their relative ratios. In the case of A. hygrophila and alligator weed, it is known that flavonoids can work as attractants (Zhou et al. 1988) but recently Li et al. (2017) demonstrated that these beetles used the ratio of two common plant volatiles, (E)-4,8-dimethyl-1,3,7-nonatriene and (Z)-3-hexenol, to discriminate between its host and non-host plants. The role these compounds, gene composition and epigenetic variation interplay in determining host–plant selection by A. hygrophila, with particular reference to discriminating amongst alligator weed plants with different ploidy levels, is unknown and warrants investigation.

A second hypothesis reviewed by Refsnider and Janzen (2010) is that the female chooses the oviposition site based on her feeding preferences to maximize its performance (Fig. 4). These two hypotheses (larval and adult performance) are not exclusive, and selecting alligator weed plants of a certain ploidy could improve the females and their progeny performance. The experiments demonstrated that females preferentially select hexaploids during oviposition and that female feeding was highest on these populations, providing support for the second hypothesis. However, it must be taken into account that in nature, the distributions of A. hygrophila and the tetraploid populations used in this experiment do not seem to overlap geographically (unpublished data). A. hygrophila may recognize some chemical clues that tetraploids (such as flavonoids) (Zhou et al. 1988) share with hexaploids, yet attack tetraploids less because they do not share a joint evolutionary history. On the other hand, the level of ploidy can affect the concentration of metabolites in the leaves. The greater the degree of ploidy, the more resources plants tend to invest in growth and survival and therefore less resources are directed towards chemical defenses against specialist herbivores (Broz et al. 2009). Another possibility would then be that the preference and increased performance of female A. hygrophila on hexaploid plants is due to lower of constitutive defenses in hexaploids relative to tetraploids.

Fig. 4
figure 4

Relationship between feeding and oviposition preferences of adult female of Agasicles hygrophila for the different populations of Alternanthera philoxeroides and Alternanthera paronynchioides. Hurlingham population (HB, black hexagon); Reconquista population (RS, black hexagon); Balcarce population (BB, grey square); Maipú population (MB, grey square); Alternanthera paronynchioides (AP, empty circle). Linear regression (solid line); 95% confidence bands (dotted line)

Full size image

The current study demonstrates female ovisposition preference–perfomance of A. hygrophila and subsequent larval performance. The observed variation in biological control efficacy using this biological control agent may also be explained by their preference for and performance on different alligator weed genotypes or cytotypes. The current “strain” of A. hygrophila would be expected to control alligator weed in China (also hexaploid), but potentially not in the USA (pentaploid)—albeit no pentaploids were tested in this study. It would be worth testing different strains of A. hygrophila from the native range that are sourced from tetraploid populations to see if similar results could be obtained as per the current study (i.e. preference–performance for plant populations with different ploidy levels), and testing different strains of A. hygrophila against populations of alligator weed from the invaded range (e.g. is the performance of A. hygrophila on Argentine hexaploid populations similar to invasive hexaploid populations in China). Further studies of genetic variation of A. hygrophila between and among native and invasive strains are required, particularly due to low genetic variation of the flea beetle found in the exotic range in China (Ma et al. 2013).

The implications of this study for weed biocontrol are applicable if the fidelity demonstrated by A. hygrophila on native alligator weed populations holds true on exotic weed populations belonging to the same cytotype. Sourcing different strains/genetic entities of A. hygrophila adapted to alligator weed differing in their ploidy levels may be required to increase the efficacy of this biocontrol agent—much like practitioners source agents from genetically-matched and/or climatically-matched source populations (Paterson et al. 2009; Robertson et al. 2008).

New genetic and cytogenetic studies are needed, particularly to compare the native and exotic ranges of alligator weed. It would then be possible to obtain a deeper insight into the relationship between the specializations of a phytophagous insect as a function of the level of ploidy of an invasive plant.