Megastigmus seed wasp damage on native Schinus terebinthifolia drupes in ecological restoration area in Brazil

Megastigmus transvaalensis Hussey (Hymenoptera: Torymidae) parasitizes drupes of Rhus genus plants in Africa and Schinus (Anacardiaceae) in South America. This exotic wasp damages Schinus terebinthifolia Raddi drupes in native forests and ecological restoration areas in Brazil. The objective of the present study was to investigate the precipitation, temperature and relative humidity effects on M. transvaalensis flight activity, and to determine the parasitism rate and sex ratio of this wasp on S. terebinthifolia plants. The study was conducted with yellow sticky traps and S. terebinthifolia drupes collected in an ecological restoration area, from August 2014 to September 2015, in the Sorocaba municipality, São Paulo state, Brazil. Megastigmus transvaalensis populations were negatively correlated with maximum and minimum temperatures and precipitation, with population peaks at the end of May 2015, with 927 insects per evaluation (48.8 adults per trap). The M. transvaalensis sex ratio was higher in the laboratory (0.42) than in the field (0.08). The parasitism rate of S. terebinthifolia drupes by M. transvaalensis ranged from zero to 36.3% under natural environmental conditions. Megastigmus transvaalensis can be monitored with yellow sticky traps. Damage by M. transvaalensis in S. terebinthifolia drupes may decrease the germination of the seeds and the establishment of this plant in native and restoration ecological areas.

In Florida, the Department of Agriculture and Consumer Services classified S. terebinthifolia as very harmful and the USA has banned the sale of this plant 8,11 . The uncontrolled increase of the area occupied by S. terebinthifolia led to studies with the parasitoids of their drupes 15,16 as candidates for the biological control of this plant, but none were effective in the USA 17 .
Megastigmus transvaalensis parasitizes drupes of Rhus angustifolia L. and Rhus laevigata L. (Sapindales: Anacardeaceae) and S. terebinthifolia and Schinus molle L. (Sapindales: Anacardeaceae) in South Africa 19 , Schinus polygamus (Cav.) Cabrera in Chile 20 and S. terebinthifolia in Paraná 21 and São Paulo state, Brazil 22 . The M. transvaalensis embryonic period is four to five days with its larval stage lasting from 20 to 25 d 23 with pupation inside the fruits, where it can remain in diapause for months with its adults emerging during S. terebinthifolia flowering and fruit formation stages 24 .
In Brazil, unlike the USA, S. terebinthifolia is important mainly in riparian forest recovery and dune stability programs and projects 25 , where M. transvaalensis may limit the development of this plant 22 .
Integrated pest management depends on insect population monitoring, mainly peaks and relationships with abiotic factors 26,27 . Capture with traps favors abundance, sampling, population dynamics and monitoring studies of insect pests [28][29][30] .
The objectives of this study were to investigate the precipitation, temperature and relative humidity effects on M. transvaalensis flight activity, and to determine the parasitism rate and sex ratio of this wasp on S. terebinthifolia plants. The correlation matrix presenting the transformed data showed a linear inverse relationship between the variable number of insects (N_Insects) and explanatory variables, minimum (T_Min) (r = −0.50) and maximum (T_Max) (r = −0.59) temperatures, precipitation (r = −0.35) and low correlation with relative humidity (r = 0.15) and drupes damaged (r = 0.08) (Fig. 1). Differences between the independent and dependent variables presented more significant effects from each other. Minimum (P = 0.0013) and maximum (P = 0.0034) temperatures influenced the dependent variables.

Numbers of
The distribution of the database was normal with a value of W = 0.4835 showing that the samples were derived from the same distribution, with non-significant results (P = 0.9438) and indicating a good fit for the model. The database was compatible and could be used to interpret and discuss the variables. The hypothesis that population variability is similar (variance homogeneity), that is, that there are differences between the M. transvaalensis populations, was rejected.
The main component analysis showed a relationship with the response variable and can be evaluated with the PC1 and PC2 axes, since they have an eigenvalue of more than one, 43.93 and 26.32%, respectively, of the data set variation (Fig. 2). The PCA retained all factors with eigenvalues greater than 1 ( Table 1), showing that the variables belong to certain axes. Broken-stick axes 1 and 2 explain most of the variability, with a steep decrease (14.51%) from axis 3. The variability ratio was low for axes 4 (8.72%), 5 (5.93%) and 6 (0.06%) ( Table 1 and Fig. 2). The correlation between the variables and the PC 1 axis was high and inversely proportional to the minimum (r = −0.93) and maximum (r = −0.89) temperatures, moderate for precipitation (r = −0.65) and low for drupes damage (r = −0.13) and relative humidity (r = −0.03) (Fig. 3). The PC1 axis showed that the variable number of insects (N_Insects) and the explanatory minimum (T_min) and maximum (T_max) temperatures and precipitation (Prec_mm) are more strongly related to the number of M. transvaalensis individuals captured. The adjusted coefficient of determination (R 2 adj. = 0.2882; n = 27) showed how the regression analysis line fits the data set (Fig. 4). Low R 2 adj. values do not always mean bad models, since the relationship between the analyzed variables and the normality of the data and the value of F (F = 31.062) should be considered. This is necessary to determine the linear relationship between the dependent and independent (Table 2) variables. The PC 2 axis was not significant (P = 0.8149).
The number of M. transvaalensis individuals captured in the field varied with temperature and precipitation. Population peaks of this insect were higher from May to August 2015, mainly in late May and early June, with inversely proportional correlation with minimum and maximum temperatures (transition from autumn to winter) and precipitation.
Relative humidity did not affect M. transvaalensis population peaks in May or June 2015, that is, did not show any association with the variable response relative humidity (P = 0.1885). The drupe damage explanatory variable showed no relation with the number of insects captured (P = 0.0506).       www.nature.com/scientificreports www.nature.com/scientificreports/ field (0.08) ( Table 3). The number of adults caught in the field traps was similar between collections (F = 1.973, df = 37.15, P = 0.1684), but it varied in the laboratory (F = 7.015, df = 15.09, P = 0.0181) (Fig. 5). Megastigmus transvaalensis males showed phenotypic variation (Fig. 6a-f).  33 . These traps are the most attractive to Hymenoptera insects 34 due to the yellow color resembling the bright leaves preferred by insects of this order as oviposition sites, as they can also be confused by appearance, which concerns the greater nitrogen quantity in the plant sap 34,35 .

Discussion
Megastigmus transvaalensis adult peaks, from early May to September 2015, coincided with minimum and maximum temperatures of around 17 and 23 °C, respectively, similar to the increase in the P. bliteus adult numbers caught in fall with minimum and maximum temperatures 31 . This inverse correlation of the M. transvaalensis populations with temperature can be explained by the direct effect on the development of this insect, as reported for L. invasa, with high survival at lower temperatures 36 . Temperature decreases reduce metabolic rate, embryonic development, larva and pupa stages and affect insect behavior 37 , being pecilothermic organisms that are sensitive to temperature changes and thermal fluctuations 38 .
The lack of impact of relative humidity on the M. transvaalensis population fluctuation suggests that the mean humidity remained stable during the study period, since variations in this parameter affected the fecundity, longevity, parasitism and progeny of Pediobius furvus Gahan, 1928 (Hymenoptera: Eulophidae) 39 in Kenya and Cotesia flavipes Cameron, 1891 (Hymenoptera: Braconidae) in Ethiopia, Africa 40 .
The inverse correlation of M. transvaalensis adult numbers captured with precipitation was similar to that reported for the decrease of Trichogramma (Hymenoptera: Trichogrammatidae) species numbers collected with increasing precipitation in Piracicaba, São Paulo state, Brazil 41 . This may be related to environmental resistance with physical and biological factors reducing insect population growth and climatic variables such as temperature and seasonal rainfall affecting these population structures 42 . The precipitation impact on M. transvaalensis is due to the mechanical control of populations of these insects, regardless of their population density 43 .
The M. transvaalensis population peak coincided with the highest number of plants with drupe (n = 13) of S. terebinthifolia, showing that this may be related to this host development 44 . This is similar to the higher activity of Oomyzus sokolowskii Kurdjumov, 1912 (Hymenoptera: Eulophidae) with increasing host numbers. The absence of viable hosts for parasitism drives the insects to search for longer periods for adequate hosts 45 , that is, the M. transvaalensis population increase depends on food availability in the field.   www.nature.com/scientificreports www.nature.com/scientificreports/ The lack of correlation between M. transvaalensis adult numbers caught and damaged S. terebinthifolia drupes is due to these individuals emerging a few months after oviposition 23 on the mature drupe that remained on the trees. The diapause period of Megastigmus wasps varies depending on the host and the number of adults emerged on food availability 46 . Hymenoptera parasitoids have different strategies to avoid adverse environmental conditions 47 , with variations in the development of immature stages, adult stage duration, male and female maturation and diapause 28 . www.nature.com/scientificreports www.nature.com/scientificreports/ The lack of effect of S. terebinthifolia fruiting on the M. transvaalensis adult numbers collected in the field during the population peak of this insect (May 20, 2015 with 927 insects), when the damage to the drupes collected was lower, reinforces the hypothesis that adults of this insect emerged from the drupes present on the trees or from previous fruits fallen on the ground. Damage in September 2014 may be related to previous flowering and fruiting periods during February to April of the same year, and the estimate of M. transvaalensis damage may have only been based on individuals who emerged from previous fruiting, since this insect presents diapause. However, S. terebinthifolia may also flower from October to December 48 , explaining damage during November and December 2014 through February 2015, with the end of one cycle and the beginning of another.
A M. transvaalensis generation was observed at 12 months at the beginning of the evaluations in 2014, when drupes present on the plants were from previous fruiting. Therefore, there was a single flowering and complete fruiting period, which was observed in 2015 and that can be explained by the flowering and fruiting differing between plants and localities, due to the wide geographic distribution and the morphological characteristics of each individual present 48 . The lack of S. terebinthifolia drupe production in the final evaluation period in the 19 trees selected may have affected the emergence of subsequent M. transvaalensis generations, since this insect may produce more than one generation per year according to flower and fruit production of this plant 24 51 . In South Africa, this wasp damaged S. molle drupes throughout the entire summer rainy season 52 . Periods of greater parasitism and differences between sites may be related to S. terebinthifolia phenology. That is, periods of flowering and fruiting influence food availability and, consequently, the incidence of the wasp. These factors may affect the incidence of M. transvaalensis, similar to reports of significant variation in damage between trees and sites in Sorocaba, São Paulo state, Brazil 22 .
The greater emergence of M. transvaalensis females in the laboratory at 25 ± 2 °C, relative humidity of 60 ± 12% and photoperiod of 12:12 h (day: night) may be due to the controlled conditions. In the field, S. terebinthifolia drupes were naturally exposed, but with the emergence of M. transvaalensis, only when the conditions were favorable to this wasp. Megastigmus transvaalensis females can control their offspring sex at oviposition 53,54 through environmental stimuli such as abiotic factors 55,56 . In addition, the greater number of males, captured on the sticky traps in the field than in the laboratory in all evaluations, can be attributed to arrhenotokous parthenogenesis 57 . The high or low sex ratio may be a response to environmental factors such as temperature and relative humidity with arrhenotokous parthenogenesis, Hymenoptera characteristics and linked to host size and age 58,59 . The sex ratio may vary with temperature as reported for Trichogramma pretiosum Riley, 1879 (Hymenoptera: Trichogrammatidae), with a higher female numbers at temperatures below 30 °C 60 . Megastigmus sp. sex ratio in laboratory was close to 1: 1 between 23 to 31 °C 49 .
Variations in body length, coloration, presence or absence of wing spots and size of the abdomen of M. transvaalensis males is similar to that reported for those emerged from S. polygamus drupes in Chile 20 . Chalcidoidea pigmentation varies between species of this superfamily and between sites and hosts, as reported for Megastigmus dorsalis Fabricius, 1978 (Hymenoptera: Torymidae) in Jordan 61 .

Conclusions
The number of M. transvaalensis individuals was negatively correlated with maximum and minimum temperatures and precipitation, presenting population peaks at the end of May 2015. The parasitism rate by M. transvaalensis in S. terebinthifolia drupes ranged from zero to 36.3%, in field conditions. Megastigmus transvaalensis sex ratio was higher in the laboratory (0.42) than in the field (0.08). Megastigmus transvaalensis can be monitored with yellow sticky traps and males of this exotic wasp presented phenotypic variation in Brazil. transvaalensis was carried out on 19 randomly selected, identified, georeferenced S. terebinthifolia plants with diameter at breast height (DBH) and tree height measured. Megastigmus transvaalensis was collected with traps www.nature.com/scientificreports www.nature.com/scientificreports/ consisting of yellow plastic cards (12 cm long x 10 cm wide) with adhesive on both sides and a capture area of 100 cm 2 each 31 , discounting the area for the card identification. Each trap was installed on an S. terebinthifolia plant, attached to a plastic coated wire and fixed with a string at an approximate height of 2.50 meters.

Material and
Twenty-seven collections were performed approximately every 15 days when the traps were replaced with new ones, wrapped in transparent plastic film to avoid damaging the insects captured and placed into paper bags identified with the tree number and collection date. The traps were sent to the laboratory and stored at 0 °C until the M. transvaalensis adults were counted. Males and females of this wasp on both sides of the traps were counted using a stereoscopic microscope (Leica Microsystems TL3000 Ergo, Wetzlar, Germany) with 10X. The maximum and minimum temperature (°C), relative humidity (%) and precipitation (mm) values were obtained from the Meteorological Database for Education and Research -BDMEP Station N° 83851, Sorocaba, São Paulo state, Brazil (23° 29′ S and 47° 26′ W and 645 m altitude) of the National Institute of Meteorology (INMET). Readings were taken daily and the average maximum and minimum temperature, average relative humidity and accumulated precipitation, by evaluation date, were used.
Schinus terebinthifolia drupe sampling. Branches with S. terebinthifolia drupes were collected at random in the middle third of the 19 plants (three branches per tree) with a pruning shear connected to a 3-meter long aluminum pole. These branches were collected, packed into paper bags when the yellow traps were collected and replaced, taken to the laboratory and stored in a Biochemical Oxygen Demand (BOD) incubator at 18 ± 2 °C to reduce the drying rate of leaves and drupes.
One hundred S. terebinthifolia drupes were chosen at random from three branches per sample point (tree) to calculate the M. transvaalensis parasitism rate. The total number of vesicular outflow holes (after the emergence of M. transvaalensis adults) was also counted under a stereoscopic microscope with 10X.
Megastigmus transvaalensis sex ratio in laboratory and field. Twenty S. terebinthifolia drupes were randomly chosen from three branches/tree/evaluation and placed into 1,300 ml plastic containers with a lid, labeled and stored in BOD at 25 ± 2 °C, relative humidity of 60 ± 12% and photoperiod of 12:12 h (day: night) in the laboratory. Megastigmus transvaalensis males and females that emerged after approximately 15 days of incubation were counted, with females identified by the presence of the ovipositor at the extremity of the abdomen 23 . The M. transvaalensis sex ratio, sampled in the S. terebinthifolia drupes and those caught in the sticky traps were calculated.

Data analysis. Megastigmus transvaalensis adult number variation.
The number of insects captured in the sticky traps was subjected to multiple linear regression analysis 64 using the meteorological variables data and the number of drupes damaged.
Statistical analyses were performed with the R Studio ® program at a significance level of 5% in variance analysis (ANOVA) and input of the variables in the regression model. The variation coefficient was tested between trees to determine damage levels between them. The data were standardized for base 10 logarithm to reduce/ equalize the value range and aid in the interpretation. This transformation was performed to evaluate the variance homogeneity and data distribution normality. A correlation matrix between the variables was generated and the data submitted to Tukey range test 65 . Shapiro-Wilk (W) normality test was used to verify if the dataset had normal distribution 66 , if there were differences between the means and if the factors could influence the dependent variable. The homogeneity of the variances was evaluated with the Levene test 67 .
The relationship between the data set variables and the trend visualization of the set, that is, the reduction or elimination of the number of variables, was verified by main component analysis (PCA). The eigenvalues, associated with a component or factor in descending order versus the number of the component or factor, were displayed in a scree plot graph of the data set with the broken-stick model 68 .
The multiple linear regression analysis was performed with the scores obtained from the PC1 and PC2 axes, adopting the model: Yi = β0 + β1xi + €i, to i = 1, … n.
The explanatory variables with the greatest contribution or predictive power were calculated with the significant axis for regression analysis PC1 (P = 0.0052) and the adjustment statistics of the model.
Schinus terebinthifolia drupe parasitism rate by Megastigmus transvaalensis. The S. terebinthifolia drupe parasitism percentage was calculated including those presenting M. transvaalensis outlet holes categorized as damaged (DD) and those without such holes as undamaged (ND). The difference between the damaged drupe values per evaluation was verified by ANOVA.
Megastigmus transvaalensis sex ratio. The sex ratio (RS) of M. transvaalensis was calculated (RS = females number ÷ insects number), submitted to variance analysis (ANOVA) and compared using the F test with 5% probability.