Predation and oviposition potential of Brazilian populations of the predatory mite Amblyseius tamatavensis (Acari: Phytoseiidae) on eggs of Bemisia tabaci (Insecta: Hemiptera)

Amblyseius tamatavensis has been reported from over 20 countries around the world, and is also widely distributed in Brazil. It has been reported as potentially useful for the control of Bemisia tabaci, a pest/vector insect of great economic importance worldwide. Some studies have reported some biological differences among phytoseiid populations. The objective of this work was to compare 14 populations of different regions of Brazil (from the states of Alagoas, Goiás, Minas Gerais and São Paulo; straight line distance between northern and southern extremes of 2,560 km) in relation to the predation and oviposition potential when fed with eggs of B. tabaci biotype B. The morphological characterization of these mites showed that the population from Olhos d'Agua das Flores (Alagoas state) and a population from mixed origin maintained in the laboratory for about 5 years were constituted by mites with larger dorsal and ventrianal shield dimensions and presented the highest rates of daily predation (respectively 7.9 and 7.0 eggs/ female) and oviposition (1.2 and 1.1 eggs/ female). A positive and significant relationship was observed between mean population predation and oviposition rates. The results suggest that subsequent efforts should focus on exploring the variability of mites of these two populations in relation to predation and oviposition potential, in order to select a new, more promising population to control B . tabaci biotype B.


Introduction
The whitefly Bemisia tabaci (Gennadius) is a pest of great economic importance attacking a variety of crops, as cotton, beans, soybeans, ornamental plants and vegetables (Brown et al . 1995;de Barro et al. 2011;Lourenção et al. 2015) and transmitting about 128 plant viruses (Hogenhout et al. 2008). The control of this pest/ vector is mainly performed using pesticides. However, due to the high reproductive potential, wide host range (allowing easy migration between crops) and numerous cases of pesticide resistance, this method is not effective enough. Moreover, there is a growing concern about maintaining or improving the quality of agricultural products and environment, searching for less harmful methods for controlling pests, including biological control. In this context, some predatory mites of the family Phytoseiidae have been used for the biological control of B. tabaci (Calvo et al. 2015).
Biological variability is expected to occur among populations of a same species, especially for the most widely distributed ones. Some authors report the impact of intraspecific variability of Phytoseiidae species on biological control efficiency (i.e. Furtado et al. 2007;Ferrero et al. 2010;Domingos et al. 2012). Furthermore, abundant information can be found on intraspecific variability in regard to resistance to pesticides (Thistlewood et al. 1995;Sato et al. 2000;Hardman et al. 2000;Salman et al. 2015), diapause (Hoy 1984;Field and Hoy 1985;Morewood and Gilkeson 1991;van Houten et al. 1995 b), or tolerance to extreme temperatures (Voroshilov 1979) and low humidity (Walzer et al. 2007;Gómez-Moya et al. 2018).
Amblyseius tamatavensis Blommers is a phytoseiid mites that seems to fit type III-b (generalist predators living on glabrous leaves) group defined by McMurtry et al. (2013). Cavalcante et al. (2017) reported this species as a promising natural enemy of B. tabaci. Experimental releases of this predator on caged plants in a screenhouse caused the reduction of the density of B. tabaci on pepper plants by up to 60-80%. It can be rather easily produced in large numbers in laboratory conditions (Massaro et al., 2018) when fed with astigmatine mites, which could facilitate the eventual mass production for augmentative biological control. This species is reported from over 20 countries around the world (Africa, Asia, America and Ocenia). Given this wide world distribution, the origin of this species is difficult to assess. However, as far as we know, it seems that it has not been introduced to Brazil on purpose, where it is reported in 12 states from the northern to southeastern regions of the country (Demite et al. 2018). The hypotheses herein tested is that populations other than that studied by Cavalcante et al. (2017) could better perform for controlling B. tabaci.
Many biological characteristics may affect the performance of potential biological control agents. Their relative importance is expected to vary according to the prevailing abiotic and biotic environmental conditions of the region where the agent is expected to be used. Thus, finding "the best" predator population to control a pest species seems to be a complex task. However, for augmentative biological control under controlled conditions, as in greenhouses, the task is facilitated, by pre-setting limits to many of the variables (as the climatic factors, plant characteristics, etc). In this framework, the characteristics that have been considered in the selection of the most effective predadory phytoseiids are predation rates and oviposition potential (van Houten et al. 1995 a).
The present work aims at comparing populations of A. tamatavensis from different regions of Brazil in relation to their predation and oviposition potential when fed with eggs of B. tabaci biotype B. An evaluation of the relation between the average size of the mites of the distinct populations and each of those factors was conducted. This study is the first step of a long process for development of a population better adapted for the control of that insect.

Material and Methods
The populations considered Aerial parts of glabrous leaf plants (preferred by A. tamatavensis) belonging to different species were collected from May 2015 to May 2016 in Brazilian regions with different eco-climatic characteristics. Mites identified as Amblyseius (with long Z5 setae and shiny dorsal shield) were collected using a stereomicroscope and then transferred to rearing units similar to those described by McMurtry and Scriven (1965). They were fed with pollen of Typha domingensis (Persoon) Steud. and different stages of Thyreophagus cracentiseta Barbosa, Oconnor and Moraes (Cavalcante et al. 2015;Massaro et al. 2016). About 15 days after the establishment of each colony, 5-10 adult females were mounted in Hoyer's medium to identify the species under a phase contrast microscope (Leica, DMLB). In total, 14 colonies were established (Table 1, Figure 1). One of these is of uncertain origin and had been maintained in the laboratory for about 5 years (population 14). The straight line distance between the most extreme sites of collection (Olho d'Água das Flores and Cananeia, sites 1 and 8 respectively) is 2,560 km.

Predation and oviposition rates
These experiments were conducted in a rearing chamber at 30 ± 1°C during the day and 23 ± 1°C at night, 75 ± 10% RH and 12-h daily photoperiod. Forty mated females were randomly collected from each colony and individualized in experimental units. Each unit consisted of an acrylic mini Petri dish (1.5 x 3.0 cm) with the basis covered by a layer of a gypsum and activated charcoal mixture (9v: 1v) (Abbatiello 1965). In each unit, 15 eggs of B. tabaci (up to 24 h of age) were added, all of which (attacked by the predator during the experiment or not) replaced daily. To ensure that the eggs were intact when offered to the predators, they were not detached from the leaves; instead, the leaf substrate next to each egg was cut, and then egg and adjacent leaf substrate were transferred to the experimental unit. The experimental units were kept moist with daily addition of distilled water and sealed with plastic film (Magipac®) to prevent mites from escaping. The study was conducted during 11 days, counting daily the number of eggs of B. tabaci killed and the number of eggs laid by the predator in each unit. The first day of evaluation was not considered in the analysis, due to the possible interference with the previous feeding. As oviposition and predation data did not satisfy the normality (Shapiro-Wilk test) and homoscedasticity (Levene test) assumptions, the non-parametric Kruskal-Wallis test was used to test the statistical difference between treatments ( P < 0.05), followed by Dunn´s test. Statistical analyzes were performed using R program (R Development Core Team 2013).
Finally, linear correlation analyses (Pearson's test) were performed to evaluate possible relationships between predation and oviposition, and mite size (estimated by dorsal shield length). All the analyses were performed using R program (R Development Core Team 2013).

Morphological characterization
Significant differences between populations were observed for few morphological parameters ( Table 2). The populations of Olhos d'Água das Flores and the laboratory population had larger dorsal and ventrianal shield dimensions as well as larger distance between the setae st4. The means of all characters of the different populations were very similar to each other. As differences among populations were minor, no attempt was made at evaluating the different levels of similarities among them.

Predation and oviposition
Significant differences between populations were observed for both mean daily predation and oviposition rates (ratios maximum/minimum = 1.58 and 1.71, respectively) ( Table 3). Daily predation was the highest for the population of Olho d'Água das Flores (7.9 eggs/ female), followed by the laboratory population (7.0 eggs/ female). The daily oviposition of the Olho d'Água das Flores (1.2 eggs/ female) population was significantly higher than for the other populations, but not for that obtained for the laboratory population (1.1 eggs/ female).
A positive and significant correlation was observed between the oviposition rates and the predation rates (y = 0.18x -0.23, where y = number of eggs of A. tamatavensis produced daily, x = number of eggs of B. tabaci consumed daily by predator; P = 0.0004, R 2 = 0.663). No significant correlations between dorsal shield length and predation or oviposition rates were Table 2 Mean lengths (in micrometers ± standard error of the mean) of structures of populations of Amblyseius tamatavensis collected in different Brazilian municipalities (May 2015 to May 2016), and mean lengths given in the original description of the species (Blommers 1974 Table 3 Daily means (± standard error of the mean) of eggs of Bemisia tabaci killed and eggs laid by adult female of populations of Amblyseius tamatavensis collected in different Brazilian municipalities.

Collection sites Predation Oviposition
Olho d'água das Flores 7.9 ± 0.7 a 1.2 ± 0.2 a observed. However, populations with the highest predation and oviposition rates (Table 3) were also those with the highest dorsal shield length (Table 2).

Discussion
Despite some small morphological differences between populations, the populations were considered belong to the same species. The measurements of the 14 populations are compatible to those reported in the original description (Blommers 1974) of A. tamatavensis. The differences herein observed are compatible with the expected intraspecific variation of a continuous structure, suggested by Tixier (2012) to be ca. 20% around the average for the phytoseiids. Significant differences in size of some morphological structures have been reported in the literature for distinct populations of a same phytoseiid species, e.g. Sourassou et al. (2011) for Neoseiulus paspalivorus De Leon, Sourassou et al. (2012) for Neoseiulus baraki Athias-Henriot, Lopes et al. (2018) for Euseius concordis (Chant) and Tixier et al. (2003) for Kampimodromus aberrans (Oudemans). In a short-term laboratory study, Lopes et al. (2018) observed marked morphological differences for colonies of E. concordis maintained under different diet or temperature conditions. Differences in temperature, diet, host plant or other factors could explain the larger size observed for the A. tamatavensis population of the Olho d'Água das Flores, the most isolated population among those evaluated in this study. The same factors could also explain the differences between morphological features reported in the original description of A. tamatavensis and those of the populations considered herein. The values obtained in the present study for predation and oviposition are similar to those reported by Cavalcante et al. (2015) in a study conducted with A. tamatavensis under approximately the same environmental conditions. In that study, daily mean consumption was 8.0 eggs and oviposition was 1.4 eggs each day.
The significant relationship between predation and oviposition seems logical. It is indeed expected that individuals that fed more would also be more prolific. It could also be expected that larger mites could prey more and probably oviposit more (Sabelis 1981). The absence of significant correlation between size and rate of predation, or between size and rate of oviposition (as observed in the present study) seems to contradict that assumption. Yet, it is noteworthy to point out that the only populations that stood out in relation to size (as indicated by dimensions of dorsal shield) in the study were from Olho d'Água das Flores and the laboratory population, and those had the highest predation and oviposition rates; the size of the other populations were not statistically different. However, the difference in size between those two and other populations was small, and size might not be a reliable trait to assess rates of predation or oviposition. However, if size is really determinant, populations of the type locality of the species (Madagascar, with dorsal shield 340 µm long) or from Africa (with dorsal shield on the average 354 µm long; Zannou et al. 2007) would be expected to have similar or better performance than the best Brazilian populations. From the available re-descriptions of this species (cited by Demite et al. 2018), specimens of the A. tamatavensis from other countries (for which measurements were provided) are not distinctly larger than specimens collected in this study.
Regardless of the cause of the highest predation or oviposition rates and focusing on the search for the most promising A. tamatavensis population for the control of B. tabaci, it seems that priority should be given to the population collected in Olho d'Água das Flores or the laboratory populations. It seems that each of them could have sufficient variability for the selection of a new, improved population (with higher potential for predation (mainly) or oviposition) when fed with B. tabaci. In this study, daily ranges of predation and oviposition rates were 5.6-12.0 and 1.1-1.5 for Olho d'Água das Flores population, and 6.2-8.2 and 0.8-1.3 for laboratory population.
Genetic improvement implies the change of genetic characteristics of a species by humans to suit their needs (Hoy 1985). Thus, this process could involve directed and purposive genetic changes to improve the efficacy of natural enemies for pest control (Hoy 1990), given that artificial selection can accelerate events that rarely occur in the field, producing results that would hardly occur naturally.
Selection processes made by humans over time have allowed the production of plant varieties and animal breeds much more adequate to meet consumption needs (Garland and Rose 2009). Studies of this type have been little explored by researchers in the use of predatory mites (Lommen et al. 2017), and the results of this study indicate the great potential of applying to these organisms similar selection processes adopted for many years for plants and animals.