In Europe and the United States, the environmen­tal cost associated with the use of chemical pesticides is considered too high. Thus, there is a general movement towards environmentally safer control and production ( Dent, 2000 ). This has been transcribed into the European Union's legislative policies with the aim of reducing the use of pesticides, removing large quantities of products, and giving renewed importance to Integrated Pest Management (IPM) ( Lefebvre et al. , 2015 ). The adoption of IPM in all member states in 2014 is the main pillar of the EU strategy to mitigate the negative impact of rapid removal of chemical pesticides from food production ( Clark & Hillocks, 2014 ).

According to the Food and Agriculture Organization ( FAO, 1966 ), IPM means the careful consideration of all available pest control techniques and subsequent integration of appropriate measures that discourage the development of pest populations and keep pesticides and other interventions to levels that are economically justified and reduce or minimize risks to human health and the environment. IPM emphasizes the growth of a healthy crop with the least possible disruption to agroecosystems and encourages natural pest control mechanisms.

The South American tomato moth Tuta absoluta (Meyrick) (Lepidoptera: Gelechiidae) is one of the main tomato pests in South American countries ( Guedes & Picanço, 2012 ). In addition, since its accidental introduction in 2006, this species has become a pest of great economic importance in the countries of the Mediterranean area and many others of Europe and Asia ( Desneux et al. , 2011 ; Campos et al. , 2017 ; Biondi et al. , 2018 ). In Spain, Tu. absoluta rapidly became a serious impediment to biological control programmes in tomato production greenhouses, requiring applications of more than 15 different insec­ticidal substances directed specifically towards Tu. absoluta ( Desneux et al. , 2011 ). The damage is caused by larval feeding mainly on leaves and fruits, but the pest can also attack stems, buds and flowers causing severe crop losses that can reach 100% if no control measures are taken. The dynamics of populations of Tu. absoluta and their consequent damage differ depending on their presence in greenhouse or outdoor crops, date of transplantation, etc., creating signifi­cant challenges for the development and successful application of biocontrol methods ( Cabello, 2009 ), and the results are not always satisfactory due to the overlap of Tu. absoluta generations and the continuous re-infestations in the crops, which motivates the need for the application of several treatments per crop cycle to adapt the population levels to the capacity of con­trol by natural enemies.

At present, pest control of Tu. absoluta is based on biological control, chemical control or a combination of both, although the most common method of control is based on the intensive use of insecticides and this constitutes the first tool in newly invaded areas ( Bielza, 2010 ; Campos et al. , 2017 ; Biondi et al. , 2018 ).

Parasitic insects of the Trichogramma genus have been widely used during the 20th century to control lepidopteran pests in maize and sugarcane crops, and subsequently extended to control many other pests in many different crops; the list of crops continues to increase ( Smith, 1996 ; van Lenteren et al. , 2018 ). Trichogramma , with approximately 200 species des­cribed, is the best-known genus in the family due to its use in the biological control of pest species in agriculture of which more than 25 species are used in pest control in 34 crops across 30 countries ( Pinto & Stouthamer, 1994 ; Querino et al. , 2010 ; van Lenteren et al. , 2018 ). Trichogrammatids can therefore play a vital role in pest control programmes by destroying the first developmental state (egg) of pest, limiting the use of pesticides and contributing to the prevention of environmental contamination ( Kumar et al. , 2013 ).

The establishment and subsequent commercia­li­zation of the parasitoid Trichogramma achaeae Na­garaja and Nagakartti (Hymenoptera: Trichogram­matoidea) has been an important advance in the control of the pest in Spain ( Cabello et al. , 2009 , 2012 ; Vila & Cabello, 2014 ). In addition, T. achaeae has been or is being used in Europe in the following countries: Germany, Belgium, Spain, France, Greece, the Ne­therlands, Romania, and Portugal, against several species of Lepidoptera in more than 15 crops both horticultural and ornamental ( Leppla et al. , 2017 ; Vila, 2017 pers. com. ; van Lenteren et al. , 2018 ).

Studying the side effects of insecticides on natural enemies is necessary to minimize any adverse impacts within the IPM programmes ( Goulart et al. , 2012 ). The integration of biological and chemical control tactics requires a thorough understanding of how pesticides affect biological control organisms ( Brunner et al. , 2001 ). Prior to the release of Trichogrammatids in an IPM system, it is essential to know their compatibility with other pest control methods, including the use of chemical pesticides. Such information will assist in the timing of parasitoid releases regarding the applica­tion of chemical pesticides ( Jalali et al. , 2016 ).

Currently, the strategy used in IPM programmes in the control of Tu. absoluta in Spanish southeast tomato greenhouses consists of the early inoculation of the omnivorous predator Nesidiocoris tenuis (Reuter) (Hemiptera: Miridae) in combination with inundative or inoculative releases of T. achaeae ( Cabello et al. , 2009 , 2012 ; Desneux et al. , 2011 ; Vila & Cabello, 2014 ).

The objective of this work was to establish the side effects of 10 insecticides commonly used in the chemical control of Tu. absoluta on the parasitoid T. achaeae . The selected materials are abamectin, azadirachtin, Bacillus thurigiensis , emamectin, flubendiamide, indo­xacarb, methomyl, chlorantraniliprole, spinosad and spiromesifen. All are included in the list of substances authorized in the South zone of Annex I of the EC Regulation ( EC, 2009 ) that covers the substances allowed in the members states of the EU.

In Europe, T. achaeae has been shown to be a suitable biological control agent against Tu. absoluta , as mentioned above. However, the control of this pest is difficult to manage alone, with the simultaneous use of natural enemies or insecticides being necessary for a satisfactory pest control ( Campos et al. , 2017 ). This is more pronounced at present; thus, the incidence of the pest has increased in tomato crops in Spain and Europe in recent years, especially in greenhouse crops (Vila, 2018, pers. com. ), possibly motivated by problems of resistance to the AIs currently used in these crops. Thus, Roditakis et al. (2018) cited several cases of resistance to emamectin, spinosad, indoxacarb, and chlorantraniliprole in that geographic area. In this sense, it has been mentioned that multiple sublethal effects, sometimes counterintuitive ones, on natural enemies have been reported for modern slower-acting insecticides and/or biopesticides. This highlights the need to revise the labelling of these products to indicate their compatibility with sustainable IPM programmes ( Biondi et al. , 2018 ).

In our work we have shown a broad picture of the side effects of the AIs used in the chemical control of Tu. absoluta on the parasitoid T. achaeae .

First, the group of AIs, abamectin, B. thurigiensis , flubendiamide, indoxacarb, and spiromesifen did not present side effects, or these were negligible, on T. achaeae for all the trials carried out under laboratory or greenhouse conditions (IOBC class 1).

It should be noted that in other studies with the AI abamectin, discrepant results have been found for other Trichogramma species. Thus, Brunner et al. (2001) found a high mortality of adults (56-60%) for T. platneri Nagakartti, when they were exposed to leaf residues less than 3 days old. Additionally, Carvalho et al. (2003) found side effects on adult emergence and the longevity and fecundity of adults when this AI was applied on the pupal stage. This was corroborated by Moura et al. (2006) for the same species and AI. On the other hand, Consoli et al. (1998) and Nornberg et al. (2009) reported that this AI had no side effects for T. pretiosum Riley when it was applied in the protected life stage (pupa). Our results seem to be intermediate those previously mentioned; thus, side effects were observed in female fecundity ( E = 16.27%, IOBC class 1) (Fig. 2), parasitism ( E = 16.79%, IOBC class 1) (Table 3) (both IOBC class 1); and adult activity in the greenhouse trial ( E = 5.42%, IOBC class 1) (Table 4). Perhaps the differences noted above may be due to different degrees of susceptibility; this has been reported for different populations of T. pretiosum by Vianna et al. (2009) . For this AI, we can highlight what was indicated by Gentz et al. (2010) that despite significant toxicity to several non-target species, abamectin was once considered suitable for use with many beneficial insects due to its short environmental persistence.

In relation to the AI B. thurigiensis , several authors agree that side effects have not been found in T. achaeae ( Saelices et al. , 2012 ; Fontes et al. , 2018 ), and in other species of the same genus: T. dendrolimi (Matsumura), T. pretiosum, T. bourarachae Pintureaeu and Babault, T. cacoeciae Marchal, and T. evanescens Westwood ( Takada et al. , 2001 ; Vianna et al. , 2009 ; Ksentini et al. , 2010 ). Our results (Fig. 2; Tables 3 and 4) corroborate the above.

Our results for the AI flubendiamide indicate that there were significant side effects on female longevity ( E = 23.97%) and female fecundity ( E = 24.59%), when the application was made on the pupal stage (Fig. 2), as well as in the exposed phase of the parasitoid (adult) ( E = 20.69%); all of them were within IOBC class 1. However, in the greenhouse trial, these side effects were lower ( E = 6.5%, IOBC class 1) (Table 4). In addition, the compatibility between the AI flubendiamide and T. achaeae found in this work corroborates the results found with other species of the same genus, such as T. chilonis Ishi and T. pretiosum . Thus, side effects were not found on adults ( Sattar et al. , 2011 ; Martins et al. , 2011 ) or on the developmental stages of T. pretiosum ( Carvalho et al. , 2005 ). The same results have been reported for T. atopovirilia Oatman and Platner ( Rezende et al. , 2005 ), that is, in the latter case with a different methodology from the one used in the present work.

Similar effects, to those previously indicated for the AI flubendiamide, were found in our work for the AI indoxacarb, both in laboratory and greenhouse trials (Fig 2, Table 4). The values found were less than a 30% reduction in pupal survival ( E = 16.38%), female longevity ( E = 9.15%), female fecundity ( E = 23.53%) (Fig. 2), and parasitism ( E = 15.75%) with respect to the control for laboratory trials (Table 3). In the greenhouse trial, the AI indoxacarb presented ( E = 8.55%) harmless side effects, grouping into IOBC class 1. Similar side effects have been found by Scholz & Zalucki (2000) for T. pretiosum and Hewa-Kapuge et al. (2003) for T. sp. nr brassicae under laboratory and field conditions. Only, Sattar et al. (2011) found a slightly harmful effect (IOBC class 2) on the adult emergence and female fecundity of T. chilonis.

Finally, in relation to the first five AIs indicated at the beginning of this discussion, the AI spiromesifen did not present harmful effects on the biological parameters analysed in the laboratory trial (Tables 2 and 3). The same result was obtained in the greenhouse trial (Table 4). This AI is highly compatible with T. achaeae . The same results have been cited by Kavitha et al. (2006) for T. chilonis .

Second, there is another group of two AIs that presented slightly harmful side effects (IOBC class 2): azadirachtin and chlorantraniliprole, in laboratory trials, but that, under greenhouse conditions, showed no side effects (IOBC class 1-harmless).

Additionally, the AI azadirachtin had side effects on pupal survival ( E =20.77%) (IOBC class 1) (Fig. 2). Similar side effects on adult emergence have been ci­ted for T. cacoeciae for this AI ( Saber et al. , 2004 ). It should also be noted that this AI presented side effects on female fecundity ( E = 31.55%) (IOBC class 2) (Fig. 2). However, in the greenhouse trial no side effects were found on parasitoid activity (IOBC class 1) (Table 4).

The AI chlorantraniliprole, in tests carried out with predatory species, presents different degrees of toxicity; from very high in some species of Coccinelids and Chrysopids, to no side effects in other species ( Stanley & Preetha, 2016 ). In relation to the species of the genus Trichogramma , the first studies have indicated that this AI is safe for T. chilonis , T. galloi (Zucchi), and T. pretiosum ( Preetha et al. , 2009 ; Brugger et al. , 2010 ; Oliveira et al. , 2013 ). Additionally, it does not affect the emergence of T. chilonis and T. pretiosum adults ( Brugger et al. , 2010 ). However, our laboratory data differ from those found in these species; thus, pupal survival ( E = 33.7%) (IOBC Class 2) and, to a lesser extent, female fertility ( E = 12.40%) (IOBC class 1) were affected by this AI (Fig. 2). Our results in relation to T. achaeae female fertility agree with those cited by Fontes et al. (2018) . Despite these effects in the laboratory, no side effects were found under greenhouse conditions (IOBC class 1) (Table 4).

For the two insecticide groups discussed above, abamectin, azadirachtin, B. thuringiensis , chlorantra­ni­liprole, flubendiamide, indoxacarb, and spiromesifen, can be considered very compatible with the use of the parasitoid T. achaeae .

On the other hand, there is a third group of AIs: emamectin, methomyl, and spinosad, which presented side effects both in the laboratory and in greenhouse trials.

The AI emamectin showed side effects on the lon­gevity of females and males (IOBC class 2) and, to a lesser extent, in the survival of pupae and fecundity of females (IOBC class 1) in the laboratory tests (Fig. 2). The decrease in fecundity of females is lower than that cited for the same species and AI by Fontes et al. (2018) (IOBC class 2). Additionally, the reduction of the values of parasitism when the females were exposed to the fresh residue of the AI (IOBC class 2) (Table 3) was lower than that cited, also for the same species and AI, by Saelices et al. (2012) . Similar results have been cited for this AI in relation to the Trichogramma species T. chilonis and T. sp. nr brassicae ( Hewa-Kapuge et al. , 2003 ; Sattar et al. , 2011 ). The detrimental effects of the AI emamectin in T. achaeae are also shown in greenhouse trials (IOBC class 2) (Table 4).

The AI methomyl has shown an important side effect on T. achaeae pupae. It represented a reduction in adult emergence ( E = 41.92%) (IOBC class 2-slightly harmful) (Fig. 2). In contrast, there were no such effects on adult longevity and female fecundity (IOBC class 1) (Fig. 2). In turn, this insecticide shows an important side effect on parasitism ( E = 80.29%) (IOBC class 3-moderately harmful) (Table 3). This value is very similar to that found by Fontes et al. (2018) for the same species. Similar results have been reported for other species of Trichogramma ( Bull & House, 1983 ; Hassan et al. , 1987 ; Scholz & Zalucki, 2000 ; Takada et al. , 2001 ; Bueno et al. , 2008 ). In the greenhouse trial, it was also shown to be moderately harmful (IOBC class 3) (Table 4). This corroborates the results found by Tipping & Burbutis (1983) and Campbell et al. (1991) for T. nubilale Ertle and Davis, T. exiguum , T. minutum and T. pretiosum .

Finally, the AI spinosad had significant side effects for T. achaeae . This AI decreased pupal survival ( E = 79.16%) (IOBC class 3-moderately harmful) (Fig. 2) and parasitism ( E = 80.29%) (IOBC class 3-moderately harmful) (Fig. 2, Table 3). These results corroborate those found by Fontes et al. (2018) for this AI in the same species. At the same time, the toxicity of the AI spinosad in other species of Trichogramma has been studied by several authors ( Suh et al. , 2000 ; Consoli et al. , 2001 ; Maia et al. , 2010 ; Liu & Zhang, 2012 ; Saljoqi et al. , 2012 ; Costa et al. , 2014 ), who have cited side effects in some of the biological parameters of Trichogramma spp. In our work, the AI spinosad also had side effects for T. achaeae in the greenhouse trial ( E = 51.97) (IOBC class 3-moderately harmful) (Table 4). This insecticide has traditionally been included in IPM programmes because of its low toxicity to mammals and birds, its slight or moderate toxicity to aquatic organisms, and its relative harmless effect for a wide range of natural enemies ( Gentz et al. , 2010 ). However, William et al. (2003) have cited that this AI shows a significant side effect on the Hymenopteran pa­rasitoid complex, both in the field and laboratory trials.

According to the values indicated above for this third group of AIs: emamectin, methomyl, and spinosad, we must mention that their use is not compatible with the release of T. achaeae in tomato crops.

Recently it has been mentioned that pesticide risk assessments for entomophagous species are being per­formed by categorizing pesticides based on mortality in laboratory and semi-field trials and reduction in field studies. Testing the pesticides under field recommended concentrations at laboratory conditions does not exactly reveal how the pesticides behave in complex field conditions ( Stanley & Preetha, 2016 ). This last point has also been previously indicated by other authors ( e.g. , Stark et al. , 1995 ).

Therefore, in the present work the tests were carried out in laboratory and field conditions indicated by the IOBC WPRS methodology, without ruling out any AI in the different stages of the sequential testing scheme methodology.

In this sense, we must indicate, on the one hand, that a very good correlation has been found between the results of laboratory tests and those carried out under greenhouse conditions. Thus, the AIs emamectin, methomyl, and spinosad presented the highest side effects, and in more tests, under laboratory conditions, they also had the highest side effects under greenhouse conditions.

On the other hand, based on the results found, we con­sider that the new methodology used in the evaluation of adult activity of T. achaeae could be more feasible under field conditions and provide reliable results to evaluate the secondary effects these conditions. This is compared to the methodology recommended by the IOBC WPRS ( Hassan, 1985 ; EPPO, 1999 ) for field tests for the evaluation of parasitism in sentinel eggs and those that could be extended to other species of Trichogramma .

Based on the results found in this work, we can conclude that there is an important group of insecticide formulations, especially those of the new generation, which present a high degree of compatibility with the use of the egg parasitoid T. achaeae . This allows a better adaptation of the use of both control systems of Tu. absoluta in tomato, both in greenhouses and open-air crops. This aligns with the recommendations indicated by Campos et al. (2017) for better control of this pest species.