Faculteit Wetenschappen
Departement Biologie
Employing Macrolophus pygmaeus as natural enemy against sweet pepper key pests in practice
Macrolophus pygmaeus inzetten als natuurlijke vijand tegen belangrijke paprikaplagen in de praktijk
Proefschrift voorgelegd tot het behalen van de graad van doctor in de Wetenschappen: biologie aan de Universiteit Antwerpen, te verdedigen door
Nathalie Brenard
Promotoren: Antwerpen, 2020
Prof. dr. Herwig Leirs
dr. Rob Moerkens
dr. Vincent Sluydts
2
Table of Contents
Summary ...... 9
Samenvatting ...... 11
Chapter 1 | General introduction ...... 15
1.1 A brief history of pest management in agriculture ...... 15
1.1.1 Ancient farming, ancient pest management ...... 15
1.1.2 The rise of synthetic pesticides ...... 17
1.1.3 Integrated pest management ...... 19
1.1.4 Biological control ...... 20
1.1.5 Pest management in greenhouse crops ...... 24
1.2 Pest management in greenhouse sweet pepper...... 26
1.2.1 Sweet pepper cultivation in Europe...... 26
1.2.2 Sweet pepper key pests and their management ...... 27
1.3 Macrolophus pygmaeus ...... 29
1.3.1 Biology ...... 29
1.3.2 Usage in pest control ...... 31
1.3.3 Potential in sweet pepper ...... 32
1.4 Population ecology in pest management ...... 34
1.4.1 Population ecology ...... 34
1.4.2 Population models: the basics ...... 34
1.4.3 Predator-prey models ...... 36
1.4.4 Examples of population models in pest management ...... 37
1.5 Thesis outline ...... 39
1.5.1 Improving M. pygmaeus release strategy ...... 39
3
1.5.2 Aphid control with M. pygmaeus ...... 40
1.5.3 Predicting pest control with predator-prey models ...... 41
Chapter 2 | Food supplementation to optimize inoculative release of the predatory bug Macrolophus pygmaeus in sweet pepper ...... 45
2.1 Abstract...... 45
2.2 Introduction ...... 46
2.3 Materials and methods ...... 48
2.3.1 Greenhouse location, crop, and climate conditions ...... 48
2.3.2 Release of Macrolophus pygmaeus ...... 50
2.3.3 Supplementary food applications ...... 50
2.3.4 Population dynamics and dispersal ...... 51
2.3.5 Statistical analysis ...... 51
2.4 Results...... 52
2.4.1 Population build-up ...... 53
2.4.2 Dispersal ...... 55
2.5 Discussion ...... 55
2.6 Conclusion ...... 59
2.7 Acknowledgements ...... 59
2.8 Supplementary material ...... 60
Chapter 3 | Biweekly supplementation with Artemia spp. cysts allows efficient population establishment by Macrolophus pygmaeus in sweet pepper ...... 63
3.1 Abstract...... 63
3.2 Introduction ...... 64
3.3 Materials and methods ...... 65
3.3.1 Cage experiment ...... 65
3.3.2 Greenhouse experiment ...... 67
4
3.3.3 Statistical analysis ...... 69
3.4 Results ...... 70
3.4.1 Cage experiment ...... 70
3.4.2 Greenhouse experiment ...... 72
3.5 Discussion ...... 75
3.6 Conclusion ...... 77
3.7 Acknowledgements ...... 78
Chapter 4 | Is leaf pruning the key factor to successful biological control of aphids in sweet pepper? ...... 81
4.1 Abstract ...... 81
4.2 Introduction...... 82
4.3 Material and methods ...... 85
4.3.1 Crop and climate conditions ...... 85
4.3.2 Release of Macrolophus pygmaeus ...... 86
4.3.3 Leaf pruning ...... 86
4.3.4 Effect leaf pruning on aphid control ...... 87
4.3.5 Effect leaf pruning on yield ...... 89
4.3.6 Statistical analyses ...... 89
4.4 Results ...... 90
4.5 Discussion ...... 97
4.6 Conclusion ...... 101
4.7 Acknowledgements ...... 102
4.8 Supplementary material ...... 102
Chapter 5 | Investigating pest control with predator-prey models in sweet pepper: Macrolophus pygmaeus preying on western flower thrips Frankliniella occidentalis ...... 105
5.1 Abstract ...... 105
5
5.2 Introduction ...... 106
5.3 Material and methods ...... 108
5.3.1 Data collection ...... 108
5.3.2 Logistic regression model ...... 112
5.3.3 Lotka-Volterra based models...... 113
5.4 Results...... 117
5.4.1 Logistic regression model ...... 118
5.4.2 Lotka-Volterra based models...... 119
5.5 Discussion ...... 126
5.6 Conclusion ...... 131
5.7 Supplementary material ...... 132
Chapter 6 | General discussion ...... 142
6.1 Establishment of Macrolophus pygmaeus in sweet pepper greenhouses ...... 143
6.2 Biological control of sweet pepper key pests with M. pygmaeus ...... 146
6.3 Monitoring and decision-making in modern pest management ...... 148
6.4 Overall conclusion...... 151
6.5 Supplementary material ...... 152
References ...... 155
Dankwoord ...... 171
6
7
8
Summary
Biological pest management is already quite successful in European sweet pepper greenhouses and many major pests can be controlled by augmentative releases of natural enemies. Aphids are a major pest for which the currently used specialist natural enemies don’t provide sufficient control and chemical interventions are often applied. Organic growers can’t use synthetic insecticides and the number of effective insecticides allowed in organic farming is very limited, therefore they occasionally suffer from severe aphid infestations. Due to their high reproductive rate and short generation time, aphids are able to quickly develop insecticide resistance. Increased attention to the role of generalist natural enemies, which attack different types of prey in contrast to specialists, in pest control has caused researchers to look more towards generalist predatory bugs to control aphids and other pests. Macrolophus pygmaeus is one of these predatory bugs with potential. They are commonly used in European tomato greenhouses against a number of pest species such as whiteflies, aphids, thrips, mites and caterpillars. This PhD research focuses on extending the use of M. pygmaeus to sweet pepper cultivation, with experiments under conditions that resemble commercial practice.
First of all, release strategies of M. pygmaeus in sweet pepper were tested with regard to food supplementation. Not all generalist natural enemies require food supplementation and different food types or supplementation strategies can affect population growth and dispersal in the crop. Supplementing food had a large positive effect and brine shrimp (Artemia spp.) cysts proved to be the best food source for this. A biweekly supplementation in a full field fashion was found to be the best strategy.
After figuring out how best to establish an M. pygmaeus population in the greenhouse, its potential as biological control agent against two sweet pepper key pests, namely aphids and thrips, was studied. The effect of M. pygmaeus on aphid control was combined with leaf pruning at four different heights in order to decrease the vertical foliage length and bring pest and predator closer together. The predatory bugs were found to successfully control aphid infestations in sweet pepper if vertical foliage length of the plants was kept no longer than 190cm.
9
Western flower thrips Frankliniella occidentalis is currently managed in sweet pepper by releasing the predatory bug Orius laevigatus. However, M. pygmaeus also feeds on thrips and may dismiss the need for O. laevigatus releases. During the three seasons of experiments for this PhD, M. pygmaeus was always able to control thrips outbreaks on its own.
In the last part of this PhD, collected times series data of thrips and M. pygmaeus population densities were used to construct predator-prey models that could predict pest control. The fitted logistic regression model was able to predict the chance of pest control one week into the future.
We developed a strategy to release M. pygmaeus in sweet pepper greenhouse where it can control thrips and aphids, though the latter requires some extra work in the form of leaf pruning. Both pests could be successfully controlled without the need for other natural enemy releases or chemical interventions. The developed population model predicts chance of thrips control and can help growers who monitor pest and natural enemy populations in their greenhouse to make better decisions on pest management measures. Together, these findings can bring growers a step closer to pesticide-poor or even pesticide-free sweet pepper cultivation.
10
Samenvatting
Biologische bestrijding van plagen is reeds een succesverhaal in de Europese paprikateelt onder glas en de meeste voorname plagen worden onder controle gehouden door het uitzetten van natuurlijke vijanden. Bladluizen vormen echter nog steeds een groot probleem en de gespecialiseerde bladluispredatoren en -parasitoïden die ertegen uitgezet worden, blijken vaak niet in staat om de plaag onder controle te krijgen waardoor naar chemische middelen wordt gegrepen. Biologische telers mogen veel van deze middelen niet gebruiken en het aantal biologische gewasbeschermingsmiddelen dat effectief is, is beperkt, waardoor telers soms grote verliezen lijden. Bovendien zijn bladluizen door hun hoge voortplantingssnelheid en korte generatietijd in staat om snel resistentie te ontwikkelen. Een grotere aandacht voor de rol van generalisten, predatoren die zich voeden met verschillende soorten prooien, in plaagbestrijding zorgde ervoor dat onderzoekers meer in de richting van roofwantsen gingen zoeken om bladluizen en andere plagen te bestrijden. Macrolophus pygmaeus is één van deze roofwantsen met veel potentieel. Ze worden veelvuldig ingezet in Europese tomatenserres tegen verschillende plagen zoals wittevliegen, bladluizen, tripsen, mijten en rupsen. Dit doctoraatsonderzoek spitste zich toe op het uitbreiden van het gebruik van M. pygmaeus naar de paprikateelt, met experimenten onder omstandigheden die nauw aansluiten bij de praktijk.
Allereerst werden uitzetstrategieën voor M. pygmaeus met betrekking tot bijvoederen getest in paprika. Niet alle generalisten hebben nood aan bijvoederen en verschillende voedseltypes of bijvoederstrategieën kunnen de populatiegroei en verspreiding in het gewas beïnvloeden. Bijvoederen bleek een positief effect te hebben en cysten van pekelkreeftjes (Artemia spp.) kwamen eruit als beste voedselbron. Tweewekelijks vollevelds bijvoederen is de beste strategie.
Nadat was uitgemaakt wat de beste manier is om een M. pygmaeus populatie te vestigen in de serre, werd onderzocht in welke mate de roofwantsen in staat zijn om twee belangrijke plagen in paprika, namelijk bladluizen en tripsen, te bestrijden. Het effect van M. pygmaeus op bladluisbestrijding werd bestudeerd in combinatie met het wegsnijden van bladeren tot op verschillende hoogtes. Dit verkleint de verticale lengte van het gebladerte aan
11
de plant en brengt zo prooi en predator dichter bij elkaar. Als deze verticale lengte korter was dan 190 cm konden de roofwantsen met succes bladluizen bestrijden in paprikaserres.
Californische trips Frankliniella occidentalis wordt momenteel bestreden door het uitzetten van de roofwants Orius laevigatus. Maar M. pygmaeus voedt zich ook met tripsen en kan zo het gebruik van O. laevigatus overbodig maken. Tijdens de drie seizoen van experimenten voor dit doctoraat was M. pygmaeus telkens op zichzelf in staat om trips onder controle te houden.
In het laatste deel van dit doctoraat werden de verzamelde tijdserie data van trips en M. pygmaeus populatiedensiteiten gebruikt om prooi-predator modellen op te stellen die het succes van plaagcontrole kunnen voorspellen. Een logistisch regressiemodel bleek in staat om op basis van monitoringsdata te voorspellen hoe groot de kans was dat de plaag één week later onder controle zou zijn.
In dit doctoraatsonderzoek hebben we een strategie ontwikkeld om M. pygmaeus uit te zetten in paprikaserres, waar deze roofwants tripsen en bladluizen kan bestrijden. Voor die laatste is nog wat extra werk vereist onder de vorm van bladeren wegsnijden. Beide plagen konden met succes bestreden worden zonder dat andere natuurlijke vijanden moesten uitgezet worden en zonder chemische interventies. Het ontwikkelde populatiemodel voorspelt de kans dat een tripsplaag onder controle zal zijn en kan zo telers die plagen en natuurlijke vijanden in hun serre monitoren helpen bij het nemen van beslissingen in zake plaagbestrijding. Tezamen kunnen deze bevindingen telers op weg helpen naar een pesticiden-arme of zelfs pesticiden- vrije paprikateelt.
12
13
14
Chapter 1 | General introduction
1.1 A brief history of pest management in agriculture
1.1.1 Ancient farming, ancient pest management
Ever since mankind started practicing agriculture some 10 000 years ago, pests have formed a continuous challenge for farmers. The list of unwanted organisms –referred to as pests– benefiting from human grown crops is lengthy: large animals such as mice, rabbits, deer and birds; smaller invertebrates such as insects, mites, snails and nematodes; fungi, bacteria, … . All compete with humans for feeding off the crops. In order to guarantee sufficient food production, farmers had to practice pest management to avoid and remove these uninvited guests in their fields and storages from the very beginning (Winston, 1999).
A wide array of pest management techniques and measures have been applied since the first agricultural societies. Examples are mechanical measures, such as trapping and manually removing (infested) plant parts; cultural measures, such as crop rotation, cultivation techniques and selection of resistant cultivars; biological measures such as releasing or facilitating natural enemies of pests; chemical measures which comprise the use of plant protection products and pesticides from different origins; and some religious and superstitious measures, such as praying and bringing offerings to God(s) or taking pests to trial in religious court. Some measures like cultivar selection and crop rotation are taken to prevent pest invasions, while others like trapping and pesticide application are performed to remove established pests (Winston, 1999; Dhaliwal et al., 2004).
The earliest record of pest management comes from the ancient Sumerians around 2500 B.C., who used sulphur compounds to control insect and mite pests. The ancient Chinese civilization provides the first record of biological pest management around 300 A.D., establishing colonies of predatory ants in citrus orchards to protect the trees from beetles and caterpillars (Winston, 1999; Van Lenteren, 2005). Examples of mechanical pest management in ancient times were found in Greece, where locusts were driven into the sea with fire and in Egypt, where drovers lined up to repel locust swarms (Zhang et al., 2008). Other pest management techniques include the use of oils, botanicals, arsenic and ashes by ancient Greek,
15
Chapter 1
Roman, Chinese and Egyptian farmers (Winston, 1999). A notable insecticide used already by the Chinese around the 1st century A.D. is pyrethrin, a botanical extracted from chrysanthemum flowers. Its use expanded to Persia during the Middle Ages and pyrethrins made their way to Europe in the 19th century via Armenian traders (Davies et al., 2007).
Not much scientific progress was made in pest management during the Middle Ages, but interest and research increased again during the scientific and agricultural revolutions and the Renaissance period (Winston, 1999). Agriculture changed drastically around the 17-19th century when it became a commercial practice (Pretty, 1991; Winston, 1999). The use of manure and fertilizers greatly increased, together with an increase in field acreage and the introduction of machines that facilitated all facets of agricultural work (Winston, 1999). This upscaling and intensification of agriculture lead to an increase in pest problems, sometimes with disastrous consequences. One of the most well-known examples is the European Potato Failure in the 1840s, also referred to as the Irish Potato Famine as it hit hardest in Ireland with over a million deaths and as many refugees. The pest at hand was a pathogen called Phytophthora infestans (Peronosporales: Peronosporaceae), which causes potato blight, and was responsible for huge yield losses throughout Europe: from 19% in France, to 43% in Belgium and even 88% in Ireland in 1846 (Vanhaute et al., 2007; Haas et al., 2009; Yoshida et al., 2013).
To combat the increase in pest incidence, research and development of resistant cultivars, biological control agents and botanical and inorganic insecticides increased drastically and with success. Remarkable examples are the grafting of European grape vines on resistant North American rootstocks to fight grape phylloxera Daktulosphaira vitifoliae (Fitch) (Hemiptera: Phylloxeridae), the release of the vedalia beetle Rodolia cardinalis (Mulsant) (Coleoptera: Coccinellidae) in Californian citrus orchards to control cottony cushion scale Icerya purchasi Maskell (Hemiptera: Monophlebidae) and the use of Paris green (mixed copper acetoarsenite) against the Colorado potato beetle Leptinotarsa decemlineata (Say) (Coleoptera: Chrysomelidae) in the 19th century (Dhaliwal et al., 2004; Oberemok et al., 2015).
Towards the end of the 19th and the beginning of the 20th century, several synthetic inorganic insecticides were developed, shifting the focus towards chemical pest management (Dhaliwal et al., 2004).
16
General introduction
The remainder of this introduction will focus on arthropod pest species of agricultural crops as they are the subject of this PhD thesis, though much the same developments and issues apply on the control of other pests.
1.1.2 The rise of synthetic pesticides
The first synthetic insecticides were inorganic compounds containing arsenic, mercury, tin and copper and were highly toxic to all living organisms. Shortly after, synthetic organic (compounds containing carbon-hydrogen-bonds) insecticides were developed as well and their impact would be much larger (Dhaliwal et al., 2004). The discovery that really paved the way for large scale chemical pest management was that of Paul Muller in 1939. He found that dichlorodiphenyltrichloroethane (DDT), synthesised in 1874 by Othmar Tseidler, had strong insecticidal properties. Few or perhaps no other synthetic substances have had such a strong impact on pest control (Davies et al., 2007; Oberemok et al., 2015). At that time, DDT was thought of as the perfect insecticide: it was very toxic to most insect pests and (thought to be) relatively harmless to humans, production was cheap, it could be made in near unlimited quantities and it retained its toxicity for an extensive period. Chlorinated hydrocarbons, such as DDT, formed the first generation of synthetic insecticides and gave rise to a new era of pest management in agriculture: one based on pesticides (Oerke, 2006). Where chlorinated hydrocarbons were mainly used to protect crops from chewing pests, organophosphorus compounds introduced in the 1940s and 1950s and carbamates in the 1960s focused more on sucking insects and other gaps in pest management (Casida & Durkin, 2013). Another important class of synthetic pesticides was introduced during the 1970s: pyrethroids, based on the natural pyrethrins already used by the Chinese almost 2000 years ago. Pyrethroids have a low toxicity to mammals, limited soil persistence and are more photostable and cheaper to produce than the natural pyrethrins (Davies et al., 2007).
Initial enthusiasm about these new relatively cheap and efficient pesticides that controlled all sorts of pests gradually made room for concerns about their impact on the environment and non-target organisms, including humans. DDT is again the most famous example. This substance and other chlorinated hydrocarbons turned out to be dangerous not only to invertebrates (killing many natural enemies), but also to larger animals such as fish, reptiles, birds and mammals, leading to massive killing of wildlife in some places (Winston,
17
Chapter 1
1999; Oberemok et al., 2015). DDT’s fat soluble properties and extremely long persistence lead to long-term accumulation in fat tissue of non-target organisms. This bioaccumulation of DDT most strongly affects aquatic organisms and animals higher up the food chain, such as predators and humans (Jensen et al., 1969; Olsson & Reutergårdh, 1986; Beard, 2006; Gerber et al., 2016). Mounting evidence eventually lead to strong restriction and even prohibition of agricultural use of DDT and related substances in most developed countries after 1970 (Beard, 2006; Davies et al., 2007). Many organophosphorous compounds, carbamates and pyrethroids are still used nowadays despite their negative effects on natural enemies or the environment (Oberemok et al., 2015). Neonicotinoids form a more recent class of pesticides, first appearing on the market in the early 1990s and still very popular, though there are big concerns about their adverse effects on pollinators and natural enemies (Godfray et al., 2014; Oberemok et al., 2015).
Harm to the environment was one matter of concern, but another arose with the massive overuse of pesticides: widespread resistance in target pest populations. The first reaction to loss of effectiveness was to use pesticides more frequently and in higher doses, which only added to the problem (Georghiou, 1972). Pest resistance to chlorinated hydrocarbons was already discovered in the 1960s. New active compounds would solve the issue temporarily, but resistance –through different mechanisms– evolves quickly in pest populations and has now been detected for almost all pesticide classes (Georghiou, 1972; Nauen & Denholm, 2005; Bass et al., 2014). This evolution poses a threat to the continued and effective use of pesticides in many management programs (Davies et al., 2007). Despite their negative side-effects, pesticides are still a fast, cheap and convenient way of dealing with agricultural pests and can be essential when all other measures have failed.
However, because of the problems caused by pesticides, general view on pest management strategies changed slowly throughout the 20th century towards an integrated approach.
18
General introduction
1.1.3 Integrated pest management
Many concepts of integrated pest management (IPM) were already used before the term was invented. Before the introduction of synthetic pesticides, crop protection specialists used a combination of different pest management techniques. A multidisciplinary approach is one of the key concepts of IPM, which aims for sustainable crop protection (Dhaliwal et al., 2004; Peshin et al., 2009). The European Commission, based on an earlier version of the Food and Agriculture Organisation (FAO) of the United Nations, defines IPM as follows (EC, 2019):
“Integrated pest management means careful consideration of all available plant protection methods and subsequent integration of appropriate measures that discourage the development of populations of harmful organisms and keep the use of plant protection products and other forms of intervention to levels that are economically and ecologically justified and reduce or minimise risks to human health and the environment. 'Integrated pest management' emphasises the growth of a healthy crop with the least possible disruption to agro-ecosystems and encourages natural pest control mechanisms.”
Note that in IPM the goal is not to fully eradicate pests, but to keep their levels below a certain acceptable threshold. This threshold depends on the pest species, the crop and the farmer’s opinion of critical damage levels. Complete eradication is neither feasible nor realistic: it would demand a tremendous amount of work, monitoring and resources, and, in the case of chemical measures, an overuse that would increase insecticide resistance and decimate natural enemy populations. In the case of biological control, a balance between pests and natural enemies at low pest levels is much more sustainable than releasing high numbers of (expensive) natural enemies repeatedly to try and kill off the whole pest population.
The first official IPM programmes were already initiated in the late 1960s, but only by the end of the 1970s, after some major projects in the USA, did IPM really gain momentum. After the initial successes, the US government submitted to implement IPM on 75% of the national crop area by 2000 and achieved it on around 70% by then (Baron, 2000 as cited in Dhaliwal et al., 2004). In 1995, the FAO implemented a successful large-scale IPM program for rice cultivation in South-East Asia. The European Union strongly increased efforts to promote IPM and published the EU Framework Directive 2009/128/EC on the sustainable use of pesticides in 2009. This Directive stated that all crop protection should be managed according to IPM principles by 1 January 2014 (Parliament and Council Directive 2009/128/EC, 2009).
19
Chapter 1
The eight principles described in this Directive can be summarized as follows (according to Annex III of Directive 2009/128/EC): (1) Prevention of harmful organisms (2) Monitoring (3) Decision-making based on monitoring and thresholds (4) Preference for sustainable biological, physical and other non-chemical control methods (5) Use specific pesticides with limited side-effects (6) Limit pesticide application to minimal required levels (7) Apply anti-resistance strategies (8) Record and evaluate success of measures
IPM principles are increasingly used around the globe, but still much work has to be done to decrease dependency on and harmful consequences of pesticide use in agriculture (Peshin et al., 2009). Increasing research and knowledge on biological control is key to expand the successful adaptation of IPM strategies.
1.1.4 Biological control
Biological control is defined as the use of organisms (natural enemies) to reduce the population densities of other organisms (pests) (Cock et al., 2010; Barratt et al., 2018; van Lenteren et al., 2018). It has always played a crucial role in pest management before the supremacy of synthetic pesticides during the 20th century. In the past decades it increasingly regained attention because the preference for non-chemical pest management measures in IPM strongly stimulates research and development of biological pest control. Biological control is even more important in organic farming than it is in farming under IPM rules. In organic farming, no synthetic pesticides are allowed. Some natural substances can be used as pesticides, but the focus lies on other means of control, such as mechanical, cultural and most importantly, biological control. Therefore, organic growers benefit even more from new research and developments in biological control.
Four major types of biological control can be distinguished: natural, conservation, classical and augmentative biological control (van Lenteren et al., 2018).
20
General introduction
Natural biological control is the suppression of pests by naturally occurring enemies of these pests, without any human interference. This ecosystem service occurs everywhere around the world and is a free contribution of biological control to agriculture and therefore the most important one from an economical point of view (van Lenteren et al., 2018). In conservation biological control, these naturally occurring enemies are stimulated by human activities that protect and promote their populations (van Lenteren et al., 2018). In classical biological control, natural enemies are collected in a certain area, usually the native distribution area of the pest, and released there where the invasive pest causes damage. These natural enemies are expected to establish populations at their new location and control the pest species on a long term. Classical biological control was the first type that was practiced on a large scale, hence the name (Cock et al., 2010; van Lenteren, 2012; van Lenteren et al., 2018). Augmentative biological control is a commercial type where natural enemies are mass- reared and released in crops to achieve pest control on a short term. Two forms of augmentative biological control are practiced, depending on the length of the crop cycle and the pest density. The first one is inundative biological control. Here, immediate control is obtained by releasing large numbers of natural enemies in a crop with a short production cycle or when pest densities are very high. The second form is seasonal inoculative biological control, where natural enemies are released in smaller numbers to control several pest generations in crops with a long production cycle. In this case, natural enemies are expected to establish populations in the crop and increase in number during the next generations (van Lenteren et al., 2018).
The first major success story of classical biological control in modern times is that of the earlier mentioned vedalia beetle R. cardinalis, an Australian species which was introduced in Californian citrus orchards in 1888 to control cottony cushion scale I. purchasi, another Australian species (DeBach, 1964 as cited in Bale et al., 2008). A more recent example is found in Africa, where cassava crops in many countries suffered from an infestation of the South American mealy bug Phenacoccus manihoti Matile-Ferrero (Hemiptera: Pseudococcidae) in the 1970s and ‘80s. A natural enemy discovered in Paraguay, the parasitoid wasp Epidinocarsis lopezi (De Santis) (Hymenoptera: Encyrtidae), was released to successfully control the pest (Bale et al., 2008).
21
Chapter 1
An important factor of these introduced biological control agents is their specificity. Introducing a predator or parasitoid in a new ecosystem poses risks to that ecosystem. Originally, safety testing of biological control agents focused only on non-target effects on agricultural systems and thus many species were deemed safe to introduce when they proved harmless to crops. Later on, safety testing was expanded to include effects on local arthropod natural enemies. Negative effects on non-target species not associated with agriculture were largely ignored until ecologists and entomologists pointed them out as a threat to natural ecosystems (Waage, 2000; Hajek et al., 2016). However, evidence of negative effects of intentional exotic natural enemy introductions is limited. van Lenteren et al. (2006) showed that more than 50% of all nonindigenous insects (both accidentally and intentionally introduced) in the continental USA are considered pests, but of the intentionally introduced insects only 1.4% cause problems (van Lenteren et al., 2006). It must be noted, however, that non-target effects of biological control agents are under-reported, especially for past introductions from the time before environmental risks were considered (Louda et al., 2003). A well-known example of an introduced natural enemy causing havoc is the Asian harlequin ladybird Harmonia axyridis (Pallas) (Coleoptera: Coccinellidae). This voracious generalist predator established well throughout Europe and North America, damaging fruit and causing decline of native coccinellid populations (Roy et al., 2016). Since the 1990s, there has been an increase in establishment and harmonization of regulations regarding the introduction and release of exotic biological control agents to guarantee rigorous safety testing and minimize risks to native species and ecosystems (Bale et al., 2008; Hajek et al., 2016).
Augmentative biological control programs came later, with some first releases of lacewings, hoverfly larvae and ladybirds against aphids in European greenhouses in the 18th and 19th centuries (Pilkington et al., 2010). The first major success story was the control of greenhouse whitefly Trialeurodes vaporariorum (Westwood) (Hemiptera: Aleyrodidae) by the parasitoid wasp Encarsia formosa Gahan (Hymenoptera: Aphelinidae) around the 1920s in UK greenhouse crops. These biological control programs were mostly abandoned after the discovery of DDT and related pesticides in the 1940s, but resumed in the ‘70s after T. vaporariorum developed resistance (Hoddle et al., 1998; Bale et al., 2008). The number of available species used in augmentative biological control rose strongly by the end of the 20th century, stimulated by the urgent need for alternatives for pesticides. In the 21st century, the
22
General introduction
amount of new natural enemies available each year decreased again because of two factors: (1) by then there was a natural enemy available for most pests, and (2) the import and registration of exotic biological control agents became much more regulated (van Lenteren, 2012). Preference shifted from exotic towards indigenous natural enemies, even for controlling exotic pests, as the latter pose far fewer environmental risks and thus have less strict evaluation and registration procedures (van Lenteren, 2012). Augmentative biological control is mostly used in greenhouses, though it is also applied on field crops such as maize and sugar cane and in citrus orchards (van Lenteren, 2012).
Success stories of conservational biological control in recent times were first found in European fruit orchards. After the overuse of pesticides in the 1940s-1960s, populations of many naturally occurring predators and parasitoids had declined drastically. Measures were taken to support the local fauna which lead to a reduced number of outbreaks and successful biological control of several pest species (Gruys, 1982; Bale et al., 2008).
For most of the 20th century, specificity was seen as key characteristic of a good biological control agent. While it is important in classical biological control programs to reduce environmental risks, it is less essential in augmentative and conservational biological control and by the start of the 21st century the importance of generalist predators became understood (Symondson et al., 2002; Janssen & Sabelis, 2015). Specialists are often more effective when fighting exotic pests on exotic crops: they focus (only) on the target pest, have a high searching efficiency and a strong numerical response to prey densities. However, the strength of generalists lies in their unique functional response: their ability to rapidly switch food sources, enabling them to persist when pest numbers are low and to attack emerging or resurgent pests, is an ideal characteristic for augmentative biological control. Control of native pests in natural areas is more often achieved by native generalist predator communities, making them the subject of conservation biological control as well (Symondson et al., 2002).
The urgent need to reduce pesticide use is recently being recognized by some of the major agricultural economies of the world. Ever growing demands of retailers and consumers for low-pesticide or pesticide-free products and strong pushes and actions by NGOs have stimulated political developments in Europe, Asia and Latin-America. Barratt et al. (2018) made a recent and extensive review on the status of biological control and mention Brazil, China and India as examples where a huge shift towards biological control is taking place, accounting for
23
Chapter 1 an enormous acreage of agricultural land. Also the obligated shift towards IPM in all EU member states as of 1 January 2014 has had a big impact on increased use of biological control (Barratt et al., 2018).
1.1.5 Pest management in greenhouse crops
Greenhouses form a unique type of farming, mainly used for vegetable and ornamental crops. By using a glass or plastic shelter, crops can be grown at times and locations where it would otherwise not be possible due to unfavourable environmental or climate conditions. The benefits of controlled climate conditions, possibility to grow crops in proximity to consumers and partly exclusion of pests outweigh the higher costs of this growing system for many farmers (Pilkington et al., 2010). There are major differences concerning pest management between greenhouse cultivation in temperate and (sub)tropic regions as well as between glass and plastic greenhouses. I describe the situation for glass greenhouses in Northwestern Europe, though some aspects apply to greenhouse cultivation in general.
Unique growing conditions mean unique pest conditions that differ strongly from field environments (van Lenteren, 2000). After the crop season, often during winter, greenhouses are cleaned and made pest free for the start of the new season. Mass immigration of pests is avoided as the greenhouse provides a barrier between the crops and the outside world. Greenhouse crops are often exotic and so are their pests, but the number of pest species that has been imported into countries with greenhouses is limited. Moreover, many of these pests can’t survive the cold conditions outside the greenhouse. On the other hand, pest survival and development are well supported by the continuous warm climate and the high supply of well fertilized and irrigated crops, often leading to higher population growth rates (van Lenteren, 2000; Pilkington et al., 2010). A factor that complicates pest control is the increasing acreage of lighted crops in Northwestern Europe. During the colder and darker winter months, crops are grown under artificial light. These cycles overlap with the normal, non-lighted crop season and thus greenhouses are never completely empty, which promotes pest survival.
As mentioned earlier, augmentative biological control was first applied systematically in greenhouses in the 1920s. Greenhouse crops form an ideal environment for pest management using biological control agents. They are simplified ecosystems with a limited number of prey and hosts on a single plant species and under controlled climate conditions
24
General introduction
(Enkegaard & Brødsgaard, 2006). Immigration from and emigration to the surrounding area is limited compared to open fields, thus released biological control agents will establish in the crop. These favourable conditions provide a firm base for successful application of biological control, often as part of IPM, in greenhouse pest management programs. Several factors stimulate growers to shift towards increased biological and reduced chemical pest control measures: increased pesticide resistance in pests, reduced availability of effective active compounds, less exposure to greenhouse workers, demands from policy makers for reduction in pesticide use and retailers and consumers requiring residue-poor or -free products (van Lenteren, 2000, 2012). Sometimes, biological control is also cheaper than chemical control: when a single natural enemy release can control a pest, while otherwise workers had to go into the greenhouse on a weekly basis for repeated pesticide sprays. This economical aspect was an important driver for the shift to biological control in the Netherlands (pers. comm. Felix Wäckers, Biobest). Because of this, the worldwide acreage of greenhouses on which biological control is applied has increased from 200ha in 1970 to around 15 000ha by the end of the 1990s and an estimated 120 000ha nowadays (van Lenteren, 2000; van Lenteren et al., 2018). It is not clear what proportion of the global greenhouse crop acreage these numbers constitute.
Despite the fact that greenhouses form ideal environments for biological pest control, there’s still a lot of pesticide use as well, though this depends strongly on the region and the pest species to be controlled. Several barriers still exist for an increased uptake of biological control (Pilkington et al., 2010; van Lenteren, 2012; Barratt et al., 2018; van Lenteren et al., 2018). One such barrier is the price of control measures. Biological control is often more expensive than chemical control, so growers tend to opt for the latter. This would be different if the true costs of pesticides would be included in the price. All costs related to human health problems and environmental damage caused by pesticides are currently paid by society and not by the users and the manufacturers who are actually responsible (van Lenteren et al., 2018). It is therefore important that policy makers change regulations regarding synthetic pesticide use. Nevertheless, biological control is increasingly applied and one can expect that the above mentioned stimuli will further promote it in the near future (van Lenteren et al., 2018).
25
Chapter 1
1.2 Pest management in greenhouse sweet pepper
1.2.1 Sweet pepper cultivation in Europe
The agricultural crop under study in this PhD research is pepper Capsicum annuum L. (Solanaceae), a plant species native to Mesoamerica and already domesticated thousands of years ago in Mexico (Kraft et al., 2014). After Columbus arrived in the New World in 1492, it was introduced in Europe and the rest of the world and is now the most widely grown and most economically important Capsicum species worldwide (Eshbaugh, 1976; Andrews, 1995; Kraft et al., 2014). Many cultivars are grown across the globe with a large variation in colour, size, shape and taste of the peppers. Favoured cultivars in tropical America, Africa and Asia are hot peppers, usually referred to as chili peppers, which are used to flavour food. In temperate regions of North America and Europe, consumers prefer non-pungent pepper cultivars of which the fruits are mostly eaten as vegetables (Pickersgill, 1997). These large, mild peppers are called sweet pepper, bell pepper, capsicum or paprika. It’s the sweet pepper cultivars that are the focus of this PhD research.
Originally a plant from tropical regions, C. annuum does not survive frost and is therefore grown as an annual crop in colder regions. In Europe, sweet pepper is usually grown in greenhouses to provide the required climate conditions. This mostly means heating under glass in Northwestern European countries such as Belgium, the Netherlands and Norway and cooling under plastic in Southern European countries such as Spain. Greenhouse sweet pepper can be grown in the soil or on a substrate such as rockwool, coconut fibre or perlite. Many different cultivars are grown and new ones emerge yearly to improve yield, ease growing requirements, provide better resistance against pests and diseases or to meet new market demands. Sweet pepper plants are self-pollinating, but pollination and fruit production can be stimulated by releasing pollinators such as bumblebees in the greenhouse (Shipp et al., 1994; Serrano & Guerra-Sanz, 2006).
The acreage of greenhouse sweet pepper in Belgium was 91ha in 2019, producing a total of 26 600 tonnes of sweet pepper that year (VLAM, 2019a; b).
26
General introduction
1.2.2 Sweet pepper key pests and their management
Greenhouse sweet pepper crops are attacked by a number of pest species. The key pests in temperate regions are western flower thrips Frankliniella occidentalis Pergande (Thysanoptera: Thripidae), two-spotted spider mites Tetranychus urticae Koch (Acari: Tetranychidae) and several aphid species, the green peach aphid Myzus persicae (Sulzer) and the foxglove aphid Aulacorthum solani (Kaltenbach) (Hemiptera: Aphididae) being the most important ones (Ramakers, 2004a). Other pests include caterpillars of the moth Chrysodeixis chalcites (Esper) (Lepidoptera: Noctuidae), tobacco whitefly Bemisia tabaci (Gennadius) and greenhouse whitefly Trialeurodes vaporariorum (Westwood) (Hemiptera: Aleyrodidae). Tobacco whitefly is a major sweet pepper pest in the Mediterranean region (Calvo et al., 2009).
Biological control is very successful in European sweet pepper greenhouses and most of the above mentioned pests can be controlled by augmentative releases of natural enemies (Ramakers, 2004a; Van der Blom et al., 2008). Two generalists are available for control of F. occidentalis: the anthocorid bug Orius laevigatus (Fieber) (Hemiptera: Anthocoridae) (Dissevelt et al., 1995; Weintraub et al., 2011) and the predatory mite Amblyseius swirskii Athias-Henriot (Acari: Phytoseiidae) (Calvo et al., 2009), while the specialist predatory mite Phytoseiulus persimilis Athias-Henriot (Acari: Phytoseiidae) controls T. urticae (Gerson & Weintraub, 2007). Amblyseius swirskii and the parasitoid wasps Encarsia formosa Gahan, Eretmocerus eremicus Rose & Zolnerowich and Eretmocerus mundus Mercet (Hymenoptera: Aphelinidae) are released against the whiteflies T. vaporariorum and B. tabaci (van Lenteren et al., 1995; Hoddle et al., 1998; Albajes & Alomar, 1999; Bolckmans et al., 2005; Castañé & Sanchez, 2006; Arnó et al., 2010). The generalist mirid bug Nesidiocoris tenuis Reuter (Heteroptera: Miridae) is used to control whiteflies in Southern Europe, but is considered a pest in Northwestern Europe because of the crop damage it causes due to plant feeding (Calvo et al., 2009; Arnó et al., 2010; Castañé et al., 2011; Pérez-Hedo & Urbaneja, 2016; Moerkens et al., 2020).
Biological control of aphids in sweet pepper is another story however. For decades, several specialist biological control agents have been available for release against aphids: the parasitoid wasps Aphidius colemani Viereck, Aphidius ervi Haliday, Aphidius matricariae Haliday (Hymenoptera: Braconidae) and Aphelinus abdominalis (Dalman) (Hymenoptera: Aphelinidae) (Jarosik et al., 1996; Mölck et al., 1999; van Schelt et al., 2011; Prado et al., 2015) and the gall midge Aphidoletes aphidimyza (Rondani) (Diptera: Cecidomyiidae) (Messelink et al., 2011a).
27
Chapter 1
Yet, effective biological control of aphids remains difficult and chemical control is often applied by growers using IPM (Bloemhard & Ramakers, 2008; Gillespie et al., 2009; Sanchez et al., 2011; Messelink et al., 2013; Messelink, 2014). Organic growers can’t use synthetic insecticides and the number of effective insecticides allowed in organic farming is very limited, therefore they occasionally suffer from severe aphid infestations (Bloemhard & Ramakers, 2008). Moreover, due to their high reproductive rate and short generation time, aphids are able to quickly develop insecticide resistance and some species, such as M. persicae, have developed resistance against most classes of insecticides (Nauen & Elbert, 2003; Bass et al., 2014).
Increased attention to the role of generalist natural enemies in biological control (Symondson et al., 2002; Janssen & Sabelis, 2015) has caused researchers to look more towards generalist predatory bugs instead of specialists to control aphids. Generalists can suppress several pest species at once, but can also survive when no pests are present in the crop and can therefore be released preventively (Symondson et al., 2002). Several generalist predatory bugs have been studied recently for their effectiveness in aphid control in sweet pepper and mirid bugs proved to be better suited for this than anthocorid bugs, with Macrolophus pygmaeus Rambur (Heteroptera: Miridae) and N. tenuis showing very promising results (Perdikis & Lykouressis, 2004; Messelink et al., 2011b; Messelink & Janssen, 2014; De Backer et al., 2015; Messelink et al., 2015; Pérez-Hedo & Urbaneja, 2015; Bouagga et al., 2018a). However, as stated earlier, the latter one is regarded as a pest species in Northwestern Europe because of its plant feeding behaviour. Macrolophus pygmaeus, on the other hand, is commonly used in European tomato greenhouses against a number of pest species such as whiteflies, aphids, thrips, mites and moths (Enkegaard et al., 2001; Perdikis & Lykouressis, 2002a; Blaeser et al., 2004; Castañé et al., 2004; Urbaneja et al., 2009; Arnó et al., 2010; van Lenteren, 2012; De Backer et al., 2014; Moerkens et al., 2015). Expanding its use to sweet pepper crops is therefore easy and can offer a solution to the lack of successful biological aphid control as well as providing additional control of other key pests (Messelink et al., 2011b; Messelink & Janssen, 2014).
28
General introduction
1.3 Macrolophus pygmaeus
1.3.1 Biology
Macrolophus pygmaeus is a zoophytophagous bug native to the Palearctic region and mostly found on plant species of the families Solanaceae, Asteraceae and Lamiaceae (Martinez- Cascales et al., 2006; Sanchez et al., 2012). The taxonomy and species identity of M. pygmaeus was a subject of much controversy in the past and often the species was mistakenly identified as Macrolophus caliginosus Wagner (Syn. Macrolophus melanotoma Costa) (Martinez-Cascales et al., 2006; Castañé et al., 2013). Distinguishing between M. pygmaeus and M. caliginosus based on morphological features is extremely difficult and molecular analysis is advised to be certain (Castañé et al., 2013). The classification of M. pygmaeus is the following:
Kingdom: Animalia Phylum: Arthropoda Class: Insecta Order: Hemiptera Suborder: Heteroptera Family: Miridae Tribe: Dicyphini Genus: Macrolophus Species: pygmaeus Macrolophus pygmaeus are bright green bugs with an adult body length of 3-4 mm (Hillert et al., 2002; Castañé et al., 2013). Notable morphological features are: dark red eyes, a black dot on the apex of the clavus, a black band-shaped macula behind each eye and a generally black or dark base of the relatively long green antennae (Perdikis & Lykouressis, 2002b; Martinez-Cascales et al., 2006; Figure 1-1). Females lay up to 100-230 eggs during their lifetime. These eggs are laid in the tissue of a host plant’s stem and are nearly invisible, only the egg’s protruding respiratory horn can be observed (Perdikis & Lykouressis, 2002a). Development of all M. pygmaeus stages is, as for most insects, strongly dependent on temperature and also varies with host plant species and the availability and quality of prey. Numbers given here are in case of presence of prey. The incubation period lasts around 10-20 days, after which emerging larvae will go through five apterous nymphal stages (Figure 1-2)
29
Chapter 1 before reaching adulthood, which takes around 14 days at 30°C and up to more than 50 days at 15°C (Perdikis & Lykouressis, 2000, 2002a; Blaeser et al., 2004; Mollá et al., 2014). Newly emerged adults are pale coloured and it takes 30 hours before normal adult colours have fully appeared (Perdikis & Lykouressis, 2002b). Young adult females will go through a preoviposition period of 5-10 days before laying their first eggs. Adult M. pygmaeus bugs live for about 40-140 days, with males having a slightly higher longevity than females (Perdikis & Lykouressis, 2002a).
Figure 1-1. Adult Macrolophus pygmaeus bug. (Photo: Koppert)
Figure 1-2. Nymphal stages of Macrolophus pygmaeus: (a) first, (b) second, (c) third, (d) fourth, (e) early fifth and (f) late fifth stage. (Photos: Roosens & De Clercq, 2014)
As typical a hemipteran feature, M. pygmaeus bugs have piercing-sucking mouthparts with which they can feed on plant tissue and thus obtain water and essential nutrients from xylem and phloem sap (Gillespie & Mcgregor, 2000; Hamdi et al., 2013). Other food sources for M. pygmaeus provided by plants are pollen and nectar (Perdikis & Lykouressis, 2000; Messelink
30
General introduction
et al., 2011b). Though an exclusive plant diet allows M. pygmaeus to sustain a population, the absence of prey reduces fecundity and longevity and prolongs nymphal development (Perdikis & Lykouressis, 2000, 2004; Lykouressis et al., 2001, 2008). Feeding on prey happens in a similar matter as plant feeding: a rupture is made in the prey’s tissue, after which saliva containing digestive enzymes is injected and the contents are sucked out, leaving only an empty skin (Fantinou et al., 2008).
1.3.2 Usage in pest control
The benefits of M. pygmaeus in pest control have been known for decades in Europe. In Southern European countries such as Spain, Italy, Greece and the South of France, M. pygmaeus naturally colonizes agricultural crops in both open fields and greenhouses during spring and summer, where it plays a major role in the control of important pests including aphids, whiteflies and thrips. Colonization success depends on the presence of natural vegetation in the area and the lack of broad-spectrum insecticide use (Riudavets & Castañé, 1998; Lykouressis et al., 2000; Castañé et al., 2004; Gabarra et al., 2004; Arnó et al., 2009; Ingegno et al., 2009; Urbaneja et al., 2012; Biondi et al., 2013).
Apart from their role as naturally occurring predator in conservation biological control, M. pygmaeus is also used in augmentative biological control. It has been commercially available as a biological control agent since the early 1990s and mostly used in European greenhouses (Sanchez et al., 2012; EPPO, 2016). Macrolophus pygmaeus is released in a number of vegetable and ornamental greenhouse crops throughout Europe and against a number of pest species such as thrips, whiteflies, aphids, spider mites and leafminers including T. absoluta (Blaeser et al., 2004; Alomar et al., 2006; Perdikis et al., 2008; Castañé et al., 2011; van Lenteren, 2012). Introducing M. pygmaeus is standard practice in Northwestern European tomato greenhouses, where it is the most important natural enemy (Perdikis et al., 2008; Arnó et al., 2009; Castañé et al., 2011; van Lenteren, 2012; Moerkens et al., 2017). Yet, at this moment, the bugs are not being introduced (commonly) in sweet pepper greenhouses.
Being a generalist, M. pygmaeus is usually introduced preventively at the beginning of the crop cycle and before pests are present. To improve population growth and establishment, supplemental food is provided during the first weeks (Put et al., 2012; Messelink et al., 2015; Moerkens et al., 2017).
31
Chapter 1
The plant feeding behaviour of M. pygmaeus can cause damage, though it is limited and mostly observed under experimental conditions with high predator density and low prey availability. Injuries include feeding marks and pits on vegetative plant parts and fruits, fruit distortion and discolouring spots on ornamental flowers (reviewed in Castañé et al., 2011). Moerkens et al. showed significant fruit damage at high M. pygmaeus densities in tomatoes when the plants are infected with Pepino mosaic virus (Moerkens et al., 2015). Still, the predator is generally considered as safe when population densities are kept at an appropriate level (by chemical interventions if necessary) and it has been widely used in biological control programs (Castañé et al., 2011; De Backer et al., 2014; Moerkens et al., 2017).
1.3.3 Potential in sweet pepper
As mentioned before, biological control in sweet pepper is widely and successfully implemented, with an exception for aphids, where releases of specialist natural enemies do not lead to expected results. Introducing a generalist predator such as M. pygmaeus, which has proven its usefulness in several other crops, can offer the solution.
Macrolophus pygmaeus is found to naturally occur on sweet pepper plants in the field, though not always or everywhere, it does not seem to be a preferred plant (Goula & Alomar Kurz, 1994; Sanchez et al., 2003). However, Ingegno et al. (2011) found sweet pepper a suitable host plant for M. pygmaeus survival and establishment and the bugs did not significantly prefer tomato leaves over sweet pepper when given the choice in laboratory experiments (Ingegno et al., 2011). Moreover, sweet pepper plants typically have large amounts of pollen which can serve as a supplemental food source for M. pygmaeus (Pilkington et al., 2010).
Several studies report the ability of M. pygmaeus to reduce or control aphid infestations on sweet pepper plants in laboratory and cage experiments (Perdikis & Lykouressis, 2004; Messelink & Janssen, 2014; De Backer et al., 2015; Messelink et al., 2015; Pérez-Hedo & Urbaneja, 2015; Bouagga et al., 2018a). Furthermore, M. pygmaeus was also shown to control whiteflies and thrips on sweet pepper in cage experiments (Messelink et al., 2011b; Messelink & Janssen, 2014; Bouagga et al., 2018b) and is considered a better choice than O. laevigatus when the crop suffers from both aphids and thrips at the same time (Messelink et al., 2011b). As M. pygmaeus is known to feed on Lepidopteran eggs such as those of T. absoluta, it may
32
General introduction
also help in controlling the caterpillars of C. chalcites when released in sweet pepper greenhouses.
There is only one study that carried out a greenhouse experiment under practical conditions and it was focused on the control abilities of O. laevigatus as M. pygmaeus was not released deliberately but came from a remaining population that was used on a previous tomato cultivation in the same greenhouse (Messelink & Janssen, 2014). Nonetheless, experiments under practical conditions are essential to gain knowledge on the optimal management strategy of aphids using M. pygmaeus and to promote the use of this natural enemy in common IPM and biological control programs in sweet pepper.
In order to determine the success of a natural enemy in pest control, populations of both pest and natural enemy should be monitored. A successful natural enemy is able to reduce and keep pest population densities at acceptable levels. The field that studies the changes within and interactions between populations is population ecology.
33
Chapter 1
1.4 Population ecology in pest management
1.4.1 Population ecology
To be able to control a pest effectively and efficiently, one should have accurate information on the pest population: its size, age-structure and interactions with the environment and other species populations are important factors that can influence pest management. Population ecology is the subfield of biology that studies the changes in species population sizes and age-structures through time and space (i.e. population dynamics) and the interactions of populations with their environment and each other. Population dynamics and interactions are often analysed through population models, which find important practical applications in conservation biology and pest management (Sinclair et al., 1998; Kozlova et al., 2005; Tonnang et al., 2009; Ghosh et al., 2017). Population models are useful tools in pest management as they can help understand the underlying processes of pest infestations. Moreover, they can be used to make predictions on pest population growth or spread and thus serve as a decision support system that helps to make adequate and timely pest management decisions. Models which include interactions between populations, such as competition, predation or parasitism can be used to predict the success of biological pest control.
To build and use population models, data on pest populations are essential. These data, along with data on natural enemies, weather and climate, etc. are gained through experiments and monitoring. Monitoring is especially important to continually provide input data to validate and use the models once they are built and is therefore one of the basic principles of IPM.
1.4.2 Population models: the basics
Now, what are models exactly? To keep it short: models are simplified representations of a complex reality and are used in almost all aspects of science (Odenbaugh, 2005; Beissinger, 2012; de Roos, 2014). In ecology, models serve different purposes of which the following five can be considered basic purposes (Odenbaugh, 2005): (1) to explore possibilities, (2) to provide simplified means to investigate complex systems, (3) to provide conceptual frameworks, (4) to generate accurate predictions and (5) to generate explanations. The third purpose, generating accurate predictions, is what is mostly desired of population models in pest management
34
General introduction
(Tonnang et al., 2017). A single model can perform only one of these purposes or several at once. When building a model, its aim should be clear up front and will aid in deciding which elements should or should not be included. Consequently, a model should only be used for what it was built and nothing more (Odenbaugh, 2005).
One can roughly distinguish two approaches to modelling: empirical (also called statistical or phenomenological) and mechanistic (or mathematical) modelling (Kendall et al., 1999; Nestorov et al., 1999; Barraquand, 2014). Though it should be noted that this distinction is relative: empirical models will often contain a mechanistic element, while mechanistic models may have some level of empiricism involved, and that these two are just the ends of a scale (Nestorov et al., 1999; Turchin, 2003).
The goal of empirical population models is to describe a certain time series of population densities and extrapolate it into the future, i.e. make predictions. Time series are treated simply as a string of numbers without any qualitative or quantitative information on the ecological system from which they originated and thus ignoring the underlying biological processes. Commonly used types of models are the linear autoregressive model and several nonlinear models of varying complexity (Royama, 1992; Kendall et al., 1999; Barraquand, 2014).
Mechanistic population models aim to understand the underlying causes of a certain phenomenon instead of a particular time series. They include the major biological processes and mechanisms that are thought to generate the observed population dynamics and thus contain information from observations or experiments outside the given time series. The models establish a mechanistic relationship between the inputs and outputs and can’t only be used to predict future dynamics, but also to study and predict the effects of changes in the underlying processes and variables (Kendall et al., 1999; Baker et al., 2018). Mechanistic models often consist of difference or differential equations. The Lotka-Volterra model and its derivatives and the Nicholson-Bailey model are examples of mechanistic population models of interacting populations. These models consist of coupled differential equations (when assuming continuous time) or difference equations (when using discrete time steps).
Empirical models are stochastic by nature, while many mechanistic models are deterministic, though stochasticity can be included (Barraquand, 2014). The advantage of empirical models is that they don’t require any extra information on mechanisms or causalities and they make few assumptions. They can handle problems with multiple space and time scales
35
Chapter 1 within data. The drawback is that they require very large datasets and they can only make predictions related to patterns within the given dataset (Kendall et al., 1999; Turchin, 2003; Baker et al., 2018). Empirical models can be preferred in highly complex systems where too many important processes are present to model all of them (Jost & Arditi, 2001). Mechanistic models on the other hand can work with small datasets and, once validated, can be used for more complex predictions on patterns not present in the original data, such as the effect of different environmental changes on populations. This allows them to be used as a replacement for experiments when the latter can’t be performed. Yet their specific nature and many simplified assumptions limit the universality of predictions by these models. Other disadvantages are the difficulties of gathering sufficient knowledge on the causal mechanisms and incorporating information from multiple space and time scales (Kendall et al., 1999; Baker et al., 2018). When comparing the advantages and disadvantages of both modelling approaches, it becomes clear that they complement each other (Baker et al., 2018).
Both types of models are constructed according to a two-stage process: a subset of the data is used to build and calibrate the model in the first step, while the rest of the data is used to validate the model in the second step, by confirming and/or refining the model to increase its accuracy (Baker et al., 2018).
The population models in which we are interested to study the effect of natural enemies on greenhouse pests are predator-prey models.
1.4.3 Predator-prey models
Models can describe the dynamics of a single species population, but can also model two or more interacting populations. Interacting species can have a positive (+), negative (-) or neutral (0) effect on each other and based on this, five general categories of pairwise species interactions are distinguished in ecology: interference competition (-,-), mutualism (+,+), commensalism (+,0), amensalism (-,0) and trophic interactions (+,-). Trophic interactions, such as predation and parasitism, are by far the most studied interactions by ecologists and are also the most relevant ones for applied population ecology in pest management (Turchin, 2003).
36
General introduction
When learning about predator-prey interactions, one usually starts with the work of Lotka (in 1925) and Volterra (in 1926) who built the first mechanistic model describing the effects of a trophic (predator-prey) interaction on population oscillations through time (Turchin, 2003). The Lotka-Volterra (LV) predator-prey model consists of two coupled ordinary differential equations (ODEs), one for the prey and one for the predator population, and time as an independent variable (Turchin, 2003). The classical LV model assumes exponential growth/decline of populations and a linear functional response, which is not very realistic. However, this overly simplified basic form serves as a framework for building more complex and realistic predator-prey models (Turchin, 2003; Kozlova et al., 2005). Adjustments to make the model more realistic are a density- dependent prey growth and predator functional response (Turchin, 2003). Other relevant modifications such as predator mutual interference (Beddington, 1975; DeAngelis et al., 1975), a prey refuge (Gause et al., 1936; Kar, 2005; Křivan, 2011; Křivan & Priyadarshi, 2015; Ghosh et al., 2017) or time-delay (Kozlova et al., 2005) can also be included. More details on the LV model and its derivatives are given in Chapter 5.
In case of a complex biological system or when the mechanisms underlying the system are poorly known, it may be better to build more empirical models (Jost & Arditi, 2001; Turchin, 2003).
1.4.4 Examples of population models in pest management
The number of population models that are currently in use to aid in pest management seems to be rather limited, yet it is not always clear from literature which models under study have in fact found a way into daily practice.
A field in which insect population models have been used frequently is that of forest entomology. Sharov (1996) gives an overview of models built between the 1960s and 1990s. Different types of models on forest insect pests, such as several species of beetles and moths, have been developed and used, including: parametric (quantitative) and non-parametric (qualitative) mechanistic models, phenological models (that are often part of other models) and life-system models, which are composed of several submodels of different ecological processes (including phenological and mechanistic models).
37
Chapter 1
The danger of large life-system models that attempt to include all ecological mechanisms influencing the population is that they become so overly complex and are expected to serve too many purposes, that in the end they are almost useless for any specific purpose. Simpler, specialized models are then preferred to fulfill specific purposes (Sharov, 1996). An example of a huge life-system model is the Gypsy Moth Life System Model (GMLSM) that was developed during the 1980s and 1990s in the United States (Sharov & Colbert, 1994 as cited in Elkinton et al., 2011). Though the GMLSM aimed to be very realistic by including all available knowledge of the gypsy moth Lymantria dispar dispar (Linnaeus) (Lepidoptera: Erebidae), it took a decade to develop and turned out to be too complex and unable to provide the expected insights. Other similar huge life-system models suffered the same problems and were therefore abandoned (Sharov, 1996; Elkinton et al., 2011).
Phenological models are a special type of models that can be simulated independently from population dynamics. They model insect development, usually based on temperature (Sharov, 1996). Models for many species are used in practice and are often included in decision support systems (DSS) such as SOPRA (Switzerland), CIPRA (Canada) and RIMpro (originally Europe, now worldwide) (Samietz et al., 2008; Trapman et al., 2008; Cormier et al., 2016; Milkovich, 2016). These DSS collect local weather data and use them as model input to make predictions on pest and disease development, which can then be used to optimize timing of monitoring and control measures. Most of these models are for orchard pests and diseases and some for other open field crops.
Mechanistic models are far simpler than life-system models. It makes them less realistic, but more useful for their specific purposes. Aside from the impractical GMLSM, a simpler version of the trophic interaction involving gypsy moth consisting of three coupled ODEs was built as well (Wilder et al., 1994). Other ODE models of agricultural pests and their enemies can be found in literature (e.g. Kozlova et al., 2005; Tonnang et al., 2009), but it is not clear if they are actually being used for predicting population dynamics in practice.
38
General introduction
1.5 Thesis outline
The subject of this PhD research is biological pest management in sweet pepper greenhouses using the generalist predatory bug Macrolophus pygmaeus. First, release strategies of M. pygmaeus in sweet pepper were tested with regard to food supplementation: is supplemental food required for M. pygmaeus establishment in sweet pepper? If food supplementation is necessary, how and how often should it be applied? After optimizing the M. pygmaeus release strategy, its potential as biological control agent against two sweet pepper key pests was studied. Biological control of aphids was carried out in combination with leaf pruning. By pruning the lower leaves of sweet pepper plants, biological control is expected to improve as less foliage may increase search efficiency of the predatory bug. The aim here is to fill the gap that aphids cause in biological pest management programs of sweet pepper. Control of thrips was studied by modelling predator-prey interactions of M. pygmaeus and this pest. Population density data collected during the three growing seasons covered by this PhD research (2016-2018) provide information on the interactions between the species and the ability of M. pygmaeus to control thrips in sweet pepper greenhouses. A better understanding of the ecology in the greenhouse will improve pest management. Models can be used to make short-term predictions on the population dynamics of both species and the success of biological control. These predictions can then be integrated in a decision support system for pest management.
The greenhouses under study are the type mostly used in Northwestern Europe: glass greenhouses where sweet pepper plants are grown on substrate such as rockwool and with sophisticated climate regulating systems. All experiments were done at Research Centre Hoogstraten (Hoogstraten, Belgium) under practical conditions in semi-commercial greenhouse compartments of a realistic scale instead of in cages or small greenhouse compartments with only a few plants.
1.5.1 Improving M. pygmaeus release strategy
Chapters 2 and 3 of this thesis deal with the release strategy of M. pygmaeus in sweet pepper greenhouses. As generalists are used for their preventive action and functional
39
Chapter 1 response rather than curative action with a fast numerical response, they are released before pests are present in the crop. This means that there’s limited food availability at first which can impede survival and reproduction and so food supplementation is often carried out when introducing generalists (Put et al., 2012; Vangansbeke et al., 2016; Moerkens et al., 2017). Quick establishment of M. pygmaeus in sweet pepper crop is desired to have a strong standing army that can attack emerging pests. Therefore the effects of different food supplementation strategies on population build-up and dispersal were studied. The first question addressed is whether M. pygmaeus requires any food supplementation when released in sweet pepper crops. Sweet pepper plants produce lots of pollen, thus the need for supplemental food applications may be reduced. Subsequent questions deal with the optimal distribution, frequency and type of supplemented food that result in quick population build-up and dispersal throughout the greenhouse. Food can be distributed locally at the release sites of M. pygmaeus or could be distributed homogeneously throughout the greenhouse. A weekly supplementation scheme is applied for M. pygmaeus in tomato crops (Moerkens et al., 2017), but the extra available pollen in sweet pepper may allow less frequent food applications. Two types of supplemental food are commonly used to facilitate natural enemy establishment in greenhouses: eggs of the Mediterranean flour moth Ephestia kuehniella Zeller (Lepidoptera: Pyralidae) have a high nutritional value, but are expensive, while cysts of brine shrimps Artemia spp. (Anostraca: Artemiidae) are a lot cheaper, but of less quality as a food source (De Clercq et al., 2005). To take the middle way, a commercial mix of these two food types exists as well. The effects of different food supplementation strategies were tested during two consecutive cropping seasons (2017-2018).
1.5.2 Aphid control with M. pygmaeus
In chapter 4 a new method for controlling aphids with M. pygmaeus is investigated. Hypothesizing that search efficiency of M. pygmaeus for aphids is hampered by the large vertical foliage length reached by sweet pepper plants during the crop season, leaf pruning is considered as a solution. By pruning the lower leaves of the plants, the amount of foliage is reduced which brings prey and predator closer together and should improve the chance of prey encounters by the predator. The lower half of the crop receives less and less light as the Leaf Area Index (LAI, the projected area of leaves over a unit of land) increases throughout the growing season. Consequently, the lower leaves consume more energy than they produce and
40
General introduction
thus pruning may even have a positive effect on sweet pepper production (Dueck et al., 2006). The effect of different pruning heights on aphid control and sweet pepper production was studied during two consecutive seasons (2017-2018). The aphid species under consideration are the green peach or peach–potato aphid M. persicae and the foxglove or glasshouse-potato aphid A. solani. Both belong to the most problematic aphid species colonizing protected sweet pepper crops worldwide (van Lenteren et al., 1997; van Schelt, 1999; Blackman & Eastop, 2000; Sanchez et al., 2007, 2011).
1.5.3 Predicting pest control with predator-prey models
In chapter 5 monitoring data of a pest and its natural enemy in the greenhouse are used to build predator-prey models. During the growing seasons of 2016, 2017 and 2018, time series data of M. pygmaeus and western flower thrips F. occidentalis population densities were collected by sampling sweet pepper flowers. Thrips is commonly the first pest species to arrive in the greenhouse and these data provide valuable information on the 1-on-1 predator-prey relationship between M. pygmaeus and F. occidentalis since no other pests were present yet. Monitoring and decision-making based on monitoring are basic principles of IPM. In order to assess the success of biological control measures, it is important to monitor not only the pest species, but the natural enemy population (predator or parasitoid) as well. It is the interaction between these two that determines the outcome of biological control programs. Two different types of models were built and parameterized using the collected time series data: a simple logistic regression (empirical) model which calculates the chance of the pest being under control and mechanistic ODE models of the LV framework which calculate population densities of both pest and predator. Four LV based ODE models were fitted and compared: (1) the basic LV model, (2) the RM model with Holling type II (specialist) functional response, (3) the RM with Holling type III (generalist) functional response and (4) model 3 extended with a prey refuge (Kar, 2005). Data of 2016 were used to parameterize the models and data of 2017 and 2018 were used to validate them. The aim was to build models that can be used to make short-term predictions about the situation in the greenhouse and serve as a decision support system: will control be achieved in the near future or are other measures required?
41
Chapter 1
With this PhD research, I wish to contribute to and improve biological pest management in greenhouse sweet pepper crops. My colleagues and me are convinced that M. pygmaeus can have a significant role here and our experiments produced very good results for introducing the predatory bug in sweet pepper crops and using it against key pest species such as aphids and thrips. This work was partly integrated and partly a spin-off of the Belgian PeMaTo and European PeMaTo-EuroPep projects that both aim to improve biological pest management in tomato and sweet pepper greenhouses.
42
43
44
Chapter 2 | Food supplementation to optimize inoculative release of the predatory bug Macrolophus pygmaeus in sweet pepper
Nathalie Brenard, Vincent Sluydts, Luc De Bruyn, Herwig Leirs & Rob Moerkens
Modified version of publication in: Entomologia Experimentalis et Applicata (2018), Vol. 166(7): 574-582. DOI: 10.1111/eea.12704
2.1 Abstract
Biological control is widespread in management of greenhouse sweet pepper crops. Several species of predatory mites, bugs, and parasitoids are used against a wide range of pest species. However, biological control of particular pests like aphids, caterpillars, and the tobacco whitefly, Bemisia tabaci Gennadius (Hemiptera: Aleyrodidae), remains problematic. Macrolophus pygmaeus Rambur (Hemiptera: Miridae) is a generalist predatory bug which is used on a large scale in Western European tomato greenhouses. It has already been demonstrated that M. pygmaeus is a valuable biocontrol option in sweet pepper crops, but it has yet to find its way into common practice. Macrolophus pygmaeus should be introduced at the start of the growing season and determining an optimal release strategy is a key step in this process. In tomato crops, M. pygmaeus requires supplemental food releases to reach sufficient population numbers and dispersal levels. In this study, the need for food supplementation in sweet pepper is investigated. Three strategies were tested: (1) no food supplementation, (2) local food supplementation, and (3) full field food supplementation. Both population numbers and dispersal rates of the second generation were higher under the third strategy. Macrolophus pygmaeus oviposits near food sources, therefore dispersal rates are higher when food is more spread out. Pest control was achieved in all treatments, but faster and at lower pest levels under the full field strategy.
45
Chapter 2
2.2 Introduction
One of the most illustrative examples of successful biological pest management is found in greenhouse sweet pepper crops in Europe (Ramakers, 2004; van der Blom et al., 2008). Inoculative releases of predatory bugs, predatory mites, and parasitoids are all used to control a variety of sweet pepper related pest species. Well known examples are the use of phytoseiid mites like Phytoseiulus persimilis Athias-Henriot (Acari: Phytoseiidae) against the two-spotted spider mite Tetranychus urticae Koch (Acari: Tetranychidae) (Gerson & Weintraub, 2007), anthocorid bugs of the genus Orius against western flower thrips, Frankliniella occidentalis (Pergande) (Thysanoptera: Thripidae) (Dissevelt et al., 1995; Weintraub et al., 2011), and the parasitoids Encarsia formosa Gahan and Eretmocerus eremicus Rose & Zolnerowich (Hymenoptera: Aphelinidae) against greenhouse whitefly, Trialeurodes vaporariorum (Westwood) (Hemiptera: Aleyrodidae) (van Lenteren et al., 1995; Hoddle et al., 1998; Bolckmans et al., 2005). Despite these successes, the biological control of aphids remains difficult in sweet pepper greenhouses and growers rely mostly on chemicals (R Moerkens, pers. comm.). Releases of the parasitoid wasps Aphidius spp. (Hymenoptera: Braconidae) (van Schelt et al., 2011; Prado et al., 2015) and Aphelinus abdominalis (Dalman) (Hymenoptera: Aphelinidae) (Jarosik et al., 1996; Mölck et al., 1999), and the gall midge Aphidoletes aphidimyza (Rondani) (Diptera: Cecidomyiidae) (Messelink et al., 2011c) are frequently applied against aphids, but control remains ineffective in European sweet pepper greenhouses (Bloemhard & Ramakers, 2008; Messelink et al., 2013). Moreover, these specialists will often disappear after reducing pest levels and are unable to establish a population in the greenhouse. Their use is limited to obtaining rapid control of an occurring pest outbreak by releasing high quantities at the right moment which requires intensive monitoring. They can’t prevent new outbreaks unless released frequently, but this is not always economically viable (Messelink et al., 2014). In protected tomato crops, aphids are successfully controlled with augmentative releases of the generalist predatory bug Macrolophus pygmaeus Rambur (Hemiptera: Miridae) (Perdikis & Lykouressis, 2002a; Urbaneja et al., 2009). It is one of the most important natural enemies in biocontrol programs of tomato crops, as it feeds on multiple species affecting the crop such as mites, thrips, and whiteflies (Enkegaard et al., 2001; Blaeser et al., 2004; Castañé et al., 2004; Alomar et al., 2006). In case of prey scarcity, M. pygmaeus will feed on plant tissue
46
Food supplementation for Macrolophus pygmaeus
while lowering its reproduction (Perdikis & Lykouressis, 2000; Lykouressis et al., 2001; Ingegno et al., 2011; Portillo et al., 2012). Therefore it can maintain its population in the greenhouse, often preventing new pest outbreaks. Several studies have shown that M. pygmaeus is able to sustain healthy populations and prevent aphid outbreaks in sweet pepper crops (van Schelt et al., 2011; Messelink & Janssen, 2014; De Backer et al., 2015; Pérez-Hedo & Urbaneja, 2015). Macrolophus pygmaeus can also control thrips in sweet pepper, although less efficient compared to Orius laevigatus (Fieber) (Hemiptera: Anthocoridae) which is commonly used against thrips (Messelink et al., 2011b; Messelink & Janssen, 2014). Messelink et al. (2011b) identified M. pygmaeus as the best choice when sweet pepper crops suffer from both aphid and thrips attacks. The mirid bug can control both pests, whereas O. laevigatus fails to control aphids. Despite these results, augmentative releases of M. pygmaeus in European greenhouse sweet pepper crops are rare. In order to successfully introduce M. pygmaeus in sweet pepper crops, an optimal release strategy should be determined. Natural enemies and supplemental food are expensive for growers and releasing them can be labour intensive. Nevertheless a grower needs a sufficient population, spread throughout the entire greenhouse. Densities should not be too high either, as cannibalism is common (Hamdi et al., 2013) and the bugs can cause fruit damage in tomato (Castañé et al., 2011). Whether M. pygmaeus can cause damage in a pepper crop is so far unknown. The right timing, number, and spreading of releases and perhaps supplemental food are very important to gain maximum results at minimum costs. In greenhouse tomato crops, M. pygmaeus is released soon after planting. Distributing the predators over more plants and supplementing food ensures a quick and sufficient population build-up throughout the crop (Moerkens et al., 2017). This supplementary food usually consists of eggs of the Mediterranean flour moth, Ephestia kuehniella Zeller (Lepidoptera: Pyralidae), often mixed with cysts of the brine shrimp, Artemia franciscana Kellogg (Anostraca: Artemiidae) (De Clercq et al., 2014). Traditionally, food is supplied directly on the release plants, but Put et al. (2012) demonstrated that uniform application of E. kuehniella eggs ensures higher population numbers and higher dispersal rates of M. pygmaeus. Quite some variation exists among growers concerning the number of supplementary food applications, ranging from weekly to biweekly for 2-6 weeks or even longer. No research has been published about the need for supplementary food applications when using M. pygmaeus in biocontrol programs of sweet pepper. In contrast to tomato crops,
47
Chapter 2 sweet pepper crops produce a lot of pollen which is available as a food source for zoophytophagous bugs. In the case of O. laevigatus, food supplementation is unnecessary as the bugs feed on nectar and pollen produced by sweet pepper plants (Cocuzza et al., 1997; Hulshof & Jurchenko, 2000). It has been hypothesized that M. pygmaeus also feeds on sweet pepper nectar and pollen (Messelink et al., 2011b; Messelink et al., 2015). Vandekerkhove & De Clercq (2010) demonstrated that M. pygmaeus could develop and reproduce on a mixed diet of pollen and E. kuehniella eggs. Macrolophus pygmaeus lays its eggs near food sources (Put et al., 2012; Moerkens et al., 2017), so if food supplementation is advantageous, spreading it throughout the greenhouse might benefit dispersal rate more than supplementing it locally on the release plants. This study investigated the need for food supplementation to boost M. pygmaeus population build-up and dispersal rate in greenhouse sweet pepper crops at the start of the growing season. Population numbers and dispersal rates were compared between three supplementary food application strategies: (1) no supplementation, (2) local supplementation on release plants, and (3) full field supplementation where food was distributed homogeneously throughout the greenhouse.
2.3 Materials and methods
2.3.1 Greenhouse location, crop, and climate conditions
The trial was carried out in six compartments (A-F) in a semi-commercial sweet pepper greenhouse at Research Centre Hoogstraten (Hoogstraten, Belgium). Greenhouse compartment surface ranged between 500 and 1 500 m² (Table 2-1). Plants were sown on 21 October 2016 and planted in the greenhouse on 7 December 2016. Plant distance was 32 cm, with 2.4 plants m-². Each plant had three stems, which resulted in 7.1 stems m-². This stem density is common practice in Belgian greenhouses. Plants were planted on a rockwool substrate (Cultilene, Tilburg, The Netherlands). The greenhouses were 7 m high and equipped with a gutter growing system (FormFlex/Metazet, Wateringen, The Netherlands). Climate conditions were automatically logged in each compartment and registered by means of an Electronic Measuring Box (Priva, De Lier, The Netherlands). Main sweet pepper Capsicum annuum L. (Solanaceae) varieties were Maduro (Enza Zaden, Enkhuizen, The Netherlands),
48
Food supplementation for Macrolophus pygmaeus
Allrounder (Rijk Zwaan, De Lier, The Netherlands), and Overture (Syngenta, De Lier, The Netherlands). Variety distribution and climate conditions are given in Table 2-1. Prior to the experiment, plants were checked and found to be pest free. No plant protection products were used in the greenhouse compartments during the course of the experiment. A single fertilization scheme representable for practical conditions and advised by a crop advisor was used in all compartments.
Table 2-1. Characteristics of the different sweet pepper greenhouse compartments. Temperature (T) and relative humidity (r.h.) during the period from M. pygmaeus release (16/12/2016) until the end of sampling (15/03/2017). Compartment Surface (m²) Variety Mean ± SE Mean ± SE Treatment T (°C) r.h. (%) A 500 Maduro 17.7 ± 0.1 67.0 ± 0.8 No food B 500 Overture 17.5 ± 0.2 68.3 ± 0.9 Full field C 1500 Maduro 17.7 ± 0.2 65.7 ± 0.8 No food D 1500 Allrounder 17.5 ± 0.2 69.5 ± 0.9 Local E 500 Maduro 17.1 ± 0.2 71.3 ± 1.0 Local F 500 Maduro 17.4 ± 0.2 69.2 ± 1.2 Full field
49
Chapter 2
2.3.2 Release of Macrolophus pygmaeus
Macrolophus pygmaeus (product name Mirical; Koppert, Berkel en Rodenrijs, The Netherlands) was released on 16 December 2016. For each 500 m² of greenhouse compartment, the contents of one Mirical tube were equally distributed across four locations, resulting in four release locations in the small compartments (500 m²) and 12 in the larger ones (1 500 m²). Each location consisted of five consecutive plants. One Mirical tube contains 500 M. pygmaeus individuals of which 90% are adults and 10% are N4-N5 nymphal instars, and some wood chips as substrate. Contents were gently mixed while holding the tube horizontally and then divided into four equal parts, containing approximately 125 individuals. Each quarter was then equally distributed over the five plants of a release location. Insects and substrate were placed in a DIBOX (Koppert; Figure S2-1) for distribution, each release plant having one DIBOX that hung on the petiole of a lower leaf. Insects were counted while placing them in a box so 25 M. pygmaeus were released per plant, resulting in an average density of one individual m-² in all greenhouse compartments.
2.3.3 Supplementary food applications
Three supplementary food application strategies were tested in six greenhouse compartments, two for each strategy. In compartments A and C no supplementary food was provided after release of M. pygmaeus. In compartments D and E supplementary food was distributed on the release plants of M. pygmaeus only. To this end, the correct amount of food was placed on a small Petri dish and carefully blown onto the plant. In compartments B and F supplementary food was applied as a full field application using a Mini-Airbug that automatically blows food on the plant (Koppert; Figure S2-2). These three strategies will be called ‘no food’, ‘local’, and ‘full field’ throughout the text. Supplementary food was provided, according to the feeding strategy, at the time of M. pygmaeus release and weekly during the next 6 weeks. As we are testing a full field supplementary food application, the costs for the growers should not be ignored. Therefore, we selected cheaper A. franciscana (product name Artefeed; Koppert) over E. kuehniella for this study. Food was provided ad libitum on the release plants under the local food strategy. The producer’s advice of 0.40 g per release plant was followed, resulting in 0.017 g m-² when considering the whole greenhouse compartment, which held 20 or 60 release plants,
50
Food supplementation for Macrolophus pygmaeus
depending on the compartment’s size (500 or 1 500 m²). More food was more homogeneously distributed in the compartments with the full field strategy (0.04 g per plant; 0.08 g m-²). In this case the dosage per plant is determined by the speed at which the user of the Mini-Airbug walks by the plants. This resulted in a total of 0.119 g m-² at a density of 2.8 g per release plant added in each compartment of the local strategy and 0.56 g m-² at a density of 0.28 g per plant in each compartment of the full field strategy by the end of the experiment.
2.3.4 Population dynamics and dispersal
The population growth of M. pygmaeus was recorded weekly from 23 December 2016 until 15 March 2017. Four count plots were created in each greenhouse compartment, each consisting of 11 consecutive plants. Two plots per compartment included release plants of M. pygmaeus and are hereafter referred to as ‘release plots’. The other two plots were selected two plant rows further and insect counts at these locations give an indication of the dispersal rate of M. pygmaeus in the greenhouse. These plots are cited as ‘dispersal plots’ throughout the text. The diagonal plot distance between a release plot and its corresponding dispersal plot was 8.4 m and no other release plants were located within a closer range than 8.4 m. In order to keep sufficient distance between a dispersal and a non-corresponding release plot, no more than four count plots fitted a compartment of 500 m². The same number of plots was created in the two large compartments. For each of the three supplementary food application strategies, there are four release and four dispersal plots divided over two compartments. A release plot and its corresponding dispersal plot are considered to be one replicate. The M. pygmaeus individuals were counted on two flowers and three random leaves per plant, divided over three plant heights (upper, middle, lower). The five nymphal instars and adults were categorized in three groups: N1-3, N4-5, and adults.
2.3.5 Statistical analysis
Of all observed M. pygmaeus individuals, 92% were found in the flowers of the pepper plants. Therefore we restricted our analysis to the count data from the flowers only. To assess the effect of supplementary food applications on the population build-up, counts of M. pygmaeus at the times of peak abundance of the first and second generation were compared. The first two generations of an introduced M. pygmaeus population do not overlap. Prior to
51
Chapter 2 the analysis, counts of the various life stages on a flower were summed to obtain a total number of M. pygmaeus. Next, M. pygmaeus totals of the 22 flowers in a plot were summed. To obtain the overall abundance per replicate, counts of the release plot and the corresponding dispersal plot were grouped. Macrolophus pygmaeus counts were analysed by constructing a generalized linear mixed model (GLMM), adding ‘compartment’ as a random factor. Treatment, generation, their interaction, and sweet pepper variety were treated as fixed effects. Residuals were assumed to follow a Poisson distribution. Dispersal rate was determined at the time of peak abundance of the second generation as the ratio of the individuals counted in dispersal plots vs. the total number of individuals in the population. Counts of M. pygmaeus individuals on the 22 flowers in each plot were summed prior to analysis. Again a GLMM was constructed, treating compartment as a random factor and treatment, life stage, their interaction, and variety as fixed effects. The number of adults counted in the crop was too low (16 over all compartments) so this life stage was omitted from the analysis. In this case a binomial distribution was used. At first, fully parameterised models were constructed. Non-significant interactions and factor effects were sequentially dropped until significance level reached 0.05 or less by using a log-likelihood ratio test (χ²). Post-hoc comparisons were performed by least square means with Tukey-adjustments of P-values. Statistical analyses were carried out in R v.3.4.1 (R Core Team, 2017), using packages lme4 (Bates et al., 2015) and lsmeans (Lenth, 2016).
2.4 Results
The experiment lasted for approximately 13 weeks in which two generations of M. pygmaeus developed (Figure 2-1). Total numbers rose steeply during the early life stages (N1- 3) and gradually lowered when the later life stages appeared. The first generation reached its maximum numbers 24 days after release on 9 January 2017, the second 81 days after release on 8 March 2017.
52
Food supplementation for Macrolophus pygmaeus
Figure 2-1. Mean (± SE) number of individuals of the various Macrolophus pygmaeus life stages [nymphal stages 1-3 (N1-3), 4-5 (N4-5), and adults] per flower in all sweet pepper greenhouse compartments combined in 2016/2017.
2.4.1 Population build-up
Population increase was highest when food was supplemented with the full field strategy and lowest when no food was supplemented (Figure 2-2, Table 2-2). This effect occurred in both generations (Table 2-2). Providing food supplements following a full field strategy increased population levels significantly as compared to both local (P = 0.015) and no food supplementation (P < 0.0001). The difference between local and no food supplementation was also strongly significant (P = 0.0001). The second generation reached significantly higher population levels compared to the first in all treatments (P = 0.003). Differences in sweet pepper variety had no effect on the population build-up of M. pygmaeus (Table 2-2).
53
Chapter 2
Figure 2-2. Mean (± SE) number of Macrolophus pygmaeus individuals per sweet pepper flower for each food supplementation strategy at two generation peaks.
Table 2-2. Results of the GLMM’s testing effects of treatment, generation, and plant variety on population build-up and dispersal. Population build-up Dispersal χ² d.f. P χ² d.f. P Treatment*generation 1.88 2 0.39 - - - Treatment*life stage - - - 0.35 2 0.84 Treatment 15.05 2 0.001 11.03 2 0.004 Generation 8.87 1 0.003 - - - Life stage - - - 0.32 1 0.57 Variety 4.43 2 0.11 0.27 2 0.87
54
Food supplementation for Macrolophus pygmaeus
2.4.2 Dispersal
Both the full field and the no food strategy caused higher dispersal rates than the local supplementation strategy (P < 0.0001 and P = 0.03, respectively) (Figure 2-3, Table 2-2); 69.6 and 60%, respectively, of the second generation population was found in dispersal plots. In case of the local supplementation strategy, only 17.5% of the population was present in dispersal plots. No significant difference in dispersal rate was found between the full field and no food strategies (P = 0.82). Sweet pepper variety and M. pygmaeus life stage had no significant effect on dispersal rate (Table 2-2).
Figure 2-3. Mean (± SE) percentage of the second generation Macrolophus pygmaeus population found in dispersal plots vs. the total population for each food supplementation strategy.
2.5 Discussion
In order to prevent pest outbreaks in greenhouse crops with augmentative releases of generalist predators, predator populations must reach sufficient numbers, well spread throughout the greenhouse, before pests emerge. If not, a first pest outbreak can’t be prevented and the use of chemicals will often be necessary. These chemicals may then harm the predator populations as well. Releasing generalist predators early in the season and
55
Chapter 2 providing supplementary food helps establishing populations in time (Messelink et al., 2011b; Messelink et al., 2015). This study found that the population build-up of an introduced M. pygmaeus population in greenhouse sweet pepper crops happens more swiftly when food is supplemented. Macrolophus pygmaeus, like O. laevigatus, is a zoophytophagous predator that feeds on plant tissue and pollen in the absence of prey (Perdikis & Lykouressis, 2000; Lykouressis et al., 2001; Ingegno et al., 2011; Portillo et al., 2012). Whereas O. laevigatus can reach sufficient numbers by feeding on sweet pepper nectar and pollen without the need for food supplementation (Cocuzza et al., 1997; Hulshof & Jurchenko, 2000), we found that this is not the case for M. pygmaeus. Female bugs oviposit near food sources (Put et al., 2012; Moerkens et al., 2017). Supplementing food locally might limit the need for dispersal of these females. Moreover, when food sources are abundant on the plants of hatching, young bugs are not forced to disperse in search for more food. Our results indicated that dispersal rate was indeed significantly lower when applying the local food strategy. When applying the full field strategy, a significantly larger proportion of the second generation was found in the dispersal plots. This was also the case when no food was supplemented. Limited food availability on release or hatching plants clearly encourages bugs to disperse. Aside from improving the dispersal rate, the full field strategy also causes the population to reach higher numbers compared to the local food strategy. Various factors may be involved. For example, females may have to compete more for oviposition spots near the locally supplemented food sources. Macrolophus pygmaeus is known to exhibit cannibalism (Hamdi et al., 2013) and too high densities on release plants due to limited dispersal rates may trigger cannibalism. Another relevant difference may be the total amount of food added – this was higher in the full field strategy. The results of this study indicate that the optimal release strategy for M. pygmaeus in sweet pepper greenhouse crops is similar to that in tomato crops (Put et al., 2012), requiring full field food supplementation to ensure sufficient population build-up and dispersal levels at the start of the season. Food supplementation should last 6-8 weeks, which is the time required to complete one generation (Moerkens et al., 2017). Macrolophus pygmaeus is used extensively in tomato crops against a number of pests, but has yet to find its way into general sweet pepper management practices. No pest outbreaks
56
Food supplementation for Macrolophus pygmaeus
were registered during the trial. Later in the season thrips invaded each greenhouse compartment, but was controlled by M. pygmaeus in all of them (see Chapter 5). A quick and easy control was achieved in both compartments with the full field strategy, even though one of these compartments has a yearly, severe thrips outbreak. In the other compartments thrips were also controlled, but often only after they reached very high numbers. This confirms that M. pygmaeus is able to control thrips outbreaks. Messelink et al. (2011b, 2014) also found that M. pygmaeus can be used against thrips in sweet pepper, either alone or together with O. laevigatus. The past 10-20 years, the tobacco whitefly, Bemisia tabaci Gennadius (Hemiptera: Aleyrodidae), has become a major problem in Western European greenhouse crops (Oliveira et al., 2001). Originally a pest in subtropical regions across the world, it has spread to temperate regions with a mild climate or with a greenhouse culture (Kirk et al., 2000). This highly resistant whitefly species (Elbert & Nauen, 2000; Palumbo et al., 2001; Horowitz et al., 2002; Nauen & Denholm, 2005) can’t be controlled by any insecticide currently on the market. Growers rely on biological control by the predatory mite Amblyseius swirskii Athias-Henriot (Acari: Phytoseiidae) (Bolckmans et al., 2005) and parasitoid wasps of the genus Eretmocerus (Stansly et al., 2005; Urbaneja et al., 2007). Releasing M. pygmaeus might be a solution for this emerging pest in sweet pepper. The bugs can feed and reproduce on B. tabaci nymphs (Sylla et al., 2016) and are already used successfully in tomato greenhouses. Bouagga et al. (2018b) showed in cage experiments that M. pygmaeus could control B. tabaci infestations on sweet pepper plants as well. Macrolophus pygmaeus is zoophytophagous and thus also capable of plant or fruit damage. The bugs feed on tomato plants which provide essential nutrients not found in prey (Moerkens et al., 2015). Severe damage occurs at high population densities and especially when there is an interaction with the Pepino mosaic virus (PepMV) in tomato (Moerkens et al., 2015). As all Belgian and Dutch tomato crops are infected with PepMV, either naturally of vaccinated, this is an important issue. Despite the risks, both the literature and our study suggest to incorporate M. pygmaeus in sweet pepper pest management programs as well. Possible complications should not be neglected, however, and more research must be done on determining the risks in sweet pepper. Growers who already use M. pygmaeus in sweet pepper crops gave notice of splitting in the heads of the plants at high population densities and they
57
Chapter 2 sometimes reported small spots on the fruits. This was also observed in some of our study plants and should be investigated further.
In practice, several different application rates are used when providing supplementary food for M. pygmaeus. The best program in tomato was found to be a weekly application for 6-8 weeks (i.e., M. pygmaeus generation time) (Moerkens et al., 2017). Supplementary food was provided for 7 weeks in this experiment, which yielded good results, but additional experiments should be conducted to optimize application rates and determine the most cost- efficient one. The choice of food source is also a topic of interest. We opted for the cheaper A. franciscana cysts in contrast to the more expensive E. kuehniella eggs. In Belgium, the average price for 500 g Artemia spp. cysts is €25, while 500 g E. kuehniella eggs costs on average €750 (pers. comm. Rob Moerkens, R&D Biobest, 2020). However, the latter might produce better results or may require fewer applications as it has a higher nutritional value (De Clercq et al., 2005; Vandekerkhove et al., 2009; Table S2-1). Our limited number of greenhouse compartments did not allow to test these factors during the same growing season, but it will be a topic of future work.
Another subject of research could be the release method of M. pygmaeus. In our study, bugs were released from four locations of five consecutive plants in each greenhouse compartment. Applying a full field strategy for bug release might also enhance dispersal rate and population growth, but will also increase labour costs.
We found that the way supplemental food is distributed in the greenhouse can have a significant impact on the population build-up and dispersal levels of the beneficial released. Most studies on supplemental food start with laboratory tests and end with cage experiments, consisting of only a few plants in a limited space. Trials in whole greenhouses or greenhouse compartments are very rare. Our findings demonstrate that testing food applications on a more realistic scale and with different distribution strategies is important and can contribute to a more successful and cost efficient release of beneficials in greenhouses.
58
Food supplementation for Macrolophus pygmaeus
2.6 Conclusion
In order to successfully release M. pygmaeus in greenhouse sweet pepper crops, food must be supplemented homogenously across the greenhouse to ensure both quick dispersal and large populations. Artemia franciscana cysts proved to be a good food source for this. With this strategy a sufficiently large population of M. pygmaeus may be obtained early in the season to prevent outbreaks of aphids, thrips, whiteflies and other pests.
2.7 Acknowledgements
This research was financed by Research Centre Hoogstraten and the Agency Flanders Innovation & Entrepreneurship (VLAIO) in the context of research project 140948. The research project 140948 was granted to Research Centre Hoogstraten (RM) in cooperation with Research Station for Vegetable Production and University of Antwerp.
59
Chapter 2
2.8 Supplementary material
Figure S2-1. A DIBOX (Koppert) was used to release Macrolophus pygmaeus in the greenhouse. Each of the five consecutive plants of a release location received one DIBOX with approximately 25 bugs.
60
Food supplementation for Macrolophus pygmaeus
Figure S2-2. A Mini-Airbug (Koppert) was used to distribute supplemental food in the full field treatment. The Mini-Airbug blows food on the plants and by slowly walking past the plant rows at a continuous pace, food can be distributed evenly throughout the greenhouse. (Photo: Koppert)
Table S2-1. Comparison of nutrient composition of fresh Ephestia kuehniella eggs (mean water content 67.7%) and dry decapsulated Artemia franciscana cysts (strain from San Francisco Bay, CA; mean water content 3.0%). After De Clercq et al. (2005). Content (g per 100 g dry weight) Component Ephestia kuehniella eggs Artemia franciscana cysts Ash 3.1 4.5 Fiber 20.4 18.0 Protein 37.4 52.6 Fat 28.5 15.4 Carbohydrate 10.6 9.5
61
62
Chapter 3 | Biweekly supplementation with Artemia spp. cysts allows efficient population establishment by Macrolophus pygmaeus in sweet pepper
Nathalie Brenard, Vincent Sluydts, Evi Christianen, Lien Bosmans, Luc De Bruyn, Rob Moerkens & Herwig Leirs
Modified version of publication in: Entomologia Experimentalis et Applicata (2019), Vol. 167(5): 406-414. DOI: 10.1111/eea.12776
3.1 Abstract
Macrolophus pygmaeus Rambur (Hemiptera: Miridae) is a generalist natural enemy that is used to control multiple pest species in a variety of horticultural crops. The bugs are released at the start of the crop cycle to allow them to establish and build up a population in the crop that can control pest infestations later in the season. To facilitate population growth and dispersal in protected sweet pepper crops, Capsicum annuum L. (Solanaceae), food should be supplemented in a full field fashion during the first 6-8 weeks after introduction. To reduce the costs of food supplementation, we investigated whether fewer applications could produce similar results in terms of population growth and dispersal within the greenhouse. First, a cage experiment was carried out in which a weekly and biweekly application rate was tested for three food sources: cysts of brine shrimps Artemia spp. (Anostraca: Artemiidae), eggs of the Mediterranean flour moth, Ephestia kuehniella Zeller (Lepidoptera: Pyralidae), and a commercial mix of the two. Artemia spp. cysts resulted in the largest M. pygmaeus populations. There was no difference in population size between the two application rates for any of the food sources. Second, a greenhouse experiment was set up to test both application rates for
63
Chapter 3
Artemia spp. cysts in practical conditions. Again, no difference in population size was observed between a weekly and a biweekly application rate. This insight is good news for growers, as they can use the least expensive food source and they need fewer applications to successfully introduce M. pygmaeus in protected sweet pepper crops.
3.2 Introduction
Macrolophus pygmaeus Rambur (Hemiptera: Miridae) is a generalist predatory bug used in biological pest control in greenhouse crops such as tomato, eggplant, and sweet pepper (Perdikis & Lykouressis, 2002a; Castañé et al., 2011; Messelink & Janssen, 2014). Naturally occurring in the Palearctic region, it has been commercially reared for biological control since the early 1990s (Sanchez et al., 2012; EPPO, 2016). Macrolophus pygmaeus is used against a wide array of pest species, including western flower thrips Frankliniella occidentalis Pergande (Thysanoptera: Thripidae), aphids such as Myzus persicae (Sulzer) (Hemiptera: Aphididae), whiteflies such as Trialeurodes vaporariorum (Westwood) and Bemisia tabaci (Gennadius) (Hemiptera: Aleyrodidae), and moths such as Tuta absoluta (Meyrick) (Lepidoptera: Gelechiidae) (Perdikis & Lykouressis, 2002a; Blaeser et al., 2004; Urbaneja et al., 2009; Messelink et al., 2015; Bouagga et al., 2018b). Macrolophus pygmaeus is a zoophytophagous insect and will also feed on plant tissue, pollen, and nectar (Perdikis & Lykouressis, 2000), as plants provide the bugs with water and essential nutrients (Gillespie & McGregor, 2000; Hamdi & Bonato, 2013; Moerkens et al., 2016). In case of shortage or absence of prey, M. pygmaeus can rely solely on plant feeding and is thus able to sustain a population in the greenhouse. An exclusive plant diet, however, results in a lower reproductive rate and prolonged nymphal development (Perdikis & Lykouressis, 2000; Lykouressis et al., 2008). Because of its zoophytophagous behaviour and long generation time, M. pygmaeus is used preventively in biological control (Messelink et al., 2015). The bugs are introduced in the crop at the beginning of the growing season and build up a viable population that attacks emerging pests throughout the rest of the season. In order to facilitate population growth when pests are absent, food is supplemented in the crop. In greenhouse tomato crops, an optimal supplementation scheme consists of weekly food applications during 6-8 weeks, as this is the time it takes for a generation of M. pygmaeus to develop (Moerkens et al., 2017). In contrast to tomato, sweet pepper crops have pollen available for the bugs, but food supplementation is
64
Optimizing food supplementation strategy
still required to reach a sufficiently high population at the start of the season and to ensure quick dispersal the food should be supplemented in a full field fashion (Brenard et al., 2018; see Chapter 2). Although full field food supplementation in sweet pepper yields the best results, growers are not always inclined to apply this method due to higher food and labour costs. If similar results can be obtained with less frequent applications than once per week, growers may be persuaded more easily to supplement introduced M. pygmaeus populations in a full field fashion. In this study, a weekly and a biweekly food supplementation scheme were compared in protected sweet pepper crops. The weekly scheme resembles the optimal application rate in tomato, with weekly food supplementation during 7 weeks. In the biweekly scheme there were only four applications of supplemental food during the 7-week period. First, a screen cage experiment was conducted in which both schemes were tested for three types of food: (1) eggs of the Mediterranean flour moth, Ephestia kuehniella Zeller (Lepidoptera: Pyralidae), (2) cysts of brine shrimps, Artemia spp. (Anostraca: Artemiidae), and (3) a commercial mix of the two food types. Then, a trial was carried out to test the two application rates under real greenhouse conditions with only the food source that performed best in the cage experiment. Population sizes of M. pygmaeus were compared among treatments in both the cage trial and greenhouse trial. The effect of application rate on dispersal of M. pygmaeus was studied in the greenhouse.
3.3 Materials and methods
3.3.1 Cage experiment
Plants and climate conditions
Cage experiments were performed at the Greenlab facilities of Biobest (Westerlo, Belgium) in a greenhouse compartment of 100 m² under controlled climate conditions. Sweet pepper plants, Capsicum annuum L. (Solanaceae) cv. Maduro (Enza Zaden, Enkhuizen, The Netherlands), were planted in 3-l pots filled with potting soil as substrate. Plants were approximately 2 months old and 60 cm tall at time of planting. All 24 plants were placed in separate screen cages (1.2 m high, 40 cm diameter). Plants were kept at 24.2 ± 0.1 °C and 50.3 ± 0.3% relative humidity. During the first 2 weeks after planting, artificial light was provided
65
Chapter 3 from 05:00 to 16:00 hours with SON-T lamps (Philips, Deurne, The Netherlands). Plants were watered twice a week and fertilizer was added once a week via the water.
Release of Macrolophus pygmaeus
Two days after planting, M. pygmaeus late instars (N4-5) and adults (Macrolophus- System; Biobest) were released on the caged plants. A Macrolophus-System tube contains 500 M. pygmaeus individuals, of which 80-90% are adults and the rest are still N4-5 nymphs, and some vermiculite as substrate. Every plant received five nymphs and 20 adults with a sex ratio of 1:1. With a cut-off pipette tip connected to a tube, individual bugs were aspirated and placed in 75-ml transparent sampling containers. These containers, each containing 25 bugs, were then hung in the top of each caged plant.
Supplementary food applications
Three types of supplemental food were tested: decapsulated Artemia spp. cysts (Artemac; Biobest), UV-sterilized E. kuehniella eggs (Nutrimac; Biobest), and a commercial mixture of both, with a 5:1 Artemia spp. to E. kuehniella ratio (Nutrimac-Plus; Biobest). Each type was tested at two application rates: weekly with a total of seven applications, and biweekly with a total of four applications. This resulted in a total of six treatments, each repeated on four individually caged plants. At each supplementary food application, a plant received 0.04 g of food according to its treatment. In a previous study, 0.04 g of supplemental food per plant, distributed by means of a Mini-Airbug (Koppert, Berkel en Rodenrijs, The Netherlands), was sufficient to establish a M. pygmaeus population in a sweet pepper greenhouse (Brenard et al., 2018; see Chapter 2). The correct amount of food was placed on a Petri dish and carefully blown from the dish onto the plant. The first instance of food supplementation was at the time of M. pygmaeus release for all treatments.
Sampling method
Sampling started 2 weeks after release, in order to give the bugs time to settle, mate, and lay their first eggs. After that, M. pygmaeus numbers were recorded twice a week by visually checking the whole plants. The individuals in each of three age categories were counted: instars 1-3 (N1-3), instars 4-5 (N4-5), and adults. The experiment lasted until 2 weeks after the last food application.
66
Optimizing food supplementation strategy
3.3.2 Greenhouse experiment
Crop and climate conditions
The trial was carried out in a semi-commercial sweet pepper greenhouse at Research Centre Hoogstraten (Belgium). Five greenhouse compartments (A-E) were used for this experiment, with sizes ranging from 300 to 1 500 m² (Table 3-1). The compartments were 7 m high and equipped with a gutter grow system (Metazet FormFlex, Wateringen, The Netherlands). Climate conditions in each compartment were automatically logged every 5 min and registered by means of a Priva Electronic Measuring Box (Priva, De Lier, The Netherlands). Mean temperature and relative humidity in each compartment during the period from M. pygmaeus release until the end of sampling are shown in Table 3-1.
Sweet pepper plants were planted on a rockwool substrate (Saint-Gobain Cultilene, Rijen, The Netherlands) in the greenhouse on 6 December 2017. The distance between plants was 32 cm and plant density was 2.37 plants m–². Each plant had three stems, which resulted in a stem density of 7.1 stems m–². This stem density is common practice in commercial Belgian greenhouses. Several sweet pepper varieties were planted in the greenhouse compartments as they were required for a study on variety comparisons later in the season. The main varieties included Maduro, Frazier (Enza Zaden), and Sardinero (De Ruiter, Bergschenhoek, The Netherlands). Composition of the varieties differed between compartments. A single fertilization scheme suggested by a crop advisor, representative for commercial practice, was used in all compartments. Plants were checked and found to be pest free prior to the experiment. There were no other biological control agents released prior to or during the experiment. No plant protection products were used in any of the greenhouse compartments.
Table 3-1. Characteristics of the greenhouse compartments. Temperature (T) and relative humidity (r.h.) during the period from M. pygmaeus release (14/12/2017) until the end of sampling (20/03/2018). Compartment Surface (m²) Mean (± SE) T (°C) Mean (± SE) r.h. (%) Treatment A 300 20.2 ± 0.05 74.2 ± 0.04 Weekly B 1500 19.1 ± 0.02 74.8 ± 0.06 Biweekly C 500 19.5 ± 0.02 72.6 ± 0.05 Weekly D 500 20.2 ± 0.02 71.2 ± 0.05 Biweekly E 500 18.9 ± 0.02 80.5 ± 0.04 Weekly
67
Chapter 3
Release of Macrolophus pygmaeus
Macrolophus pygmaeus (Mirical; Koppert) was released on 14 December 2017. Mirical contains a wood chip substrate and 500 M. pygmaeus individuals, of which approximately 90% are adults and 10% are N4-N5 instars. Four release locations of five consecutive plants were chosen in the 300 and 500 m² greenhouse compartments. The contents of one Mirical tube were equally distributed among those four locations. To this end, the Mirical was divided into four equal parts after gently mixing the contents. Each quarter was then equally distributed over the five plants in a release location. Insects and substrate were placed in a DIBOX (Koppert) which was hung on the petiole of a lower leaf on each of the release plants. Twelve release locations were chosen in the 1 500 m² greenhouse compartment and three Mirical tubes were used.
Supplementary food applications
Two treatments of supplementary food applications were tested. In the first treatment, supplementary food was provided 7×: at the time of M. pygmaeus release and then weekly during the next 6 weeks. In the second treatment, food was provided only 4×: also at the time of release and then 3× at 2-week intervals. Hereafter the treatments will be referred to as the weekly and biweekly treatment, respectively. The weekly treatment was carried out in compartments A, C, and E, whereas the biweekly treatment was carried out in B and D (Table 3-1).
Decapsulated Artemia spp. cysts (Artefeed; Koppert) were given as a supplementary food source in both treatments. Food was supplied in a full field fashion using a Mini-Airbug (Koppert). Every compartment received 0.04 g cysts per plant (0.09 g m–²) at each supplementation occasion. During the course of the experiment, a total of 0.28 g per plant (0.63g m–²) was added to compartments of the weekly treatment and 0.16 g per plant (0.36 g m–²) to compartments of the biweekly treatment.
Sampling method
The M. pygmaeus population was sampled weekly from 22 December 2017 until 20 March 2018. In each of the small compartments (300-500 m²), two sample units were created, whereas four units were chosen in the large (1 500 m²) compartment. Each unit consisted of two sample plots of 10 consecutive plants. One of the sample plots in every unit corresponds
68
Optimizing food supplementation strategy
to a release location of M. pygmaeus and is hereafter referred to as release plot. The other plot was selected two plant rows further, with a diagonal distance of 8.4 m from the nearest release plot. These plots are called dispersal plots throughout the text as they give an indication of the predatory bug’s dispersal rate. To ensure a diagonal distance of more than 8.4 m between a dispersal plot and the release plot of another unit, no more than four sample plots fit in the small greenhouse compartments. In total, six units were created for each of the two treatments.
Every week, two flowers were sampled on every plant in each plot. No leaves were sampled, as a previous experiment showed that during the first months after introduction, more than 90% of M. pygmaeus reside in the flowers (Brenard et al., 2018; see Chapter 2). Occasionally there were not enough flowers present and fewer flowers were sampled. Bugs in each of three age categories were counted: instars 1-3 (N1-3), instars 4-5 (N4-5), and adults.
3.3.3 Statistical analysis
The effect of various food types and application rates on M. pygmaeus population size in the cage experiment was assessed in the first generation. All life stages were summed to obtain a total count on each sampling day. Peak abundance of the first generation was then used for analyses. A generalized linear model (GLM) with a Poisson error distribution was constructed, handling ‘food type’ and ‘application rate’ as fixed effects and ‘M. pygmaeus counts’ as response variable. To determine the effect of application rate on population size in the greenhouse experiment, the number of individuals of the first and second generation were compared. The first two generations of an introduced M. pygmaeus population are very distinct in time with hardly any overlap, which makes comparison feasible. For each treatment, the peak abundances of each generation were chosen as representative. Again, counts of the various M. pygmaeus life stages were summed prior to the analysis. Per sample plot, the total counts on all flowers were summed to avoid an excess of zeros in the data. Counts in a release plot and its corresponding dispersal plot were then grouped to obtain an overall abundance for each unit. Population size was analysed by means of a generalized linear mixed model (GLMM) with a Poisson error distribution or a negative binomial distribution when overdispersion occurred. ‘Treatment’, ‘generation’, and their interaction were used as fixed effects and ‘M. pygmaeus
69
Chapter 3 counts’ as response variable. ‘Greenhouse compartment’ was added as random factor, so the differences in sweet pepper variety composition are also accounted for. The number of flowers in each unit was used as an offset. The effect of food application rate on dispersal rate in the greenhouse experiment was based on the peak abundance of the second generation. Again, the total number of M. pygmaeus individuals was used. A GLMM was constructed with residuals assumed to follow a Poisson distribution or a negative binomial distribution in case of overdispersion. ‘Macrolophus pygmaeus counts’ were used as response variable and the interaction between ‘treatment’ and ‘plot type’ (release or dispersal) as fixed effect. ‘Greenhouse compartment’ was added as a random factor and the number of flowers as offset. Fully parameterised models were constructed initially, followed by sequential dropping of non-significant interactions and factor effects until the significance level reached 0.05 or less by using a log-likelihood ratio test. Statistical analyses were carried out in R v.3.5.1 (R Core Team, 2018), using package lme4 (Bates et al., 2015).
3.4 Results
3.4.1 Cage experiment
The cage experiment lasted for 56 days, during which the first generation developed completely in all treatments and some nymphs of the second generation already emerged (Figure 3-1). The peak abundance of the first M. pygmaeus generation occurred 20 days after release in all treatments except the weekly Artemia spp. treatment (25 days after release). A significant effect of food type was found on the population size of M. pygmaeus, whereas there was no effect of application rate (Table 3-2). Artemia spp. cysts yielded higher population sizes than E. kuehniella eggs (P = 0.001) and the mix Nutrimac-Plus (P = 0.002; Figure 3-2). There was no difference between the populations fed with E. kuehniella eggs and those fed with Nutrimac-Plus (P = 0.90).
70
Optimizing food supplementation strategy
Table 3-2. Log-likelihood ratio test (χ²) of the generalized linear model (GLM). Population size cage
χ² d.f. P
Application rate 0.736 1 0.39
Food type 14.676 2 0.001
Food type*Application rate 0.233 2 0.89
Figure 3-1. Mean (± SE) number of individuals per plant of Macrolophus pygmaeus life stages [N1-N3: nymphal stages 1-3, N4-N5: nymphal stages 4-5, adults, and their sum (total population)] under the six treatments of the cage experiment: (A) weekly Artemia spp. cysts, (B) biweekly Artemia spp. cysts, (C) weekly Ephestia kuehniella eggs, (D) biweekly E. kuehniella eggs, (E) weekly Nutrimac-Plus, and (F) biweekly Nutrimac-Plus.
71
Chapter 3
Figure 3-2. Mean (± SE) number of first-generation Macrolophus pygmaeus individuals found on sweet pepper plants under the two application rates (weekly or biweekly) for each of three food sources (Artemia spp. cysts, Ephestia kuehniella eggs, and their mixture, Nutrimac-Plus) in the cage experiment.
3.4.2 Greenhouse experiment
Two generations of M. pygmaeus developed during the course of the experiment, which lasted for approximately 13 weeks. On 9 January 2018, 26 days after release, the first generation reached its highest abundance (Figure 3-3). For the second generation, this occurred on 14 March 2018, which was 100 days after release. Peak abundances occurred at the same time in both treatments. The analysis indicated no significant effect of food application rate on the population size of M. pygmaeus (Table 3-3, Figure 3-4). Population numbers of the second generation were significantly higher than those of the first generation (Table 3-3). There was no effect of food application rate on the dispersal rate of M. pygmaeus (Table 3-3, Figure 3-5).
72
Optimizing food supplementation strategy
Table 3-3. Log-likelihood ratio tests (χ²) of the generalized linear mixed models (GLMMs). Population size greenhouse Dispersal rate
χ² d.f. P χ² d.f. P
Treatment 0.021 1 0.88 - - -
Generation 7.579 1 0.006 - - -
Treatment*generation 0.015 1 0.90 - - -
Treatment*plot type - - - 0.075 1 0.78
Figure 3-3. Mean (± SE) number of individuals per flower of Macrolophus pygmaeus life stages [N1-N3: nymphal stages 1-3, N4-N5: nymphal stages 4-5, adults, and their sum (total population)] in sweet pepper greenhouse compartments under two application rates of Artemia spp. cysts, (A) weekly and (B) biweekly.
73
Chapter 3
Figure 3-4. Mean (± SE) number of Macrolophus pygmaeus individuals found in sweet pepper flowers at two generation peaks under two Artemia spp. cysts application rates (weekly and biweekly) in the greenhouse experiment.
Figure 3-5. Mean (± SE) number of second-generation Macrolophus pygmaeus individuals found in sweet pepper flowers in release and dispersal plots under two Artemia spp. cysts application rates (weekly and biweekly) in the greenhouse experiment.
74
Optimizing food supplementation strategy
3.5 Discussion
In the past, augmentative releases of specialists were favoured over generalist predators for biological control, because they would only attack the targeted pest species (Symondson et al., 2002). However, specialist natural enemies will die when the pest is eradicated and repeated introductions are required when new pest outbreaks occur (Messelink et al., 2011b). Since the 1990’s, generalist predators have gained favour. Anthocorid bugs such as Orius spp. (Hemiptera: Anthocoridae), mirids such as M. pygmaeus and Nesidiocoris tenuis Reuter (Hemiptera: Miridae), phytoseiid mites such as Amblyseius swirskii (Athias-Henriot) (Acari: Phytoseiidae), and many other species are now used on a large scale in pest management (Weintraub et al., 2011; Messelink & Janssen, 2014; Janssen & Sabelis, 2015; Bouagga et al., 2018a). The advantage of generalists is that they can suppress populations of several pest species at the same time and can remain in the crop by switching to alternative prey or plant-provided food sources when pest levels are low (Symondson et al., 2002). Therefore, they require only a single introduction at the beginning of the season and can prevent outbreaks as they are already present in the crop when pests arrive. Most of these generalists receive food supplementation during the first weeks after introduction to boost their population numbers, as prey is scarce or absent at that time (Nomikou et al., 2002; Urbaneja et al., 2005; Janssen & Sabelis, 2015; Messelink et al., 2015). In greenhouse tomato and sweet pepper crops, M. pygmaeus ideally receives food supplementation during 6-8 weeks after introduction (Moerkens et al., 2017; Brenard et al., 2018 [see Chapter 2]). Dispersal is enhanced when food is applied in a full field fashion (Put et al., 2012; Brenard et al., 2018 [see Chapter 2]). The two most common food sources for supplementation are E. kuehniella eggs and Artemia spp. cysts. Ephestia kuehniella eggs have a higher nutritional value than Artemia spp. cysts, containing lower or similar amounts of protein, but similar amino acid patterns and up to 3× more fatty acids (De Clercq et al., 2005; see Table S2-1 in Chapter 2). Biochemical analyses on M. pygmaeus carcasses by Vandekerkhove et al. (2009) indicated that amino acid contents did not differ between M. pygmaeus adults raised on E. kuehniella eggs and those raised on Artemia spp. cysts. However, fatty acid contents did differ between the two groups with Artemia-fed M. pygmaeus showing deficiencies in linoleic, palmitic, and oleic acids. Fatty acids are necessary as signal molecules and as elements in membrane phospholipids besides serving
75
Chapter 3 as energy storages (Vandekerkhove et al., 2009). In the cage experiment, food supplementation with Artemia spp. cysts led to higher population sizes of M. pygmaeus on sweet pepper plants compared to a diet of E. kuehniella eggs or a mixed diet (Nutrimac-Plus), despite the food source’s lower nutritional value. A potential explanation for this is that E. kuehniella eggs are more prone to fungal contamination. After a few days the eggs start to turn bad and become unavailable as a food source for the predatory bugs. Artemia spp. cysts are known to have a longer shelf life (Messelink, 2016) and also remained fresh longer in our experiment. The Nutrimac-Plus mix contains more Artemia cysts than E. kuehniella eggs, but still did not perform well in our experiments. In part, it is likely to suffer from the same problem as the pure E. kuehniella diet. The fact that Artemia spp. cysts are nutritionally inferior to E. kuehniella eggs is undoubtedly of much importance in mass rearing multiple generations of predators, as was shown by De Clercq et al. (2005). The difference in nutritional value will be less important when the cysts only serve as a supplemental food source to a single generation in the greenhouse, where plants and eventually other prey are available for feeding. Different application rates did not yield differences in population sizes for any of the three food sources. The same result was obtained from the greenhouse experiment, where a weekly and a biweekly application of Artemia spp. cysts resulted in similar M. pygmaeus population sizes and dispersal levels. Biweekly application during the food supplementation period will lower the costs for growers as they can buy less food and have lower labour costs. Moreover, Artemia spp. cysts cost less (€25 compared to €750 for 500 g in Belgium) and have a longer shelf life than E. kuehniella eggs (De Clercq et al., 2005; Messelink, 2016; Vangansbeke et al., 2016; pers. comm. Rob Moerkens, R&D Biobest, 2020). A treatment without food supplementation was not tested, because a study in the previous season already showed that M. pygmaeus population numbers stayed very low when no alternative food source was provided (Brenard et al., 2018; see Chapter 2). Although the initial density in compartment A (weekly treatment) was higher than in the other compartments with 500 individuals released in 300 m² instead of 500 or 1 500 individuals in 500- and 1 500-m² compartments, respectively, it did not result in a larger M. pygmaeus population during the course of the experiment. Artemia spp. cysts from different companies were used in this study, Artemac (Biobest) for the cage experiment and Artefeed (Koppert) for the greenhouse experiment. Previous research indicated that there may be a considerable difference in quality between
76
Optimizing food supplementation strategy
cysts from different sources and locations (De Clercq et al., 2005; Vangansbeke et al., 2016). Artemia spp. cysts of lower quality may lead to lower absolute M. pygmaeus numbers compared to higher-quality cysts. The origin of the commercial products used here is not known. However, the aim of this research was to compare two application rates and not absolute M. pygmaeus population numbers and both the cage and greenhouse experiment produced the same results. A cheaper release strategy for M. pygmaeus in sweet pepper should encourage growers to use the predatory bug in their greenhouses. Sweet pepper cultivation in Europe is known for its extensive and successful use of biological control in pest management programs (Ramakers, 2004; van der Blom et al., 2008). Most pests are successfully controlled by releasing natural enemies, but aphids remain a problem for which either very expensive repeated releases or chemical interventions are required (Bloemhard & Ramakers, 2008; Sanchez et al., 2011). Macrolophus pygmaeus preys on aphids and its potential on sweet pepper has been shown in several studies (Messelink et al., 2013; Messelink & Janssen, 2014; De Backer et al., 2015; Pérez-Hedo & Urbaneja, 2015). More studies under commercial greenhouse conditions should be performed to develop an efficient method for aphid control with M. pygmaeus so the use of chemicals in sweet pepper can be further reduced.
3.6 Conclusion
Our study shows that a biweekly supplemental food scheme with Artemia spp. cysts during 6 weeks after release suffices for the successful introduction of M. pygmaeus in sweet pepper greenhouses at the start of the crop season. This insight may reduce the costs for growers as they not only can use the cheapest food source, but they also need fewer applications to obtain benefits.
77
Chapter 3
3.7 Acknowledgements
This research was financed through two projects, the first (PeMaTo – nr. 140948) by the Agency Flanders Innovation & Entrepreneurship (VLAIO), granted to Research Centre Hoogstraten (RM) in cooperation with Research Station for Vegetable Production and the University of Antwerp, and the second (PeMaTo-EuroPep – nr. 160427) by VLAIO (C-IPM), granted to RM in cooperation with Research Station for Vegetable Production, Wageningen University and Research, University of Antwerp, and the Andalusian Institute of Agricultural and Fisheries Research and Training. The cage experiment was carried out by EC at Greenlab, Biobest, in function of a bachelor thesis (Thomas More, Campus Geel, Belgium).
78
79
80
Chapter 4 | Is leaf pruning the key factor to successful biological control of aphids in sweet pepper?
Nathalie Brenard, Lien Bosmans, Herwig Leirs, Luc De Bruyn, Vincent Sluydts & Rob Moerkens
Modified version of publication in: Pest Management Science (2020), Vol. 76: 676-684. DOI 10.1002/ps.5565
4.1 Abstract
Aphids (Hemiptera: Aphididae) are a problematic pest in global sweet pepper cultivation. Control of aphids often relies on insecticides, leading to widespread resistance. Biological control of aphids is mainly based on releasing specialist natural enemies, but they often fail to control outbreaks. Macrolophus pygmaeus Rambur (Hemiptera: Miridae) is a zoophytophagous generalist which attacks several sweet pepper pests, including aphids. Previous research showed that M. pygmaeus is capable of strongly reducing aphid populations in sweet pepper, but complete control was seldom achieved. Sweet pepper plants continue to grow during the season, reaching more than 3 m high in Belgian and Dutch greenhouses. Dense foliage and large vertical distance from the flowers to the lower leaves impede the search efficiency of the predator. Leaf pruning may improve aphid predation by M. pygmaeus by increasing the probability of encountering prey. Four and five treatments (foliage range: 100 cm to full length) respectively were tested in a semi-commercial sweet pepper greenhouse in 2017 and 2018. Aphid populations in pruned treatments grew more slowly than in the control and M. pygmaeus was eventually able to control aphids in all pruned treatments in 2018. There was no difference in aphid control between the pruned treatments. Sweet pepper production was lower in the treatments with the shortest foliage lengths. Leaf pruning up to 160 or 190 cm foliage length improves aphid control by M. pygmaeus in sweet pepper without affecting production.
81
Chapter 4
4.2 Introduction
Aphids (Hemiptera: Aphididae) are a problematic pest in sweet pepper cultivation all over the world (Bloemhard & Ramakers, 2008; Gillespie et al., 2009; Sanchez et al., 2011; Matsuura & Nakamura, 2014; Mdellel & Ben Halima Kamel, 2014). They can cause severe crop damage, directly through plant feeding and indirectly through the production of honeydew and the transmission of plant viruses (Sanchez et al., 2011). Honeydew on sweet pepper leaves disrupts photosynthesis and respiration and serves as a substrate for sooty molds, which further disrupt photosynthesis (Ebert & Cartwright, 1997; Singh & Singh, 2016). Honeydew on the fruits has to be cleaned before selling, costing extra time and money (Bloemhard & Ramakers, 2008). Aphid outbreaks occasionally result in severe yield losses in sweet pepper cultivation (Bloemhard & Ramakers, 2008; Gillespie et al., 2009).
Two of the most problematic aphid species that colonize protected sweet pepper crops worldwide are the green peach or peach-potato aphid Myzus persicae Sulzer and the foxglove or glasshouse-potato aphid Aulacorthum solani (Kaltenbach) (van Lenteren et al., 1997; van Schelt, 1999; Blackman & Eastop, 2000; Sanchez et al., 2007, 2011). Myzus persicae is highly cosmopolitan and the most economically important aphid pest worldwide, infesting over 400 plant species (Blackman & Eastop, 2000). More than 100 plant viruses are transmitted by M. persicae, making its function as a vector one of its most problematic features, together with honeydew excretion and its consequences (Blackman & Eastop, 2000). Aulacorthum solani has a near worldwide distribution nowadays (Miller et al., 2009) and has been reported on at least 95 plant species (Kim et al., 1991 as cited in Jandricic et al., 2013). At least 45 plant viruses are transmitted through A. solani (Chan et al., 1991 as cited in Miller et al., 2009). In contrast to M. persicae, most damage caused by this aphid species arises as a direct result of its plant feeding behaviour: its salivary secretions are highly toxic to sweet pepper plants and result in yellow leaf veins, leaf deformation and necrosis and spotted fruits even at low population densities. At high densities they cause defoliation and stop plants from producing (Sanchez et al., 2007; Bloemhard & Ramakers, 2008).
Control of aphids often relies on the extensive use of insecticides which lead to widespread resistance against many active ingredients (Bass et al., 2014). The development of
82
Aphid control by Macrolophus pygmaeus
multiple forms of resistance in aphids is facilitated by their short generation time and high reproductive rate enabled by parthenogenesis (Bass et al., 2014). Myzus persicae is one of the most resistant pest species in the world, having acquired resistance against most insecticide classes, including neonicotinoids, pyrethroids, organophosphates, carbamates and cyclodienes (Devonshire et al., 1998; Nauen & Elbert, 2003; Bass et al., 2014; Mottet et al., 2016). Aside from pest resistance, problems with residues and environmental contamination also arise when using insecticides. The demand for non-chemical pest management is further driven by more strict regulations on maximum residue levels (MRL) (Andriukaitis, 2018) and stronger consumer awareness. Aphids are also biologically controlled through the introductions of natural enemies in the greenhouse (Sanchez et al., 2011; Prado et al., 2015). Several biological control agents have been commercially reared and sold for decades (van Lenteren, 2012). Parasitoid wasps such as Aphidius colemani Viereck, Aphidius ervi Haliday and Aphidius matricariae Haliday (Hymenoptera: Braconidae) and Aphelinus abdominalis (Dalman) (Hymenoptera: Aphelinidae) are introduced in greenhouses to control M. persicae, A. solani and other aphid species (Acheampong et al., 2012; van Lenteren, 2012; Prado et al., 2015). Female wasps oviposit in aphids, using them as hosts for the developing larvae and turning them into so-called mummies. The gall midge Aphidoletes aphidimyza Rondani (Diptera: Cecidomyiidae) is a predator that has been used in the biological control of aphids since 1973 (Markkula et al., 1979 as cited in Havelka & Zemek, 1999). Females of A. aphidimyza will oviposit near aphid colonies after which the hatched larvae will prey on the aphids (Jandricic et al., 2013; Boulanger et al., 2018).
All of these natural enemies are aphid specialists that are unable to survive or reproduce once the aphid population is eradicated (Messelink et al., 2011b). Because of this, repeated releases with high numbers are required throughout the year, making it costly for growers (Bloemhard & Ramakers, 2008). Moreover, the success of these specialists in practice is very variable and growers are often forced to turn to insecticides (Bloemhard & Ramakers, 2008; Gillespie et al., 2009; Messelink, 2014). Organic growers can’t use synthetic insecticides and there is only a limited number of biological insecticides available. These growers occasionally suffer severe fruit contamination and yield losses due to aphid infestations (Bloemhard & Ramakers, 2008).
83
Chapter 4
These obstacles in biological control led researchers to look for other natural enemies that can be introduced in the greenhouse prior to aphid infestations and will remain in the crop after the aphids have been eradicated. The predatory bug Macrolophus pygmaeus Rambur (Hemiptera: Miridae) is a natural enemy that meets these requirements. It is a generalist zoophytophagous predator and will attack a range of pest species in sweet pepper, including aphids (Perdikis & Lykouressis, 2002a, 2004; Messelink et al., 2011b; De Backer et al., 2015; Bouagga et al., 2018a). When pests are scarce or absent, it will turn to plant feeding and thus sustain a population that can prevent pest outbreaks (Perdikis & Lykouressis, 2000; De Backer et al., 2015). Introducing M. pygmaeus at the beginning of the season accompanied by a correct food supplementation strategy is sufficient to establish a population in sweet pepper greenhouses (De Backer et al., 2015; Brenard et al., 2018, 2019 [see Chapters 2 & 3]). Previous research has shown M. pygmaeus is capable of strongly reducing aphid population growth in sweet pepper, but complete control has not always been achieved (Messelink et al., 2011b; Messelink & Janssen, 2014; De Backer et al., 2015).
The difficulty in practical conditions is that sweet pepper plants continue to grow during the whole crop season, reaching more than 3m high in Belgian and Dutch greenhouses with a leaf area index (LAI) of up to 8 (Dueck et al., 2006). Dense foliage and large vertical distance from the flowers, where M. pygmaeus feeds on thrips and pollen, to the lower leaves can hamper search efficiency and predation of aphids. Leaf pruning on the lower parts of sweet pepper plants will reduce LAI and force aphids to occupy higher parts of the plants, which may improve search efficiency and predation by M. pygmaeus. Currently, leaf pruning is not common practice in protected sweet pepper cultivation.
In this study, the effect of leaf pruning on aphid predation by M. pygmaeus in a protected sweet pepper crop was tested during two consecutive seasons. Four and five leaf pruning heights respectively, were tested under commercial greenhouse conditions. We hypothesized that aphid populations will grow more slowly and reach a lower maximum when plant foliage is kept shorter because this will improve search efficiency and thus predation by M. pygmaeus. In addition, sweet pepper yield was measured and compared to estimate the effect of leaf pruning on sweet pepper production.
84
Aphid control by Macrolophus pygmaeus
4.3 Material and methods
4.3.1 Crop and climate conditions
A semi-commercial sweet pepper greenhouse compartment with a surface area of 300 m² at Research Centre Hoogstraten, Belgium was used for this experiment in the cropping seasons of 2017 and 2018. The compartment was 7 m high and was equipped with a gutter growing system (FormFlex/Metazet, Wateringen, the Netherlands) for planting and a Priva Electronic Measuring Box (Priva, De Lier, the Netherlands) that logged and registered climate conditions every 5 minutes. Mean temperature and relative humidity during the harvest period are shown in Table 4-1.
Sweet pepper Capsicum annuum L. (Solanaceae) variety Maduro (Enza Zaden, Enkhuizen, the Netherlands) sowing and planting dates are shown in Table 4-1. Plants were planted in the greenhouse using rockwool (Cultilene, Rijen, the Netherlands) as a substrate. The greenhouse compartment has 13 plant rows of 17m long, each containing 52 sweet pepper plants. Distance between plants was 32 cm, with 2.37 plants m-². Three stems were grown per plant, resulting in a stem density of 7.1 stems m-², which is common in commercial Belgian greenhouses. The two side rows are close to the wall and hold fewer plants and stems, so they were not used in the experiments. The centre row was not used either as the plants there receive less light because it encompasses the supporting pillars of the greenhouse. A similar fertilization scheme was followed in both years, advised by a crop advisor and representative for practical conditions.
85
Chapter 4
Table 4-1. Climate and sweet pepper crop parameters in the greenhouse compartment during the two seasons of the experiment. Season 2016-2017 2017-2018 Sowing date 21 October 2016 19 October 2017 Planting date 7 December 2016 5 December 2017 Start harvest 21 March 2017 3 April 2018 End harvest 29 August 2017 25 September 2018 Mean temperature (°C) during harvest 23.2 (15.2-34.6) 22.6 (12.8-35.7) (min-max) Mean relative humidity (%) during harvest 78.6 (42.2-99.0) 78.4 (25.0-100.0) (min-max)
4.3.2 Release of Macrolophus pygmaeus
Macrolophus pygmaeus (Mirical, Koppert Biological Systems, Berkel en Rodenrijs, the Netherlands) was released in the greenhouse compartment on 16 December in 2016 and 14 December in 2017. A Mirical tube contains 500 M. pygmaeus individuals of which most (± 90%) are adults and the rest are N4-N5 nymphal instars. One tube was used for the compartment, equally dividing the contents over four release locations, each consisting of five consecutive plants. A DIBOX (Koppert Biological Systems) hung on the petiole of a lower leaf was used to release the mirids on the sweet pepper plants. Macrolophus pygmaeus received full field food supplementation of 0.09g m-² Artemia spp. cysts (Artefeed, Koppert Biological Systems) on the day of release and then weekly for another 6 weeks, according to the protocol as described in Brenard et al. (2018; see Chapter 2) Food was distributed throughout the greenhouse by means of a mini-airbug (Koppert Biological Systems).
4.3.3 Leaf pruning
In 2017, three different leaf pruning treatments were applied in which all leaves, starting at the base of the plant, were manually pruned (removed from the plant) until a certain length of foliage (160, 130 and 100 cm) remained on the plant (Figure S4-1). A fourth treatment in which no leaf pruning occurred was used as a control, and reached a foliage length of 250- 300 cm by the end of the season. Each of the treatments was repeated in four plots, spread
86
Aphid control by Macrolophus pygmaeus
over eight plant rows. All eight plant rows held two plots, one covering the first half of the row and the other the second half, each consisting of 26 plants. Plot locations in the greenhouse were randomized. In 2018, five treatments were tested, adding pruning up to 190 cm as a treatment. A total of 20 plots was created on 10 plant rows using the same method as the year before. Leaf pruning started when the plants reached their treatment length: pruning in the 100 cm treatment started first in March, 130 cm a few weeks later and so on. Every two weeks, leaves were pruned according to the treatment and LAI was measured in every plot with an LAI-2200C Plant Canopy Analyzer (LI-COR, Lincoln, NE, USA). Light intensity was measured above the canopy and ten times at the bottom of the plant after which an average LAI was calculated.
4.3.4 Effect leaf pruning on aphid control
Aphids did not come naturally into the greenhouse in either year. An infestation of A. solani emerged in one of the other sweet pepper compartments at the Research Centre in 2017 and these aphids were released into the study compartment (Table 4-2). A sweet pepper leaf from the other compartment with approximately 100 aphids, was placed in the centre part of every plot. The leaf was hung between the lower leaves of a plant. It took three weeks of sampling before the first aphids were counted and they were found on only three leaves. Therefore, a second release took place four weeks later. A similar release method was used in 2018, but with M. persicae from a laboratory culture on young sweet pepper plants as no natural infestations occurred that year. The first release took place in May 2018, but this time two leaves carrying 100 aphids were hung on two separate plants in the centre of each plot. Again, no aphids were found in the following weeks and a second release according to the method of 2017 took place four weeks later (Table 4-2).
87
Chapter 4
Table 4-2. Events and interventions regarding the aphid populations during both experiments. Season 2016-2017 2017-2018 Aphid release 1 26 June 2017 22 May 2018 Aphid release 2 24 July 2017 19 June 2018 Start sampling 3 July 2017 12 June 2018 First aphids counted 18 July 2017 26 June 2018 Pirimicarb intervention 1 27 September 2017 11 July 2018 Pirimicarb intervention 2 5 October 2017 19 July 2018 End sampling 9 October 2017 22 August 2018
In 2017, weekly sampling of aphid and M. pygmaeus numbers was done from July to October (Table 4-2). The ten plants in the centre of every plot were sampled on each occasion, leaving eight plants on every side to minimize edge effects. Aphids and M. pygmaeus were counted on three randomly chosen leaves, one in the lower, middle and upper part of the foliage and two flowers of each plant. M. pygmaeus counts were categorized in three age classes: early nymphal stages (N1-N3), late nymphal stages (N4-N5) and adults. Originally, no chemical interventions were planned, but when aphids caused severe honeydew contamination of fruits in the control plots, these plots and plant rows not part of the experiment were treated with pirimicarb (Pirimor, 50% WG, Syngenta Crop Protection N.V., Bergen op Zoom, the Netherlands) at a rate of 0.32 g ha-1 . At this point, aphid densities in the control plots were so high that mass migration into other plots had occurred after which the population densities rapidly increased there as well. To avoid migration from the control plots in 2018, a threshold was set at an average aphid density of 10 aphids leaf-1. A chemical intervention with Pirimor (as mentioned above) was performed when average aphid density exceeded this threshold. All control plots and the three rows that were not part of the experiment were then sprayed. Sampling in 2018 started on 12 June and lasted until the aphid populations in the pruned plots were controlled. The same sampling method was used.
88
Aphid control by Macrolophus pygmaeus
4.3.5 Effect leaf pruning on yield
The effect of leaf pruning on yield was measured in all treatments on the 20 plants in the centre of each plot and this for the full harvesting season. The start and end dates of the harvest period are shown in Table 4-1. Harvesting happened three times per fortnight: on Wednesday in the first week and on Tuesday and Friday in the second week. All harvested sweet peppers were weighed individually. At the end of the season, the total number of peppers, total weight harvested per m² and average fruit weight were calculated for each treatment.
4.3.6 Statistical analyses
To test the effect of different leaf pruning heights on aphid control by M. pygmaeus, the aphid population density and growth rate were compared between treatments. Growth rates for the aphid populations were determined for the period of exponential growth before the first chemical intervention. Growth rates were estimated using a linear mixed model with a normal error distribution. Log transformed aphid counts were used as the response variable and regressed against time (in days) and treatment (Vehrs et al., 1992), whereas plot was added as a random effect to account for possible differences between plots. Growth rate was determined as the slope of the regression. Population densities on the last sampling date before the first intervention with pirimicarb were analysed. A generalized linear mixed model was constructed with treatment as a fixed effect, plot as a random effect and aphid counts as the response variable. Counts on leaves were summed per plant prior to analyses and the number of leaves was used as an offset. A Poisson error distribution was used and in case of overdispersion, a negative binomial distribution was preferred. To assess the effect of leaf pruning on production, two indicators of yield were analysed: total production during the whole season in kg m-² and average fruit weight. A one- way analysis of variance (ANOVA) was used to compare the effects of the different treatments on these indicators. To cope with heteroscedasticity, variance was set to differ between treatments. For all models, a log-likelihood ratio test (χ2) was used to determine if factor effects were significant (p ≤ 0.05). Post-hoc comparisons were performed with Tukey’s HSD test.
89
Chapter 4
Statistical analyses were carried out in R v.3.5.1 using packages ‘lme4’, ‘nlme’, ‘multcomp’ and ‘emmeans’ (Bretz et al., 2010; Bates et al., 2015; Pinheiro et al., 2018; R Core Team, 2019; Lenth et al., 2019).
4.4 Results
Important dates and interventions regarding the aphid populations in the experiments are given in Table 4-2. No chemical interventions were planned in 2017, but when aphid densities rose steeply to almost 55 aphids leaf-1 in control plots on 27 September, they were sprayed with pirimicarb.
Aphid densities under the different treatments in both years are shown in Figure 4-1. After the first aphids were counted in 2017, densities remained very low for another two weeks after which exponential growth became visible. In 2018, densities started rising exponentially immediately after the first aphid sightings. Figure 4-2 shows the vertical distribution of the aphids sampled during the experiments. Aphids were found on the lower, middle and higher leaves of the sweet pepper plants in all treatments during both years.
90
Aphid control by Macrolophus pygmaeus
Figure 4-1. Mean (± SE) number of aphids per leaf in the different treatments in 2017 and 2018, starting from the date the first aphids were counted (18 July 2017; 26 June 2018) until the end of the experiment. Vertical dashed lines mark pirimicarb sprays.
91
Chapter 4
Figure 4-2. Vertical distribution of aphids sampled on sweet pepper plants in the different treatments in 2017 and 2018.
Growth rates were determined for the period starting on the first sampling date where aphid densities differed from zero in all treatments, until the last sampling date before spraying with pirimicarb. For 2017, this period lasted from 10 August to 20 September and for 2018 it lasted from 26 June to 9 July. In both years, there was a significant interaction between treatment and time, meaning that the slopes and thus aphid population growth rates differed between treatments (2017: df = 3, χ² = 30.16, p < 0.0001; 2018: df = 4, χ² = 39.29, p < 0.0001). Figure 4-3 shows the growth rates for all treatments during the period of exponential growth before spraying with pirimicarb. In both years the control treatment had a significant higher growth rate compared to the experimental treatments (p = 0.0008 for 160 cm in 2017; all other p-values < 0.0001), while no differences were found between the experimental treatments. The growth rate in the control treatment was higher in 2018 compared to 2017 (Figure 4-3). Aphid densities were compared for the last sampling date before pirimicarb was sprayed and a significant difference between treatments was found for both years (2017: df = 3, χ² = 8.73, p = 0.03; 2018: df = 4, χ² = 18.07, p = 0.001). In 2017, aphid densities on leaves were significantly higher in control plots than in plots that were pruned up to 100 (p = 0.008) and 130
92
Aphid control by Macrolophus pygmaeus
cm (p = 0.050). Densities in plots with a foliage length of 160 cm did not differ from the control plots. No difference was found between the different pruning lengths (Figure 4-4). All four pruned treatments differed significantly from the control in 2018 (p = 0.0001, p = 0.001, p = 0.002 and p = 0.002 for 100, 130, 160 and 190 cm respectively). Again, no differences were found between the pruned treatments (Figure 4-4).
Figure 4-3. Mean (± SE) aphid growth rates in the different leaf pruning treatments in 2017 and 2018.
Figure 4-4. Mean (± SE) number of aphids per leaf in the different leaf pruning treatments on the last sampling date before chemical intervention with pirimicarb in 2017 and 2018.
93
Chapter 4
Average greenhouses densities of M. pygmaeus throughout the experiments are shown in Figure 4-5 (2017) and Figure 4-6 (2018). Densities in flowers decreased during summer in both years. In 2017, a maximum of ±2.5 bugs flower-1 was reached after two weeks after which the numbers decreased steadily. The maximum density in flowers was 1.2 bugs flower-1 in 2018 on the day the first aphids were counted. A week before the first chemical intervention in 2017, M. pygmaeus densities on leaves started to rise strongly until they reached 0.5 bug leaf-1 by the end of the experiment. In 2018 M. pygmaeus densities increased again on both flowers and leaves after the pirimicarb sprays, but this increase was much stronger on leaves. Densities on leaves decreased again by the end of the experiment. Population densities of M. pygmaeus were lower in 2018 compared to 2017.
Figure 4-5. Mean (± SE) number of individuals per flower and per leaf of Macrolophus pygmaeus life stages [N1-N3: nymphal stages 1-3, N4-N5: nymphal stages 4-5, adults, and their sum (total population)] in the whole greenhouse compartment in 2017, starting from the date the first aphids were counted (18 July 2017) until the end of the experiment. Vertical dashed lines mark pirimicarb sprays. Note y-axes have different scaling.
94
Aphid control by Macrolophus pygmaeus
Figure 4-6. Mean (± SE) number of individuals per flower and per leaf of Macrolophus pygmaeus life stages [N1-N3: nymphal stages 1-3, N4-N5: nymphal stages 4-5, adults, and their sum (total population)] in the whole greenhouse compartment in 2018, starting from the date the first aphids were counted (26 June 2018) until the end of the experiment. Vertical dashed lines mark pirimicarb sprays. Note y-axes have different scaling.
The analyses of total sweet pepper production showed a significant effect of leaf pruning height in 2018 (df = 4, χ² = 19.3, p = 0.0007; Figure 4-7.A), while there was no effect in 2017 (df = 3, χ² = 7.09, p = 0.07). Total production was higher in the 160 cm and 190 cm treatments compared to the shorter pruned treatments 100 cm (p = 0.018 and p = 0.009) and 130 cm (p = 0.006 and p = 0.003). No significant difference was found between the control and the pruned treatments. In 2017, there was a significant effect of leaf pruning height on fruit weight (df = 3, χ² = 7.66, p = 0.05; Figure 4-7.B): fruit weight in the 130 cm treatment was significantly lower than in the 160 cm treatment (p = 0.032). There was no significant difference between treatments in 2018 (df = 4, χ² = 7.4, p = 0.12).
95
Chapter 4
Figure 4-7. (A) Total sweet pepper production (mean ± SE) and (B) mean fruit weight (± SE) in 2017 and 2018 under the different pruning treatments.
LAI measurements were carried out every two weeks after pruning and results are shown in Table 4-3 to illustrate what the differences in pruning height cause in terms of light penetration.
Table 4-3. Leaf area index (LAI) measurements (mean ± SE) in the different leaf pruning treatments in both years. Treatment 2017 2018 Control 5.34 ± 0.30 5.36 ± 0.15 100 cm 3.50 ± 0.15 3.47 ± 0.12 130 cm 4.25 ± 0.31 3.99 ± 0.04 160 cm 4.68 ± 0.22 4.43 ± 0.06 190 cm - 4.82 ± 0.10
96
Aphid control by Macrolophus pygmaeus
4.5 Discussion
Our results show that leaf pruning improves aphid control by M. pygmaeus in greenhouse sweet pepper. The aphid population growth rates were significantly larger in control treatments compared to pruned treatments in both years. M. persicae (2018) had a higher growth rate in the control treatment than A. solani (2017), whereas growth rates in the other treatments are similar between the years and thus between the species. It is possible that the effect of lower M. pygmaeus densities in 2018 was stronger in control plots. Macrolophus pygmaeus was eventually able to control the aphid populations in all the pruned treatments in 2018. Aphids reached their maximum densities 10-11 weeks after the first introduction and declined to zero after 12 weeks. The growth rates obtained in our study are not comparable to the results from studies on the intrinsic rate of natural increase (rm) of M. persicae and A. solani on sweet pepper and other host plants (Jarosik, Honek, Lapchin, et al., 1996; Nikolakakis et al., 2003; Satar et al., 2008; Jandricic et al., 2010). The rm describes the exponential growth of an unlimited aphid population, without the effects of competition, natural enemies, limited food availability and so on. Furthermore, Guldemond et al. (1998) emphasize that it is important to know when the exponential growth phase is over to properly estimate rm. As we worked under practical conditions in a semi-commercial greenhouse, it was not possible to allow aphid densities to become so high.
To regulate high aphid densities in and avoid migration from the control plots, a chemical intervention with pirimicarb was performed. Pirimicarb belongs to the insecticide class of carbamates and is an active ingredient used in insecticides against aphids. It works on the nervous system as inhibitor of acetylcholinesterase (AChE), blocking neurotransmission which will lead to death (Nabeshima et al., 2003). Although it is regarded as a selective pesticide, it has some side effects on M. pygmaeus. Rahmani et al. (2016) found that for LC30, which they determined at 2013.4 mg L-1, pirimicarb prolongs pre-adult development and reduces fecundity in M. pygmaeus. The authors consider pirimicarb safe and without real short- term risks for M. pygmaeus at recommended concentrations. The recommended field concentration for aphid control in sweet pepper, which we used in our experiments, is 500 mg L-1 and thus much lower than the LC30 found in Rahmani et al. Pirimicarb was only sprayed locally in the control plots and plant rows not part of the experiment as discussed in the Material and Methods. Because of its local application, restricted concentration and limited
97
Chapter 4 effects on M. pygmaeus, we are confident that it did not have a significant effect on the biological control of aphids in the other treatments.
Aphid densities immediately before spraying differed significantly from the control in two out of three pruning lengths in 2017, whereas all pruned treatments differed from the control in 2018. This discrepancy may be due to the fact that a chemical intervention was originally not planned in 2017 and pirimicarb was used only in the control treatment when aphids there reached a density of 55 aphids leaf-1. Consequently, pirimicarb sprays in control plots were planned and applied much earlier in 2018, when densities reached > 10 aphids leaf- 1, to avoid extremely high densities and migration to other plots. At this point, differences between pruned and non-pruned treatments were still greater. In practical conditions this is more relevant as growers will not wait to intervene until an aphid density of 55 aphids leaf-1 is reached. Different plots were not separated by any barrier because the established M. pygmaeus population had to be able to move through the whole greenhouse. For this reason, aphids were released in the centre of the plots, leaving 10-12 plants on either side to keep a distance from neighbouring plots. Only the 10 centre plants of each plot were sampled to ensure a minimum effect of aphid migration within plant rows. All plots received the same number of aphids and all populations had plenty of sweet pepper leaves available, so there was no immediate pressure to disperse, though it can’t be completely avoided. Only when aphid populations reach very high densities will significant immigration occur as seen in 2017. This was timely tackled by spraying with pirimicarb in 2018.
After release on the lower leaves, aphids moved vertically to reside in all levels of the foliage during the experiment and this in all treatments during both years. Some aphids on the lower leaves will have been physically removed when leaf pruning was performed, but this was never visible in the evolution of aphid population densities. For example, the highest density in the pruned treatments in 2018 was reached three days after a leaf pruning instance. The next leaf pruning occurred only after densities had been decreasing for two consecutive sampling dates.
Macrolophus pygmaeus densities were higher in the flowers than on the leaves in both years. This is not surprising as sweet pepper plants have many more leaves than flowers at any given time, leading to lower concentrations on the leaves. Moreover, flowers provide pollen as a food source, attracting the bugs when prey are scarce. Macrolophus pygmaeus population
98
Aphid control by Macrolophus pygmaeus
densities were higher in 2017 than in 2018. This may be due to the warmer and sunnier weather in 2018, leading to higher mortality rates of the predatory bugs. Martínez-García et al. (2017) found that temperatures above 31 °C cause excessive mortality of M. pygmaeus. Temperatures exceeded 31 °C twice as often in 2018 compared to 2017. A week before chemical intervention in 2017 M. pygmaeus densities in the flowers decreased further, although there was a strong increase on the leaves from 0.1 to 0.5 bugs leaf- 1. This increase was mainly caused by higher numbers of young nymphs, but the densities of N4-N5 nymphs increased as well. Older nymphs tended to reside more in the flowers during the first weeks of the experiment, but moved to the leaves when aphid densities were > 10 aphids leaf-1 for a few weeks. In 2018, the M. pygmaeus densities decreased throughout the experiment until the week after the last pirimicarb spray. After that there was an overall increase, especially on the leaves where high densities of young nymphs were counted. This seems to be a numerical response of M. pygmaeus following higher aphid densities and thus prey availability in the weeks before.
Using leaf pruning in combination with M. pygmaeus as a management strategy against aphids in sweet pepper can reduce the amount of insecticide used against this pest. Aphids can already appear in spring, so it is important to have an established M. pygmaeus population by then. This requires a release strategy that is both effective and efficient. In a previous study, we developed an optimal release strategy for M. pygmaeus in sweet pepper (Brenard et al., 2018, 2019; see Chapters 2 & 3). It consists of introducing the bugs a week after planting and providing supplemental food in a full field fashion to ensure sufficient population growth and dispersal (Brenard et al., 2018; see Chapter 2). The optimal food supplementation scheme involves providing Artemia spp. cysts every two weeks during the development of the first generation (7 weeks) (Brenard et al., 2019; see Chapter 3). After that, M. pygmaeus is able to reside in the crop during the rest of the season without the need for certain prey or multiple releases as required by specialist natural enemies. This management strategy is especially interesting for organic growers, who are constantly in search of insecticide-free ways of controlling aphids. Growers using integrated pest management (IPM) might consider leaf pruning too expensive because it requires extra work. But pruning once every month instead of every two weeks should be enough to receive the benefits of this method. No difference was found between the 160 and 190 cm treatments, so a grower could prune up to 160 cm and then let the plants grow until they reach 190 cm
99
Chapter 4 before pruning again. It is a good alternative for IPM growers as the successful use of pesticides becomes limited due to aphid resistance, fewer active compounds allowed and demands on fewer pesticide residues by retailers and consumers. On top of the above proposed management strategy, growers could release specialist natural enemies curatively when aphids are detected to further improve biological control, as suggested by Messelink et al. (2013). The gall midge A. aphidimyza is the best candidate to control high densities. Another option is the cheaper parasitoid wasp A. colemani which has a strong and fast numerical response in comparison with M. pygmaeus and hardly suffers from intraguild predation by the latter (Martinou et al., 2009; Messelink et al., 2013). Decreasing the foliage length through leaf pruning can increase the search efficiency of these natural enemies as well. When used along with M. pygmaeus, the specialists can be applied efficiently and only when required, instead of performing repeated preventive releases that are often in vain.
Because sweet pepper plants grow taller throughout the season, LAI increases and light penetration towards the lower leaves decreases, which in turn decreases their photosynthesis. Dueck et al. (2006) showed that the lower half of the crop made a 0.5% negative contribution to photosynthesis and a 10% positive contribution to transpiration annually. In other words, lower leaves consume energy instead of producing it, while the plants lose water through them. This extra humidity then has to be regulated by the ventilation system, consuming extra energy. The authors suggest leaf pruning in the lower part of the plants might improve production as a result of decreased maintenance respiration. Through leaf pruning the LAI decreased by 10% (190 cm) to 35% (100 cm) compared to the reference plants. In our study, the effect of leaf pruning on total sweet pepper production differed between years. Plants with shorter foliage lengths (100 and 130 cm) produced significantly less than those with longer foliage lengths (160 and 190 cm) in 2018. This suggests that pruning too many leaves has a negative impact on sweet pepper production as it probably reduces the photosynthetic capacity too much. In 2017, there was no significant effect of leaf pruning on total sweet pepper production although there was a trend towards lower production in the 100 cm treatment. No difference was found between the control and experimental treatments in 2018, but the variability within the control treatment was rather large. Fruit weight was only affected by leaf pruning in 2017, where fruit weight in the 160 cm treatment was lower than in the 130 cm treatment. It is not entirely clear why the average
100
Aphid control by Macrolophus pygmaeus
fruit weight in 2018 was much lower than in 2017 for all treatments. It may be due to differences in climate or light conditions between the years. Harvesting in 2017 stopped earlier than usual due to very high aphid pressure. Fruits suffered excessively from honeydew and sooty mould and plants were removed earlier to avoid aphids spreading to nearby greenhouse compartments. In general, leaf pruning seems to have limited effects on production. Cutting too many leaves (up to 100 or 130 cm foliage left) and thus lowering the LAI to 3.5-4 can be disadvantageous for total production and is not recommended.
The two main obstacles for full biological pest control in Belgian and Dutch sweet pepper crops are aphids and caterpillars of the moth Chrysodeixis chalcites (Esper) (Lepidoptera: Noctuidae) (Messelink et al., 2011b; Messelink, 2014). Chrysodeixis chalcites are highly dispersed in the crop, reside on the underside of leaves and have developed resistance against several insecticides making chemical control very difficult. Effective larval parasitoids are known, but not commercially produced because it is too expensive (Messelink, 2014). Macrolophus pygmaeus are known to feed on eggs of Lepidoptera and may offer a solution (Urbaneja et al., 2009, 2012). The combined effect of M. pygmaeus and leaf pruning on caterpillar control was not tested in these experiments, but a benefit through improved search efficiency may be expected. Further research on the topic is required.
4.6 Conclusion
In this study we developed a new management method consisting of early release of the generalist predatory bug M. pygmaeus in combination with pruning of the lower leaves to control aphids in sweet pepper crops. Pruning up to 160-190 cm is sufficient to increase search efficiency by M. pygmaeus and thus significantly improve aphid control while keeping production levels equal.
101
Chapter 4
4.7 Acknowledgements
This research was financed through two projects, the first (PeMaTo – nr. 140948) by the Agency Flanders Innovation & Entrepreneurship (VLAIO), granted to Research Centre Hoogstraten (RM) in cooperation with Research Station for Vegetable Production and the University of Antwerp, and the second (PeMaTo-EuroPep – nr. 160427) by VLAIO (C-IPM), granted to RM in cooperation with Research Station for Vegetable Production, Wageningen University and Research, University of Antwerp, and the Andalusian Institute of Agricultural and Fisheries Research and Training.
4.8 Supplementary material
Figure S4-1. Visual representation of sweet pepper plants and their foliage lengths under the different leaf pruning treatments.
102
103
104
Chapter 5 | Investigating pest control with predator-prey models in sweet pepper: Macrolophus pygmaeus preying on western flower thrips Frankliniella occidentalis
Nathalie Brenard, Rob Moerkens, Herwig Leirs & Vincent Sluydts
Manuscript
5.1 Abstract
Predator-prey models can be used to aid with pest management decisions because they can predict a future state of the system. Time series data on population densities of the predatory bug Macrolophus pygmaeus Rambur (Hemiptera: Miridae) and the sweet pepper key pest Western flower thrips Frankliniella occidentalis Pergande (Thysanoptera: Thripidae) were collected in sweet pepper greenhouses during three consecutive years. Each year, the predator was able to control thrips infestations. Two modelling approaches were used to model the interaction between M. pygmaeus and thrips: (1) an empirical model consisting of a simple logistic regression that can predict the chance of control but not actual population densities and (2) a mechanistic model of the Lotka-Volterra framework that can predict future population densities of both prey and predator. One week ahead predictions were made and validated as this time span is useful in practice. The logistic regression model was able to predict chance of control well, while the best mechanistic model, a Rosenzweig-MacArthur model with Holling type III functional response, was unable to predict population densities accurately.
105
Chapter 5
5.2 Introduction
Western flower thrips Frankliniella occidentalis Pergande (Thysanoptera: Thripidae) is a common pest species in Dutch and Belgian greenhouse sweet pepper cultivation. Thrips can cause distortion, discolouring, bronzing and silvering of fruits and leaves and even abortion of flowers and young fruits (Bosco et al., 2008; Reitz, 2009). Malformed and discoloured fruits have a reduced market value and feeding injuries on leaves can affect their size and plant vigour in general and eventually reduce yield (Shipp et al., 1998; Reitz, 2009). Damaged or aborted flowers can’t grow into fruits, which also results in direct yield loss (Reitz, 2009). Because of the damage it causes, its high colonization capacities and quick development of insecticide resistance, western flower thrips is considered a key pest of sweet pepper and many other crops, but one that can be controlled through the use of natural enemies (Reitz, 2009). The generalist predatory bug Orius laevigatus (Fieber) (Hemiptera: Anthocoridae) is commonly released in greenhouses to effectively and reliably control thrips (Dissevelt et al., 1995; Weintraub et al., 2011). However, when introducing Macrolophus pygmaeus Rambur (Hemiptera: Miridae) in sweet pepper against other pest species such as aphids, releasing O. laevigatus may become obsolete as M. pygmaeus also feeds on thrips (Messelink et al., 2011b; Messelink & Janssen, 2014). The potential of M. pygmaeus to successfully control thrips infestations in sweet pepper greenhouses can be examined through modelling both the species’ populations and their interactions.
Mechanistic models have since long been used by ecologists to understand, explore and predict the complex processes, interactions and systems of the living world (May, 1981; Odenbaugh, 2005). One such type of ecological models are predator-prey models, which describe the interactions between prey and predator populations. Predator-prey models can be used in applied ecology for pest management and conservation biology (Sinclair et al., 1998; Kozlova et al., 2005; Tonnang et al., 2009; Ghosh et al., 2017). The classical and most basic mechanistic predator-prey model was developed by Lotka and Volterra in 1926. The Lotka-Volterra (LV) model is a second-order dynamical process: it consists of two coupled ordinary differential equations (ODEs), with time as the independent variable (Turchin, 2003; Eq. 1). The first equation describes the change in population density of the prey over time and the second equation describes this for the predator population. Prey density increases through reproduction and decreases through predation by the predator. The
106
Modelling thrips control by Macrolophus pygmaeus
predator population grows by turning consumed prey into offspring and declines by a certain mortality rate of predators. The functional response of the predator, i.e. the rate at which prey are killed by predators as a function of prey density, forms the link between the two equations. The classical LV model is not very realistic: it assumes exponential growth/decline of populations and a linear functional response. Though too simplistic in its basic form, it serves as a framework for building more complex and realistic predator-prey models. One of the first adjustments towards a more realistic model is to drop the assumption of density-independent prey growth. In 1931, Volterra proposed an expansion of the LV model with logistic prey growth (Turchin, 2003). Another important modification is the use of a different functional response than the linear one. Holling was the first to propose a basic classification of functional responses in 1959 and Rosenzweig and MacArthur added the Holling type II (hyperbolic) functional response to the LV model in 1963 (Turchin, 2003; Barraquand, 2014; Eq. 2). The Rosenzweig- MacArthur model is, according to Turchin (2003), the simplest model that can be applied to real-life systems. Other elements, for example predator mutual interference (Beddington, 1975; DeAngelis et al., 1975) or prey refuge (Gause et al., 1936; Huang et al., 2006; Křivan, 2011; Křivan & Priyadarshi, 2015), can also be added to the above models to account for relevant real-life interactions.
Predator-prey models of the LV framework have been used to study population dynamics from a theoretical perspective: the effect of changes in densities of prey or predator populations, the equilibrium states of the system, cyclic behaviour, … (Wollkind et al., 1988; Hanski et al., 1991; Jost & Arditi, 2001; van Rijn et al., 2002; Kar, 2005; Odenbaugh, 2005; Kozlova et al., 2005; Huang et al., 2006; Křivan, 2011). Parameters are directly estimated by fitting the model to time series data of predator and prey population densities or independently derived from experiments and observations on, for example, life tables or feeding behaviour (Kendall et al., 1999). However, it can be difficult to get correct information on the model parameters from observations or experiments. Parameter estimation through model fitting on time series data is a good alternative to this (Kendall et al., 1999). To our knowledge, this type of models have not been used as part of a decision support system in greenhouse crop management. But, in case of practical conditions in agriculture or horticulture, a LV based model approach might not be the best choice. Direct parameter estimation from time series requires lots of data, even more when models are more complex and contain more parameters. Data availability in these cases is often insufficient to parameterize complex models. Moreover,
107
Chapter 5
M. pygmaeus is a generalist, a zoophytophagous generalist even, while well-predicting examples of these models are most often about specialists predators (Bernstein, 1985; Kozlova et al., 2005; Chivers et al., 2014) and the story becomes much more complicated when several food sources are available (Smout et al., 2010; Tornberg et al., 2013). Empirical (or statistical) models may therefore be preferred in these situations. Moerkens et al. (2020) proposed a simple logistic model to predict biological control of several tomato pests, based on earlier pest and predator densities. This model does not calculate future population densities, but predicts the probability of pest control being achieved at the next time step. Pest control is then defined as the period of decline after the peak density of the prey population. No underlying biological or other processes are described in this model.
Using three years of time series data of western flower thrips and M. pygmaeus densities in a semi-commercial sweet pepper greenhouse, we fitted and validated the logistic regression model of Moerkens et al. (2020) and the Lotka-Volterra predator-prey model and its derivatives. To validate the models, we checked if they were able to predict future (one-week ahead) chance of pest control, in case of the logistic regression model, or future population densities, in case of the LV models. These short-term model predictions can serve as a decision- support tool for pest management. Growers receive information on the future state of pest control based on their monitoring efforts and can then decide if they need to take action.
5.3 Material and methods
5.3.1 Data collection
During three consecutive years, only M. pygmaeus was released in several compartments of a semi-commercial sweet pepper greenhouse. During that time, different experiments were performed in the greenhouse, such as testing the ideal food supplementation method for the predatory bug (Brenard et al., 2018, 2019; see Chapters 2 & 3). All compartments suffered from yearly thrips infestations in varying degrees and so during or after the experiments, from February-March until June, M. pygmaeus and thrips population densities were counted. This information is useful to verify if M. pygmaeus can control thrips outbreaks and the time series data can be used to fit predator-prey population models.
108
Modelling thrips control by Macrolophus pygmaeus
Data of prey and predator populations were gathered in seven compartments (300- 1500m²) of a semi-commercial greenhouse at Research Centre Hoogstraten, Belgium, during the seasons of 2016, 2017 and 2018 (Table 5-1).
Sweet pepper Capsicum annuum L. (Solanaceae) varieties differed between compartments and years. All crops were grown under natural light conditions and climate conditions within years were very similar between greenhouse compartments (Table 5-2). Temperature and relative humidity were automatically logged and registered every 5 minutes by a Priva Electronic Measuring Box (Priva, De Lier, The Netherlands).
Pest control in all greenhouse compartments was done following IPM (Integrated Pest Management) principles, except for compartment A in which only biological pest control measures were taken. At the beginning of the growing season, M. pygmaeus (Mirical, Koppert Biological Systems, Berkel en Rodenrijs, the Netherlands) was released in all compartments at a density of 1 bug m-2, except in compartment A where it was released at a density of 1.7 m-2 because a whole Mirical tube of 500 bugs was used in this smaller (300m²) compartment as well. No other predatory bugs or generalists were introduced. Food supplementation schemes for M. pygmaeus differed between years and compartments due to experiments on this subject (Brenard et al., 2018, 2019; see Chapters 2 & 3). Frankliniella occidentalis emerges in the greenhouse compartments every year around February-March.
Several plots of 10 consecutive sweet pepper plants on a row were marked in each greenhouse compartment. Sampling of predator and prey populations was performed weekly by counting all individuals on two randomly chosen flowers per plant in each plot. Table 5-1 shows the start and end dates of data collection and the number of compartments and plots per compartment monitored during each year.
109
Chapter 5
Table 5-1. Properties of sampled greenhouse compartments and sample period per year. Compartment letter codes are the same across years. Year Compartments no. of plots First sampling Last sampling 2016 B 12 05/02/2016 16/06/2016 D 6 2017 B, C (1500m²) 9 each 02/03/2017 06/06/2017 D, E, F, G (500m²) 4 each 2018 A (300m²) 4 07/03/2018 27/06/2018 B, C 9 each D, E 4 each
110
Modelling thrips control by Macrolophus pygmaeus
Table 5-2. Climate conditions in the monitored greenhouse compartments during the three years of sampling. Mean temperature (T) and relative humidity (r.h.) during the sampling period are shown per year and greenhouse compartment. 2016 2017 2018 Compartment Mean (± SE) Mean (± SE) Mean (± SE) Mean (± SE) Mean (± SE) Mean (± SE) T (°C) r.h. (%) T (°C) r.h. (%) T (°C) r.h. (%) A - - - - 22.25 ± 0.02 81.07 ± 0.05 B 21.12 ± 0.02 86.10 ± 0.04 22.4 ± 0.1 76.2 ± 0.5 22.37 ± 0.02 82.37 ± 0.06 C - - 22.6 ± 0.1 70.5 ± 0.4 22.60 ± 0.02 73.47 ± 0.05 D 20.94 ± 0.02 83.97 ± 0.04 22.1 ± 0.2 76.2 ± 0.4 22.52 ± 0.02 79.76 ± 0.05 E - - 22.5 ± 0.2 74.0 ± 0.5 22.64 ± 0.02 77.72 ± 0.06 F - - 21.6 ± 0.2 82.3 ± 0.5 21.82 ± 0.02 84.74 ± 0.05 G - - 22.0 ± 0.2 78.4 ± 0.5 21.78 ± 0.02 83.26 ± 0.06
111
Chapter 5
In 2018 a preventive chemical intervention with pymetrozine was performed on 24 April against aphids. This useless intervention was commissioned by the greenhouse manager and is an example of the often persistent conservative view on pest control. Data after this date were omitted from the dataset as this substance is harmful for M. pygmaeus and thus impacts the population.
5.3.2 Logistic regression model
The method developed by Moerkens et al. (2020) comprises a simplified representation of predator-prey interactions by means of a logistic regression. It is used to predict the probability of the pest being under control based on prey and predator population densities at a certain point in time. For this analysis, a mean prey and predator density per flower was calculated for each sampling plot on each sampling date.
First, a score was given manually to each sampling point, determining if pest control was achieved or not at that time. This was done by means of a binomial variable. The variable was given the value ‘0’ for all data points before the prey population peak: no pest control, the population density will increase after this point; and the value ‘1’ for the peak itself and all the data points after the peak: pest control achieved, the population density will decrease after this point.
Second, a generalized linear model (GLM) was used to study the relationship between prey and predator population densities and the realization of pest control. The data of 2016 were used to estimate model parameters. Achievement of pest control as defined by the 0/1 score was used as response variable, while thrips and M. pygmaeus counts were treated as fixed effects. Insect counts were log transformed prior to the analysis. A fully parameterised model was built at first and non-significant interactions and factor effects were sequentially dropped until a significance level of 0.05 or less was reached as determined with a log- likelihood ratio test (χ²). This analysis was done using the package ‘lme4’ in R v.3.6.1 (R Core Team, 2019).
The model was validated on the data of 2017 and 2018 grouped together as the dataset of 2018 was very small after omitting the data following the pymetrozine treatment. The parameters estimated from the GLM were used to predict chance of control (CC), which is
112
Modelling thrips control by Macrolophus pygmaeus
a value between 0 and 1. This predicted CC was compared with the 0/1 score manually assigned to each time point in these datasets, using the same method as Frans et al. (2018) and Moerkens et al. (2020). For this, the predicted CC values were rounded and grouped in one of 11 categories (from 0 to 1 by steps of 0.1). For each category the proportion of corresponding manual ‘1’ scores, i.e. observed pest control, was determined and regressed against the predicted CC values. It was then verified if the intercept and slope of this regression differed from zero and unity respectively.
The predator and prey density pairs for which the predicted CC is 0.5, i.e. an equal chance (50%) for the prey population to increase or decrease, correspond with the prey isocline. The prey isocline is the line on a graph of prey density (x-axis) against predator density (y-axis) that joins the pairs of predator and prey densities that lead to an unchanging prey population (dN/dt = 0). Predator-prey density pairs on one side of the isocline lead to an increase in prey density, while combinations on the other side lead to a decrease in prey density (Begon et al., 2006). To obtain the prey isocline for this M. pygmaeus-thrips system, the predator density required to achieve CC = 0.5 was calculated for each prey density. These predator-prey density pairs where then plotted on a graph and the curve through them, the prey isocline, was calculated.
5.3.3 Lotka-Volterra based models
A more complex population model framework consists of coupled ODEs derived from the basic LV system (Eq. 1). Here we use a form of the Rosenzweig-MacArthur (RM) model to fit to the data. Four models, including the basic LV model, were fitted and compared to find out which described the data best. The second model (Eq. 2) comprises the basic RM model with a Holling type II functional response, thus describing the feeding behaviour of a specialist predator. The third model (Eq. 3) uses the Holling type III functional response, thereby accounting for a generalist predator’s feeding behaviour. Thrips may be able to find some sort of hiding place within the sweet pepper flowers where it is harder for M. pygmaeus to find or catch them. Or it may just be more difficult for the predatory bugs to detect thrips when population numbers are low. To account for possible reduced predation success at low densities, a factor describing a refuge, i.e. a number or
113
Chapter 5 proportion of prey that is safe from predation, is added to the fourth model (Eq. 4). The refuge is modelled according to the method used by Kar (Kar, 2005; Huang et al., 2006). All parameters used in the models are explained in Table 5-2.
Lotka-Volterra model:
푑푁 = 푟푁 − 푎푁푃 푑푡
푑푃 = 푒푎푁푃 − 푚푃 푑푡 (1)
Rosenzweig-MacArthur model with Holling type II (specialist) functional response:
푑푁 푁 푐푁 = 푟푁 ( 1 − ) − 푃 푑푡 퐾 푑 + 푁
푑푃 푐푁 = 푒 푃 − 푚푃 (2) 푑푡 푑 + 푁 푐푁 The factor describes the functional response of the predator. 푑+푁
Rosenzweig-MacArthur model with Holling type III (generalist) functional response:
푑푁 푁 푐푁² = 푟푁 ( 1 − ) − 푃 푑푡 퐾 푑² + 푁²
푑푃 푐푁² = 푒 푃 − 푚푃 (3) 푑푡 푑² + 푁²
Rosenzweig-MacArthur model with generalist functional response and prey refuge:
푑푁 푁 푐(1 − 푢)²푁2 = 푟푁 ( 1 − ) − 2 2 푃 푑푡 퐾 푑 + (1 − 푢)²푁
푑푃 푐(1 − 푢)²푁² = 푒 푃 − 푚푃 (4) 푑푡 푑² + (1 − 푢)²푁²
114
Modelling thrips control by Macrolophus pygmaeus
Table 5-2. Parameters used in predator-prey models of the Lotka-Volterra type. Parameter Description N Prey population density P Predator population density r Intrinsic rate of natural increase of the prey a Attack rate of the predator e Assimilation efficiency of the predator (i.e. turning prey into new predators) m Predator mortality rate K Carrying capacity of the environment for the prey c Maximum feeding rate of the predator d Prey density at which half of maximum feeding rate occurs u Refuge parameter ϵ [0, 1]
Model fitting was performed using the Bayesian Monte Carlo Markov Chain (MCMC) simulation approach. Euler’s method was used to solve the ODEs with a precision of 0.01 time step. Prey (N) and predator counts (P) were modelled by a Poisson distribution. Data of 2016 were used for parameter estimation. Prior to the analyses, insect counts on all flowers were summed for each plot per sampling date. Time series were shortened to 10 sampling points, removing the first dates as practically no insects were found in the greenhouse at those times. Exponential functions of parameters K and d were used in the model function for MCMC simulation to reduce the difference in magnitude between these two parameters and the others.
Prior parameter information was based on literature (Montserrat et al., 2000; Blaeser et al., 2004; Maselou et al., 2015) or deduced from our own observations in the greenhouse. A log normal distribution was used as prior distribution for all parameters except m and u, which were taken from a uniform distribution to have them bound between 0 and 1. Parameters for which there was little information were given a broader distribution. Initial values of N and P were also estimated and allowed to vary between the 18 sample plots. This way, every sample plot is viewed as a dataset with separate initial values of N and P, but all other model parameters are shared between the plots. More details on the prior distributions are given in Table 5-3.
115
Chapter 5
Bayesian model fitting was performed in R v.3.6.1 (R Core Team, 2019) using the package ‘rjags’ (Plummer et al., 2019). This package provides an R interface for JAGS which is a library for Bayesian data analysis using MCMC.
Table 5-3. Prior parameter distributions for the MCMC simulations. Parameter Distribution Mean Precision (1/variance) Initial N Log normal ln(10) 3 Initial P Log normal ln(2) 3 r Log normal ln(2) 5 a Log normal ln(1) 3 e Log normal ln(1) 3 K Log normal 4.5 3 c Log normal ln(1.5) 1 d Log normal 3 3 Minimum Maximum m Uniform 0.001 0.5 u Uniform 0 1
Validation of the selected model was done on the data of 2017 and 2018. As we are interested in one-week ahead predictions of densities, which is useful from the growers’ perspective, we used the observations to predict one week ahead and compared those predictions with the actual observations at each time point. The parameter estimates from the best model obtained through MCMC simulation were used to calculate predictions of prey and predator densities. Predicted values were plotted against observed values for both prey and predator, after which the intercept and slope of this regression were checked to be different from zero en unity respectively.
Isoclines of the prey and predator for the selected model were calculated using the R package ‘deSolve’ (Soetaert et al., 2010) and visualized using the package ‘phaseR’ (Grayling, 2014).
116
Modelling thrips control by Macrolophus pygmaeus
5.4 Results
Population densities of thrips and M. pygmaeus during the three years of sampling are shown in Figure 5-1. Population densities in the different plots are represented by thin lines in different shades of red (prey) and blue (predator). The dashed lines show the population means across all sampled plots that year, with a band marking the standard error of the mean. Intervention with pymetrozine on 24 April 2018 is also marked on the graph.
Figure 5-1. Population densities of thrips (red) and Macrolophus pygmaeus (blue) per flower during the three years of monitoring. Thin lines in different shades represent sampling plots, thicker dashed lines represent the overall population means and the standard error of the mean is shown by coloured bands. The black dotted line on the graph of 2018 marks the pymetrozine spray.
Thrips populations peaked each year in April after which they sharply decreased and remained low. Macrolophus pygmaeus populations increased most strongly after the thrips population peak in all years and remain present even months after the thrips population has crashed. In 2018 it took longer for the M. pygmaeus population to begin its strong increase, possibly because of the pymetrozine intervention.
117
Chapter 5
5.4.1 Logistic regression model
The interaction between prey and predator density had a significant effect (Table 5-4), so the full model was retained. The parameter estimations from the GLM are shown in Table 5-4.
Table 5-4. Parameter estimates and test statistics of the GLM. Estimate Standard error Z-value P-value Intercept 1.4610 0.3307 4.418 < 0.0001 Prey 0.1704 0.2238 0.761 0.446 Predator 3.1208 0.4303 7.252 < 0.0001 Prey*Predator -1.0682 0.3070 -3.480 0.0005
Figure 5-2 shows the result of the validation of the logistic regression model on the grouped data of 2017 and 2018. The prediction of CC is quite good, with R² = 0.92.
Figure 5-2. Validation of the logistic regression model on the data of 2017 and 2018 grouped together. The x-axis shows the predicted chance of control (CC), while the y-axis shows the observed achievement of control as a proportion of total plots sampled.
118
Modelling thrips control by Macrolophus pygmaeus
The calculated prey isocline, where the predicted CC is 0.5, is shown in Figure 5-3. For pairs of predator and prey population densities on this line there’s an equal chance for the prey population to increase or decrease. For example: when the prey density is 5 thrips per flower, ± 0.2 M. pygmaeus bugs per flower are required to have a CC of 0.5.
Figure 5-3. Prey isocline calculated as the prey/predator combinations that lead to a predicted CC of 0.5.
5.4.2 Lotka-Volterra based models
The statistics of the MCMC fitting procedure are shown in Table 5-5. The multivariate potential scale reduction factor (mpsrf) gives an indication of how well the MCMC chains have converged. A commonly used cut-off value for the mpsrf is 1.05. For values above that, convergence is not considered successful (Denwood, 2016). The deviance information criterion (DIC) gives an indication of the quality of a model relative to other models for a given set of data. A model is preferred based on how well it fits the data, but it gets penalized for complexity, i.e. the number of parameters. DIC is a generalization of the well-known Akaike information criterion (AIC) and often used when posterior distributions of model parameters are obtained through MCMC simulation (Spiegelhalter et al., 2002).
119
Chapter 5
Table 5-5. Statistics of the MCMC fitting procedure using rjags. Model # parameters mpsrf DIC 1. Lotka-Volterra 6 1.01 4428 2. Rosenzweig-MacArthur type II 8 1.16 3963 3. Rosenzweig-MacArthur type III 8 1.04 3973 4. Rosenzweig-MacArthur type III + refuge 9 1.40 3973
Model 3, the Rosenzweig-MacArthur model with a Holling type III functional response, gave the best results. Model 2 has a slightly lower DIC value, but the mpsrf indicates that the MCMC chains failed to converge well. The DIC value of model 4 is equal to model 3, but again convergence was not successful. Parameter estimates of model 3 are shown in Table 5-6.
Table 5-6. Parameter values of the best model: the Rosenzweig-MacArthur model with Holling type III functional response. Parameter Value (mean ± SD) r 0.919 ± 0.024 e 0.250 ± 0.024 K 158.698 ± 1.054 c 3.133 ± 0.221 d 30.846 ± 1.087 m 0.045 ± 0.031
The observed population densities in 2016 are plotted next to the model fit in Figure 5-4 for compartment B and Figure 5-5 for compartment D. Thrips and M. pygmaeus population densities are represented by red and green dots and lines respectively.
120
Modelling thrips control by Macrolophus pygmaeus
Figure 5-4. Population densities of thrips (red) and Macrolophus pygmaeus (green) per plot in compartment B in 2016. Observed data are represented by filled dots and full lines, the simulations of the Rosenzweig-MacArthur model with type III functional response are represented by empty dots and dashed lines. 95% confidence intervals of the simulated densities are indicated by the coloured bands.
121
Chapter 5
Figure 5-5. Population densities of thrips (red) and Macrolophus pygmaeus (green) per plot in compartment D in 2016. Observed data are represented by filled dots and full lines, the simulations of the Rosenzweig-MacArthurmodel with type III functional response are represented by empty dots and dashed lines. 95% confidence intervals of the simulated densities are indicated by the coloured bands.
Figure 5-6 shows the results of the validation of the selected ODE model on the data of 2017 and 2018. Predicted values are plotted against the actual observed values at each sampling date. The predicted values are closer to the observed values for M. pygmaeus (R² = 0.67) than for thrips (R² = 0.61) in 2017. The same is true for validation on the data of 2018, but the predicted values differ much more from the observed values for that year, with an R² value of 0.27 and 0.35 for thrips and M. pygmaeus respectively.
122
Modelling thrips control by Macrolophus pygmaeus
Figure 5-6. Validation of the Rosenzweig-MacArthur model with type III functional response on the data of 2017 (top graphs) and 2018 (bottom graphs). Observed thrips (left) and Macrolophus pygmaeus (right) numbers as total per plot are shown on the y-axis, predicted values on the x-axis. Symbols represent greenhouse compartments, colours represent different plots per compartment. Note M. pygmaeus axes have different scaling between years.
Figure 5-7 shows the validation data as well but here the predicted value at time t+1 is compared to the observed value at time t. Thus it compares a predicted value to the observed value from which it was calculated instead of to the observed value it should predict. This comparison leads to high and even very high R² values: 0.93 and 0.97 for thrips and M. pygmaeus respectively in 2017, and 0.88 and 0.93 in 2018.
123
Chapter 5
Figure 5-7. Comparison of observed values at time t and predicted values at time t+1. Predicted values were calculated using the Rosenzweig-MacArthur model with type III functional response on the data of 2017 (top graphs) and 2018 (bottom graphs). Observed thrips (left) and Macrolophus pygmaeus (right) numbers as total per plot are shown on the y-axis, predicted values at time t+1 on the x-axis. Symbols represent greenhouse compartments, colours represent different plots per compartment. Note M. pygmaeus axes have different scaling between years.
Observed and predicted population densities of both thrips and M. pygmaeus are shown in Figure 5-8. This is only a selection of the data, namely the nine plots of compartment B in 2017. Figures of the population densities in other compartments and those of 2018 can be found in the supplementary material (Figure S5-1 - S5-10). Predicted densities are similar to observed densities one week earlier.
124
Modelling thrips control by Macrolophus pygmaeus
Figure 5-8. Comparison of observed (full lines/filled dots) and predicted (dashed lines/empty dots) population densities of thrips (red) and Macrolophus pygmaeus (green) in compartment B in 2017. Predicted values were calculated using the Rosenzweig-MacArthur model with type III functional response.
The prey and predator isoclines of the selected model are shown in Figure 5-9. The green predator isocline is vertical, with a sharp bend near zero. The red prey isocline is more or less L-shaped, but with a wide turn from the vertical to the horizontal part. When looking at the prey isocline for example: at 20 thrips per plot, which equals 1 thrips per flower, ± 17 M. pygmaeus bugs per plot or ± 0.85 bugs per flower are required to keep the prey population equal. The predator population however is able to increase with that amount of prey available and when it exceeds the prey isocline, prey density will fall. This in turn will lead to prey shortage and a decrease in predator density and eventually the prey population can increase again. Arrows indicate this phase portrait of the system: a graphical description of the dynamics over the entire state space, i.e. the trajectory that is expected to be followed by the combined predator-prey dynamics over time. Both populations will move towards the stable equilibrium point of the system. This stable equilibrium point occurs where the prey isocline intersects with the predator isocline, which is at x = 7.6, y = 37 in this case.
125
Chapter 5
Figure 5-9. Prey (red) and predator (green) isoclines of the Rosenzweig-MacArthur model with type III functional response. The black line and grey arrows represent the phase portrait of the system.
5.5 Discussion
Two different modelling approaches were used to model predator-prey interactions of western flower thrips and M. pygmaeus in sweet pepper greenhouses. The simple logistic regression model does not predict actual population densities, but rather a chance of the prey population to be under control by the predator. The validation of this model gave good results, with an R² value of 0.92, the predicted chance of control was very close to the observed control in 2017 and 2018. Though the model does not predict any population numbers, the prey isocline gives an indication on how many M. pygmaeus bugs are required to control a certain thrips density. Modelling the predator-prey interactions in a more mechanistic way using LV derivatives, results in actual population density predictions. Four models were chosen, starting from the most simple one, the original LV model and then three RM model variants: one with
126
Modelling thrips control by Macrolophus pygmaeus
a specialist’s functional response, one with a generalist’s functional response and the fourth model includes a prey refuge in the generalist’s functional response. The LV model was not expected to give good results, as this is an unrealistic and purely theoretical model. It served as a control situation, to which the more complex models could be compared (Hassell, 1978). The predator M. pygmaeus is a generalist, feeding on several prey species as well as plants, so models with the Holling type III functional response were expected to perform better. Thrips tend to reside and feed in concealed areas of flowers and fruits (Reitz, 2009), making a model with a prey refuge a plausible candidate. The model that emerged from the analyses as the best fit was the RM model with generalist functional response, though the difference with the other RM models in terms of DIC was only small or non-existing. Only the LV model was clearly of inferior quality for our data set, with a much larger DIC value. The RM specialist model and the model with a refuge did not converge well in the MCMC simulation. The specialist RM model has a slightly lower DIC value (2963) compared to the generalist RM (2973) model, which is rather surprising with M. pygmaeus’ generalist feeding behaviour. Though this may be explained by the lack of alternative prey available in the greenhouse to exhibit prey switching behaviour. Validation on the 2017 and 2018 data showed that the generalist RM model was not very good at predicting population densities one week ahead. The regression slope was far from 1, with R² values between 0.27 (thrips 2018) and 0.67 (M. pygmaeus 2017). One of the reasons for this lack of predictive capacity could be that the dataset was too small to fit models with this number of parameters and level of complexity. The whole dataset is not particularly large to begin with and only part of it could be used for fitting as the rest needed to be saved for validation. Larger datasets may be able to improve parameter estimation of the RM models. The high R² values shown in Figure 5-7 reveal that a predicted population density at time t+1 is very similar to the observed density at time t, i.e. the value from which it was calculated. The same can be concluded from comparing the observed with the predicted population densities in Figure 5-8: there seems to be a shift of one week in the prediction. This suggests that the initial population density at time t is the main determiner of densities at time t+1 and that the other parameters in the model have hardly any effect. Why this is the case is not clear. Other authors rarely use their models to make and validate short-term predictions on new data so it is hard to decide on the real practical value of these models (Bernstein, 1985; Kozlova et al., 2005; Tonnang et al., 2009; Křivan & Priyadarshi, 2015). Fitting a model to
127
Chapter 5 experimental data to study system properties and making reliable predictions with that model are two very different things. In our case, this modelling approach does not seem to be the most suitable one to result in useful predictions for thrips control.
The shape of the prey isocline corresponds more or less with what can be expected for low consumption rates at lower prey densities, as is the case with generalists (Begon et al., 2006). However the turn in the isocline is quite sharp, resulting in an L-shape. This L-shaped prey isocline has been described as typical for models that assume a prey refuge (Křivan, 2011; Křivan & Priyadarshi, 2015). Yet in our case, the isoclines are drawn from the results of the RM generalist model without refuge. A remarkable difference with the isoclines found for refuge models, is that the intersection of the predator and prey isocline occurs on the vertical part of the prey isocline and thus at high predator densities in our case: the equilibrium point is found at 37 predators for 7.6 prey. For the refuge models, the intersection (equilibrium point) occurs on the horizontal part of the prey isocline and thus at relatively low predator densities. This is an odd result, with such a high predator to prey ratio, there would be no equilibrium in reality: thrips would be extinct quite quickly. The two prey isoclines we obtained show an L-shaped curve, indicating that their dynamics are similar. Yet they differ strongly in the amount of predatory bugs required to control a certain thrips density. The prey isocline from the logistic model, for example, requires ± 0.2 M. pygmaeus bugs per flower when 5 thrips are present, while the one from the RM generalist model requires ± 0.85 M. pygmaeus for 5 thrips per flower. This is more than triple the amount of bugs required compared to the logistic model. Both prey isoclines are not only obtained from results from different analyses, but are also calculated in a different way. Though the concept of a prey isocline remains the same and results should be similar if both models would predict well, which is not the case for the RM model.
Our data indicate that M. pygmaeus alone was sufficient to control thrips populations in sweet pepper greenhouses during all three seasons and no releases of O. laevigatus or other control measures were required. This is important news for growers, because releasing natural enemies is costly and M. pygmaeus is already more expensive than O. laevigatus: around €55- 65 and €30-40 for 500 individuals respectively (pers. comm. Rani Mertens, RCH). Thrips appeared as always around February/March and after an initial rise in population density, the population reached a peak in April after which it declined sharply. Each year, the peak in thrips
128
Modelling thrips control by Macrolophus pygmaeus
population density was followed by a peak in M. pygmaeus population density, but height and timing of the peak varies between years. In 2018, it took more time before the steeper population increase of M. pygmaeus occurred. It is possible that the pymetrozine spray had something to do with this. This product was found to be light toxic (25-50% mortality) when irrigated and medium toxic (50-75% mortality) when sprayed for M. pygmaeus nymphs and adults by Biobest NV (Biobest, 2012), though another study found it to be harmless in the lab (Tedeschi et al., 2002). At Biobest, side effects are studied on treated plants (pers. comm. Rob Moerkens, R&D Biobest), while Tedeschi et al. kept the bugs on glass with spray deposits (Tedeschi et al., 2002). In any case, no aphid infestations were present at the time of spraying so the preventive application could only harm the greenhouse ecosystem.
After the thrips population crashed, the M. pygmaeus’ did not, but remained at similar densities during the following weeks and months. This clearly shows the difference between the generalist M. pygmaeus and a specialist. The bugs will use other prey and plant food as alternative food sources. There were no control treatments, i.e. greenhouse compartments without natural enemies, available to study the development of thrips populations without the influence of a predator. This is the downside of working under practical field conditions where large yield losses can’t be tolerated. Yet the effect of F. occidentalis on sweet pepper plants is well known: damage is caused by feeding, oviposition in the plant tissue and virus transmission.
There were no problematic levels of thrips damage on the sweet pepper plants and fruits during any of the seasons, suggesting that the population did not reach densities at which its infamous pest problems arise. When no predators are present, F. occidentalis can reach mean densities of 22 individuals per flower in sweet pepper (Bolckmans et al., 2005). Except for one plot out of 34 in 2017 and one in 2018 that reached a mean density of 11 and 15 thrips per flower respectively, the population densities in our experiments peaked at ≤ 10 individuals per flower. This is less than half the population density in the control treatment of Bolckmans et al. (2005). Because of this and the fact that no other control measurements were taken, no other predators against thrips were present and no competing pest species occurred in significant numbers during the sampling periods, we conclude that the decline of the thrips population is mainly caused by the predatory activities of M. pygmaeus. This is in agreement with Messelink et al. (Messelink et al., 2011b; Messelink & Janssen, 2014) who found that M. pygmaeus is able to control thrips in sweet pepper, though O. laevigatus is more efficient at it.
129
Chapter 5
They suggest that when the crop suffers from both thrips and aphid infestations, M. pygmaeus is the better choice.
In order to use models in practice, frequent and sufficient monitoring is required to supply model input. Monitoring is also important for making correct and timely pest management decisions and is therefore one of the basic principles of IPM (EU Parliament and Council Directive 2009/128/EC, 2009). It provides information on the pest species that are present, their densities and severity of the damage they cause. When using natural enemies, monitoring pests alone is not sufficient. It is the ratio of pest to natural enemy that determines if pest control is at hand or if other measures should be taken. The population densities of both actors are required as input for the predator-prey models. But monitoring has a financial cost: investments in monitoring tools and systems and workers that spent their time monitoring. The benefits obtained through the gathered information or the model predictions should outweigh these extra costs if it were to be implemented in practical conditions.
Unlike sampling in our experiments, monitoring of M. pygmaeus in greenhouses by growers is conducted through the use of yellow sticky traps (YSTs). These yellow plates covered with glue are hung in the greenhouse and catch all flying insects that touch them, including adults of both M. pygmaeus and F. occidentalis. Checking the plates every 1 or 2 weeks gives growers an idea on the densities of flying pests and beneficials in the greenhouse so they can adjust pest management measures accordingly. However, manually counting bugs on the traps is very time-consuming. Automating (parts of) the monitoring process will speed it up and lower labour costs, allowing for more intensive monitoring of thrips and M. pygmaeus with YSTs and opening up the possibility to apply population models, such as our thrips – M. pygmaeus model, in greenhouse pest management (Moerkens et al., 2019).
130
Modelling thrips control by Macrolophus pygmaeus
5.6 Conclusion
We found that the predatory bug M. pygmaeus alone is able to keep western flower thrips infestations in sweet pepper greenhouses at acceptable levels. Two modelling approaches were used to describe predator-prey interactions of thrips and M. pygmaeus in the greenhouse. A logistic regression model was able to accurately predict chance of control by the predator, while mechanistic ODE models of the Lotka-Volterra type did not perform well and probably require a larger dataset for reliable parameter estimation. To be able to use the model in practice, thorough monitoring of pest and natural enemy is required.
131
Chapter 5
5.7 Supplementary material
Figure S5-1. Comparison of observed (full lines/filled dots) and predicted (dashed lines/empty dots) population densities of thrips (red) and Macrolophus pygmaeus (green) in compartment C in 2017. Predicted values were calculated using the Rosenzweig-MacArthur model with type III functional response.
132
Modelling thrips control by Macrolophus pygmaeus
Figure S5-2. Comparison of observed (full lines/filled dots) and predicted (dashed lines/empty dots) population densities of thrips (red) and Macrolophus pygmaeus (green) in compartment D in 2017. Predicted values were calculated using the Rosenzweig-MacArthur model with type III functional response.
133
Chapter 5
Figure S5-3. Comparison of observed (full lines/filled dots) and predicted (dashed lines/empty dots) population densities of thrips (red) and Macrolophus pygmaeus (green) in compartment E in 2017. Predicted values were calculated using the Rosenzweig-MacArthur model with type III functional response.
134
Modelling thrips control by Macrolophus pygmaeus
Figure S5-4. Comparison of observed (full lines/filled dots) and predicted (dashed lines/empty dots) population densities of thrips (red) and Macrolophus pygmaeus (green) in compartment F in 2017. Predicted values were calculated using the Rosenzweig-MacArthur model with type III functional response.
135
Chapter 5
Figure S5-5. Comparison of observed (full lines/filled dots) and predicted (dashed lines/empty dots) population densities of thrips (red) and Macrolophus pygmaeus (green) in compartment G in 2017. Predicted values were calculated using the Rosenzweig-MacArthur model with type III functional response.
136
Modelling thrips control by Macrolophus pygmaeus
Figure S5-6. Comparison of observed (full lines/filled dots) and predicted (dashed lines/empty dots) population densities of thrips (red) and Macrolophus pygmaeus (green) in compartment A in 2018. Predicted values were calculated using the Rosenzweig-MacArthur model with type III functional response.
137
Chapter 5
Figure S5-7. Comparison of observed (full lines/filled dots) and predicted (dashed lines/empty dots) population densities of thrips (red) and Macrolophus pygmaeus (green) in compartment B in 2018. Predicted values were calculated using the Rosenzweig-MacArthur model with type III functional response.
138
Modelling thrips control by Macrolophus pygmaeus
Figure S5-8. Comparison of observed (full lines/filled dots) and predicted (dashed lines/empty dots) population densities of thrips (red) and Macrolophus pygmaeus (green) in compartment C in 2018. Predicted values were calculated using the Rosenzweig-MacArthur model with type III functional response.
139
Chapter 5
Figure S5-9. Comparison of observed (full lines/filled dots) and predicted (dashed lines/empty dots) population densities of thrips (red) and Macrolophus pygmaeus (green) in compartment D in 2018. Predicted values were calculated using the Rosenzweig-MacArthur model with type III functional response.
140
Modelling thrips control by Macrolophus pygmaeus
Figure S5-10. Comparison of observed (full lines/filled dots) and predicted (dashed lines/empty dots) population densities of thrips (red) and Macrolophus pygmaeus (green) in compartment E in 2018. Predicted values were calculated using the Rosenzweig-MacArthur model with type III functional response.
141
142
Chapter 6 | General discussion
6.1 Establishment of Macrolophus pygmaeus in sweet pepper greenhouses
Introducing natural enemies in a greenhouse is a costly affair for growers and should happen as efficiently as possible. Because M. pygmaeus is already commonly used in tomato greenhouses, we could rely on an established strategy to start from. Macrolophus pygmaeus is a generalist with a relatively long generation time and therefore it was evident that introductions should take place early in the season, in our case a week after planting, which is also common practice in tomato crops. This way M. pygmaeus is present before pests arrive and ample time is allowed for the predator population to build up. The goal of our experiments was to determine an optimal food supplementation scheme to facilitate these introductions. Supplementing food is common practice when relying on arthropod natural enemies for biological pest control and especially when introducing them preventively in the crop (Wade et al., 2008; Messelink et al., 2015; Oveja et al., 2016; Vangansbeke et al., 2016; Moerkens et al., 2017; Seko et al., 2019).
The results of the first experiment showed that food supplementation had a large positive effect on population growth of M. pygmaeus in sweet pepper. Furthermore, a full field supplementation strategy significantly improved dispersal of the predatory bugs. From the second experiment we learned that food supplementation could be accomplished much more cost-efficient: Artemia spp. cysts gave better results than the more expensive Ephestia kuehniella eggs and providing them biweekly proved to be equally successful as a weekly supplementation.
In tomato crops, growers release M. pygmaeus at a density of 0.5 to 2 individuals m-². We did not test the effects of different M. pygmaeus release densities, but opted for a density of 1 individual m-². However, the smallest greenhouse compartment we used in the experiments in 2018 received a higher density because it has a surface area of 300 m² and we released a standard dose of 500 bugs. This led to a higher release density of 1.7 individuals m²,
143
Chapter 6
but it didn’t result in a difference in population density by the time the second generation appeared. Obviously one replication is insufficient to draw conclusions on this subject.
Choosing several release locations spread out through the greenhouse is likely to further quicken the dispersal of the population in the greenhouse. With four well spread release locations of five plants per 500 m² and full field food supplementation, an adequately dispersed population can be achieved by the time the second generation of M. pygmaeus peaks as shown in Chapter 2. It must be considered that choosing more release locations will require more time and will therefore impose an additional financial cost on growers. Figure S6-1 shows that with four release locations per 500 m², the second generation covers most, if not all, of the greenhouse. There is already some overlap between colonization areas and plant rows not covered by them are few. Furthermore it is likely that the flying M. pygmaeus adults disperse further than the distance from release locations to dispersal plots during the 8 weeks between the first and second generation peaks. As the releases happen preventively, it is not necessary that the first M. pygmaeus generation is already present throughout the whole greenhouse and therefore four release locations per 500 m² are a good recommendation.
All experiments were performed under field conditions in semi-commercial greenhouse compartments, which means they are very close to practical conditions. The greenhouse compartments at Research Centre Hoogstraten (RCH) are built according to the same standards and equipped with systems such as those in commercial greenhouses in Northwestern Europe. Even though the greenhouses are used for experiments and research, the greenhouse manager still aims to optimize yield and fruits are sold when the experiments allow it. The crops are therefore managed in the same manner as in commercial greenhouses. The major difference between RCH’s greenhouse compartments and commercial greenhouses is the size: while RCH’s are 300 – 1500 m², commercial sweet pepper greenhouses in Belgium are usually a few hectares in size. The downside of doing experiments under these field conditions is that there are fewer replicates as the number of greenhouse compartments of that size is obviously limited. Also, because of the higher number of uncontrolled variables, there’s more room for variation in the data compared to experiments under controlled conditions. Yet, the similarity of the experimental conditions to those in practice makes the results and conclusions more relevant and directly applicable for growers.
144
General discussion
Macrolophus pygmaeus is an omnivorous bug and will also feed on sweet pepper plants. At densities present in our experiments, no visual damage to sweet pepper plants or fruits was reported. Moreover, this plant feeding behaviour of M. pygmaeus positively affects biological control of certain sweet pepper pests. Zhang et al. (2018) found that phytophagy by M. pygmaeus before pest infestation induces plant defences in sweet pepper, thereby reducing reproduction of two-spotted spider mite Tetranychus urticae and western flower thrips Frankliniella occidentalis when they colonize the plants afterwards. No effects were found for the green peach aphid Myzus persicae. Hence, the predatory bugs carry out pest control both directly through predation and indirectly through induced plant defences. Bouagga et al. (2018c) found that the volatile organic compounds sweet pepper plants release after phytophagy by M. pygmaeus not only repel the pests F. occidentalis and Bemisia tabaci, but also attract other natural enemies such as the whitefly parasitoid Encarsia Formosa. These positive effects of phytophagy by M. pygmaeus can be exploited to vaccinate plants in the nursery (Bouagga et al., 2018c). Pre-plant releases of M. pygmaeus and Nesidiocoris tenuis in nurseries have two big advantages: (1) to obtain an established predator population in the crop immediately at the time of planting and (2) to activate the plants’ defences through phytophagy so they repel pests and attract natural enemies, this is called vaccinating the plants. Pre-plant releases in nurseries are carried out in Spain, where a warm climate causes high pest pressures at the very beginning of the season and a previously established predator population is of high value (Pérez-hedo et al., 2020). In Northwestern Europe, it’s much colder and sweet pepper plants that grow under natural light are planted in winter. This ensures ample time for the predators to establish a population during the period between planting and emergence of the first pest infestations as shown in Chapter 2 and 3. Pre-plant releases in the nursery have disadvantages as well: predator populations can’t grow too large as they could damage the young plants and spreading to other plants where they are not wanted must be avoided. This may involve chemical interventions. In Spain and other Mediterranean countries, the benefits will outweigh the costs, but in Northwestern Europe this is not necessarily the case. A second positive effect of phytophagy by zoophytophagous predatory bugs is that the induced plant defences can reduce the accumulation of plant viruses. Sweet pepper plants that were punctured by M. pygmaeus or N. tenuis suffered significantly less from Tomato Spotted Wilt Virus (Bouagga et al., 2020).
145
Chapter 6
Aside from positive effects on pest and virus control, M. pygmaeus’ phytophagy also affects performance of the sweet pepper plants themselves (Zhang et al., 2019). Plants exposed to the predatory bugs had less leaves and flowers than unexposed plants, but produced an equal amount of fruits. More importantly: time between the first flower and the first fruit was significantly reduced when M. pygmaeus fed on the plants. The dry weight of leaves plus stems and fruits did not differ between exposed and unexposed plants, but fruits from exposed plants contained more seeds suggesting that M. pygmaeus positively affected pollination.
With the release strategy developed in this PhD research, M. pygmaeus can be successfully introduced in sweet pepper greenhouses and the population will be established well throughout the crop by the time pests such as thrips and aphids arrive. The bugs will perform biological pest control by inducing plants defences before pests arrive and by preying on pests when they infest the crop. All products required for a successful establishment of M. pygmaeus in sweet pepper are already commercially available.
6.2 Biological control of sweet pepper key pests with M. pygmaeus
Earlier studies on caged plants found that M. pygmaeus was able to control populations of sweet pepper key pests such as aphids, thrips and whiteflies on sweet pepper plants (Perdikis & Lykouressis, 2004; Messelink et al., 2011b & 2011c; Messelink & Janssen, 2014; De Backer et al., 2015; Messelink et al., 2015; Pérez-Hedo & Urbaneja, 2015; Bouagga et al., 2018a, 2018b).
From our experiments on aphid management it becomes clear that aphids can indeed be controlled with M. pygmaeus released preventively in sweet pepper greenhouse crops. In order to achieve successful control, however, foliage length should be reduced by pruning the lower leaves of the sweet pepper plants. Aphid populations in control treatments where no leaf pruning occurred grew a lot faster and reached densities that required chemical interventions in order to avoid great damage to the crop. During the experiments, M. pygmaeus showed a tendency to move from the flowers to the leaves when aphid densities rose. There was also a numerical response from M. pygmaeus to rising aphid densities. Pruning leaves is
146
General discussion
currently not practiced in sweet pepper cultivation and growers might need some persuasion in order to adopt this new idea in their management schemes. Because aphid control has always been difficult in sweet pepper, growers might be more inclined to try this new method. It is also a relatively easy measure: it requires no extra cost for tools, chemicals or other pest control materials, though it will increase labour costs. Greenhouse workers can cut off the lower leaves once a month when passing through the plant rows with a cart, which they regularly do for practices such as crop management and harvesting.
During and after the two experiments on food supplementation we sampled populations of M. pygmaeus and western flower thrips F. occidentalis in the greenhouse compartments. Data on population densities of M. pygmaeus and thrips were also collected in 2016. Contrary to the usual management schemes of RCH, no releases of Orius laevigatus were carried out in the respective greenhouse compartments during those three seasons. This anthocorid bug is currently widely used in commercial sweet pepper crops because it successfully controls thrips infestations, but it is unable to sufficiently manage aphids (Messelink et al., 2011b; Messelink & Janssen, 2014). Thrips appeared each year around February/March and reached its highest population density in April. Peak thrips density was followed by a peak in M. pygmaeus population density. Because the thrips infestations never reached problematic levels and no other control measures were applied, we consider predation by M. pygmaeus as the main cause of thrips control. This corresponds with earlier findings in literature of M. pygmaeus’ ability to successfully control infestation of F. occidentalis on sweet pepper (Messelink et al., 2011b; Messelink & Janssen, 2014).
Something we did not study in this PhD research, is the combined use of natural enemies. Messelink & Janssen (2014) already found that both thrips and aphid control was better when M. pygmaeus and O. laevigatus were both present in the greenhouse. A study by Bouagga et al. (2018a) showed that control of thrips, whiteflies and aphids in sweet pepper was successful when combining M. pygmaeus with the generalist predatory mite Amblyseius swirskii. The mites are mainly used against whiteflies, but also control thrips and some mite pests, while M. pygmaeus attack the aphids as well. In this study, lab experiments showed that A. swirskii populations suffered from intraguild predation (IGP) by the different mirid bugs, but in the consequent greenhouse experiment, the combination of A. swirskii and N. tenuis or M.
147
Chapter 6 pygmaeus proved to be the most successful combination to achieve biological control (Bouagga et al., 2018a). This means that the negative effects of IGP were not large enough to significantly affect A. swirskii’s pest control abilities under conditions resembling those in the greenhouse. Aside from combining natural enemies to control an array of pests, combinations may also improve the control of a single pest. Macrolophus pygmaeus is a generalist natural enemy and has a different effect on pest populations than specialist natural enemies. At high pest densities, it is useful to introduce specialists in the greenhouse as they have a fast numerical response and will focus solely on the targeted pest species. Exploring the effects of combining a preventive M. pygmaeus introduction with curative specialist introductions at higher pest densities would be an interesting next step. Classical aphid specialists such as the gall midge Aphidoletes aphidimyza and Aphidius spp. wasps may help to reduce aphid populations faster, yet IGP may nullify the positive effect of an extra natural enemy or even disrupt biological pest control. For example, A. swirskii can’t be used in combination with the specialist aphid predator A. aphidimyza. The mites prey on the eggs of A. aphidimyza, impeding successful aphid control (Messelink et al., 2011a; Messelink et al., 2012). Different combinations of M. pygmaeus and specialist natural enemies should be tested for their interspecies interactions and the effect on biological control of aphids.
An important thing to keep in mind when using natural enemies, such as M. pygmaeus, is to check their compatibility with other management measures. The use of pesticides can harm M. pygmaeus populations through direct mortality, reduced reproductive success or altered behaviour (Martinou et al., 2014; Sukhoruchenko et al., 2015), which results in a reduction of biological control and a loss of investment.
6.3 Monitoring and decision-making in modern pest management
Growers aim to optimize their yield and profit and therefore continuously try to improve their pest management programs. Monitoring is an important step in Integrated Pest Management (IPM) but also very time consuming. All systems that aid growers in speeding up the process are more than welcome and (partly) automating monitoring is therefore a hot topic in pest management research.
148
General discussion
This PhD research was partly funded by a national project called PeMaTo (Pest Management Tool for tomato crops) and a European project called PeMaTo-EuroPep (Pest Management Tool for tomato and pepper in Europe). The aim of both projects was to improve biological pest control through automating parts of the monitoring process and to facilitate decision-making based on the monitoring data by visualizing them and using them in predator- prey models to make predictions on pest control. An important point in these projects is the focus on an ‘ecological monitoring system’ instead of a pest monitoring system: not only pests, but also natural enemies should be monitored in order to obtain relevant information on the state of pest control.
Within the scope of the projects, we constructed a semi-automated detection, counting and identification system for the whiteflies Trialeurodes vaporariorum and B. tabaci on yellow sticky traps (YSTs) using the open source software ImageJ (Moerkens et al., 2019). YSTs are a very common and useful tool to monitor flying pest and beneficial species in both greenhouses and open fields, yet counting insects on the traps is very labour intensive. ImageJ can automatically draw contours around whiteflies on grey-scaled standardized pictures of YSTs and then immediately gives the number of whiteflies on the YST. Using a threshold based on amount of pixels in a contour and knowing that B. tabaci are on average smaller than T. vaporariorum, the number of whiteflies from each species on YSTs containing at least 200 individuals (if less, pictures can be grouped) can be calculated with a formula. With this system, whitefly monitoring was greatly accelerated at RCH. Furthermore, an automated spider mite damage detection system was developed as well as an algorithm that identifies and counts several insect species on YSTs (Nieuwenhuizen et al., 2018). The detection system for spider mite damage consists of a greenhouse cart mounted with a high resolution camera that takes pictures at certain intervals when driving past rows of tomato plants. Next, the pictures are used by a program that automatically detects spider mite damage. In a further step, this cart could be equipped with a kind of GPS system that knows at which location in the greenhouse each picture was taken. The algorithm for counting insects on YSTs is currently able to fully automatically identify and count whiteflies, M. pygmaeus and Nesidiocoris tenuis, though more input is needed to improve N. tenuis identification. Counting thrips proved more difficult because of their slender bodies and non-contrasting colour and will require more work before it can be used in practice. The ability of the algorithm to identify both M. pygmaeus and N. tenuis makes
149
Chapter 6 it particularly interesting as the predatory bugs look very similar and therefore require a well trained eye to count them on YSTs. Moreover, N. tenuis is becoming more and more abundant in Western Europe, but contrary to the Mediterranean region, is considered a pest that damages crops through plant feeding. Its presence should be detected through monitoring. The algorithm is currently unable to distinguish between the two whitefly species, but this would be an important expansion as the whiteflies require different pest management measures. The training and validation of the algorithm was done based on annotated YST pictures taken with a Scoutbox (AgroCares, Wageningen, The Netherlands). This ensured standardized and qualitative pictures with a good resolution, but not all growers are interested in or willing to buy or lease a Scoutbox. The system becomes truly efficient if it works on pictures taken with a smartphone: growers or workers could easily walk through the greenhouse, take a quick picture and instantly upload it for the algorithm to count. First results on smartphone pictures were promising, but more fine tuning is still needed.
During user committee meetings of the projects, growers and companies frequently indicated their interest in these automated monitoring systems. With modern technologies such as robots, drones, high resolution cameras and artificial intelligence, fully automated and trustworthy monitoring systems are within reach and are unmistakably the future of pest management.
Monitoring and gathering data is only the first step, next, data should be visualized and processed in order to support decision-making in pest management. In the PeMaTo and PeMaTo-EuroPep projects we completed this whole pipeline from an ecological monitoring system to a decision-support system: starting from standardized YST pictures, followed by automatic counting and visualization and using the collected data for short-term predictions based on population models (Figure S6-2). The visualization of the monitoring data and model outputs were included in a web front-end demo and can be used by growers or crop advisors to make well-informed decisions on pest management (Figure S6-3). Apps and software for visualizing monitoring data, preferably for mobile devices, are becoming increasingly popular. For example, the Dutch company 30MHz developed a data platform called ZENSIE on which growers can integrate several types of data collected by different means and even share them with other growers. Koppert Biological Systems also developed a platform for data centralization and analyses for which it won the 2019 GreenTech
150
General discussion
Innovation Award. Including population models such as the ones developed in this PhD ensures that we make the most of the collected data.
Two modelling approaches were applied on the data of western flower thrips and M. pygmaeus: (1) a simple logistic regression model that predicts chance of control by the predator, and (2) mechanistic models of the Lotka-Volterra type that predict population densities of both prey and predator. The logistic regression model proved to be able to predict chance of pest control quite well. The downside of this model is that it does not predict population densities. A pest can be under control at high densities, which is of course unacceptable for growers. The mechanistic model with the best fit was a Rosenzweig- MacArthur model with a Holling type III (generalist) functional response. However, predictions made with this model didn’t match the observed values very well. All parameters were estimated from the time series data and our dataset may have been too small to successfully fit these models. Automated monitoring systems will provide researchers with much more data that can be used to build useful population models in the future. Depending on the species, the crop and available monitoring data, different types of models may be more successful or appropriate.
6.4 Overall conclusion
Macrolophus pygmaeus’ popularity in greenhouse tomato pest management can be extended to sweet pepper crops. With the correct food supplementation strategy developed in this PhD, the predatory bugs can establish a population in sweet pepper greenhouses that can control thrips. This can be done with products already on the market. Moreover, when applying leaf pruning –a new tactic in sweet pepper management– M. pygmaeus can also manage aphid infestations, for which successful control measures are highly searched after. Through monitoring and the use of population models, growers can get short term predictions on the state of pest control, on which they can base pest management decisions. Gathering more data will improve the models and their predictions and will increase their importance in future pest management. This will bring us ever closer to a pesticide-free food production from which both humans and our planet will benefit.
151
Chapter 6
6.5 Supplementary material
Figure S6-1 shows a simplified map of a 500m² greenhouse compartment at Research Centre Hoogstraten. The 500m² compartments have 19 rows of 13 rockwool slabs, with 4 plants (3 on the side rows) growing on each slab. As plants grow taller, they are led to grow three main stems, one stem to one side of the row and two to the other. The direction of the two stems alternates between consecutive plants. This stem divergence ensures that every row of rockwool slab consists of two plant rows, that are considered and numbered separately in crop management and research. A total of 36 plant rows are present in the 500m² compartments, the two side rows only have one plant row. The dash-dotted line indicates the location of the gutters in the roof. Plants underneath receive a little less light. Four release locations are indicated in dark-green and the plots that are at the same distance (8,4m diagonally) from the release locations as the dispersal plots in Chapter 2 are shown in orange. The light-green area between them will certainly be colonized by M. pygmaeus by the time the 2nd generation peaks as seen in Chapter 2.
Figure S6-1. Simplified map of 500m² greenhouse compartment. Dark-green rectangles indicate release plants of Macrolophus pygmaeus, while orange rectangles indicate locations that are at a 8.4m diagonal distance. Light-green rectangles indicate plants that are closer to release plants than the orange ones.
152
General discussion
Figure S6-2. Schematic view of the pipeline of the ecological monitoring system and decision-support system for pest management in tomato and sweet pepper greenhouses.
Figure S6-3. Screenshot of the web front-end demo showing data visualization of whitefly and Macrolophus pygmaeus counts on yellow sticky traps in the greenhouse.
153
154
References
Acheampong S, Gillespie DR & Quiring D (2012) Survey of parasitoids and hyperparasitoids (Hymenoptera) of the green peach aphid, Myzus persicae and the foxglove aphid, Aulacorthum solani (Hemiptera: Aphididae) in British Columbia. Journal of the Entomological Society of British Columbia 109: 12– 22.
Albajes R & Alomar Ò (1999) Current and potential use of polyphagous predators. Integrated Pest and Disease Management in Greenhouse Crops. Developments in Plant Pathology, vol. 14. (ed by R Albajes, M Lodovica Gullina, JC van Lenteren & Y Elad) Springer, Dordrecht, pp 265–275.
Alomar O, Riudavets J & Castañe C (2006) Macrolophus caliginosus in the biological control of Bemisia tabaci on greenhouse melons. Biological Control 36: 154–162.
Andrews J (1995) Peppers: the domesticated Capsicums. University of Texas Press, Austin, Texas, USA.
Andriukaitis V (2018) Factsheet - Pesticides authorisation and use in the European Union. European Commission, Food Safety: 2.
Arnó J, Castañé C, Riudavets J & Gabarra R (2010) Risk of damage to tomato crops by the generalist zoophytophagous predator Nesidiocoris tenuis (Reuter) (Hemiptera: Miridae). Bulletin of entomological research 100: 105–15.
Arnó J, Gabarra R, Estopà M, Gorman K, Peterschmitt M, Bonato O, Vosman B, Hommes M & Albajes Garcia R (2009) Implementation of IPM programs on European greenhouse tomato production areas: tools and constraints. Edicions de la Universitat de Lleida, Lleida, Spain.
Arnó J, Gabarra R, Liu T-X, Simmons AM & Gerling D (2010) Natural enemies of Bemisia tabaci: Predators and parasitoids. Bemisia: Bionomics and Management of a Global Pest. (ed by PA Stansly & SE Naranjo) Springer Science, pp 1–540.
Arnó J, Sorribas R, Prat M, Matas M, Pozo C, Rodríguez D, Garreta A, Gómez A & Gabarra R (2009) Tuta absoluta, a new pest in IPM tomatoes in the northeast of Spain. IOBC/WPRS Bulletin 49: 203–208.
De Backer L, Megido RC, Haubruge E & Verheggen FJ (2014) Macrolophus pygmaeus (Rambur) as an efficient predator of the tomato leafminer Tuta absoluta (Meyrick) in Europe. A review. Biotechnologie, Agronomie, Societe et Environnement 18: 536–543.
De Backer L, Wäckers F, Francis F & Verheggen F (2015) Predation of the peach aphid Myzus persicae by the mirid predator Macrolophus pygmaeus on sweet peppers: effect of prey and predator density. Insects 6: 514–523.
Baker RE, Peña JM, Jayamohan J & Jérusalem A (2018) Mechanistic models versus machine learning, a fight worth fighting for the biological community? Biology Letters 14: 1–4.
Bale JS, Van Lenteren JC & Bigler F (2008) Biological control and sustainable food production. Philosophical Transactions of the Royal Society B: Biological Sciences 363: 761–776.
Barraquand F (2014) Functional responses and predator-prey models: A critique of ratio dependence. Theoretical Ecology 7: 3–20.
Barratt BIP, Moran VC, Bigler F & van Lenteren JC (2018) The status of biological control and recommendations for improving uptake for the future. BioControl 63: 155–167.
155
References
Bass C, Puinean AM, Zimmer CT, Denholm I, Field LM, Foster SP, Gutbrod O, Nauen R, Slater R & Williamson MS (2014) The evolution of insecticide resistance in the peach potato aphid, Myzus persicae. Insect Biochemistry and Molecular Biology 51: 41–51.
Bates D, Maechler M, Bolker B & Walker S (2015) Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software 67: 1–48.
Beard J (2006) DDT and human health. Science of the Total Environment 355: 78–89.
Beddington JR (1975) Mutual interference between parasites or predators and its effect on searching efficiency. Journal of Animal Ecology 44: 331–340.
Begon M, Townsend CR & Harper JL (2006) Ecology: from individuals to ecosystems. Blackwell Publishing Ltd, Oxford, UK.
Beissinger SR (2012) Foreword. Bayesian population analysis using WinBUGS - A hierarchical perspective. (ed by M Kéry & M Schaub) Academic Press, Oxford, UK, pp 11–12.
Bernstein C. (1985) A simulation model for an acarine predator-prey system (Phytoseiulus persimilis- Tetranychus urticae). Journal of Animal Ecology 54: 375–389.
Biobest (2012) Side effect manual. Accessible via https://www.biobestgroup.com/en/side-effect-manual [accessed on 27/06/2020]
Biondi A, Chailleux A, Lambion J, Han P, Zappalà L & Desneux N (2013) Indigenous natural enemies attacking Tuta absoluta (Lepidoptera: Gelechiidae) in Southern France. Egyptian Journal of Biological Pest Control 23: 117–121.
Blackman RL & Eastop VF (2000) Aphids on the world’s crops: an identification and information guide. John Wiley, Sons Ltd., Chicherter, UK.
Blaeser P, Sengonca C & Zegula T (2004) The potential use of different predatory bug species in the biological contol of Frankliniella occidentalis (Pergande) (Thysanoptera: Thripidae). Journal of Pest Science 77: 211–219.
Bloemhard C & Ramakers P (2008) Strategies for aphid control in organically grown sweet pepper in the Netherlands. IOBC/WPRS Bulletin 32: 25–28.
Van der Blom J, Robledo A, Torres S, Sánchez JA & Contreras M (2008) Control biológico de plagas en almería: revolución verde después de dos décadas. Phytoma 198: 42–48.
Bolckmans K, Van Houten Y & Hoogerbrugge H (2005) Biological control of whiteflies and western flower thrips in greenhouse sweet peppers with the phytoseiid predatory mite Amblyseius swirskii Athiashenriot (Acari: Phytoseiidae). Second International Symposium on Biological Control of Arthropods 2: 555–565.
Bosco L, Giacometto E & Tavella L (2008) Colonization and predation of thrips (Thysanoptera: Thripidae) by Orius spp. (Heteroptera: Anthocoridae) in sweet pepper greenhouses in Northwest Italy. Biological Control 44: 331–340.
Bouagga S, Urbaneja A, Depalo L, Rubio L & Pérez-Hedo M (2020) Zoophytophagous predator-induced defences restrict accumulation of the tomato spotted wilt virus. Pest Management Science 76: 561–567.
Bouagga S, Urbaneja A & Pérez-Hedo M (2018a) Combined use of predatory mirids with Amblyseius swirskii (Acari: Phytoseiidae) to enhance pest management in sweet pepper. Journal of Economic Entomology 111: 1112–1120.
156
References
Bouagga S, Urbaneja A & Pérez-Hedo M (2018b) Comparative biocontrol potential of three predatory mirids when preying on sweet pepper key pests. Biological Control 121: 168–174.
Bouagga S, Urbaneja A, Rambla JL, Flors V, Granell A, Jaques JA & Pérez-Hedo M (2018c) Zoophytophagous mirids provide pest control by inducing direct defences, antixenosis and attraction to parasitoids in sweet pepper plants. Pest Management Science 74: 1286–1296.
Boulanger F, Jandricic S, Bolckmans K, Wäckers FL & Pekas A (2018) Optimizing aphid biocontrol with the predator Aphidoletes aphidimyza, based on biology and ecology. Pest Management Science 75: 1479–1493.
Brenard N, Sluydts V, De Bruyn L, Leirs H & Moerkens R (2018) Food supplementation to optimize inoculative release of the predatory bug Macrolophus pygmaeus in sweet pepper. Entomologia Experimentalis et Applicata 166: 574–582.
Brenard N, Sluydts V, Christianen E, Bosmans L, De Bruyn L, Moerkens R & Leirs H (2019) Biweekly supplementation with Artemia spp. cysts allows efficient population establishment by Macrolophus pygmaeus in sweet pepper. Entomologia Experimentalis et Applicata 167: 406–414.
Bretz F, Hothorn T & Westfall P (2010) Multiple comparisons using R. CRC Press, Boca Raton, USA (FL).
Calvo FJ, Bolckmans K & Belda JE (2009) Development of a biological control-based Integrated Pest Management method for Bemisia tabaci for protected sweet pepper crops. Entomologia Experimentalis et Applicata 133: 9–18.
Calvo J, Bolckmans K, Stansly PA & Urbaneja A (2009) Predation by Nesidiocoris tenuis on Bemisia tabaci and injury to tomato. BioControl 54: 237–246.
Casida JE & Durkin KA (2013) Anticholinesterase insecticide retrospective. Chemico-Biological Interactions 203: 221–225.
Castañé C, Agustí N, Arnó J, Gabarra R, Riudavets J, Comas J & Alomar Ó (2013) Taxonomic identification of Macrolophus pygmaeus and Macrolophus melanotoma based on morphometry and molecular markers. Bulletin of Entomological Research 103: 204–215.
Castañé C, Alomar O, Goula M & Gabarra R (2004) Colonization of tomato greenhouses by the predatory mirid bugs Macrolophus caliginosus and Dicyphus tamaninii. Biological Control 30: 591–597.
Castañé C, Arnó J, Gabarra R & Alomar O (2011) Plant damage to vegetable crops by zoophytophagous mirid predators. Biological Control 59: 22–29.
Castañé C & Sanchez JA (2006) Amblyseius swirskii (Acari: Phytoseiidae) as a biological control agent of the tobacco whitefly Bemisia tabaci (Hom.: Aleyrodidae) in protected sweet pepper crops in Southern Spain. IOBC wprs Bulletin 29: 77–82.
Chan CK, Forbes AR & Raworth DA (1991) Aphid-transmitted viruses and their vectors of the world. Technical Bulletin Agriculture Canada/Research Branch 3E: 1– 216.
Chivers WJ, Gladstone W, Herbert RD & Fuller MM (2014) Predator-prey systems depend on a prey refuge. Journal of Theoretical Biology 360: 271–278.
De Clercq P, Arijs Y, Van Meir T, Van Stappen G, Sorgeloos P, Dewettinck K, Rey M, Grenier S & Febvay G (2005) Nutritional value of brine shrimp cysts as a factitious food for Orius laevigatus (Heteroptera: Anthocoridae). Biocontrol Science and Technology 15: 467–479.
De Clercq P, Coudron TA & Riddick EW (2014) Production of heteropteran predators. Mass Production of Beneficial Organisms: Invertebrates and Entomopathogens. (ed by JA Morales-Ramos, MG Rojas & DI Shapiro-Ilan) Elsevier, Amsterdam, the Netherlands, pp 57–100.
157
References
Cock MJW, Lenteren JC Van, Brodeur J, Haas F & Mason PG (2010) Do new Access and Benefit Sharing procedures under the Convention on Biological Diversity threaten the future of biological control ? Conference of Parties. : 199–218.
Cocuzza GE, De Clercq P, Van De Veire M, De Cock A, Degheele D & Vacante V (1997) Reproduction of Orius laevigatus and Orius albidipennis on pollen and Ephestia kuehniella eggs. Entomologia Experimentalis et Applicata 82: 101–104.
Cormier D, Chouinard G, Pelletier F, Vanoosthuyse F & Joannin R (2016) An interactive model to predict codling moth development and insecticide application effectiveness. IOBC-WPRS Bulletin 112: 65– 70.
Davies TGE, Field LM, Usherwood PNR & Williamson MS (2007) DDT, pyrethrins, pyrethroids and insect sodium channels. Life 59: 151–162.
DeAngelis D, Goldstein R & O’Neill R (1975) A model for trophic interaction. Ecology 56: 881–892.
Denwood M (2016) runjags: An R package providing interface utilities, model templates, parallel computing methods and additional distributions for MCMC models in JAGS. Journal of Statistical Software 71: 1–25.
Devonshire AL, Field LM, Foster SP, Moores GD, Williamson MS & Blackman RL (1998) The evolution of insecticide resistance in the peach-potato aphid, Myzus persicae. Philosophical Transactions of the Royal Society B: Biological Sciences 353: 1677–1684.
Dhaliwal G, Koul O & Arora R (2004) Integrated pest management: retrospect and prospect. Integrated pest management: potential, constraints and challenges. (ed by O Koul, G Dhaliwal & G Cuperus) CABI Publishing, Wallingford, UK, p 336.
Directive 2009/128/EC of the European Parliament and of the Council on establishing a framework for Community action to achieve the sustainable use of pesticides (2009) Official Journal 309: 71–86.
Dissevelt M, Altena K & Ravensberg W (1995) Comparison of different Orius species for control of Frankliniella occidentalis in glasshouse vegetable crops in the Netherlands. Mededelingen - Faculteit Landbouwkundige en Toegepaste Biologische Wetenschappen, Universiteit Gent 60: 839– 845.
Dueck TA, Grashoff C, Broekhuijsen G & Marcelis LFM (2006) Efficiency of light energy used by leaves situated in different levels of a sweet pepper canopy. Acta Horticulturae 711: 201–205.
Ebert TA & Cartwright B (1997) Biology and ecology of Aphis gossypii Glover (Homoptera: Aphididae). Southwestern Entomologist 22: 116–153.
EC (2019) Integrated Pest Management (IPM). Accessible via https://ec.europa.eu/food/plant/pesticides/sustainable_use_pesticides/ipm_en [accessed on 01/08/2019]
Elbert A & Nauen R (2000) Resistance of Bemisia tabaci (Homoptera: Aleyrodidae ) to insecticides in southern Spain with special reference to neonicotinoids. Pest Management Science 56: 60–64.
Elkinton JS, Trotter RT & Paradis AF (2011) Simulations of population dynamics of Hemlock woolly adelgid and potential impact of biological control agents. Implementation and status of biological control of the Hemlock woolly adelgid. (ed by B Onken & R Reardon) Morgantown, WV, USA, pp 15–24.
Enkegaard A & Brødsgaard HF (2006) Biocontrol in protected crops: is lack of biodiversity a limiting factor? An ecological and societal approach to biological control. Springer, Dordrecht, Netherlands, p 320.
158
References
Enkegaard A, Brødsgaard HF & Hansen DL (2001) Macrolophus caliginosus: Functional response to whiteflies and preference and switching capacity between whiteflies and spider mites. Entomologia Experimentalis et Applicata 101: 81–88.
EPPO (2016) EPPO Standards - List of biological control agents widely used in the EPPO region. Paris, France.
Eshbaugh WH (1976) Genetic and biochemical systematic studies of chili peppers (Capsicum-Solanaceae). Bulletin of the Torrey Botanical Club 102: 396–403.
Fantinou AA, Perdikis DC, Maselou DA & Lambropoulos PD (2008) Prey killing without consumption: does Macrolophus pygmaeus show adaptive foraging behaviour? Biological Control 47: 187–193.
Frans M, Moerkens R, Van Gool S, Sauviller C, Van Laethem S, Luca S, Aerts R & Ceusters J (2018) Modelling greenhouse climate factors to constrain internal fruit rot (Fusarium spp.) in bell pepper. Journal of Plant Diseases and Protection 125: 425–432.
Gabarra R, Alomar Ò, Castañé C, Goula M & Albajes R (2004) Movement of greenhouse whitefly and its predators between in- and outside of Mediterranean greenhouses. Agriculture, Ecosystems & Environment 102: 341–348.
Gause G, Smaragdova N & Witt A (1936) Further studies of interaction between predators and prey. Journal of Animal Ecology 5: 1–18.
Georghiou GP (1972) The evolution of resistance to pesticides. Annual review of ecology and systematics1 3: 133–168.
Gerber R, Smit NJ, Van Vuren JHJ, Nakayama SMM, Yohannes YB, Ikenaka Y, Ishizuka M & Wepener V (2016) Bioaccumulation and human health risk assessment of DDT and other organochlorine pesticides in an apex aquatic predator from a premier conservation area. Science of The Total Environment 550: 522–533.
Gerson U & Weintraub PG (2007) Review: Mites for the control of pests in protected cultivation. Pest Management Science 63: 658–676.
Ghosh J, Sahoo B & Poria S (2017) Prey-predator dynamics with prey refuge providing additional food to predator. Chaos, Solitons and Fractals 96: 110–119.
Gillespie DR & Mcgregor RR (2000) The functions of plant feeding in the omnivorous predator Dicyphus hesperus: water places limits on predation. Ecological Entomology 25: 380–386.
Gillespie DR, Quiring DJM, Foottit RG, Foster SP & Acheampong S (2009) Implications of phenotypic variation of Myzus persicae (Hemiptera: Aphididae) for biological control on greenhouse pepper plants. Journal of Applied Entomology 133: 505–511.
Godfray HCJ, McLean AR, Blacquière T, Field LM, Hails RS, Potts SG, Raine NE & Vanbergen AJ (2014) A restatement of recent advances in the natural science evidence base concerning neonicotinoid insecticides and insect pollinators. Proceedings of the Royal Society B: Biological Sciences 281.
Goula M & Alomar Kurz O (1994) Míridos (Heteroptera: Miridae) de interés en el control integrado de plagas en el tomate. Guía para su identificación. Boletín de sanidad vegetal. Plagas 20: 131–143.
Grayling MJ (2014) PhaseR: An R package for phase plane analysis of autonomous ODE systems. R Journal 6: 43–51.
Gruys P (1982) Hits and misses: the ecological approach to pest control in orchards. Entomologia Experimentalis et Applicata 31: 70–87.
159
References
Guldemond JA, Van Den Brink WJ & Den Beider E (1998) Methods of assessing population increase in aphids and the effect of growth stage of the host plant on population growth rates. Entomologia Experimentalis et Applicata 86: 163–173.
Haas BJ, Kamoun S, Zody MC, Jiang RHY, Handsaker RE, Cano LM, Grabherr M, Kodira CD, Raffaele S, Torto- Alalibo T, Bozkurt TO, Ah-Fong AM V., Alvarado L, Anderson VL, Armstrong MR, Avrova A, Baxter L, Beynon J, Boevink PC, … Nusbaum C (2009) Genome sequence and analysis of the Irish potato famine pathogen Phytophthora infestans. Nature 461: 393–398.
Hajek AE, Hurley BP, Kenis M, Garnas JR, Bush SJ, Wingfield MJ, Lenteren JC Van & Cock MJW (2016) Exotic biological control agents: A solution or contribution to arthropod invasions? Biological Invasions 18: 953–969.
Hamdi F & Bonato O (2013) Relation entre sources trophiques et capacité de survie chez Macrolophus pygmaeus (Hemiptera : Miridae). Canadian Entomologist 146: 285–290.
Hamdi F, Chadoeuf J & Bonato O (2013) Functional relationships between plant feeding and prey feeding for a zoophytophagous bug. Physiological Entomology 38: 241–245.
Hamdi F, Chadoeuf J, Chermiti B & Bonato O (2013) Evidence of cannibalism in Macrolophus pygmaeus, a natural enemy of whiteflies. Journal of insect behavior 26: 614–621.
Hanski I, Hansson L & Henttonen H (1991) Specialist predators, generalist predators, and the microtine rodent cycle. Journal of Animal Ecology 60: 353–367.
Hassell MP (1978) The dynamics of arthropod predator-prey systems. Princeton University Press, Princeton, New Jersey, USA.
Havelka J & Zemek R (1999) Life table parameters and oviposition dynamics of various populations of the predacious gall-midge Aphidoletes aphidimyza. Entomologia Experimentalis et Applicata 91: 481– 484.
Hillert O, Jäckel B & Plate HP (2002) Macrolophus pygmaeus (Rambur 1839) (Heteroptera, Miridae) - ein interessanter Nützling im biologischen Pflanzenschutz. Gesunde Pflanzen 54: 66–73.
Hoddle MS, Van Driesche RG & Sanderson JP (1998) Biology and use of the whitefly parasitoid Encarsia formosa. Annual review of entomology 43: 645–69.
Horowitz AR, Kontsedalov S, Denholm I & Ishaaya I (2002) Dynamics of insecticide resistance in Bemisia tabaci: A case study with the insect growth regulator pyriproxyfen. Pest Management Science 58: 1096–1100.
Huang Y, Chen F & Zhong L (2006) Stability analysis of a prey-predator model with holling type III response function incorporating a prey refuge. Applied Mathematics and Computation 182: 672–683.
Hulshof J & Jurchenko O (2000) Orius laevigatus in a choice situation: thrips or pollen? Mededelingen - Faculteit Landbouwkundige en Toegepaste Biologische Wetenschappen, Universiteit Gent 65: 351– 358.
Ingegno BL, Pansa MG & Tavella L (2009) Tomato colonization by predatory bugs (Heteroptera: Miridae) in agroecosystems of NW Italy. IOBC/WPRS Bulletin 49: 287–291.
Ingegno BL, Pansa MG & Tavella L (2011) Plant preference in the zoophytophagous generalist predator Macrolophus pygmaeus (Heteroptera: Miridae). Biological Control 58: 174–181.
Jandricic SE, Wraight SP, Bennett KC & Sanderson JP (2010) Developmental times and life table statistics of Aulacorthum solani (Hemiptera: Aphididae) at six constant temperatures, with recommendations on the application of temperature-dependent development models.
160
References
Environmental Entomology 39: 1631–1642.
Jandricic SE, Wraight SP, Gillespie DR & Sanderson JP (2013) Oviposition behavior of the biological control agent Aphidoletes aphidimyza (Diptera: Cecidomyiidae) in environments with multiple pest aphid species (Hemiptera: Aphididae). Biological Control 65: 235–245.
Janssen A & Sabelis MW (2015) Alternative food and biological control by generalist predatory mites: the case of Amblyseius swirskii. Experimental and Applied Acarology 65: 413–418.
Jarosik V, Honek A, Lapchin L & Rabasse J-M (1996) An assessment of time varying rate of increase of the green peach aphid, Myzus persicae: its importance in IPM of commercial greenhouse peppers. Ochrana Rostlin 32: 269–276.
Jarosik V, Honek A, Rabasse JM & Lapchin L (1996) Life-history characteristics of the aphid parasitoid Aphelinus abdominalis reared on Macrosiphum euphorbiae. Ochrana Rostlin 32: 83–88.
Jensen S, Johnels AG, Olsson M & Otterlind G (1969) DDT and PCB in Marine Animals from Swedish Waters. Nature 224: 247–250.
Jost C & Arditi R (2001) From pattern to process: identifying predator-prey models from time-series data. Population Ecology 43: 229–243.
Kar TK (2005) Stability analysis of a prey-predator model incorporating a prey refuge. Communications in Nonlinear Science and Numerical Simulation 10: 681–691.
Kendall BE, Murdoch WW, Turchin P, Ellner SP, McCauley E, Nisbet RM & Wood SN (1999) Why do populations cycle? A synthesis of statistical and mechanistic modeling approaches. Ecology 80: 1789–1805.
Kim DH, Lee GH, Park JW & Hwang CY (1991) Occurrence aspects and ecological characteristics of foxglove aphid Aulacorthum solani Kaltenbach (Homoptera: Aphididae) in soybean. Research report RDA (Crop protection) 33: 28–32.
Kirk AA, Lacey LA, Brown JK, Ciomperlik MA, Goolsby JA, Vacek DC, Wendel LE & Napompeth B (2000) Variation in the Bemisia tabaci s. l. species complex (Hemiptera: Aleyrodidae) and its natural enemies leading to successful biological control of Bemisia biotype B in the USA. Bulletin of entomological research 90: 317–327.
Kozlova I, Singh M, Easton A & Ridland P (2005) Twospotted spider mite predator-prey model. Mathematical and Computer Modelling 42: 1287–1298.
Kraft KH, Brown CH, Nabhan GP, Luedeling E, De Jesús Luna Ruiz J, D’Eeckenbrugge GC, Hijmans RJ & Gepts P (2014) Multiple lines of evidence for the origin of domesticated chili pepper, Capsicum annuum, in Mexico. Proceedings of the National Academy of Sciences of the United States of America 111: 6165–6170.
Křivan V (2011) On the Gause predator-prey model with a refuge: A fresh look at the history. Journal of Theoretical Biology 274: 67–73.
Křivan V & Priyadarshi A (2015) L-shaped prey isocline in the Gause predator–prey experiments with a prey refuge. Journal of Theoretical Biology 370: 21–26. van Lenteren JC (2000) A greenhouse without pesticides: fact or fantasy ? Crop Protection 19: 375–384. van Lenteren JC (2012) The state of commercial augmentative biological control: Plenty of natural enemies, but a frustrating lack of uptake. BioControl 57: 1–20.
161
References
Van Lenteren JC (2005) Early entomology and the discovery of insect parasitoids. Biological Control 32: 2– 7. van Lenteren JC, Bale J, Bigler F, Hokkanen HMT & Loomans AJM (2006) Assessing risks of releasing exotic biological control agents of arthropod pests. Annual Review of Entomology 51: 609–634. van Lenteren JC, Bolckmans K, Köhl J, Ravensberg WJ & Urbaneja A (2018) Biological control using invertebrates and microorganisms: plenty of new opportunities. BioControl 63: 39–59. van Lenteren JC, Drost YC, Roermund HJW & Posthuma-Doodeman CJAM (1997) Aphelinid parasitoids as sustainable biological control agents in greenhouses. Journal of Applied Entomology 121: 473–485. van Lenteren JC, van Roermund HJW & Sütterlin S (1995) Biological control of greenhouse whitefly (Trialeurodes vaporariorum) with the parasitoid Encarsia formosa: How does it work? Biological Control 6: 1–10.
Lenth R V. (2016) Least-Squares Means: The R Package lsmeans. Journal of Statistical Software 69: 1–33.
Lenth R, Singmann H, Love J, Buerkner P & Herve M (2019). emmeans: Estimated Marginal Means, aka Least‐Squares Means. R package version 1.4, https://CRAN.R-project.org/package=emmeans
Louda SM, Pemberton RW, Johnson MT & Follett PA (2003) Nontarget effects - the Achilles’ heel of biological control? Retrospective analyses to reduce risk associated with biocontrol introductions. Annual Review of Entomology 48: 365–396.
Lykouressis D, Giatropoulos A, Perdikis D & Favas C (2008) Assessing the suitability of noncultivated plants and associated insect prey as food sources for the omnivorous predator Macrolophus pygmaeus (Hemiptera: Miridae). Biological Control 44: 142–148.
Lykouressis Ρ, Perdikis DC & Chalkia CΑ (2000) The effects of natural enemies on aphid populations on processing tomato in Central Greece. Entomologia Hellenica 13: 35–42.
Lykouressis D, Perdikis D & Michalaki M (2001) Nymphal development and survival of Macrolophus pygmaeus Rambur (Hemiptera : Miridae) on two eggplant varieties as affected by temperature and presence/absence of prey. Biological Control 20: 222–227.
Markkula M, Tiitanen K, Hamalainen M & Forsberg A (1979) The aphid midge Aphidoletes aphidimyza (Diptera, Cecidomyiidae) and its use in biological control of aphids. Annales Entomologici Fennici 45: 89–98.
Martinez-Cascales JI, Cenis JL, Cassis G & Sanchez JA (2006) Species identity of Macrolophus melanotoma (Costa 1853) and Macrolophus pygmaeus (Rambur 1839) (Insecta: Heteroptera: Miridae) based on morphological and molecular data and bionomic implications. Insect Systematics & Evolution 37: 385–404.
Martinez-Cascales JI, Cenis JL & Sanchez JA (2006) Differentiation of Macrolophus pygmaeus (Rambur 1839) and Macrolophus melanotoma (Costa 1853) (Heteroptera: Miridae) based on molecular data. IOBC wprs Bulletin 29: 223–227.
Martínez-García H, Sáenz-Romo MG, Aragón-Sánchez M, Román-Fernández LR, Sáenz-de-Cabezón E, Marco-Mancebón VS & Pérez-Moreno I (2017) Temperature-dependent development of Macrolophus pygmaeus and its applicability to biological control. BioControl 62: 481–493.
Martinou a. F, Milonas PG & Wright DJ (2009) Patch residence decisions made by Aphidius colemani in the presence of a facultative predator. Biological Control 49: 234–238.
Martinou AF, Seraphides N & Stavrinides MC (2014) Lethal and behavioral effects of pesticides on the insect predator Macrolophus pygmaeus. Chemosphere 96: 167–173.
162
References
Maselou D, Perdikis D & Fantinou A (2015) Effect of hunger level on prey consumption and functional response of the predator Macrolophus pygmaeus. Bulletin of Insectology 68: 211–218.
Matsuura A & Nakamura M (2014) Development of neonicotinoid resistance in the cotton aphid Aphis gossypii (Hemiptera: Aphididae) in Japan. Applied Entomology and Zoology 49: 535–540.
May RM (1981) The role of theory in ecology. American Zoologist 21: 903–910.
Mdellel L & Ben Halima Kamel M (2014) Effects of different varieties of pepper (Capsicum annum L.) on the biological parameters of the green peach aphid Myzus persicae Sulzer (Hemiptera, Aphididae) in Tunisia. European Journal of Environmental Sciences 4: 102–105.
Messelink GJ (2014) Persistent and emerging pests in greenhouse crops : Is there a need for new natural enemies? IOBC-WPRS Bulletin 102: 143–150.
Messelink GJ (2016) Food sprays for supporting predatory bugs in greenhouse crops. BioGreenhouse Fact Sheet: 1–2.
Messelink GJ, Bennison J, Alomar O, Ingegno BL, Tavella L, Shipp L, Palevsky E & Wäckers FL (2014) Approaches to conserving natural enemy populations in greenhouse crops: Current methods and future prospects. Biological Control 59: 377–393.
Messelink GJ, Bloemhard CMJ, Cortes J a., Sabelis MW & Janssen A (2011a) Hyperpredation by generalist predatory mites disrupts biological control of aphids by the aphidophagous gall midge Aphidoletes aphidimyza. Biological Control 57: 246–252.
Messelink GJ, Bloemhard CMJ, Hoogerbrugge H, van Schelt J, Ingegno BL & Tavella L (2015) Evaluation of mirid predatory bugs and release strategy for aphid control in sweet pepper. Journal of Applied Entomology 139: 333–341.
Messelink GJ, Bloemhard CM, Kok L & Janssen A (2011b) Generalist predatory bugs control aphids in sweet pepper. Integrated Control in Protected Crops, Temperate Climate 68: 115–118.
Messelink GJ, Bloemhard CMJ, Sabelis MW & Janssen A (2013) Biological control of aphids in the presence of thrips and their enemies. BioControl 58: 45–55.
Messelink GJ, Bloemhard CMJ & Vellekoop R (2011c) Biological control of aphids by the predatory midge Aphidoletes aphidimyza in the presence of intraguild predatory bugs and thrips. Acta Horticulturae 915: 171–178.
Messelink GJ & Janssen A (2014) Increased control of thrips and aphids in greenhouses with two species of generalist predatory bugs involved in intraguild predation. Biological Control 79: 1–7.
Messelink GJ, Sabelis MW & Janssen A (2012) Generalist predators, food web complexities and biological pest control in greenhouse crops. Integrated Pest Management and pest control: current and future tactics. (ed by S Soloneski & M Larramendy) Books on Demand, Norderstedt, Deutschland, p 684.
Milkovich M (2016) RIMpro technology can help manage apple scab, other pests. Fruit Growers News.
Miller GL, Favret C, Carmichael A & Voegtlin DJ (2009) Is there a cryptic species within Aulacorthum solani (Hemiptera: Aphididae)? Journal of Economic Entomology 102: 398–400.
Moerkens R, Berckmoes E, Van Damme V, Ortega-Parra N, Hanssen I, Wuytack M, Wittemans L, Casteels H, Tirry L, De Clercq P & De Vis R (2015) High population densities of Macrolophus pygmaeus on tomato plants can cause economic fruit damage: Interaction with Pepino mosaic virus? Pest Management Science 72: 1350–1358.
163
References
Moerkens R, Berckmoes E, Van Damme V, Wittemans L, Tirry L, Casteels H, De Clercq P & De Vis R (2017) Inoculative release strategies of Macrolophus pygmaeus Rambur (Hemiptera: Miridae) in tomato crops: population dynamics and dispersal. Journal of Plant Diseases and Protection 124: 295–303.
Moerkens R, Brenard N, Bosmans L, Reybroeck E, Janssen D, Hemming J & Sluydts V (2019) Protocol for semi-automatic identification of whiteflies Bemisia tabaci and Trialeurodes vaporariorum on yellow sticky traps. Journal of Applied Entomology 143: 652–658.
Moerkens R, Janssen D, Brenard N, Reybroeck E, Tellez M, Rodríguez E, Bosmans L, Leirs H & Sluydts V (2020) Simplified modelling enhances biocontrol decision making in tomato greenhouses for three important pest species. Journal of Pest Science. Published Online https://doi.org/10.1007/s10340- 020-01256-0
Moerkens R, Pekas A, Bellinkx S, Hanssen I, Huysmans M, Bosmans L & Wäckers F (2020) Nesidiocoris tenuis as a pest in Northwest Europe: Intervention threshold and influence of Pepino mosaic virus. Journal of Applied Entomology 144: 566–577.
Mölck G, Micha S & Wyss U (1999) Attraction to odour of infested plants and learning behaviour in the aphid parasitoid Aphelinus abdominalis. Journal of Plant Diseases and Protection 106: 557–567.
Mollá O, Biondi A, Alonso-Valiente M & Urbaneja A (2014) A comparative life history study of two mirid bugs preying on Tuta absoluta and Ephestia kuehniella eggs on tomato crops: Implications for biological control. BioControl 59: 175–183.
Montserrat M, Albajes R & Castañé C (2000) Functional response of four heteropteran predators preying on greenhouse whitefly (Homoptera: Aleyrodidae) and western flower thrips (Thysanoptera: Thripidae). Environmental Entomology 29: 1075–1082.
Mottet C, Fontaine S, Caddoux L, Brazier C, Mahéo F, Simon J-C, Micoud A & Roy L (2016) Assessment of the dominance level of the R81T target resistance to two neonicotinoid insecticides in Myzus persicae (Hemiptera: Aphididae). Journal of Economic Entomology 109: 2182–2189.
Nabeshima T, Kozaki T, Tomita T & Kono Y (2003) An amino acid substitution on the second acetylcholinesterase in the pirimicarb-resistant strains of the peach potato aphid, Myzus persicae. Biochemical and Biophysical Research Communications 307: 15–22.
Nauen R & Denholm I (2005) Resistance of insect pests to neonicotinoid insecticides: Current status and future prospects. Archives of Insect Biochemistry and Physiology 58: 200–215.
Nauen R & Elbert A (2003) European monitoring of resistance to insecticides in Myzus persicae and Aphis gossypii (Hemiptera: Aphididae) with special reference to imidacloprid. Bulletin of Entomological Research 93: 47–54.
Nestorov I, Rowland M, Hadjitodorov ST & Petrov I (1999) Empirical versus mechanistic modelling: Comparison of an artificial neural network to a mechanistically based model for quantitative structure pharmacokinetic relationships of a homologous series of barbiturates. AAPS PharmSci 1: 1–9.
Nieuwenhuizen A, Hemming J & Suh H (2018) Detection and classification of insects on stick-traps in a tomato crop using Faster R-CNN. Proceedings of the Netherlands Conference on Computer Vision. pp 1–4.
Nikolakakis NN, Margaritopoulos JT & Tsitsipis JA (2003) Performance of Myzus persicae (Hemiptera: Aphididae) clones on different host-plants and their host preference. Bulletin of Entomological Research 93: 235–242.
164
References
Nomikou M, Janssen A, Schraag R & Sabelis MW (2002) Phytoseiid predators suppress populations of Bemisia tabaci on cucumber plants with alternative food. Experimental and Applied Acarology 27: 57–68.
Oberemok VV, Laikova KV, Gninenko YI, Zaitsev AS, Nyadar PM & Adeyemi TA (2015) A short history of insecticides. Journal of plant protection research 55: 221–226.
Odenbaugh J (2005) Idealized, inaccurate but successful: A pragmatic approach to evaluating models in theoretical ecology. Biology and Philosophy 20: 231–255.
Oerke E-C (2006) Crop losses to pests. Journal of Agricultural Science 144: 31–43.
Oliveira MRV, Henneberry TJ & Anderson P (2001) History, current status, and collaborative research projects for Bemisia tabaci. Crop Protection 20: 709–723.
Olsson M & Reutergårdh L (1986) DDT and PCB pollution trends in the swedish aquatic environment. Ambio 15: 103–109.
Oveja MF, Riudavets J, Arnó J & Gabarra R (2016) Does a supplemental food improve the effectiveness of predatory bugs on cucumber? BioControl 61: 47–56.
Palumbo J, Horowitz A & Prabhaker N (2001) Insecticidal control and resistance management for Bemisia tabaci. Crop Protection 20: 739–765.
Perdikis D, Kapaxidi E & Papadoulis G (2008) Biological control of insect and mite pests in greenhouse solanaceous crops. The European Journal of Plant Science and Biotechnology 2: 125–144.
Perdikis D & Lykouressis D (2000) Effects of various items, host plants, and temperatures on the development and survival of Macrolophus pygmaeus Rambur (Hemiptera: Miridae). Biological Control 17: 55–60.
Perdikis DC & Lykouressis DP (2002a) Life table and biological characteristics of Macrolophus pygmaeus when feeding on Myzus persicae and Trialeurodes vaporariorum. Entomologia Experimentalis et Applicata 102: 261–272.
Perdikis C & Lykouressis DΡ (2002b) Description of the egg and nymphal instars of Macrolophus pygmaeus Rambur (Hemiptera: Miridae). Entomologia Hellenica 14: 32–40.
Perdikis DC & Lykouressis DP (2004) Myzus persicae (Homoptera: Aphididae) as suitable prey for Macrolophus pygmaeus (Hemiptera: Miridae) population increase on pepper plants. Environmental Entomology 33: 499–505.
Pérez-hedo M, Riahi C & Urbaneja A (2020) Use of zoophytophagous mirid bugs in horticultural crops: current challenges and future perspectives. Pest Management Science [Online].
Pérez-Hedo M & Urbaneja A (2015) Prospects for predatory mirid bugs as biocontrol agents of aphids in sweet peppers. Journal of Pest Science 88: 65–73.
Pérez-Hedo M & Urbaneja A (2016) The zoophytophagous predator Nesidiocoris tenuis: A successful but controversial biocontrol agent in tomato crops. Advances in Insect Control and Resistance Management. (ed by A Horowitz & I Ishaaya) Springer, Cham, Switzerland, p 347.
Peshin R, Bandral RS, Zhang W, Wilson L & Dhawan AK (2009) Integrated Pest Management: a global overview of history, programs and adoption. Integrated Pest Management: innovation- development process. (ed by R Peshin & AK Dhawan) Springer Netherlands, Dordrecht, p 689.
Pickersgill B (1997) Genetic resources and breeding of Capsicum spp. Euphytica 96: 129–133.
165
References
Pilkington LJ, Messelink G, van Lenteren JC & Le Mottee K (2010) “Protected Biological Control” - Biological pest management in the greenhouse industry. Biological Control 52: 216–220.
Pinheiro J, Bates D, DebRoy S, Sarkar D & R Core Team (2018). nlme: Linear and Nonlinear Mixed Effects Models. https://CRAN.R-project.org/package=nlme
Plummer M, Stukalov A & Denwood M (2019) rjags: Bayesian graphical models using MCMC.
Portillo N, Alomar O & Wäckers F (2012) Nectarivory by the plant-tissue feeding predator Macrolophus pygmaeus Rambur (Heteroptera: Miridae): Nutritional redundancy or nutritional benefit? Journal of Insect Physiology 58: 397–401.
Prado S, Jandricic S & Frank S (2015) Ecological interactions affecting the efficacy of Aphidius colemani in greenhouse crops. Insects 6: 538–575.
Pretty JN (1991) Farmers’ extension practice and technology adaptation: agricultural revolution in 17-19th century Britain. Agriculture and human values 8: 132–148.
Put K, Bollens T, Wäckers FL & Pekas A (2012) Type and spatial distribution of food supplements impact population development and dispersal of the omnivore predator Macrolophus pygmaeus (Rambur) (Hemiptera: Miridae). Biological Control 63: 172–180.
Rahmani S, Azimi S & Moghadasi M (2016) LC30 effects of thiamethoxam and pirimicarb, on population parameters and biological characteristics of Macrolophus pygmaeus (Hemiptera: Miridae). Arthropods 5: 44–55.
Ramakers PMJ (2004a) IPM program for sweet pepper. Biocontrol in Protected Culture. (ed by KM Heinz, V Driesche & MP Parrella) Ball Publishing, Batavia, IL, USA, pp 439–455.
Ramakers PMJ (2004b) IPM Program for sweet pepper. Biocontrol in protected culture. (ed by KM Heinz & MP Van Driesche, R. G. Parrella) Ball Publishing, Batavia, pp 439–455.
R Core Team (2019) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.
Reitz SR (2009) Biology and ecology of the western flower thrips (Thysanoptera: Thripidae): the making of a pest. Florida Entomologist 92: 7–13. van Rijn PCJ, van Houten YM & Sabelis MW (2002) How plants benefit from providing food to predators even when it is also edible to herbivores. Ecology 83: 2664–2679.
Riudavets J & Castañé C (1998) Identification and evaluation of native predators of Frankliniella occidentalis (Thysanoptera: Thripidae) in the Mediterranean. Environmental Entomology 27: 86– 93. de Roos AM (2014) Modeling Population Dynamics. University of Amsterdam, Amsterdam, the Netherlands.
Roosens E & De Clercq P (2014) Predatiecapaciteit van de roofwants Macrolophus pygmaeus op de invasieve tomatenmineermot Tuta absoluta.
Roy HE, Brown PMJ, Adriaens T, Berkvens N, Borges I, Richard SC, Patrick FC, Rene DC, Estoup A, Evans EW, Facon B, Gardiner MM, Gil A, Grez AA, Guillemaud T, Haelewaters D, Herz A, Honek A, Howe AG, … Vigla S (2016) The harlequin ladybird, Harmonia axyridis: global perspectives on invasion history and ecology. Biological Invasions 18: 997–1044.
Royama T (1992) Analytical Population Dynamics. Chapman & Hall, London, UK.
166
References
Samietz J, Graf B, Hoehn H, Hoepli HU & Schaub L (2008) SOPRA: Phenology modelling of major orchard pests - from biological basics to decision support. VIII International symposium on modelling in fruit research and orchard management. INT SOC HORTICULTURAL SCIENCE, PO BOX 500, 3001 LEUVEN 1, BELGIUM, pp 35–42.
Sanchez JA, Cánovas F & Lacasa A (2007) Thresholds and management strategies for Aulacorthum solani (Hemiptera: Aphididae) in greenhouse pepper. Journal of Economic Entomology 100: 123–30.
Sanchez JA, La-Spina M, Michelena JM, Lacasa A & Hermoso de Mendoza A (2011) Ecology of the aphid pests of protected pepper crops and their parasitoids. Biocontrol Science and Technology 21: 171– 188.
Sanchez JA, Martinez-Cascales JI & Lacasa A (2003) Abundance and wild host plants of predator mirids (Heteroptera: Miridae) in horticultural crops in the Southeast of Spain. IOBC-WPRS Bulletin 26: 147–152.
Sanchez JA, La Spina M & Perera OP (2012) Analysis of the population structure of Macrolophus pygmaeus (Rambur) (Hemiptera: Miridae) in the Palaearctic region using microsatellite markers. Ecology and Evolution 2: 3145–3159.
Satar S, Kersting U & Uygun N (2008) Effect of temperature on population parameters of Aphis gossypii Glover and Myzus persicae (Sulzer) (Homoptera: Aphididae) on pepper. Journal of Plant Diseases and Protection 115: 69–74. van Schelt J (1999) Biological control of sweet pepper pests in the Netherlands. IOBC Bulletin 22: 217– 220. van Schelt J, Hoogerbrugge H, Becker N, Messelink GJ & Blockmans K (2011) Comparing Aphidius colemani and Aphidius matricariae on Myzus persicae ssp. nicotianae in sweet pepper. IOBC/WPRS Bulletin 68: 169–172.
Seko T, Abe J & Miura K (2019) Effect of supplementary food containing Artemia salina on the development and survival of flightless Harmonia axyridis in greenhouses. BioControl 64: 333–341.
Serrano AR & Guerra-Sanz JM (2006) Quality fruit improvement in sweet pepper culture by bumblebee pollination. Scientia Horticulturae 110: 160–166.
Sharov A (1996) Modelling forest insect dynamics. Caring for the forest: research in a changing world. pp 293–303.
Sharov AA & Colbert JJ (1994) Gypsy Moth Life System Model. Integration of knowledge and a user’s guide. Virginia Polytechnic Institute and State University, Blacksburg, VA, USA.
Shipp JL, Binns MR, Hao X & Wang K (1998) Economic injury levels for western flower thrips (Thysanoptera: Thripidae) on greenhouse sweet pepper. Journal of Economic Entomology 91: 671–677.
Shipp JL, Whitfield GH & Papadopoulos AP (1994) Effectiveness of the bumble bee, Bombus impatiens Cr. (Hymenoptera: Apidae), as a pollinator of greenhouse sweet pepper. Scientia Horticulturae 57: 29– 39.
Sinclair ARE, Pech RP, Dickman CR, Hik D, Mahon P & Newsome AE (1998) Predicting effects of predation on conservation of endangered prey. Conservation Biology 12: 564–575.
Singh R & Singh G (2016) Aphids and Their Biocontrol. Ecofriendly Pest Management for Food Security. Academic Press, Cambridge, Massachusetts, USA, p 762.
Smout S, Asseburg C, Matthiopoulos J, Fernández C, Redpath S, Thirgood S & Harwood J (2010) The functional response of a generalist predator. PLoS ONE 5.
167
References
Soetaert K, Petzoldt T & Setzer RW (2010) Solving Differential Equations in R: Package deSolve. Journal of Statistical Software 33: 1–25.
Spiegelhalter DJ, Best NG & Carlin BP (2002) Bayesian measures of model complexity and fit. Journal of the Royal Statistical Society: Series B (Statistical Methodology) 64: 583–639.
Stansly PA, Calvo FJ & Urbaneja A (2005) Augmentative biological control of Bemisia tabaci biotype “Q” in Spanish greenhouse pepper production using Eretmocerus spp. Crop Protection 24: 829–835.
Sukhoruchenko GI, Belyakova NA, Pazyuk IM & Ivanova GP (2015) The toxic effect of greenhouse insecticides on the predatory bugs Nesidiocoris tenuis Reuter and Macrolophus pygmaeus H.-S. (Heteroptera , Miridae). Entomological Review 95: 1166–1173.
Sylla S, Brévault T, Diarra K, Bearez P & Desneux N (2016) Life-history traits of Macrolophus pygmaeus with different prey foods. PLoS ONE 11: 1–8.
Symondson WOC, Sunderland KD & Greenstone MH (2002) Can generalist predators be effective biocontrol agents? Annual review of entomology 47: 561–594.
Tedeschi R, Tirry L, Van de Veire M & de Clercq P (2002) Toxicity of different pesticides to the predatory bug Macrolophus caliginosus (Heteroptera: Miridae) under laboratory conditions. IOBC/WPRS Bulletin 25: 71–80.
Tonnang HEZ, Hervé BDB, Biber-Freudenberger L, Salifu D, Subramanian S, Ngowi VB, Guimapi RYA, Anani B, Kakmeni FMM, Affognon H, Ndjomatchoua FT, Pedro SA, Nana P, Johansson T, Nedorezov L V., Tanga CM, Nana P, Fiaboe KM, Mohamed SF, … Borgemeister C (2017) Advances in crop insect modelling methods — Towards a whole system approach. Ecological Modelling 354: 88–103.
Tonnang HEZ, Nedorezov L V., Ochanda H, Owino J & Löhr B (2009) Assessing the impact of biological control of Plutella xylostella through the application of Lotka-Volterra model. Ecological Modelling 220: 60–70.
Tornberg R, Lindén A, Byholm P, Ranta E, Valkama J, Helle P & Lindén H (2013) Coupling in goshawk and grouse population dynamics in Finland. Oecologia 171: 863–872.
Trapman M, Helsen H & Polfliet M (2008) Development of a dynamic population model as a decision support system for Codling Moth (Cydia pomonella L.) management. Proceedings of the 13th International Conference on Cultivation Technique and Phytopathological Problems in Organic Fruit-Growing. pp 247–251.
Turchin P (2003) Complex population dynamics: a theoretical/empirical synthesis. Princeton University Press, Princeton, New Jersey (USA).
Urbaneja A, González-Cabrera J, Arnó J & Gabarra R (2012) Prospects for the biological control of Tuta absoluta in tomatoes of the Mediterranean basin. Pest Management Science 68: 1215–1222.
Urbaneja A, Montón H & Mollá O (2009) Suitability of the tomato borer Tuta absoluta as prey for Macrolophus pygmaeus and Nesidiocoris tenuis. Journal of Applied Entomology 133: 292–296.
Urbaneja A, Sánchez E & Stansly PA (2007) Life history of Eretmocerus mundus, a parasitoid of Bemisia tabaci, on tomato and sweet pepper. BioControl 52: 25–39.
Urbaneja A, Tapia G & Stansly P (2005) Influence of host plant and prey availability on developmental time and surviorship of Nesidiocoris tenius (Het.: Miridae). Biocontrol Science and Technology 15: 513– 518.
Vandekerkhove B & De Clercq P (2010) Pollen as an alternative or supplementary food for the mirid predator Macrolophus pygmaeus. Biological Control 53: 238–242.
168
References
Vandekerkhove B, Parmentier L, Van Stappen G, Grenier S, Febvay G, Rey M & De Clercq P (2009) Artemia cysts as an alternative food for the predatory bug Macrolophus pygmaeus. Journal of Applied Entomology 133: 133–142.
Vangansbeke D, Nguyen DT, Audenaert J, Gobin B, Tirry L & Clercq P De (2016) Establishment of Amblyseius swirskii in greenhouse crops using food supplements. Systematic and Applied Acarology 21: 1174–1184.
Vanhaute E, Paping R & Ó Gráda C (2007) The European subsistence crisis of 1845‐1850: a comparative perspective. When the potato failed. Causes and effects of the ‘last’ European subsistance crisis (CORN Publication Series 9). (ed by C Ó Gráda, R Paping & E Vanhaute) Brepols Publishers, Turnhout, Belgium, pp 15–40.
Vehrs SLC, Walker GP & Parrella MP (1992) Comparison of population growth rate and within-plant distribution between Aphis gossypii and Myzus persicae (Homoptera: Aphididae) reared on potted Chrysanthemums. Journal of Economic Entomology 85: 799–807.
VLAM (2019a) Groenteareaal (2010-2019).
VLAM (2019b) Groenteproductie (2010-2019).
Waage JK (2000) Indirect ecological effects in biological control: the challenge and the opportunity. Evaluating indirect ecological effects of biological control. (ed by E Wajnberg, J Scott & P Quimby) CABI Publishing, p 288.
Wade MR, Zalucki MP, Wratten SD & Robinson KA (2008) Conservation biological control of arthropods using artificial food sprays: Current status and future challenges. Biological Control 45: 185–199.
Weintraub PG, Pivonia S & Steinberg S (2011) How many Orius laevigatus are needed for effective western flower thrips, Frankliniella occidentalis, management in sweet pepper? Crop Protection 30: 1443– 1448.
Wilder JW, Voorhis N, Colbert JJ & Sharov A (1994) A three variable differential equation model for gypsy moth population dynamics. Ecological Modelling 72: 229–250.
Winston ML (1999) Nature Wars: People vs. Pests. Harvard University Press, Cambridge, Massachusets, USA.
Wollkind DJ, Collings JB & Logan JA (1988) Metastability in a temperature-dependent model system for predator-prey mite outbreak interactions on fruit trees. Bulletin of Mathematical Biology 50: 379– 409.
Yoshida K, Schuenemann VJ, Cano LM, Pais M, Mishra B, Sharma R, Lanz C, Martin FN, Kamoun S, Krause J, Thines M, Weigel D & Burbano HA (2013) The rise and fall of the Phytophthora infestans lineage that triggered the Irish potato famine. eLife 2013: 1–25.
Zhang H, Chen L & Nieto JJ (2008) A delayed epidemic model with stage-structure and pulses for pest management strategy. Nonlinear Analysis: Real World Applications 9: 1714–1726.
Zhang NX, Messelink GJ, Alba JM, Schuurink RC, Kant MR & Janssen A (2018) Phytophagy of omnivorous predator Macrolophus pygmaeus affects performance of herbivores through induced plant defences. Oecologia 186: 101–113.
Zhang NX, Messelink GJ, Verdonkschot S & Janssen A (2019) Plant feeding by an omnivorous predator affects plant phenology and omnivore performance. Biological Control 135: 66–72.
169
170
Dankwoord
Juni 2020 en ik schrijf de, voorlopige, laatste zinnen van mijn doctoraatsthesis. Voorlopig, want het proces is nog niet rond en er zal nog veel verbeterwerk aankomen, maar toch voelt het al een beetje “af”. Ook voor mezelf is dit toch een beetje een verrassing, want de weg tot hier was zeer hobbelig en werd verschillende keren onderbroken, maar telkens verscheen er wel een te volgen omleiding…
Juni 2014 en ik had net de laatste zinnen van mijn masterthesis geschreven. Examens volop bezig en wat daarna: een doctoraat starten of toch maar een “echte” job zoeken? Ik had de knoop nog niet doorgehakt, maar toeval hielp een handje. Op weg naar huis van m’n laatste examen riep Herwig mij vanop zijn fiets aan de overkant van de Groenenborgerlaan “Ah, Nathalie!”. En zie daar, een voorstel om een doctoraat te starten bij EVECO, ingefluisterd door Luc, promotor van mijn masterthesis en mijn eeuwige raadsman voor hulp bij geleedpotigen, statistiek, algemene onderzoeksfilosofie, levensfilosofie en koetjes en kalfjes.
Halsoverkop een beursaanvraag schrijven in de zomervakantie, want er was maar geld voor een klein jaar. De wereld van landbouw, perenbladvlo en fluweelmijten ging voor mij open. Het IWT kende me nipt geen beurs toe, maar door de hoge ranking gaf de UAntwerpen één jaar extra geld. En zo start het verhaal met twee jaar onderzoek in biologische plaagbestrijding in de perenteelt. Ongetwijfeld een grote plottwist voor zij die net deze thesis gelezen hebben en er de afgelopen zes jaar niet bij waren.
Hoewel er weinig output is voortgekomen uit die twee eerste jaren, heb ik er wel bergen ervaring en kennis aan overgehouden. Ik ben dankbaar voor alle mensen die mijn pad kruisten: Jonas Reijniers, om mij met veel geduld de eerste stappen van het modelleren aan te leren en mij met raad en daad bij te staan bij de eerste projectmeetings; Tim Beliën en de andere collega’s van Proefcentrum Fruitteelt vzw, waar ik tevens mijn laatste drie maanden als onderzoeker in de perenteelt doorbracht (en waardoor ik voor twee jaar Limburger werd); Johan Witters van het ILVO, die mij als mijtenspecialist op weg zette en me aanraadde om naar Polen te gaan, waar prof. Joanna Mąkol, Magda, Hannah en de andere Poolse onderzoekers van de Wroclaw University of Environmental and Life Sciences mij een week met open armen ontvingen en me alles leerden over fluweelmijten en aanverwanten. Vervolgens kwam Magda
171
een maand mee de mijten in de Belgische perenboomgaarden bestuderen en een jaar laten konden we samen onze resultaten presenteren in Beijing.
Juni 2015 en ik wist dat het perenbladvlo-project niet lang genoeg liep om een heel doctoraat te maken. Gelukkig was daar Rob en dit keer ging de tuinbouwwereld voor mij open. Perenboomgaarden maakten plaats voor paprikaserres en tripjes naar Sint-Truiden gingen nu richting Hoogstraten, de tweede IWT-beursaanvraag werd geschreven. Meer tijd, meer voorbereiding en goede begeleiding mochten helaas niet baten, dus werkte ik het perenbladvlo-project verder af en zwaaide eind 2016 de collega’s van pcfruit vzw uit.
Na drie maanden pauze kon ik terug aan de slag bij EVECO en dan toch in de paprikateelt, met Vincent en Rob als extra promotoren. Vincent, bedankt dat je me toen hebt aangenomen, want we kenden elkaar eigenlijk niet. Je zorgde er steeds voor dat er middelen waren voor een contractverlenging waardoor ik een volwaardig doctoraatsonderzoek kon uitvoeren. Jouw hulp en inzicht bij de populatiemodellen was ook cruciaal, anders zou hoofdstuk 5 van deze thesis weinig om het lijf hebben. Rob, bedankt dat je me niet was vergeten na die beursaanvraag en bedankt om mij al die tijd bij te staan met goeie raad, om altijd zo vlot bereikbaar te zijn, om je kennis en voortgang te delen, om mee te denken over allerhande problemen en naar mijn verzuchtingen te luisteren als ik vast zat. Herwig, bedankt om het zaadje te planten dat is uitgegroeid tot een doctoraat. Bedankt voor jouw vertrouwen doorheen de jaren en de vrijheid die je gaf, waardoor ik alle kansen kon grijpen.
Bedankt aan de collega’s van Proefcentrum Hoogstraten, met in het bijzonder Lien en Rani: altijd paraat om me snel te helpen en mee na te denken. De andere projectpartners van PeMaTo en PeMaTo-EuroPep: Dirk, Jochem, Suh, Ard, Eva en Els.
Bedankt aan de EVECO collega’s om de afgelopen zes jaar te vullen met goede raad, hulp en plezier. In het bijzonder grote Natalie, bedankt om het labo open te stellen voor een mijtenkweek, om “de kleine vrienden” te verzorgen en voor de fijne babbels. Bureaugenootjes Sophie en Anneleen (CGB) en Myrthe en Omid (CDE), want een toffe sfeer op het werk is goud waard!
172
Mama en papa, jullie moedigden mijn interesse voor de levende wereld al aan van toen ik klein was en mijn beste insectendeterminatie bij “emelbees” strandde. Jullie gaven me de kans om te studeren en dit te bereiken. Oma en opa, hoeveel uren heb ik wel bij jullie doorgebracht, pratend, lachend, klagend, zuchtend, gefrustreerd of hoopvol. Jullie luisterden altijd weer en bleven me aanmoedigen en in mij geloven. Tante Tanja en Diane, jullie gingen me voor en die gedachte heeft me altijd dat extra duwtje in de rug gegeven.
De bio-vrienden: Marjolein, Lore, Winnie, Cosette, Stephanie, Carlynn, Kelly, Leandro, Tim, Lieven, Matthias, Philip, Jonathan, … Al tien jaar lang amusement en steun verzekerd en het voelde goed om compagnons te hebben die in hetzelfde schuitje zaten. Ook na het afsluiten van het UAntwerpen-hoofdstuk zullen we er nog vele decennia bij doen, bedankt om de beste vrienden te zijn die ik me kan wensen!
Wim, jij krijgt een alinea voor jezelf. September 2009 was het toen ik de ondergrondse bioruimte ging verkennen en jij daar in jouw aquarium zat. Meteen een leuke babbel en al gauw werd de bioruimte één van de favoriete plekken op de campus. Je hebt me elk stapje in mijn opleiding tot bioloog zien zetten, van de eerste wormen-dissectie, tot de Tanzania stage en het verdedigen van mijn masterthesis. Daarna bleef ik nog student en zo bleef ik ook de (nieuwe) bioruimte bezoekjes brengen. Bedankt voor alle gezellige momenten, de wandelingen in de biotuin, het Nachtegalenpark of Fort VI als ik mijn hart moest luchten, de gezellige lunchpauzes. Je bent er altijd geweest, altijd mijn rots in de branding op ’t unif, kortom, held!
Als laatste wil ik mijn collega’s van het team Online Banking bij Bank van Breda bedanken, met speciale vermelding voor Gert en Christophe. Een doctoraat afwerken in combinatie met een fulltime job is niet gemakkelijk en de rare corona-tijden maakten het er niet beter op. Maar door een fantastische werksfeer en veel begrip, steun, ruimte en interesse heb ik de combinatie dit jaar kunnen maken en met succes afronden, bedankt!
Foto fluweelmijt: Inge Hofmans
173
174