Introduction
Plants produce a wide variety of secondary metabolic compounds such as alkaloids that mediate plant–animal interactions (Schoonhoven et al., Reference Schoonhoven, Van Loon, van Loon and Dicke2005). These phytochemicals are found in nectar and in plant tissues at different concentrations and, although the dose-dependent role of some of them in attracting and rewarding pollinators has been demonstrated (Wright et al., Reference Wright, Baker, Palmer, Stabler, Mustard, Power, Borland and Stevenson2013; Stevenson et al., Reference Stevenson, Nicolson and Wright2017), they are generally regarded as defence mechanisms to reduce damage from non-adapted phytophagous animals, because they are bitter tasting and toxic (Szentesi and Wink, Reference Szentesi and Wink1991; Adler, Reference Adler2000; Thomson et al., Reference Thomson, Draguleasa and Tan2015; Muñoz et al., Reference Muñoz, Schilman and Barrozo2020; Mustard, Reference Mustard2020). In the context of evolutionary arms race, the complex mechanisms of secondary metabolite production evolved by plant species to resist herbivory have, in return, made herbivores adapt a wide range of mitigation processes of their negative effects (Dearing et al., Reference Dearing, Foley and McLean2005). For example, some generalist herbivore species, such as the aphid Myzus persicae (Hemiptera: Aphididae), show strain-specific adaptations to plant defences, conferring them resistance to lupanine or to nicotine (Cardoza et al., Reference Cardoza, Wang, Reidy-Crofts and Edwards2006; Ramsey et al., Reference Ramsey, Elzinga, Sarkar, Xin, Ghanim and Jander2014).
Evidence is accumulating that plant secondary metabolites can also mediate multitrophic interactions, well beyond plant–herbivore two-way interactions (Gols, Reference Gols2014; Harvey and Gols, Reference Harvey and Gols2018; Ode, Reference Ode2019; Sedio, Reference Sedio2019). Herbivore-induced plant volatiles (HIPVs) are known to impact the structure of species interactions in food-webs, because herbivores, pollinators, their natural enemies and competitors all respond to HIPVs (Vet and Dicke, Reference Vet and Dicke1992; Dicke and Baldwin, Reference Dicke and Baldwin2010). Concerning secondary metabolites that are ingested from the plant tissues, some herbivorous insects have evolved mechanisms of sequestration in a way that compounds can be transferred to higher trophic levels (Duffey, Reference Duffey1980; Szentesi and Wink, Reference Szentesi and Wink1991; Schmidt et al., Reference Schmidt, Brack, Rommé, Tyrell and Gehrt2000; Erb and Robert, Reference Erb and Robert2016). Predators and parasitoids usually suffer from reduced fitness when consuming herbivores sequestering secondary metabolites, or even just feeding on plants that produce them (Barbosa et al., Reference Barbosa, Saunders, Kemper, Trumbule, Olechno and Martinat1986; Gols, Reference Gols2014). For example, the caterpillar Manduca sexta (Lepidoptera: Sphingidae) can sequester nicotine, creating toxic conditions for the development of the endoparasite Apanteles congregatus [syn. Cotesia congregata] (Hymenoptera: Braconidae) (Thurston and Fox, Reference Thurston and Fox1972). As a textbook example, Campbell and Duffey (Reference Campbell and Duffey1979) demonstrated that the corn earworm Heliothis zea (Lepidoptera: Noctuidae) fed on artificial diet containing α-tomatine was toxic to its endoparasitoid Hyposoter exiguae (Hymenoptera: Ichneumonidae). Secondary consumer had lower pupal eclosion rates and was smaller in size, with potential detrimental effects on biological control strategies against this pest of tomatoes (Campbell and Duffey, Reference Campbell and Duffey1979).
Caffeine (1,3,7-trimethylxanthine) is a purine alkaloid present in various families of plants such as coffee (Gentianales: Rubiaceae) and tea plants (Theales: Theaceae), mostly in fruits and seeds, pollen, leaves, but also in the phloem and xylem sap (Mazzafera and Gonçalves, Reference Mazzafera and Gonçalves1999; Gonthier et al., Reference Gonthier, Witter, Spongberg and Philpott2011; van Breda et al., Reference van Breda, van der Merwe, Robbertse and Apostolides2013; Mustard, Reference Mustard2020). It has been detected in the floral nectar and pollen of Citrus (Sapindales: Rutaceae) and other plants from sub-tropical regions (Kretschmar and Baumann, Reference Kretschmar and Baumann1999; Sano et al., Reference Sano, Kim and Choi2013; Wright et al., Reference Wright, Baker, Palmer, Stabler, Mustard, Power, Borland and Stevenson2013), but also in the yaupon Ilex vomitoria (Celastrales: Aquifoliacae) (Power and Chesnut, Reference Power and Chesnut1919) and the linden Tilia spp. (Malvales: Tiliaceae) (Naef et al., Reference Naef, Jaquier, Velluz and Bachofen2004) in temperate regions. Caffeine effects are well documented on the animal brain and body, including on arthropods (Mustard, Reference Mustard2014; Thomson et al., Reference Thomson, Draguleasa and Tan2015). Its role in protecting plants against herbivores and pathogens (Ashihara et al., Reference Ashihara, Sano and Crozier2008; Sano et al., Reference Sano, Kim and Choi2013), and in influencing herbivore and pollinator behaviours is well understood (Wright et al., Reference Wright, Baker, Palmer, Stabler, Mustard, Power, Borland and Stevenson2013; Couvillon et al., Reference Couvillon, Al Toufailia, Butterfield, Schrell, Ratnieks and Schürch2015; Prado et al., Reference Prado, Collazo, Marand and Irwin2021). In Drosophila fruit flies (Diptera: Drosophilidae), for example, caffeine reduces sleep duration and increases locomotor activities at both day and night (Hendricks et al., Reference Hendricks, Finn, Panckeri, Chavkin, Williams, Sehgal and Pack2000; Nall et al., Reference Nall, Shakhmantsir, Cichewicz, Birman, Hirsh and Sehgal2016; Keebaugh et al., Reference Keebaugh, Park, Su, Yamada and Ja2017). At high doses, caffeine is toxic for phytophagous insects, it can shorten life span, and deter pollinators from visiting plants and reduce their memory (Ashihara et al., Reference Ashihara, Sano and Crozier2008; Nikitin et al., Reference Nikitin, Navitskas and Nicole Gordon2008; Mustard, Reference Mustard2014). At low doses in laboratory studies, caffeine enhances pollinator memory of reward and increases pollination (Wright et al., Reference Wright, Baker, Palmer, Stabler, Mustard, Power, Borland and Stevenson2013; Thomson et al., Reference Thomson, Draguleasa and Tan2015).
Studying the effects of caffeine on insects could help deepen our understanding of how such molecule mediates plant–insect interactions, through direct effects on the biology of primary consumer, but also how it could have knock-on effects on higher trophic levels and could modify interspecific interactions. Species interactions in crop systems with high concentrations of caffeine are mostly governed by caffeine resistance evolved by the different protagonists (Damon, Reference Damon2000), but what happens in trophic systems with low-caffeine plants with regards to the effect of the alkaloid remains to be studied. Ecological aspects of caffeine–insect interactions are also relevant in applied ecology; it has potential for manipulating the foraging behaviour of natural enemies of insect pests, by providing them with caffeinated food sources, but also for developing plants resistant to phytophagous animals (Cardoso et al., Reference Cardoso, Martinati, Giachetto, Vidal, Carazzolle, Padilha, Guerreiro-Filho and Maluf2014). For example, transgenic chrysanthemum (Asterales: Asteraceae) have been produced to express the caffeine biosynthetic pathway and produce caffeine in tissues, which helps the plants resisting aphids (Kim et al., Reference Kim, Lim, Kang, Jung, Lee, Choi and Sano2011). However, unlike some other secondary metabolites, caffeine has not received much attention with regard to its role in host–parasitoid interactions, even when focusing on specialized phytophagous insects on coffee plants (Damon, Reference Damon2000; Jaramillo et al., Reference Jaramillo, Borgemeister and Baker2006; Vega et al., Reference Vega, Infante, Castillo and Jaramillo2009; Green et al., Reference Green, Davis, Cossé and Vega2015).
Within both this fundamental and applied context, we used aphids and parasitoids as a model system to explore potential cascading effects of caffeine intake across trophic levels. Parasitoids are essential components of most terrestrial ecosystems and provide important ecosystem services, such as the regulation of herbivorous pests. Endoparasitoids are interesting models because they feed on the entire body of their hosts during development (Godfray, Reference Godfray1994). Parasitoids may thus directly encounter and ingest toxic metabolites in the haemolymph or tissues of their hosts, and ingestion of plant toxins by the herbivore may reduce the suitability of the host or alter its sensitivity to parasitism (Gols, Reference Gols2014; Ode, Reference Ode2019). In some cases, plant quality may reduce the host's immune response (e.g., encapsulation), and benefit the parasitoid (Bukovinszky et al., Reference Bukovinszky, Poelman, Gols, Prekatsakis, Vet, Harvey and Dicke2009). Finally, parasitoids have developed many ways to assess host quality which modifies host handling behaviour and foraging decisions (Godfray, Reference Godfray1994; Wajnberg, Reference Wajnberg2006). Optimal choices of hosts and nutritional resources have to be made to increase offspring quality and survival rates. In particular, we questioned how parasitoids could adapt their behavioural response to the host's diet, and how the oviposition decision on hosts fed on caffeine diet could ultimately influence the parasitism success, and the fitness of the offspring generation. We hypothesized that if caffeine negatively affected the primary consumers' fitness, it would also affect the secondary consumer's host handling behaviour, and would cascade to the next parasitoid generation by negatively affecting their survival and life-history trait values.
Materials and methods
Biological material
We used the generalist and major pest aphid M. persicae as a model for the primary consumer compartment, and the generalist parasitoid Aphidius matricariae (Hymenoptera: Braconidae) as a model for the secondary consumer. For a fundamental study, this is a convenient system to work with because interactions between M. persicae and A. matricariae are well explored, because artificial diets are already developed for the aphid, and because M. persicae is already known to develop alkaloid-resistant strains when shifting host plants (Cardoza et al., Reference Cardoza, Wang, Reidy-Crofts and Edwards2006; Ramsey et al., Reference Ramsey, Elzinga, Sarkar, Xin, Ghanim and Jander2014). Studying the effects of alkaloids in generalist herbivores such as M. persicae is important because they feed on diverse families of plants and are thus likely to be exposed to a great variety of secondary metabolites. Aphidius matricariae is a generalist aphid parasitoid species that is widely used for the biological control of aphid pests (Hance et al., Reference Hance, Kohandani-Tafresh, Munaut, van Emden and Harrington2017). Parasitoids and aphids were purchased from the Viridaxis SA company (Charleroi, Belgium), and no reported resistance or tolerance to alkaloids exists in these strains. Experiments were done in the laboratory at 20 ± 2°C, 60 ± 10% relative humidity and 16:8 h light:dark photoperiod.
Effect of caffeine on primary consumers
Two treatments were applied: artificial diet alone, provided by Viridaxis SA and specifically developed for M. persicae (e.g., on other aphid species: Cambier et al., Reference Cambier, Hance and De Hoffmann2001; van Emden and Wild, Reference van Emden and Wild2020), and artificial diet with caffeine at 0.1 mg ml−1 (0.5 mM). Note that during preliminary assays, aphids exposed to a 1 mg ml−1 (5 mM) caffeine treatment all died before reaching the reproductive period (i.e., null fecundity), and before turning into third-instar larvae (mean longevity 2.2 ± 0.2 days). Comparisons were thus only done between the control and the caffeine 0.1 mg ml−1 treatments. Because the concentration of caffeine in the sap of most plants on which this aphid feeds is unknown, these concentrations were chosen according to preliminary results on aphid parasitoids (Tougeron et al., unpublished data), and to previous work on Drosophila melanogaster (Nikitin et al., Reference Nikitin, Navitskas and Nicole Gordon2008; Keebaugh et al., Reference Keebaugh, Park, Su, Yamada and Ja2017) that are comparable in size with large aphids and aphid parasitoids. To feed aphids, 600 μl of the solution were deposited on the surface of a small sterilized plastic petri dish (Ø 5 cm) and it was immediately covered with a piece of stretched parafilm to create a thin membrane through which aphids could feed (Pirotte et al., Reference Pirotte, Lorenzi, Foray and Hance2018; van Emden and Wild, Reference van Emden and Wild2020). Thirty adult female aphids were taken from the cultures and placed on a fresh Brassica rapa (Brassicales: Brassicaceae) leaf until they deposited larvae. Among newborn aphids, 20 of them per treatment were placed individually on artificial feeders and monitored daily for their longevity, and fecundity (daily mean offspring produced by adult female and total offspring number). Newborn larvae were removed from the artificial diet feeder as they were counted. Aphids were gently moved to new feeders every 3 days to ensure good quality of the artificial diet.
Cascading effects on secondary consumers
Mated parasitoid females (<48 h old, at their egg-load peak) and fed with a 50% honey dilution were exposed to 15 third-instar aphids, which is the favourite host stage of this species (Rezaei et al., Reference Rezaei, Talebi, Fathipour, Karimzadeh and Mehrabadi2019), and that had been allowed to settle for 20 min on a B. rapa leaf placed in a glass petri dish (Ø 10 cm) on a 1.5% agar substrate, to avoid any potential effect of the artificial diet feeder on parasitoid behaviour. This experiment was repeated ten times (with N = 10 parasitoid females per treatment) for aphids exposed to the control treatment and the 0.1 mg ml−1 caffeine treatment since they were born (i.e., for around 4 days). No visible effect of caffeine was detected on the aphid development rate up to the third larval instar.
For each arena, we monitored by direct observation through binocular microscope the proportion of contacted aphids each parasitoid actually oviposited in (acceptance rate), and the time spent from host encounter (usually followed by antennal evaluation) to leaving the aphid with or without sting (mean handling time) (Gerling et al., Reference Gerling, Roitberg and Mackauer1990). The mean number of kicks by aphids (defence strategy) was recorded for each contacted aphid. The experience ended after 15 min or after the parasitoid had made antennal contact with all 15 aphids at least one time. Aphids were kept by groups of 15 on B. rapa leaves and the parasitism rate was estimated as being the number of aphids that turned into mummies (dead aphid containing a parasitoid pupa)/the total number of aphids in which oviposition occurred. Emergence rate was evaluated for each clutch as the number of mummies from which a parasitoid emerged/the total number of mummies. After the emergence of the following generation, parasitoids were sexed and individually kept in plastic tubes (1.5 ml) with honey and water, and their longevity was monitored. The day they died, we measured parasitoid fresh mass and the size of the left hind leg tibia, as standard proxy traits of parasitoid fitness (Godfray, Reference Godfray1994).
Finally, we used a dissection experiment to complete data on parasitism success and try to detect whether parasitoid mortality occurs mainly during the early immature stages or during pupation. Four additional batches of 15 parasitized aphids, fed on either treatment (caffeine or control), were each exposed to one different parasitoid female. The same protocol as for the main experiment was used, and each aphid in which oviposition occurred was placed on a fresh B. rapa leaf, until we obtained 15 parasitized aphids per assay. Dissections were done under stereo microscope on a total of 40 surviving parasitized aphids (ten per batch) on the seventh day of their development; before the potential formation of the mummy, but late enough in the parasitoid development cycle to observe if late instar larvae were alive (fresh and moving) or dead (dry, dark-coloured and not moving).
Statistical analyses
Generalized Linear Models (GLMs) were fit to the data to analyse the effect of the caffeine treatment on total and daily offspring produced, assuming quasi-Poisson distribution, and a Cox model was used to analyse aphid longevity data. Generalized Linear Mixed Models (GLMMs) were used to compare the handling time of aphids by parasitoids between treatments assuming a Gaussian distribution, and to compare the number of kicks, assuming a Poisson error distribution, using the identity of the parasitoid female as a random effect. Acceptance rates, parasitism rates and emergence rates were compared between treatments using GLMMs assuming binomial error distributions. Fresh mass and tibia size of the new generation were compared between treatments and sexes using GLMMs assuming Gaussian error distributions, longevity was compared using a Cox survival model fitted with the package coxme (Therneau, Reference Therneau2015), and using the mother identity as a random effect for both types of models. Models were tested using type II ANOVA from the package car (Fox and Weisberg, Reference Fox and Weisberg2011). All statistical analyses were carried out using the R software (R Core Team, 2020).
Results
Effect of caffeine on primary consumers
Aphids had a lower longevity and produced fewer nymphs when exposed to caffeine, but the caffeine treatment did not affect the mean daily offspring number produced by aphid females (table 1).
Life-parameter table and statistical results of Cox model (longevity) or GLMs (total and daily offspring), for Myzus persicae fed on either artificial diet alone (control) or added to 0.1 mg ml−1 caffeine
Mean daily offspring number is provided for the reproductive period. N = 20 female aphids per treatment. DF = 1 for all tests.
Cascading effects on secondary consumers
There were no differences in host handling time by parasitoids, nor in kicking occurrence by aphids between the two feeding treatments. Acceptance rate and emergence rate were similar between treatments. Parasitism rate (i.e., successful mummy formation) was 16% lower for parasitoids exposed to aphids fed with caffeine compared to parasitoids exposed to control aphids (table 2). After dissection at day 7 (before mummy formation), we found 60% survival in parasitoids in control aphids (N = 15), and 46% survival for those parasitizing aphids with caffeine diet (N = 13).
Behavioural parameter table and statistical results of GLMMs, for ten Aphidius matricariae females per treatment, each exposed to 15 Myzus persicae fed either on artificial diet alone (control) or added to 0.1 mg ml−1 caffeine
DF = 1 for all tests. N represents the total number of aphids on which each parameter was evaluated.
Sex ratio (♂:♀) was (1:1.5) for control treatment and (1:1.3) for the caffeine treatment. Parasitoid longevity (Cox model, z = −0.67, DF = 1, P = 0.50), fresh mass (GLMM, χ2 = 2.18, DF = 1, P = 0.14) and tibia size (GLMM, χ2 = 0.04, DF = 1, P = 0.83) did not differ between sexes so data were pooled for the rest of the analyses. There was a marginally significant effect of the aphid diet treatment origin on parasitoid longevity (χ2 = 3.4, DF = 1, P = 0.048). Parasitoids that have developed on aphids fed on control diet lived on average 4.5 days longer than on aphids fed with 0.1 mg ml−1 caffeine. There were no differences in parasitoid fresh mass (z = −2.13, DF = 1, P < 0.05) and parasitoid tibia length (χ2 = 2.8, DF = 1, P = 0.09) between treatments (fig. 1, control: N = 35, and caffeine: N = 23 parasitoids of the next generation).
Survival curves (±95% CI), fresh mass and mean tibia size of Aphidius matricariae parasitoids of the next generation which developed on aphids that had either fed on control artificial diet (black) or on artificial diet with 0.1 mg ml−1 caffeine (grey). Dotted lines represent 50% parasitoid survival. N = 35 (control) and 23 (caffeine). NS, not significant.
Discussion
Our results confirm the historical observations that alkaloids and other secondary metabolites can be involved in plant–herbivore–secondary consumer tritrophic interactions, in addition to directly affecting plant–herbivore and plant–pollinator relationships (Campbell and Duffey, Reference Campbell and Duffey1979; Ode, Reference Ode2019). On this point however, previous work has focused on a limited set of secondary plant metabolites, mostly acutely toxic compounds of specific plant families, and some molecules such as caffeine have been understudied within the context of host–parasitoid interactions. We report a caffeine effect in a cosmopolitan generalist herbivore and parasitoid species, in which no particular resistance, detoxification or sequestration mechanisms to caffeine have been reported. Myzus persicae feeds on many plant families, and A. matricariae attacks many aphid genera (Hullé et al., Reference Hullé, Chaubet, Turpeau and Simon2020), so they both could be exposed to caffeine, although at relatively low concentrations because these species are not related to tea or coffee crops. This study model opens interesting perspectives to understand both how caffeine affects host–parasitoid interactions, and how this molecule may be used in the future for the manipulation of insects in biological control strategies.
We observed a diminution of aphid longevity, therefore reducing their reproductive period and total progeny, but not their fecundity, as mean daily offspring production was not affected. Aphid defence behaviour (kicking) was not affected by caffeine intake, suggesting that they keep responding to environmental stimuli such as visual cues of a parasitoid approaching, or conspecific alarm pheromones. In parasitoids, we initially hypothesized that both the behaviour of the parental generation, and the survival and traits of the offspring generation would be influenced by caffeine intake by the host. However, we observed that parasitoid behaviour (host handling time and aphid acceptance rate) was not affected by the caffeine treatment, at the tested dose. It is possible that aphids feeding on even higher concentrations of caffeine would be rejected by parasitoids after antennal or ovipositor contact (allowing detection of chemical cues). However, it would require the parasitoid to be able to detect caffeine within the host prior to oviposition, for which we have no evidence.
Nevertheless, the decision made by the parental generation to oviposit in a host fed on caffeine diet affected the following generation of parasitoids. The lower parasitism rates recorded in aphids exposed to caffeine intake, but similar emergence rates, as also supported by dissection results, are indicative of lower success and higher mortality primarily of early parasitoid developmental stages (egg to prepupa, before mummy formation), which may be due to caffeine toxicity (Mustard, Reference Mustard2020). Of course, the caffeinated diet of the host may still affect parasitoid pupa formation to some extent, and additional deaths may occur in parasitoids at this critical stage of metamorphosis. We finally report an effect of aphid diet on parasitoid adult longevity, a fitness indicator, either because they parasitize low-quality aphids, or because they are directly affected by caffeine when they feed on aphid tissues. Parasitoids are likely at risk of toxicity transfer from their host, as shown, for example, for the snowdrop lectin entomotoxin transferred from the diet of M. persicae to tissues of Aphidius ervi (Hymenoptera: Braconidae) endoparasitoid pupae (Couty et al., Reference Couty, Down, Gatehouse, Kaiser, Pham-Delegue and Poppy2001). The next steps of the work will be to perform LC-MS or HPLC analyses (e.g., Kim et al., Reference Kim, Lim, Kang, Jung, Lee, Choi and Sano2011; Mendes et al., Reference Mendes, Coelho, Tomé, Cunha and Manadas2019) to try detecting if caffeine – or its main metabolites – has been transferred from aphid to parasitoid tissues. More generally, host quality (size, age, nutritional content, immune defences) is crucial for proper parasitoid development and survival (Harvey, Reference Harvey2000), but can be altered by the plant quality (Harvey and Gols, Reference Harvey and Gols2018). When feeding caffeine directly to parasitoids in their nectar diet, we observed that their longevity decreased (Tougeron et al., unpublished data). It is thus likely that what we observed in the present study can arise from direct effects on the parasitoid larvae being surrounded by toxic compounds in the host, and not only due to aphids being less suitable hosts, having been poisoned by the alkaloid intake (Gols, Reference Gols2014).
Although the aim of our study was not to mimic the presence of caffeine as it occurs in natural environments, defining a concentration to test in the lab was not straightforward. Indeed, caffeine concentrations vary a lot among species of the same genera and among varieties of the same species (Wright et al., Reference Wright, Baker, Palmer, Stabler, Mustard, Power, Borland and Stevenson2013; Pham et al., Reference Pham, Ismail, Mishyna, Appiah, Oikawa and Fujii2019), and each plant organ has variable concentrations (Kretschmar and Baumann, Reference Kretschmar and Baumann1999; Dado et al., Reference Dado, Asresahegn and Goroya2019). Alkaloid levels in the xylem or phloem sap of most plant species have rarely been quantified (Mazzafera and Gonçalves, Reference Mazzafera and Gonçalves1999), and as concentrations may differ between leaf tissues and sap, it is possible that their effects could vary between leaf consumers and sap-feeding animals (Mazzafera and Gonçalves, Reference Mazzafera and Gonçalves1999; Gonthier et al., Reference Gonthier, Witter, Spongberg and Philpott2011; van Breda et al., Reference van Breda, van der Merwe, Robbertse and Apostolides2013). More generally, studying caffeine – and other alkaloids – concentrations in more plant species and in various organs is needed. Caffeine has been detected at the concentrations of ~0.25 and 0.10 mM in floral nectar of Coffea canefora and Coffea liberica, respectively, and the concentration was 0.02 mM in Citrus paradisi (Wright et al., Reference Wright, Baker, Palmer, Stabler, Mustard, Power, Borland and Stevenson2013). Therefore, the tested concentrations of caffeine in our study (0.1 mg ml−1, 0.5 mM) were higher than those typically found in nature, but we tested within the millimolar range in which most experimental studies have found caffeine effects on both vertebrates and invertebrates (Nikitin et al., Reference Nikitin, Navitskas and Nicole Gordon2008; Mustard, Reference Mustard2014; Thomson et al., Reference Thomson, Draguleasa and Tan2015).
Myzus persicae is a generalist herbivore and some strains have expanded their host range to include the lupine, Lupinus angustifolius (Fabales: Fabaceae), and have become more resistant to the lupine-specific alkaloid lupanine than non-adapted lineages (Cardoza et al., Reference Cardoza, Wang, Reidy-Crofts and Edwards2006). Lineages that have expanded their host range to include tobacco also often have elevated nicotine tolerance (Zepeda-Paulo et al., Reference Zepeda-Paulo, Simon, Ramirez, Fuentes-Contreras, Margaritopoulos, Wilson, Sorenson, Briones, Azevedo and Ohashi2010), due to the production of detoxifying enzymes (Ramsey et al., Reference Ramsey, Elzinga, Sarkar, Xin, Ghanim and Jander2014). Myzus persicae is active on Citrus spp. trees (Ghosh et al., Reference Ghosh, Das, Lepcha, Majumdar and Baranwal2015) and on other caffeine-producing plant species, but the extent to which they are actually exposed to caffeine on these plants, and how strains could adapt resistances to caffeine remains to be studied. Ramsey et al. (Reference Ramsey, Elzinga, Sarkar, Xin, Ghanim and Jander2014) reported that a low concentration (0.01 mM) of nicotine stimulated the fecundity of the nicotine-tolerant M. persicae strain, showing hormetic effects that were also reported in bees (Cutler and Rix, Reference Cutler and Rix2015). In our study on caffeine, the 5 mM dose was lethal for the strain of M. persicae that we used, while 0.5 mM allowed aphids to survive, although with fitness costs. Lower doses of caffeine might be necessary to unravel any hormesis effect on this aphid species. In any case, further studies should focus on assessing the resistance threshold of herbivorous insects and their natural enemies to different types of alkaloids, and also determining variations in resistance between insect species and lineages. In the context of host–parasitoid evolutionary arming race, it is crucial to understand the relative costs and benefits of exposure to plant-secondary metabolites, for each of the interacting species (Gols, Reference Gols2014), which could depend on the concentration and on the nature of the secondary metabolite.
To conclude, using a host-parasitoid insect model, we described how caffeine, a plant secondary metabolite overlooked in non-coffee or tea plant systems, has detrimental effects on two trophic levels. It highlights the importance of such molecules not only regarding plant–animal interactions, but also regarding food-web and ecosystem functioning. The effects of caffeine on insect circadian activities, behaviour and physiology have been extensively studied (Damon, Reference Damon2000; Mustard, Reference Mustard2014; Keebaugh et al., Reference Keebaugh, Park, Su, Yamada and Ja2017), but how they translate to species interaction and community-level ecological effects, for example, via sleep disruption, remains overlooked (Tougeron and Abram, Reference Tougeron and Abram2017). Of course, the study of host–parasitoid interactions in coffee production systems is also interesting, because caffeine concentrations are high and herbivores have developed detoxification mechanisms (Infante, Reference Infante2018), and because not all parasitoids have proven successful to control these phytophagous insects (Jaramillo et al., Reference Jaramillo, Borgemeister and Baker2006). Manipulation of caffeine-producing plants for pollination, or for pest control by boosting the predatory activity of natural enemies or by increasing plant resistance are interesting perspectives (Kim et al., Reference Kim, Lim, Kang, Jung, Lee, Choi and Sano2011; Cardoso et al., Reference Cardoso, Martinati, Giachetto, Vidal, Carazzolle, Padilha, Guerreiro-Filho and Maluf2014; Thomson et al., Reference Thomson, Draguleasa and Tan2015), but one first needs to assess potential trade-offs with the survival of upper trophic levels and other organisms in the interacting network.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0007485321000687
Data
Raw data can be found in the Supplementary material file.
Acknowledgements
We thank L. Dhondt for her help in preparing the artificial diet feeders, and R. Hoffmann for his participation in the caffeine project. This article is No BRC 371. KT was supported by the F.R.S.-FNRS. We thank two anonymous reviewers and the editor for their work on our manuscript.
Author contributions
KT conceived the study, designed the experiment, collected and analysed data and wrote the manuscript. TH secured funding and revised the article.
Conflict of interest
None.
