Effects of elevated atmospheric carbon dioxide on insect–plant interactions

Abstract

The rarefied atmosphere of US politics has been heating up recently. During March, President George W. Bush contributed significantly to the accumulation of hot air over Washington and elsewhere by announcing two major policy shifts related to the emission of CO2 into the atmosphere. First, President Bush reneged on a campaign pledge to regulate carbon dioxide emissions from U.S. power plants. Soon afterwards, his administration explicitly opposed the Kyoto Protocol, the international agreement setting country-by-country limits on emissions of greenhouse gases. These two policy positions ensure that little will change in the near future to reduce the amounts of CO2 entering the atmosphere from US sources. In short, it is a great time to be in the CO2 business. Entomologists have recognized for some time that elevated concentrations of atmospheric CO2 may influence the distribution, abundance and performance of insects that feed on plants (Lincoln et al., 1984, 1986; Fajer et al., 1989). Major questions remain, however, on the relative importance of changes in weather, changes in plant quality and changes in predation pressure on the dynamics of insect herbivore populations under conditions of elevated CO2. If you grow crops or trees for a living, these questions boil down to one simple concern; will yields increase or decrease as CO2 levels continue to rise? The problem is that we do not really know yet, and it is going to cost significant sums of money to find out. There are too many interacting variables to make simple predictions about changes in pest damage to forestry and agricultural commodities. CO2–mediated changes in temperature or precipitation may affect insects directly and may influence the geographical ranges of agricultural and natural plant communities (Cannon, 1998). The predators, parasites and pathogens that maintain some level of control over insect populations may also be affected by global climate change or changes in plant phenotype (Stiling et al., 1999). Changes in the nutritional and defensive characteristics of host plants may drive changes in levels of insect damage to plants (Bezemer & Jones, 1998) and all of these ecological effects may interact with other sources of environmental variation including drought, nutrient availability and light (Arnone et al., 1995; Roth et al., 1997; Haettenschwiler & Schafellner, 1999; McDonald et al., 1999). In the longer term, elevated CO2 may influence the fundamental ecosystem properties upon which all plant productivity depends (Ball & Drake, 1997; Jones et al., 1998; Kampichler et al., 1998; Kandeler et al., 1998; Hungate et al., 1999; Strand et al., 1999). While presidents and policy makers play Russian roulette with the climate, ecologists and entomologists are exploring with ever-increasing accuracy and complexity the potential ramifications of elevated CO2 for plant–herbivore interactions. Without long-term studies of the crucial variables, we will simply be unprepared for the enriched CO2 atmosphere that is developing (Coviella & Trumble, 1999). Atmospheric CO2 concentrations have already risen by about 25% since the industrial revolution and are expected to increase from current ambient levels of 350–360 p.p.m. (or µL/L) to around 600 p.p.m. by the end of the century (Houghton et al., 1995). All of the potential consequences of elevated CO2 concentrations are too great to cover in detail here. In this paper, I focus upon what we know about changes in plant quality under elevated CO2 and how changing food quality might interact with other ecological variables to alter the performance and abundance of insects on plants. As the principle source of carbon for photosynthesis, it should be no surprise that changes in concentrations of CO2 have marked effects upon the phenotype of plants (Lincoln, 1993; Ceulemans & Mousseau, 1994; Curtis & Wang, 1998). For example, elevated CO2 generally results in increased rates of photosynthesis (Drake et al., 1997; Norby et al., 1999), increased rates of growth (Saxe et al., 1998) and increased biomass (Leadley et al., 1999; Owensby et al., 1999). Assuming no concurrent changes in nutrient availability, the accumulation of biomass under elevated CO2 dilutes concentrations of nitrogen in tissues by 15–25% (Lincoln et al., 1993; Lindroth et al., 1995), thereby increasing C : N ratios (Ceulemans & Mousseau, 1994; Wilsey, 1996; Hughes & Bazzaz, 1997) and the allocation of carbon to some carbon-rich secondary metabolites (Lindroth et al., 1995; Agrell et al., 2000). Elevated concentrations of CO2 may sometimes (Agrell et al., 2000) but not always (Thompson & Drake, 1994; Bezemer & Jones, 1998) decrease the water content of foliage and increase rates of leaf abscission and plant senescence (Paez et al., 1983; Houpis et al., 1988; Baxter et al., 1994; Sicher & Bunce, 1997). Of course, not all plant species respond identically to elevated concentrations of CO2 (Lindroth et al., 1993). For example, elevated CO2 results in reduced foliar nitrogen levels and increased condensed tannin levels in paper birch but not in white pine (Roth & Lindroth, 1994). In further studies with paper birch, quaking aspen and sugar maple (Roth et al., 1998; Agrell et al., 2000), all species show increases in foliar concentrations of condensed tannins under elevated CO2. However, the foliage of quaking aspen also expresses higher concentrations of phenolic glycosides, and sugar maple is the only species to show elevated foliar concentrations of hydrolysable tannins. In a study contrasting a C3 sedge with a C4 grass in marsh habitat, Thompson & Drake (1994) reported CO2-mediated declines in foliar nitrogen only in the sedge. The grass, in contrast, exhibited increases in foliar water concentrations and concomitant increases in fungal infection. Despite the predictions of the carbon-nutrient balance hypothesis (Chapin, 1980; Bryant et al., 1983) that all carbon-rich secondary metabolites should increase under elevated CO2, this appears not to be the case. For example, foliar concentrations of the iridoid glycosides in Plantago lanceolata are unaffected under CO2 enrichment (Fajer et al., 1989, 1991). Likewise, the volatile terpenoids of peppermint (Lincoln & Couvet, 1989), big sagebrush (Johnson & Lincoln, 1990) and loblolly pine (Williams et al., 1997a) do not vary with experimental increases in CO2. Because nitrogen concentrations in foliage are diluted by the increased C : N ratio of plant tissues under elevated CO2, there is the potential for reduced efficacy of nitrogen-based plant defences. In an interesting twist to this hypothesis, Coviella et al. (2000) reported that transgenic cotton grown under elevated CO2 expressed reduced concentrations of Bt protein. In bioassays with Spodoptera, larval performance increased under elevated CO2 because increased larval consumption in response to low nitrogen levels did not compensate for reductions in Bt protein. Before you dash out and sell your biotech stocks, it is obviously far too early to say that the use of Bt transgenes will be compromised under elevated CO2, but it certainly merits further study. The Bt-lovers amongst us may be gratified to learn that the efficacy of conventional topical applications of Bt may be enhanced under elevated CO2. Experiments suggest that increased consumption by insects to compensate for high C : N ratios results in greater exposure to Bt and higher levels of mortality (Coviella & Trumble, 2000). We are dealing with complex organisms and there will be no perfect generalities for the way that plant phenotype changes under elevated CO2. Nonetheless, in nearly every case examined to date, foliar nitrogen concentrations decline under elevated CO2 and, when present, foliar concentrations of condensed tannins increase (Fajer et al., 1989, 1991; Johnson & Lincoln, 1991; Lincoln et al., 1993; Lindroth et al., 1995). This level of generality is somewhat heartening and allows us to predict that overall decreases in foliar quality should induce at least some insect herbivores to eat more. And that prediction usually holds true. Lower levels of nitrogen and higher C : N ratios in plants under elevated CO2 have generally been associated with compensatory feeding and subsequent increases in levels of damage or defoliation (Lincoln et al., 1984, 1986; Fajer et al., 1989; Lincoln et al., 1993; Lindroth et al., 1993, 1995; Salt et al., 1995; Docherty et al., 1996; Kinney et al., 1997; Williams et al., 1997a). Leaf-chewing insects such as grasshoppers (Johnson & Lincoln, 1990, 1991) and caterpillar larvae (Lindroth et al., 1993, 1995) generally consume more leaf area when they are fed plants that have been grown under elevated CO2. Likewise, the area damaged by leaf-mining insects may also increase (Salt et al., 1995). For example, the area of leaf mines on Quercus myrtifolia increased by over 25% under elevated CO2, apparently because nitrogen concentrations fell by over 11% (Stiling et al., unpublished data). However, this is where the first complicating factor arises: simply because per capita consumption of foliage by insects increases under elevated CO2, it does not mean that plants suffer more damage overall. Two additional effects that mediate the ultimate level of damage that plants receive are CO2-induced increases in plant biomass and changes in insect density. It is well established that many plants accumulate more biomass under elevated CO2 (Leadley et al., 1999; Owensby et al., 1999) and that such direct effects of CO2 on plant growth can more than compensate for increases in defoliation (Caulfield & Bunce, 1994). For example, even though per capita rates of consumption by insects on Q. myrtifolia increase with CO2 enrichment, the proportion of leaves damaged by mining and chewing insects actually declines (Stiling et al., unpublished data). The leaf area index on Q. myrtifolia increases by 26% under elevated CO2 but this ignores the impact of reductions in damaged leaf area. Calculations suggest that undamaged leaf area actually increases by 38% when effects on insects are considered. Similarly, leaf area increases of 1.6-fold on milkweed under elevated CO2 jump to 3.6-fold increases in undamaged leaf area when the effects on herbivorous thrips are accounted for (Hughes & Bazzaz, 1997). Ultimately, the effects of increases in atmospheric CO2 on damage by insect pests will depend upon changes in insect performance at the individual and population levels. The ability of insects to compensate for CO2-mediated reductions in foliage quality is key to understanding long-term effects on herbivore population dynamics and the injury that will be inflicted upon hosts of economic importance. If eating more allows insects to compensate fully, then defoliation levels will rise while insect fitness remains constant. The question then becomes whether CO2-mediated increases in plant productivity are sufficient to offset increases in defoliation levels and which effect is more important to the part of the crop that is harvested for human use. Some insects can certainly compensate well when foliage quality declines. For example, red-headed pine sawfly larvae increase nitrogen utilization efficiency in response to CO2-mediated declines in foliar nitrogen in loblolly pine (Williams et al., 1994). The result is that their rates of nitrogen accumulation remain unchanged. Potential mechanisms by which insect herbivores may compensate for CO2-mediated changes in plant quality are diverse. For example, the activity of detoxification enzymes may be stimulated by increased concentrations of secondary metabolites in foliage (Lindroth et al., 1993). The good news, if you grow plants for a living, is that most insects appear to be unable to compensate fully for CO2-mediated reductions in plant quality. For example, buckeye butterflies on Plantago lanceolata exhibit both higher rates of mortality and increased development time when fed on plants grown under elevated CO2 (Fager et al., 1989; Fajer et al., 1991). Higher rates of insect mortality have been associated with nutritional deficiency that results from reduced foliar nitrogen concentrations under elevated CO2 (Brooks & Whittaker, 1999; Stiling et al., 1999). However, direct effects of changes in plant quality on insect performance are not always dramatic. For example, Lindroth et al. (1995) explored the performance of three species of saturniid moths feeding on paper birch under elevated CO2. Birch leaves were lower in nitrogen (23%), higher in condensed tannin (two-fold increase) and foliar C : N ratios increased from 12.7 to 28.1. Despite these significant reductions in foliage quality, survival of first-instar larvae declined only marginally, while fourth-instar larvae exhibited moderate increases in rates of consumption and decreases in rates of growth, development and food processing efficiency. Brooks & Whittaker (1998, 1999) have studied multiple generations of insects reared on plants under elevated CO2. In their first experiment (Brooks & Whittaker, 1998), Gastrophysa leaf beetles grown on Rumex plants for three consecutive generations exhibited relatively minor effects of elevated CO2 on performance, despite measurable declines in indices of foliage quality. Fecundity and egg size were reduced by the end of the second generation, which led to fewer, smaller larvae in the third generation. In the second study (Brooks & Whittaker, 1999), they reported reductions in the survival of nymphal spittlebugs in two sequential generations under elevated CO2. There were also declines in the rate of development in consecutive years. Such multigenerational studies are crucial if we are to develop any kind of realistic predictions of long-term population dynamics (Williams et al., 1997a). Perhaps more dramatic effects upon insect performance will be mediated by the third trophic level. Given that rates of insect growth also seem to decline under elevated CO2 (Fajer et al., 1989; Lindroth et al., 1995; Smith & Jones, 1998), we might expect that the risk of mortality from natural enemies will increase. However, two laboratory studies have failed to provide evidence for greater mortality imposed by natural enemies under elevated CO2 (Roth & Lindroth, 1995; Bezemer et al., 1998). For example, effects of parasitism by Cotesia melanoscela on the performance of gypsy moth larvae do not differ between ambient and elevated CO2 treatments (Roth & Lindroth, 1995). Nonetheless, field studies using open-top chamber technology have demonstrated that rates of leaf-miner parasitism increase under elevated CO2 (Stiling et al., 1999; Stiling et al., unpublished). Nitrogen concentrations in the foliage of two dominant oak species, Quercus myrtifolia and Q. geminata, decline under elevated CO2 and densities of leaf-mining insects are lower because of the combined effects of reduced foliage quality and increased rates of attack by parasitoids. Overall, death from plant effects increases by 50% and death from parasitoids by over 50% under elevated CO2. We have barely begun to study the effects of elevated CO2 on the efficacy of natural enemies and much work needs to be done. Effects of CO2 enrichment on insect–pathogen interactions have rarely been considered, although links between foliar phenolic concentration and viral infection of caterpillars are well established (Keating & Yendol, 1987; Hunter & Schultz, 1993). However, in at least one study (Lindroth et al., 1997), susceptibility of gypsy moth larvae to NPV was unaffected by CO2-mediated changes in foliar phenolics. Effects upon parasites, predators and pathogens under field conditions would appear to be a priority for future research. If declines in foliar nitrogen and increases in foliar C : N ratios are generally predictable responses of plants to elevated concentrations of CO2, there seems to be less predictability in the responses of insects. Although levels of consumption by insects usually rise under elevated CO2, additional effects upon performance appear to be somewhat idiosyncratic. For example, reductions in gypsy moth performance on aspen are associated with CO2-induced increases in phenolic glycosides (McDonald et al., 1999). Although nutrient and secondary chemistry of birch and maple are also affected by CO2, there are no parallel changes in gypsy moth performance. In other words, CO2-mediated effects on insect herbivores will depend both on the species of plant and the species of insect under study (Lindroth et al., 1995; Traw et al., 1996; Coviella & Trumble, 1999). In experiments with white marked tussock moth, Agrell et al. (2000) have shown that it is possible to rank host plants based upon their deleterious effects on insects under elevated CO2. Effects on larvae were most pronounced when fed upon quaking aspen, followed by paper birch, and least pronounced on sugar maple. Moreover, effects of elevated CO2 may vary seasonally as both leaves and herbivores age. On oaks, increases in foliar C : N ratios become more pronounced as leaves age, yet negative effects on caterpillars are more pronounced in early instars on younger foliage. Older larvae appear to be better able to compensate for reductions in foliar quality than are young larvae (Williams et al., 1998). On evergreen species such as loblolly pine, elevated concentrations of CO2 can interact with natural between-year differences in foliage quality to influence the palatability of leaf resources for insect larvae (Williams et al., 1997b). Finally, not all insects respond negatively to the changes in plant phenotype that are mediated by elevated concentrations of CO2. At least some phloem-feeding insects exhibit increases in performance when provided with plants grown under elevated CO2 (Awmack et al., 1997; Bezemer & Jones, 1998). As before, however, there appears to be qualitative variation in the responses of phloem-feeding insects to changes in plant quality. For example, the same clone of the aphid Aulacorthum solani responds differently to elevated CO2 on two different plant species (Awmack et al., 1997). On bean plants, the daily rate of nymph production increases by 16%, whereas rates of development are unaffected. In contrast, aphids on tansy exhibit faster rates of development and no change in reproductive rate. Overall, aphid responses are positive under elevated CO2 on both host plants, but the mechanisms differ between hosts. What this suggests to me is that we are a long way from being able to predict changes that may occur in the population dynamics of important crop pests under elevated concentrations of atmospheric CO2. Initial laboratory studies under controlled conditions were necessary to determine the general responses of plant phenotype to elevated CO2 and their potential effects on insect herbivores. However, most laboratory studies provide grossly artificial environments in which plants are not limited by nutrient availability or light, where communities are simplified to two or three interacting species, where behavioural choices of herbivores are limited, and where stochastic fluctuations in the environment are eliminated (Lincoln et al., 1993; Arnone et al., 1995). In other words, they are free of any ecological complexity. In reality, the effects of elevated CO2 on plant phenotype, and subsequent insect responses, will be mediated by the availability of resources to plants such as water (Roth et al., 1997), light (McDonald et al., 1999) and nutrients (Arnone et al., 1995; Haettenschwiler & Schafellner, 1999) and modified by climatic and biotic variability. Interactive effects of light and CO2 on tree growth and secondary chemistry have been studied by McDonald et al. (1999). Reductions in insect performance on aspen grown under high light and elevated CO2 were dramatic. When tussock moth were reared on treatment foliage for the entire larval period, survival fell by 62% with concomitant decreases in growth rate and pupal weight. Reductions in gypsy moth performance on aspen were associated with both CO2- and light-induced increases in phenolic glycosides. The key finding of these studies is the degree to which phenotypic changes in plants under elevated CO2, and the subsequent effects on insect herbivores, depend upon the availability of light (McDonald et al., 1999; Agrell et al., 2000). This suggests that successional stage and community composition will influence the response of plants, and their insect herbivores, to elevated CO2. Nutrient availability is also likely to affect plant and insect responses to atmospheric change (Arnone et al., 1995). At the very least, nutrient limitation is likely to set limits on the gains in plant biomass that generally result from elevated CO2 environments (Johnson & Lincoln, 1991; Saxe et al., 1998). Given natural variation in nutrient availability and uptake by plants, and increasing levels of nitrogen deposition from anthropogenic sources, we should be aware of potential interactions between elevated CO2 and nutrient availability. Indeed, nitrogen deposition may mitigate the effects of elevated CO2 on insect performance. In one experimental study, Haettenschwiler & Schafellner (1999) exposed larvae of the nun moth to spruce trees grown under three levels of nitrogen deposition and three levels of CO2. The effects of the treatments on plant phenotype were generally opposite. Nitrogen deposition caused reductions in starch, condensed tannins and total phenolics and increases in sugar and nitrogen concentrations in spruce needles − the direct opposite of responses to elevated CO2. As a consequence, nitrogen deposition was able to mitigate, in part, the deleterious effects of elevated CO2 on nun moth performance. In contrast, Kinney et al. (1997) found relatively few interactions between nutrient availability and CO2 concentration in their studies of gypsy moth performance on aspen, oak and maple. Effects of the treatments were strongly host-specific rather than predictably opposing. In some systems, the consequences of nitrogen deposition may greatly outweigh those of elevated CO2. For example, Kerslake et al. (1998) have demonstrated that the defensive and nutritional phenotype of heather, Calluna vulgaris, does not change after 20 months of growth under elevated CO2. In contrast, simulated nitrogen deposition results in decreases in foliar C : N ratio and increases in shoot growth. Unlike other studies (Lindroth et al., 1995; Haettenschwiler & Schafellner, 1999; Agrell et al., 2000), neither elevated CO2 nor nitrogen deposition influenced the foliar phenolics of heather. Consistent with phenotypic changes in the plants, the performance of winter moth larvae was increased by nitrogen deposition and unaffected by elevated CO2. One dominant consequence of elevated levels of atmospheric CO2 is the predicted increase in global temperature. Yet studies that combine effects of CO2-enriched plants and elevated temperatures on insect performance are still remarkably rare. In one such study (Dury et al., 1998), a 3 °C increase in temperature reduced the nutritional quality of oak leaves by reducing foliar nitrogen concentrations and increasing foliar concentrations of condensed tannin. In other words, elevated temperature resulted in phenotypic changes in plants that were broadly similar to those mediated by elevated CO2. In combination with CO2-mediated reductions in the quality of primary and secondary leaf flushes, the elevated temperatures associated with global warming may significantly reduce the quality of plant food for insect herbivores. However, complex interactions among temperature, plant quality, insect performance and range expansion make predictions of pest problems in future global environments extremely difficult (Cannon, 1998). For example, the reduced rates of insect growth that have been observed under elevated CO2 may disappear as temperature increases in response to rising levels of atmospheric CO2 (Fajer et al., 1991). The bottom line? We need long-term, multifactorial experiments under field conditions to have any hope of predicting the interactive effects of CO2 and other ecological variables on the insect pests of crops and trees. Given this complexity, can we make any generalizations about the effects of elevated CO2 on insects that feed on plants? While there are always going to be effects that are species- and system-specific (Traw et al., 1996; Kinney et al., 1997; Roth et al., 1998; Coviella & Trumble, 1999; Agrell et al., 2000), we should not shy away from synthesis when justified (Fajer & Johnson, 1993). For example, Bezemer & Jones (1998) analysed data from 61 plant–herbivore combinations and found some compelling patterns. First, they confirmed the general decreases in foliar nitrogen concentration (15%) and increases in carbohydrate (47%) and phenolic-based secondary metabolites (31%) reported in many individual studies. Second, consumption by herbivores was related primarily to changes in nitrogen and carbohydrate levels. Third, no differences were found between CO2-mediated herbivore responses on woody and herbaceous plant species. Fourth, leaf-chewing insects generally increased their consumption of foliage (30%) under elevated CO2 to compensate for reduced nutritional quality and suffered no adverse effects upon pupal weights. Fifth, leaf-mining insects could only partially compensate by increased consumption and their pupal weights did decline. Finally, phloem-feeding and whole-cell-feeding insects responded positively to elevated CO2, with increases in population size and decreases in development time. Of course, there will be exceptions to these general patterns, but they provide a benchmark from which to develop hypotheses for future studies. To proceed further, we require more comparative studies of a large variety of plant–insect combinations in diverse ecosystems under field conditions (Saxe et al., 1998). There remain fundamental gaps in our understanding of plant and insect responses to elevated concentrations of atmospheric CO2. For example, we know very little about the responses of below-ground herbivores to increases in atmospheric CO2. Although the number of studies addressing CO2-mediated changes in the dynamics of soil communities is increasing (Ball & Drake, 1997; Jones et al., 1998; Kampichler et al., 1998; Kandeler et al., 1998), effects on insects that feed on plant roots are poorly understood. In at least some cases, elevated CO2 results in increases in root growth (Day et al., 1996) and increases in root nodule production of nitrogen-fixing plants (Hungate et al., 1999). However, the quality of these resources for insects that feed below ground are generally unknown. Likewise, we know almost nothing about the effects of CO2-mediated changes in plant quality on aquatic insects. Questions of water temperature aside, many aquatic insects in heterotrophic streams depend upon litter inputs as sources of carbon to drive food web dynamics. Changes in litter quality as the result of elevated CO2 have the potential to influence decomposition processes and resource availability in streams. However, in the one study I could find to date, litter from oak and birch grown under elevated CO2 had equivalent effects on mosquito growth and reproduction, as did litter from ambient atmospheric conditions (Strand et al., 1999). We desperately need more information on CO2-mediated changes in litter quality and subsequent effects on stream food webs. In reference to forest trees, we do not really know whether the studies of seedlings and saplings that have dominated the literature reflect accurately the responses that we should expect from mature trees. The increasing availability of open-top chamber (OTC) and free-air carbon enrichment (FACE) technologies should provide us with the ability to conduct long-term experiments on mature plants and the insects that they support. Field studies that challenge insects with quasi-natural communities of plants, animals and microbes under fluctuating environmental conditions may be more likely to reveal future effects of elevated CO2 on insect distribution and abundance (Fajer & Johnson, 1993; Stiling et al., 1999). I have already mentioned how little we know about the responses of natural enemies to CO2 enrichment. OTC and FACE studies have considerable potential to address the kinds of complex multitrophic interactions that are pervasive in both natural and production ecosystems. Effects of natural enemies (Stiling et al., 1999), mutualists (Marks & Lincoln, 1996) and complex plant communities (Arnone et al., 1995) on insect behaviour and performance cannot be understood in the ecological vacuum of the laboratory. Perhaps most important of all, we need long-term studies that integrate multiple generations of both insects and plants. Such studies are prerequisites for assessing responses such as parental effects on offspring quality (Rossiter, 1994) and adaptation by herbivore species to a CO2-enriched atmosphere. Long-term studies under field conditions are, of course, expensive. But they represent a fraction of the cost that big business is willing to pay to elect the politicians who will close their eyes to unacceptable levels of CO2 emissions. I would like to thank the National Institute of Global Environmental Change and the Department of Energy for supporting our work on the ecological consequences of elevated concentrations of atmospheric CO2.