Effects of nitrogen fertilization on tritrophic interactions
- First Online:
- Cite this article as:
- Chen, Y., Olson, D.M. & Ruberson, J.R. Arthropod-Plant Interactions (2010) 4: 81. doi:10.1007/s11829-010-9092-5
Tritrophic interactions (plant—herbivore—natural enemy) are basic components of nearly all ecosystems, and are often heavily shaped by bottom-up forces. Numerous factors influence plants’ growth, defense, reproduction, and survival. One critical factor in plant life histories and subsequent trophic levels is nitrogen (N). Because of its importance to plant productivity, N is one of the most frequently used anthropogenic fertilizers in agricultural production and can exert a variety of bottom-up effects and potentially significantly alter tritrophic interactions through various mechanisms. In this paper, the potential effects of N on tritrophic interactions are reviewed. First, in plant-herbivore interactions, N availability can alter quality of the plant (from the herbivore’s nutritional perspective) as food by various means. Second, nitrogen effects can extend directly to natural enemies through herbivores by changes in herbivore quality vis-à-vis the natural enemy, and may even provide herbivores with a defense against natural enemies. Nitrogen also may affect the plant’s indirect defenses, namely the efficacy of natural enemies that kill herbivores attacking the plant. The effects may be expressed via (1) quantitatively and/or qualitatively changing herbivore-induced plant volatiles or other plant features that are crucial for foraging and attack success of natural enemies, (2) modifying plant architecture that might affect natural enemy function, and (3) altering the quality of plant-associated food and shelter for natural enemies. These effects, and their interactive top–down and bottom-up influences, have received limited attention to date, but are of growing significance with the need for expanding global food production (with accompanying use of fertilizer amendments), the widening risks of fertilizer pollution, and the continued increase in atmospheric CO2.
Studies on the impacts of nitrogen on tritrophic interactions have basic and applied implications, and yield information to help: (1) enhance the understanding of bottom-up forces in shaping tritrophic interactions; (2) maximize bottom-up effects on management of insect pests by increasing compatibility of plant resistance and natural enemies; and (3) indirectly predict the impact of elevated atmospheric CO2 concentration on tritrophic interactions. Relative to the third point, it is evident that global CO2 levels are rising. The global concentration of atmospheric CO2 has increased to a present level of 386 ppm (National Oceanic & Atmospheric Administration (NOAA) 2009) from 270 to 280 ppm at the beginning of the industrial revolution (Houghton et al. 1996). Although the accurate prediction of future atmospheric CO2 concentrations is difficult and the predictions vary greatly, most analyses anticipate levels will rise to over 700 ppm (Sundquist 1993). Short-term elevation of atmospheric CO2 increases the photosynthetic rates of C3 plants (Lee et al. 2001 and references therein) and carbon-based secondary compounds (Koricheva et al. 1998), but also affects the ability of plants to acquire nitrogen. Plants grown under enriched CO2 typically have a lower percentage of total nitrogen in their dry mass, and higher carbon (C) to N ratios (Rogers et al. 1996; Lawler et al. 1997). Therefore, bottom-up effects of nitrogen may become increasingly important as atmospheric CO2 rises. However, how plants will respond to increases in CO2 and N availability over the long term is not clear. For example, Lee et al. (2001) found that 13 perennial species representing 4 functional groups (C3 grasses, C4 grasses, legumes and non-leguminous forbs) showed pronounced photosynthetic acclimation over 2 years resulting in minimal stimulation of photosynthesis, and this did not depend on the level of nitrogen supplied. Other studies involving various species found neutral or greater photosynthetic responses at higher N under elevated CO2, and still other species have greater photosynthetic responses at lower N (reviewed in Lee et al. 2001). Thus, predicting plant species community responses to elevated CO2 and N over the long term will require an understanding of the extent that species acclimate photosynthesis and how N availability affects this response under prolonged elevated CO2 conditions.
This review considers the various direct and indirect effects on tritrophic interactions of altering N available to plants in hopes of generating greater interest in this important area. Anthropogenic N is becoming increasingly abundant in managed and natural systems, and it has the potential to significantly modify ecosystem structure and function. Thus, understanding these interactions has important implications for agriculture and conservation biology.
N alters suitability of plants as herbivore hosts
Nutritional quality of a food plant
The nutritional quality of plant tissue varies with spatial location within the plant, plant developmental stage (ontogenetic variation), species and genotype (between plant variation), and external factors associated with geographic location and season. Within an individual plant the N levels can vary from 0.03 to 7.0% of dry weight, with higher N content in young and expanding plant parts or reproductive structures (e.g., seeds; Mattson 1980). For example, in the early developmental stages, cotton leaf tissue contains ca. 4% N, but it decreases to less than 3% shortly after flowering (Bassett et al. 1970). N also varies between plant species. Many plants have evolved under N-limited conditions, but may differ in their growth strategy depending on their habitats. For example, Grime (1977) categorized plants that typically have rapid growth, higher non-sequestered N content, and occur in habitats with higher resource availability as C- and R-selected strategist, in comparison to S-selected plants that experience slower growth and were characterized as occurring in habitats with more limited resources. Grime (1977) suggested that plants that exhibit more vigorous growth and are not so readily limited by resources would be generally more palatable for herbivores than S-selected plants. C-selected plants minimize herbivory through selective spatial and temporal herbivore resistance factors. R-selected plants compensate for herbivory through rapid completion of their life cycle and maximization of seed production. S-selected plants, on the contrary, are predicted to be less palatable overall, presumably in response to slower growth rates and intensive natural selection for resistance to herbivores. These strategies can also differentially affect the responses of herbivores.
Phytophagous insects that feed on diets or host plants of lower nutritional quality typically exhibit lower growth rates, lower efficiency of conversion of ingested food, and lower fecundity (Dixon 1970; Mattson 1980; Weibull 1987; Karowe and Martin 1989; Lindroth et al. 1995; Awmack and Leather 2002; Chen et al. 2004), although the degree of response to N variation can be dependent on herbivore feeding guild and specific herbivore-plant interactions. For example, addition of N to white sagebrush, Artemisia ludoyiciana Nutt. (Asteraceae), increases performance of seed- and phloem-feeding insects but not leaf chewing insects (Strauss 1987). The abundance of the leaf feeding cereal aphid Metopolophium dirhodum (Walker) (Hemiptera: Aphididae) is greater on fertilized wheat Triticum aestivum L. (Gramineae) and barley Hordeum vulgare L. (Gramineae) compared to unfertilized wheat and barley, whereas the performance of the ear-feeding grain aphid Sitobion avenae F. is unaffected by fertilization (Honek 1991).
Given choices, many insect herbivores can distinguish host plants of high nutritional quality from those of low quality. Females of two Pieris butterflies, Pieris rapae crucivora and P. canidia canidia (Lepidoptera: Pieridae) (Chen et al. 2004) and buckeye butterfly, Junonia coenia Hübner (Lepidoptera: Nymphalidae) (Prudic et al. 2005) prefer fertilized over unfertilized host plants for oviposition. Similarly, Chen et al. (2008a) found that female Spodoptera exigua (Hübner) preferentially oviposited on cotton plants receiving higher levels of N. In a small-scale field study, the diamondback moth, Plutella xylostella (L.) (Lepidoptera: Plutellidae), was more abundant in high N plots than in low N plots (Fox et al. 1990).
To compensate for low N availability in N-stressed plants, insects tend to adjust their total food consumption by increasing consumption rates, prolonging feeding periods, or a combination of the two, or by adjusting their nutrient processing efficiency, for example through changes in food residence time or digestive enzyme levels (Mattson 1980; Barbehenn et al. 2004). Paper birch, Betula papyrifera (Betulaceae), grown under elevated CO2 environments had decreased N content (Lindroth et al. 1995). Fourth-instar saturniid caterpillars, Hyalophora cecropia L., Actias luna L., and Antheraea polyphemus Cramer (Lepidoptera: Saturniidae), grown on these birch plants consumed more plant material than on those grown under ambient atmospheric CO2 (Lindroth et al. 1995). Insect herbivores typically need to reach a certain size before molting to the next stage of development, and N availability will influence this process. Furthermore, the nutritional indices, such as approximate digestibility index (AD), efficiency of conversion of ingested food (ECI) and/or efficiency of conversion of assimilated or digested food (ECD) of insects feeding on low N food are typically decreased (Mattson 1980; Chen et al. 2004; but see Barbehenn et al. 2004). This means more low-N food is needed by many insects to complete their development, which may be exacerbated by the general increase in induction of anti-feedants and toxins in lower N plants. Population densities will likely decrease for these herbivore species, especially in temperate regions where delayed growth rates in nutritionally challenged herbivores may hamper escape in time from environmental exigencies; for example, they may not reach their critical stage for winter diapause. Delays in attainment of reproductive age also slow population growth.
Both increased consumption rates and prolonged feeding periods may also increase exposure and/or susceptibility of herbivores to potential predators, parasitoids, and pathogens, and result in greater herbivore mortality, as predicted by the slow-growth-high-mortality (SG-HM) hypothesis. The SG-HM hypothesis states that slower developing herbivores would be expected to suffer higher mortality from enemies (Feeny 1976; Augner 1995; Häggström and Larsson 1995; Benrey and Denno 1997; Fordyce and Shapiro 2003), although the validity of the hypothesis might depend on the system of study and the underlying assumptions (Clancy and Price 1987; Williams 1999). Outcomes also may vary with the life history of herbivorous insects and their natural enemies, and the extent to which plant characteristics that impair herbivore growth also interfere with the foraging efficiency of, or host/prey suitability for the natural enemies (Benrey and Denno 1997), because plant traits that confer resistance to herbivores are not always compatible with the functioning of natural enemies of the herbivores (Cortesero et al. 2000; see Hare 2002 for a review). For example, adverse effects of plant morphological traits, such as glandular trichomes and trichome density, on parasitic insect foraging have been noted in tobacco (Nicotiana tabacum L.) (Rabb and Bradley 1968; Kantanyukul and Thurston 1973; Elsey and Chaplin 1978), cotton (Gossypium hirsutum L.) (Treacy et al. 1986), wild potato (Solanum berthaultii Hawkes) (Obrycki and Tauber 1984; Obrycki 1986), alfalfa (Medicago sativa L.) (Lovinger et al. 2000), and soybean (Glycine max L.) (McAuslane et al. 1995). However, in relation to plant nitrogen levels, reduced food quality and resulting developmental delays and impaired vigor of herbivores would be generally expected to lead to greater herbivore mortality and dampening of population growth. Williams (1999) pointed out that the SG-HM hypothesis appears to apply consistently to generalist natural enemies, which are less likely to be adversely affected by suboptimal prey/hosts because of their capacity to switch food types.
Direct resistance traits of food plants
Instead of being helpless, plants have innate capacities for resistance to herbivores, with traits that can be broadly grouped into direct and indirect resistance. Indirect resistance includes any plant traits that increase fitness through interactions with organisms other than herbivores, for example, attracting entomophagous enemies of herbivores. In contrast, direct resistance refers to morphological (e.g., glandular trichomes) and chemical traits (e.g., terpenes) that directly exert negative effects on herbivores. Direct resistance can be further divided into constitutive and induced. Besides maintaining diverse constitutive morphological structures and plant secondary metabolites independent of herbivory, plants can be induced to manufacture a larger array of defensive compounds and structures in response to herbivory. These nitrogen-containing (e.g., alkaloids, non-protein amino acids) and non-nitrogen-containing (e.g., flavonoids, phenolics, tannins, and terpenes) plant secondary metabolites had previously been considered waste products because they were thought to have no clear functions in plant survival (Seigler and Price 1976). However, more evidence is emerging of diverse ecological, physiological and biochemical roles of these chemicals (Seigler and Price 1976; Bennett and Wallsgrove 1994; Constabel and Ryan 1996; Zangerl and Rutledge 1996; Simmonds 2003; Wink 2003; Zagrobelny et al. 2004), although, the distribution of these compounds among plants appears highly idiosyncratic (Berenbaum and Zangerl 2008), and there is as yet no unifying theory to explain how and why plants produce, transport, and store such a diverse array of chemicals (see Firn and Jones 2000; Dudareva et al. 2004; Peñuelas and Llusià 2004; Owen and Peñuelas 2005, 2006a, b; Firn and Jones 2006a, b; Pichersky et al. 2006 for discussion). Furthermore, there are several well-studied examples of herbivores that have developed detoxification mechanisms, and these mechanisms are highly idiosyncratic in distribution among herbivore taxa even for those feeding on the same plant (Berenbaum and Zangerl 2008). Berenbaum and Zangerl (2008) suggest that using genomic tools that have been developed in studies of the relatively few plant families used as models over the last several decades may clarify our understanding of the ecologically idiosyncratic nature of production and detoxification of plant defense compounds. Given the breadth of secondary compounds, the range of possible functions, and the inconsistent pattern of responses to secondary compounds by herbivores, predictions of ecosystem- and community-level outcomes for N changes are difficult.
Expression of constitutive and induced allelochemicals in a wide range of plant species is significantly influenced by soil nutrient availability (Dudt and Shure 1994; Koricheva et al. 1998; Stout et al. 1998; Darrow and Bowers 1999; Cipollini and Bergelson 2001; Coviella et al. 2002; Hol et al. 2003; Orians et al. 2003; Wall et al. 2005), although the magnitude of their expression may increase, remain neutral, or decline depending on the study systems. For example, total concentration of the carbon-based iridoid glycoside from Plantago lanceolata L. (Plantaginaceae) was decreased by fertilization (Darrow and Bowers 1999; Prudic et al. 2005). Nitrogen addition also lowered constitutive phenolics in tomato plants, Lycopersicon esculentum Mill. (Solanaceae) (Stout et al. 1998), polyphenols in Solanum carolinense L. (Solanaceae) (Wall et al. 2005), and condensed tannins in quaking aspen Populus tremuloides (Salicaceae) (Hemming and Lindroth 1999). However, fertilization had no effect on the phenolics of tulip poplar, Liriodendron tulipifera L., and dogwood, Cornus florida L. (Dudt and Shure 1994). Proteinaceous trypsin inhibitor concentrations in Brassica napus L. (Brassicaceae) seedlings (Cipollini and Bergelson 2001) and in tobacco Nicotiana attenuata Torr. ex S. Watson (Solanaceae) (Lou and Baldwin 2004), and nicotine content in tobacco (Lou and Baldwin 2004) were enhanced by nutrient fertilization. Proteinase inhibitor levels of tomato (L. esculentum) plants grown under low, medium, and high N conditions remained at the same levels, although leaflet total protein concentrations increased as N availability went from low to high (Stout et al. 1998). In stinking willie, Senecio jacobaea L. (Asteraceae), the total amount of the N-based defensive compound pyrrolizidine alkaloid was not affected by addition of nutrients, although concentrations were decreased because of higher plant mass due to more rapid growth (Hol et al. 2003). These authors suggest that there is no need for additional defense as long as plant growth is faster than biomass removal by herbivory. Plants may, therefore, invest in more rapid growth when this strategy allows them to escape herbivory in time. A growth-escape strategy would be expected to have little or no adverse effect on herbivores, but may be beneficial to individual and population growth of herbivores.
Besides the effects on C- and N-based constitutive chemicals discussed above, N may also affect plants’ induced defense at the time of herbivory. For example, N fertilization increased the degree of induced resistance in poplar (Populus nigra L.) after continuous feeding of gypsy moth (Lymantria dispar L.) for 72 h (Glynn et al. 2003). Similarly, the magnitude of induced trypsin inhibitor in the high nutrient treatment was greater than in the low nutrient treatment in Brassica napus L. following mechanical damage (Cipollini and Bergelson 2001). However, there appear to be upper thresholds of N quantity above which induced responses of plants to herbivory are reduced. For example, Olson et al. (2009) found that cotton (G. hirsutum) plants that were grown with twice the recommended N levels and those plants grown with no nitrogen had increased feeding damage on leaf tissue by Spodoptera exigua when compared to plants grown with recommended levels of nitrogen, presumably because of reduced induction of terpenoid aldehydes (Olson et al. 2009). Conversely, Chen et al. (2008b) found that terpenoid aldehyde induction was increased in low-N (42 ppm) cotton plants experiencing herbivory by S. exigua relative to plants receiving more N. Therefore, there is likely a range of nitrogen concentrations that is optimal for production of N-responsive defensive secondary compounds. Above this range the plant may be at greater risk for herbivory, but in a nutrient-rich environment the plant may be able to outgrow herbivory with minimal investment in chemical defense.
N alters herbivore suitability as prey/host of natural enemies
Nutritional quality of prey/host
The development time of immature parasitoids is typically positively related to host size, although the relationship can be neutral and negative in some cases (e.g., King 1987; Sequeira and Mackauer 1992). The dependency of development time upon host size differs between idiobiont (parasitoids that terminate host development at the time of oviposition) and koinobiont parasitoids (parasitoids that allow hosts to continue developing after oviposition) (Salt 1941; Vinson and Iwantsch 1980; Kouamé and Mackauer 1991; Godfray 1994). Size of herbivore hosts is, in turn, closely and directly related to the nutritional quality of their host plants, and herbivore size can be used as a proxy for plant quality.
Adult parasitoids also may be affected by differences in host quality. Host-feeding parasitoid adults are restricted to the insect order Hymenoptera and 140 species from 17 families were noted to have this behavior (Jervis and Kidd 1986). The fecundity of host-feeding parasitoids is affected by the hosts on which they feed (Jervis and Kidd 1986; Thompson 1999). For example, fecundity of hymenopteran parasitoids, such as Bracon hebetor Say (Braconidae), Aphytis lingnanensis Compere (Aphelinidae), and Pimpla turionellae (L.) (Ichneumonidae) was greatest when supplied with hemolymph of their hosts, compared to those starved or provided only water (Edwards 1954; Debach and White 1960; Benson 1973; Lum 1977), because they obtain amino-nitrogen for egg development (Jervis and Kidd 1986). Therefore, qualitative changes in herbivores due to plant N may directly affect parasitoid reproduction.
The fitness of predators can also be affected by their diets (Jervis and Kidd 1986; Li and Jackson 1997; Thompson 1999; Mayntz and Toft 2001). For example, when jumping spiders, Portia fimbriata (Araneae: Salticidae), were provided with prey composed of intraguild spiders, they had greater survival, in comparison to those supplied with N-poor phytophagous insects (Li and Jackson 1997). Compared to predaceous stink bugs (Podisus maculiventris) reared on caterpillars fed on diets made of mature-leaf powder, their conspecifics reared on caterpillars fed on diets made of new-leaf powder grew faster (Strohmeyer et al. 1998). The higher growth rate of P. maculiventris when feeding on caterpillars reared on a young leaf diet was attributed to higher nutrients in the caterpillars, even in the presence of higher amounts of iridoid glycosides, which are known feeding deterrents to generalist herbivores (Strohmeyer et al. 1998). Thus, the fitness of predators may depend on the quality of the predator’s prey, which in turn may depend on the quality of the prey’s host.
It is likely that entomopathogens are also affected by N changes in plants. As noted above, reduced plant quality often leads to increased consumption by herbivores. Some entomopathogens (bacteria, fungi, and viruses) can persist on the phylloplane as infective units (e.g., spores), and some of these (bacteria and viruses) must be ingested to infect the host. Thus, increased consumption by herbivores may increase the probability of consuming infective propagules (Cory and Hoover 2006). Increased movement also may expose the herbivore to more fungal spores, thereby increasing the risk of infection. Host herbivores also may be weakened as a result of inappropriate plant N levels, leading to reduced resistance to infection. Indeed, Lee et al. (2006) observed that dietary protein levels were highly influential in determining success of nuclear polyhedrosis virus infection in the caterpillar Spodoptera littoralis (Boisduval), and infected caterpillars actively modified their N intake to address the infection. Changes in plant architecture resulting from N availability also may indirectly affect entomopathogen survival and infection success by altering the microhabitat (most notably humidity and UV irradiation).
Herbivore defense against natural enemies
Lower nutritional quality of host plants may lower an herbivore’s encapsulation ability. The herbivore’s chances of encapsulating invading parasites or pathogens is generally correlated with the herbivore’s developmental stage (instar), physiological condition, and capacity for defensive behavior (Salt 1968; Smith and Smilowitz 1976; Blumberg and Debach 1981; van Driesche and Bellows 1988), which, in turn, may be influenced by N availability (Chen et al. 2008a).
Many plant allelochemicals that function as defensive compounds are sequestered by various herbivorous insects in the hemolymph. The predators and host-feeding parasitoids that feed on those insects, and larval offspring of parasitic wasps that live part of their life time inside such insects will in many cases suffer in terms of developmental time and survivorship (Campbell and Duffey 1979; Duffey et al. 1986; van Emden 1995; Kester and Barbosa 1991; for a review, see Turlings and Benrey 1998; but see Schuler et al. 1999). The adverse effect of the antibiotic tobacco compound nicotine absorbed in tobacco hornworm, M. sexta, hemolymph on parasitism and survival of the gregarious parasitoid Cotesia congregata (Say) is a good example (Morgan 1910; Gilmore 1938; Thurston and Fox 1972). Manduca sexta is a specialist herbivore in tobacco and can process nicotine effectively mostly through excretion. However, some amount of nicotine is sequestered in the M. sexta hemolymph without any ill-effect to the herbivore (Self et al. 1964). The parasitic wasp C. congregata, on the other hand, is more sensitive to nicotine, which reduces their survival (Parr and Thurston 1972; Thorpe and Barbosa 1986; Barbosa et al. 1991).
The effects of N on herbivore defense, and natural enemies may vary with the plant species or the type of allelochemical produced. For example, Lou and Baldwin (2004) noted that N addition increased tobacco nicotine production, however, Baldwin (1999) found that M. sexta is resistant to nicotine. In separate studies, Thorpe and Barbosa (1986), and Parr and Thurston (1972) found lower survival of C. congregata on M. sexta larvae that had fed on tobacco plants with nicotine and artificial diets containing nicotine compared to cotton plants and artificial diets without nicotine. Therefore, addition of N to tobacco plants may adversely affect the performance of C. congregata. In contrast, as shown previously, the quantities of many constitutive defensive plant secondary metabolites are negatively related to N levels. Consequently, in such cases predators and parasitoids that consume herbivores that are grown on host plants of higher N levels may perform better.
N affects plant indirect resistance/defense incurred through natural enemies
The attraction of entomophagous natural enemies by plants is referred to as plant indirect defense. Because the relationship can appear mutualistic, these natural antagonists of herbivores are sometimes called ‘plant bodyguards’ (Dicke and Sabelis 1988; Whitman 1994; Cortesero et al. 2000). Herbivore-induced volatile organic compounds (VOCs) that natural enemies rely on when foraging, as well as food and shelter of natural enemies, may be altered by plant N status. In the study by Olson et al. (2009), cotton plants grown in zero nitrogen that were induced by feeding of Spodoptera exigua (Hübner) were significantly less attractive to Microplits croceipes (Hymenoptera: Braconidae) in flight choice bioassays than damaged plants grown in recommended N levels. Interestingly, parasitoids were also more favorably responsive to damaged plants receiving recommended N rates than to plants grown in twice the recommended N similarly damaged by S. exigua. This indicates the potential for decreased attraction of this and likely other natural enemy species in cotton fields that have too little or too much N. Therefore, fluctuations in N due to resource availability or acquisition capacity may contribute to significant changes in population dynamics of herbivores and their natural enemies.
N changes volatile release pattern (orienting cues)
Plants release a blend of volatile chemicals following wounding by herbivores. Some of these induced volatiles are released around the actual feeding site, while others are systemically released from plant tissue distal to and above the wounded site. Green leaf volatiles (GLVs) (e.g., (Z)-3-hexenal, (Z)-3-hexenol, and (Z)-3-hexenyl acetate), some acyclic monoterpenes, sesquiterpenes, homoterpenes, and indole are among the typical locally induced volatile organic compounds (VOCs) in cotton (Loughrin et al. 1994; McCall et al. 1994; Turlings et al. 1995; Paré and Tumlinson 1997, 1998). (Z)-3-hexenyl acetate, some acyclic monoterpenes, sesquiterpenes and homeoterpenes can be systemically induced (Loughrin et al. 1994; Röse et al. 1996; Paré and Tumlinson 1997, 1998). Many of these herbivore-induced plant-originated VOCs provide foraging natural enemies essential cues to locate potential hosts/prey. Both parasitoids and predators have been observed to respond actively to VOCs. For example, the parasitoids Cotesia marginiventris (Cresson) (Röse et al. 1998), Microplitis croceipes (Cresson) (Röse et al. 1998) and Cardiochiles nigriceps Viereck (De Moraes et al. 1998) fly more frequently to host-damaged plants. The predatory mite Phytoseiulus persimilis (Acari: Phytoseiidae) and two insect predators, Scolothrips takahashii (Thysanoptera: Thripidae) and Oligota kashmirica benefica (Coleoptera: Staphylinidae), were attracted to spider mite (Tetranychus urticae)-infested lima bean plants (Dicke et al. 1990; Shimoda et al. 2002; Choh et al. 2004). These VOCs also have repellent effects on ovipositing conspecific herbivores (De Moraes et al. 2001), which would appear to benefit both the plant by reducing herbivore load and the herbivore by reducing intra-specific competition.
Nitrogen levels can alter the production and release of these volatiles. Depending upon the plant, positive, negative and no effects have been observed. In corn (Zea mays var Delprim), the peak of volatile release was detected when N concentration in the nutrient solution was the lowest, both after mechanical wounding and addition of volicitin (an elicitor isolated from oral secretion of beet army worm, Spodoptera exigua (Hübner) (Schmelz et al. 2003). Low N availability also increased production of the main sesquiterpenes ((E)-α-bergamotene, β-caryophyllene and (E)-β-farnesene) to a greater extent after volicitin application, compared with mechanical damage. In addition, reduced N levels made the concentration of jasmonic acid (a chemical messenger thought to be crucial to the induction of volatiles) wane at a slower rate when compared to those levels in higher N level plants. Jang et al. (2008) found decreased levels of jasmonic acid in rice plants receiving higher rates on N fertilization in all three of the cultivars tested. Likewise, in a second system studied, celery with additional N had a lower quantity of volatile compounds (Van Wassenhove et al. 1990). Nevertheless, Gouinguené and Turlings (2002) found that unfertilized corn plants (Zea mays var Delprim) emanated less volatiles when compared with those that had received a complete nutrient solution. The role of N was not implied in this study as all the nutrients were varied (Schmelz et al. 2003). In tobacco (Nicotiana attenuata), oral secretion from tobacco hornworm Manduca sexta (L.) and methyl jasmonate (MeJA) induced volatile release was not affected by N, though low N availability attenuated the jasmonate and salicylate levels and reduced two N-containing anti-herbivore defense compounds, nicotine and trypsin proteinase inhibitors (Lou and Baldwin 2004). Chen et al. (2008b) found that cotton plants with the lowest N had substantially higher induced VOC’s than those plants with higher N. GC–MS analyses indicated that nitrogen affected the amount and/or rate of volatiles released, not the induction per se, in cotton plants grown with no nitrogen and those grown with twice the recommended nitrogen, compared with those grown in recommended nitrogen (Olson et al. 2009). No other studies on VOC release patterns are available to date. However, the studies to date suggest that the effects of N on the release pattern of VOCs might be system- or species-specific. Plants generally increase VOC emission under stress from low nitrogen, unless the plant has evolved more plastic responses to herbivory and defenses, such as found in tobacco plants; these plants have VOC production that is independent of nutrient availability, and have a major herbivore, Manduca sexta, which has developed resistance to nicotine (Baldwin 1999). Placing more emphasis on VOC emission than secondary compound production would be advantageous to the plant when herbivores are less affected by the secondary metabolites, or if the fitness costs of nicotine production are too high as a result of N limitation (Baldwin 1999). This underscores the need to understand how plants and their herbivores have co-evolved (e.g., Berenbaum and Zangerl 2008). It would be of interest to determine if plants in the families Brassicaceae and Apiaceae that have herbivores that have evolved detoxification methods (Berenbaum 2001; Li et al. 2007), also exhibit VOC production that is independent of nutrient availability.
Plants as food and shelter of natural enemies
Many insect predators feed on pollen as supplemental food, whereas the prevalence of pollen-feeding in parasitoids seems to be rather uncommon (reviews in Wäckers 2005). Pollen is primarily a source of nitrogenous compounds (proteins and amino acids), but also contains starch, lipids and some sterols. It is likely that with increasing N this food source will increase in value and/or abundance to those species that feed upon it.
Both predators and parasitoids feed on floral and extra-floral nectar and various fitness correlates of many natural enemies such as longevity, movement and fecundity are increased by feeding on these plant foods (Hagley and Barber 1992; Wäckers and Swaans 1993; Olson and Nechols 1995; Morales-Ramos et al. 1996; Baggen and Gurr 1998; Eijs et al. 1998; Jervis and Kidd 1999; Irvin and Hoddle 2007). Male and female parasitoids Diachasmimorpha longicaudata (Ashmead) (Hymenoptera: Braconidae), of the tephritid fruit fly for example, lived up to 15 and 28 days, respectively, when cotton extrafloral nectaries were available (Sivinski et al. 2006). Conversely, with provision of only water male and female parasitoids lived a maximum of 7 days. Trissolcus basalis (Wollaston) (Hymenoptera: Scelionidae), an important egg parasitoid of southern green stink bug (Nezara viridula (L.)) (Hemiptera: Pentatomidae), lives longer when floral nectars are available (Rahat et al. 2005). Provision of food sources can attract more natural enemies and increase the mortality of herbivorous insects. For instance, parasitism of the gypsy moth, Lymantria dispar L. (Lepidoptera: Lymantriidae), was higher on plants with extrafloral nectaries, although the parasitoid species richness between nectaried and nectariless plants was not different (Pemberton and Lee 1996). More bollworm, H. zea, eggs were parasitized by Trichogramma pretiosum Riley (Hymenoptera: Trichogrammatidae) in cotton plants with extrafloral nectaries than those without nectar (Treacy et al. 1987).
Overall N effects on nectar production appear to vary among plant species. Burkle and Irwin (2009) found that increasing N fertilization increased nectar production of Ipomopsis aggregata (Pursh) V.E. Grant (Polemoniaceae), but had no effect on nectar production by Linum lewisii Pursh (Linaceae). Similarly, Ryle (1954a, 1954b) reported that increasing N (as nitrates) fertilization led to decreased nectar production in apple trees and mustard plants, but enhanced nectar production in buckwheat and had no effect on clover. However, she also noted that there was an interaction between N and the other nutrients relative to nectar production, underscoring the need to consider multiple variables. Nitrogen changes may also affect the amino acid content of nectar, which may affect herbivore quality (Mevi-Schütz and Erhardt 2005) for natural enemies, or N availability may alter the abundance of nectar through influencing changes in numbers of available nectaries (e.g., through changes in flower abundance), possibly reducing or increasing competition for resources among natural enemies. Studies of reduced N in conjunction with increased CO2 levels also noted increased carbohydrates (Koricheva et al. 1998), and one study (Fischer et al. 1997) found that the concentration of sugars, but not their composition in floral nectar of Gentianella germanica increased by 36% under elevated CO2. Thus, over the longer term as CO2 levels rise, plant carbohydrate availability may increase and ameliorate any possible short-term negative effects of reduced N accessibility.
Plants not only provide natural enemies with food, but also shelter. Many plant structures such as leaf domatia and leaf veins can provide shelter and overwintering sites to various natural enemies (Karban et al. 1995; Walter 1996; Hance and Boivin 1993; Whitman 1994; Corbett and Rosenheim 1996; Elkassabany et al. 1996; Maschwitz et al. 1996; Agrawal and Karban 1997). Despite the importance of these plant structures, there are no studies of the effects of N availability on their growth and development and ultimately, their ability to provide shelter to natural enemies. However, at the plant community scale, increased nitrogen availability substantially increases the density and changes the composition of plant communities (e.g., Manning et al. 2009). This is likely to extend over the long-term as atmospheric N levels continue to rise in response to increased N from fertilization and combustion of fossil fuels (Vitousek et al. 1997). Plant communities with increased N have increased net primary productivity and decreased plant biodiversity (Gough et al. 2000; Suding et al. 2005). Increases in plant biomass and reduced plant diversity may profoundly affect plant species’ ability to acquire nutrients or to evade attack. For example, increases in plant density (biomass) means more plant surface area and higher edge to surface area ratios for natural enemies to forage and this could have negative effects on predators and parasitoids in locating their hosts and prey (e.g., reviews in Olson and Andow 2006 for parasitoids and Rutledge and O’Neil 2005 for predators). Beckerman et al. (1997) found that the generalist leaf-chewing grasshopper Melanoplus femurrubrum shifted habitat from grasses to more complex herbaceous species as predation risks from the hunting spider, Pisaurina mira increased. Reductions in plant community biodiversity could also remove needed refuge habitats for herbivore species, making them more vulnerable to predation. Therefore, increased plant density and biomass, along with the previously discussed reduction in VOCs in higher N plants is expected to negatively affect the foraging efficacy of natural enemies. Decreased plant diversity in higher N plant communities is expected to have a negative effect on herbivores that are unable to switch habitats while feeding on a particular plant species, but this would depend on how complex the structure of the vegetation becomes in the original habitat by increased N. In the latter case, N increases with concomitant increases in complexity would likely decrease predation rates overall, and increase herbivory on the plants.
Predation/parasitism rates changed by N
Natural enemies (predators, parasitoids, and pathogens) of herbivores employ chemical, visual, and vibrational cues (both from hosts/prey and food plants of hosts/prey) to search for and/or attack potential preys/hosts. Chemical cues (also called semiochemicals) are, in most cases, the most important cues used by predators and parasitoids to locate hosts/prey (Mattiacci et al. 2001; Wäckers and Lewis 1994; Röse et al. 1998; Olson et al. 2000; Dicke et al. 1990; Shimoda et al. 2002; Choh et al. 2004). Nitrogen has been shown to affect chemical attractiveness of plants for foraging enemies of herbivores.
In small-plot studies, parasitoids were more attracted to plots with higher N plants and exerted greater control on herbivores on such plants. Encarsia formosa Gahan (Hymenoptera: Aphelinidae), a parasitoid of the whitefly Bemisia argentifolii Bellows and Perring (Hemiptera: Aleyrodidae), was more frequently observed on N-fertilized and whitefly-infested poinsettia plants, Euphorbia pulcherrima Willd. ex Klotzsch, than on whitefly-infested but unfertilized plants in choice-tests (Bentz et al. 1996). Significantly more whiteflies were parasitized per leaf in the higher N treatment than in the lower N treatment. The mean counts of whitefly per leaf (sum of parasitized, fed upon and unparasitized) were about the same across the treatments (see Table 1 of Bentz et al. 1996), so the possibility that the parasitoids were responding to greater sucking damage or host density alone could be excluded. In a study of the impact of collard plant (B. oleracea) quality on parasitism rate and sex ratio of the diamondback moth parasitoid Diadegma insulare (Cresson) (Hymenoptera: Ichneumonidae), Fox et al. (1990) found more parasitoids in the well-fertilized treatment and parasitism rates were lowest under regimes without application of fertilizer, although foliar N level and protein concentration were marginally positively correlated with parasitism rate. Additionally, parasitoids that had emerged from high N treatments were more female-biased. Loader and Damman (1991) also found that parasitism rates were higher on cabbage butterfly, Pieris rapae (L.) (Lepidoptera: Pieridae), developing on collards with higher N. All of these studies were carried out with potted plants where differences in plant density in the plot stands were controlled. Small parasitoids that are weaker fliers and more wind-borne may not rely on chemical and visual cues to the extent that stronger fliers can. With no apparent increased plant surface area or increased edge to surface area ratios in these plots, the parasitoids may have located hosts equally well on high and low N plants, even though they had higher success developing in herbivores that had fed on higher N plants. Larger parasitoids, with the exception of D. insulare, responded as predicted; higher parasitism was found on low N plants. Diadegma insulare are known to have variable sex ratios, but also produce more females on high N plants and within larger hosts (Fox et al. 1990) which may explain its higher success on high N plants. Therefore, parasitioids and predators may increase their overall fitness developing in higher N hosts due to increased suitability. However, their ability to locate hosts will likely be reduced in higher N plant communities because of reduced phytochemical cues and greater foraging areas in foliage.
The cues that natural enemies respond to in the studies discussed remains unknown. The crop systems utilized to investigate VOC release patterns differ from those selected to examine natural enemy effects. Based on limited information available at this point, it is hard to draw conclusions on whether or not the observed parasitism/predation patterns are consistent with variable rates of VOC release. Other orienting cues such as visual cues may also play a role in some of the cases because plants with low and high N availability not only often differ in height, but also in color, architecture, density, and community composition. Plant morphological traits also interact with foraging efficiency of natural enemies, and mutualistic, antagonistic, and neutral relationships between plant trichomes, and other structural features, and natural enemies of herbivores have been documented (e.g., Elsey and Chaplin 1978; Price et al. 1980; Obrycki 1986; Treacy et al. 1986; Kauffman and Kennedy 1989; McAuslane et al. 1995; Sutterlin and van Lenteren 1997; Bottrell et al. 1998; Cortesero et al. 2000; Lovinger et al. 2000; Gassmann and Hare 2005; Simmons and Gurr 2005; Olson and Andow 2006; Styrsky et al. 2006).
N fertilization may exert profound bottom-up influences on ecosystems— interactively extending across trophic levels and influencing outcomes at the individual, population, and community levels. These influences, and their interactive top–down and bottom-up effects, have received limited attention to date, but are of growing significance with the need for expanding global food production (with accompanying use of fertilizer amendments), the widening risks of fertilizer pollution, and the continued increase in atmospheric CO2. The biomass loss of low N plants due to reduced growth and compensatory consumption of herbivores appears to be compensated for at least in part by increased direct plant defenses, and by greater indirect defenses through enhanced natural enemy recruitment and reduced foraging areas due to decreased plant size and complexity. On the other hand, high N availability to plants promotes plant biomass production and the increased biomass might be offset by increased herbivory resulting from greater recruitment of herbivores to a more nutritious plant, and reduced natural enemy recruitment because of reduced chemical cues. However, plants may respond by providing increased food (e.g., nectar and pollen) and shelter resources for natural enemies. Further, plant life history (e.g., perennial vs. annual) may alter the relative contributions of induced defensive and volatile compounds in response to herbivory and N availability such that the low N/high defense and volatile pattern observed in the few examples studied may not hold true. Ultimately, plants must balance N utilization against the action of herbivores and their natural enemies, as well as the metabolic requirements of constitutive and induced defenses, in their management of herbivory. Maintaining this balance will likely become more complicated with increasing environmental contamination by anthropogenic N and CO2.
We appreciate funding support from the Georgia Cotton Commission and Cotton Incorporated. We also appreciate the valuable comments of the anonymous reviewers and editor on the manuscript.