Aboveground Feeding by Soybean Aphid, Aphis glycines, Affects Soybean Cyst Nematode, Heterodera glycines, Reproduction Belowground

Heterodera glycines is a cyst nematode that causes significant lost soybean yield in the U.S. Recent studies observed the aphid Aphis glycines and H. glycines interacting via their shared host, soybean, Glycine max. A greenhouse experiment was conducted to discern the effect of A. glycines feeding on H. glycines reproduction. An H. glycines-susceptible cultivar, Kenwood 94, and a resistant cultivar, Dekalb 27–52, were grown in H. glycines-infested soil for 30 and 60 d. Ten days after planting, plants were infested with either zero, five, or ten aphids. At 30 and 60 d, the number of H. glycines females and cysts (dead females) and the number of eggs within were counted. In general, H. glycines were less abundant on the resistant than the susceptible cultivar, and H. glycines abundance increased from 30 to 60 d. At 30 d, 33% more H. glycines females and eggs were produced on the resistant cultivar in the ten-aphid treatment compared to the zero-aphid treatment. However, at 30 d the susceptible cultivar had 50% fewer H. glycines females and eggs when infested with ten aphids. At 60 d, numbers of H. glycines females and cysts and numbers of eggs on the resistant cultivar were unaffected by A. glycines feeding, while numbers of both were decreased by A. glycines on the susceptible cultivar. These results indicate that A. glycines feeding improves the quality of soybean as a host for H. glycines, but at higher herbivore population densities, this effect is offset by a decrease in resource quantity.


Introduction
Crop production is at risk for yield loss from both aboveground and belowground herbivores that can occur concurrently and interact through a shared host plant [1], [2]. Recent reviews of aboveground-belowground herbivore interactions have hypothesized that both plant nutrients and common plant defense pathways are important mediators of herbivore interactions [3][4][5]. General hypotheses are proposed including the importance of study location (i.e. field versus greenhouse) [1], feeding guild similarity [3], herbivore arrival time [1], and infestation intensity [3] in determining the outcome of the interaction for each herbivore.
Belowground plant-parasitic nematodes are important yieldreducing pathogens of all major field crops produced in the U.S. [6]. However, their effect on aboveground insects has been sparingly studied, and the reciprocal effect of aboveground insects on nematodes is even less well studied [1]. Johnson et al. [1] conducted a meta-analysis that included 123 observations that investigated the interaction between aboveground and belowground herbivores, of which only 11 observations included nematodes. Overall, plant-parasitic nematodes had no observable effect on the performance of aboveground insects, and the reciprocal effect (aboveground herbivores on nematodes) was not examined in the meta-analysis. To what extent the general pattern of above-and belowground herbivores predicts the interaction between nematodes and aboveground herbivores is not known.
Aboveground-belowground herbivore interactions are of particular importance for soybean, Glycine max (L.) Merrill, because the crop is challenged by a belowground herbivore, the soybean cyst nematode, Heterodera glycines Ichinhoe, and a diverse community of aboveground insect herbivores. Heterodera glycines is the leading yield-reducing pathogen of soybean both in the U.S. and worldwide [7], [8]. In the U.S., H. glycines is widely distributed throughout all major soybean-producing regions and causes an estimated yield loss of $1.8 billion each year [7]. The population density of H. glycines eggs in the soil at the beginning of the growing season is the strongest predictor of yield loss [9][10][11][12]. Population densities are managed by growing non-hosts (i.e. crop rotation) or H. glycines-resistant soybean cultivars. However, crop rotations often consist of two-year rotations, and H. glycines-resistant soybean cultivars are mostly derived from a single source of resistance, PI 88788 [13]. Therefore, populations of H. glycines persist within the agroecosystem, with infestations of H. glycines occuring in 47-83% of fields in the major soybean-producing region of the Midwestern U.S. [14].
Previous research observed that aboveground lepidopteran herbivores were capable of increasing H. glycines reproduction on soybean [15], [16]. These studies, however, were conducted using only H. glycines-susceptible cultivars and lepidopteran herbivores, which occur rarely as aboveground pests of soybean in the Midwestern U.S. [17]. In this region, the invasive soybean aphid, Aphis glycines Matsumura, colonizes fields during early vegetative stages of soybean development with population densities increasing through the reproductive stages of the crop, leading to yield losses of up to 40% in outbreak years [18]. The co-occurrence of H. glycines and A. glycines is an intriguing system to study as plantparasitic nematode and aphid infestations both result in changes in induced defense responses and primary plant metabolites [19].
Research exploring the co-occurrence of A. glycines and H. glycines on soybean suggests that an interaction may occur, but the results of these studies have been incomplete [11], [20][21][22]. Heterodera glycines infections are proposed to increase [22], decrease [20], and have no effect on A. glycines populations [11], [21]. Furthermore, H. glycines infections reduce A. glycines alate preference for soybean plants [11], [22]. In addition, A. glycines infestations are suggested to both increase [20] and have no effect [21] on H. glycines reproduction. Discrepancies in these reports may be due to differences in field versus laboratory settings of the experiments [11], [22], pest population densities, and the inclusion of other pest species in the experimental treatments [20], [21].
McCarville et al. [20] and Heeren et al. [21] used similar field micro-plots to investigate the effect of A. glycines feeding on H. glycines reproduction over the course of the entire season.
McCarville et al. [20] [20] did not include a treatment in which plants were exposed to only H. glycines and A. glycines, and, therefore, could not discern whether the increase in H. glycines reproduction was due solely to A. glycines feeding.
Heeren et al. [21] included treatments in which soybean plants were exposed to H. glycines alone or to both H. glycines and A. glycines, thus making direct observations on the interaction between A. glycines and H. glycines possible. They did not observe an effect of A. glycines feeding on H. glycines reproduction. However, their study utilized much lower pest population densities than McCarville et al. [20], and in the case of H. glycines, densities were often below the limit of detection. Given the discrepancies in pest treatments studied (A. glycines alone or in combination with C. gregata) and pest population densities utilized between McCarville et al. [20] and Heeren et al. [21], our goal was to determine whether A. glycines feeding by itself could affect H. glycines reproduction. In addition we explored whether the population densities of both A. glycines and H. glycines, which vary widely across the North Central U.S., affect the outcome of the interaction.

Materials and Methods
In a greenhouse, we manipulated the density of H. glycines populations through the use of resistant and susceptible soybean cultivars and A. glycines populations through the use of different initial infestation densities. In addition to H. glycines-resistant and susceptible cultivars and differential A. glycines infestation densities (both described below), pest densities examined were also manipulated by conducting the experiment for different lengths of time. Half of all plants were harvested at 30 d to measure treatment effects on a single generation of H. glycines reproduction, and the remaining plants were harvested at 60 d to measure treatment effects after two generations of H. glycines reproduction.
For these experiments, a modified version of the Standard Cyst Evaluation-2008 (SCE-08) protocol was utilized [23], in which 125-ml cone-tainers (Stuewe & Sons, Tangent, OR) were arranged in 7.5-l sealed plastic buckets filled with construction sand. The buckets were kept in a water bath to maintain a constant soil temperature between 26.7uC and 28.9uC, which allows for the completion of a single generation of H. glycines in approximately 25 d [24].
Cone-tainers were filled with 100 ml of a soil-sand mixture created by adding construction sand to H. glycines-infested Eolian Sand type soil. The H. glycines population was HG type 0, which is defined by having less than 10% reproduction on all published sources of H. glycines resistance (i.e. avirulent to all H. glycines resistance genes) [25] and was chosen for its limited ability to reproduce on the PI 88788-derived resistant cultivar utilized in our experiment. Eolian Sand type soil (a fine silt type soil with a high sand content) was used as it consistently permits high H. glycines reproduction in the field [26] and is easily washed from soybean roots permitting efficient collection of H. glycines females and cysts. The soil was diluted with construction sand to obtain a soil-sand mixture with an approximate population density of 1,000 eggs 100 ml 21 of soil. This population density was selected to reduce the likelihood of competition among H. glycines females for the nutritional resources of soybean plants. Plants were grown under natural lighting supplemented with 16:8 (L:D) 400 W highpressure sodium growth lamps and watered as needed.
Two soybean cultivars were used for the experiment, Kenwood 94 and Dekalb 27-52. Kenwood 94 is a H. glycines-susceptible cultivar and Dekalb 27-52 is a PI 88788-derived H. glycinesresistant cultivar that was used in the field experiment by McCarville et al. [20]. In addition to the two soybean cultivars, we used three initial aphid population densities in both the 30 d and 60 d time periods. Aphid treatments were defined by the initial population of A. glycines added to each plant (zero aphids, five aphids, and ten aphids). The treatment factors of soybean cultivar and aphid density were fully crossed to create six total treatment combinations per time period. These treatments were arranged in a split-plot design, with the whole plots arranged in a randomized complete block design. It was possible to prevent A. glycines from moving between buckets but not between cone-tainers within a bucket, so the whole plot was an individual 7.5-l bucket, with the treatment factor of aphid density assigned to the whole plot. Each bucket contained eight cone-tainers, four per soybean cultivar. Each of the four cone-tainers per soybean cultivar was randomly assigned to one of the two time periods (i.e. 30 d and 60 d). Data were analyzed separately for each time period. Therefore, the split-plot was considered a group of two cone-tainers from the same time point containing the same soybean cultivar, with each individual cone-tainer considered a subsample (two subsamples per split-plot). We conducted three separate runs of the experiment with eight blocks in each of the first two runs, and four blocks in the third. In the third run, all eight cone-tainers in each bucket were allocated to the 30 d group as sufficient statistical power had been achieved in the first two runs of the experiment to test our hypotheses involving the 60-d treatments.
Aphid-density treatments were applied to whole plots when plants reached the first trifoliate or V1 stage [27], which occurred 10 d after planting. Mixed-aged apterous A. glycines were transferred from a greenhouse biotype-1 colony (i.e. avirulent to all known A. glycines resistance genes, Hill et al. [28]) to each plant assigned to the five-aphid and ten-aphid treatments. Each whole plot bucket was then covered with a modified paint strainer (Trimaco, Morrisville, NC) to prevent the movement of aphids among whole plots. Aphis glycines populations were then allowed to increase for the remainder of the experiment.
Cone-tainers in the 30-d group were harvested from each whole plot at 30 d after planting, and data were collected as described below. Plants assigned to the 60-d group were transferred with all the soil within their respective 125-ml cone-tainers to 650-ml conetainers (Stuewe & Sons, Tangent, OR) after 30 d. The new conetainers then were filled to 650 ml with the addition of H. glycinesinfested soil-sand mixture and placed back into the water bath. These larger cone-tainers prevented soybean roots from becoming tangled and pot bound before the plants were harvested at 60 d, allowing for easier extraction of H. glycines females and cysts from roots.
All A. glycines, both nymphs and adults, were counted for each plant before the root mass of each plant was soaked in water to dislodge the soil. Roots were sprayed with pressurized water to dislodge H. glycines females and cysts, which were captured on a 250-mm-pore sieve positioned below a 850-mm-pore sieve. The total number of females and cysts recovered from each plant was counted under a dissecting microscope. Females and cysts were then ground on a 250-mm-pore sieve using a motorized rubber stopper [29], and released eggs were recovered on a 25-mm-pore sieve nested below a 75-mm-pore sieve. Eggs were suspended in 100 ml of water, and the number of H. glycines eggs present in a representative 1-ml sample of solution was counted under a dissecting microscope. The total number of H. glycines eggs recovered from each plant was calculated.

Data analyses
Data collected from the 30-d and 60-d groups of plants were analyzed separately using analysis of variance (ANOVA) with a mixed effects model. The model included the fixed effects of experimental run, block, aphid density, and soybean cultivar. The interactions of run*block, run*aphid density, block*aphid density, and aphid density*soybean cultivar were included as fixed effects. The whole-plot error term of run*block*aphid density was included as a random effect, along with the effect of subsample (i.e. plant nested within aphid density*soybean cultivar). This model allowed us to assess the effects of soybean cultivar, aphid density, and their interaction on the total number of H. glycines females and cysts and eggs produced plant 21 .
The number of H. glycines females and cysts plant 21 and eggs plant 21 were log transformed to meet the assumptions of ANOVA (non-transformed data are presented in all figures). These data were analyzed to determine if soybean cultivar, aphid density, or their interaction affected the number of H. glycines females and cysts present or the number of eggs they produced.
Based on the results of our initial analyses, we hypothesized that the effect of A. glycines on H. glycines reproduction varied with the population density of H. glycines. Despite our attempts to limit competition by using an initial low H. glycines population density, the numbers of H. glycines measured at 30 d and 60 d were high enough to suggest that competition may have occurred among H. glycines females. We hypothesized that the competition among H. glycines females would be increased with the addition of A. glycines. To test this hypothesis, we plotted the effect of A. glycines feeding on H. glycines population densities across the average H. glycines population density in the three aphid treatments. We calculated the effect of A. glycines feeding as the percent change in H. glycines population densities between the ten-aphid treatment mean and zero-aphid treatment mean for each combination of cultivar and time period (4 total data points). The ten-aphid treatment was selected because it generally represented the strongest effect of A. glycines feeding on H. glycines reproduction. We plotted the H. glycines females and cysts plant 21 and eggs plant 21 data separately.

Aphis glycines populations
Mean A. glycines population densities plant 21 (6 SEM) among the ten-, five-, and zero-aphid density treatments were 278624, 225616, and 361, respectively for the 30-d group. Upon transfer of the 60-d group plants from the 125-ml cone-tainers to the 650ml cone-tainers, there were only a few aphids on plants in the zeroaphid treatment. These aphids were removed before the nets were placed back over the buckets. At the conclusion of the 60-d group, A. glycines population densities in the ten-and five-aphid treatments had declined to 54610 aphids plant 21 and 99614 aphids plant 21 , respectively.

Heterodera glycines population density at 30 d
Numbers of H. glycines females plant 21 for the 30-d group varied significantly by experimental run (F = 6.27; df = 2,33; P = 0.0049), cultivar (F = 619.06; df = 1,18; P ,0.0001), and the interaction of aphid density*cultivar (F = 6.18; df = 2,18; P = 0.0090). Consequently, the analysis was performed by cultivar to discern the effect of aphid density on the number of H. glycines females plant 21 . On the H. glycines-susceptible cultivar, the aphid density treatment factor had a marginally significant effect (F = 3.59; df = 2,9; P = 0.0715) on numbers of females plant 21 with H. glycines population densities decreasing with increasing aphid density (Fig.  1). On the resistant cultivar, numbers of females plant 21 varied significantly by experimental run (F = 11.06; df = 2,11; P = 0.0023) and by aphid density (F = 4.57; df = 2,9; P = 0.0428). The number of H. glycines females plant 21 increased as aphid density increased on the resistant cultivar, with a 28% increase in numbers of females between the zero-aphid density and ten-aphid density treatments (Fig. 1).

Effect of Heterodera glycines population density
Our data summary analyses of both numbers of H. glycines females and cysts plant 21 and eggs plant 21 revealed that the effect of A. glycines feeding on H. glycines reproduction was highly dependent on the population density of H. glycines (Fig. 3). The trend suggested that as H. glycines population densities increased due to either soybean cultivar (susceptible versus resistant) or number of generations (60 d versus 30 d), increasingly negative effects of A. glycines feeding on H. glycines reproduction were observed. However, at the lowest H. glycines population (resistant cultivar at 30 d), A. glycines feeding increased H. glycines reproduction.

Discussion
In our experiment, A. glycines feeding significantly affected reproduction of H. glycines. However, the outcome of this interaction varied significantly with the cultivar and length of experiment. In the 30-d experiment, we observed increased H. glycines reproduction on the H. glycines-resistant cultivar and decreased reproduction on the susceptible cultivar in response to A. glycines feeding. In the 60-d experiment, we again observed decreased H. glycines reproduction in response to A. glycines feeding on the susceptible cultivar, however we did not observe any effect on the resistant cultivar. We believe the differences in the effect of A. glycines feeding on H. glycines reproduction to be due to differences in overall pest population densities as mediated by soybean cultivar and experiment length. Support for this conclusion can be found both in the results of the final regression analyses and the A. glycines population density data. Our summary analyses (Fig 3.) indicate that higher numbers of H. glycines females increased the severity of competition experienced by females upon the addition of A. glycines to plants. Competition for limited plant resources may also explain the decline in A. glycines population densities from 30 d to 60 d in both the five-and ten-aphid treatments. Soler et al. [3] predicted that the population density of herbivores, especially phloem-feeders, a feeding guild that includes aphids and nematodes, would affect the outcome of interactions with other herbivores. More specifically, Soler et al. [3] predicted that facilitation would occur at lower herbivore densities and competition at higher densities. The results of this experiment provide evidence supporting this hypothesis. In contrast, Johnson et al. [30] found that increasing durations of aphid infestations, and therefore increasing population densities, did not diminish the positive effect of aphid feeding on belowground wireworms. This discrepancy may be a result of the aphid population densities in the Johnson et al. [30] experiment not reaching a threshold to induce competition among the wireworms, or it may be due to a difference in how belowground chewing herbivores (i.e. wireworms) and belowground piercing-sucking herbivores (i.e. H. glycines) respond to increasing aphid population densities.
Aboveground lepidopteran herbivores are reported to affect belowground plant-parasitic nematodes in soybean [15], [16], [31], with the strength of the effect influenced by insect population density [16], [31]. Generally, these studies reported increasing nematode reproduction for both H. glycines and root-knot nematodes (Meloidogyne spp.) in response to increasing insect density or damage. This effect is counter to our observation of a variable response of the nematode to increasing aphid population density. This difference in trends may be due to differences in pest population densities in the experiments and the magnitude of their subsequent effect on plant quality, or it may be due to a difference in the resources utilized by the different insect feeding guilds. Both A. glycines and H. glycines feed from vascular plant tissue, increasing the likelihood for resource competition to occur, whereas lepidopteran herbivores feed on foliage. Therefore H. glycines and A. glycines could affect each other's performance both through the removal of shared nutritional resources and activation of related defense pathways [32], [33].
In more recent research, conflicting results concerning the effect of A. glycines feeding on the reproduction of H. glycines are reported.
McCarville et al. [20] found that simultaneous infestations of A. glycines and the causal agent of brown stem rot disease, C. gregata, increased H. glycines reproduction. However, Herren et al. [21] reported that H. glycines reproduction was unaffected by the presence of A. glycines. Both of these experiments used small, field micro-plots to measure H. glycines reproduction in response to artificial infestations of A. glycines. Therefore, it is worth comparing these two experiments to frame the results of our current greenhouse experiment.
McCarville et al. [20] observed H. glycines reproduction to be 5.24x greater on both H. glycines-resistant and susceptible cultivars when plants were also co-infected with A. glycines and C. gregata compared to plants infected with H. glycines alone. This observation was taken from soybean plants infected with C. gregata at planting and later infested with A. glycines at the early vegetative V3 stage and then comparing end-of-season H. glycines egg population densities to beginning-of-season population densities. Therefore, this increase in H. glycines reproduction was measured across an entire growing season. In our current experiment, we measured H. glycines egg production to be 1.34x greater in the presence of A. glycines on the resistant cultivar after 30 d. The 30-d period was a measurement of a single generation of H. glycines reproduction. In the U.S., H. glycines can complete three to six generations per year [34]. If the 1.34x increase we observed after 30 d occurred across all six generations in the field, we would expect to see a 5.79x increase for the entire year, which is consistent with the findings reported by McCarville et al. [20], suggesting that A. glycines feeding was primarily or solely responsible for the observed increase in H. glycines reproduction in that field micro-plot experiment. It is also noteworthy that, although H. glycines resistant and susceptible cultivars supported significantly different H. glycines populations in McCarville et al. [20], these populations responded similarly to A. glycines feeding (i.e. H. glycines population densities increased). Therefore, A. glycines-mediated competition for resources with H. glycines may not occur in the field due to the lower H. glycines population densities present in field environments. Supporting this conclusion are the H. glycines egg population densities we observed in the current experiment, 55,941 and 91,209 eggs 100cc soil 21 in the 30-d and 60-d SCN-susceptible cultivar treatments, respectively, and the average end-of-season H. glycines egg population densities in Iowa soybean fields, 2,438 eggs 100cc soil 21 (maximum 34,975 eggs 100cc soil 21 ) [13], [35][36][37][38][39][40][41]). This conclusion is consistent with the findings of Johnson et al. [1], specifically that negative effects of aboveground herbivores on belowground herbivores are more likely to be observed in laboratory studies than field studies.
Heeren et al. [21] manipulated the population densities of both H. glycines and A. glycines using a full factorial treatment arrangement of resistant and susceptible lines (i.e. susceptible to both, resistant to both, resistant to A. glycines, and resistant to H. glycines). They did not detect an effect of A. glycines feeding on H. glycines reproduction on any of the soybean lines. This result may be due, at least in part, to the extremely low pest population densities present in their study, including ,100 H. glycines eggs 100cc soil 21 and ,100 cumulative aphid days (i.e. ,10 aphids plant 21 for ,10 d) for some soybean lines.
Given the results of our current experiment and the previous results of McCarville et al. [20] and Heeren et al. [21], we propose the following model to explain the effect of A. glycines on H. glycines reproduction. Aphis glycines feeding increases the quality of soybean as a host for H. glycines through the manipulation of plant defenses [33] and/or a change in nutrient content [42]. An estimated 28-56% of H. glycines juveniles that penetrate susceptible plants reach adulthood [43], [44]. We propose that A. glycines feeding increases the percentage reaching adulthood irrespective of the cultivar's resistance to H. glycines. At the 30-d time point in our experiments, H. glycines females which reached adulthood would have established their feeding site before aphids were added to plants. Therefore, A. glycines did not affect juvenile H. glycines penetration or feeding site establishment. Aphis glycines increased numbers of H. glycines females and eggs, but had no effect on fecundity or eggs female 21 (analysis not shown). Therefore, the effect of increased H. glycines reproduction observed in our experiment was likely due to an increased number of females. This increase could be through  21 as affected by H. glycines population density. The effect of A. glycines on H. glycines reproduction was calculated as the ratio of the mean of the ten-aphid treatment divided by the zero-aphid treatment. The average number of H. glycines females and cysts plant 21 and eggs plant 21 was calculated as the average of the ten-, five-, and zero-aphid treatment means. Aphis glycines increased H. glycines reproduction at the lowest H. glycines population density, with competition occurring at higher population densities. Asterisks denote data points in which the effect of aphid density was significant at P = 0.10 (see figures 1 and 2). doi:10.1371/journal.pone.0086415.g003 an increased ability of the nematodes to obtain nutrients from the feeding site (i.e. change in nutrient content) or to sustain the feeding site (i.e. change in plant defenses). If an increase in numbers of H. glycines females is due to a change in plant defense, this is likely due to a suppression by A. glycines of a broad-based, general plant defense to nematodes that is not mediated by rhg genes. This interaction, however, is density dependent, with A. glycines increasing H. glycines reproduction at all pest densities except at very low A. glycines population densities (,10 aphids plant 21 ), where aphid feeding has no effect on H. glycines reproduction (see [21]), or at high pest population densities (see Fig. 3), where A. glycines and H. glycines compete for limited nutritional resources.
Going forward, it will be essential to determine under what range of field conditions A. glycines feeding leads to an increase in H. glycines reproduction or competition with H. glycines. It is also necessary to determine whether abiotic factors, such as drought, soil pH, or soil nutrient content can affect the outcome of the interaction indirectly by mediating host plant quality. Finally, given the widespread distribution of both A. glycines and H. glycines and the economic significance of both pests, it will be important to explore the need for an integrated management approach that mitigates yield reductions that occur both from A. glycines removing plant nutrients and from increasing the population density of H. glycines. Therefore, a multi-location field study is warranted to investigate this potentially significant aboveground-belowground interaction across a diversity of aphid population densities and infestation timings, nematode population densities, and abiotic conditions.