The Role of Natural Enemy Foraging Guilds in Controlling Cereal Aphids in Michigan Wheat

Insect natural enemies (predators and parasitoids) provide important ecosystem services by suppressing populations of insect pests in many agricultural crops. However, the role of natural enemies against cereal aphids in Michigan winter wheat (Triticum aestivum L.) is largely unknown. The objectives of this research were to characterize the natural enemy community in wheat fields and evaluate the role of different natural enemy foraging guilds (foliar-foraging versus ground-dwelling predators) in regulating cereal aphid population growth. We investigated these objectives during the spring and summer of 2012 and 2013 in four winter wheat fields on the Michigan State University campus farm in East Lansing, Michigan. We monitored and measured the impact of natural enemies by experimentally excluding or allowing their access to wheat plants infested with Rhopalosiphum padi (L.) and Sitobion avenae (F.) (Hemiptera: Aphidae). Our results indicate that the natural enemy community in the wheat fields consisted mostly of foliar-foraging and ground-dwelling predators with relatively few parasitoids. In combination, these natural enemy groups were very effective at reducing cereal aphid populations. We also investigated the role of each natural enemy foraging guild (foliar-foraging versus ground-dwelling predators) independently. Overall, our results suggest that, in combination, natural enemies can almost completely halt early-season aphid population increase. Independently, ground-dwelling predators were more effective at suppressing cereal aphid populations than foliar-foraging predators under the conditions we studied. Our results differ from studies in Europe and the US Great Plains where foliar foraging predators and parasitoids are frequently more important cereal aphid natural enemies.


Introduction
populations. However, when foliar-foraging Coccinellidae predators were added to the system, the effect of both predator groups on aphid suppression was higher than the sum of each community. The reason for this synergetic effect was that Coccinellidae foraging caused aphids to drop from the vegetation onto the ground, where they were consumed by Carabidae [32].
While the complex of aphid natural enemies in cereal crops has been studied in parts of North America [33] e.g. South Dakota [7], eastern Washington [34,35], and Colorado [2,36], to our knowledge, the community structure or impacts of cereal aphid enemies has not been described in Michigan or the Midwest United States. To investigate the role of natural enemies in controlling cereal aphid populations in this region we conducted exclusion cage field studies in 2012 and 2013 in East Lansing, Michigan. The overall goals of these studies were to characterize the natural enemy community in Michigan winter wheat fields, determine their community-wide effect on aphid population growth, and determine the relative impact of foliar-foraging versus ground-dwelling predators on suppression of cereal aphid populations.

Study sites
Two independent studies were conducted on the Michigan State University campus, East Lansing, Michigan in 2012-13 using natural enemy exclusion cages to explore the impact of natural enemy communities on cereal aphid population growth. The first study was conducted in four different winter wheat fields in 2012 (2 fields) and 2013 (2 fields). The second study was conducted only in 2013 using the same sites (2 fields) as the first study. For both studies, the wheat was sown in October of the prior year and utilized for experimentation in May of the following year. Each field received typical commercial applications of herbicides, fungicides and fertilizers at rates determined by the university farm manager ( Table 1). No insecticides were applied in the fields in either year. In each field, a 30620 m area was delineated at least 35 m from the field edge. Individual plots within this area were established 5 m equidistant from each other in a completely randomized design.

Study 1: Community-wide Natural Enemy Impacts
The first study contrasted cereal aphid population growth using; open plots, which allowed unrestricted access of the ambient natural enemy community, closed plots which attempted to exclude all natural enemies, and sham cages which allowed natural enemy access but controlled for cage effects. Open plots consisted of circular area with a diameter of 0.36 m, without any barrier to natural enemies. Closed plots consisted of 1 m tall, 0.36 m diameter wire tomato support cages (Figure 1), covered with sewn sleeves of no-see-um mesh (approx. 625 holes per 6.45 cm 2 , Skeeta Inc., Bradenton, FL). The bottoms of the cages were buried 15 cm into the ground to prevent access by ground-dwelling predators. The top of the sleeves were tied with nylon cord to prevent flying predators from entering. The sham cages were identical to closed plots with the exception that the sleeves had multiple 10 cm slits on each side and at the bottom of the mesh allowing entry of predators and parasitoids, including foliar-foraging and ground-dwelling natural enemies. Cage treatments were initiated on May 18, 2012 and May 14, 2013, and were replicated five times per field in a completely randomized design.

Study 2: Guild-specific Natural Enemy Impacts
The second study contrasted cereal aphid population growth in the presence of selected natural enemy guilds in 161 m plots. Treatments included: exclusion of foliar-foraging predators and parasitoids (hereafter called -F), exclusion of ground-dwelling predators (-G), exclusion of both foliar and ground-dwelling natural enemies (-F-G), and fully open plots (O), which were exposed to all natural enemies. Each of the four treatments consisted of a 1 m 3 PVC frame cage, erected around the plots, with the legs buried in the soil. In the plots excluding ground-dwelling predators (-G), and in the plots excluding all natural enemies (-F-G), a 30 cm tall corrugated plastic barrier was erected around the PVC frame. The bottom 10 cm of this barrier was buried in the soil so that 20 cm was left above ground to restrict access by ground-dwelling predators (Figure 2A, B). To exclude all natural enemies (-F-G) and foliar-foraging predators and parasitoids (-F), the top and all sides of the PVC frame were additionally covered with no-seeum mesh, to prevent flying predators from entering the cage. To exclude all natural enemies (-F-G), the bottoms of the mesh were buried 5 cm into the ground while in -F plots, the bottoms of the mesh were raised 2 cm above the ground to allow access by ground-dwelling predators ( Figure 2C). Finally, open plots (O) consisted of 161 m area demarcated with flags, without any barrier to natural enemies ( Figure 2D). All treatments were initiated on May 14, 2013 and were replicated five times in each field.

Natural enemy sampling
To characterize the overall natural enemy community in the wheat fields, we used three sampling methods in both studies. Natural enemies were sampled weekly (Study 1 from May 29 to June 19 of 2012 and May 21 until June 5 2013, and Study 2 from May 21 until June 5 of 2013). To sample the ground-dwelling predator community, pitfall traps (plastic Solo cups, 11 cm in diameter and 14 cm in depth) were placed at each study site. Four pitfall traps per site were used for Study 1 and twenty per site for Study 2. Pitfall traps were filled 1/3 of the way with 40% propylene glycol solution. The pitfall traps for Study 1 were established next to the plots and 4 m apart from one another, and for Study 2 were established inside the cages 5 m apart from one another. To sample foliar foraging predators, 23628 cm yellow sticky card traps (PHEROCON AM, Great Lakes IPM, Vestaburg, Michigan) were placed next to the pitfall traps (Study 1: n54; Study 2: n520). The yellow sticky cards were hung on a plastic step-in fence posts (Zareba Systems, Lititz, PA), and positioned just above the plant canopy. Finally, on each sample date, natural enemies were counted by visual observation for a fixed time (five minutes) in each plot. Any natural enemies found in closed plots were manually removed during the sampling. All the predators were identified in the field or returned to the laboratory for identification. Araneae and Opiliones were identified to order while most other organisms were identified to family. Due to their potential importance in aphid control, Coccinellidae were identified to species. Coccinellidae that were difficult to identify in the fields or were missing identifying features on sticky cards were categorized as ''other Coccinellidae.'' The overall mean abundance (¡ SEM) of each natural enemy taxa was calculated by field and by year.

Aphid Infestation and Population Growth
On May 18, 2012 and May 14, 2013 all plots were infested with 50 (Study 1) and 100 (Study 2) laboratory reared R. padi of mixed adult and nymphal stages, when wheat plants in the field were at the six-seven Feekes growth stage [37,38]. In 2012, the initial establishment of R. padi in the plots failed so all plots were reinfested with 50 additional aphids on May 23. Prior to aphid infestation, natural enemies were removed from each plot by hand and vacuuming the plot area using a modified leaf blower [39]. The aphids were transferred to the field treatment plants by cutting leaves infested with aphids from the cultured greenhouse plants and placing them between the leaves of the middle wheat plant of the plot. Aphid abundance was assessed once per week after infestation, by counting all aphids on all plants within the plots and recording number of wheat tillers present in the plots. Both alate and apterous aphids were recorded. On the first sampling date, naturally occurring S. avenae were also observed, so from that date on, counts of S. avenae were also recorded. To control for sampling effects, plants were manipulated to the same degree in all plots to ensure that all plants, aphids, and natural enemies received the same amount of disturbance.

Statistical analyses
For both studies, ANOVA procedures (R version 3.0.2) [40] were used to analyze within year populations of each species of per tiller of wheat by cage treatment, site, and sample day. A negative binomial error structure was used in all ANOVA analyses, and Rao's efficient score test was used in lieu of the F statistic because it is appropriate for models with non-normal error structure [41]. If statistically significant treatment differences were observed, means were compared by pairwise t-tests that had been Holm-adjusted for multiple comparisons to compare treatments. The code used for analysis, in the form of an R script, as well as all data used in analysis are available in a GitHub repository at https://github.com/ cbahlai/Safarzoda_2014.

Study 1: Community-wide Natural Enemy Impacts
Using all three sampling methods, we collected a total of 4,065 natural enemies representing 13 taxa ( Table 2). Overall, seven taxa of ground-dwelling natural enemies were captured in pitfall traps over both years. In 2012, the most common family in Field 1 were Formicidae followed by Araneae, Carabidae and Opiliones. In contrast, in Field 2 the most common natural enemies were Araneae followed by Carabidae. In 2013, in Field 3 and Field 4 Carabidae were the most common taxa followed by Araneae and Staphylinidae. Yellow sticky cards captured six families of flying predators. In 2012, in Field 1 the most common family was Dolichopodidae followed by Chrysopidae and Syrphidae while in Field 2 the most common family was Syrphidae followed by Chrysopidae, Dolichopodidae and the coccinellid Harmonia axyridis. In 2013, Syrphidae were the most common natural enemies in both fields followed by Nabidae in Field 3, Cantharidae and the coccinellid Coleomegilla maculata in Field 4.
During the visual sampling, six natural enemy families were observed. In 2012, Coccinella septempunctata and Syrphidae were the most commonly observed in Field 1. In contrast, Araneae were the most commonly observed taxa in Field 2, followed by Coccinellidae larvae and Syrphidae adults. In 2013, the most common taxa were adult C. maculata in both fields, following by adult H. axyridis, Syrphidae, Carabidae and Anthocoridae in Field 3, and adult C. septempunctata, H. axyridis and Syrphidae in Field 4.  Yellow sticky card  , open (exposed to all natural enemies), and sham (to control for cage effect) plots in Study 1. ANOVA was used to test statistical differences. Different letters next to the treatments indicate statistically significant differences among aphids per tillers at a50.05.

Study 2: Guild-specific Natural Enemy Impacts
Predator sampling Using all three methods of sampling we captured a total of 4,567 individual natural enemies representing 11 taxa (Table 3). No exclusion method completely prevented the occurrence of natural enemies. Pitfall traps captured seven taxa of natural enemies. In the -F-G plots, the most frequently captured taxa were Staphylinidae followed by Carabidae, Araneae and Formicidae. In contrast, in -G, -F, and O plots, the Carabidae were the most common taxon, followed by Araneae, and Staphylinidae. Overall, the numbers of ground-dwelling predators captured in the -F-G and -G plots were less than they were in the -F and O plots.
On yellow sticky cards, we captured five families of flying predators. In -F-G plots, a single Coccinellidae was captured. In the -G plots Syrphidae were the most common family captured, followed by Cantharidae. In -F plots, the Coccinellid C. maculata, followed by Syrphidae were the most common taxa captured, and in O plots, Syrphidae followed by C. maculata and Cantharidae were most frequently captured, together representing more than half of insects captured.
In visual observations, we occasionally observed adult C. maculata, C. septempunctata, Syrphidae, Formicidae, Araneae and Opiliones in the -F -G plots (i.e. 1-5 individuals of each taxa over the course of the study). In -G plots, C. maculata followed by Syrphidae, were most commonly observed, representing almost 75% of observed insects. In the -F plots, C. maculata followed by Araneae and Carabidae were most commonly observed, although fewer insects were observed, in general, in these plots. In the O plots, the most common natural enemy was C. maculata, at just under 40% of observations, followed by Syrphidae and Carabidae, representing 20 and 10% of observed insects respectively.
For Carabidae and Coccinellidae, often the most important predators of cereal aphids, the exclusion techniques sufficiently reduced their numbers to allow us to examine their effect on aphid population growth: Coccinellids were significantly reduced in -F-G and -F plots compared to the open plots, and Carabids were significantly reduced in -F-G and -G plots ( Figure 4).

Aphid population growth
Treatments manipulating natural enemies had different effects on population growth of R. padi and S. avenae. R. padi populations varied significantly between treatments (Rao 3, 116 54610 7 , p,0.001), and days (Rao 2, 113 53610 7 , p,0.001), but not sites (Rao 1, 84 53610 6 , p50.066). In -F-G plots, R. padi increased from 0.1 aphids per tiller to a mean of just over 5 aphids per tiller over the three week interval ( Figure 5). Aphid increase was significantly lower when foliar-foraging predators had access to plots (-G plots). Aphid population increases were even lower when ground-dwelling predators alone (-F plots) had access to plots. Finally, in plots where all natural enemies had access (O plots), aphid populations remained steady at approximately 0.03 aphids per tiller during the 3 week period indicating that the natural enemy community as a whole was very effective in preventing R. padi increases.
S. avenae populations were lower than R. padi. Although ANOVA suggested that populations varied significantly between treatments (Rao 3, 116 5159250, p,0.001), days (Rao 2, 113 568233, p,0.001), and sites (Rao 1, 84 5233, p,0.001), however, post hoc tests did not find any statistically significant differences, although the general pattern of population growth by treatments was largely similar to that of R. padi ( Figure 5).

Collection methods Plots
Pitfall traps Yellow sticky card

Discussion
From both studies we observed that the intact natural enemy community effectively controlled aphid population densities in wheat in all sites in both years. Similar to other studies in North America [2,33,42,43] the most common taxa of foliar-foraging predators that we found included adult Syrphidae, adult and larval Coccinellidae, adult Chrysopidae, Cantharidae and Nabidae, and the most common ground-dwelling predators were Carabidae, Staphylinidae, Araneae and Opiliones. Unlike other North American studies [44], we observed very low numbers of parasitoid wasps. In addition, the overall abundance of grounddwelling predators in our system was higher than that of the foliar-foraging predators.  Communities of ground-dwelling and foliar-foraging natural enemies, acting either independently or together, can reduce aphid population densities [25,26,30,45,46]. In fact, in both of our studies, when both predator groups were present (in the open plots), aphid population growth was very low. This significantly lower aphid population growth provided by the combination of foliar-foraging and ground-dwelling predators suggests a synergy between the different natural enemy foraging groups. Losey and Denno [32] showed the synergistic interaction of ground-dwelling and foliar-foraging predators on pea aphid in alfalfa. The aphids, in response to foraging Coccinelidae, dropped from the alfalfa canopy to the ground where they were consumed by ground-dwelling predators. Our results confirm that in combination these two foraging groups better suppress aphid populations than when acting alone.
Numerous studies show that adult Coccinellidae can suppress cereal aphid populations [42,47]. It has also been shown that certain Carabidae consume R. padi and S. avenae [24,48,49] and can reduce aphid density [30] by climbing the plant [4] or by preying on aphids when they drop from plants. In our study, ground-dwelling predators alone were more effective than foliar-foraging predators. These results are in contrast to previous European studies [25,46] where foliar-foraging predators and parasitoids were shown to contribute more to biological control of cereal aphids. Schmidt et al. [25] conducted an experiment in the early-season (May) in Germany and found that parasitoid wasps provided more effective aphid control than ground-dwelling predators. Thies et al. [46] conducted a similar experiment in five European regions and their results suggested that parasitoids and foliar-foraging predators were more important in controlling cereal aphids than ground-dwelling predators, but the relative importance of parasitoids and foliar-foraging predators greatly differed among European regions.
The increased efficacy of ground-dwelling predators that we observed in the present study might be due to the early-season importance of ground-dwelling predators. Previous work has suggested that the effects of ground-dwelling predators on suppressing aphids are strongest in early May [24,50,51], when aphid densities in cereals are low and reproduction is slower than in summer. In contrast, flying predators like Coccinellidae which primarily feed on aphids [7] usually become more important once aphid population densities increase [52].
Natural enemies are important for Michigan wheat production because they can keep aphid populations below damaging levels. The average numbers of S. avenae we found in the closed plots reached an average of 1-5 aphids per tiller, while in the open plots numbers of S. avenae never exceeded 0.03 aphids per tiller on average in either year. The average numbers of R. padi in our study reached nearly 11 aphids per tiller in the closed plots in 2013, although it did not exceed 0.3 aphids per tiller in 2012. The open plot reached a maximum average of just over 1 aphid of R. padi per tiller in 2013, which is much less than the economic thresholds. The economic threshold in Michigan for both species is 12-15 aphids per tiller [53], although in adjacent regions, economic damage has been observed with aphid numbers as low as 5 per tiller [54].
Under the conditions we studied, natural enemies provided sufficient and reliable aphid suppression, keeping cereal aphid populations below economically damaging levels. It is well established that the use of preventive insecticides may be harmful to existing and effective natural enemy communities [17,55], and result in pest resurgence [56]. If prophylactic or unwarranted insecticide sprays are used in Michigan wheat, they have the potential to disrupt the ecosystem services provided by the existing natural enemy community. Relying on natural biological control provided by aphid predators and only using chemical control when necessary will help insure more economical and sustainable insect pest management in Michigan wheat.