Predators of monarch butterfly eggs and neonate larvae are more diverse than previously recognised

Conserving threatened organisms requires knowledge of the factors impacting their populations. The Eastern monarch butterfly (Danaus plexippus L.) has declined by as much as 80% in the past two decades and conservation biologists are actively seeking to understand and reverse this decline. While it is well known that most monarchs die as eggs and young larvae, few studies have focused on identifying what arthropod taxa contribute to these losses. The aim of our study was to identify previously undocumented predators of immature monarchs in their summer breeding range in the United States. Using no-choice feeding assays augmented with field observations, we evaluated 75 arthropod taxa commonly found on the primary host plant for their propensity to consume immature monarchs. Here we report 36 previously unreported monarch predators, including representatives from 4 new orders (Orthoptera, Dermaptera, Lepidoptera and Opiliones) and 11 taxa (Acrididae, Gryllidae, Tettigoniidae, Forficulidae, Anthocoridae, Geocoridae, Lygaeidae, Miridae, Nabidae, Erebidae and Opilliones). Surprisingly, several putative herbivores were found to readily consume immature monarchs, both in a targeted fashion or incidentally as a result of herbivory. This work expands our understanding of the monarch predator community and highlights the importance of unrecognized predation on insects of conservation concern.

widespread adoption of herbicide resistant corn and soybean and subsequent use of broad-spectrum herbicides 26 . Accompanying the loss of milkweed in the summer breeding range is a shift in the habitat where remaining milkweed stems occur. With the near elimination of milkweed from annual crop fields, most remaining monarch breeding habitat in the Midwestern US occurs in perennial grasslands 27,28 . Previous studies show that grasslands also support large and diverse populations of arthropod predators 29 and predation of Lepidoptera eggs in these habitats frequently exceeds those in annual croplands 21,30 . Therefore, the effect of milkweed loss from croplands in the Midwestern US on monarch declines may be exacerbated by the potential for increased risk of predation in remaining grassland habitat 20 .
Like many Lepidoptera, survival of monarch eggs and early-instars is quite low 31 . Some mortality is due to interactions with milkweed defenses 32 or extreme weather events 33 , but predation is also a key factor in monarch mortality 31 . One study reported 78% mortality of monarch eggs and 59% mortality of first instars, and noted that on occasion ants removed 100% of eggs and larvae from individual plants 34 . Another study reported approximately 98% mortality of sentinel monarch eggs after 7 d and high rates of loss on plants with ants and aphids present in a Wisconsin, US old-field 35 . A study conducted in Minnesota, US considered cumulative proportion survival of monarchs in a restored prairie and found only 20% of eggs survived to hatching, with <10% survival to 2nd instar, and <2% to 3 rd instar 31 . However, despite this evidence, predation was seldom directly observed and most eggs and young larvae were reported to simply disappear.
To more fully understand the impact of predation on monarchs, we need information on which arthropods contribute to the loss of eggs and young larvae. In 2015, a literature review identified 12 arthropod taxa as predators of monarch eggs and/or larvae 36 ; these included members of the following taxa: Chrysopidae 36 , Formicidae 31,35,[37][38][39][40] , Coccinellidae 34,41,42 , Araneae 31,37,43 , Vespidae 43,44 , Pentatomidae 31,43 and Mantidae 45 . However, many other arthropods are commonly found on milkweeds, causing us to ask if other taxa also contribute to predation losses. Finally, while many of the milkweed-specialist herbivores have been studied in detail [46][47][48][49] , the wider community of generalist herbivores and omnivores that frequent milkweed stems has received less attention. To our knowledge, no one has directly examined the potential for these arthropods to consume monarchs; they could do so intentionally, or incidentally while eating leaf material where monarch eggs or larvae are present.
Given the likely importance of predation in limiting monarch population growth, the aim of this study was to identify which of the many arthropods that visit common milkweed have the potential to prey on monarch eggs and young larvae. We used field observations to determine which arthropods visit milkweed, then tested their propensity to consume monarch eggs and neonate larvae in laboratory no-choice trials. Finally, we used additional observations to confirm predation under field conditions. Based on previous observations, we predicted that most predatory and omnivorous taxa would consume eggs and larvae, and that at least some of the putatively herbivorous taxa would do so as well.

Results
We collected a total of 779 individual arthropods from 75 taxa across 11 orders and 33 families and tested their predation potential under no-choice laboratory conditions. From these we found 34 unique taxa representing 8 orders which consumed monarch eggs and 30 taxa across 8 orders that consumed neonate monarchs (Table 1). These include 4 orders of arthropods not previously reported to consume immature monarchs (Orthoptera, Dermaptera, Lepidoptera, and Opiliones), including 11 new families and 25 species (excluding at least 11 other distinct taxa not identified to the species level). Monarch eggs were consistently consumed (defined here as predation in >50% of trials with n ≥ 3) by 16 taxa including: Melanoplus differentialis (Thomas) and other Acridide spp., several Oecanthus and Allonemobius species, Nabis americoferus (Caryaon), Podisus maculiventris (Say), Neoconocepjalus spp. and other katydids in the family Tettigoiidae, Chysopidae, Forficula auricularia (Linnaeus), Plagiognathus spp., Coleomegilla maculata (DeGeer), Crematogaster cerasi (Fitch) and Tapinoma sessile (Say). Monarch neonates were consistently consumed by 11 taxa including: several Oecanthus and Allonemobius species, N. americoferus, adult P. maculiventris, Chysopidae larvae, F. auricularia, various Coccinellidae and Formicidae (Table 1). We also found 22 taxa that consumed eggs or neonates occasionally (at least once but in less than 50% of trials, or with n < 3), and 9 taxa that consumed monarchs in every trial but which were observed with very limited replication (n < 3).
Independent field observations confirmed that many of the potential predators identified by our lab trials also prey on monarch eggs and/or neonates under natural field conditions 50 . These include: Oecanthus fultoni (Walker), species in the family Tettigoniidae, and Chrysopidae, F. auricularia, Lygaeus kalmii (Stål), Plagiognathus spp., P. maculiventris, a Nabidae, Harmonia axyridis (Pallas), Formica subsericea (Say), and two families of arachnid, Salticidae and Opilliones. In addition, individuals in the following families that we did not test in lab trials were observed to consume monarch eggs or larvae in the field: Carabidae, Cantharidae, and Trombididae.
The number of milkweed-visiting taxa that did not eat monarch eggs and or larvae in our trials was approximately equal to the number that consumed them ( Table 2). These include several taxa that have previously been reported as monarch predators (e.g., Tenodera aridifolia sinensis, Polistes dominulus), or are part of known predaceous groups (e.g., Pentatomidae, Coccinellidae and Formicidae).

Discussion
The decline of the Eastern migratory monarch overwintering population has sparked concern from citizens and scientists alike. While many factors likely contribute to the decline, predation is one of the most significant sources of mortality for eggs and neonates 31 , and may be exacerbated by the monarch's increased reliance on perennial grasslands where predator populations are diverse and abundant 20,29,30 . A recent modeling study predicts that as little as a 4% increase in survival of breeding monarchs in the North Central US could potentially lead to recovery of the overwintering population 24 ; therefore, understanding which arthropods prey on monarchs is an important step toward designing and managing monarch-friendly habitats. Since field-observed predators were not always identified to the same taxonomic resolution as those in the lab, some observations are noted at higher taxonomic levels (e.g., Chrysopidae, Opiliones). Superscript bracketed numbers represent previously published studies in which taxa were listed as predatory on immature monarchs.
www.nature.com/scientificreports www.nature.com/scientificreports/ There are several reports of monarch predation in the literature; however, most of these are anecdotal or based on stochastic events which do not capture the breadth of potential predators in milkweed habitats. In addition, all predation events to-date have been observed in daylight hours yet a recent study suggests that a significant portion of predation occurs at night 50 , further suggesting current observations are lacking. In the most complete list of monarch predators prior to our work, the majority of observations focused on predators of adults and only 12 predators of monarch eggs and neonates were described 36 . Our results more than double the number of predators of immature monarchs and show they are far more diverse than previously reported. We found 30 new egg predators and 25 new larval predators, including representatives from 11 families and 4 orders that were not previously reported to prey on monarchs. Despite the monarch being a classic example of defense sequestration leading to protection from higher trophic levels we still see a considerable amount of predation 51,52 . Such findings highlight the importance of evaluating the breadth of predation for this insect and the many other specialists thought to have escaped such top-down interactions by commandeering host plant defenses.
Ants have long been implicated as important egg and larval predators of monarchs, and are common on milkweed plants 34 . For example, in one recent study they comprised 69% of all predatory arthropod individuals found on milkweeds in grasslands 20   www.nature.com/scientificreports www.nature.com/scientificreports/ in ca. 90 min in a grassland setting 35 . Fire ants (Solenopsis invicta Buren) have also been implicated as a driver of immature monarch mortality in Texas 38,39 , and in other studies ant abundance was negatively related to immature monarch survival 40 . Many of the ant species we tested also fed on monarchs, although they differed in their behaviours and preferences. For example, Tetramorium caespitum occasionally removed both eggs and larvae, while Tapinoma sessile removed eggs but not larvae, and Formica vinculans (Wheeler) removed one of two eggs but attacked larvae aggressively and consistently. To allow ants to forage naturally in our trials, we gave entire colonies access to arenas with monarch eggs or neonates. However, we observed that some colonies foraged much more actively than others and removal rates are in part a function of colony activity levels. We also noted interesting differences in foraging behaviour. For example, T. sessile only foraged nocturnally, removing eggs in 4 out of 8 trials. In contrast, in the acrobat ant C. cerasi foraged diurnally in large groups and removed eggs in all trials, sometimes within a just a few minutes of gaining access to the container. However, C. cerasi removed only 2 of 5 larvae despite frequently encountering and antennating them. Finally, Lasius neoniger (Linnaeus) which has been reported as a voracious predator of other Lepidopteran eggs in turf grass 53 did not attack eggs or larvae in our trials.
The presence of spiders on milkweed has also been associated with increased monarch mortality 31 .
In our experiments, we tested three spider families: Aranidae (orb-web spiders), Salticidae (jumping spiders) and Thomosidae (crab spiders). All three families were found to consume monarch neonates, but only spiders in the family Thomosidae consumed monarch eggs (and did so in only 1 of 34 trials). Spiders may have favoured neonates over eggs because they often rely on prey movement in their foraging behaviour 54 . Since spiders are abundant and diverse in grassland habitats, a larger survey of spiders including ground-dwelling taxa which were not tested here could provide additional clarity on the role of these predators in monarch mortality.
Individuals from both Forficulidae and Coccinellidae are commonly observed on milkweed plants and may be important predators of monarch eggs and neonates 21 . In our study, F. auricularia were consistently predaceous, consuming eggs in 8 out of 10 trials and neonates in 13 out of 14 trials. F. auricularia are predominantly nocturnal and in the field we often observed them resting during the day hidden in the newly forming leaves at the apex of milkweed plants, a position also favored by 1 st and 2 nd instar monarch larvae 31 . In addition, we tested 9 species of Coccinellidae; of which, the adult forms of all species except Brachiacantha ursina (Fabricus) consumed monarch eggs or larvae. In contrast, immature Coccinellidae were generally less likely to consume monarchs, although replication for this life stage was low. We note that while we tested taxa from 8 Coleopteran families, Coccinellidae were the only family to consume monarchs. In particular, H. axyridis was the most consistent predator from the family Coccinellidae in our trials. In the larval form they consumed eggs and neonates in all trials (n = 8 and 11, respectively); adults consumed eggs in 8 of 10 trials and neonates in 10 of 10 trials. It has been previously demonstrated that H. axyridis is capable of imposing strong predation pressure on immature monarchs in controlled laboratory and field trials 42,55 . However, when H. axyridis were presented with alternative aphid prey (Aphis nerii Boyer de Fonscolombe), monarch consumption declined with increasing aphid populations 55 . Additional studies in open-field settings will help to elucidate the role of lady beetles on immature monarch survival.
In addition to known predatory or omnivorous taxa, we also examined the consumption potential of several common, but putatively herbivorous, arthropods. While not previously reported as monarch predators, all 12 of the Orthopteran taxa we tested consumed monarch eggs or neonates, and some did so quite consistently (although for Melanoplinae and individuals from the family Tettigoniidae, results differed between immature and adult stages). Interestingly, most of the incidences of predation by these herbivores occurred without herbivory, i.e. they selectively removed eggs directly off the foliage without consuming the foliage itself. In other instances, Acrididae individuals and Euchaetes egle (Drury) ate eggs along with the foliage they consumed. At least one previous study has documented monarch eggs removed from plants as a result of plant defoliation by insect and non-insect herbivores 43 . While the frequency of herbivores encountering monarch eggs and neonates in the field is unknown, some milkweed specialists (especially, E. egle and L. kalmii) can have large populations and may frequently encounter monarch eggs in the field. Finally, some species we tested remained strict herbivores. For example, Tetraopes tetrophthalmus (Forster) and Popillia japonica (Newman) failed to consume eggs or larvae in all trials.
In our assays, we used a no-choice laboratory arena to test for predation potential and consequently the behaviors we documented may not be representative of a field setting. Prey acceptance may increase under no-choice conditions due to starvation and increased encounter rates 56 . For example, in the field a given predator may forage on a different part of the milkweed plant than where monarch eggs or neonates are typically found, or might prefer alternative prey. In a prior study Polistes dominulua (Christ) readily consumed monarch larvae in no-choice assays, yet when provided with a choice between the toxic late-instar monarch larvae and a less toxic Pieris rapae (Linnaeus) or Trichoplusia ni (Hübner) they preferentially consumed the alternate prey 44 . Likewise, some arthropods that are important predators in a field setting might not eat monarchs in the artificial environment of a lab-based no-choice trial, and would need to be discovered using other methods. Two of the predators we tested (Tenodera aridifolia sinensis Saussure, Polistes spp.) which did not consume monarchs in our assay were previously reported as predators in the literature 36 . It is possible that the no-choice arena used in our assays simply did not facilitate normal foraging for these species. For these reasons it is best to confirm lab results against field observations. In this regard, representative species from eight of the nine orders we found to consume monarchs in lab no-choice tests were independently observed to prey on those stages under field conditions (Table 1). Additional field observations are needed to determine if the other taxa we observed to consume monarchs in the lab also do so in the field.
Evidence suggests most monarch eggs and neonates in summer breeding habitats succumb to predation 31,35 , and predation may be more prominent as monarchs now use milkweed in grassland habitats where predator abundance and diversity is high 29 . Therefore, in addition to filling gaps in the natural history of a well-studied organism, identifying monarch predators could provide knowledge that proves useful to conservation efforts. (2019) 9:14304 | https://doi.org/10.1038/s41598-019-50737-5 www.nature.com/scientificreports www.nature.com/scientificreports/ Reducing predator prevalence in important monarch habitats, or prioritizing habitats where predation pressure is lower, could allow more monarch eggs and neonates to reach adulthood and help to stabilize the overwintering population.

Methods
We tested a wide range of arthropod taxa found on milkweeds to determine whether they would consume monarch eggs and/or larvae. For each arthropod we began with the null hypothesis that it would not consume monarchs of either life stage. Any observation of predation during the feeding trials caused us to consider that taxon a potential predator and to seek confirmation that they attack monarchs under field conditions. Collection and identification of potential predators. We limited the range of arthropods we tested to those observed or collected on common milkweed, since they are most likely to encounter immature monarchs in a field setting. We excluded potential aerial predators not found foraging on milkweed stems, as well as parasitoids 36,57 . Arthropods used in the experiment were field collected from A. syriaca patches using sweep nets, aspirators, or hand collection in Ingham County, Michigan, USA and State College, Pennsylvania, USA during the summers of 2017 and 2018 and used in trials within 24 h of field collection. Once trials were completed, arthropods were frozen, placed in 70% ethanol, and identified to the lowest possible taxonomic level. In a few cases arthropods were only identified to coarse taxonomic groups (e.g., family or genus); this was particularly true for spiders and immature stages of some orders. Therefore, some test groups could potentially contain multiple species, and we refer to the group as "various spp. " To be conservative, we count each of these groups as a single predatory taxon even though it is possible it contains more than one species. www.nature.com/scientificreports www.nature.com/scientificreports/ egg predation trials. To determine which milkweed-visiting arthropods can consume monarch eggs, we performed no-choice assays in 473 mL (16 oz.) deli-cup arenas (Solo Bare DM16R-0090). Larger predators (e.g., Mantodea and Opiliones) were placed in 946 mL (32 oz.) deli-cup containers (Solo Bare DM32R-0090) to allow for more natural movement. Field collected leaves of common milkweed were rinsed in tap water to remove any naturally occurring arthropods or debris. Late in the season when common milkweed was senescing, swamp milkweed (Asclepias incarnata L.) leaves of the same general shape and size were field collected and used in trials. Leaves of approximately 10 cm in length were placed diagonally against the side of the deli-cup arena to allow potential predators full access to forage on the top and bottom of leaf surfaces (Fig. 1). The petiole of each leaf was placed in a damp cotton ball to avoid desiccation during the experiment. A single monarch egg, obtained from a colony of wild monarchs, was lightly glued (Elmers ® Glue-All) to the bottom side of each leaf, with a fine tipped paintbrush, to mimic the natural placement of eggs in nature. Field trials confirmed that naturally foraging predators readily consumed eggs attached in this fashion 21 . A single predator was placed in each deli-cup; then the cup was sealed with a perforated lid. Cups were placed in a climate-controlled growth chamber at 27 °C and 16:8 light cycle. Egg presence/absence was recorded at 48 h and each egg was also examined under a dissecting microscope for evidence of egg content removal by sucking arthropods. If an individual predator died during the assay, that replicate was discarded. neonate trials. Following the same basic procedure, deli-cup predation arenas were used to assess potential predators of freshly hatched monarch neonates. Field collected A. syriaca or A. incarnata leaves were placed diagonally in the arena and a single neonate was placed on each leaf. Neonates were transferred to leaves with a fine-tipped paintbrush and observed under a microscope to ensure they were not damaged. Following placement, the caterpillars were left to acclimate for 10-20 minutes before a potential predator was added. Arenas were placed in the growth chamber, and neonate presence/absence and condition (alive or dead) was recorded at 48 h. egg and neonate trials with ants. Because individual ants do not forage normally when displaced from the colony, ant predation was assessed by linking a predation arena (described above) to an ant colony held in the lab. Colonies of six different ant species and associated soil/litter were collected from locations in and around East Lansing, Michigan and placed in 20 × 21.5 × 11 cm (4 L) containers with Fluon (#2871C Insect-a-slip) applied to the inner top 2.5 cm to keep ants from escaping. Colonies were provided food and sugar water 2x per week and starved for 24 h prior to use in feeding trials to encourage foraging. As described above, monarch eggs or neonates were placed individually on field-collected A. syriaca or A. incarnata leaves in 473 mL deli-cups. We then connected ant colonies to the deli-cups using clear, flexible PVC tubing (0.64 cm ID, Model 702165 Home Depot; Fig. 1). We applied Fluon to the inner top 2.5 cm of each deli-cup to keep ants from escaping through the perforated lid. To initiate a trial, a wood coffee stirrer (3 mm width) was placed to connect the soil surface in the ant colony to the tube, allowing ants access to the test arena. The colony and arena assembly were then placed in the growth chamber and the egg/neonate presence was recorded at 48 h. field observations of predation. Field studies were conducted in 2017-18 to identify factors influencing monarch oviposition and survival in different crop and non-crop habitats 21 . During these trials, 1581 sentinel monarch eggs were placed on milkweed stems and observed every 2 h for 24 h and again at 48 and 72 h. During these observations any incidences of predation on monarch eggs and larvae were recorded and the predator was identified to the lowest possible taxonomic level without disturbing its ongoing behaviour 21 . In addition, video surveillance cameras were used to determine the fate of 152 sentinel monarch eggs on milkweed in grassland habitats 50 . Here we use these observations to determine if taxa tested in our no-choice trials also prey on monarchs under field conditions.

Data Availability
All data generated or analysed during this study are included in this published article.