Do adult Magicicada (Hemiptera: Cicadidae) feed? Historical perspectives and evidence from molecular gut content analysis

Abstract

The periodical cicadas in the genus Magicicada are remarkable for their unusual life histories and dramatic synchronized emergences every 13 or 17 years. While aspects of their evolution, mating behaviors, and general biology have been well-characterized, there is surprising uncertainty surrounding the feeding habits of the short-lived adult stage. Despite a tentative scientific consensus to the contrary, the perception that adult Magicicada do not feed has persisted among the general public, and recent studies are lacking. We directly investigated the feeding behavior of Magicicada spp. through high-throughput sequencing (HTS)-based dietary analysis of nymphs, freshly molted (teneral) adults, and fully sclerotized adults collected from orchard and wooded habitats during the 2021 emergence of Brood X. Identifiable plant DNA (trnF, ITS amplicons) was successfully recovered from nymphs and adults. No plant DNA was recovered from teneral adults, suggesting that all DNA recovered from sclerotized adults was ingested during the post-teneral adult stage. Both nymphs and adults were found to have ingested a range of woody and herbaceous plants across 17 genera and 14 families. Significantly more plant genera per individual were recovered from adults than from nymphs, likely reflecting the greater mobility of the adult stage. We hypothesize that the demonstrated ingestion of plant sap by Magicicada adults is driven by a need to replace lost water and support specialized bacteriome-dwelling endosymbionts that cicadas depend upon for growth and development, which constitutes true feeding behavior.

periodical cicada, xylem feeding, metabarcoding, gut content analysis

Introduction

Periodical cicadas in the genus Magicicada W. T. Davis (Hemiptera: Cicadidae) are among the most well-studied and charismatic insects in North America, emerging in tremendous numbers every 13 or 17 years in locally synchronous broods. A rich body of scientific literature investigating the life history, reproductive biology, and evolutionary relationships of the 7 species in this complex has emerged over the past 150 years (Williams and Simon 1995, Simon et al. 2022). Despite this research interest, there is surprising uncertainty about a basic feature of the brief adult life of these insects: do adult Magicicada feed? This question has been the subject of considerable historical debate, and, despite a sometimes-ambiguous scientific consensus to the contrary, it remains widely held among the general public that Magicicada do not feed as adults.

The historical belief that adult cicadas in general do not feed on plants can likely be attributed to their lack of recognizable mandibular mouthparts and the inconspicuous nature of cicada feeding damage (Myers 1928). This Old World perception was apparently extended to Magicicada spp. upon the European colonization of North America (Myers 1928), with the earliest literature repeating that adult cicadas do not feed at all (Hildreth 1830), feed only on dew (Jaeger and Preston 1854), or feed only rarely (Marlatt 1898). Further complicating the subject, one early work (Burnett 1851) reported a distinction between the sexes in gut morphology and feeding habits, concluding that adult female Magicicada feed but males do not.

Feeding behavior of Magicicada was first approached with empirical rigor at the turn of the 20th century. Noting that adult mouthparts are robust enough to be functional, Quaintance (1902) directly observed feeding in a young pome orchard and confirmed the insertion of the stylets into branches by snipping the rostrums of feeding individuals and sectioning the plant tissue. Quaintance also observed Magicicada crops apparently distended with ingested plant sap and noted the presence of a complete alimentary canal in both sexes. Concluding that adults feed, he also reported temporal patterns in feeding behavior and widespread leakage of sap from feeding injury sites in orchards. Marlatt subsequently concluded in his 1907 revision of his earlier work (1898) that adults “feed in the typical Hemipteran fashion” (Marlatt 1907). Moreover, he added that Magicicada adults die if separated from plant material for a few days and that feeding appears to be necessary. Also noting the robust mouthparts and distended digestive tracts, Snodgrass (1921) agreed with these assessments and concluded that adults “feed abundantly by sucking the sap from trees and bushes”.

In contrast, Hargitt (1903) examined the internal anatomy of adult Magicicada and reported a degeneration of the digestive tract in both sexes. He concluded that adults feed little if at all and hypothesized that the guts of adults are converted into an “organ of absorption” to metabolize fat reserves in lieu of feeding. While he later accepted the evidence for a functional digestive tract in both sexes (Hickernell 1920), Hargitt maintained that feeding was extremely rare in adult Magicicada and that the utilization of the fat reserves was analogous to processes found in adult Ephemeroptera and spawning salmon (Hargitt 1923).

Several decades later, Alexander and Moore (1958) demonstrated that caged Magicicada adults quickly died when given access only to plant cuttings, but readily survived when caged directly onto living branches. They also reported large-scale feeding behavior of adults in the field and noted that “the defecation of a treeful of adults is like a continuous sprinkling rain”. Analysis of this liquid excrement in 3 nonperiodical cicadas and histological sectioning of plant material with stylets in situ indicated that adult cicadas specifically are xylem feeders (Cheung and Marshall 1973). By 1974, it was widely accepted that adult Magicicada feed (Dybas and Lloyd 1974), and detailed characterizations of feeding sites and behaviors were published (Dybas and Lloyd 1974, Maier 1982). Despite this general scientific consensus (Williams and Simon 1995, Simon et al. 2022), the belief that adult Magicicada do not feed has persisted among the general public. The most recent official source of this claim appears to be a short USDA publication published in 1978, in which it was simply stated that “control of the adult cicada is difficult because it does not feed after emergence” (Pierce 1978); no supporting data were provided.

Molecular gut content analysis can provide direct evidence of feeding activity in individual insects. In this approach to investigating feeding behavior, ingested plant barcoding genes are amplified via polymerase chain reaction (PCR), sequenced, and identified to genus or species. Variations of this approach that employ high-throughput sequencing (HTS) methods are especially powerful ecological tools as they allow the detection of multiple ingested plant species within a single sample (Pompanon et al. 2012). Such HTS-based methods have been used to infer the dietary histories and landscape movement patterns of sap-feeding insects in several Hemipteran families, including Psyllidae (Cooper et al. 2019), Triozidae (Cooper et al. 2019, Reyes Corral et al. 2021), Aphalaridae (Cooper et al. 2019), Liviidae (Cooper et al. 2019), Cicadellidae (Cooper et al. 2022), and Fulgoridae (Cooper et al. 2022, McPherson et al. 2022). Here, we present gut content data obtained from nymphal, teneral, and adult Magicicada spp. collected from orchard and wooded habitats during the 2021 emergence of Brood X in Maryland, Virginia, and West Virginia and discuss the implications of current evidence for feeding behavior in the adult insect. To our knowledge, this is the first application of gut content analysis to the Cicadidae and the first application of HTS-based dietary analysis to a xylem-feeding insect.

Materials and Methods

Insect Collection

Nymphal, teneral, and sclerotized adult Magicicada spp. were collected during the 2021 emergence of Brood X in Washington and Frederick Counties, Maryland, Frederick County, Virginia, and Berkeley and Jefferson Counties, West Virginia, USA, from apple orchards, a wooded public park, and wooded residential properties. Nymphs were collected before they emerged in early May from under rocks and logs or were dug up from just under the soil surface. Teneral adults were collected as they molted on tree trunks in late May, and sclerotized adults were collected by hand mid-May through mid-June in flight, from vegetation, or as they crawled on manmade structures. All insects were collected live into 50 ml centrifuge tubes filled with 100% molecular grade ethanol. Samples were either stored temporarily at −4 °C before being transferred for storage at −80 °C, or were stored immediately at −80 °C.

In total, 75 Magicicada spp. individuals were examined, including 30 late-stage nymphs (15 orchard, 15 wooded), 30 post-teneral (sclerotized) adults (15 orchard, 15 wooded), and 15 teneral (newly eclosed, unsclerotized) adults (orchard and wooded). Sclerotized adult insects were sexed and identified to species based on the descriptions provided by Alexander and Moore (1962); all three 17-year species were represented. The composition of adults collected from orchards was: 6 male, 7 female M. septendecim; and 2 female M. cassini. The composition of adults collected from wooded habitats was: 2 male, 1 female M. septendecim; 3 male, 8 female M. cassini; and 1 male M. septendecula. Teneral adults were sexed but were not identified to species due to the lack of pigmentation; there were 8 males and 7 females. No attempt was made to determine the sex or species of nymphs. A summary of sample compositions can be found in Table 1.

Table 1.

Summary of Magicicada gut content samples (1 insect per sample) submitted for sequencing and yielding identifiable plant DNA (“recovered”)

Lifestage	Habitat	Species	Sex	# Submitted	# Recovered	% Recovered	Avg. reads/recovered sample
Nymph	Wooded	.	.	15	4	26.7	90
	Orchard	.	.	15	2	13.3	27
Teneral	.	.	Male	8	0	0	0
			Female	7	0	0	0
Adult	Wooded	septendecim	Male	2	0	0	0
			Female	1	0	0	0
		cassini	Male	3	3	100	377
			Female	8	2	25.0	502
		septendecula	Male	1	0	0	0
	Orchard	septendecim	Male	6	1	16.7	239
			Female	7	3	42.9	113
		cassini	Female	2	2	100	198
			Total	75	17	22.7	207

DNA Extraction and Amplification

Prior to dissection, all insects were individually surface-sterilized by swirling 1–2 s in 70% molecular grade ethanol and 1–2 s in sterile deionized water, followed by a 1–2 min soak in a 1% bleach solution and two 1–2 s rinses in sterile deionized water (Cooper et al. 2019, Hepler et al. 2021). Approximately half of the samples were soaked in a bleach solution that was likely degraded with age (>24 hr old) and thus less effective at degrading surface DNA contaminants. However, we believe the combination of successive rinses and the selective inclusion of only internal digestive tissues was sufficient to exclude DNA contamination from the insect exterior. Wings and legs were removed in the final water rinse dish and the abdominal cavity was opened. Each insect was then transferred to a 5 ml microcentrifuge tube containing a solution of 4 ml phosphate buffer saline (PBS, 1×) and 250 µl molecular grade glycerol and allowed to soak for 5–60 min; longer soak times were found to be necessary to increase the pliability of delicate digestive tissues and facilitate their removal. Following this soak, the contents of each tube were emptied into a sterile Petri dish for dissection using flame-sterilized fine tipped forceps. While we attempted to remove entire digestive tracts intact, this was not always possible due to their small size, frequently indistinct morphology, and fragility. All tube-like structures were considered to be digestive in nature and were removed. When no obviously digestive structures were apparent, the most likely candidate tissues were included based on their position within the abdominal cavity. While this likely resulted in the inclusion of some respiratory and reproductive structures, we decided it was necessary to include potentially ambiguous tissues to ensure the entire digestive tract was represented. Following their removal, gut tissues were blotted dry on a clean KimWipe to remove surplus buffer and transferred to 1.5 ml microcentrifuge tubes held on a chill block. A blank was also included as a negative control sample; forceps were dipped in the surface sterilization and PBS + glycerol solutions, blotted dry on a clean KimWipe, and inserted into a 1.5 ml microcentrifuge tube such that the shanks of the forceps contacted the interior sides of the tube.

Total DNA was extracted using Qiagen DNeasy Blood and Tissue Kits (Qiagen, Hilden, Germany) following the manufacturer’s protocol, with the following adaptation. Following the incubation at 56 °C and the subsequent addition of Buffer AL and ethanol, only the top 350 µl of supernatant was transferred to the spin columns. We found this to be necessary to avoid column overloading with surplus tissue and the resulting potential reduction of DNA yield. Extracted DNA concentration was quantified on a NanoDrop, and all samples were diluted with water to 50 ng/µl.

Plant DNA in the trnF chloroplast gene region and the ITS2 gene (internal transcribed spacer) were amplified separately by PCR using universal primers. TrnF primers were B49873-e and A50272-f (Taberlet et al. 1991) and primers for ITS were ITS2F and ITS3R (Chen et al. 2010). Each sample was assigned a unique asymmetrically barcoded forward/reverse primer pair to permit the pooling and sequencing of all samples on a single PacBio SMRT cell. Each 50 µl reaction contained Invitrogen AmpliTaq Gold 360 Master Mix (Invitrogen, Carlsbad, CA), 250 nM Fwd and Rev primer (Invitrogen, Carlsbad, CA), and 250 ng of template DNA. PCR conditions were: an initial polymerase activation step of 94 °C for 5 min, 40 cycles of 94 °C for 30 s, 58 °C (trnF) or 56 °C (ITS) for 30 s, and 72 °C for 45 s, and a final extension at 72 °C for 5 min. PCR controls included 75 ng DNA extracted from cochineal scale insects [(Dactylopius sp.)(Hemiptera: Dactylopiidae)] collected from Opuntia Mill. sp. (prickly pear cactus), or 5 µl sterile deionized water. We opted to use extracted DNA from female Dactylopius collected from Opuntia as positive controls because: (i) Dactylopius is a sap-feeding insect (Mann 1969) and thus an approximate practical analogue for Magicicada; (ii) Dactylopius feed exclusively on members of the family Cactaceae (predominantly Opuntia and related genera) (Perez-Guerra and Kosztarab 1992, Claps and de Haro 2001, Chávez-Moreno et al. 2011,Chávez-Moreno et al. 2011) and are therefore of known feeding history; and (iii) Opuntia cacti are not typically found in mid-Atlantic deciduous woodland habitats and are unlikely to be encountered or fed upon by Magicicada, allowing any contamination of samples during PCR and misbinning during sequencing to be easily identified. Dactylopius DNA extracts were prescreened for the reliable production of trnF and ITS product bands. A positive barcoded control (75 ng Dactylopius DNA) was also included, and the blank sample served as a barcoded negative control. PCR products (400–600 bp) were visualized on a 2% agarose gel stained with GelRed fluorescent dye (Biotium, Fremont, CA) and purified with Qiagen QIAquick PCR Purification Kits (Qiagen, Hilden, Germany). Purified products were pooled into a single vial for sequencing according to observed band intensity: 10, 20, or 30 µl was contributed if a strong, weak, or absent band was observed, respectively. To avoid unbalancing the pooled samples with overly concentrated positive control DNA, only 1 µl of each barcoded positive control product was contributed.

Sequencing

Pooled samples were sent to Washington State University’s Laboratory for Biotechnology and Bioanalysis in Pullman, WA, for sequencing. Barcoded PCR products were concentrated by vacuum centrifuge and purified according to the manufacturer’s instructions using 1.0× AMPure XP beads (Beckman Coulter, IN, USA). PacBio SMRTbell sequencing libraries were prepared using the SMRTbell Express Template Kit 2.0 (Pacific Biosciences, Menlo Park, CA) following Pacific Bioscience’s provided Procedure and Checklist—Preparing SMRTbell Libraries Using PacBio Barcoded Universal Primers for Multiplexing Amplicons. The resulting library was purified again with 1.0× AMPure beads and further purified via size selection using a Blue Pipin instrument (Sage Sciences, Beverly, MA). The SMRTbell library was selected for a size range of 400–600 bp using the 0.2% DF agarose cassette and V1 size marker following the manufacturer’s instructions. Finally, the SMRTbell library was further concentrated using 1.0– AMPure beads.

SMRTbell libraries were annealed and bound to sequencing polymerase following the instructions within SMRT Link 9.0 software using the Sequel Binding Kit 3.0 (Pacific Biosciences, Menlo Park, CA). The bound SMRTbell library was sequenced on a PacBio Sequel with the Sequel Sequencing Kit 3.0 reagents and SMRTcell 1M V3 for 10-hour data collection time. Demultiplexing and circular consensus sequencing (CCS) analysis was performed using SMRT Link 9.0. Data were demultiplexed using a minimum barcode score of 70, and CCS analysis was subsequently performed on each demultiplexed data set with the minimum number of passes equal to 4 and 0.9999 minimum predicted accuracy.

Data Analysis

Analyses of sequence data were performed using Geneious Prime v. 2022.2.2 (Biomatters Ltd., Auckland, New Zealand) analysis software. Sequences were first filtered by size to remove nontarget and chimeric reads <400 bp and >700 bp. Sequences 400–700 bp in length were grouped into operational taxonomic units (OTUs) using the DeNovo Assembly tool. To reduce the impact of misbinning during the sequencing process, all OTUs with <5 reads were excluded from analyses, and remaining OTUs were trimmed to remove overhangs. One consensus sequence was generated for each OTU and identified with the Basic Local Alignment Search Tool (BLAST) function (Altschul et al. 1990) of the National Center for Biotechnology Information (NCBI) integrated with Geneious Prime. Putative identifications were accepted based on their % pairwise identities with reference sequences. For trnF OTUs, putative identifications were accepted to family level if % pairwise identity ≥90%, to genus level if ≥95%, and to species level if ≥98%. For ITS OTUs, putative identifications were accepted to genus level if % pairwise identifications ≥90%, and to species level if ≥95%. Identifications meeting these criteria were further evaluated for plausibility by verifying their regional occurrence using the species range maps feature of the USDA PLANTS database (USDA-NRCS 2022). OTUs identified as species or genera not occurring in the sample collection region or as unlikely congeners of a common species were re-analyzed using a modified BLAST search with maximum hits set to 10. The most probable identity from these top 10 closest matches was selected based on regional occurrence, % pairwise identity, % query coverage, and quality score. An identity was assigned at a conservative taxonomic level in the rare cases when this procedure did not result in a satisfactory identification.

The mean number of plant genera detected per recovered insect by lifestage was calculated and compared using Welch’s t-test due to unequal sample size and variance (PROC TTEST, SAS Institute 2019).

Results and Discussion

In total, 4,436 reads were recovered (3144 ITS; 1292 trnF). Of these, 914 ITS reads were identified as Opuntia sp. (prickly pear cactus) and were attributed to the barcoded positive control submitted for sequencing with the pooled samples. No reads from the trnF barcoded positive control or the barcoded negative controls were recovered. Of the remaining 3,522 reads, 2,494 (1896 ITS; 598 trnF) were recovered from samples collected in wooded habitats, and 1,028 reads (334 ITS; 694 trnF) were recovered from samples collected from apple (Malus domestica) orchards. Recovered samples (Table 1) included 6 nymphs (4 wooded, 2 orchard) and 11 adults (5 wooded, 6 orchard); 4 adults were identified as M. septendecim and 7 were identified as M. cassini. No plant DNA was recovered from the single M. septendecula submitted for sequencing or from any of the teneral adults. The mean number of plant taxa detected per individual differed significantly by lifestage (Welch’s t-test, t = 3.11, df = 10, P = 0.011). Significantly more plant genera per individual were detected in adults (mean ± SE genera/individual: 1.82 ± 0.26) than in nymphs (1.00 ± 0.00).

While ingested plant DNA was successfully recovered from nymphal and adult Magicicada spp., sequence read counts (Table 1) were substantially lower than figures reported from similar studies of other Hemipteran taxa (Cooper et al. 2019, 2023, Reyes-Corral et al. 2021b, 2021a). Nymphs were poorly represented in the sequence data, with plant DNA recovered from only 6 (6/30, 20% of submitted samples) individuals and comprising only 10% (414/3,522) of total reads. By contrast, plant DNA from 11/30 (36.7%) submitted adult samples were recovered, accounting for the remaining 88.2% of reads. No plant DNA was recovered from teneral adult samples. Several additive factors likely contributed to these results. First, sap feeding likely results in the ingestion of especially small quantities of host plant DNA in comparison to foliage feeding by chewing insects, as plant sap is not thought to contain high concentrations of DNA. This challenge has been noted in gut content studies of phloem-feeding insects (e.g., Cooper et al. 2016), which are speculated to only ingest host DNA incidentally released from parenchyma or companion cells ruptured during the probing of stylets in search of living phloem vessels. Xylem feeding by Magicicada spp. likely results in the ingestion of similarly small quantities of host DNA released during stylet probing. Second, differences in the timing of feeding behavior relative to collection likely contribute to observed differences in plant DNA recovery. While adults appear to take food regularly as they move across the landscape, it is unknown whether significant feeding occurs in mature nymphs in the weeks prior to their emergence during which they remain just below the surface or build turrets (Williams and Simon 1995). Unfortunately, due to the challenges inherent in working with adult Magicicada, a signal decay curve could not be developed through controlled feeding studies to determine how long ingested plant DNA persists in the insect gut. However, as it is known that ingested plant DNA degrades with time (e.g., Hepler et al. 2021), the small amount of DNA recovered from nymphs could represent temporally distant feeding events. Lastly, some plant DNA may have been lost during the dissection process when gut tissues of some samples were inadvertently ruptured, potentially resulting in lower recovery rates and read counts.

For nymphs, OTUs assembled from individuals collected in wooded habitats were identified as Carya Nutt. sp., Juglans L. sp., Lonicera maackii (Rupr.) Herder, and Trifolium repens L. (for a reference table of scientific and common names, see Table 2). OTUs recovered from nymphs collected in apple orchards were identified as Malus domestica auct. non Borkh and M. sylvestris (L.) Mill. only with no other plant taxa identified. OTUs assembled from adults collected in wooded habitat were identified as Acer negundo L., Cannabis sativa L., Festuca L. sp., Fraxinus pennsylvanica Marshall, Fraxinus L. sp., Malus domestica, Prunus serotina Ehrh., Pyrus calleryana Decne., Pyrus communis L., Quercus rubra L., Quercus L. sp., and Syringa vulgaris L., while those recovered from adults collected from apple orchards were identified as Ailanthus altissima (Mill.) Swingle, Fraxinus sp., Lactuca serriola L., an unknown member of Lauraceae (likely spicebush, Lindera Thunb. sp.), Malus domestica, Nyssa sylvatica Marshall, Populus L. sp., and Quercus sp. While the above species-level identifications are believed to be generally reliable, there is a possibility of cross-matching between congenerics due to similarities in barcoding gene regions. Due to the potential ambiguity this introduces for some taxa (e.g., Carya spp., Malus spp., Fraxinus spp., Pyrus spp., and Quercus spp.), we have elected to report our results at the genus level for all subsequent analyses (Fig. 1).

Table 2.

Scientific and common names of ingested plant taxa identified from the digestive tracts of Magicicada spp

Scientific name	Order: family	Common name
Acer negundo	Sapindales: Aceraceae	Boxelder
Ailanthus altissima	Sapindales: Simaroubaceae	Tree of heaven
Cannabis sativa	Urticales: Cannabaceae	Marijuana
Carya sp.	Juglandales: Juglandaceae	Hickory
Festuca sp.	Cyperales: Poaceae	Fescue
Fraxinus pennsylvanica	Scrophulariales: Oleaceae	Green ash
Fraxinus sp.	Scrophulariales: Oleaceae	Ash
Juglans sp.	Juglandales: Juglandaceae	Walnut
Lactuca serriola	Asterales: Asteraceae	Prickly lettuce
Lindera sp.a	Laurales: Lauraceae	Spicebush
Lonicera maackii	Dipsacales: Caprifoliaceae	Amur honeysuckle
Malus domestica (syn. M. pumila)	Rosales: Rosaceae	Paradise apple
Malus sylvestris	Rosales: Rosaceae	European crab apple
Nyssa sylvatica	Cornales: Cornaceae	Blackgum
Populus sp.	Salicales: Salicaceae	Cottonwood
Prunus serotina	Rosales: Rosaceae	Black cherry
Pyrus calleryana	Rosales: Rosaceae	Callery pear
Pyrus communis	Rosales: Rosaceae	Common pear
Quercus rubra	Fagales: Fagaceae	Northern red oak
Quercus sp.	Fagales: Fagaceae	Oak
Syringa vulgaris	Scrophulariales: Oleaceae	Common lilac
Trifolium repens	Fabales: Fabaceae	White clover

^aProbable identity, though sequence data inconclusive at genus level.

Fig. 1.

Dietary histories of individual Magicicada labeled by lifestage (nymph, adult), habitat (wooded, orchard), species (cassini, septendecim), and sex (male, female). Shaded boxes indicate plant DNA was identified from gut contents.

Nymphs of Magicicada spp. feed on the xylem of host plant roots (White and Strehl 1978, Lloyd and White 1987). Their limited subterranean mobility (Marlatt 1907) constrains them to feed upon the species with roots in the soil directly below their hatching site and likely precludes frequent host switching. It is, therefore, unsurprising that all recovered nymph samples each contained plant DNA from only a single host species. Magicicada nymphs appear to be broadly polyphagous, with feeding reported on fruit orchard trees (Morris 1848, Asquith 1954, Hamilton 1961), broadleaf forest trees (Dybas and Lloyd 1974, White and Strehl 1978), conifers (White et al. 1982, Lloyd and White 1987, Sahli and Ware 2000), and herbaceous taxa (Lloyd and White 1987). The 4 recovered nymphs from wooded habitats contained OTUs identified as 2 common deciduous trees (Carya and Juglans) and 2 understory species (Lonicera and Trifolium). All plant DNA recovered from the 2 nymphs collected from an apple orchard was identified as Malus spp. This evidence of feeding on apple supports historical reports and more recent studies of Magicicada nymphs as orchard pests capable of inducing general decline and reduced growth in fruit trees when present at high densities (Morris 1848, Banta 1960, Hamilton 1961, Hamilton and Cleveland 1964, Karban 1982) and is consistent with a lack of significant dispersive movement through the soil during the nymphal stage.

The absence of recoverable plant DNA in the teneral adult samples strongly suggests that all plant DNA recovered from fully sclerotized adults was ingested during the post-teneral adult stage. Insect digestive tracts are likely mostly voided during the molting process, as the linings of the foregut and hindgut are epidermal in origin and are thus shed (Klowden 2002). This voiding, in combination with a possible lack of recent feeding by pre-emergent nymphs, provides a satisfactory explanation for the lack of recovered plant DNA from teneral adults.

The only OTUs identified in adults recovered from both orchard and wooded sites were Fraxinus, Malus, and Quercus. Taxa detected only in adult M. septendecim included Lactuca, unkn. Lauraceae, Nyssa, and Populus. Acer, Ailanthus, Cannabis, Festuca, Malus, Prunus, Pyrus, and Syringa were recovered only from adult M. cassini. OTUs identified as Quercus and Fraxinus were recovered from adults of both species. No other patterns of host use were detected, partially due to the unbalanced species distribution of samples submitted and recovered from orchard vs. wooded sites (Table 1).

In general, adults collected from both orchard and wooded habitats fed on a broad range of hosts including 13 identifiable genera and 11 families. There is an apparent relationship between this polyphagy and the generalist ovipositional habits (Maier 1982) of Magicicada females. As feeding behavior has been reported to occur during oviposition (Maier 1982) and after male chorusing (presumably nearby) ceases for the day (Dybas and Lloyd 1974), this overlap in host use is to be expected. Indeed, of the plant taxa detected in adult gut contents, all woody genera are also known ovipositional hosts: Acer (Forsythe Jr. 1976, White 1980), Ailanthus (Miller and Crowley 1998), Fraxinus (White 1980, 1981), Malus (Forsythe Jr. 1976, White 1980), Nyssa (Forsythe Jr. 1976), Populus (Forsythe Jr. 1976), Prunus (Forsythe Jr. 1976, White 1980, 1981), Pyrus (Hamilton 1961), Quercus (Forsythe Jr. 1976, White 1980, 1981), and Syringa (Miller and Crowley 1998). Moreover, feeding behavior has been directly observed on Fraxinus (Dybas and Lloyd 1974), Malus (Fig. 2), and Quercus (Quaintance 1902). These 3 genera were detected in individuals collected from both orchard and wooded habitats. No other feeding patterns were detected; observed differences in detected taxa likely reflect differences in the source plant communities. With the exception of Cannabis, Festuca, and Lactuca, all detected plant taxa were deciduous woody plants known to occur in wooded and ornamental settings in the Mid-Atlantic.

Fig. 2.

Adult Magicicada sp. feeding on apple branch during the 2021 emergence of Brood X.

The presence of herbaceous taxa was surprising, as adult feeding on nonwoody hosts is only known from a single brief report of feeding on a weed in the genus Aster (Maier 1982). Festuca was identified from a single adult collected in wooded habitat and likely represents opportunistic feeding on ornamental lawn grass. Lactuca was detected in a single individual collected from an apple orchard and is a common agricultural weed species. The presence of Cannabis, however, was particularly unexpected. The source adult was collected in a residential wooded backyard, and subsequent surveying revealed Cannabis was present ~100 m away. This distance falls well within the reported estimated maximum adult dispersal distance of 300 m (Williams and Simon 1995), though most individuals remain within 50 m of their emergence site (Karban 1981).

Over half (54.5%) of recovered adults contained identifiable DNA from multiple plant taxa, implying frequent movement among hosts and regular feeding on plant sap. Further evidence of frequent dispersal by Magicicada adults is the diversity of taxa detected in individuals collected from apple orchards. Interestingly, Malus DNA was identified from only a single orchard-collected individual (16.7% of recovered orchard samples). This suggests that most Magicicada adults observed in apple orchards may have emerged in adjacent habitat and subsequently moved into the orchard. Indeed, this is the pattern reported in the literature (e.g., USDA 1919, Hamilton and Cleveland 1964, Maier 1982).

While sample sizes were too small to make species-level distinctions in feeding behavior between M. septendecim and M. cassini, overall trends can be observed across the genus. Together with the wide range of ingested plant taxa identified, our data suggest that Magicicada spp. are broadly polyphagous and mobile as adults and regularly ingest plant sap as they move from plant to plant to call, mate, and oviposit. This general polyphagy appears to include opportunistic feeding on herbaceous taxa encountered in the landscape.

Although the gut content data presented here confirm that adult Magicicada of both sexes do ingest plant sap, it does not necessarily follow that the purpose is nutritive. Cicadas as a group appear to have highly permeable cuticles that permit considerable water loss (Toolson 1987, Toolson and Toolson 1991). In desert species, this water loss is leveraged for thermoregulation though evaporative cooling; passive evaporation is even augmented by the active extrusion of water through cuticular ducts distributed across the dorsal abdomen and mesothorax (Toolson 1987, Sanborn 2002). These high water loss rates in a hot, dry environment are sustained through regular ingestion of plant sap (Toolson 1987). While the warm and humid climate typically encountered by Magicicada spp. reduces the need for and efficacy of evaporative cooling as a thermoregulatory mechanism (Toolson and Toolson 1991), both passive and active water loss may still be important factors. An investigation of the thermobiology of M. tredecim revealed a highly permeable cuticle facilitating passive water loss rates able to reduce body temperature below ambient through evaporative cooling. Additionally, though fewer in number and more localized in distribution than in desert species, the cuticular ducts in M. tredecim may be placed such that the heart is preferentially cooled when water is extruded during “thermal emergencies” (Toolson and Toolson 1991). Regardless of its role in thermoregulation, water lost across the cuticle certainly must be replaced through feeding or quasi-feeding behaviors at some point during the 2–6-week lifespan of adult Magicicada (Williams and Simon 1995), as has been suggested (Maier 1982).

A nutritive purpose for the ingestion of plant sap can perhaps be inferred by the presence of endosymbiotic bacteria in the gut tissues of Magicicada spp. Cicadas ingest the xylem fluid of roots, stems, and branches, a nutritionally poor resource providing carbohydrates such as sucrose but lacking nitrogenous compounds. Accordingly, cicadas have coevolved with the obligate bacteriome-dwelling copartner endosymbionts, Sulcia muelleri and Hodgkinia cicadicola, that provide their hosts with amino acids and vitamins essential for survival and development (McCutcheon et al. 2009). Microbiome analysis of adult Magicicada spp. identified both Sulcia and Hodgkinia as core members of the gut microbiome, suggesting these bacteria may play additional roles in host metabolism (Brumfield et al. 2022) and implying that some metabolization of ingested plant sap for energy may be occurring. While the fat bodies accumulated during the nymphal stage are clearly of primary importance for adult maturation and energy (Hargitt 1923), it seems possible or even likely that adult feeding may supply supplemental nutrients to prolong the brief adult life of Magicicada spp.

Overall, our gut content data corroborate observations of feeding behavior by adult Magicicada spp. and confirm that plant sap is ingested by the adult insect. The high rates of water loss across the adult cuticle strongly suggest that water intake in some form is required on a regular basis. While replacing lost water is clearly important, the presence of essential amino-acid- producing endosymbionts in the gut tissues of adult Magicicada imply that some nutrition is also being derived. We conclude that the ingestion of xylem fluids by Magicicada adults is driven by a combination of water replacement and the acquisition of resources to support the obligate gut microbiome and prolong the brief adult stage of the insect, and thus constitutes true feeding behavior.

Acknowledgments

We thank Alyssa Kloos and Joseph Wirts for their technical assistance, and we thank Kyle Brumfield for his helpful suggestions in the preparation of this manuscript.

Funding

This work was funded by USDA-ARS-CRIS Project No. 8080-21000-032-000D.

Author Contributions

James Hepler (Data curation [Equal], Formal analysis [Equal], Investigation [Lead], Methodology [Equal], Writing—original draft [Lead], Writing—review & editing [Equal]), William Cooper (Conceptualization [Equal], Methodology [Equal], Writing—review & editing [Equal]), John P. Cullum (Investigation [Supporting], Resources [Equal]), Chris Dardick (Writing—review & editing [Equal]), Liam Dardick (Investigation [Equal], Resources [Equal]), Laura Nixon (Conceptualization [Equal], Methodology [Equal], Resources [Equal]), Derek Pouchnik (Data curation [Equal], Formal analysis [Equal], Investigation [Equal], Writing—original draft [Supporting]), Michael Raupp (Investigation [Equal], Resources [Equal], Writing—review & editing [Equal]), Paula Shrewsbury (Investigation [Equal], Resources [Equal], Writing—review & editing [Equal]), and Tracy Leskey (Conceptualization [Equal], Funding acquisition [Equal], Methodology [Equal], Project administration [Lead], Supervision [Equal], Writing—original draft [Equal], Writing—review & editing [Equal])