PosterPDF Available

Facultative ant mutualism in the rare hops azure butterfly, Celastrina humulus (Lycaenidae)



Hops azure (Celastrina humulus) larvae feed on male inflorescences of hops. Like other lycaenids, hops azure larvae can form a mutualism with several species of ants. Ants protect larvae from natural enemies in exchange for larval secretions. However, field surveys reveal that roughly one-third of hops azure larvae are not tended in this manner. Our objective was to assess the factors associated with ant tending of this species. We found that the likelihood of ant tending increased with (1) the density of larvae on the bine, (2) the area of the hops patch, and (3) the phenological stage of the flowers.
Figure 3 Male hops inflorescences at varying phenological stages; at stage 1, buds have
elongated into oval, but anthers remain inside of sepals; by stage 6, anthers have begun to
Figure 1 Female hops azure laying eggs on an early-stage male inflorescence (A),
panicle of male flowers of hops (B), and an ant tending a larvae while it feeds on male
flowers (C); in panel C, flowers show characteristic herbivory of pollen inside sepals.
Facultative ant mutualism in the rare hops azure butterfly,
Celastrina humulus (Lycaenidae)
Emily Mooney1, Abbey Swift1, Hailee Nolan2, Robert Schorr3,
1University of Colorado Colorado Springs, 2Colorado State University, 3Colorado Natural Heritage Program
Introduction to the Study System
Conservation of rare species relies on understanding how biotic and abiotic factors
shape abundance [1]. The hops azure butterfly (Celastrina humulus, Family
Lycaenidae) is ranked as a G2 Imperiled species by NatureServe [2]. Butterflies lay
eggs on male flowers of native hops, Humulus lupulus var. neomexicanus (Fig. 1A).
Larvae develop quickly feeding on the pollen-rich male inflorescences (Fig. 1B). Like
other lycaenids, hops azure larvae can form associations with several species of
ants [3], which walk over the caterpillars and drink nectar from specialized organs
(Fig. 1C). Kubik and Schorr (2018) found hops azure larvae tended by ants in 10 of
15 sites along Monument Creek near Colorado Springs, CO [3]. Given the well
documented benefits of mutualism with ants [4], it is unclear why many larvae do
not form associations with ants. Our objective was to assess which biotic and
abiotic factors were associated with ant tending.
Photo by A. Swift
Methods In the Field
We surveyed larvae on 134 unique bines (flowering vines) along a 1.5 km section of Monument Creek in Colorado Springs, CO. Once larvae were found, we used
replicate surveys to record the number of larvae on the bine, larval developmental stage, ant attendance, ant species, and floral phenology (Fig. 2). We ranked
phenology of male flowers using a numerical score (Fig 3). To account for variation among host plants and their environments, we measured light (PAR) above each
bine using a meter and patch area. We surveyed bines within the same host plant population in 2020 and 2021.
Ordinal date
Larval stage
Larvae per bine
Ant tended?
Ant species
Phenological stage
Patch area
We used a model selection approach to determine which of the
factors we measured best explained variation in likelihood of ant
tending (Y/N). We fit mixed effects models with binomially distributed
errors using the ‘glmer()’ in the package lmerTest [5]. The model set
included univariate, additive, and 2-way interactive models (Table 1).
After initial analysis failed to support models that included an effect of
larval stage, we did not include this factor in the final model set. We
obtained AICc and ΔAIC scores from the AICcmodavg package [6]. All
analyses were performed in R version 4.1.1 [7].
Model Selection and Results
Table 1 Models and fit statistics
Model Weight ΔAICc
Ants? ~ Larvae + Flowerstage + Area + (1 | Year) 0.567 0
Ants? ~ Larvae + Flowerstage + Area+ Light + (1 | Year) 0.752 0.590
Ants? ~ Larvae + Area + Light + (1 | Year) 0.884 1.644
Ants? ~ Larvae + Flowerstage * Light + Area + (1 | Year) 0.946 2.705
Ants? ~ Larvae + Flowerstage * Area + (1 | Year) 0.987 3.949
Ants? ~ Larvae + Flowerstage * Area + Light + (1 | Year) 0.998 3.965
Ants? ~ Larvae + Area * Light + (1 | Year) 0.999 4.439
Ants? ~ Larvae + Flowerstage + (1 | Year) 14.923
Ants? ~ Larvae + (1 | Year) 18.849
What’s Associated with Ant Tending?
Figure 4 Greater densities of
larvae per bine was associated
with increased likelihood of ant
tending. More larvae may
simply attract more ants. Such
density dependence can be
common in ant mutualisms [6]
Figure 5 Larvae on bines with
later stage flowers were more
likely to be ant tended.
Flowering phenology can
mediate associations between
herbivores and ants [9]. Nectar
from larvae feeding on later
stage flowers may be more
attractive to ants [10].
Figure 6 Larvae on bines in
larger patches had increased
likelihood of ant tending. Patch
area can be a signal of site
quality perhaps due to greater
nitrogen, a key factor in these
mutualisms [11].
L.R. χ2= 7.632, P= 0.006
L.R. χ2= 5.450, P= 0.005
L.R. χ2= 9.275, P= 0.010
1. Wagner DL, Grames EM, Forister ML, Berenbaum MR, Stopak D (2021) Insect decline in the Anthropocene: Death by a thousand cuts. Proceedings of the
National Academy of Sciences, 118(2): e2023989118
2. Celastrina humulus | NatureServe Explorer.
3. Kubik TD, Schorr RA (2018) Facultative Myrmecophily (Hymenoptera: Formicidae) in the Hops Blue Butterfly, Celastrina humulus (Lepidoptera:
Lycaenidae). Entomological News, 127(5):490498.
4. Ness J, Mooney K, Lach L (2010) Ants As Mutualists. Ant Ecology, 9780199544:432.
5. Kuznetsova A, Brockhoff PB, Christensen RHB (2017) lmerTest Package: Tests in Linear Mixed Effects Models. Journal of Statistical Software, 82(13):126.
6. Mazerolle, MJ (2020) AICcmodavg: Model selection and multimodel inference based on (Q)AIC(c). R package version 2.3-1. https://cran.r-
7. R Foundation for Statistical Computing (2021) R: A language and environment for statistical computing. :R version 4.1.1 (2021-08-10)-- “Kick Things.”
8. Peterson MA (1995) Unpredictability in the facultative association between larvae of Euphilotes enoptes (Lepidoptera: Lycaenidae) and ants. Biological
Journal of the Linnean Society, 55(3): 209223.
9. Patricelli D, Barbero F, Morgia V La, Casacci L P, Witek M, Balletto E, Bonelli S (2011) To lay or not to lay: Oviposition of Maculinea arion in relation to
Myrmica ant presence and host plant phenology. Animal Behaviour, 82(4):791799.
10. Pierce NE, Nash DR, Baylis M, Carper ER (1991) Variation in the attractiveness of Iycaenid butterfly larvae to ants. Ant-Plant Interactions, :131142.
11. Billick I, Brown R, Reithel JS (2005) Importance of fertilization of host plants to ant tending and growth rates in Glaucopysche lygdamus (Lepidoptera:
Lycaenidae). Annals of the Entomological Society of America, 98(4):491495.
This project received support from the Colorado Native Plants Society Marr Grant. Funding to A. Swift was provided by
the UCCS Undergraduate Research Academy (2020) and the Department of Biology (2021). Funding to H. Nolan was
provided by the Linda Hamilton Conserving Colorado Award at CSU Fort Collins. We also thank C. Hamilton, E.
Callahan, and L. Callahan for additional help with field surveys.
Stage 1 Stage 6Stage 2 Stage 4
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
The choices made by ovipositing female butterflies play a key role in the survival of their offspring and consequently in the persistence of butterfly populations. These choices are even more crucial in the case of obligate myrmecophilous organisms such as Maculinea butterflies with larvae that, after a phytophagous period, need to be adopted by Myrmica ants to complete their life cycle. Because the worker ants' foraging range is limited, selecting an 'ideal' oviposition site requires that both the phenological stage of the larval food plant (short-term larval fitness) and the presence of suitable host ants (long-term larval fitness) are taken into account. Whether the female's selection of a valuable oviposition plant is influenced by the closeness of a Myrmica nest is unclear. We studied the oviposition behaviour of a Maculinea arion population exploiting Origanum vulgare as a host plant. By following females, we collected phenological data on the visited plants that were either 'chosen' for oviposition or 'avoided' (flowers were visited and evaluated, but received no eggs), and we assessed the presence of Myrmica ants in the vicinity of each plant. Results suggest that plants are selected by M. arion females on the basis of their bud phenology and the presence of host ants and not of other environmental features. We thus hypothesize the evolution of an adaptive mechanism that affords females of this strictly myrmecophilous butterfly the ability to ensure the long-term survival of their brood by selecting host plants growing near a Myrmica nest.
The hops blue butterfly (Celastrina humulus) is a rare North American butterfly, found in the Rocky Mountain region of the United States in areas where its host plant, wild hops (Humulus lupulus), grows in abundance. As a member of the Lycaenidae we considered it likely that myrmecophiles may exist with ants found in C. humulus habitats. To better understand the ecology of this species and guide conservation efforts, we investigated and documented myrmecophily with multiple ant species. We found seven ant species with close associations with C. humulus larvae.
Environmental conditions such as nutrient availability may modify the strength and nature of interactions between mutualists. We examined the extent to which variability in nitrogen modified the interaction between ants and larvae of Glaucopsyche lygdamus Doubleday. Whereas nitrogen addition had no effect on larval growth rates, the number of larvae, or the number of ants tending larvae 2 wk after the beginning of the experiment, by 4 wk nitrogen addition had increased tending rates from 2.5 to 4.1 ants per larva. Despite the effect of nitrogen on tending rates of larvae, there was no effect of nitrogen on host plant growth, herbivory, or seed production. Unlike a previous study of this population of butterflies on a different host plant species, there was no indication that larvae feeding on pods attracted larger ant guards. Thus, although the amount of nitrogen available to plants did have a detectable effect on the ant–lycaenid mutualism, there was no evidence that plant substrate played a strong role in mediating tending rates. Because our finding that nitrogen addition affected tending rates parallels findings for a very different ant–lycaenid system, we suggest that nitrogen availability to host plants may be of general importance to ant–lyceanid interactions.
Larvae of the lycaenid butterfly, Euphilotes enoptes, are facultatively associated with ants. To determine whether variation in this association is predictable, I documented the proportion of larvae tended by ants and the identity of those ants over a three-year period in two study areas in central Washington state. Each study area contained numerous patches of Eriogonum compositum, the sole host plant of E. enoptes in this region. Using complete censuses and random subsampling, I estimated the dispersion and densities of E. enoptes larvae occupying each patch. The results of these observations indicated significant spatial and temporal variation in the identities of ants tending larvae and the frequency of ant attendance. In addition, early instar larvae were tended less frequently and by smaller ant species, than fourth instar larvae. Although larvae were aggregated at two spatial scales: individual plants within a patch and individual inflorescences on a plant, attendance rates did not vary consistently with aggregation size at either scale. Similarly, attendance rates were independent of the patch-wide density of larvae. I discuss the potential sources of such unpredictable variation and its implications for the evolution of further specialization in this facultative association.
  • J Ness
  • K Mooney
  • L Lach
Ness J, Mooney K, Lach L (2010) Ants As Mutualists. Ant Ecology, 9780199544:432.
Package: Tests in Linear Mixed Effects Models
  • A Kuznetsova
  • P B Brockhoff
  • Rhb Christensen
Kuznetsova A, Brockhoff PB, Christensen RHB (2017) lmerTest Package: Tests in Linear Mixed Effects Models. Journal of Statistical Software, 82(13):1-26.
Variation in the attractiveness of Iycaenid butterfly larvae to ants
  • N E Pierce
  • D R Nash
  • M Baylis
  • E R Carper
Pierce NE, Nash DR, Baylis M, Carper ER (1991) Variation in the attractiveness of Iycaenid butterfly larvae to ants. Ant-Plant Interactions, :131-142.