Acta zoológica lilloana 63 (1): 1–13, 7 de junio de 2019
Mariana E. Carro
, Mariela V. Lacoretz
, M. Cecilia de rsico
Gustavo J. Fernández
Grupo de Investigación en Ecofisiología de Fauna Silvestre (GIEFAS), CCT Patagonia Norte-CO-
NICET, Asentamiento Universitario San Martín de los Andes (AUSMA), Universidad Nacional del
Comahue, Pje de la Paz 235, (8370) San Martín de los Andes, Neuquén, Argentina.
Departamento de Ecología, Genética y Evolución-IEGEBA CONICET, Facultad de Ciencias Exactas
y Naturales. Universidad de Buenos Aires, Ciudad Universitaria, (1428) Ciudad Autónoma de
Buenos Aires, Argentina.
* Email:
Especificidad de la planta hospedadora y preferencia alimentaria
de larvas de Morpho epistrophus argentinus (Lepidoptera:
Nymphalidae) en bosques costeros de Argentina
ä Ref. bibliográfica: Carro, M. E.; Lacoretz, M. V.; de Mársico, C.; Fernández, G. J. 2019. “Host plant specific-
ity and feeding preferences of Morpho epistrophus argentinus (Lepidoptera: Nymphalidae) larvae in coastal
woodlands of Argentina”. Acta zoogica lilloana 63 (1): 1-13. Fundación Miguel Lillo, Tucumán, Argentina.
ä Recibido: 15/05/19 Aceptado: 31/05/19
ä URL de la revista:
ä Algunos derechos reservados. Esta obra está bajo una Licencia Creative
Commons Atribución No Comercial Sin Obra Derivada 4.0 Interna-
The selection of suitable host plants for larval development is critical for most
herbivorous insects. In this study, we examined host plant preferences of larvae from
Epistrophus white morpho butterfly, Morpho epistrophus argentinus (H. Fruhstorfer),
a specialized butterfly that oviposits on a few plant species. In many species, the
selection of host plant for larvae development is under the control of ovipositing
females and the role of larvae in host selection is minimal. At field, we characterized
larval host plant use in coastal woodland patches of its southernmost distribution
range (36º 25’ S, 56º 57’ W; Argentina). All sampled M.e. argentinus larvae groups
(n = 40) were found on Scutia buxifolia trees, although the number of larval groups
did not correlate with cover of this plant at patch scale. Translocation experiments
showed that M.e. argentinus larvae translocated to unsuitable host plants had a shorter
residence time than those translocated to suitable (i.e. control) hosts. Choice assays
Host plant specificity and feeding
preferences of Morpho epistrophus
argentinus (Lepidoptera: Nymphalidae)
larvae in coastal woodlands of Argentina
M. E. Carro et al.: Host plant and feeding preferences of Morpho epistrophus argentinus
In plant feeding insects, the selection of suitable plants for neonate larvae devel-
opment is critical as it affects their growth rates and ultimately, their fitness. It has
been proposed that females of these species should choose hosts where larval perfor-
mance is optimal («mother knows besprinciple; Thompson, 1988). Accordingly,
in many species, the selection of host plant species for larvae development is under
the control of ovipositing females (Wiklund, 1975; Futuyma, 1990; Thompson and
Pellmyr, 1991; Nylin and Janz, 1993; Price, 2003; Gripenberg et al., 2010). In such
species, the role of larvae in host selection is minimal and usually constrained by the
difficulty of larvae to move across plants (Prager et al., 2014). In the other extreme,
ovipositing females of some species play a minor role in host plant selection, and
made also indicated strong feeding preferences of wild larvae for S. buxifolia leaves.
The results highlight the high degree of host specificity of M.e. argentinus by S.
buxifolia, and reveal that growing larvae preference for host plant matches that of
ovipositing females.
Palabras clave Butterfly, host plant selection, larval behavior, larval preferences.
La selección de plantas hospedadoras adecuadas para el desarrollo larvario es
crítica para la mayoa de los insectos herbívoros. En este estudio, examinamos las
preferencias de la planta hospedadora de larvas de la mariposa Bandera Argenti-
na, Morpho epistrophus argentinus (H. Fruhstorfer), una mariposa especializada que
ovipone en pocas especies de plantas. En muchas especies, la selección de plantas
hospedadoras para el desarrollo de las larvas está bajo el control de las hembras y
el papel de las larvas en la selección de hospedadores es mínimo. En el campo, ca-
racterizamos el uso de la planta hospedadora por parte de las larvas en parcelas de
bosques costeros de su área de distribución más meridional (36º 25’ S, 56º 57’ W;
Argentina). Todos los grupos de larvas de M.e. argentinus muestreados (n = 40) fueron
encontrados en árboles de la especie Scutia buxifolia, aunque elmero de grupos de
larvas no se correlacionó con la cobertura de esta planta a escala de parche. Experi-
mentos de translocación demostraron que las larvas de M.e. argentinus translocadas
a plantas hospedadoras inadecuadas tuvieron un tiempo de residencia más corto
que las translocadas a hospedadores adecuados (i.e. controles). Ensayos de elección
realizados tambn indicaron fuertes preferencias de alimentación de las larvas por
hojas de S. buxifolia. Los resultados resaltan el alto grado de especificidad de M.e.
argentinus por el hospedador S. buxifolia y revelan que la preferencia de las larvas por
la planta hospedadora coincide con la de las hembras cuando oviponen.
Keywords Mariposa, selección de planta hospedadora, comportamiento de las larvas, pre-
ferencias de la larva.
Acta zoológica lilloana 63 (1): 1–13, 7 de junio de 2019
larvae must be able to find suitable hosts for their development (Chew, 1977; Bernays
and Chapman, 1994; Berdegué et al., 1998, Crespo and Castelo, 2008).
Many butterfly species have a wider range of suitable host plants for larval de-
velopment than that they actually use for oviposition (Thompson and Pellmyr, 1991).
This reflects mother’s preferences and it would be the result of females maximizing
their own performance rather than seeking to improve offspring development and
survival with their choices (Mayhew, 1997; Prager et al., 2014). In this scenario,
larvae may be restricted in host plant use by its mother’s preference, and their per-
formance will depend on selection of high-quality hosts by its mother. However,
larvae may not be entirely dependent on female’s decisions. In some butterfly spe-
cies, oviposition is not selective with respect to host plant, and larvae must to locate
the suitable host (Wiklund, 1984). Then, larval host selection can act as a reinforc-
ing (or correcting) mechanism of maternal host choice, through which immature
offspring can circumvent eventual female’s oviposition errors (Gamberalle-Stille
et al., 2014). In fact, it has been found in the comma butterfly (Polygonia c-album
Linnaeus) that larvae can show similar host preference than ovipositing females
(Nylin and Janz, 1996; Nylin, Bergström and Janz, 2000; Gamberalle-Stille et al.,
2014). Therefore, host-plant choice is not necessarily restricted to female or larvae
decisions, and can imply hierarchical mechanisms where both female and larvae
decisions are involved.
In this study, we examined preferences of the butterfly larvae Morpho epistrophus
argentinus (H. Fruhstorfer) for host plant, in coastal woodlands of central-east Ar-
gentina. This conspicuous and specialized Morpho butterfly oviposits on a few plant
species (Ackery, 1988; Canals, 2000; Beccaloni et al., 2008). The larvae feed on plant
where they hatch, but Orlandi (2016) found that larvae of Morpho epistrophus can be
feed with other plant species than those where they were collected. At our study site
in the southernmost distribution range of M.e. argentinus, females laid eggs on only
one tree species, Scutia buxifolia (Rosales:Rhamnaceae). However, female choice of
this perennial tree species may be restricted because the offspring that overwinter
in the stage, need protection and shelter, and therefore, may not reflect the larval
preferences. In that sense, S. buxifolia is the most abundant perennial native species
in the southernmost distribution of the butterfly. However, in the last decades, native
woodlands at this range have been severely reduced which threaten the biota that
depend on this habitat type (Goya et al., 1992; Arturi and Goya, 2004). Therefore,
considering that M.e. argentinus is at risk due to habitat loss, determine the ability
of larvae to use alternative host-plant species is important for conservation purposes.
Then, our objectives were to characterize plant use by larvae and to assess their host
plant preferences using both a translocation experiment and feeding choice assays.
We expected that if larvae preferences match those of ovipositing females, they must
to prefer to stay on S. buxifolia trees and feed on their leaves avoiding unusual host
plants as substrate and food.
M. E. Carro et al.: Host plant and feeding preferences of Morpho epistrophus argentinus
Species studied
Morpho epistrophus argentinus is a big and conspicuous nymphalid butterfly
(wingspan: 90–110 mm) that inhabits riparian and xerophytic forests along North
and East Argentina. The conservation status of this species has not been evaluated
yet, but at their southernmost distribution range it is considered as threatened due
to the degradation and loss of suitable habitat (Nuñez-Bustos, 2009). At this range,
adults are seen flying between late January and March (Gallardo, 1908; Nosswitz,
1927). The females lay small eggs (~2 mm diameter) under the leaves of specific
plant species such as S. buxifolia, Lonchocarpus nitidus (Fabales: Fabaceae) and Inga
vera (Fabales: Fabaceae) (Gallardo, 1907; Canals, 2000; Nuñez Bustos, 2009). At
the southernmost distribution range of M. e. argentinus, only S. buxifolia is present.
Larvae emerge after about two weeks, remaining on the host plant (Bourquín, 1944).
They do not eat and stay on diapauses gregariously over winter months (Gallardo,
1908; Otero and Marigo, 1990). During September, they resume feeding activity.
After first molt, the larvae exhibit an intense reddish color that presumably makes
them highly aposematic (García et al., 2014). They rest gregariously during daylight
hours, forming clusters of up to 30 individuals. Feeding occurs mostly during night
when the cluster breaks up (Turner, 1936). Larvae reach a size of 55-70 mm at the
final instars (Bourquín, 1944). Early in summer (December-January), the larvae
migrate from their host plants and pupate.
Field methods
From October to early December 2015 (austral spring), we searched for larvae
clusters at woodland patches at General Lavalle (36º 25’S, 56º 57’ W), Buenos Aires
province, Argentina. The site comprises patches of deciduous woodland of Celtis
ehrenbergiana and S. buxifolia, with a mean vegetation height < 5m, surrounded by
pastures (Lacoretz, 2018).
We quantified host-plant specificity through field observations and censuses.
We searched for larvae clusters at 10 randomly selected woodland patches in which
the preferred host plant (S. buxifolia) of M.e. argentinus larvae was absent and at other
10 patches where the host plant was present. We surveyed the patches using a 4 x 40
m strip transect, walking through them and carefully searching for M.e. argentinus
larvae. Once localized, we identified the host plant species and counted the number
of larvae in the cluster. For each larvae cluster, we measured the body length (nearest
2 mm) from four arbitrarily selected larvae. Body length was used to estimate the
instar to which each group belonged following Orlandin (2016).
We also estimated species cover in each of the 20 woodland patches using eight
randomly located sampling points within a 20m radius area from the starting point
of the strip transect. At each point, a sampling rod (7 m height) marked off in 1m
intervals was held vertically on the ground and the plant species contacting the rod
Acta zoológica lilloana 63 (1): 1–13, 7 de junio de 2019
were recorded. Cover for each plant species was estimated as the percentage of sam-
pling points at each woodland patch where the species was present (i.e. contacted
the rod).
In one woodland patch, we also searched for M.e. argentinus larvae within a
plot of 50 x 50 m, and when localized, we marked the location of the cluster with
a colored and numbered tape for cluster identification. These larval groups (n =
12) were revisited after 30 days to note the permanence in the site where they were
first found.
Host plant specificity experiments
To determine host specificity of M.e. argentinus under field conditions, we per-
formed experimental translocations in groups of larvae during late November. Twen-
ty larvae groups were removed from host plants where they were found, and moved
to trees located more than 30 m apart. Ten randomly selected of these groups were
placed on branches of S. buxifolia (its natural host plant), and the other 10 groups
on branches of C. ehrenbergiana. The latter is the dominant deciduous tree in the
woodlands and thus broadly available for consumption by larvae. Larval size of these
groups ranged from 35 to 45 mm (presumably corresponding to 3
instars) and
group size was 8–19 individuals. We monitored these groups every 3–4 days up to
17 days after the beginning of the experiment. On each visit, we recorded the num-
ber of larvae present at each group and inspected neighbor trees to detect missing
individuals or groups.
Feeding preferences
We also tested the feeding preferences of M.e. argentinus larvae using a choice
assay at our field station. We collected 20 individuals (40–52 mm in length; pre-
sumably 4
instar) from 5 larvae clusters (4 individuals/group) at evening hours
(18:00–20:00 h) during November. Each larva was placed individually on a plastic
box (approximately 15 x 25 x 10 cm) with the bottom covered with a filter paper
slightly moistened with a few drops of water. Boxes were closed with perforated lids
and kept in the shade under natural photoperiods and environmental temperatures.
After collection, caterpillars were deprived of food for about 12 h. Then we offered
leaves of the two most common tree species at our study area to individual larva.
Five 1x1 cm pieces of fresh S. buxifolia leaves and 5 of C. ehrenbergiana leaves were
arranged alternately along the edges of the boxes. Trials started at 8:00 hs. After 12
h, we collected the remaining leaf pieces from each box and estimated the individual
consumption of each species as the number of quarters of eaten leaf pieces. Then,
the assay was replicated to estimate nocturnal consumption. After the 24 h trial, 10
of these larvae were returned to their original clusters in the field and the remaining
10 individuals were subjected to a second experiment. At this, the larvae were kept
in boxes and were induced to feed only on green leaves of C. ehrenbergiana during 3
M. E. Carro et al.: Host plant and feeding preferences of Morpho epistrophus argentinus
consecutive days. After this period, a choice experiment as described above was con-
ducted during night hours. After testing, the larvae were returned to their original
field locations. No larval mortality was recorded during the experiments.
Data analyses
We used descriptive statistics to quantify host plant specificity and the charac-
teristics of plant hosts. The occurrence of M.e. argentinus larvae in woodland patches
where S. buxifolia plants was present and absent was compared using a Fisher’s ex-
act test. Relationship between the number and size of larvae and the size of group
were assessed using Spearman rank correlations. We used a Cox proportional hazard
mixed regression model to analyze the results of the translocation experiment. The
model included the residence time, defined as the number of visits in which larvae
were seen at the translocation site, as the response variable, host plant treatment as a
fixed factor and larvae group identity as a random variable. Analysis was performed
using the COXME package, v2.7.1 (Therneau, 2015) in the R environment (v3.3.0,
R Core Team, 2016). Feeding preferences of M.e. argentinus larvae were tested using
generalized linear mixed models (GLMM) with a Poisson structure and a log-link
function. We performed these analyses using the lme4 package, v1.1-17 (Bates et al.,
2015) for R. First, we tested for differences in leaf consumption of each tree species
by individuals recently collected from the field. For this model, the response variable
was the number of leaf quarters consumed, and time of day (day or night), plant
species and the interaction term as predictors. Individual identity was included as
a random variable. In a second analysis, we compared the feeding preferences of
individuals before and after being induced to feed on C. ehrenbergiana. The model
included sampling time (before or after induction) and plant species as predictors,
and larvae identity as a random factor. The interaction was not assessed because
no larval feeding was detected on C. ehrenbergiana before the induction treatment.
For all these models, we used residual and normal probability plots to check model
assumptions. We assessed the global contribution of predictors to the response vari-
ables by comparing the deviance of nested models (i.e., with and without the factor)
using likelihood-ratio tests.
We report means with ± SE for all data (unless otherwise specified). All tests
are two-tailed, and differences were considered significant at P < 0.05.
The occurrence of M.e. argentinus larvae was strongly related to the presence of
S. buxifolia plants in the woodland patches (Fisher’s exact test, P < 0.01). Morpho
epistrophus argentinus larvae were recorded in seven out of 10 woodland patches with
S. buxifolia, but in none of 10 patches where S. buxifolia was absent. Sampled wood-
land patches were mostly dominated by C. ehrenbergiana (mean cover percentage:
58.7 %; SD: 30.1 %). In woodland patches with presence of S. buxifolia, its mean
Acta zoológica lilloana 63 (1): 1–13, 7 de junio de 2019
cover was 56.7 % (SD: 22.1 %) and was co-dominant with C. ehrenbergiana (mean
cover percentage of this last species: 42.5 %; SD: 27.8). Other tree species (e.g.,
Schinus longifolius, Jodina rhombifolia) had a vegetation cover of less than 5 % at the
studied patches.
We found a total of 40 larvae groups, all located on branches of S. buxifolia.
Larval size of groups recorded varied between 15 and 34 mm, corresponding to 1
instars. Larvae size was not related to group size (Spearman rank correlation, R
= 0.07, P = 0.79), and the number of larval groups was not related to S. buxifolia
cover at patch scale (Spearman rank correlation, R = 0.19, P = 0.59). Furthermore,
surveyed larvae groups at the 50 x 50 m plot (n= 12) remained in the same branches
where first observed after 30 days.
Host plant specificity experiments
From the 10 larvae groups translocated to C. ehrenbergiana, only one remained
on this host plant after 17 days. Three other larvae groups migrated to neighbor S.
buxifolia trees and the remaining six were missing. The C. ehrenbergiana plant which
the larvae group remained showed slight signs of herbivory on its leaves. In contrast,
nine of the larvae groups translocated to S. buxifolia branches remained in their new
location. Only one group was missing and it was not observed in the neighboring
trees. As expected, residence time was significantly lower for individuals translocated
to C. ehrenbergiana than to S. buxifolia trees (Cox regression model:
= 16.77, P
< 0.01; Fig. 2).
Figure 1 Frequency of group sizes of M. e. argentinus larvae (n = 40). Larval body size varied from
15 to 35 mm in length (1
M. E. Carro et al.: Host plant and feeding preferences of Morpho epistrophus argentinus
Feeding preferences
The choice experiment showed a marked preference of M.e. argentinus for feed-
ing upon S. buxifolia leaves (
= 305.47, P < 0.01; Fig. 3a). Larval feeding activity
was highest during the night (
= 19.20, P < 0.01; Fig. 3a), but preferences did
not change with the time of day (interaction term:
= 0.17, P = 0.68). Although
the larvae increased slightly the consumption of C. ehrenbergiana leaves after being
induced to feed on this host plant (Fig. 3b), they fed preferentially on S. buxifolia
leaves both before and after that treatment (plant species:
= 208.31, P < 0.01,
sampling time:
= 0.28, P = 0.59; Fig. 3b).
It is widely accepted that herbivorous insects should prefer host plant spe-
cies that will maximize their fitness. This concept can be expanded to include the
selection of oviposition sites, as females should choose those host plants that are
optimal for the development of their offspring (Thompson, 1988; Valladares and
Lawton, 1991; Scheirs et al., 2000; Gripenberg et al., 2010). Morpho epistrophus ar-
gentinus oviposits on a few plant species (Ackery, 1988; Canals, 2000; Beccaloni et
al., 2008) and, in its southernmost distribution range, it uses exclusively S. buxifolia
for oviposition. Consistent with this, all larvae groups in this study were found
attached to branches of S. buxifolia, in woodland patches that had this tree as a co-
dominant species. Conversely, we could not find M. e. argentinus larvae in patches
were S. buxifolia was absent. These observations suggest a high degree of specializa-
Figure 2 Proportion of M.e. argentinus larvae that remained in clusters translocated to C. ehren-
bergiana (CE; n = 10; white boxes) and S. buxifolia trees (SB; n = 10; sptriped boxes) along the
experiment. Median (dots), interquartil range (boxes) and minimum and maximum values (whiskers)
values are shown.