You are on page 1of 10

Entomologia Experimentalis et Applicata 96: 289–298, 2000.

© 2000 Kluwer Academic Publishers. Printed in the Netherlands.


289

Butterfly nectaring flowers: butterfly morphology and flower form

Sarah A. Corbet
Department of Zoology, Downing Street, Cambridge CB2 3EJ, UK
Present address: I St Loy, St Buryan, Penzance TR19 6DH, UK

Accepted: May 23, 2000

Key words: tongue, proboscis, mass, wing loading, nectar, corolla depth, inflorescence, foraging, selectivity

Abstract
The profitability of butterfly foraging depends in part on the corolla depth and clustering of flowers, and the tongue
length, body mass and wing loading of butterflies. Interactions among these attributes of flowers and butterflies
were investigated, using data from a field study in Cornwall and from Porter et al. (1992). The maximum corolla
depth from which a butterfly can feed depends on tongue length, which correlates with the more easily measured
attributes of body mass and wing loading. Small, short-tongued butterflies did not visit deep flowers. The quantity
of nectar sugar per flower necessary for profitable foraging depends on foraging costs, which are expected to
correlate with wing loading. Butterfly species with a high wing loading generally confined their visits to flowers that
were clustered or very nectar-rich. Butterfly species with a low wing loading included solitary and less nectar-rich
flowers in their diet. Body mass and wing loading affect a butterfly’s load-carrying capacity (limiting the distance
between fuelling stops) and cooling rate (limiting the distance between stops for basking or endothermic warming),
and will therefore influence the capacity for floral selectivity and for migration and dispersal. Body mass, wing
loading and tongue length characterised families or subfamilies of butterflies. For example vanessine nymphalids,
with their long tongues and high wing loading, visited the deep, massed flowers of Buddleja davidii, but lycaenids,
with their short tongues and low wing loading, did not. These often visited members of the Asteraceae. Eupatorium
cannabinum, with massed flowers offering abundant and accessible nectar, was visited by butterflies of all tongue
lengths and both high and low wing loading. These findings may help to inform habitat management for butterfly
nectaring flowers.

Introduction on the energetic cost of flight and other foraging activ-


ities and the proportion of foraging time allocated to
Butterfly species differ in the species of flowers that each activity. In general, an insect makes a foraging
they visit (Feber & Smith, 1995). The causal basis profit if:
of these differences is poorly understood. Insofaras (energy content of mean standing crop per flower) ×
nectar consumption increases fitness (Porter, 1992), a (number of flowers probed per unit time) > (energy
butterfly should select nectaring flowers in a way that consumption per unit time during foraging).
maximises either its energetic efficiency or its net rate I therefore consider two attributes of flowers that
of energy gain, that is, the difference between the gross govern foraging costs and benefits: corolla depth, and
rate of gain and the rate of energy expenditure in forag- the extent to which flowers are massed in inflores-
ing. The potential gross gain rate depends on the sugar cences.
content of the nectar in each flower (the standing crop) The standing crop of sugar per flower at a given
and the number of flowers emptied per unit time. The time depends on the balance between the secretion
rate of energy consumption during foraging depends rate, which correlates with flower size (Harder &
on the forager’s body mass and the mass-specific en- Cruzan, 1990), and the rate of depletion by insects.
ergy consumption rate during foraging, which depends A deep corolla restricts the nectar clientele to long-
290

tongued species, reducing the rate of depletion and shallow flowers. Corbet et al. (1995) referred to
increasing the probability that a long-tongued visitor this constraint as the depth threshold.
will find a large standing crop in the flower (Cor- (3) Butterfly species with relatively high flight costs
bet et al., 1995; Opler & Krizek, 1984). Among forage on plant species that give a relatively high
flower species, the standing crop of nectar generally energetic reward per foraging flight. Butterflies
varies positively with corolla depth (Harder, 1985; with high wing loading forage mainly on massed
May, 1988; Prŷs-Jones, 1982). The number of flowers flowers and forage on solitary flowers only if these
probed per unit time depends on handling time (de- are unusually nectar-rich.
termined in part by the relationship between tongue
length and corolla depth; Harder, 1988) and the time
spent travelling between flowers. Interflower travel Materials and methods
times (and costs) are relatively low if flowers are
massed in inflorescences, allowing foragers to move The main study took place in West Cornwall (grid ref.
between flowers by walking, rather than flying. Costs SW 422233) during August and September 1998. A
of travel between inflorescences are low if inflores- few additional records of butterfly feeding were made
cences are densely crowded. at other sites within 20 km: Crean Mill, Lamorna and
I consider three attributes of butterflies that gov- Ludgvan. The main study site was small (250 m di-
ern foraging costs and benefits: tongue length, body ameter) and it is assumed that all plant species were
mass and wing loading (mass per unit area of wing). available to all butterfly species. The few records from
The rate of energy consumption during flight is not other sites involved plant and butterfly species that
well documented for different butterfly species, but it were also present at the main site, except for the
increases with wing loading among species of moths Heliotropium hybrid.
(Heinrich, 1993) and euglossine bees (Casey et al., For each plant species, I measured the corolla of
1985). If butterfly species with high wing loading ten flowers when possible, taking each from a differ-
similarly consume energy rapidly, they can forage ent plant or ramet. Corolla depth was measured from
profitably only if they have access to large aliquots of the most proximal point to which a butterfly might
nectar per interflower flight. They should visit flowers push the base of its tongue (in sympetalous flowers,
that are massed, allowing them to move from flower to often the base of the deepest cleft in the corolla) to
flower without flying; or flowers that are individually the supposed location of the nectar at the corolla base.
nectar rich. In Asteraceae, insects probably take nectar from just
In butterflies, tongue length limits the range of above the stamen insertions, where traces of nectar
flowers from which nectar can be extracted (Erhardt, were sometimes found. I therefore measured from
1995; Opler & Krizek, 1984). Similarly, among bum- the base of the deepest cleft in the corolla to the re-
blebee species, correlates of flower choice include gion where the stamens inserted and the corolla flared
tongue length as well as flight costs (Comba et al., out, at the distal end of the very narrow part of the
1999; Corbet, 1997; Corbet et al., 1995; Harder, 1982, corolla tube that is almost filled by the style. In the
1983; Prŷs-Jones, 1982). Tongue length increases with tribe Cardueae in particular, the distance measured
body size in some species of flower-visiting insects was much less than the total corolla length. All plant
(with wing length among species of hawkmoth (Bul- species visited by butterflies in this study are peren-
lock & Pescador, 1983) and with mass within and nials (including the monocarpic perennials known as
among species of bumblebee (Prŷs-Jones, 1982)). biennials) except radish (Raphanus sativus), which is
The following predictions are tested in this pa- an annual (Table 3). Plant nomenclature follows Stace
per. (1997).
(1) In an assemblage of butterfly species, tongue I found butterflies of 15 species (Table 1). They
length correlates with body mass and wing load- represent the following subfamilies: family Pieri-
ing. dae, subfamilies Pierinae (Pieris spp.) and Coliadinae
(2) Butterflies do not forage on flowers with a corolla (Colias); family Nymphalidae, subfamily Vanessinae
depth (or, more precisely, an effective depth to (Cynthia, Aglais, Inachis, Polygonia, Vanessa); family
nectar) that exceeds their tongue length, so that Satyridae (Maniola, Pyronia, Pararge, Lasiommata);
small butterflies confine their visits to relatively family Lycaenidae, subfamily Lycaeninae (Lycaena,
Polyommatus); family Hesperiidae, subfamily Hes-
291
Table 1. Tongue length, mass and wing loading index for fifteen species of butterfly found in this study. The table gives mean ± s.e.m.
where appropriate, with the sample size in brackets. Column four shows tongue lengths from Porter et al. (1992)

Species Family Tongue length, Tongue length, Body mass, Wing loading
mm (this study) mm mg index,
(Porter et al., 1992) N m−2

1. Common blue Lycaenidae 8.19 ± 0.20 (5) 8 26 ± 2 (5) 0.66 ± 0.015 (5)
Polyommatus icarus
(Rottemburg)
2. Small copper Lycaenidae 7.34 ± 0.07 (5) 7 34 ± 3 (5) 0.86 ± 0.04 (5)
Lycaena phlaeas (L.)
3. Meadow brown Satyridae 10.45 ± 0.19 (10) 10 70 ± 9 (10) 0.67 ± 0.054 (10)
Maniola jurtina (L.)
4. Gatekeeper Satyridae 7.81 ± 0.12 (9) 8 40 ± 7 (9) 0.58 ± 0.066 (9)
Pyronia tithonus (L.)
5. Speckled wood Satyridae 7.90 ± 0.14 (8) 8 41 ± 4 (8) 0.47 ± 0.028 (8)
Pararge aegeria (L.)
6. Small white Pieridae, 12.18 ± 0.14 (5) 15 67 ± 6 (5) 0.56 ± 0.040 (5)
Pieris rapae (L.) Pierinae
7. Large white Pieridae, 14.72 ± 0.39 (10) 16 128 ± 9 (10) 0.83 ± 0.039 (10)
Pieris brassicae (L.) Pierinae
8. Green-veined white Pieridae, 10.53 ± 0.47 (4) 11 56 ± 5 (4) 0.59 ± 0.052 (4)
Pieris napi (L.) Pierinae
9. Clouded yellow Pieridae, 14.09 ± 0.21 (5) 15 96 ± 5 (5) 0.79 ± 0.026 (5)
Colias croceus Coliadinae
(Geoffroy)
10. Small skipper Hesperiidae 15.00 (2) 14 61 (2) 2.34 (2)
Thymelicus sylvestris (Poda)
11. Comma Nymphalidae 12.94 ± 0.98 (2) 12 192 (2) 1.63 (2)
Polygonia c-album (L.)
12. Painted lady Nymphalidae 15.10 ± 0.31 (4) 14 218 ± 29 (4) 1.54 ± 0.224 (4)
Cynthia cardui (L.)
13. Red admiral Nymphalidae 15.11 ± 0.28 (9) 13 276 ± 28 (9) 1.71 ± 0.151 (9)
Vanessa atalanta (L.)
14. Small tortoiseshell Nymphalidae 14.17 ± 0.42 (5) 14 136 ± 13 (5) 1.20 ± 0.095 (5)
Aglais urticae (L.)
15. Peacock Nymphalidae 16.19 ± 0.41 (3) 17 275 ± 31 (3) 1.60 ± 0.112 (3)
Inachis io (L.)

periinae (Thymelicus). Butterfly nomenclature follows by comparing counts of flowers visited and flowers
Dennis (1977), Thomas & Lewington (1991) and available.
Pullin (1995). Flower visits were recorded only if a butterfly def-
Many studies of foraging insects involve regular initely probed a flower rather than just resting on it,
systematic counts of visits and of flowers in a limited and they usually involved sustained feeding. Even
area (e.g., Corbet et al., 1995). Counting visits and so, some recorded interactions may represent one-off
flowers was not practicable here because butterflies sampling visits rather than regular associations, be-
were so sparse that it was necessary to monitor a very cause I did not quantify visit frequency. My findings
large area. Selectivity is revealed by differences be- are compared with those of Porter et al. (1992), who
tween butterfly species in flowers visited, rather than list flower-visiting records of British butterfly species.
They do not state that their list is limited to vis-
292

its involving sustained probing rather than sampling,


settling or basking.
To measure butterflies, I captured them and within
a few hours (usually two) weighed them to the near-
est 2 mg on an Ohaus CT portable electronic balance.
When possible, I sampled five individuals of each sex,
but for some species fewer individuals were available
(Table 1). The head was removed and the proboscis
was unrolled and measured with Mitutoyo digital
calipers (Mitutoyo (UK) Ltd, Andover, UK) from the
tip to the front of the compound eye. Occasionally a
proboscis was distorted and apparently nonfunctional,
and could not be measured.
Wing loading in N m−2 is mg/S where m is
the body mass in kg, g is the gravitation constant
(9.80665 m s−2 ), and S is the effective area of the
wings during flight, in m2 . Wing areas of all four
wings were measured using a Delta T leaf area meter
(Delta T, Cambridge, UK) and summed for each spec-
imen. To the extent that wings overlap during flight,
this summed area measurement overestimates the ef-
fective wing area, resulting in underestimation of wing
loading. I therefore refer to this underestimate of wing
loading as the wing loading index. Any damaged wing
was assumed to have the same surface area as its pair.
If both wings of a pair were incomplete, I did not use
the wing area measurement.
Statistical analysis was performed using SPSS 9.0.
The data set consisted of measurements for up to
ten individuals of each of 15 species. Analysis was
based, not on data for individual butterflies, but on Figure 1. Mean measurements for fifteen species of butterfly be-
species means or subfamily means. A Kolmogorov- longing to five families: Satyridae (white circles); Pieridae (black
triangles); Nymphalidae, Vanessinae (white triangles); Lycaenidae
Smirnov test showed that all variables were normally
(black diamonds); and Hesperiidae (black dot), showing (a) tongue
distributed after log transformation of body mass and length versus log body mass, and (b) wing area versus log body
wing loading index. Preliminary analysis of variance mass.
showed that any effect of sex on the measured at-
tributes of the butterflies, whether log transformed or
tongue length correlated significantly (P<0.001) with
not, was not significant at the 5% probability level, so
log body mass, wing area and log wing loading index
the sexes were pooled for further analysis.
(Table 2). Omission of the mean for the small skip-
per improved the correlation of tongue length with log
body mass and wing area, but not with log wing load-
Results
ing index (Table 2). The significant correlations were
not all due to inclusion of several species in each sub-
Mean values of tongue length (Figure 1a) and wing
family. When I used the means of all the species means
area (Figure 1b) of the 15 species increased with body
in each subfamily, again omitting the skippers, tongue
mass. The small skipper (Thymelicus sylvestris), the
length correlated significantly with log mass and wing
sole representative of the Hesperiidae, had a longer
area but not with log wing loading index (Table 2).
tongue and a smaller wing area relative to its body
No butterflies were seen to probe flowers deeper
mass than the other species (Figures 1a, b). With
than the mean tongue length of their species (Ta-
each species mean treated as an independent data
ble 3). Potential interactions between a butterfly and a
point (a ‘naïve’ interspecific analysis; Grafen, 1989),
293
Table 2. Pearson correlation coefficients for tongue length versus log body mass,
total wing area and log wing loading index, based on species means of 2–10
individuals in each of 15 species, and means of the species means in six subfam-
ilies (total: 86 individuals). Correlations are improved by omitting the skippers
(Hesperiidae, Hesperiinae), leaving 14 species in 5 subfamilies

Log mass Wing area Long wing loading index

All species (n = 15) 0.862∗∗∗ 0.672∗∗ 0.733∗∗


Species, skipper 0.933∗∗∗ 0.933∗∗∗ 0.724∗∗∗
omitted (n = 14)
Subfamilies, skippers 0.932∗ 0.935∗ n.s.
omitted (n = 5)

∗ P < 0.05; ∗∗ P < 0.01; ∗∗∗ P < 0.001

cosus agg.), which produces large and often clustered


flowers that can be nectar-rich. The standing crop of
nectar sugar in flowers of a given species fluctuates
markedly over time (Corbet et al., 1995), but that of
R. fruticosus is often so large that this has been de-
scribed as a cornucopia species sensu Mosquin (1971)
(Yeboah Gyan & Woodell, 1987a,b). The only single-
flowered species with a long corolla was red campion
(Silene dioica). Seven butterfly species have tongues
longer than the corolla of this species, five with wing
loading indices greater than 1.2 N m−2 (four vanes-
sine nymphalids and a hesperiid) and two pierids with
wing loading indices below 0.85 N m−2 . Only the two
species with low wing loading (Coleus croceus and
Pieris brassicae) probed red campion flowers.
Figure 2. Relation between the mean effective corolla depth of the
plants visited by 15 butterfly species and the butterflies’ mean wing
loading index (N m−2 ), for plant species with solitary (black dots) Discussion
or massed (open circles) flowers. Each point represents at least one
visit. The line marks a wing loading index of 0.9 N m−2 . Butterflies
with wing loading greater than this generally confined their visits to In this assemblage of butterfly species, tongue length
massed or nectar-rich flowers. correlates with body mass and wing loading, support-
ing hypothesis 1.
My observations support hypothesis 2, in that but-
flower with a corolla longer than its tongue are marked
terflies did not feed on flowers with a mean corolla
X in Table 3. I saw no such interactions. Although
depth greater than their mean tongue length. Hypoth-
plants with deep corollas were visited by only the
esis 2 is also supported by the findings of Porter et al.
few long-tongued butterfly species, some effectively
(1992), with two apparent exceptions. These authors
short-tubed plants such as Centaurea nigra and Eupa-
recorded two butterfly species, small whites (Pieris
torium cannabinum received visits from a wide range
rapae) and green-veined whites (P. napi), visiting red
of butterfly species.
campion (Silene dioica), for which my mean corolla
Pierids, satyrids and lycaenids all had wing load-
length measurement (13.30 mm in female flowers) ex-
ing indices below 0.9 N m−2 , and visited both massed
ceeds the mean tongue length I measured for small
and solitary flowers. The vanessine nymphalids and
whites (12.18 mm) and green-veined whites P. napi
the hesperiid all had wing loading indices exceeding
(10.53 mm). Given the variability in the lengths of
1.2 N m−2 , and did not generally visit solitary flow-
both tongues (small whites, range 12.01 to 12.72 mm
ers (Figure 2). The exception was a visit by a red
(n = 5); green-veined whites, range 9.30–11.51 mm,
admiral (Vanessa atalanta) to bramble (Rubus fruti-
n = 4) and flowers (range 11–15 mm, n = 10), some
Table 3. Plant species on which at least one butterfly visit was recorded in this study, with corolla depth (mean ± s.e.m.) and flower attributes. Column 5 shows which flowers
294

are massed in inflorescences, and column 6 shows which are members of the important family Asteraceae. Flowers present also in the list of Porter et al. (1992) are in bold.
Butterfly visits in that list are shown as asterisks. Plant and butterfly species in the list of Porter et al. but not in my list are omitted. Potential butterfly/flower combinations
in which the corolla depth would exceed the tongue length are marked X

mean corolla depth, mm


± s.e.m.
no. flowers measured
no. plants [or ramets]
are flowers massed?
Asteraceae
no. butterfly spp. (this study)
no. butterfly spp. (Porter)
Lycaena phlaeas
Pyronia tithonus
Pararge aegeria
Polyommatus icarus
Maniola jurtina
Pieris napi
Pieris rapae
Polygonia c-album
Coleus croceus
Aglais urticae
Pieris brassicae
Thymelicus sylvestris
Cynthia cardui
Vanessa atalanta
Inachis io

Mean tongue length, mm 7.34 7.81 7.90 8.19 10.45 10.53 12.18 12.94 14.09 14.17 14.72 15.00 15.10 15.11 16.19

Rubus fruticosus agg. 0.00 0 6 9 Y∗ Y∗ Y∗ Y ∗ ∗ Y ∗ Y


Jasione montana 0.00 0 Y 1 Y
Escallonia sp. nr. macrantha 0.00 0 1 Y
Geranium sp. nr. endressii 0.00 0 1 Y
Taraxacum officinale agg. 0.33 0.03 10 2 Y Y 1 7 ∗ ∗ ∗ Y∗ ∗ ∗ ∗

Sedum spectabile 2.05 0.06 10 2 Y 6 Y Y Y Y Y


Hebe cv. A 2.69 0.07 10 1 Y 1 Y
Eupatorium cannabinum 2.71 0.08 10 10 Y Y 8 10 Y Y Y Y Y ∗ Y Y Y∗ ∗

Leucanthemum × superbum 2.99 0.10 10 [10] Y Y 5 Y Y Y Y


Erigeron karvinskianus 3.31 0.04 10 [10] Y Y 6 Y Y Y Y Y Y
Senecio jacobaea 3.72 0.09 10 10 Y Y 6 9 Y Y Y Y ∗ ∗ Y ∗

Hebe cv. B 3.93 0.08 10 1 Y 2 Y Y


Mentha aquatica 4.11 0.16 10 10 Y 6 8 ∗ Y Y Y Y∗ Y Y ∗

Heliotropium hybrid 4.20 0.10 10 1 Y 4 Y Y Y


Origanum vulgare 4.25 0.17 10 10 Y 2 8 Y ∗ ∗ Y ∗ ∗ ∗ ∗

Centaurea nigra 4.79 0.23 10 10 Y Y 11 12 Y∗ Y∗ Y Y ∗ Y∗ ∗ Y∗ Y∗ Y∗ Y Y


Ligustrum sp. 5.65 0.19 2 2 Y 1 9 ∗ ∗ ∗ ∗ ∗ Y ∗ ∗ ∗

Buddleja globosa 6.01 0.16 10 1 Y 3 Y Y Y


Geranium robertianum 6.12 0.19 10 10 2 Y Y
Prunella vulgaris 7.22 0.22 10 10 Y 3 4 ∗ Y Y
Agastache urticifolia 7.57 0.13 10 1 Y 3 X Y Y Y
Raphanus sativus 7.69 0.13 10 10 1 X Y
Buddleja davidii 9.20 0.22 10 10 Y 5 X X X X Y Y Y Y Y
Agapanthus praecox 12.53 0.23 10 10 Y 3 X X X X X X X Y Y Y
Silene dioica 13.30 0.42 10 10 2 4 X X X X X X∗ X∗ X Y Y
295

individuals of both pierid species could reach the nec- The potential duration of uninterrupted flight de-
tar in at least some S. dioica flowers. Thus hypothesis pends on maintenance of a high enough body tem-
2 is supported without exceptions. perature for flight. Body mass is positively associated
Hypothesis 3, that butterfly species with high wing with the capacity for endothermy in bees (Stone &
loading forage mainly on flowers that are massed or Willmer, 1989) and probably in butterflies as well
unusually nectar-rich, is supported by my findings, (Heinrich, 1993). Small insects have relatively high
and by the subset of those of Porter et al. (1992) in- rates of convective cooling and tend to cool during
volving the same species, in that no high-wing-loading flight. Heinrich (1993) has suggested that at low am-
species fed on solitary flowers other than the nectar- bient temperatures, insects weighing less than about
rich bramble (Rubus fruticosus). However, the total 200 mg probably cannot warm up metabolically fast
list of Porter et al. includes apparent exceptions among enough to achieve flight temperature or to maintain or
species not represented in my study. Eight of the 38 elevate thoracic temperature during flight. Small in-
butterfly-visited plant species that they examined have sects can extend flight duration by pre-flight basking to
solitary flowers. As in my study, Rubus fruticosus elevate the thoracic temperature well above the flight
received visits from many butterfly species including threshold, but they must stop flying when the thoracic
six nymphalids and two hesperiids, and red campion temperature falls below this threshold. Hence at low
(Silene dioica) was visited only by pierids. The re- ambient temperatures, only butterflies weighing more
lated ragged robin (Lychnis flos-cuculi), with a shorter than about 200 mg can maintain continuous uninter-
corolla and a higher midday nectar secretion rate than rupted flight (Heinrich, 1993). Presumably smaller
red campion (Comba et al., 1999), was visited not species can migrate when the ambient temperature
only by six pierid species but also by three species (or, more precisely, a black globe temperature that
of hesperiids, one satyrid and two species of fritillary. allows for radiative warming in the sun: Pivnick &
Fritillaries are nymphalids of the subfamilies Argyn- McNeil, 1987) equals or exceeds the flight threshold.
ninae and Melitaeinae, for which the wing loading Of the species considered here, nine (all the pierids
index has not been measured but is probably lower and vanessine nymphalids) are classified as migrants
than that of the vanessine nymphalids considered here. by Thomas (1984 and pers. comm.). Only three of
Fritillaries were the only nymphalids that visited tor- these weigh more than 200 mg. Presumably the lim-
mentil (Potentilla erecta) and buttercups (Ranunculus ited endothermic capacity of the other six allows them
species), but cuckoo flower (Cardamine pratensis) and to sustain uninterrupted flight only during warm sunny
greater stitchwort (Stellaria holostea) were visited by weather.
vanessine nymphalids (a peacock Inachis io and a Sustained flight also depends on the amount of fuel
small tortoiseshell Aglais urticae respectively). In the carried. Load-carrying capacity depends on the ratio
list of Porter et al. (1992) the only apparent exceptions of flight muscle mass to total body mass. This ratio
to hypothesis 3 are these and buttercups visited by is expected to be high in species that frequently need
hesperiids. Those visits may have involved sampling maximal power output for lifting loads such as large
rather than sustained probing. Thus my findings sup- nectar meals or batches of unlaid eggs (Marden, 1987).
port hypothesis 3 and those of Porter et al. probably The ability to sustain flight has implications for
do not falsify it. the population dynamics and conservation prospects
The assemblage of attributes that influence flower of butterflies. The relation between mobility and mor-
choice also influence other aspects of butterfly biol- phology is illustrated in Figure 3. With the exception
ogy. Attributes that influence flower choice by af- of the hesperiid, all the species considered here that
fecting flight costs and rate of energy intake will, in Thomas (1984 and pers. comm.) classifies as seden-
conjunction with fuel capacity, determine how far a tary species with closed populations have tongues less
butterfly can fly between stops. This has implications than 10.5 mm long and a wing loading index below
for flower selectivity and migration, both of which de- 0.9 N m−2 . They appear in the lower left quadrant
pend on both the potential duration of uninterrupted of Figure 3. All the species that Thomas classifies
flight (determining how far an insect can fly without as migratory or having open populations have longer
stopping to warm up), and load-carrying capacity (de- tongues, and the vanessine nymphalids (but not the
termining how far an insect can fly between refuelling pierids) also have a higher wing loading index.
stops). The ability to sustain flight also has implications
for flower selectivity. An insect capable of making
296

Figure 3. Mean tongue length of 15 butterfly species, with family symbols as in Figure 1, plotted against wing loading index. Line A at
0.9 N m−2 and line B at 10.5 mm delimit a lower left quadrant in which all species (1–5 in Table 1) form closed populations (Thomas, 1984 and
pers. comm.). Above line B, all species (6–9 and 11–15) except the small skipper (10 in Table 1) are classified as forming open or migratory
populations (Thomas, 1984 and pers. comm.).

long flights and carrying heavy loads can be highly nuals in relation to their relative abundance, as do
selective, visiting only nectar-rich flowers or inflo- bumblebees (Fussell & Corbet, 1992a, b). In the total
rescences, and taking meals that are large and few. list of Porter et al. (1992) all 38 plant species visited
An insect capable only of brief flights and of car- by butterflies were herbaceous or woody perennials,
rying light loads might be less selective, bypassing not annuals. In my study the only annual visited was
fewer flowers when foraging, and taking smaller, more radish (Raphanus sativus, Brassicaceae). The related
frequent meals. May (1992) found a difference in se- Sinapis arvensis is visited by Pieris napi, P. rapae and
lectivity between two syntopic butterfly species with P. brassicae on conservation headlands in arable fields
mean body weights of 195 and 231 mg respectively (Dover, 1989). On annually-ploughed land, which in-
(May, 1988). The heavier species, Phoebis sennae L. evitably yields few perennials, Sinapis is one of a
(Pieridae), achieved a higher net rate of energy gain small group of annuals rich enough in nectar to permit
than the lighter Agraulis vanillae L. (a non-vanessine butterflies to forage profitably, and even this is appar-
nymphalid) by selecting flowers with more nectar. ently useful only for pierids with their combination of
May (1988) asked why the smaller species did not a long tongue with a moderate wing loading.
forage as profitably as the larger species. Perhaps the For an assemblage of butterfly species in Britain
paradox of its lower selectivity and hence apparently in late summer, this paper has shown a relationship
suboptimal foraging can be interpreted in terms of between the tongue length and wing loading index of
thermoregulatory and load-carrying constraints. a butterfly species and features of the flower species
Nearly all the plant species that butterflies visited that it probes for nectar. Although many of the plant
for nectar flowers are perennials. Presumably, few an- species studied here are exotic, the generalisations de-
nuals contain enough nectar to make a visit profitable. rived from them are broadly supported by comparison
Feber & Smith (1995) and Dover (1996) found that with the more comprehensive set of records of Porter
butterflies visited perennials more frequently than an- et al. (1992), and may be expected to illuminate flower
297

choice by butterflies in seminatural habitats as well as Corbet, S. A., N. M. Saville, M. Fussell, O. E. Prŷs-Jones & D.
gardens. M. Unwin, 1995. The competition box: a graphical aid to fore-
casting pollinator performance. Journal of Applied Ecology 32:
These generalisations may help inform the man- 707–719.
agement of vegetation for butterflies. In general, Dennis, R. L. H., 1977. The British Butterflies. Their Origin and
species with shorter tongues, notably lycaenids and Establishment. E. W. Classey Ltd, Faringdon.
some satyrids, do not visit deep flowers (including the Dover, J. W., 1989. The use of flowers by butterflies foraging in
cereal field margins. Entomologist’s Gazette 40: 283–291.
well-known butterfly flower Buddleja davidii). Provi- Dover, J. W., 1996. Factors affecting the distribution of satyrid
sion for these shorter-tongued species should include butterflies on arable farmland. Journal of Applied Ecology 33:
Asteraceae such as Eupatorium, Senecio and Leucan- 723–734.
themum. Deep flowers, with their potentially abundant Erhardt, A., 1995. Ecology and conservation of alpine Lepidoptera.
In: A. S. Pullin (ed), Ecology and Conservation of Butterflies.
nectar, were much visited by butterflies with tongues Chapman & Hall, London, pp. 258–275.
long enough to reach to the corolla base. Pierids, Feber, R. E. & H. Smith, 1995. Butterfly conservation on arable
which combine a relatively long tongue with a mod- farmland. In: A. S. Pullin (ed), Ecology and Conservation of
erately low wing loading (Figure 1b), can include in Butterflies. Chapman & Hall, London, pp. 84–97.
Fussell, M. & S. A. Corbet, 1992a. Flower usage by bumble-bees: a
their diet deep solitary flowers such as Silene dioica basis for forage plant management. Journal of Applied Ecology
and even annual brassicaceous arable weeds. The 29: 451–465.
vanessine nymphalids, long-tongued butterflies with a Fussell, M. & S. A. Corbet, 1992b. Forage for bumble bees and
honey bees in farmland: a case study. Journal of Apicultural
high wing loading, usually visited flowers massed in
Research 30: 87–97.
dense inflorescences, such as Buddleja davidii, Hebe Grafen, A., 1989. The phylogenetic regression. Philosophical Trans-
species and Asteraceae of the tribe Cardueae. Proba- actions of the Royal Society B 326: 119–157.
bly they cannot forage profitably on solitary flowers Harder, L. D., 1982. Measurement and estimation of functional pro-
boscis length in bumblebees (Hymenoptera: Apidae). Canadian
unless, like Rubus fruticosus, these are very rich in Journal of Zoology 60: 1073–1079.
nectar. Many important nectaring plants are members Harder, L. D., 1983. Flower handling efficiency of bumble bees:
of the family Asteraceae, and Eupatorium and nectar- morphological aspects of probing time. Oecologia 57: 274–280.
rich Cardueae such as Centaurea nigra host large Harder, L. D., 1985. Morphology as a predictor of flower choice by
bumble bees. Ecology 66: 198–210.
numbers of butterfly species spanning a wide range of Harder, L. D., 1988. Choice of individual flowers by bumble bees:
morphological types. interaction of morphology, time and energy. Behaviour 104: 60–
77.
Harder, L. D. & M. B. Cruzan, 1990. An evaluation of the
physiological and evolutionary influences of inflorescence size
Acknowledgements and flower depth on nectar production. Functional Ecology 4:
559–572.
I thank Andreas Erhardt and Jeremy Thomas for pro- Heinrich, B., 1993. The Hot-Blooded Insects. Springer-Verlag,
Berlin.
viding unpublished data and ideas, Andrew Barron,
Marden, J. H., 1987. Maximum lift production during takeoff in
Charlie Ellington, Lawrence Harder and Robin Woot- flying animals. Journal of Experimental Biology 130: 235–258.
ton for comments on an earlier draft, Paul Freeman May, P. G., 1988. Determinants of foraging profitability in two
for help with equipment and John Hobart and Philip nectarivorous butterflies. Ecological Entomology 13.
May, P. G., 1992. Flower selection and the dynamics of lipid re-
Corbet for help with field recording.
serves in two nectarivorous butterflies. Ecology 73: 2181–2191.
Mosquin, T., 1971. Competition for pollinators as a stimulus for the
evolution of flowering time. Oikos 22: 398–402.
References Opler, P. A. & G. O. Krizek, 1984. Butterflies East of the Great
Plains. The Johns Hopkins University Press, Baltimore.
Pivnick, K. A. & J. N. McNeil, 1987. Diel patterns of actvity of
Bullock, S. H. & A. Pescador, 1983. Wing and proboscis dimensions
Thymelicus lineola adults (Lepidoptera: Hesperiidae) in relation
in a sphingid fauna from western Mexico. Biotropica 15: 292–
to weather. Ecological Entomology 12: 197–207.
294.
Porter, K., 1992. Eggs and egg laying. In: R. L. H. Dennis (ed),
Casey, T. M., M. L. May & K. R. Morgan, 1985. Flight energetics
The Ecology of Butterflies in Britain. Oxford University Press,
of euglossine bees in relation to morphology and wing stroke
Oxford, pp. 46–72.
frequency. Journal of Experimental Biology 116: 271–289.
Porter, K., C. A. Steel & J. A. Thomas, 1992. Butterflies and com-
Comba, L., S. A. Corbet, L. Hunt & B. Warren, 1999. Flow-
munities. In: R. L. H. Dennis (ed), The Ecology of Butterflies in
ers, nectar and insect visits: evaluating British plant species for
Britain. Oxford University Press, Oxford, pp. 138–177.
pollinator-friendly gardens. Annals of Botany 83: 369–383.
Prŷs-Jones, O. E., 1982. Ecological studies of foraging and life
Corbet, S. A., 1997. Role of pollinators in species preservation,
history in bumblebees. PhD, Department of Applied Biology,
conservation, ecosystem stability and genetic diversity. Acta
University of Cambridge.
Horticulturae 437: 219–229.
298

Pullin, A. S., 1995. Ecology and Conservation of Butterflies. Thomas, J. & R. Lewington, 1991. The Butterflies of Britain and
Chapman & Hall, London. Ireland. Dorling Kindersley, London.
Stace, C., 1997. New Flora of the British Isles. Cambridge Univer- Yeboah Gyan, K. & S. R. J. Woodell, 1987a. Flowering phenol-
sity Press, Cambridge, UK. ogy, flower colour and mode of reproduction of Prunus spinosa
Stone, G. N. & P. G. Willmer, 1989. Warm-up rates and body tem- L. (blackthorn); Crataegus monogyna Jacq. (hawthorn); Rosa
perature in bees: the importance of body size, thermal regime canina L. (dog rose); and Rubus fruticosus L. (bramble) in
and phylogeny. Journal of Experimental Biology 147: 303–328. Oxfordshire, England. Functional Ecology 1: 261–268.
Thomas, J. A., 1984. The conservation of butterflies in temper- Yeboah Gyan, K. & S. R. J. Woodell, 1987b. Nectar production,
ate countries: past efforts and lessons for the future. In: R. I. sugar content, amino acids and potassium in Prunus spinosa L.,
Vane-Wright & P. R. Ackery (eds), The Biology of Butterflies. Crataegus monogyna Jacq. and Rubus fruticosus L. at Wytham,
Academic Press, London, pp. 333–353. Oxfordshire. Functional Ecology 1: 251–259.

You might also like