102Four hypotheses are suggested to explain the use of lesssuitable hostplants by
Hypothesis l:Females are unable to distinguish differenthostplant species.Hypothesis 2: Insufficient time has elapsed for evolution toproduce discrimination.Hypothesis 3 : The apparent lack of selectivity is a strategyin time and/or space against resource unpredictability.Hypothesis 4: Females are in fact maximizing reproductiveeffort, laying eggs on poor hostplants as a consequence of timeconstraints.
Stanton (1979), in research on chemotactile stimulation and ovi-position of legume-feeding Pieridae, emphasised the need to re-cognise physiological limitations to discrimination by female but-terflies amongst foodplants. Polyphagy is unsurprising if foodplants are indistinguishable from one another. However, in anumber of different crucifer-feeding species, insects have beenshown to distinguish between different glucosinolates, each cru-cifer species typically having a unique profile of these precursorsof the primary defence chemicals, the mustard oils (Nair et al.1973; Hicks 1974; Sehgal and Vjagir 1977; Nielsen 1978). Rod-man and Chew (1980) have argued that Pierinae females arewell able to distinguish between Cruciferae by such glucosinolateprofiles, except for certain cases where different species sharesimilar profiles.
H. matronalis and B. vulgaris
contain quite dif-ferent glucosinolates to
(Kjaer 1976). AdditionallySalmon (in prep.) shows convincingly that larvae of two Pierinaespecies are well able to distinguish amongst Co. Durham Cruci-ferae by chemosensory means. Thus it seems reasonable to as-sume that discrimination by females does or could exist in
if selection is favourable. Hypothesis l is not favouredon available evidence.Chew (1977a, b) showed that oviposition on inappropriatehostplants by Nearctic Pierinae was in part explainable by histor-ical factors. She argued that
which kills larvaeof North American
(L.), and which is an intro-duced species, has not been present in the butterfly's environmentlong enough for avoidance behaviour to evolve. Maynard Smith(1978) has detailed other similar examples of 'evolutionary lag',such as the inappropriate clutch size of some seabirds (Nelson1964). Of the two hostplants species poor for
is an introduced species in Britain.The introduction is probably of long standing, and the hostplanthas been known for
since Newman (1869) re-corded both
amongst the four host-plants most commonly used. One specimen of
taken in Co. Durham in the year 1855 was found to bear pollenof
when examined microscopically; thus it seemsthat the butterfly and crucifer have co-existed in the study areafor at least 125 insect generations, arguably long enough foravoidance behavior to evolve ((Jones 1977) has shown how food-plant location behavior of
has changed in populationsderived from common stock in the 19th Century). On the conti-nent of Europe,
is native and has probably beenin proximity with
for millenia; in Italy, at least,the plant is regularly used as a host (C.J. Hill, pers. comm.).'Evolutionary lag', Hypothesis 2, does not appear tenable.Several authors have argued that the 'polyphagous type'of oligophagy is practised by Pierinae as a response to the unpre-dictability of larval foodplants and related mortality factors ineither space or time, or in both.
(Chew 1977a) and
(Wiklund and Ahrberg 1978)are all held to adopt this strategy in order to overcome fluctua-tions in resource availability and suitability. Chew emphasisedthe importance of variation in foodplant species in space; thisdoes not appear to be a satisfactory explanation for
in northern England. In 30 populations on a transectspread across some 75 km distance of the northern Pennines,taking in several habitat types,
was always asso-ciated with the foodplants
L. (Table 1). These two foodplants are amongst the best availablefor larval survival (Courtney t981) and are those typically re-garded as the hosts of the butterfly in both Britain and Sweden.Of 48 British correspondents, only one has recorded an
population which does not feed on either of these species.
females which oviposited only on these two cru-civers would not therefore experience substantially less predicta-bility of foodptant species in space. Foodplant unpredictabilityin time is also unlikely to favour polyphagic strategies in British
At one major study site, Durham, the numberof available hostplants was counted annually for 1977 to 1980(Table 2). It is seen that although some variance occurs, in thefour years of study,
dominated the crucifer commun-ity of the riverbank study site. However, in Sweden more dramat-ic variations in crucifer abundance may occur in early succes-sional habitat (C. Wiklund, pers. comm.).Wiklund and Ahrberg (1978) have emphasised that unpredic-tability of some foodplant associated mortality (such as drowningon marsh species) will lead to selection against specialisationon single hostplant species. Present evidence from northern En-gland suggests strongly that most differences in larval mortalityon different hostplants are however predictable. Based on a studyof the survival of 1,798 individuals in the wild, I have arguedthat most differences in mortality reflect differences in the qualityof the Cruciferae themselves as larval food, with mechanicaland chemical defences killing many larvae on
(Courtney 1981). Similarly
is seen to bea poor foodplant for
spp. not only in Co. Durham,but also in Sweden (C. Wiklund, pers. comm.) and the U.S.A.(Chew 1977b). In the long term, evolution against the use ofsuch a consistently poor foodplant should have occurred. Fe-males of
avoid ovipositing on
which contains cardenolides, allelochemicals which seemto kill 100% of larvae grazing on the plants of this genus (Chew1977; Wiklund and Ahrberg 1978). Additional problems alsooccur with the hypothesis of Wiklund and Ahrberg, and other'strategic' formulations, not least the difficulty of investigatinglong-term processes. A major objection is that in situations ofat least short term predictability (as at Durham), 'polyphagoustype' oligophagy will not be evolutionarily stable against inva-sion by monophagous strategists. Polymorphism for the differentstrategies would then be expected (since monophagy will alwayslead to higher fitness in predictable conditions); Chew (1977a)has shown that this does not occur in oligophagous Pierinae.The maintenance of a pure 'polyphagous type' strategy by thewhole population may then require group selection scenariosunder this model. One further problem with the use of strategicexplanations is that they explain even results exactly contraryto those predicted, for instance in the study of Rausher (1979),and used
they cease to have any claim to testability.Hypothesis 3 is not corroborated by available evidence. Theremaining hypothesis holds that the behaviour of femlaes, inovipositing on foodplants as they are found, is adaptive whentime in which to locate hosts is limited. This hypothesis suggeststhat polyphagic strategists will always be favoured and hence