THE PHENOLOGY OF PIERIS NAPI MICROSTRIATA ( LEPIDOPTERA : PIERIDAE ) DURING AND AFTER THE 1975-77 CALIFORNIA DROUGHT , AND ITS EVOLUTIONARY SIGNIFICANCE

Interest in the theoretical basis of insect phenology has increased very markedly in recent years (Bradshaw, 1974; Cohen, 1970; Giesel, 1976; Levins, 1969). The evolution of phenological "strategies" via natural selection is of interest to ecologists and applied entomologists alike. Insect phenology involves responses to both "normal" and "abnormal" weather. The developmental plasticity displayed by a population may determine its survival in seasons of unusual meteorological stress and in turn may reflect a history of selection by recurrent exposure to that stress. The 1975-77 California drought was a short-term climatic anomaly with no equal in the meteorological records of that state. Any event of such magnitude would be expected to affect both phenology and reproductive success of a great variety of organisms, especially annual or ephemeral species. Since 1972 the phenology of entire butterfly faunas has been monitored along a transect paralleling Interstate Highway 80 from sea level at Suisun Bay to 2750 m at Castle Peak, north-central California. This has permitted year-toyear comparisons within localities as well as within-year comparisons among localities. This long-term study involving over 150 species includes the 1975-77 drought and the tremendous rains of winter 1977/78. Certain butterfly populations showed very pro-


INTRODUCTION
Interest in the theoretical basis of insect phenology has increased very markedly in recent years (Bradshaw, 1974; Cohen, 1970; Giesel,  1976; Levins, 1969).The evolution of phenological "strategies" via natural selection is of interest to ecologists and applied entomolo- gists alike.Insect phenology involves responses to both "normal" and "abnormal" weather.The developmental plasticity displayed by a population may determine its survival in seasons of unusual mete- orological stress and in turn may reflect a history of selection by recurrent exposure to that stress.
The 1975-77 California drought was a short-term climatic anom- aly with no equal in the meteorological records of that state.Any event of such magnitude would be expected to affect both phenol- ogy and reproductive success of a great variety of organisms, especially annual or ephemeral species.Since 1972 the phenology of entire butterfly faunas has been monitored along a transect parallel- ing Interstate Highway 80 from sea level at Suisun Bay to 2750 m at Castle Peak, north-central California.This has permitted year-to- year comparisons within localities as well as within-year comparisons among localities.This long-term study involving over 150 species includes the 1975-77 drought and the tremendous rains of winter 1977/78.Certain butterfly populations showed very pro- *Manuscript received by the editor June 30, 1979.venosa is normally double-to partially triple-brooded, with two seasonal phenotypes (Shapiro, 1975, 1977)under photoperiod and temperature control.P. n. microstriata is normally univoltine, with only a spring phenotype (Shapiro, 1975(Shapiro, , 1976a(Shapiro, , 1977)).It is capable of producing a summer phenotype and may have a partial second brood in some localities in some years.This most often occurs in Coast Range localities subject to occasional maritime influence, as around the Napa Valley.
Populations of P. n. microstriata occur at two of the regular sampling stations along 1-80: Gates Canyon, an east-facing canyon on the east slope of the Vaca Hills (Inrier Coast Range) near Vaca- ville, Solano Co. (50-600 m), and Lang Crossing of the South Yuba River in the mixed-conifer belt of the Sierran west slope, Nevada Co. (1350-1500 m).These are matched with nearby U.S. Weather Bureau stations of record at Vacaville and Blue Canyon Airport, respectively.Some aspects of the host-plant relations of the butter- flies are described in Shapiro, 1974 and 1976b.Both populations fluctuated during the most anomalous year (1977), but in opposite directions.Both were quantified by direct census of both adults and immatures.

GATES CANYON
The history of this population since 1972 is given in Table 1.The 1972 estimate is unreliable because only one sampling day is involved, but all of the others are based on weekly to biweekly visits through the flight season.
January-March 1977 were statistically unremarkable for tempera- ture.The adult population of P. n. microstriata was almost non- existent, despite the great burst of reproduction the previous year.
Four males were seen--one each on four days--and at least one Year female must have flown because a single egg was found 11 April.It must have failed to develop since there was no damage at all to the plant on the 28th.The biomass of Barbarea was reduced by roughly an order of magnitude relative to 1976, and the number of plants decreased very markedly.This facilitated search for early stages, and host plant coverage was especially good in 1977.It is thus uncertain that any successful reproduction by P. n. microstriata took place in Gates Canyon this year.
Winter 1977/78 brought record heavy rainfalls (Table 2).Rather than being extinct, as would be expected for an obligate annual, P. n. microstriata was back at pre-drought numbers.Most of the emer- gence was early: females were already present on 14 March and eggs were found on the less-preferred host Dentaria (Barbarea was late).By 9 April there were eggs, small larvae, and large, larvae on most of the Barbarea plants and some on Dentaria.One male each was seen on 21 April and 7 May (none on 29 April).By 7 May only a handful of large larvae remained.On 20 and 28 May single males of the rare second generation (castoria phenotype)--the first ever recorded in the Vaca Hills--were found.

LANG CROSSING
Here there are no 1976 data, and the irregular topography makes coverage of potential hosts less accurate.However, the 1977 events make a striking counterpoint to those in Gates Canyon (Table 1).
The 1976/77 winter produced very little snow at mid-elevations.
Although it was a mild winter, cumulative chilling experienced by diapausing pupae near ground level was undoubtedly enhanced by lack of snow cover.Lang Crossing was completely snow-free on 17 April and 20 species were flying, including male P. n. microstriata.
By 22 April females were flying and laying on Barbarea and on Nasturtium officinale R. Br.April temperature was below normal but the weather was fair until the very end of the month, after most 1979] Shapiro Phenology of P. napi microstriata May.On 13 May, with north-facing slopes snow-covered, as many plants as were exposed were censused.43 eggs and 2 small larvae were found on Barbarea.On 21 May the same plants, along with others and several Arabis glabra (L.) Bernh., were again censused but not one egg or larva was found.Larval feeding damage was limited to the pinholes inflicted by first-instar larvae.Insofar as could be determined., the entire 1977 reproductive output was lost.
Other than a soft-winged female found 13 May, no more adults were seen.Harcourt (1966) found that rain was a major mortality factor for eggs and larvae of Pieris rapae (L.) on cabbage in Canada.
Winter 1977 78 saw a return of heavy snowfall, but there was also so much warm rain that snow packs were not good, and Lang was snow-free 12 April: 10 species, not including P. n. microstriata, were flying. 32cm of snow fell 15 April, with cold rain 24-25 and 28 April.On 3 May one male was seen.Snow fell again 23-24 May.On 29 May and 2 June P.n. microstriata were common, with eggs on the latter date on (in descending order of preference), Arabis, Nasturtium, and Barbarea.The last adult seen was 9 June; on 15 June 3 mature larvae were still found on Arabis.

DISCUSSION
There is little in the literature on catastrophic extinctions of natu- ral butterfly populations (Ehrlich et al., 1972).Both populations of P. n. microstriata under study flirted with extinction in 1977, for Psyche [March different reasons.As an early spring insect, P. n. microstriata trades off lateness of emergence (to reduce the probability of catastrophic weather-induced mortality) against earliness (to match the phenol- ogy of the vernal Crucifers).The former factor would seem more compelling at Lang, the latter at Gates.How do P. n. microstriata "decide" when to emerge?
Under controlled laboratory conditions populations show an astonishing intrapopulational variance in the "chilling requirement" to break pupal diapause.There seem to be interpopulational differ- ences as well, but for the purposes of Table 3 pupae from four Table 3. "Chilling requirement" to break diapause in P. n. microstriata.Dormant pupae were held at 3 C and tested at intervals at 20 to ascertain whether diapause had been broken.Pooled data for several broods (see notes), 1972 through 1978.defined as individuals which diapaused under continuous light at 20C+.From 14 females ex 3 foothill populations and mid-elevation Sierran population.
-defined as individuals which diapaused under inducing photoperiods at 20 C+.Since these broods included some genetic diapausers which would have diapaused anyway, these are included in the tally.From 4 females ex 2 foothill populations.3time elapsed does not include pre-testing time and testing time to assess dormancy at 20 1979] Shapiro Phenology of P. napi microstriata populations are pooled.There is a difference between genetic "obli- gate" diapausers, which are refractory to photoperiod, and faculta- tive diapausers which have been reared on inducing regimes, as well as a large spread of developmental times in each group.How this variability is expressed afield is unclear.Except for some pupae which do not develop at all in the first year, all individuals probably come out of diaause in January and subsequent development is temperature-dependent.The accumulated chilling in continuous cold storage is much greater than the wild pupae experience in fluctuating temperatures, so the "chilling requirement" is really a more complex chilling-time interaction.Except for pupae which carry over to a second or subsequent year, the effect of intrapopula- tional variance may be largely masked in cold, wet springs and maximally expressed in warm, dry ones.The carryover pupae provide the only reasonable explanation for the large 1978 population at Gates after the 1977 disaster.The numbers, however, suggest that a large number of carryover pupae may have been involved.This in turn suggests that something about the 1976/77 season cued a carryover response in a larger-than-usual fraction of the population.The temperature pattern is conveniently similar to 1977/78; the striking difference is rainfall.
Carryover pupae occur in other Pierids, including Anthocaris sara and A. cethura (Felder & Felder), A. lanceolata Lucas, Euchloe hyantis (Edw.),E. ausonides Lucas, and Pieris sisymbrii Bdv. in western North America.They also occur in Papilio rudkini Comstock (Papilionidae), a desert swallowtail noted for its aseasonality, correlation of flights with heavy rains, and diapause pupae which may carry over for 6 years or more (Emmel and Emmel, 1973 and pers.comm.).There is strong circumstantial evidence for this spe- cies,.that a physiological response to water initiates post-diapause development--the insect behaves like a desert annual plant with a water-soluble seed-germination inhibitor.Of the Pierid species listed above, some occur in xeric and some in more mesic habitats, but not enough data are available to say whether the frequency of carryover pupae is correlated with rainfall uncertainty in the habitat.
In any case, P. n. microstriata is a surprising insect to have the carryover response, given that it is almost certainly of Arcto-Tertiary mesic origin (Shapiro, 1975(Shapiro, , 1977)).Years of 25 cm or less of rain have occurred 14 times at Sacramento since 1849 and 14 times Psyche [March at Davis, Yolo County, since 1871.For both only the 1975/76 and 76/77 water years stand as consecutive severe doughts.At both stations 1930/31 and 32/33 were dry, separated by an average year, and at Davis 1917/18 and 1919/20 were dry, separated by a wet year.Is the long-run rainfall variance in central California adequate to select for a carryover response, or is it a preadaptive property of Pierids generally?Data on populations from other climates are needed.
It is dangerous to argue that because an aspect of the biology of an animal is adaptive in a particular situation, it evolved as a response to that situation in evolutionary time.In the case of P. n. microstriata at Gates it is not clear that a tenfold reduction in host biomass would have adversely affected fitness, had the insect attempted to reproduce.(Compare Murdoch, 1966 for a Carabid beetle case.)We simply do not know if intra-or interspecific compe- tition for food would have occurred.Because P. n. microstriata is usually out of phase with Dentaria at Gates, it uses that plant mainly at the beginning of the flight and in early years, and most of its eggs are placed on the phenologically better-matched Barbarea, even though Dentaria is more abundant.Its eggs are strongly conta- giously distributed, and it does not assess egg load on individual hosts, so that single plants sited such that females find them easily may receive overloads of eggs.It can thus be argued that with fewer and smaller plants intraspecific competition would have been aggravated--but we do not know if oviposition behavior might have been modified.P. n. microstriata is also positively associated with A. sara on individual hosts.This butterfly seems to assess its own egg load but not that of P. n. microstriata.The Pieris is a leaf-, the Anthocaris a silique-feeder.In 1977 there was no conspicuous reduction in numbers of A. sara adults or eggs at Gates, even though it is able to carry over (indeed, most captive pupae do); its 1978  performance was outstanding.A major reduction in host biomass, then, failed to adversely affect it, perhaps because it distributes eggs more evenly.
We do not know where the 1978 butterflies at Lang came from.Some may have been carryovers, but here the destruction of the 1977 egg crop may have been more apparent than real.Many hosts at Lang are inaccessible for censusing, and precisely these--on steep, wooded slopes--may have afforded more benign microcli- mates to eggs and larvae than those at and near canyon bottoms.In 1979] Shapiro Phenology of P. napi microstriata 9 most years 75-80% of potential hosts are censused at Gates, but only 50% at Lang.
There are two major modeling approaches to insect phenology.One, the physiological-time or degree-day approach, began with Shelford's (1927) study of the codling moth, Las_peyresiapomonella (L.) (Tortricidae).This is still the best-quantitated species (Riedl,  Croft and Howitt 1976; Riedl and Croft, 1978).Here the emphasis is on the development of predictive algorithms from empirical data.The other approach proceeds from Darwinian principles and is exemplified by the treatment of Levins (1969) or Cohen (1970) who derive strategies from survival and reproduction parameters.The two approaches are complementary: responses to real meteorologi- cal events are the proximate consequences of selection for a genetic blueprint of development.The fusion of the two approaches into a comprehensive theory of insect phenology will require data on how seasonality contributes to resource utilization, competition, and survival.

Table 1 .
Populations of Pieris napi microstriata at study sites.

Table 2 .
Precipitation at nearest weather stations to study sites.
of the egg-laying had been completed.Cold rains then began and turned to snow: on 5 May, 15 cm of snow fell; by 8 May this had increased to 33 cm.There were daily rain showers thereafter until 18