UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Food plant spacing and the dispersal tendency of the cinnabar moth larva Campbell, Barbara Jane 1975

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


831-UBC_1975_A6_7 C34_6.pdf [ 5.37MB ]
JSON: 831-1.0099933.json
JSON-LD: 831-1.0099933-ld.json
RDF/XML (Pretty): 831-1.0099933-rdf.xml
RDF/JSON: 831-1.0099933-rdf.json
Turtle: 831-1.0099933-turtle.txt
N-Triples: 831-1.0099933-rdf-ntriples.txt
Original Record: 831-1.0099933-source.json
Full Text

Full Text

FOOD PLANT SPACING AND THE DISPERSAL TENDENCY OF THE CINNABAR MOTH LARVA by BARBARA JANE CAMPBELL B.Sc, University of Alberta, 1970 L. es-Sciences Naturelles, Universite Claude Bernard, 1972 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in the Department of Plant Science We accept this thesis as conforming -to the required standard THE UNIVERSITY OF BRITISH COLUMBIA February 1975 In presenting t h i s t h e s i s in p a r t i a l f u l f i l m e n t of the requirements f o r an advanced degree at the U n i v e r s i t y of B r i t i s h Columbia, I agree that the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r reference and study. I f u r t h e r agree that permission for extensive copying of t h i s t h e s i s f o r s c h o l a r l y purposes may be granted by the Head of my Department or by h i s r e p r e s e n t a t i v e s . It i s understood that copying or p u b l i c a t i o n of t h i s t h e s i s f o r f i n a n c i a l gain s h a l l not be allowed without my w r i t t e n permission. Department of The U n i v e r s i t y of B r i t i s h Columbia Vancouver 8, Canada ABSTRACT The tendency of larvae of the cinnabar moth, Tyria , jacobaeae (L.)> to disperse from their food plant, before i t i s defoliated and before they are ready to pupate, was investigated in 13 subpopulations of ragwort, Senecio jacobaea L. Dispersal of larvae from their original food plant to a new food plant was assumed to be acted upon by selection. The purpose of this thesis was to examine the likelihood that larvae in any given subpopulation could locate a new plant, and to examine the dispersal behaviour those larvae displayed. The research hypothesis, relating dispersal tendency to plant spacing and dispersal success, was that larval dispersal should be favoured in subpopulations with closer plant spacing, where i t is more l i k e l y to be successful. Field measures of ragwort density and spacing, and larval distribu-tion, were made at each of the 13 subpopulations. The behaviour of larvae from each subpopulation was studied in the f i e l d , or on dispersal plots of variable plant spacing, built at south campus, U.B.C. The basic method was to start from a known number of larvae of a given age on a centre plant, record the number that had disappeared after a given number of days, and, of those missing, the number that had arrived on new plants in the v i c i n i t y . The f i r s t measure was called disappearance, and the second, dispersal success. Of the 13 subpopulations, 8 were classified as being closely spaced on the basis of a median rank of 6 separate measures of plant density and spacing. F i f t h instar larvae were found to disperse with a probability of success that varied from .74 when the probability was .33 that plant aggregations were separated by ) 1 m (the more closely spaced populations) to .44 when p ) 1 m was .75 (the more widely spaced populations). Thus the differential survival, necessary to any selection for different i i behavioural types, was demonstrated. And, in 2 out of 3 measures of larval distribution from the f i e l d , there was significant indication that larvae from widely spaced populations disperse less. However, disappearance, as measured on the south campus plots for larvae from areas of different plant spacing, offered no support to the hypothesis. These data were thought to be unreliable due to the influence of other factors on disappearance. Given the marked differential survival of dispersing larvae i n closely spaced and widely spaced plant populations, i t was concluded that selection could produce different behavioural types in these populations. The f i e l d evidence supports the hypothesis that dispersal tendency i s influenced by the probability of dispersal success. If the higher dispersal mortality associated with wider plant spacing selects against dispersal behaviour, then dispersal mortality may be constant in populations with different plant spacing. Dispersal mortality w i l l not act diffe r e n t i a l l y as a population regulating mechanism in populations with different plant spacing. Lack of a relationship between larval density and disappearance suggests that dispersal tendency is influenced less by larval numbers on the original food plant than by the probability of dispersal success. There was no indication, from mean egg batch size, that selection favoursd smaller egg batches and thus fewer overcrowded plants in those areas with wider plant spacing. i i i Table of Contents Page Abstract i Table of Contents i i i List of Tables . • v List of Figures v i Acknowledgements v i i General Introduction 1 History and Description of Study Sites 10 History Description 13 Field Sampling of Plant and Larval Density and Distribution Methods 16 Plant Density - Quadrat Sampling 17 Plant Density - Transect Sampling 19 Plant Distribution in Space 23 Egg Batch and Larval Distribution 26 Digression on an Index of Dispersion 27 Determination of Tyria Density from Egg Batch Distribution . . . . 28 Distribution of Larvae on Ragwort Stems 30 Larval Behaviour Introduction . . . 37 U.B.C. South Campus Experiments 38 Disappearance 43 Success 51 Mean/moves/larva/day 57 Multiple Moves 57 « Directionality of Successful Moves 61 Dispersal in Natural Tyria Populations 63 Dispersal Success and Plant Spacing - Summary 68 General Discussion of Hypotheses Relating Dispersal Tendency to Plant Spacing . . 74 iv Page Dispersal, Fecundity and Population Regulation of Tyria . . . . 78 Summary 86 Literature 88 Appendix - Egg Batch Order and Disappearance 91 V L i s t of Tables Page I H i s t o r y of Study S i t e s 12 I I D e s c r i p t i o n of Study S i t e s 14 I I I P l a n t Density from Quadrat Sampling 18 IV Conversion of Transect Values to Density 21 2 V Conversion of P l a n t s / t r a n s e c t to Plants/m 22 VI Median Ranking of Populations f o r P l a n t Spacing 24 V I I S e l e c t i o n f o r Rosettes, Stems, Multistems i n Egg Laying . . 29 V I I I T y r i a Density from Egg Batch D i s t r i b u t i o n 31 IX Indices of D i s p e r s i o n f o r D i s t r i b u t i o n s of Larvae 33 X F i e l d Measures of L a r v a l Behaviour 36 XI O r i g i n of Larvae Tested at South Campus 42 X I I P r o p o r t i o n Leaving by Geographic O r i g i n of Larvae . . . . . 46 X I I I 5th Disappearance and Centre P l a n t L a r v a l Density 49 XIV Success and P l a n t Spacing - A l l Larvae 52 XV Success and P l a n t Spacing - Geographic O r i g i n of Larvae . . 54 XVI Success and Geographic O r i g i n of Larvae 55 XVII Moves/Larva/Day and P l a n t Spacing 58 XVIII D i r e c t i o n a l i t y of S u c c e s s f u l Moves 62 XIX P r o p o r t i o n M i s s i n g A f t e r Two Days 66 XX Success and Distance Moved - Cultus Lake 67 XXI Success i n F i e l d T r i a l s 69 XXII Pupal Weights of D i s p e r s e r s and Non-dispersers 80 X X I I I Batch S i z e , Stem Biomass and P l a n t Spacing 81 XXIV Changes at Cultus Lake 1973 to 1974 83 94 XXV P r o p o r t i o n M i s s i n g , Environment Room Experiment -v i List of Figures Page 1 Indices of Larval Dispersion and Plant Spacing 34 2 Sketch of South Campus UBC Plots 39 3 Photograph of South Campus 40 4 Regression of 5th Instar Disappearance and Elapsed Time 1974 . . . 45 5 5th Instar Disappearance, Maximum Temperature and Elapsed Time, 1974 47 6 5th Disappearance and Centre Plant Larval Density 50 7 Dispersal Success and South Campus Plant Spacing 53 8 Mean Moves/Larva/Day and Plant Spacing 59 9 Multiple Moves and Plant Spacing 60 10 Photographs of Dispersal Boxes - Cultus Lake 64 11 Dispersal Success and Log^g p ) 1 m Plant Spacing 70 12 Dispersal Success and Plant Spacing 72 13 Photographs of Cultus Lake 84 14 Environment Room Dispersal Box 92 v i i ACKNOWLEDGEMENTS This thesis i s the result of the work and support of many people, as my use of the pronoun "we" in the text suggests. Sincere thanks to Rosemarie Iyer, Jayne Freitag, Pat Slobodzian, Kathleen Fry and Balbir Khunkhun, for their capable help with the f i e l d work, often seven days a week; to Fred Wilkinson and Dr. P. Harris of the Canada Department of Agriculture, Malcolm Neary of N.S.A.C., Truro, Nova Scotia, Dr. W.Nagel of Oregon State University, and Dennis Isaacson of Corvallis, Oregon, for showing us the locations of the Tyria populations and for their valuable information on the history of the sites; to Mrs. Marge Gossard for her hospitality in Fort Bragg, California; and to the people at the Institute of Animal Resource Ecology, UBC for their ideas and criticisms of my work, especially Dr. Wren Green, who, after ..all, started the cinnabar project and suggested my thesis topic.I was very grateful,too, to the members of my thesis committee, Dr. V.C. Runeckles, Dr. W.G. Wellington and Dr. A.J. Renneyfor their comments on the f i r s t draft of the thesis. It is however to my thesis supervisor, Dr. Judy Myers, I'hat I owe the largest vote of thanks for the constant encouragement, willing help and valuable advice she offered through a l l phases of the work - from planning the experiments to guiding my attempts to analyze the data. Without the support of such a willing supervisor I doubt that this thesis would have been possible. 1 GENERAL INTRODUCTION That animal populations survive over time in environments which are both spatially and temporally variable i s evidence of their success in evolutionary terms. As va r i a b i l i t y i s the rule in nature, any movement of animals either within local areas of a subpopulation, or between sub-populations, must play a key role in allowing a population to monitor this v a r i a b i l i t y and make best use of i t for long term persistence. Such movement is broadly defined as dispersal. However, within this broad definition, there i s controversy over the meaning of such terms as migration, 'goal-directed movement' and t r i v i a l movement to describe dis-placements undertaken for different reasons. Variability of a habitat may mean patchy distribution of a necessary resource within a local area, as well as a patchy distribution of local areas. It seems logical to me, therefore, to include within a definition of dispersal that behaviour which is directed toward finding the necessary resource, even though this has been dismissed as mere wandering or t r i v i a l movement by some (eg. Johnson 1969). Such movement is no less important to persistence on a local level than are longer term dispersal flights or migrations between subpopulations to persistence in a larger sense. I include within my definition of dispersal, therefore, movement which i s goal-directed and of a relatively short duration. v If dispersal i s important to the survival of populations, tendency of animals to disperse must be maintained by selection. When both the probability of surviving dispersal and the probability that any given habitat w i l l disappear are high, dispersal w i l l be favoured by individual selection. Conversely, with a low probability of surviving dispersal, individual selection w i l l oppose the behaviour and group selection may be 2 invoked to account for any " a l t r u i s t i c " tendency to disperse. Van Valen (1971) suggests that individual selection against dispersal . "may be mitigated somewhat by the favouring of dispersal within a local population. If local populations become extinct, new populations w i l l be founded by dispersers, and this i s the selective advantage for dispersal." We must know the probability of an individual surviving dispersal, the probability of a local population going extinct, and the variation in tendency to disperse among members of the population in order to under-stand the selection pressure acting to maintain dispersal, and i t s role in population regulation. I have chosen to investigate one aspect of the selection acting on dispersal in populations of a lepidopteran larva which feeds on one food plant. The system is that of the cinnabar moth, Tyria jacobaeae , and i t s foodplant, ragwort, Senecio jacobaea. Green's (1974) work on this herbivore/plant system suggested that some cinnabar larvae did disperse. He elaborated a hypothesis of population regulation after experimenting with the relationship between larval dispersal success and ragwort den-sity. It predicted that cinnabar populations in habitats of low density ragwort would regulate their numbers when dispersing larvae suffered high mortality. By the same logic, outbreaks of cinnabar moth could only occur i n habitats of high density ragwort, where dispersing larvae a l l survive and contribute to the next generation. Green's hypothesis is based on a short study of larvae from one Tyria population near Nanaimo, British Columbia. I found his ideas thought provoking, and decided to embark upon a more detailed investigation of cinnabar larval dispersal and ragwort distribution. A brief description of the natural history of cinnabar moth and rag-wort w i l l make clear their s u i t a b i l i t y for such an investigation. Tyria 3 jacobaeae is a moth of the family Arctiidae. The bright red and grey-black moth emerges from overwintering pupae through the months of May, June and July. Females are mated shortly after emergence, then lay several clusters of 30-50 eggs on the underside of broad leaves of d i f -ferent ragwort plants. The eggs hatch in about one week. Fi r s t and second instar larvae then feed in a large group near the hatched egg cases, and third instar larvae move together to the top of the plant. In the fourth and f i f t h instars, they spread themselves out over the entire plant and, very often, leave their original home plant for another ragwort. The larval instars vary in duration with the ambient temperature. In general, though, the f i r s t four instars are about 5 days long, and the f i f t h from 7 to 12 days long. F i f t h instar larvae which have attained pupation weight leave ragwort plants and move over the ground in search of pupa-tion sites. These sites are usually dark and enclosed, either in the punky wood of rotting logs, or under the s o i l and in the leaf l i t t e r . The ragwort plants are of two types: rosettes, or f i r s t year plants, and flowering stems or second year plants. It is not known how important seed is in the spread of ragwort, as new plants bud readily from the root crowns of stems, and much of i t s growth pattern can be associated with this vege-tative habit. This results in a clumped ragwort distribution. We have, therefore, in the cinnabar/ragwort system a clumped distribution of plants, a variable distribution of egg batches on plants, and larvae with a ten-dency to disperse over the ragwort population. The purpose of this thesis is to study variation in the tendency of fourth and f i f t h instar larvae to leave their foodplant before i t i s de-foliated and before they are ready to pupate; that i s , the variation in their likelihood to disperse. I assume that such variation in dispersal tendency is the outcome of natural selection. Whether or not selection 4 favours larval dispersal in any given subpopulation w i l l depend on the relative advantages of staying on the original food plant and of trying to locate a new food plant. The object i s , of course, to attain pupation weight, pupate successfully and contribute to the next generation. The factors to be weighed are thus the likelihood that the original food plant w i l l be overcrowded, and the likelihood that a new food plant can be successfully located. The f i r s t is an expression of Tyria density re-lative to plant biomass; and the second of larval search a b i l i t y and the spacing of the ragwort resource. This thesis i s concerned largely with the second of these factors - food plant spacing - although the role of larval density i s also investigated. The rate at which these two factors change between Tyria and Senecio generations - from one year to the next - in any subpopulation w i l l affect the nature and the extent of any selection acting on larval dispersal. Can a subpopulation change quickly from high Tyria and Senecio densities to low densities? If so, what is the corresponding change in the selection regime acting on larval dispersal from one generation to the next? These questions can only be answered by study of discrete sub-populations over several years. Such a study was unfortunately beyond the scope of this Masters thesis. I have chosen instead to examine the relationship of food plant spac-ing to both the dispersal tendency and the search a b i l i t y of cinnabar larvae. One aspect of this work is an attempt to quantify "low" and "high" ragwort densities with respect to dispersal success. How far can larvae travel successfully between ragwort plants? How different with respect to ragwort spacing are discrete ragwort subpopulations? By testing the behaviour of larvae from geographically distinct sub-populations for which I have measures of ragwort spacing, I can test the 5 prediction that larval dispersal tendency should be higher where ragwort plants are found close together (at "high" densities). In subpopulations where ragwort plants are widely spaced, mortality of dispersing larvae should be high enough to maintain selection against dispersal. These predictions do not take into account the plant spacing conditions ex-perienced by previous generations of larvae; nor do they deal with past or present conditions of larval density. They do, however, provide a useful starting point for consideration of the problem. I am also interested in the mechanism by which selection can act on variation in dispersal tendency in a cinnabar population. As mentioned earlier, I am working from the evolutionary tautology that larval dis-persal i s maintained in the population because i t conveys an advantage in enabling a population to persist. It is not clear what causes dispersal behaviour in cinnabar larvae. One could hypothesize a dispersal gene or genes whereby those larvae with the gene dispersed and those without i t did not. Or one could hypothesize a gene for "tendency to disperse under certain larval density conditions" whereby larval density on any one rag-wort plant could trigger the behaviour. By keeping the progeny of d i f -ferent females separate and testing their dispersal behaviour at different larval densities per plant, I can get an idea of genetic variation in dis-persal between females, and of the effect of density on dispersal tendency. A correlation between the phenotype of the parents and that of the offspring i s necessary for selection and evolution to proceed. However, the correlation need not always be due to chromosomal inheritance. Cyto-plasmic inheritance could be an important factor, as one would expect i f the determinant of dispersal were related to egg batch order. The fact that cinnabar moths lay their eggs in several discrete batches enables me to investigate the effect of egg batch order, or within female variation, 6 on dispersal tendency. Perhaps early laid clusters, born of a moth that had been a disperser as a larva would contain maternal factors that would give rise to dispersing progeny. This would be a system like Wellington (1965) suggested for the tent caterpillar, Malacosoma pluviale, where "unequal partitioning of the maternal food reserves during egg production produced offspring of d i f f e r -ent qualities." The necessary correlation between parent and offspring would be explained by the fact "that the differences in feeding rate and food capacity dis-played by the different types of females during their own larval stage affect the proportions of the various types of progeny per egg mass, as well as the v i a b i l i t y of consecutive groups of eggs within the mass" (Wellington 1965). Restated for the cinnabar moth, a moth from a dispers-ing larva might produce a range and quality of egg batches different from those produced by a moth that had been a non-dispersing larva. While considerable time and energy have been devoted to study of the cinnabar moth and to development of l i f e tables (Dempster 1961, Bornemissza 1966, Hawkes 1968, Isaacson 1973), i t was not u n t i l Green (1974) that the role of larval dispersal in the population dynamics of Tyria was examined in any depth. The cinnabar population at Weeting Heath in England studied by Dempster showed wide fluctuations in numbers over several years, lead-ing him to conclude that i t was not regulated to i t s food supply at a l l . "The upper size of a Tyria population is limited by density dependent com-petition for food." Van derMeijden (1970) f e l t too that the influence of ragwort on Tyria was considerable at high moth densities in the populations he studied in Holland. He observed that, through lack of food, larvae suffer the mortality of starvation, or are exposed to ant predation when they must leave defoliated food plants and search for more food. Thus the populations in both England and Holland showed wide 7 fluctuations in numbers from one year to the next. Whether or not this means they were not regulated i s debatable. Way and Cammell (1970) in describing aphid populations with unstable numbers say "while there i s the appearance of i n s t a b i l i t y , i t i s also a fact that the species is succeeding in violently changing circumstances; this implies the existence of self regulating mechanisms that are perhaps more sensi-tive than those possessed by species with so-called stable populations"^ It is clear though that cinnabar populations frequently exceed the carry-ing capacity of their habitat. Isaacson (1973) documented this for a population in Linn County, Oregon. While Dempster observed delayed re-duction in fecundity after density related crowding of larvae, Isaacson did not (pers. comm.). The reduction of fecundity before the carrying capacity was reached would constitute a true self-regulatory system -like that described by Shaw (1970) for aphids. The only evidence that a similar mechanism might be operating in a Lepidopteran species was pre-sented in 1966 when Klomp wrote that "the pine looper, Bupalus piniarius, larvae interfere (with each other) at densities far below the food capa-city of the pine trees, thus giving rise to an inhibition of larval growth and a decreased fecundity at higher levels. It is suggested that the larvae are stimulated to disperse by their encounters with other individuals. Moreover, there i s strong evidence that the v i a b i l i t y of eggs and larvae of the next generation is affected dis-advantageously." The relationship of larval dispersal to fecundity of the next genera-tion of moths has never been investigated for the cinnabar moth. It is clear that i f dispersing larvae contribute less to the following genera-tion than non-dispersing larvae, dispersal could be acting indirectly to regulate the population. The test for this is to compare pupation success, pupal size and fecundity of dispersing and non-dispersing larvae. It w i l l be noted that Gruys (1970), reporting further on the pine 8 looper work, retracts the hypothesis of self regulation, and says, i n -stead, that mutual interference of larvae i s "rather an adaptation for avoiding by dispersal (of the subsequent smaller, lighter moths) the effect of density related mortality." The problem remains open to i n -vestigation. HISTORY AND DESCRIPTION OF STUDY SITES The f i e l d work for this thesis was done in the summers of 1973 and 1974. Since neither Tyria nor Senecio are native to North America, the f i e l d work was of necessity limited to those locations where ragwort has spread accidently and where the cinnabar moth has subsequently been re-leased. Senecio jacobaea is considered a noxious weed by agricultural scientists and farmers alike. Ragwort has followed European settlement to North America, Australia and New Zealand. On this continent i t is re-stricted to the east and west coasts, except for a small stand at Guelph, Ontario (Harris 1974). It i s most often seen in poor pastures, along roadsides and in disturbed habitats like forest clearcuts. At high den-s i t i e s , ragwort can virtu a l l y take over pasture land and reduce the area in grass. Then, and even at lower densities, i t can be a problem to grazing cattle who develop cirrhosis of the liver after prolonged inges-tion of the alkaloids present in ragwort foliage. It i s not known just how poisonous these alkaloids are, how l i k e l y cattle are to eat ragwort i f i t is in their pasture, or how many cattle hkve actually died from ragwort poisoning. Palfrey eit a l . (1967) were able to correlate ragwort poisoning and a lack of minerals in cattle, suggesting that cattle eat ragwort, as they eat many other things, i f their mineral intake i s insufficient. This suggests that the best way to protect cattle from ragwort poisoning may be to ensure that they are getting adequate minerals, and therefore graz-ing in good pasture, rather than on the marginal land on which ragwort flourishes. Be that as i t may, some agricultural scientists and biolo-gists in North America have been interested in the biological control of tansy ragwort with cinnabar moth for the last twenty years, and have made numerous releases of the moth on the east and west coasts. 10 HISTORY The f i r s t release of Tyria was made in 1959 in a ragwort-infested f i e l d right-by the Pacific Ocean at Fort Bragg, California, with stock imported from near Paris, France (Frick and Holloway 1964). In 1961, more larvae were released, and, by 1963, the population had become well established. In that year, larvae from Fort Bragg were released at Comptche, about 20 miles from the coast on a ranch in the mountains. The Fort Bragg population was so large that on August 4, 1964, a f i e l d day was held and over 50,000 larvae were given out to people who took them back to Coos County, southern Oregon, among other places. Quite apart from the original Fort Bragg release, biologists at Oregon State Univer-sity had, in 1960, imported cinnabar moths from France and made an intro-duction on the property of a farmer named Silbernagel in the foothills of the Cascade Mountains, Linn County, north-central Oregon. In both counties the releases were quite successful. In Canada, releases of the cinnabar moth have been made since 1961 -f i r s t with stock imported from south Sweden, the France/Switzerland border, and Fort Bragg - and later with larvae from successful colonies in Canada (Harris 1974). In general, "the success in establishing colonies im-proved when stock from the region was used, instead of that imported and reared in the laboratory." In British Columbia, larvae were released at Abbotsford in the Fraser Valley in 1962, with l i t t l e success. In 1964, larvae from France were released on a newly cleared f i e l d near Nanaimo, on Vancouver Island. When their numbers failed to increase substantially, a supplementary release was made with Fort Bragg larvae. The population started to build up in 1968-1969, but by 1972-1973, showed underutiliza-tion of plant biomass and less than 100% defoliation (P. Harris pers. comm.). This is the f i e l d population that Green (1974) studied from 1968 to 1972. A f i e l d day was held here in 1968 or 1969, and, from Nanaimo, larvae were carried to farms in various parts of Vancouver Island. In 1971, a woman from a farm near Cultus Lake (in the Chilliwack area) re-quested a cinnabar release to control ragwort on her land, and larvae from Nanaimo were quite successfully introduced there (Harris 1974). On Canada's east coast, ragwort is a problem particularly in Pictou County, Nova Scotia, and on Prince Edward Island. Larvae were released at a site near Durham town, on the East River in Pictou County in 1963. This population had become well established by 1968, and served as a source for many other releases in north-eastern Nova Scotia and in Prince Edward Island. In the summer of 1973, I did f i e l d work on the populations at Nanaimo and Cultus Lake, British Columbia. By 1974, the Nanaimo sites had been bulldozed and the populations disturbed, although the Cultus population was s t i l l available for study. During June and July 1974, I travelled to California, Oregon and Nova Scotia to observe plant and larval density and distribution, and to bring larvae back to Vancouver for behaviour testing. Table I summarizes the study sites visited. The names given to different subpopulations within an area are li s t e d . Not a l l these releases have been monitored in the same way, or for the same period of time. Harris et^ a l . (1971) have regularly sampled plant density and estimated cinnabar numbers in Canada, and w i l l soon be pub-lishing a history of cinnabar releases on the east and west coasts (Harris 1974). Hawkes (1966) reported on the early history of the California populations, but no systematic ragwort or cinnabar sampling program has been undertaken there recently. In Oregon, the original Linn County re-lease went unmonitored for ten years, before Isaacson (1973) started his investigation of the cinnabar l i f e table. The Coos County populations 12 TABLE I HISTORY OF THE STUDY SITES Population Year of 1st release Cinnabar Origin Fort Bragg, California Comptche, California Linn County, Oregon Silbernagel Neal Creek-1 Neal Creek-0 1959 1963 1960 Paris, France Fort Bragg France (via rearing New Jersey) Coos County, Oregon 1964 Eckley Creek Eckley Creek H i l l Powers Catching Creek Nanaimo, British Columbia 1964 Topfield Junkyard Pylon Cultus Lake, British Columbia 1971 Field 1 Field 2 Stickum Field Fort Bragg Delemont, Switzerland: Fort Bragg Nanaimo Durham, Nova Scotia Mackinnon Cowfield Sylvester 1963 France 13 also received l i t t l e attention u n t i l 1970, when Nagel and Isaacson (1974) started following them. The quantitative data on these popula-tions are thus patchy, but they provide some information about the areas I studied, particularly with regard to plant density. DESCRIPTION Tansy ragwort grows well in the disturbed habitats that result from man's activities - frequently on agricultural land and logged clearcut areas.. The important environmental determinants of ragwort success, apart from the effect of Tyria, seem to be weather factors and competi-tion with grasses, according to a preliminary report from Van der Meijden (1970). Table II l i s t s the study areas with a brief description of each. Most of the study sites i n any given geographic area were separated from each other by at least a couple of miles. The exception i s Cultus Lake, where Field 1, 2 and Stickum f i e l d are a l l close together. The original Tyria release was made in Field 1, and the moth has not yet spread to Field 2 and Stickum, thus providing an excellent opportunity to observe the effect of larvae on plant density, height etc. by comparing the fields. 14 TABLE II DESCRIPTION OF THE STUDY SITES Fort Bragg Comptche Silbernagel Neal Creek«0 Neal Creek -1 Powers (Site I, Nagel 1974) Eckley Creek H i l l (Site II,ibid.) Eckley Creek (Site III, ibid.) Catching Creek (Site IV, ibid.) Cultus, Field 1 Cultus, Field 2 Cultus, Stickum Nanaimo, Topfield Abandoned pasture at sea level; no grazing at present; dense plant cover of grasses, ferns, some ragwort. Improved pasture; overgrazed; cropped grass, ragwort, poison oak. Unimproved cleareut, presently used as pasture; rotten logs, stumps, grasses, small Douglas f i r , ragwort. Unimproved cleareut in Cascade foot-h i l l s ; enclosed by Douglas f i r ; shrubs, thi s t l e , ragwort. Unimproved, infrequently used roadway; off logging road, Cascade fo o t h i l l s ; no grazing; ragwort, grasses, herba-ceous plants. Improved pasture; 700 f t . elevation; grasses, forbs, ragwort; no grazing. Large natural prairie-like pasture; 650 f t . elevation; grasses, forbs, ragwort. Cleared pasture; 160 f t . elevation; grasses, forbs, poison oak, ragwort; heavy grazing. Semicleared pasture; 500 f t . elevation; same plant cover as Site III above. Unimproved cleareut on ridgetop; 100m by 25 m; light grazing pressure; dense grasses, ferns, small alder trees; surrounded by Douglas f i r , hemlock, Western red cedar. East of Field 1; separated from i t by 10 m of trees and shrubs. South of Field 1; separated from i t by 15 m of trees. Large unimproved pasture on ridgetop; logged in the '50's; logs and brush piles; quite heavily grazed; sparse grasses, ragwort. 15 TABLE II continued Nanaimo, Junkyard Nanaimo, Pylon Mackinnon Cowfield Sylvester Improved pasture one-half mile from Topfield; on lower ground; overgrazed by cattle; grasses, nettles, ragwort, shrubs. Improved pasture on another ridgetop; one mile from Topfield; occasionally grazed; grasses and ragwort. Improved pasture; occasionally grazed; grasses, thistles, ragwort. Improved pasture; very overgrazed; grasses and ragwort; sea level. Unimproved land, ridgetop; no grazing; grasses, ragwort, th i s t l e , shrubs. 16 FIELD SAMPLING OF PLANT AND LARVAL DENSITY AND DISTRIBUTION METHODS Information describing ragwort density and distribution, egg batch size and distribution, and larval numbers and distribution can be col-lected by a f i e l d sampling program in which the sample unit i s an area of ground. In 1973, data of this type were collected at Nanaimo from two 20 f t . square (6.15 m square) grids of 10 20-by-2-foot strips, one at Topfield and one at Pylon. This method restricted sampling to a very limited area of each f i e l d , and did not do justice to the heterogeneity of the area. That summer at Cultus Lake, we collected similar data by randomly choosing one-half meter square areas with a system of number co-ordinates, after surveying Field 1 and laying out a grid. Sampling proved to be very time consuming because of the high number of rosettes, and the dense ground cover. In 1974, we sampled 1 meter x 15 meter strips of land at Fort Bragg, examining every plant within the given sur-face area. It soon became obvious that this was as time consuming as i t had been at Cultus Lake the previous'year. We therefore decided upon a quicker sampling method; arbi t r a r i l y choosing a transect site, then laying down a 15 meter measuring tape in a straight line over the area to be sampled. Each plant that touched the tape was recorded as being on the transect. For each transect plant, we located the nearest neighbour plant, and measured the distance in centi-meters between the two. For both transect plant and neighbour plant the following data were recorded: 1. Type of plant - rosette, single stem or multi-stem. 2. Plant biomass - wet weight in grams of the entire plant cut off at ground level. 3. Number and size of egg batches, both hatched and unhatched. 4. Total number of larvae of each instar found on the plant. 5. Extent of defoliation. In some cases, the nearest neighbour to a transect plant was a plant which was also on the transect. If so, both plants were called recipro-cals and identified as such so that they would not count twice when the analysis was done. The position in meters along the length of the 15 m tape was recorded for each transect plant, except for those done in California. This transect method was used in California, Oregon, British Columbia and Nova Scotia and forms the core of my descriptive data from the f i e l d sites. We transferred the information to computer cards, and analyzed i t with a program written by Dr. J. H. Myers. Depending on the time re-quired per transect in different study sites, we did between 1 and 9 transects at each place. The data presented, unless otherwise stated, combine a l l transects in any one study site. PLANT DENSITY - QUADRAT SAMPLING The data collected in 1973 and at Fort Bragg in 1974 from the grids or quadrats are presented in Table III. As can be seen, only the data from Cultus Lake were collected from small enough unit areas (one-half meter squares) at 39 randomly chosen locations within Field 1, to provide a really accurate measure of the distribution of ragwort. In the other three areas, the sample area was too large in comparison with the 'grain' of the habitat. Nevertheless, the estimates made at Nanaimo, correspond f a i r l y well with those of Green (1974) who, in 1971, measured a density 2 2 of 3 stems/m at Topfield, and 6 stems/m at Pylon. The data from Cultus Lake are suitable for an estimate of the dis-2 tribution of stems in randomly selected .25 m units. I tested the f i t TABLE III PLANT DENSITY FROM QUADRAT SAMPLING No. Sample Area Sample ' A 2 + 2 + * Units Unit Ros./m - S.E. Stems/m - S.E. Plants/m' + 26 + Nanaimo'Pylon 10 3.9 sq.m 8. ,28 1. 4, .20 .80 12. .48 Nanaimo•Topf ield 10 3.9 sq.m 16. ,15 + 1. 85 2. .30 + .65 18, .45 Cultus Field 1 39 .25 sq.m 95. ,25 + 10. 78 10, .56 + 1.46 105. .81 Fort Bragg (1974) 5 3 @ 15 sq.m 14. ,01 + 6. 98 10. .19 + 4.62 24. .20 1 @ 3 sq.m 1 @ 5 sq.m A Rosettes 2 '• * Ros./m & Stems/m' of the observed distribution to the expected values of a negative binomial distribution, finding a mean/variance ratio of 2.167 for a mean number of 2 stems per .25 m of 2.67. The observed distribution f i t s the theoretical negative binomial distribution: and the k value is 1.44 (p=.05). Accord-ing to Southwood (1971) "the smaller the value of k, the greater the ex-tent of aggregation, ... a large value (over about 8) indicates that the distribution i s ... virtu a l l y random." Thus I conclude that the d i s t r i -bution of ragwort stems at Cultus Lake Field 1 in 1973 was strongly clumped. PLANT DENSITY - TRANSECT SAMPLING It may appear that by sampling plants along a linear transect, rather than plants within an area of ground, I have prevented the collection of just that information in which I said I was interested; plant density and distribution. I was certainly not able to measure these parameters directly, but could estimate them by comparing transect values with those values reported in the literature by other workers. The usual approach to plant density in studies of cinnabar/ragwort interaction has been to use 2 the number of stems/m . By comparing our measured stems per transect with 2 the reported stems/m for identical study sites, I was able to get an es-timate of how closely one approximates the other. Five of the seven possible comparisons were made using the density data reported by Isaacson (1973) and Nagel and Isaacson (1974). In no case were both values of a pair of data to be compared collected in the same year, and the estimates are based on a much larger number of meter square quadrats than transects. Transect positions were chosen arbit-r a r i l y , but areas having no plants were undoubtedly under-represented. The usual method was for two people to walk around with their eyes closed 20 and to place the meter tape on the ground without looking at the amount of ragwort in that particular place. This method was complicated in areas with poison oak. The method used by Nagel and Isaacson was to establish two transects perpendicular to each other and crossing at the original point of release of the moths. Quadrat samples were then taken every 100 feet along each transect. 2 A correlation of stems/m and stems per transect for a l l seven pos-sible pairs (Table IV) is not significant (r = 0.28 p (.05). i f however one eliminates the Eckley Creek H i l l pair, where the transect estimate differs^vastly from the one given by Nagel, the correlation becomes signi-ficant (r = 0.87, p ^ .05). I think this i s j u s t i f i e d , given the problems of choosing transect positions, and the fact that so few transects were done at Eckley Creek H i l l . From a regression line derived from the six 2 pairs of data, an estimate of the stem density per m can be made from the transect data (Table IV) . A similar analysis was performed using the data available for f i r s t year plants and for the total plants (stems and ros-ettes Table IV),. An estimate of the densities of plants in a l l the subpopulations was derived from an equation based on the analysis of both stems and rosettes 2 (Table V). Van der Meijden (1970) used an estimate of biomass/m as a measure of ragwort resource available to Tyria. Peter Harris (pers. comm. to J. Myers) suggests that since the rosettes are an important source of available food material, an estimate of plant biomass based on both ros-ettes and stems i s superior to any estimate based on stems alone. The proportion of rosettes in each of the ragwort populations is given in Table V; as well as the proportion of biomass represented in rosettes. It can be seen that frequently both proportions are high. There was no cor-relation between the estimated density of plants and either the proportion 2 1 TABLE IV CONVERSION OF TRANSECT VALUES TO DENSITY Study Site Ros./ Ros/ Stem/ Stem/ Total Plants Total trans. m^  trans. m^  / trans. Plants/m' Fort Bragg 5. ,23 14. ,01 6. ,5 10. ,2 11. ,0 24. ,0 Cultus ' 1 37. ,00 95. ,30 5. ,7 10. ,6 42. ,7 105. ,9 A Eckley Creek 3. ,7 5. .3 2. ,6 1. ,2 7. ,0 6. ,5 A Eckley Cr. H i l l 3. .5 9. .4 * * 18. .5 9. .8 Catching Creek A 7, .5 13. ,2 0. .5 0. ,4 8, .0 13, .6 Powers A 12, .0 11. .2 . 5. ,0 2. ,8 17, .0 14, .0 Silbernagel A 1, .0 0. .6 0.97 2.0 0.87 1.34 0.93 2.07 A Density data from Nagel and Isaacson (1974) and Isaacson (1973) * Data eliminated from calculation of r £ xy o Slope based on equation b = 2 — which forces the line through the origin ^ x 22 TABLE V ESTIMATED PLANTS / m FROM PLANTS / TRANSECT Study Site Proportion Prop. Biomass Plants/ Estimated Rosettes in Rosettes Transect Plants/m 2 Fort Bragg .49 .21 11 .00 + 2. 64 22 .8 Comptche .12 .007 9 .88 + 0. 97 20 .5 Silbernagel .89 . .60 8, .60 + 3. 07 17 .8 Neal Creek*1 .66 .20 27, .80 + 8. 83 57 .5 Neal Creek ••.0 .24 .07 1, .88 + 0. 53 3 .9 Eckley Creek .59 .19 7, .00 + 1. 71 14 .5 Eckley Creek H i l l .28 .04 18, .50 + 0. 49 38 .3 Powers .76 .25 17, .00 35 .2 Catching Creek .97 .66 8, .00 + 6. 99 16 .6 Cultus Lake-l .87 .62 42, .66 + 4. 09 88 .3 Cultus Lake-2 .71 .25 18. ,00 + 1. 99 37 .3 Stickum .65 .22 44. ,50 + 3. 49 92 .1 Mackinnon .49 .09 16. ,50 + 1. 44 34 .2 Cowfield .56 ;16 5. .00 + 1. 45 10 .4 Sylvester .78 .24 12. ,80 + 1. 65 26 .5 r (prop, rosettes and density) =0.27 r (biomass rosettes and density ) = 0.21 23 of plants which were rosettes or the proportion of biomass i n rosettes. PLANT DISTRIBUTION IN SPACE As described earlier, the distribution of stems at Cultus Lake 1973 was found to be quite clumped, f i t t i n g the negative binomial distribution. Using Lloyd's index, Nagel and Isaacson (1974) report that the four Coos County, Oregon, plant populations are also aggregated. I think i t i s safe to assume that a l l ragwort populations are clumped in space due to their reproductive biology, and perhaps to the effects of sustained Tyria attack (Nagel and Isaacson 1974). Plant density i s an expression of three aspects of this aggregation -the mean number of plants per aggregation (R. Jones, pers. comm.), the mean distance between plants in an aggregation, and the mean distance be-tween aggregations. It i s these three factors which together determine the distance a dispersing larva w i l l have to travel before encountering a plant. The a b i l i t y of larvae to travel different distances w i l l actually de-termine the effective size of an aggregation, or the "grain" of hetero-geneity which is meaningful to any discussion of dispersal mortality. However, the measured mean distance between a transect plant and i t s nearest neighbour should indicate the average spacing of plants within an aggregation as observed in the f i e l d (Table VT). In 10 of the 13 subpopu-lations studied, the mean was less than 30 cm, and in only one case (Neal Creek*0) was i t higher than 35 cm. The measures of plant spacing most relevant to dispersing larvae w i l l be the likelihood of completing dispersal within an aggregation, and, for those that leave an aggregation, the distance from the edge of one aggrega-tion to the edge of the next. The data for the latter come from measures TABLE VI MEDIAN RANKING OF POPULATIONS FOR PLANT SPACING P > lm p> 2m p> 5m Plants/m^ x Dist. Between Plants cm x Dist. Between Clumps m Median Rank Neal Creek*0 .75 .75 .25 3.9 68.4 5.6 1 Cowfield .59 .38 .09 10.4 31.8 3.9 2 Eckley Creek .54 .32 .03 14.5 28.5 3.3 3.5 Sylvester .40 .26 .04 26.5 19.6 2.9 4.5 Mackinnon .39 .06 .02 34.2 24.5 2.2 5.5 i-17.9-.45 x=34.56-8.70 x=3.58-.58 Cultus'2 .33 .07 0 37.3 17.1 1.7 8 Powers .31 .06 0 35.2 21.4 1.1 8 Silbernagel .27 .12 0 17.8 15.4 2.1 7.5 Eckley Creek H i l l .24 .09 0 38.3 32.6 1.5 9.5 Catching Creek .21 .15 0 16.6 13.1 . 3.4 7.5 Neal Creek*1 .07 .03 .009 57.5 9.8 1.7 11 Cultus'1 .05 .02 0 88.3 10.6 1.4 12 Stickum .05 .01 0 92.1 10.4 1.2 13 x=47.89-10. 29 x=16.30-2.72 x=1.76-.26 t = -2.1584 t = 2.4197 t = 3.2808 d.f. = 11 d.f. = 11 p<.05 d.f. = 11 p <.01 25 of transect plant spacing along the 15 m transects, where distinct aggre-gations were defined as plants separated by more than 30 cm. From the frequency distribution of the distances between aggregations, I calculated the probability that the distance from the edge of one aggregation to the next would be greater than 1 m, 2m, and 5 m respectively. The probabi-l i t i e s are in Table VI along with values for plants per meter squared, mean distance in centimeters between transect plants and their neighbours, and the mean distance in meters between aggregations. If the various sub-populations are ranked for each of these six measures, and the median of the six ranks i s calculated, the median rank values listed in TableVI are obtained. The populations can be divided by their rank into 'more widely spaced' (Neal Creek-0, Cowfield, Sylvester, Eckley Creek, Mackinnon) and 'less widely spaced' (the remaining eight). Of this second group, Neal Creek'1, Cultus*1 and Stickum are consistently the most closely spaced. Mean plant density i s not significantly different between the widely spaced populations (17.9 - .45) and the more closely spaced ones (47.9 - 10.29). Thisjlack of difference shows the danger of using density alone as the measure of spatial v a r i a b i l i t y in plant distributions. For both the distances between plants, and between aggregations, widely spaced populations have significantly larger means than closely spaced populations (p <( .05 and p <( .01 respectively). Thus i t would seem that the wide spac-ing of some Senecio populations i s a product of greater distances both within and between aggregations; and not solely of one or the other. For testing the hypothesis that larval dispersal should be lower in widely spaced plant populations, I propose to use the values for the pro-bability that spacing between aggregations i s greater than 1 meter. Un-fortunately this measure does not reflect the likelihood that a larva w i l l complete dispersal within an aggregation. However, the ranking of 26 populations according to this measure is most like that of the median rank. Furthermore, as w i l l be seen later, larval dispersal success de-creases significantly when plant spacings are greater than 1 meter, but does not change further with wider plant spacings. EGG BATCH AND LARVAL DISTRIBUTION ON PLANTS Information about the number and size of egg batches and the number of larvae of different instars present per plant i s available from both 1973 and 1974 data. In 1973, the populations at Nanaimo were sampled three times through June and early July. In 1974, study sites were sampled at best over a period of a few days, and often just on one day. The data available from the different study sites depend largely on the age of the Tyria population at the time of sampling, since not a l l instars were pre-sent in every population. For analysis I considered the distribution of eggs and larvae on both transect plants and their nearest neighbours. A l l the transects in any one area were combined. I t a l l i e d an estimate (keeping rosettes, stems and multi-stems separate) of the proportion of plants that had egg batches, both hatched and unhatched, and larvae of different ages. I also t a l l i e d the frequency distribution of egg batches on rosettes, stems and multi-stems; and of numbers of larvae of different instars on each of these three plant types. I hoped to calculate an index of dispersion which would enable me to compare behaviour of cinnabar larvae between study sites. For those plants with larvae of the 3rd, 4th and 5th instars, I calculated propor-tions having 2 or less larvae. The change in proportion with changing age of the larvae should give another indication of dispersal behaviour of different instars, and at different study sites. I was able to relate some of these measures of larvae present per plant 27 to the biomass present per plant, for the data collected in 1974. The computer program t a l l i e d the proportion of plants that were already de-foliated at the time of sampling. It also compared, for transect plants with 4th and 5th instars, whether the neighbour plant would be more or less crowded than the plant the larvae were on. Digression on an Index of Dispersion Considerable thought was given to choosing an index of dispersion that was best suited to the data. The ideal index, as described by E l l i o t (1971) should provide real values over the range from regularity through randomness to contagion. It should not be influenced by the size of the sampling unit, number of sampling units, sample mean, or total number in sample. It should be easy to calculate and should enable differences to be tested for significance. No index satisfies a l l these c r i t e r i a . The k of a negative binomial as an index of dispersion requires agreement with the negative binomial, increases with the mean, and is not independent of the size of the sample unit (Southwood 1971). It i s , however, independent of the number of sample units (N. Gilbert, pers. comm.). In terms of our sampling program, the sample unit is one ragwort plant. The size of the sample unit is the biomass represented by one plant. The number of sam-pling units i s the number of plants sampled in any one study site. The sample mean is the mean number of egg batches and larvae of different i n -stars per plant, while the total number in a sample i s the total number per plant. I decided to use the k value for those distributions that f i t the negative binomial, with the following precautions. I made the size of the sample units as similar as possible by keeping the data for rosettes, stems, and multi-stems separate. I compared k values only where the means were similar; that is only among populations of the same age and Tyria density. 28 These precautions restrict sweeping comparisons, but make the results more reliable. Determination of Tyria Density from Egg Batch Distribution It i s necessary to classify the populations sampled as to Tyria density to avoid making unwarranted comparisons between indices of disper-sion. Since both hatched and unhatched egg batches were recorded, the distribution of egg batches on both rosettes and stems is a good indica-tion of density as i t is independent of the age of the population. The mean number of egg batches laid on stems within a ragwort population i s always higher than that laid on rosettes. This selection by female moths in egg laying behaviour for stems over rosettes has been documented by other workers (Green 1974). It appears that female moths lay their eggs independently of each other. That i s , they do not select against plants that already have an egg batch when choosing an egg laying site. They also choose large plants over small ones. This selection of large plants is quantified in Table VII. Because of the preponderance of egg batches and thus larvae on stem plants, at least before larval dispersal occurs, I shall confine further discussion of egg batch and larval distributions to stem plants. Whether or not the egg laying behaviour of cinnabar moths results in a clumped distribution of egg batches on stems depends upon the density of the moths vis a vis the density of ragwort stems. The observed distribu-tions of egg batches on stems are either clumped, and f i t the negative b i -nomial, or are not clumped at a l l but random. Given the egg laying be-haviour of the cinnabar moth, the f i r s t case implies a higher density of moths to plants where i t is very l i k e l y that more than one moth w i l l lay eggs on most of the stems in the population. The second case implies a TABLE VII SELECTION FOR ROSETTES-STEMS • MULTI-STEMS IN EGG LAYING BEHAVIOUR Study Site n Egg Batch Prop.Ros. in Pop. Prop.Egg B. on Ros. Prop.Stem in Pop. Prop.Egg B. on Stem Prop. Multi-st. Prop. Eg] Oh Mult: Fort Bragg 111 .49 .24 .50 .68 .01 .08 Comptche 84 .12 0 .83 .68 .05 .32 Silbernagel 27 .89 .52 .11 .37 .01 .11 Neal Creek*1 111 .66 .33 .34 .67 Neal Creek*0 200 .24 .09 .76 .91 Eckley Creek 97 .59 .15 .41 .85 Eckley Creek H i l l 87 .28 .03 .72 .97 Catching Creek 20 .97 .60 .03 .40 Powers 48 .76 .38 .24 .62 Cultus"1 12 .87 .42 .13 .58 Cultus"2 1 .71 0 .27 0 .02 1.00 St ickum 3 .65 .33 .35 .67 MacKinnon 21 .49 .05 .50 .95 .01 0 Cowfield 42 .56 .21 .41 .62 .03 .17 Sylvester 20 .78 .30 .21 .45 .01 .25 Pylon (1973) 28 .56 .18 .44 .82 30 very low density of moths per stem, since the probability that two egg batches w i l l be laid on the same stem i s low. If each moth lays her egg batch on a different stem, the resulting egg batch distribution w i l l be random. We can therefore divide the populations into those which do not have a clumped egg batch distribution, and those which do, separating these latter populations further into classes based on mean numbers of egg batches/plant (Table VIII). The negative binomial series has two parameters, the mean and an ex-ponent k which is a measure of dispersion or aggregation. Successive terms of the distribution are obtained by expansion of the expression (l -p)k where p = and q = 1 + p (Waters 1959). Generally, values of k are in the region of 2; as they become larger the distribution approaches randomness (k = 8) while fractional values indicate an approach to the ex-treme aggregation of the logarithmic series. The k values of Table VIII do not vary greatly within groups II, III, and IV. The fact that the egg batch distributions f i t the negative bino-mial provides, however, a starting point from which to consider the larval distributions. Distribution of Larvae on Ragwort Stems The classification of populations as to Tyria density based on egg batch distribution and mean egg batches per stem is retained for the com-parison of indices of dispersion of the larval distributions. The i n -dices for 3rd, 4th and 5th instar dispersion were calculated for those populations where the instars were present in sufficient numbers. Since the k values were again a l l i n the region of 2, agreement with the nega-tive binomial is indicated simply by k. Random distributions are 31 TABLE VIII TYRIA DENSITY FROM EGG BATCH DISTRIBUTION Egg Batch Dist. Not Clumped Group I Stickum Mackinnon Sylvester Pylon(Nanaimo) TopfieId(Nanaimo) Index of Dispersion x Egg B./Stem i < < .04 .41 .39 .19 .06 Egg Batch Dist. Clumped Group II x K. 1/stem Fort Bragg Comptche * Cultus Lake Neal Creek*1 Cowfield Junkyard .44 .49 1.20 .63 1.62 1.27 .53 .43 .63 .87 .74 .71 Group III x = 1-5/stem Silbernagel Eckley Creek Eckley Creek H i l l Powers .66 .62 1.34 .57 1.11 2.39 1.62 3.75 Group IV x > 5/stem Neal Creek*0 c) Catching Creek 2.52 7.32 8.00 c) only 1 stem * data from PLOT A Field 1 substituted for transect data 32 indicated by ( (less clumped than the negative binomial), while } means that a distribution i s more clumped than the negative binomial (Table IX) . While distribution of egg batches most often f i t s the negative binomial, the distribution of 3rd instar larvae is usually overclumped. This is because the recording unit for egg batches i s one egg batch, whereas for 3rd instar larvae i t is one larva. The distribution of individual eggs w i l l always be overclumped because the eggs are laid in clusters. Given that the distribution of eggs i s always overclumped, the fact that a larval distribution f i t s the negative binomial at a l l is in i t s e l f indica-tive of a move toward randomness, which i s indicative of dispersal ten-dency or of high mortality. Using only k as a symbol to indicate agreement with the negative b i -nomial eliminates the necessity to group the data by populations with similar Tyria density. The symbols from Table IX can be plotted in Figure 1 where the plant populations are classified as to the probability that two plants are separated by more than 1 meter as a test of the hypothesis that there should be more larval movement where aggregations are closer to-gether . One would expect that the overclumped distributions of 4th and 5th instars would be found to the right of the figure, where the probability of having to travel more than 1 meter is high. And that random distribu-tions indicating much movement would be found where this probability i s low. Both results are observed, although the sample size for a random distribution i s only 1. Overclumped distributions overlap the two which f i t the negative binomial, but they are located to the right of the figure as predicted by the hypothesis. Using the classification of plant populations from p ) 1 m, we can 33 TABLE IX INDICES OF DISPERSION FOR 3 rd , 4th AND 5th INSTAR LARVAE 3rd 4th 5th Group I Stickum MacKinnon Sylvester Pylon (Nanaimo) Topfield (Nanaimo) k > > > > k > Group II Fort Bragg Comptche Neal Creek'1 Cultus Field'1 Cowfield Junkyard > > < < > k < Group III Group IV Silbernagel Eckley Creek Eckley Creek H i l l Powers Neal Creek'0 Catching Creek > > k > k k > > k k 34 > >> > > K < K K < g Q ti 3 = ) 0 < > O O uJ u u">Q- 2 C n LD U Z i i i i i i 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 3 0.9 1.0 P L A N T S P A C I N G P > 1 M . 4 T H I N S T A R S : > < K 5TH I N S T A R S :>< K F I G . 1 . I N D I C E S O F L A R V A L D I S P E R S I O N A N D P L A N T S P A C I N G 35 now compare other measures of larval activity between populations of different plant spacings. The f i r s t of these i s the proportion of 5th instar groups on stems that have 2 or less larvae. The higher the level of dispersal activity the greater should be the proportion of these small groups of larvae per plant (Table X). A t-test indicates that there i s no significant difference between the mean proportion of small groups in widely spaced and closely spaced plant populations. We have compared larval consumption of foliage to ragwort biomass of rosettes, stems and multi-stems, assuming that total consumption / larva during the last two instars is 5 grams wet weight. The total number of 4th and 5th instars multiplied by 5 grams was compared to the biomass of both the transect plant and i t s nearest neighbour. The proportion of transect plants from which larvae should move was calculated where the criterion for moving was that the transect plant biomass did not equal the estimated ragwort consumption, but that of the neighbour plant did. This proportion should be higher in widely spaced ragwort populations, where a low dispersal tendency would result in many overcrowded plants. The proportion of larvae that should move i s , as predicted, significantly + higher for widely spaced (.47 - .04) than for closely spaced populations (.27 - .02). 36 TABLE X FIELD MEASURES OF LARVAL BEHAVIOUR Study Site Prop. Groups with < 2 larvae Prop. 4ths and n n 5ths should move 4ths 5ths Neal Creek-0 Cowfield Eckley Creek Sylvester Mackinnon .59 .56 .51 .27 .42 .39 .50 .44 .42 .59 155 112 143 173 178 27 31 89 142 85 x - .47 - .06 x = .47 - .04 Silbernagel Catching Creek Cultus-2 Powers Eckley Creek H i l l Neal Creek*1 Cultus'1 Stickum ,68 .33 1.00 .67 .64 ,26 ,30 .25 41 0 0 66 6 0 21 13 74 0 0 31 1 0 15 3 x = .66 - .11 x = .27 - .02 t = -1.6057 d.f. = 8 t = 4.0811 p < .005 * data from J. Myers LARVAL BEHAVIOR 37 INTRODUCTION Information regarding the tendency of larvae to disperse and their a b i l i t y to locate ragwort plants was sought through three different ex-perimental approaches. Most of the data came from experiments done at UBC where larvae from different locations were tested on plots of d i f f e r -ent ragwort densities in both 1973 and 1974. The second approach was to observe the behaviour of larvae in their own home ragwort population by marking them with paint and following them for as long as possible. I did this at Cultus Lake in 1973, and at Cultus Lake, Comptche, Linn County, Oregon and Nova Scotia in 1974. The third method was a purely a r t i f i c i a l one. Larvae from Oregon and Cultus Lake were reared in a constant en-vironment room and tested in a plastic dispersal box to observe their tendency to leave a food supply. The type of data gathered from each of these three methods was similar. In each case, starting from a known number of larvae of a given age on a plant, I measured the number that had disappeared from the o r i -ginal plant after a certain number of days or hours. Of those larvae miss-ing from the original plant, I counted the number that had appeared on other ragwort plants in the v i c i n i t y after the same number of days. The f i r s t measure I called disappearance; and the second, dispersal success. The disappearance value was calculated at the end of the instar or, in the case of 5th instars, just before the estimated date of pupation. Because the success measure contains no indication of larvae that move from plant to plant after they have successfully dispersed once, I also calculated the mean number of moves per larva per day. I recorded the distance between the original plant and any plant to which a larva moved, 38 as well as the compass direction of the movement. Wherever possible, I used larvae of a known female parent and of a known egg batch, so that I was able to look at variation in both disappearance and success within the progeny of one female, and between the progeny of different females. U.B.C. SOUTH CAMPUS EXPERIMENTS These experiments occupied most of the time devoted to testing larval behaviour in the summers of 1973 and 1974. In 1973, we laid out plots A-R as shown in Figure 2. They were l e f t in place over the winter. The plots were of three sizes; 3 m square, 6 m square and 9 m square. Each plot had 9 ragwort plants numbered and arranged according to Figure 2, which gave ragwort spacings of 1 m, 2 m and 3 m respectively. In 1974, plots S and T were added, being 15 m square and giving a ragwort spacing of 5 m. The plots were fenced with 1/4 inch plywood, 6 inches high, held in place with wooden stakes and interlocking corners. Clear vinyl strips 6 cm wide were sprayed on one side with fluon (similar to teflon). The strips were held at right angles to the inside of the fence, fluon-sprayed side down, with staples and much masking tape. The purpose of the fluon-sprayed plastic was to keep dispersing larvae from crawling out of the plots, as they cannot crawl across fluon, especially when they are upside down (Dr-TJCram, pers. comm.). Ragwort plants were transported from Abbotsford in May of '73 and '74 and planted in the plots in the described pattern. We planted clover as ground cover between the ragwort plants, and devoted considerable time and effort both years to keeping the clover from getting too t a l l , and from growing over the plot fences. These ef-forts were often in vain as the clover grew so quickly. Figure 3 shows a photograph/ of the south campus plots taken i n August 1974. 39 TRUE NORTH DETAIL OF PLOT WITH I M . SPACING TRUE NORTH 1M.SPACING 2M.SPACING 3M.SPACING F IG.2 .SKETCH O F SOUTH C A M P U S U B . C . P L O T S 40 I PLOT H ; 1M SPACING, AUGUST 1974 F I G . 3 . P H O T O G R A P H O F S O U T H C A M P U S The method of testing larval behaviour was simply to put a known number of larvae on the centre plant (C) and then check every day to see how many were s t i l l on the centre plant, and how many had reached any of plants 1 - 8 . Each of plants 1 - 8 was associated with a colour. When a larva moved to a plant, i t received a dot of airplane model paint (Testors Pla) of the appropriate colour on i t s back. The paint did not seem to affect the larvae, although they would try to eat i t off i f the dot was low enough on the dorsal surface to be within their reach. The colour coding system enabled us to keep track of larvae that made multiple moves, and to separate dispersers from non-dispersers at pupation time. In 1973 we ran two experiments; one in June-July with larvae from Nanaimo Junkyard, and one in August with larvae from Cultus Lake. In 1974 we ran many experiments over the period from June to early September and tested larvae from Topfield, Cultus Lake, Fort Bragg, Comptche, Neal Creek, Powers, Silbernagel, Mackinnon, Sylvester and Cowfield, as well as some descendents of the 1973 Junkyard larvae that had spent the winter as pupae in a cold room. (See Table XI). The larvae tested were procured in many different ways. They are listed i n Table XI, along with the number of plots on which larvae from each study site were tested. We tried to test larvae from as many different study sites as possible, and had to sacrifice numbers of replicates sometimes in order to do this. We en-countered problems in getting larvae back to British Columbia from California, and this contributed to the low number of replicates for Fort Bragg and Comptche. The time of the summer during which larvae were tested i s also listed in Table XL, as the aging of ragwort through summer, 'particularly i n August and September, undoubtedly affected the quality of the foliage, and i t s appeal to larvae. No measurements of plant quality were made. 42 TABLE XI ORIGINS OF LARVAE TESTED AT SOUTH CAMPUS Area Origin of Larvae Number Month Tested of Plots 1973 1974 Junkyard Cultus Lake Nan*Topfield Cultus Lake Fort Bragg Comptche Coldroom (Junk:1973) Neal Creek*1 Powers Silbernagel July August June July July Progeny of £'s caught 8 June 1 Progeny of $'s caught 20 June 29 - July 6 Pupae collected May 2 20 1. Progeny of ?scaught 14 May 31 2. Larvae collected ran- 6 domly in f i e l d Progeny of ¥scaught June 16-20 3 Eggs mailed to Vancouver Progeny of ¥'s caught June 16-18 Eggs mailed to Vancouver 6 July Pupae from 1973 experiment 3 Late July Progeny of s caught June 23 Eggs mailed to Vancouver 8 July'Aug. Progeny of £'s caught July 1 ¥ brought to Vancouver 2 July'Aug. Progeny of £ caught June 23 Eggs mailed to Vancouver 2 JulyAug. Mackinnon Sylvester Cowfield Aggregations of 1st, 2nd 3rd collected in f i e l d ; batches kept separate, sent air freight to Vancouver 7 10 8 Late Aug. Late Aug. Late Aug, 43 I planned to look at the effect of egg batch order, female parent, larval age, larval density and geographic origin on disappearance; and egg batch order, female parent, larval age, plant spacing and geographic origin on dispersal success. Work done in the environment room in the spring of 1974 with Cultus Lake and Linn County Oregon larvae suggested that egg batch order was not important in disappearance. In view of the many other variables to contend with in analyzing the data, I have assumed this to be the case in these experiments. The environment room data are presented in the Appendix. My approach in analyzing the effect of the re-maining variables has been f i r s t to separate out the larval age effect by compiling separate estimates of disappearance and success for 4th and 5th instars. If larvae were put out in the f i e l d as 4th instars, I calculated 5th instar disappearance from the number of 4ths that moulted to the 5th instar on the centre plant. I have controlled for larval densities by using the following classification of the original number of larvae on the centre plant; ^5, 6-15, 16-25, 25-30, > 30. Densities of over 30 were used in 1973 only. Disappearance The mean proportion of 5th instar larvae leaving the centre plant (.74 - .03) is significantly higher (,"p < .05) than that for 4th instars (.54 - .04). These means are for a l l larvae tested in 1974 from a l l areas. I plotted the rate of larval disappearance (expressed as proportion miss-ing) for both 4th and 5th instars against the time elapsed from June 12, 1974 (Day 1 of the south campus experiments) to see how i t varied with the time of the summer and the age of the ragwort plants. There was no sig-nificant correlation of elapsed time from Day 1 and 4th instar disappear-ance, but a trend was obvious for 5th instars. The r value for a corre-44 lation of elapsed time and the angular transformation (arc sin square root) of the proportion missing i s 0.2643, barely significant at p = .05 (Figure 4). There was a significant tendency for larvae tested later in the summer to leave the centre plants more readily than those tested early i n the summer, but this tendency was slight i n comparison to the variation among the measures for larvae from different populations. Larvae from Cultus Lake were tested in August 1973 and in June 1974, and there was no difference in the mean proportion missing between the two tr i a l s (Table XII). The mean proportions missing - S.E. are plotted on Figure 5 above daily maximum temperature values recorded at Vancouver International Airport. The abscissa of the upper part of Figure 5 is not to be matched day for day with that of the lower figure; i t merely describes the period of the summer during which larvae from any one geographic area were tested in 1974; for example, Cultus Lake larvae were tested between days 20 and 35. From this figure i t i s obvious that there are large differences in tendency to leave the centre plant for larvae from different geographic areas. But given the significant correlation of disappearance and elapsed time, i t i s impossible to separate these real differences from those caused by temperature effects and by the age of the ragwort plants. Note that late July and late August were the two warmest parts of the summer. It is unclear whether there are any real differences in dispersal tendency between areas as predicted from their respective plant spacings. No test can be made of the hypothesis using a l l these data together. However, one can compare proportion leaving and plant spacing rank within the groups tested in midsummer and those tested in late summer (Table XII). There i s no support for the hypothesis here. Nova Scotia larvae should a l l have had a low level of dispersal, with Cowfield the lowest. Larvae from Neal Creek-1 o J I I I ' ' I 1 1 1 1 1 1 1 1 1 5 10 15 20 25 30 35 40 A5 . 50 55 60 65 70 75 SO JUNE I JULY 1 AUGUST ELAPSED TIME IN DAYS FROM JUNE 12,1974 F I G . 4 . 5 T H I N 8 T A R D I S A P P E A R A N C E : S O U T H C A M P U S , 1 9 7 4 TABLE XII PROPORTION LEAVING CENTRE PLANT BY GEOGRAPHIC ORIGIN Proportion Leaving n replicates + 10 1973 Junkyard .78 - .32 Cultus Lake .63 - .06 18 Early Summer 1974 Topfield'Nanaimo .63 - .06 15 Cultus Lake .64 + .07 13 Midsummer Neal Creek*1 .72 1 .08 7 Comptche .71 ± .13 2 Silbernagel .96 t .04 2 Coldroom (Junk. '73) .83 - .06 3 Late Summer Sylvester .78 ± .09 3 Mackinnon .93 + .04 3 Cowfield .81 t .03 6 1.0 0.9 o 0.8 < 0.7 cr 0.6 o CL o I 0.5 o < < 0) 0) UJ — i - XT X uJ U ( ) uJ 2- - i _ i O U <r < z 8 § 4> CC CD (D _ J V) cc uJ (— I/) L U > 4) 4) o < Q _ J uJ O O I 80L < E A R L Y S U M M E R MID S U M M E R LATE S U M M E R N A N A I M O C O M P T C H E F O R T B R A G G M A C K I N N O N S Y L V E S T E R C U L T U S L A K E | N E A L C R E E K . C O L D R O O M | C O W F I E L D S I L B E R N A G E L X X X X X X X X 5 10 J U N E 15 20 25 4 0 45 50 30 35 J U L Y I E L A P S E D T IME IN DAYS FROM J U N E 12,1974 55 60 65 70 A U G U S T 75 80 85 FIG.5. S O U T H C A M P U S 1974: D I S A P P E A R A N C E OF 5TH I N S T A R S , MAXIMUM T E M P E R A T U R E A N D E L A P S E D TIME 48 should have dispersed more readily than those from Silbernagel. 2 Using a Chi test of homogeneity, or in some cases the Fisher Exact Probability test, there was no significant difference in performance of larvae from different egg batches of the same female, or different t r i a l s of the same egg batch. Similarly, there was no significant difference in performance of larvae from different females within a geographic area. The similar behaviour of larvae from the same female may indicate herita-b i l i t y of dispersal tendency. But i f this were so, one would expect varia-tion between offspring of different females to be the source of v a r i a b i l -ity upon which selection could act. The larval densities on the centre plant varied in 1973 from 10 - 75 and in 1974 from 5 - 3 0 . It i s impossible to detect differences i n the probability that larvae w i l l leave a plant over the range from 5 to 30 2 larvae using Chi tests of homogeneity for larvae from one geographic area. If a l l the data from 1973 and 1974 are grouped together for a l l areas, no clear positive effect of larval density on disappearance is vi s i b l e (Table XIII-. ,- Figure 6). Most dispersal occurred in t r i a l s with densities > 30, but no trend of increasing disappearance with increasing density up to 30 is obvious. The distribution of early and late 1974 replicates in these different larval density categories i s undoubtedly influencing the observed variation. This is suggested by the fact that larvae from the same geo-graphic area, tested during the same part of the summer, show no effect of density. Larval numbers per plant may have to be related more directly to plant biomass in order to understand their importance in the trade-off be-tween staying and leaving a food plant. However, failure to observe any density effect suggests that other factors may be more important than larval numbers in determining dispersal behaviour. 49 TABLE XIII 5th INSTAR DISAPPEARANCE AND CENTRE PLANT LARVAL DENSITY Density Proportion Leaving Replicates ^ 5 .71 - .09 9 6 - 1 5 .64 - .05 30 16 - 25 .78 - .03 27 25 - 30 .65 - .07 11 > 50 .89 - .05 4 50 1.0,. 0.9 0.8 < LU o 0.7 cc o Q_ O cc • Q- 0.6 3> 3> 0) 0.5 _L <5 6 - 1 5 1 6 - 2 5 26-30 C E N T R E PLANT LARVAL D E N S I T Y >50 F I G . 6 . 5 T H I N S T A R D I S A P P E A R A N C E A N D L A R V A L D E N S I T Y O N C E N T R E P L A N T 51 Success Success was defined as the arrival for the f i r s t time of a larva on one of plants 1 - 8 ; and was calculated as a proportion of those larvae missing from the centre plant. I looked f i r s t at the relative proportions of successful 4th and 5th instar larvae on plots with plant spacings of 1 meter, 2 meters, 3 meters and 5 meters. These data are summarized in TableXIV and Figure 7. F i f t h instar larvae are significantly more successful than 4th instars at plant spacings of 1 m and 3 m but not at 2 m and 5 m . The small sample size at 5 m should be noted. F i f t h instar larvae are more successful at 1 m than at 2 m , but disperse with equal success at plant spacings ranging from 2 m to 5 m . Fourth instars behave a bit differently. They disperse with equal success at 1 m and 2 m spacing, but do much worse at 3 m and 5 m . I then separated the 5th instar data into the geographic areas (Table XV). The proportion successful is significantly higher at 1 m than at 2 m for Cultus Lake and Sylvester larvae, but not for Topfield, Junkyard or Cowfield larvae. This implies that the pattern of search of larvae at these latter three areas may be more suited to locating plants that are widely spaced. In order to test for significant differences in success between d i f -ferent areas, I combined mean estimates of success made at different plant spacings where the differences were not significant (Table XVI). I then compared proportion of successful dispersal for the different geographic areas tested at 1 m and 3 m spacing respectively. If larval search pat-terns are indeed suited to local plant spacing conditions, then larvae with a low probability of having to travel further than 1 m should do 52 TABLE XIV DISPERSAL SUCCESS AND PLANT SPACING - SOUTH CAMPUS 5th INSTAR LARVAE 1 m 2 m 3 m 5 m n .31 x + 54 - .18 11 x , 28 .37 - .04 n x , 25 .39 - .06 n x , 3 .38 - .16 t-tests 1 m and 2m t = 2.46 2 m and 3m t = 336 3 m and 5m t = .0562 4th INSTAR LARVAE 1 m 2 m 3 m 5 m n X n X n x n x 15 .43 t .26 14 .37 - .06 16 .18 - .06 3 .17 - .17 t-tests 1 m and 2 m t = - .7522 2 m and 3 m t = -2.1640 3 m and 5 m t = .1001 4th and 5th comparison - t-tests 1 m t = -1.1674 2m t = .0417 3m t = -2.4567 5 m t = ,9424 53 1.0 0.9 0.8 ui 0.7 to LU o £ 0.5 Q 0.4 o g 0-3 CL O CC °- 0.2 0.1 0 5 T H 5 4 T H S 4> _L=L 2 3 4 S O U T H C A M P U S P L A N T S P A C I N G IN M E T E R S F I G . 7 . D I S P E R S A L S U C C E S S A N D P L A N T S P A C I N G : L A R V A E F R O M A L L A R E A S 54 TABLE XV DISPERSAL SUCCESS AND PLANT SPACING BY GEOGRAPHIC ORIGIN OF LARVAE l_m 2 m 3 m + + * + Cultus Lake .65 - .07 .38 - .06 .29 - .07 Cowfield .27 - .10 .40 - .18 Sylvester .75 - .17 .22 - .16 Junkyard .63 - .10 .50 -"'.10 .57 - .25 Topf ield .34 - .12 .22 - .16 .42 - ,'13 Neal Creek .39 - .11 Silbernagel .60 - .25 * values joined by are not significantly different 55 TABLE XVI DISPERSAL SUCCESS AND GEOGRAPHIC ORIGIN OF LARVAE 1 m 3 m Cultus-1 .65 - .07 .34 » .05 Cowfield .31 - .08 Topfield .34 - .26 .34 - .26 Junkyard .57 - .08 .57 - .08 Sylvester .75 - .17 Silbernagel .60 - .25 t-test not sig. different Cowfield - Topfield -0.2252 Cultus - Sylvester -0.4823 Cultus - Junkyard .6886 t-test not sig. different Cultus - Topfield -0.0513 Junk.- Silbernagel -0.1253 Cultus Junkyard Sylvester Topfield Cowfield Junkyard \ Cultus•1 Silbernagel ' Topfield 56 better on 1 m plots than those with a correspondingly high probability. On 1 m plots, Cultus-1 (p > 1 m = .05), Junkyard (p > 1 m not available) and Sylvester (p > 1 m = .40) larvae are more successful than larvae from Topfield (p > 1 m not available) and Cowfield (p > 1 m = .59). Similarly, larvae with a high probability of having to travel further than 2 m should do better on 3 m plots than those with a correspondingly low probability. Junkyard (p > 2 m not available) and Silbernagel (p y 2 m = .12) larvae are more successful than Cultus-1 (p )> 2 m = .02) and Topfield (p y 2 m not available) larvae. There is thus some evidence to suggest that larval search a b i l i t y is suited to local plant spacing conditions. This should be investigated further. I examined the variation in dispersal success of 5th instars within 2 the progeny of one female using Chi tests of homogeneity. For a l l but 2 out of 14 females, i t was possible to accept the hypothesis of homogeneity of dispersal success for their progeny, even i f they were tested at d i f -ferent plant spacings. This i s not too surprising given that i n s i g n i f i -cant differences in success at south campus plant spacings were observed for larvae from Cowfield, Junkyard and Topfield. I also tested the variation in 5th instar success for the progeny of different females, controlling for plant spacing of the t r i a l s . In only 11 out of 18 cases was I able to accept homogeneity of success among pro-geny of different females. Failure to accept a test of homogeneity may indicate real differences in dispersal success or search behaviour among progeny of different females, or i t may indicate variation between r e p l i -cates in mortality not associated with dispersal behaviour at a l l . It may be worth pursuing the possibility that there are real differences in search success for progeny of different females. Such differences could be acted upon by selection to adjust dispersal search behaviour to local plant 57 conditions as suggested above. Mean moves/larva/day The mean number of successful moves/larva/day i s a combined estimate of both the extent and rate of disappearance and success. I used Green's (1974) method for calculating'this s t a t i s t i c . It standardizes larval ac-t i v i t y on a per day basis and includes an indication of multiple moves. A move i s defined as the arrival on a new plant of a larva which had pre-viously been on the centre plant, or on one of plants 1 - 8 . This fact makes the measure valuable in the investigation of dispersal success and plant spacing (Table XVII). The values for 5th instars suggest that larvae of this age are more successful at a plant spacing of 1 m than at wider spacings, but that over the range from 2 m to 5 m a substantial number of larvae s t i l l manage to move around. This agrees with the information on dispersal success and plant spacing from the previous section, and further separates the rela-tive a b i l i t i e s of 5th over 4th instars (Figure 8). Multiple Moves and Plant Spacing The mean proportion of successful dispersers that move again success-f u l l y i s .11 - .02. Of these, most of the multiple moves occur at plant spacings of 1 m and 2 m and very few at spacings higher than 3 m . The distribution of multiple moves on plots of different plant spacings i s shown in Figure 9. Plant spacing would seem to be more important than the geographic origin of the larvae in determining the incidence of multiple moves. The mean proportions of successful dispersers that move twice, three, four and five times for larvae from a l l areas are .098 - .02 , .0055 - .0035 , .0006 and .0006 respectively. 58 TABLE XVII SUCCESSFUL MOVES/LARVA/DAY AND PLANT SPACING LARVAE FROM ALL AREAS Plant Spacing Moves/larva/day - S.E. plots 5th Instar 1 m 2 m 3 m 5 m 31 30 26 3 1153 - .0174 ,0728 - .0115 ,0793 - .0174 ,0612 t .0298 4th Instar 1 m 2 m 3 m 5 m 18 15 16 3 ,0839 - .0173 ,0375 - .0089 .0143 - .0043 .0028 - .0028 59 .01 i l i i ^ 1 2 3 4 5 SOUTH CAMPUS PLANT SPACING IN METERS FIG.8 .MEAN M O V E S P E R L A R V A P E R DAY A N D PLANT S P A C I N G : S O U T H C A M P U S 6 0 3 0 _ o n uJ 2 5 o y 2 0 CL Z) o cr LU CQ 1 5 10 1 2 3 A 5 SOUTH C A M P U S P L A N T S P A C I N G IN M E T E R S F I G . 9 . M U L T I P L E M O V E S A N D P L A N T S P A C I N G : S O U T H C A M P U S 6 1 Distance Effect and Directionality within Plots Figure 2 illustrates the arrangement of plants within the south campus plots. A l l 8 are not equidistant from the centre; plants 1-2-3-8 are closer to the centre plant than plants 1-3-5-7. Plants 1-2-3-8 are to the East of the centre plant and plants 4-5-6-7 are to the West. We noticed in 1973 a tendency of larvae to arrive on plant 2, while in 1974 i t seemed that plants 2 and 8 received more than a random share of suc-cessful dispersers. Table XVIII presents the number of larvae f i r s t arriv-2 ing on plants 1-8 ( a l l areas are combined) and the results of a Chi test which hypothesizes that there are equal numbers of larvae on each plant. For 5th instars, more larvae arrive on plants 2-4-6-8 (x = 151 .35135) than on plants 1-3-5-7 (x = 67.8 - 9). The same is true of 4th instars where plants 2-4-6-8 receive x = 28.5 - 4, and plants 1-3-5-7 receive x = 12.25 1 1. This would be expected i f larvae were moving i n a random walk rather than a straight path in a random direction, when searching for plants. They would be more li k e l y to arrive at those plants closer to the centre plant, than at plants further away. Straight path travel would populate the outer plants with equal probability. Within both the inner c i r c l e of plants (2-4-6-8) and the outer c i r c l e (1-3-5-7) for 5th instars, there is a marked directional effect. Plant 2 receives more larvae than plants 4-6-8, and plants 1-3-7 receive more larvae than plant 5. Thus the order i s 2 ) 4-6-8 > 1-3-7 ) 5, and for both rings of plants, the eastern ones receive significantly more than the western ones. Plant 7 is an exception, being west of centre. Wellington (1955) has shown that insects orient with reference to the sun as a heat source. That i s , the basic photic orientation is a function of their i n -ternal temperature. If the cinnabar larvae move towards the sun in the morning when they are cool, and away from i t in the afternoon when they 62 TABLE XVIII DIRECTIONALITY OF SUCCESSFUL MOVES Plants 1 2 3 4 5 6 7 8_ 5ths 85 251 70 116 42 91 74 149 4ths 15 23 12 28 10 22 12 41 2 Chi Calculations: 5ths 4ths Plants 1-8 272 d.f.=7 reject 37.78 d.f.=7 rej ect Plants 2-4-6'8 97.7 d.f.=3 reject 8.04 d.f.=3 reject Plants 2-8 4-6 26.01 3.02 d.f.=1 d.f.=1 rej ect accept 5.06 .72 d.f.=1 d.f.=1 reject accept Plants 1-3-5-7 14.83 d.f.=3 reject 1.04 d.f.=3 accept Plants 1-3 5'7 1.45 8.83 d.f.=1 d.f.=1 accept reject 63 are warmer, they w i l l always move generally eastward. This could account for the observed directionality in dispersal success, since i t would mean that more larvae were looking for plants in the eastern part of the plot than in the western. 4th instar larvae show the same distance effect as 5th instars. The directionality of their movement i s different in that plant 8 receives the most larvae of the inner ring, and there i s no directionality dis-played in the outer ring. Therefore the decreasing order of larvae re-ceived is 8 > 2-4-6 > 1-3-5-7. DISPERSAL IN NATURAL TYRIA POPULATIONS While the south campus experiments provide useful comparative data about larvae from different populations, the data were collected at plant densities very different from those observed in the study sites. The hypothesis relating larval disappearance and plant spacing is really best tested by observing dispersal behaviour of larvae in their local setting. I used this approach at Cultus Lake in 1973 and 1974, and in Linn County Oregon, and Nova Scotia in 1974. The basic method was to place a known number of larvae (identified with paint) on a ragwort stem, number the surrounding stems, measure their distance and direction from the centre plant, and then record larval movement to these numbered plants. At Cultus Lake, the local patch of numbered plants was enclosed with a 6 inch plywood fence l i k e that at South Campus, and the ground cover between the ragwort plants (grasses and ferns) was cut back. See Figure 10 for photographs of dispersal boxes at Cultus Lake. (In 1973, the plywood fences also had fluon-sprayed plastic). In Oregon and Nova Scotia, no fences were used to enclose the plants. I had planned to get more information from 13 experiments set up at Linn 64 1M x 1M P L O T 1973 5Mx 5M PLOT 1974 F I G . 1 0 . P H O T O G R A P H S O F D I S P E R S A L B O X E S C U L T U S L A K E 65 County, Oregon, but there was such severe ant predation on the cinnabar larvae at the site I chose, that results of only two of the experiments can be used. Ants were observed at other Linn County sites but they did not seem to affect the larvae as much as at the site of the experiments. Disappearance , The length of time I was able to follow larval behaviour varied greatly with the number of days I could spend at each study site. The shortest period of observation was 2 days. I compared disappearance after 2 days for the different study sites classified as before into widely spaced and more closely spaced plant populations (Table XIX) . The mean proportion missing was not significantly higher for populations whose plants were found close together, although the mean number missing certainly varied in the direction predicted by the hypothesis. Success The data for dispersal success in the f i e l d as calculated at Cultus Lake are very reliable as the larvae were followed through the 5th instar to pupation, a period of 12 days (Table XX). The average distance of a successful move measured from the centre plant was significantly lower for 4th instars (.38 - .05 m) than for 5th instars (1.04 - .18 m). Fourth instars were often found on rosettes. Larvae of the 4th instar dispersed + + with almost as high a rate of success (.82 - .04) as 5th instars (1.00-0) under a f i e l d situation which had been modified by removing grasses and ferns from between the ragwort stems. The fact that a l l the 5th instars could be accounted for as successful dispersers was striking and repeat-able. A very high proportion of the 5th instars dispersed more than once, as evidenced by a mean moves/larva/day of 0.3353 1 .0006, much higher than that from south campus 1 m spacing of 0.1556 1 .03. 66 TABLE XIX PROPORTION MISSING AFTER 2 DAYS p > lm Prop.Missing after 2 Days Replicates Neal Creek-0 .75 .23 2 Cowfield .59 .23 6 Sylvester .40 .03 6 Mackinnon .39 0 3 + x = .12 - .06 Silbernagel .27 .24 2 Cultus-1 .05 .65 9 x = .45 - .20 t = 2.0599 67 TABLE XX- DISPERSAL SUCCESS AT CULTUS LAKE FIELD 2 4ths 5ths x Dist. Successful Prop. Mean Dist. Successful Prop. Plot Move Success Move -Success D .49 - .06 .76 E .33 - .05 .80 G .34 - .05 .91 -H .46 - .18 1.00 I 1.03 - .35 1.00 F a. b. 1.50 - .24 1.30 ± .23 1.00 1.00 C .89 - .22 1.00 x = . 38 - .05 x = 1.04 - .18 Distance moved t = -2.6927 4ths and 5ths p .05 * Plot F had 2 centre plants 68 The data from the other f i e l d locations are less reliable since they were not collected in a consistent manner or are based on small sample sizes (Table XXI). At Cowfield, however, the dispersing larvae were f o l -lowed closely and were easily seen because the grass cover was so closely grazed. The proportion of successful larvae i s not available for Sylvester, Mackinnon or Silbernagel for the above mentioned reasons. From the average distance of a successful move, i t seems that 5th instar larvae disperse mainly to those plants nearest the plant they l e f t . Where the probability of finding another plant within 1 m was high, the mean distance moved was less than 1 meter. Mean distances moved were not significantly different in widely spaced and closely spaced plant populations. There was no significant directional tendency of dispersing 4th or 5th instar larvae either at Cultus Lake or the other f i e l d locations. SUMMARY - DISPERSAL SUCCESS AND PLANT SPACING The values for dispersal success from the f i e l d t r i a l s at Cultus Lake and Cowfield can be combined with those values for dispersal success for a l l 5th instar larvae tested (1974 and 1974) at south campus. If these proportions are plotted against the probability that plant spacing is greater than 1 meter, we should get an indication of what proportion of successful dispersal to expect under different f i e l d conditions (Figure 11). The south campus plots are entered at their respective posi-tions on the abscissa; small plots having 4 of the 8 plants at a distance of 1.41 m rather than 1 m from the centre (See Figure 2). The 2 m, 3 m and 5 m plots have, of course, a probability of 1 that the plants are separated by more than 1 meter. The 5 points at p. ) 1 n = .33 on the abscissa are from f i e l d t r i a l s at Cultus Lake, and the 6 points at 69 TABLE XXI 5th INSTAR DISPERSAL SUCCESS IN FIELD TRIALS AT STUDY SITES n S u c c e s s f u l Mean d i s t . p *L 1 m Prop. Larvae success-move Success Cowfield 12 1.00 - .17 .41 .46 - ,07 S y l v e s t e r 6 .54 - .08 .60 Mackinnon 2 .44 - .06 .61 Cultus-2 75 1.04 - .18 .67* 1.00 - 0 S i l b e r n a g e l 8 .80 - .20 .73 h a b i t a t modified before t r i a l s run; so p ( 1 ra i n d i s p e r s a l boxes i s probably lower. 70 r = -0.4423p<.01 Y = -80229x+.3572 LOG 1 Q P>1M.PLANT SPACING F f G . 1 1 . P R O P O R T I O N S U C C E S S A N D L O G 1 Q P > 1 M P L A N T S P A C I N G p ) 1 in = .59 are from Cowfield. A l l the others are from south campus. These points can be described by a curvilinear function, which is straightened to a linear function by the transformation of p > 1 m to log^Q p y 1 m (r = -0.4423, p <C .01 , Figure 11). The curvilinear function is plotted in Figure 12. The various subpopulations studied are entered along the abscissa at the appropriate points to enable predic-tion of the proportion of successful dispersal which might be expected under their plant spacing conditions. It is useful at this time to en-quire whether the south campus conditions approximated f i e l d conditions well enough to enable such predictions to be made from Figure 12. As mentioned above, the Cultus plots were cleared of grass cover, whereas those at UBC often had very dense clover between the ragwort, which pro-bably made dispersal more d i f f i c u l t . Thus one might expect dispersal success in the f i e l d to be even higher than that of south campus. On the other hand, Green (pers. comm.) has suggested that the "dense clover cover might have provided shade that reduced dessication, trapped micro-level moisture, and generally aided the survival of 5th instars, which could then survive longer in their quest for food." If this were true, then one might expect the proportion of successful dispersers to have been higher at south campus than in the f i e l d . Whatever the case, Figure 12 suggests that there i s differential survival of dispersing larvae under different conditions of plant spacing. Success would vary, according to the curvilinear regression, from .74 at Cultus*2 to .44 at Neal Creek*0 . I am reluctant to extend the function past .33 on the abscissa, although this is the region in which most of the subpopulations f a l l . Figure 12 shows that the differential mortality necessary for selection exists. It i s d i f f i c u l t to compare the f i e l d measures of dispersal success 33 O CO O 5 ^ PROPORTION S U C C E S S o "0 m o co o o rn co co > o ~ D T J > ~0 > o T J O o co o 1> o "cn o o o Lo p Ko p 'oo 1 1 S T I C K U M C U L T U S : 1 N E A L CREEK:1 C A T C H I N G C R E E K E C K L E Y CR.HILL S I L B E R N A G E L P O W E R S CULTUS:2 M A C K I N N O N S Y L V E S T E R S. C A M P U S E C K L E Y C R COWFIELD N E A L CREEK- .O L© S. C A M P U S o CO o "co with those reported by Green (1974) since he classified plant populations by density and not by any measure of the spatial distribution of plants. However his figure relating probability of 5th instar larvae locating a 2 plant on a grazed pasture suggests that at a density of 1 stem/m success is only about .35, and that i t drops to .25 and then to .02 as density drops below this level. If these values are placed on Figure 12 at positions on the abscissa for south campus plots of similar density to those Green reports, his values seem very low. The bottom line repre-sents a linear regression of the points he reports. It has an r of -0.7172 which is not significant due to the small sample size. Thus the slope is not significantly different from zero. It is however much lower than the top line, which i s the curvilinear function described in Figure 11. Green's estimates were made under considerably different conditions from mine. He used 5th instar larvae that had been starved for 30-40 minutes, released them from a petri dish and recorded their success at finding plants for at most one day. My estimates are from larvae that l e f t a centre plant on their own i n i t i a t i v e , their success being re-corded over the f u l l length of the instar. Green's larvae may not have been interested in finding ragwort plants i f they were near the end of the 5th instar, or they may have taken longer than one day to locate a plant. On the south campus plots, larvae making multiple moves were often missing for several days before showing up on another plant. 74 GENERAL DISCUSSION OF HYPOTHESES RELATING DISPERSAL TENDENCY TO PLANT SPACING I propose to briefly review the hypotheses concerning environmental va r i a b i l i t y and dispersal before discussing the implications of this thesis work. As Roff (1974b) explains, "the way in which we view popula-tion regulation may depend on the 'level of resolution' at which we look at the problem..." We may concern ourselves only with the average sub-population in an attempt to understand the greater dynamics of a whole population; or we may' analyze the spatial and temporal v a r i a b i l i t y of specific individual subpopulations. My work has been at the latter level of resolution. Understanding of the spatial and temporal arrangement of the resources' in specific subpopulations, and their likelihood to dis-appear over time may be a necessary prerequisite to a consideration of dispersal and regulation at a higher level. Green's (1974) hypothesis of Tyria population regulation states that, under some conditions, populations of the cinnabar moth can regulate their own numbers before they overuse the plant resource, through the mortality suffered by dispersing larvae. The necessary condition for such regula-tion i s low ragwort density. This hypothesis really pertains to the indi -vidual Tyria subpopulation, and contributes to an understanding of Tyria regulation at the population level by suggesting that subpopulations are less l i k e l y to disappear since they have this mechanism for persistence. This hypothesis is part of a larger discussion of dispersal selection and plant spacing, i n which Green recognizes the d i f f i c u l t y of predicting the selection regime in any given environment. He mentions the importance of larval density as well as plant spacing in calculating the relative ad-vantage of leaving a food plant; and the importance of the historical plant and insect densities to the selection regime of the present. 75 Myer's ideas (pers. comm.) about the role of dispersal in the regula-tion of insect populations concern the fate of whole populations - what she calls the geographic or temporal level - as well as the fate of individual subpopulations. She predicts that, at both levels, dispersal should i n -crease population s t a b i l i t y , and should be favoured where i t is more li k e l y to be successful. My work has taken place within this general hypothesis. Selection should favour dispersal in more closely spaced plant populations. From this hypothesis, I have investigated several measures of disper-sal behaviour and looked for significant differences between larval popula-tions from closely and widely spaced plant populations. The measures of larval groups having 2 or less larvae, the indices of dispersion of 4th and 5th instar larvae on stem plants, and the proportion of 4th and 5ths that should move from the plant they are on. In two out of these three measures, there was significant indication that larvae from widely spaced plant popu-lations disperse less. The distribution of larvae in areas of wide plant spacing was overclumped, and significantly more larvae in these areas should have moved to the neighbour plant. I also used values for dispersal ten-dency of larvae, from different areas, obtained from the south campus ex-periments and from the f i e l d t r i a l s . There was no indication that larvae from populations that were, in 1974, widely spaced dispersed more. A l -though i t should be recognized that the main body of data, from the south campus, was confounded by the aging of the plants and perhaps by the weather, i t is particularly d i f f i c u l t to explain away the observation that larvae from Nova Scotia dispersed so readily. By the hypothesis, those from Cowfield and Sylvester particularly should have-been very reluctant to leave"" the centre plant. I have attempted to quantify the relationship between plant spacing 76 and dispersal success, to make the predictions of the general hypothesis more specific. The fact that there is lower survival of dispersing larvae under conditions of wide plant spacing at south campus is significant. This differential survival is a necessary condition of any selection regime acting on dispersal. According to genetic theory, the differential need not be large in order to significantly affect the quality of the population. It must be remembered here though that for larvae from 3 of the areas studied (Topfield, Junkyard and Cowfield) dispersal success at south campus was not significantly different at lm 2m and 3m spacing, and thus the differential mortality does not seem to exist. From Figure 12, larvae in closely spaced plant populations should have a proportion survival of .74 or higher. Those dispersing in the mid range of plant spacing, a proportion success of between .74 and .60, and the five widely spaced populations, between .60 and .44. Such differential survival could be sufficient to produce different behavioural types in widely spaced and closely spaced plant populations. Two f i e l d measures of larval dispersal support the hypothesis that dispersal tendency is influenced by the probability of dispersal success. On the other hand, the south campus results do not add any weight to the hypothesis. This may be due to the influence of factors such as plant quality or weather. Or i t may be, as Green (pers. comm.) has suggested, that the plant spacing conditions in the areas studied had not been sufficiently extreme for a long enough period of time to produce different larval behavioural types. If the higher dispersal mortality associated with wider plant spacing selects against dispersal behaviour, then dispersal mortality may be constant in populations with different plant spacing. It w i l l 77 not act differentially as a population regulating mechanism in populations with different plant spacing as Green (1974) has suggested. If dispersal were acting to regulate population numbers, one would expect dispersal tendency to increase with the number of larvae on the original plant. This was not observed for larval densities of less than 30, which are the most usual in the f i e l d . This suggests that dispersal tendency is influenced less by larval numbers on the original food plant than by the probability of dispersal success. But the relationship between larvai numbers and food plant biomass must play some role in this process, since the object is to attain pupation weight and pupate successfully. One could speculate that in those populations where plants were far apart, selection would favour smaller egg batches . This w i l l be duscussed in the next section. My examination of these 13 Tyria subpopulations suggests that larval dispersal would be favoured where the probability of surviving dispersal was high. But that in more widely spaced plant populations, dispersal could be selected against. If larval dispersal contributes to the st a b i l i t y of local subpopulations by enabling the cinnabar moth to make best use of the tansy ragwort resource, then an interesting question is raised. That of the population dynamics of those areas with wide plant spacing where disperser types are not favoured. This is an area that should be investigated before we can speak with any certainty of the role larval dispersal plays in Tyria population regulation on a larger scale. 78 DISPERSAL, FECUNDITY AND POPULATION REGULATION OF TYRIA JACOBAEAE Regulation in insect populations is more often defined as their a b i l i t y to prevent complete exhaustion of their food resource than simply as their a b i l i t y to persist over time. When Dempster (1961) says that Tyria populations in England are not regulated, he means that their numbers fluctuate wildly and many larvae die of starvation in some years. Any density-dependent reduction i n fecundity i s insufficient to prevent these great fluctuations. Given that lighter larvae produce smaller pupae, smaller pupae i produce smaller moths and smaller moths lay fewer eggs (Dempster 1961, Van der Meijden, 1970) i t is reasonable to ask whether dispersing larvae are lighter, produce smaller pupae and thus contribute less to the next generation than non-dispersers. If so, dispersal could act to regulate Tyria populations through a feedback to fecundity. It i s unlikely that this would ever prevent population crashes, but i t could be a factor. The data for consideration of this hypothesis come from south campus 1973. That year, both dispersing and non-dispersing larvae were weighed while they were s t i l l on the plants. The subsequent dates of larval pupation were highly synchronized in both July and August and i t was pos-sible to gather pupating larvae from the inside of the plot walls, and separate them by their paint dots into dispersers and non-dispersers. In 1974 no such mass movement of pupating larvae to the edge of the plots was seen, and we have no datum on comparative pupal weights. The extreme va r i a b i l i t y in the weather.during the early part of the summer of 1974 certainly helped destroy any synchrony of development time among the larvae. The plot walls were also not as larvae-proof in 1974 as in 1973 and this may be the real reason for the difference. Results from the environment room suggest that pupal weights of 79 progeny of different females are significantly different (See Appendix). Therefore I have examined variation in the larval weights of dispersing and non-dispersing larvae only within progeny of the same female. This w i l l avoid the confusing influence of between-female variation, and also possible age differences. In no case tested were there significant larval weight differences between dispersers and non-dispersers as mea-sured by a t-test, p = .05. It i s not surprising then that no difference was observed between pupal weights of dispersing and non-dispersing larvae; (progeny of d i f -ferent females were not separated). These data are presented in Table XXII. Either pupal weight or pupal width can be used as an index of ex-pected fecundity, as the two are correlated with an r value of .76. Pupal widths are not significantly different between dispersers and non-dispersers either, as measured by a t-test. It does not seem then that successful dispersers contribute any more or less to the next generation than non-dispersers. Obviously though, those dispersers that starve while looking for a plant do not contribute anything. Dispersal does not act through fecundity as a local population regulator. Data collected from a l l the study sites in 1973 and 1974 enable us to consider whether Tyria populations are regulated through fecundity at a l l . If they were, then one would expect differences in mean egg batch size be-tween different local areas, as selection adjusts the mean number of eggs per batch to the available plant biomass/egg. The mean estimates of batch size of unhatched egg batches for which there are also estimates of stem biomass are presented in Table XXIII. Mean egg batch size varies be-tween different study sites; but there is no correlation between the mean number of eggs per batch and the average biomass of stems (r = -.01). This lack of any relationship between measured fecundity and available biomass 80 TABLE XXII PUPAL WEIGHTS OF DISPERSING AND NON-DISPERSING LARVAE Dispersers Non-dispersers Junkyard .1265 .1235 t= non sig. Cultus Lake .1338 .1323 t= .234 d.f.= 44 PUPAL WIDTHS OF DISPERSING AND NON-DISPERSING LARVAE Dispersers Non-dispersers Junkyard Cultus Lake .4736 .4853 .4781 .4821 t= non sig. t= .362 d.f.= 44 81 TABLE XXIII EGG BATCH SIZE, STEM BIOMASS AND PLANT SPACING x Egg Batch x Stem Biomass Neal Creek-0 25.41 117.54 Cowfield 37.25 49.55 Eckley Creek 37.34 53.25 Sylvester 65.56 Mackinnon 57.30 63.88 x= 39.33 - 9.90 x= 69.96 - 12.28 Silbernagel 41.00 Catching Creek 27.43 36.00 Cultus-2 Powers 34.93 109.00 Eckley Creek H i l l 26.50 137.00 Neal Creek*1 24.10 18.00 Cultus-1 29.45 10.15 Stickum 28.71 x= 28.48 - 1.83 x= 54.27 - 21.81 t= -2.2144 t= 0.6446 not significant not significant 82 suggests that local Tyria populations do not anticipate shortage of food resource with any decline in fecundity. In other words, they are not self-regulatory and fast fluctuations of larval numbers are the result. Egg batches were not significantly smaller in widely spaced populations, indicating no avoidance of dispersal mortality through a feedback to fecundity. The Cultus Lake area, which was followed for two years, provides a good case study of this lack of regulation (See Table XXIV;;, Figure 13). In 1973, stem plants were la r g e r d e n s e r and had more biomass than in 1974. In 1973, there was serious defoliation in only one part of the f i e l d , while in 1974 defoliation took place earlier and was much more ex-tensive, spreading to parts of the f i e l d where few plants had previously been attacked by larvae. The fact that pupae collected in late 1974 were lighter and a higher proportion of them were dead indicated that 1974 was a year when this Tyria subpopulation really overran i t s food resource; and where a delayed density-dependent feedback to fecundity w i l l probably be observed next year. This subpopulation was not prepared for the fact that i t s own success at defoliating 1973 plants would mean smaller plants in 1974 and thus less available resource. Cultus Lake f u l f i l l s the conditions necessary for an outbreak as described by Green - high plant density and low dispersal mortality - so his hypothesis would predict that this subpopulation could not have avoided the larval starvation of 1974. The larvae that were best off as far as food supply at Cultus Lake in 1974 were probably descendents of early-emerging moths that were search-ing for ragwort foliage in mid June when there was s t i l l some available. This i s the situation Dempster (1971) described when he said: 83 TABLE XXIV CHANGES AT CULTUS LAKE BETWEEN 1973 AND 1974 Cultus-1 1974 Plants/m Stem Height Pupal Weight 88.3 51.63-3.76 .11561.001 Prop. Pupae Dead Cultus-1 1973 105.81 59.59-3.5 .13981.005 ,45-.16 (Aug.18 Sept.9) Cultus-2 1974 37.3 Stickum 1974 92.1 61.671.33 66.25l .02 no larvae few larvae 84 FIELD 1,1973 F I G . 1 3 . P H O T O G R A P H S O F C U L T U S L A K E F I E L D 1, A U G U S T 1974 P A S T U R E E A S T OF S T I C K U M F IELD,AUG. 1974 8 5 "The only factor which appears to buffer the population against extinction in years when food runs out early in the season, i s the heterogeneity within the moth and the ragwort populations. The earliest individuals manage to obtain sufficient food in those patches of ragwort which survive longest." Local cinnabar populations do not seem to be regulated in the sense of avoiding years of mass larval starvation, but they do have an a b i l i t y to persist. It i s this a b i l i t y which might now be attributed to an average cinnabar subpopulation in any discussion of Tyria regulation and per-sistence at the geographic level of resolution. 86 SUMMARY The purpose of this thesis was to study variation in the tendency of cinnabar larvae to leave their food plant, tansy ragwort, before i t is defoliated and before they are ready to pupate. Dispersal tendency was assumed to be a product of natural selection, and thus influenced by the probability of dispersal success. The research hypothesis was that larval dispersal should be favoured in areas with closer plant spacing and selected against where plants are more widely spaced. Plant spacing and dispersal tendency were measured in 13 subpopulations of cinnabar moth. The influence of plant spacing on dispersal success was measured in experimental plots with 1,2,3 and 5 metres between plants. The results were as follows: 1. Eight out of thirteen subpopulations of tansy ragwort were classified as closely spaced on the basis of measures of plant density and distribution. 2. Based on the plant spacing measured in the thirteen subpopulations, the expected probability of dispersal success was estimated to be from .44 in widely spaced populations to .74 in more closely spaced populations. 3. Such differential survival could be sufficient to produce different behavioural types in widely spaced and closely spaced plant populations. 4. The distribution of fourth and f i f t h instar larvae on stems in widely spaced populations was more clumped than the negative binomial, and that in a closely spaced population, less clumped. This suggests reduced larval movement in populations with wide plant spacing. 5^ The proportion of fourth and f i f t h instar larvae that would have had more food available i f they had moved to the nearest plant was significantly higher in widely spaced plant populations. Therefore, 87 in these populations, larvae weren't adjusting their density on the plants to the available food. 6. Dispersal tendency as measured by the proportion of larval groups with two or less larvae was not significantly higher in populations with close plant spacing. 7. Thus, in 2 out of 3 measures of larval distribution in the f i e l d , there was significant indication that larvae from widely spaced populations disperse less, and that the hypothesis relating dispersal tendency to plant spacing is supported. 8. Since higher dispersal mortality associated with wider plant spacing seems to select against dispersal behaviour, I suggest that the total amount of dispersal mortality may be constant in populations with different plant spacing. Dispersal mortality w i l l not act d i f f e r e n t i a l l y as a population regulating mechanism in populations with different plant spacing. .8,8 LITERATURE Boer, P.J. den. 1968 Spreading of risk and stabilization of animal numbers. Acta biotheoret. (Leiden) 18:165-194. Bornemissza, G.F. 1966 An Attempt to Control Ragwort in Australia with the Cinnabar Moth, Callimorpha jacobaeae (L.) (Arctiidae: Lepidoptera) Australian J. Zool. 14:201-243. Bucher, G.E. and P. Harris. 1961 Food-Plant Spectrum and Elimination of Disease of Cinnabar moth Larvae, Hypocrita jacobaeae (L.) (Lepidoptera: Arctiidae) Can. Ent. XCIII:931-936. Cameron, Ewen. 1935 A Study of the Natural Control of Ragwort (Senecio jacobaeae L.) J. Ecol. 23:265-322. Dempster, J.P. 1961 The analysis of data obtained by regular sampling of an insect population. J. Anim. Ecol. 30:492.432. Dempster, J.P. 1970 A Population Study of the cinnabar moth, Tyria (Callimorpha) jacobaeae L. Proc. Adv. Study Inst. Dynamics Numbers Popul. (Oosterbeek, 1970) 380-389. Dempster, J.P. 1971 The Population Ecology of the Cinnabar Moth. Oecologia 7:26-67. Dethier, V.G. 1959 Food Plant Distribution and Density and Larval Dispersal as Factors Affecting Insect Populations. Can. Ent. 91:581-596. E l l i o t t , J.M. 1971 Some Methods for the St a t i s t i c a l Analysis of Samples of Benthic Invertebrates. Freshwater Biological Assoc. Scientific Pub. No. 25, Westmorland, Eng. Frick, K.E. and J.K. Holloway. 1964 Establishment of the cinnabar moth, Tyria jacobaeae on tansy ragwort in the Western United States. J. Econ. Entom. 57:152-154. Gadgil, M. 1971 Dispersal: population consequences and evolution. Ecology 52:253-261. Green, Wren Q. 1974 An Antagonistic Host/Plant System: The Problem of Persistence. PhD. Thesis, UBC, Vancouver. Gruys, P. 1970 Growth in Bupalus piniarius (Lepidoptera: Geometridae) in relation to larval population density. I. The influence of some abiotic factors. II. The effect of larval density. Verhandelingen No. 1 of the Research Institute for Nature Management. 127pp. Gruys, P. 1970 Mutual interference i n Bupalus piniarius (Lepidoptera: Geometridae) Proc. Adv. Study Inst. Dynamics Numbers Popul. (Oosterbeek, 1970) 199-207. 89 Harcourt, D.G. 196b. Spatial pattern of the Imported Cabbageworm, Pieris rapae (L.) (Lepidoptera: Pieridae), on Cultivated Cruciferae. Can. Ent. XCIII:945-952. Harris, P. 1972 Insects in the population dynamics of plants. Symposia of the Royal Entom. Sco. of London. No. 6. Blackwell Scientific Pub. Harris, P. 1974 History of Cinnabar Moth Releases in Canada. MS. Harris, P., Wilkinson, A.T.S., Neary, M.E. and Thompson, L.S. 1971 Senecio jacobaeae L. tansy ragwort (Compositae) In: Biological control programmes against insects and weeds in Canada - Tech. Commun. Commonwealth Inst. Biol. Control, (No. 4), Farnham Royal Eng. 97 - 104. Hawkes, R.B. 1968 The cinnabar moth, Tyria jacobaeae for control of tansy ragwort. J. Econ. Entom. 61:499-501. Isaacson, D.L. 1973 A Life Table for the Cinnabar Moth, Tyria jacobaeae, in Oregon. Entomophaga, 18:291-303. Johnson, G.C. 1969 Migration and Dispersal of Insects by Flight. Metheun and Co., London. Klomp, H. 1966 The dynamics of a f i e l d population of the pine looper, Bupalus piniarius L. (Lepidoptera:Geometridae) Adv. Ecol. Res. 3: 207-305. Long, D.B. 1953 Effects of Population Density on Larvae of Lepidoptera Trans. R. Ent. Soc. London 104:543-585. Long, D.gj. 1955 Observations on SubSocial Behaviour in Two Species of Lepidopterous Larvae, Pieris brassicae L. and Plusia gamma L. Trans. R. Ent. Soc. London 106:421-437. Nagel, W.P. and Isaacson, D.L. 1974 Tyria jacobaeae and Tansy Ragwort in Western Oregon. Jour. Econ. Ent. 67:494-496. Roff, Derek. 1974a Spatial Heterogeneity and the Persistence of Popula-tions. Oecologial 15:245-258. Roff, Derek. 1974b The Analysis of a Population Model demonstrating the importance of dispersal in a heterogenous environment. Oecologia 15:259-275. Roff, Derek. Population Stability and the Evolution of Dispersal i n a Heterogenous Environment. MS. Southwood, T.R.E. 1971 Ecological Methods. Chapman and Hall, London. Shaw, M.J.P. 1970c Effects of Population Density on alenicolae of Aphis fabae Scop. Ann. appl. Biol. 65:205-212. 90 Van der Meijden, E. 1970 Senecio and (Callimorpha) Tyria in a Dutch dune area. A study on an interaction between a monophagous consumer and i t s host plant. Proc.adv. Study Inst. Dynamics Numbers Popul. (Oosterbeek, 1970) 390-404. Van Valen, L. 1971 Group selection and the evolution of dispersal. Evolution 25:591-598. Waters, William E. 1959 A Quantitative Measure of Aggregation in Insects. J. Econ. Entom. 52:1180-1184. Way, M.J. and Cammell, M.E. 1970 Self regulation in aphid populations Proc. Adv. Study Inst. Dynamics Numbers Popul. (Oosterbeek, 1970) 232-242. Wellington, W.G. 1955 Solar Heat and Plane Polarized Light versus the Light Compass Reaction in the Orientation of Insects on the Ground. Annals Entom. Soc. Am. 48:67-76. Wellington, W.G. 1957 Individual differences as a Factor in Population Dynamics: The Development of a Problem. Can J. Zool. 35:293-323. Wellington, W.G. 1965 Some Maternal Influences on Progeny Quality in the Western Tent Caterpillar, Malacosoma pluviale (Dyar). Can.Ent. 97:1-14. Wilkinson, A.T.S. 1965 Release of Cinnabar Moth, Hypocrita jacobaeae (L.), (Lepidoptera:Arctiidae) on Tansy Ragwort in British Columbia. Proc. Entom. Soc. B.C. 62:10-13. APPENDIX Environment Room Experiments - Egg Batch Order and Disappearance Tyria females lay their eggs in 4-5 clusters or batches of about 40 eggs. This partitioning of a female's reserves among early-laid and late-laid eggs lends i t s e l f to a qualitative comparison of progeny from early-laid and late-laid eggs, as suggested by Wellington (1965). Since indications from some 1973 work done on between batch variation in tend-ency to disperse were inconclusive, I decided to investigate the i n f l u -ence of egg batch order on dispersal behaviour during the spring of 1974. Attempts to bring pupae collected at Cultus Lake December 18, 1973 out of their winter diapause failed. It was not u n t i l late February with Cultus Lake pupae collected February 4, that we succeeded in break-ing diapause, and moths began emerging. In early March we received pupae from Silbernagel site, Oregon and they produced moths late that month. Thus the environment room rearing program and dispersal experiments were carried out with larvae from these two study sites. The method used was as follows. Mated female moths were allowed to lay as many egg clusters as they could. Progeny of these clusters were reared in cardboard containers on foodplant from the greenhouse, and were identified as to female parent and egg batch number. When the larvae reached the third instar, they were separated into containers of 5 each. At least once during the 5th instar, larvae from each container were tested in a plastic dispersal box as shown in Figure 14. The five larvae to be tested were placed on 'plant' 1 either on the ragwort leaf or in the space above the foam plug. After 2 1/2 or 3 hours, the number of larvae s t i l l on plant 1, the number on plant 2 and the number wandering around in the box or completely missing were recorded. 92 PLANT 1 1 2 " PLANT 2 FIG. 14. ENVIRONMENT ROOM DISPERSAL BOX 93 The data for each of the 6 females i s i n Table XXV. There was no s i g n i f i c a n t d i f f e r e n c e i n mean proportion missing between progeny of the d i f f e r e n t Cultus Lake females. ( A l l batches were combined.) Nor were there differences between the successive batches when a l l data from the four Cultus Lake females were combined. The larvae from Oregon tended to leave the plant s i g n i f i c a n t l y more than those from Cultus Lake, but there was no s i g n i f i c a n t d i f f e r e n c e between successive Oregon batches 2 as measured by a Chi homogeneity t e s t . The f a c t that there i s no s i g -n i f i c a n t d i f f e r e n c e i n d i s p e r s a l performance of larvae from d i f f e r e n t batches can, I think, be s a f e l y assumed i n the analysis of disappearance. T r i a l s under more natural conditions than these could however prove d i f f e r e n t . The proportion of larvae that were found on plant 2 i s not r e l i a b l e . The l e a f covering the v i a l and enabling larvae to make contact with f o l i a g e rather than with glass as they approached plant 2 was not used fo r e a rly t r i a l s of the Cultus Lake larvae, so i t i s d i f f i c u l t to compare the r e s u l t s . TABLE XXV DISAPPEARANCE AND EGG BATCH ORDER Cultus + 8 Lake * 13 •? 26 + 32 Means for a l l Cultus females Oregon 1 .10 .25 .10 .16 .15 1 .04 .56 2 .12 .06 0 .22 .10 t .05 .50 3 .11 0 0 .04 1 .04 .37 4 .10 .11 .10 .10 1 .003 .35 5 .20 .10 .15 1 .05 .50 6 0 .15 .30 7 .10 x=.13^.02 x= » . 0 8 - . 0 6 x - .09- .02 x- .12- .05 x - . 4 3 l . 0 4 No significant differences No significant differences 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items