mill Will/ll? ( l 3 1293 00675 9934 W l Will :‘7 This is to certify that the thesis entitled A BEHAVIORAL ANALYSIS OF OVIPOSITIONAL DISCRIMINATION IN THE CABBAGE BUTTERFLY (PIERIS RAPAE (L.)) presented by BETH ANN BISHOP has been accepted towards fulfillment of the requirements for MASTER OF SCIENCE degree in ENIQMQIQGX Date //////35 /Z4M/%%I //—-/end) and visiting (seq-->V) for different plant species. , Number of visiting butterflies alighting (V-->1-c) and ending (V-->end) for different plant species. Number of alighting butterflies curling (l-c-->cu) and ending (l-c-->end) for different plant species. Number of butterflies entering the soil space that entered the plant space (s-->p) and that ended (s-->end) for different plant species. Number of butterflies entering the plant space visiting (p-->V) and ending (p-->end) for different plant species. 42 46 49 63 70 78 80 84 87 90 93 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. xi Number of butterflies entering the target space via the soil space (5) and from above (Ap) for different plant species. Number of butterflies fluttering, contacting and landing as the first act in visits to different plant species. Number of butterflies contacting, landing and ending following an initial flutter above different plant species. Number of butterflies contacting and landing on their initial alight on different plant species. Number of butterflies ending and continuing after their first alight on different non-host species. Number of butterflies performing a flutter, a contact, a land, a land-curl, or an end on steps 1 through 4 of visits to the broccoli target (comparison I). Number of butterflies performing specified transitions to all possible behaviors following a flutter, a contact, and a land as a function of step number. Number of butterflies performing specified transitions to all possible behaviors following a flutter, a contact, a land, and a land-curl as a function of "experience". Chi-Square test for quasi-independence of preceding behavior from following behavior on table including an impossible transition (flutter-to-flutter). Percentage of butterflies making the same transition between behaviors for the same plant (broccoli or tansy) in different comparisons (different sampling dates). Mean length of time Of sampling intervals (time taken to record sequences) for different target plant species. Summary of the hypotheses, predictions, and results from the first section of the results for hosts vs non-hosts. 96 105 109 112 125 134 135 137 142 174 184 191 24. 25. 26. xii Summary of the hypotheses, predictions, and 195 results from the first section of the results for hosts of differing acceptabilities. Summary of the hypotheses, predictions, and 199 results from the first section of the results for different non-host species. Summary of the hypotheses, predictions, and 203 results from first section of the results for repellent vs non-repellent plants. 1A. LIST OF FIGURES Female Pieris rapae. 18. Male Pieris rapae. 2. 4A. 4B. Hydrolysis reaction of thioglucosides to isothiocyanates a) R=CH2=CH-CH2--Sinigrin, b) R:p-HOC6H4CH2--Sinalbin. (From Kjaer 1963, 1976) Map of the Field Cage Interior. -T=target plant, P=plant area, s=soil area, RH=reserve host, F=cut flowers. Spatial representation of hypothetical sequence of behaviors of Pieris rapae in target area. s=soil space, p=plant space, A=enter target area from above, v=visit, f=f1utter, l=land, c=contact, cu=curl, end=leave target area. Diagramatical representation of hypothetical sequence of behavior of Pieris rapae in target area. s=soil space, p=plant space, A=enter target area from above, Vevisit, f=flutter, l=land, c=contact, cu=curl, end=leave target area. Organization of the analysis in first part of the Results. Data is analyzed on successive Levels (I through IV). Each subdivision of a level represents one test. Statistical test used is given in lower right corner of box (G = G test, ANOVA = Analysis of variance). Seq=sequence, v=visit, end=leave target area, l=land, c=contact, f=flutter, cu=curl, Ap=enter target area from above, s=soil space, p=plant space, cont=continue (no curl but reland). xiii 13 13 17 43 52 53 65 10. 11. 12. 13. 14. xiv Distribution of the number of eggs laid per sampling interval on the broccoli target plant (comparison I). Percentage of butterflies entering the plant area from above (Ap) and via the soil space (s) that visited broccoli, tansy, sage and mustard. Percentages shown under the same letter are not significantly different at the .05 level. Percentage of butterflies entering the soil space that left the target area without visiting the plant, for the broccoli and tansy target plants (comparison I). Number of butterflies performing each behavior in the alight stage of oviposition for the broccoli target (comparison I). The width of the boxes is proportional to the number of butterflies performing the designated behavior (this number is given in the box). The width of the arrows is proportional to the transitional probabilities (shown next to the arrow). The total area of the boxes is insignificant. \ Percentage of butterflies curling, continuing and ending on .successive alights on broccoli (comparison I). Percentage of butterflies performing their first curl on the first through fifth alight on broccoli (comparison I). Percentage of alighting butterflies that rested only on differpent plant species. Percentages shown under the same letter are not significantly different at the .05 level. Br=broccoli, T=tansy, S=sage, M=mustard, L=lettuce, Sy=soybean, BT=broccoli-tansy, =data not analyzed due to the low number of visits occurring. Number of butterflies curling, continuing and ending after the first three alights on broccoli and tansy (comparison I). The number in parentheses is the percentage of butterflies making that particular alighting that performed the designated behavior. Percentage of butterflies curling, continuing, and ending after the first three alights on broccoli and mustard (comparison III). 75 99 102 114 116 118 120 121 123 15. 16. 17. 18. XV Percentage of butterflies curling, continuing, and ending after the first three alights on broccoli and broccoli-tansy (comparison V). Percentage of butterflies curling, ending, and continuing after making their first alight on the broccoli leaves, and on the tansy leaves of the broccoli-tansy target. Percentage of butterflies fluttering, contacting, landing, and landing with a curl as the first behavior: 1) in a visit, 2) following the first alight, 3) following the first curl, and 4)following the second curl. The visit sequence to the broccoli target (comparison I). The size of the boxes is proportional to the relative frequency of the behavior. The width of the arrows is . proportional to the transitional probability 19. between behaviors (given next to the arrow). The relative proportion of the behavior as the first act in a visit is shown by the arrow in the upper right corner of each box. The visit sequence to the tansy target (comparison I). The size of the boxes is . proportional to the relative frequency of the 20. 21. 22. behavior. The width of the arrows is proportional to the transitional probability between behaviors (given next to the arrow). the relative proportion of the behavior as the first act in a visit is shown by the arrow in the upper right corner of each box. The visit sequence to the mustard target (comparison I). The size of the boxes is proportional to the relative frequency of the behavior. The width of the arrows is proportional to the transitional probability between behaviors (given next to the arrow). The relative proportion of the behavior as the first act in a visit is shown by the arrow in the upper right corner of each box. Percentage of butterflies ending (not making additional curls) after curling on broccoli and mustard as a function of curl number. Percent communication from plant to butterfly during different stages of the ovipositional sequence. 126 128 139 144 145 146 150 157 23 24 26 27 28. 29. 30. 23. 24. 25. 26. 27. 28. 29. 30. xvi Percent communication from plant to butterfly during different steps in the ovipositional sequence. Percent relative discrimination between hosts and non-hosts, hosts of differing acceptabilities, and different species of non-hosts during different stages in the ovipositional sequence. Percent relative discrimination between hosts and non-hosts, hosts of differing acceptabilities, and different species of non-hosts during different steps in the ovipositional sequence. Eggs laid per sampling interval vs visits per sampling interval for broccoli (comparison I). r=.758 Eggs laid per sampling interval visits per sampling interval for mustard (comparison III). r=.856 Eggs laid per sampling interval vs visits per sampling interval for broccoli-tansy (comparison V). r=.866 Percentage of butterflies performing ending behaviors in the feeding, oviposition, and ending categories as a function of how far they progressed through the sequence. Percentage of butterflies exhibiting ending behavior classed as "ovipositional" that directed their ending behavior toward hosts rather than toward non-hosts as a function of how far they progressed in the ovipositional sequence. 160 164 166 170 171 172 180 182 LIST OF TERMS Term Definition Alight Stage A sequence stage. Behaviors performed after the first alight up to the first abdominal curl Curl Stage A sequence stage. Behaviors performed after the first abdominal curl Movement Stage A sequence stage. Behaviors performed before the visit. Plant Space The inner cylinder of space in the target area. Reserve Hosts Host plants that were placed in the field cage among the wild vegetation to provide the butterflies with suitable ovipositional sites and to collect eggs for maintenance of the butterfly culture. Response Stage A sequence stage. Behaviors performed from the first visit behavior up to the first alight (land or contact). Sampling Date A day on which sequences were observed and recorded. sampling Interval The period of time during which behavioral sequences performed in response to a single target plant were observed and recorded. xvii xviii Sequence Stage A group of consecutive behaviors in the ovipositional sequence. Four sequence stages were identified: the movement stage, the response stage, the alight stage and the curl stage. Sequence Step Corresponds to the performance of one behavior in the ovipositional sequence. Soil Space The outer cylinder of space in the target area. Target Area The area inside the field cage in which the behavior of butterflies was observed and recorded. Target Plant The plant used during an experiment. Butterfly behavior in response to the target plant was recorded. rn INTRODUCTION To a phytophagous insect, different species of plants are not equally acceptable as food. Even insect species that feed on a variety of plants consume some plant species heavily, others very little, and still others not at all. For example, larvae of the Colorado potato beetle (Leptinotarsa decemlineata [Say]) feed mainly on plants in the family Solanaceae, but will accept some species in the families Caesalpiniaceae, Asclepiadaceae, Cruciferae and Compositae. Other plants are not eaten (Hsiao and Fraenkel 1968). Even insects considered to be generalist feeders, such as grasshoppers, are selective in diet. For example, the sulfur-winged grasshopper (Arphia sulphurea [Fab.]) feeds mainly on grasses. Although A; sulphurea may consume some herbaceous species, others are not accepted (Gangwere 1965). Because insects are often apart from acceptable food, they must be able to find (Miller and Strickler 1984) and distinguish acceptable from non-acceptable plants. By so doing they discriminate among plants. Discrimination among objects is based on perception. All information about the external environment that an animal has access to is received by its sensory organs and interpreted by its nervous system. If the information 2 received from two objects differs, and if the animal registers this difference physiologically, then the animal's perceptions of the two objects differ. The perceptual basis of observed discrimination patterns in the Insecta has long intrigued humans. There are at least two reasons why understanding an insect's perceptions may be desirable. First, many species of phytophagous insects are pests of crops. By understanding the perceptual basis for discrimination, humans have the potential for manipulating relevant plant stimuli and thus changing the acceptability of a plant (Beck 1965). Secondly, the realization that: 1) An animal's perceptions are influenced by the structure of its sensory and neurological equipment, and 2) that each species, by virtue of its unique evolution, possesses slightly different equipment, leads to the inference that the perceptions of our fellow animals may be fundamentally different from our own (von Uexkell 1957, Dethier 1969). An understanding of the perceptions of insects allows us to expand our own world by making theirs partially accessible to us. Humans are incapable of knowing directly another animal's perceptions: but, we may gain knowledge about these perceptions by observing behavioral discrimination (e.g., differences in Observed behavior toward two Objects). Dethier (1969) loosely defines perception as "How the world looks", or "What things are the same and what things are different." He explains that we may infer perception when 3 an animal behaves "as though" Objects are the same or "as though" they are different. For example, if a phytophagous insect exhibits identical behavior toward two different plant species then the possibility exists that they are perceived as essentially the same for that insect at that time. Any qualitative or quantitative differences in stimuli perceived by humans (either unaided or with the help of instruments) are either not perceived by the insect or do not influence behavior. Conversely, if the responses toward the two plant species differ, then we may assume that they are perceived differently by the insect. By examining characteristics that differ between the plant species, one may at least speculate as to the stimuli responsible for perceptual differences. In this way discrimination gives us a "window" on the perceptual world of phytophagous insects. Discrimination between plants by insects is often assessed by comparing criteria (such as tissue damage or number of individuals per unit) that are the net result of all responses made by many individual insects of a given species toward a plant. As many authors have pointed out, the behavior that results in plant use is, in reality, a sequence of separate responses, each mediated by perceptions (Dethier 1953, Thorsteinson 1960, Beck 1963, 1965, Kennedy 1965). It is important to remember this sequence when investigating the perceptual basis of discrimination. One cannot assume that an insect's perceptions remain constant throughout the sequence. Certain stimuli may become more or 4 less important as the sequence progresses (Kennedy 1965). Because an animal's perceptions may change as the sequence progresses, so too may the observed degree of discrimination. The present study investigates discrimination between plants during the ovipositional sequence of the cabbage butterfly (Pieris rapae [L.]), an oligophagous butterfly. Oligophagous butterflies pose several challenges to development of an understanding of the perceptual basis of discrimination. The first challenge lies with the nature of the animal. Because butterfly larvae are relatively small and immobile, survival depends, to a large extent, on a female's ability to place her eggs on the proper food plants. Since adults are nectar feeders, they usually do not maintain an association with larval food plants except during oviposition. Adults are relatively mobile: during daily flight a female usually encounters a great variety of plants, yet she is able to distinguish those species crucial to offspring survival. The second challenge to understanding the perceptual basis of discrimination in oligophagous butterflies lies with the nature of oligophagy. Oligophagous insects feed on a limited range of plants, often taxonomically related (Thorsteinson 1958). Species of host plants are often diverse, appearing (to a human observer) to be morphologically as different from one another as they are from non-hosts. Yet there are sharp differences in the 5 number of eggs laid on hosts as compared with the number laid on non-hosts. g; £2222: for example, colonizes mostly plants within the family Cruciferae. Plants in this family are extremely diverse, yet a "mistake" by a g; rapae female (i.e., an egg laid on a non-host) is rare. This pattern implies that the butterflies discriminate behaviorally between hosts and nOn-hosts at one or more points in the ovipositional sequence. The ability to respond appropriately to a diverse group of objects, such as host plants, must be due to a "class recognition" based upon the ability to perceive the constant and unique class characteristics despite variation in these characteristics and amid large amounts of extraneous information. Brues (1920) explained it this way: "Undoubtedly there is some attribute of such plants that insects can recognize in a general way and not as a specific characteristic of some single plant species or genus." Understanding just what this characteristic (or group of characteristics) is lies at the heart of understanding the perceptual basis of host plant discrimination by Oligophagous insects. For g; rapae this "class characteristic" has long been thought to be thioglucosides: secondary plant chemicals produced by hosts (Verschaffelt 1911). However, some recent research has suggested that a E; rapae female is able to perceive these host-specific chemicals only after alighting on a plant (Hovanitz and Chang 1964, Renwick and Radke 1983, 6 Traynier 1979). If so, then this question is raised: Is ovipositional behavior before alightment non-discriminatory? Does g; rapae alight randomly on plants, distinguishing hosts from non-hosts only after alighting has occurred? The present study was designed to address the above questions. Emphasis was placed on the sequential nature of oviposition. The approach taken was to Observe the ovipositional sequence of E; rapae in a semi-natural setting, with the goal of identifying the responses in the behavioral sequence where discrimination is exhibited, and analyzing the role played by discrimination during each response in determining the overall acceptability of the . plant. The specific objectives of the study were: 1. To describe the sequence of behaviors leading to P; rapae oviposition on a normal host plant. The temporal order of behaviors would then serve as a basis for comparing sequences performed in response to different plant species. 2. To determine where in the ovipositional sequence discrimination occurs. Discrimination was assessed between the following plant types: 1) hosts vs non-hosts, ii) hosts of different acceptabilities, iii) different species of non- hosts, and iv) repellent vs non-repellent plants. 3. Comparison of the relative degree of discrimination: i) occurring between hosts and non-hosts at different points of the ovipositional sequence, and 7 ii) occurring among different plant types at the the same point in the ovipositional sequence. In summary, the purpose of this study was to assess discrimination by comparing behaviors toward different plants. The nature of the stimuli influencing the performance of such behaviors is not directly addressed. However, a better understanding of discrimination patterns is the first step to discovering the nature of the stimuli involved. LITERATURE REVIEW The Animal The cabbage butterfly, Pieris rapae (L.), is a member of the cosmopolitan family Pieridae, which includes mostly small to medium-sized butterflies. The family is represented by two North American subfamilies: Colianae, the sulfurs, most species of which feed as larvae on legumes, and Pierinae, the whites, most species of which are larval specialists on crucifers and capers (Opler and Krizek 1984). P; rapae belongs to the later subfamily, which includes a large number of genera and species (see Ehrlich and Raven 1964). As mentioned previously, most temperate species are crucifer specialists (Cruciferae) although plants in the related families, Caparidaceae, Resedaceae, and Tropaeolaceae are sometimes utilized. Eleven species of Pierinae are found in the Eastern U.S. alone. Where the 8 geographical distribution of different species overlaps, niche separation occurs by a variety of behavioral and ecological mechanisms including: adult habitat, range and acceptability of different species of potential host plants, and differences in the part of the plant utilized by larvae (Shapiro 1974, 1975: Chew 1977, Opler and Krizek 1984). 21.52222 is endemic to Eurasia and North Africa, but has been introduced to many other parts of the world, including New Zealand (Moss 1933), Australia, Bermuda, Hawaii, and other Pacific islands (Opler and Krizek 1984). The animal was accidentally introduced to the North American continent in 1860, near Quebec, Canada and its spread was rapid (Scudder 1887). By the early 1900's the invasion was so complete that one W. Virginia agricultural bulletin not only reported a serious effect of the larvae (i.e., "cabbage worms") on cole crops where none existed previously, but also reported that the invader was displacing and restricting native Pierinae species (Rumsey and Brooks 1909). The present day range of nggapae in N. America extends from coast to coast and from central Canada to northern Mexico. Pieris rapae has had a significant impact on cultivated cruciferous crops. Prior to the advent of modern insecticides, at least one tenth of the total cabbage crop was lost to larval feeding (Twinn 1924). Although today direct crop losses are probably minimal due to the use of insecticides, the cost of control is high, resulting in 9 significant increases in production costs. For example, in 1979 89,000 acres of cabbage were planted in the U.S. for the fresh market, 67,000 of which were treated for insect control (Ferguson 1984). Over 193,400 pounds of insecticides were used. At 1984 prices, this translates to a insecticide cost of approximately $3.5 million for cabbage alone (derived from data supplied by E. Grafius, personal communication). The primary damage to cabbage is done by the late—instar larvae which burrow into the head of the plant and deposit large amounts of green frass, thereby reducing market value (Harcourt 1963). Although the success of species introduced into new areas without their biological control agents is well known, there is perhaps an additional reason accounting for the success of 2;.52232 as a colonizer. Despite a narrow range of acceptable food plants as larvae, the adult butterfly is very adaptable. Extremes in temperature seem to affect it much less than other butterfly species. Where occurring seasonally, g; rapae is among the first butterflies to emerge in the spring and among the last to disappear before winter: e.g., they occur between the middle of May and the first of September in Michigan's upper peninsula, from late March to early November in Virginia, and year round in Mississippi (Opler and Krizek 1984). Where 3; rapae occurs seasonally there is usually sufficient time for a number of generations to occur. Development from egg to adult requires between 37 days 10 (at 17 degrees C) to 22 days (at 27 degrees C) (Richards 1940). Females lay eggs as long as temperatures and the presence of hosts permit. In some regions of the Southern United States as many as 6 to 7 generations a year may occur (Opler and Krizek 1984). During the average 3 weeks or 156 degree days (base 10 degrees C) (Gossard and Jones 1977), a female has the potential to lay a large number of eggs: laboratory counts of the total number of eggs laid range from 300 to 500 (Moss 1933, Rahman 1968, Twinn 1924). Gossard and Jones (1977) measured egg production in a field cage and found that it ranged up to fifty eggs per female per day, depending on age and temperature. As a result of this high fecundity, in combination with the propensity of females to cover up to 700 meters per day (Jones et al. 1980), an individual female is able to distribute her eggs over a wide area. Females are reported to fly in a great diversity of habitats, being rare or absent only in canopied forests and in open areas not supporting crucifers (Opler and Krizek 1984). Adaptability in food and host requirements is probably responsible for this. ‘g;‘rgp§g is reported to visit a large number of flower species, even those not believed to possess nectar (Opler and Krizek 1984). Additionally, eggs are laid on a large number of host species. In this respect, 3; EEEEE is less restricted than most other Pieris species which 11 oviposit on a limited number of crucifer species (Opler and Krizek 1984). Given the mobility of females, the high reproductive potential, and the relative non-specificity of adult food and habitat requirements, it is easy to understand the success this insect has achieved as a colonizer. The life cycle of P; rapae has been described by several authors including: Twinn (1924), Richards (1940), and Harcourt (1963). Eggs are laid singly on host plants. Oogenesis requires 4 to 9 days, depending on temperature (Moss 1933, Richards 1940). There are 5 larval instars. First instar larvae are less than 4 mm long and feed primarily on the outer leaves of plants. The last two instars feed mainly on younger leaves, burrowing into the heads of plants such as cauliflower, cabbage, and broccoli. Older larvae are capable of moving to another plant upon consumption of the original host. ApprOximately one day prior to pupation, the caterpillar stops feeding and begins to move to a pupation site. Although some may pupate on the same host plant, larvae more commonly move to another object, such as a fence post, outbuilding, or another plant, climb it, and then pupate. Pupae are green or brown, pointed on either end with three ridge-like projections on the dorsal and lateral surface, and are attached to the substrate by a silken 12 girdle. Pupation may take from 6 to 11 days depending on temperature (Richards 1940). Adults are mostly white, with black markings on the tips of the forewings, one black spot on the dorsal side of the hind wing, and one (males) or two (females) black spots on the dorsal surface of the forewing (Figure 1). The underside of the wings is speckled with yellow. Females usually mate and begin to oviposit within 24 hours of emergence (weather permitting). Males patrol a home range, mating with females emerging from pupation in this area (W.G. Wellington, personal communication in Jones et al. 1980). Males are probably stimulated to approach females, at least from close range, by visual stimuli (Obara and Hidaka 1968, Obara 1970). Egg laying begins shortly after mating. A female lands on a plant, curls the tip of her abdomen downward until it touches the (usually) underside of the leaf, and then withdraws, leaving a single egg glued to the leaf surface. Approximately 85% of the eggs are laid on the underside of leaves (Richards 1940). Most of the time, successive ovipositions are separated by at least a short intervening flight. Jones et al.(l980) and Root and Kareiva (1984) both found that individual females were highly directional, tending to fly in the same general direction over the course of a day's flight. An individual's "preferred" direction Figure 1A. Female Pieris rapae. Figure 1B. Male Pieris rapae. 14 varied from day to day and was not dependent on any environmental cues such as sunlight or wind direction. Host Plants The host plant range of £;.£EEEE includes a large number of morphologically diverse plants, mostly within the families Capparidaceae, Resedaceae, Tropaeolaceae and Cruciferae (Twinn 1924). Members of the latter family are the usual hosts of most temperate Pierinae species (including 2; rapae). Tropical Pierinae species mostly utilize capers (Capparidaceae). The family Cruciferae is large and diverse, containing approximately 350 genera and 2,500 species, most of which are annual or bienniel herbs distributed in the cooler regions of the northern hemisphere (Lawrence 1951). Most crucifers, as members of this family are commonly known, are adapted to disturbed habitats. They include many common weeds of roadsides and cultivated fields. In addition, several species have been selectively bred and cultivated by humans. The most well-known cultivated species include varieties of Brassica oleraceae: cauliflower (B; oleracea var. botrytis), collards (B; oleracea var. acephala), kale (B; oleracea var. acephala), cabbage (B; oleracea var. 23215333), broccoli (B; oleracea var. italica), brussel sprouts (E; oleracea var. gemmifera), and other species including: mustard (B; juncea), turnip (E; campestris, radish (Rhaphanus §pp), and watercress (Nasturtium 15 officianale). Varieties of B; oleraceae have been cultivated for hundreds, perhaps thousands of years (Nieuhof 1969). Long-term selective breeding programs have resulted in the creation of "infraspecific complexes and artificial hybrids", making taxonomic relationships difficult to decipher (Kjaer 1976). To humans crucifers share few morphological characteristics: only the anatomy of the flowers is common to all species. Most species also have simple, alternate leaves and many are characterized by the presence of forked or stellate unicellular hairs (Lawrence 1951). Biochemically, all crucifers are characterized by the presence of thioglucosides in living tissue. Thioglucosides are sulfur-containing chemicals synthesized by the plant (for biosynthesis, see Kjaer 1976) and present in parenchyma cells (Stevens 1924). Since their discovery over 100 years ago, more the? 70 naturally-occurring thioglucosides have been identifi;',(Kjaer 1976). Thioglucosides are considered to be present‘lw . - plants in the families Cruciferae, Resedaceae, C2-azridaceae, and Moringaceae: and in at least some species of several other families, including Tropaeolaceae (Kjaer 1963). A typical piece of cruciferous tissue contains a complex of different thioglucosides, the composition of which varies from organ to organ (Josefsson 1967). There is evidence that the qualitative pattern of thioglucosides is under genetic control (Hemingway et al. 1969). 16 Thioglucoside pattern varies between species, and between varieties in the case of cultivated varieties of B; oleracea (Josefsson 1967, Nair et al. 1976, Rodman and Chew 1980). Environmental and developmental factors may also affect thioglucoside composition quantitatively. Younger leaves contain higher concentrations of thioglucosides than older leaves (van Emden and Bashford 1969). Nutrient regimes deficient in sulfur and those deficient in nitrogen reduce and increase thioglucoside concentration, respectively (Wolfson 1980). When plant tissue is damaged, thioglucosides may undergo enzymatic hydrolysis to glucose, sulfuric acid and a volatile isothiocyanate (Kjaer 1963) (see Figure 2). Each thioglucoside yields, upon hydrolysis, a unique iso- thiocyanate (Josefsson 1967). The reaction is catalyzed by the enzyme myrosinase, which is contained in specialized idioblasts (Esau 1977) and is known to be present in all thioglucoside-containing plants (Bjorkman 1976). The enzyme has also been identified in a fungus (Aspergillus sydowii), a bacterium (Paracorobactrium aeroqenoides), the cabbage aphid (Brevicoryne brassicae), and mammalian tissue (Bjorkman 1976). A number of different myrosinase isoenzymes have been identified, but individual enzymes do not appear to be adapted to particular thioglucosides (Bjorkman 1976). As with the parent compounds, different isoenzymes co-occur in plant tissue, forming complexes that vary from one organ to 17 C"OS 3 °\N OH __ / HO H H OH Thioglucoside Enzyme-induced detachment of glucose. /N/ —— < HS C R Molecular rearrangement. Isothiocyanate R - NCS q. 9 H 804 Figure 2. Hydrolysis reaction Of thioglucosides to isothiocyanates . a) R: CH2=CH-CH2—-Sinagrin, b) Ft= p—HOC6H4CH2—-Sinalbin. (from Kjaer 1963, 1976) 18 the next (Macgibbon and Allison 1970). Nutrition and habitat seem to influence the abundance of myrosinase- containing idioblasts a plant possesses. When development is checked, the abundance of these cells increases (Solereder 1909). In addition, there may be a genetic component to enzyme activity, as it varies from species to species (Bjorkman 1976). Because of the multitude of factors resulting in differences between plants in both substrate and enzyme, major differences in isothiocyanate production are not surprising. Different parts of a single plant may vary in the amount of isothiocyanate produced (Bjorkman 1976). Growing conditions may affect isothiocyanate production; cabbage plants grown earlier in the season produce over 100 times more allyl isothiocyanate (the hydrolysis product of the thioglucoside sinigrin) than plants grown later in the season. Also, brussel sprout and cabbage plants grown 1 ft apart produce 20 times more allyl isothiocyanate than those spaced 3 ft apart (MacLeod 1976). In summary, the total thioglucoside picture is exceedingly complex. Different plant species may have unique gentically- determined chemical profiles subject to modification by numerous environmental factors. Yet all plants serving as hosts for g; rapae contain thioglucosides. As might be expected, thioglucosides play an important role in the host specificity of 21.59292 and other insects specializing on crucifers. Over the years, the relationship 19 between crucifer chemistry and crucifer-specialist insects has been well studied and documented. Thioglucosides in crucifers are one of the the best supported examples of a plant secondary compound originally evolving as a defense against predators and pathogens and later becoming a releasing stimulus to insect species able to overcome the defensive barrier. As evidence for a defensive function, thioglucosides (and/or isothiocyanates) have been found to function as a fungicide, an antibiotic, a mammalian digestional irritant (Feeny 1977), and an insect feeding inhibitor (Eriksonand Feeny 1974). Furthermore, some cruciferous species (Lepidium spp and Thlaspi spp) contain atypical enzymes that degrade thioglucosides to the geometrical isomers of isothiocyanates, the thiocyanates (Kjaer 1976). Such plants typically support poor growth or are lethal to some Pieris species (Chew 1975). A number of insects have overcome the thioglucoside defensive chemical barrier. For example, the green peach aphid, Myzus persicus, appears able to tolerate these compounds, although its growth is negatively correlated with the total allyl isothiocyanate content of the leaves (van Emden 1972). Other crucifer specialist insects actually require thioglucosides or isothiocyanates to initiate feeding or colonization. Besides influencing the ovipositional behavior and larval feeding behavior in Pieris rapae, thioglucosides and/or isothiocyanates have been found to play a role in the host specificity of cabbage aphids 20 (Brevicoryne brassicae, Wensler 1962), cabbage flea beetles (Phyllotreta cruciferae and g; striolata, Feeny et al. 1970), the cabbage root fly (Delia brassicae, Finch 1978), the diamondback moth, Plutella maculipennis (Thorsteinson 1952, Gupta and Thorsteinson 1960), and the large white butterfly, Pieris brassicae (David and Gardiner 1962, 1966). Response to Plant Stimuli Basic to understanding discrimination behavior in g; rapae is an understanding of the perceived plant characteristics that influence the behavior of ovipositing butterflies. Many studies have investigated the role played by selected stimuli in g; rapae ovipositional behavior. Most investigators use the number of eggs laid as the measured result. In such cases, it may be impossible to infer correctly which responses in the ovipositional sequence are involved (Jones 1977). Nevertheless, such studies form the basis of our knowledge. The following discussion considers the influence of plant stimuli of two basic types-~visual stimuli and chemical stimuli--on 2;.EEEEE ovipositional behavior. Vision potentially plays a role in all phases of oviposition, but traditionally has been considered to be more important in the earlier stages (Dethier 1953, Beck 1965). Furthermore, visual stimuli were considered to play a relatively minor role in host-plant specificity because visual aspects of host plants, as compared with 21 chemical aspects, were believed to be too variable to enable insects to distinguish hosts from non-hosts. More recently, emphasis has been placed on the interaction of several stimuli in determining host specificity (Dethier 1976, 1982, Schoonhoven 1977). Accordingly, vision has been accorded a more important role in ovipositional behavior. Prokopy and Owens (1983) state that "Numerous examples exist where both visual and chemical stimuli, operating sequentially or simultaneously, play a role in host plant location by a variety of insects." Three different visually perceived characteristics of plants potentially influence 2;.EEEEE ovipositional behavior: plant color, plant size, and plant shape. Color influences courtship behavior (Obara 1970, Obara and Hidaka 1968) and feeding behavior (Miyakawa 1976) of g; rapae. That color has a role in the ovipositional behavior of .2; rapae was shown by Hovanitz and Chang (1964) who found that most eggs were laid on blue-green artificial ovipositional substrates. Because of their use of the end result of oviposition (eggs) as a criterion for comparisons, Hovanitz and Chang's (1964) study did not reveal which behavior(s) in the ovipositional sequence were influenced by color. Later studies suggest that color affects g; rapae movement behavior, post-alighting behavior, and the probability of landing on a plant. Miyakawa (1976) found that butterflies were more likely to approach and fly over green backgrounds 22 than non-green backgrounds. In addition, the flight paths of ngrapgg butterflies flying over vegetation were characterized by considerably more turning than those over bare ground. This may have been due, in part, to a color difference between the two backgrounds. The influence of color on the post-alighting behavior of 2;.23222 has been demonstrated by Fox (1966). When cabbage butterflies were held on differently colored pieces of paper, tarsal drumming only occurred on green paper. The influence of color on the probability of alighting by g; £9222 was first suggested by Ives (1978) and later confirmed by Traynier (1984). Ives (1978) found that more 2; rapae eggs were laid on young plants than on older plants of the same size, and also on plants grown under low light intensities. This difference was in part explained by an increased probability of landing on the younger plants (Jones 1977). Ives noted that the plants grown under low light conditions were similar to the younger plants, having "fewer, larger leaves that were lighter in color, thinner and less fibrous." Thus, the difference in the probability of landing between young and old plants could have been due partially to a difference in color. Recently, Tranier (1984) has confirmed with laboratory experiments that alighting by P; rapae is influenced by color. The size of a host plant also affects the number of g; rapae eggs a species or variety receives (Ives 1978, Latheef and Irwin 1979). Jones (1977) found that plant size 23 affected the probability of landing on a host, but not the probability of laying an egg after landing. The probability of landing was also affected by the variety or species of host. The response to variety was found to override the response to size in that bigger varieties did not necessarily elicit more landings (Ives 1978). g; rapae ovipositional behavior may also be influenced by shape: either the overall shape of the plant (Prokopy and Owens 1983), or the shape of the leaves. Leaf shape is known to influence oviposition in the butterfly Battus philenor (Rausher 1978) and is believed to influence oviposition in other butterfly species (Stanton 1982, Feeny et a1. 1983). Miyakawa (1976) found that shape affected 2;.22222 feeding behavior, but the effect of shape on oviposition has not yet been investigated. The influence of chemicals on P; rapae ovipositional behavior has been intensely investigated. Particular emphasis has been placed on host-specific thioglucosides and isothiocyanates. Verschaffelt (1911) was the first to demonstrate a relationship between a phytophagous insect and the chemicals of its hosts when he noted that plants fed on by g; rapae larvae and the larvae of the closely related 3; brassica contained thioglucosides. He also demonstrated that larvae could be induced to feed on normally unacceptable food if it was smeared with juice from a host plant or with solutions of sinigrin (a thioglucoside). While Verschaffelt could not determine the relative roles of 24 gustation (of soluble thioglucosides) and olfaction (of volatile isothiocyanates), he postulated that olfaction would be relatively more important for host reCognition by ovipositing adults. Since thioglucosides are soluble, they are presumably perceived only after contact. Butterflies are known to possess tarsal contact chemoreceptors that are usually more numerous in females than in males (Fox 1966). Presumably, these sense organs facilitate ovipositing females in identification Of their host plants via characteristic chemicals perceived after landing. Perception of these chemicals may be enhanced by the drumming behavior often performed by females after landing (Ilse 1956). Since the tarsal chemoreceptor organs are always located beneath a tarsal spine, drumming presumably abrades the plant tissue, thereby releasing chemicals (Fox 1966). The physiological effect of thioglucosides on tarsal chemoreceptors of the large white butterfly (Pieris brassicae) was studied by Chun and Schoonhoven (1973). They found two types of trichoid sensilla located on the tarsi. A cell in type "B" sensilla was found to be responsive to GTA, a thioglucoside. Pieris rapae possesses tarsal hairs similar to g; brassicae (Fox 1966, Traynier 1979). Although the effect of thioglucosides on these hairs has not yet been investigated, it is assumed that they serve a similar function since both host plant extracts and commercial preparations of sinigrin 25 have been shown to elicit PL_rgpgg oviposition (Hovanitz and Chang 1964, Renwick and Radke 1983, Traynier 1984), and because thioglucosides are not perceived via abdomenal contact chemoreceptors in either 2; brassicae (Klijnstra 1982) or Pieris rapae (Traynier 1979). The number of g; rapae eggs laid is influenced by the concentration of host-plant extract or sinigrin. The findings of Renwick and Radke (1983) suggest that, although a single thioglucoside, such as sinigrin, stimulates oviposition, other chemicals are probably involved in determining the overall ovipositional response. Since host plant tissue typically contains a complex of different thioglucosides, it is possible that it is the particular combination and relative quantities of thioglucosides and/or other constituents that determine overall response (Dethier 1976), but this has not yet been investigated for g; rapae. Perception of host—specific chemicals via tarsal chemoreceptors affects Pieris rapae behavior for up to 72 hours after contact. Traynier (1979) found that g; rapae females that had previously contacted host foliage were more likely to land on plants and were also more likely to lay eggs on non-hosts than individuals that had, instead, previously contacted non-host foliage. In contrast to thioglucosides, isothiocyanates are volatile compounds which butterflies are potentially able to perceive prior to landing. Perception of isothiocyanates could result in host-specific behavior prior to landing. 26 There is circumstantial evidence that the large white butterfly (g; brassicae) uses host volatiles in its host finding (Mitchell 1978, Rothschild and Schoonhoven 1977, Behan and Schoonhoven 1978). Results for gL_r§p§g, however, have largely been negative. Traynier (1979) suggested that host plant volatiles played no role in 21.29292 oviposition because his caged butterflies were Observed to fly directly toward any green object placed in the cage (host foliage, non-host foliage, or yellow and green index cards). Allyl isothiocyanate (AITC) applied to artificial ovipositional substrates was found to elicit no more eggs than water (Hovanitz and Chang 1964). Ovipositional substrates possessing a source of host volatiles (cabbage leaves obscured from the butterflies' view) received no more eggs than those without (Renwick and Radke 1983). Additionally, high concentrations of AITC (Hovanitz and Chang 1964) and macerated cabbage (Renwick and Radke 1983) reduced the number of g; rapae eggs laid on artificial ovipositional substrates. The conclusion that volatiles play no role in g; rapae ovipositional behavior is based on results Obtained using butterflies confined to small cages. In such cases, butterflies are often deprived of normal releasing stimuli: this results in threshold lowering and a concomittent reduction in the degree of discrimination exhibited (Lorenz 1981, Singer 1982). In addition, both Hovanitz and Chang (1964) and Renwick and Radke (1983) Obtained their results 27 by counting eggs, not by observing behavior. As a result, the behavior can only be inferred and such inferences may be wrong (Jones 1977). Consequently, whether host volatiles affect g; rapae ovipositional behavior has not been established, but any such effect is not likely to be strong. Acceptability of Host Plants An ovipositional "mistake" is relatively rare in g; rapae. However, plants within the host-plant range may differ in the number of eggs they receive. A number of studies have compared the number of eggs received by different species or varieties or plants, or by plants differing in intrinsic characteristics such as age or size. In reporting results of these studies the term acceptability, which refers to the relative number of P; £3232 eggs a plant tends to receive in a choice situation, will be used. A highly acceptable host, on the average, receives more eggs than a host of lower acceptability, under similar conditions. Different species or varieties of host may differ in acceptability. Varieties of Brassica oleraceae usually are more acceptable than other host species (Dickson and Eckenrode 1975, Radcliffe and Chapman 1966a, Ives 1978). In some cases, the acceptabilities of different B; oleracea varieties differ (Harrison and Brubaker 1943, Ives 1978). However, the rank order of acceptability among different varieties is not consistent among studies. Furthermore, 28 Radcliffe and Chapman (1966a), using several cultivars of each variety, found that acceptability was more dependent upon the particular cultivar used than on variety. For example, several kale cultivars had low acceptabilities (relative to other E; oleracea varieties), while other kale cultivars were highly acceptable. In addition, results varied from year to year (Harrison and Brubaker 1943, Latheef and Irwin 1979a), from a greenhouse to a field situation (Dickson and Eckenrode 1975), and within a single growing season (Radcliffe and Chapman 1966b, Latheef and Irwin 1979a). Jones (1977) found that both the probability of landing on a particular host and the probability of laying an egg after landing was influenced by the variety of host involved. Thus, differences in acceptability are the result of differences in both pre-alighting and post-alighting ovipositional behavior. Red cabbages are reported to be less acceptable than green cabbages, presumably because of the color of their foliage (Opler and Krizek 1984). Results of several studies confirm this (Dickson and Eckenrode 1975, Radcliffe and Chapman 1966a, 1966b). Other results, however, refute this finding (Harrison and Brubaker 1943, Radcliffe and Chapman 1965, Latheef and Irwin 1979a). In any case, even when shown to be less acceptable than green cabbages, red cabbages still received more eggs than other species of 29 (green) hosts, such as mustard (Radcliffe and Chapman 1966a). This lack of consistency in results has led some investigators to suggest that overall host acceptability may be determined more by intrinsic plant characteristics such as size, maturity, height, or amount of foliage than by the variety of host (Harrison and Brubaker 1943, Latheef and Irwin 1979a). Both size and age have been found to affect acceptability. Latheef and Irwin (1979a) found a significant positive correlation between plant diameter and the number of eggs received by cabbages. Ives (1978) found that bigger plants (size was measured as total leaf surface area), received more eggs than smaller plants of the same variety. Other measures of plant size such as height and diameter were not associated with acceptability. The influence of plant size on acceptability, however, was valid only for comparisons within a single host variety. For example, kale plants, although bigger than cabbages, received fewer eggs. This finding is in agreement with results from Radcliffe and Chapman (1966a) who, when using different varieties of hosts, found that total leaf surface area was not correlated with acceptability. The effect of age on acceptability is inverse to that of size. Younger plants tend to receive more eggs than do older plants of the same size (Ives 1978, Latheef and Irwin 1979a). 30 Deterrents Plant chemicals functioning as deterrents are believed to play a major role in the host specificity of many insect species (Jermy 1966). The most extensively studied have been those chemicals acting as feeding inhibitors (Munakata 1977). While less attention has been paid to chemicals inhibiting oviposition, some non-host plant extracts, when applied to host foliage, have been found to reduce the number of eggs laid by P; rapae and other Pieris species (Lundgren 1975, Rothschild and Schoonhoven 1977, Rothschild and Fairbairn 1980). Certain plants, mostly aromatic herbs, have traditionally been considered to have the ability to deter insects when planted near crop plants. Several of these so-called companion plants, including tansy, rosemary, tomatoes, sage, nasturtium, catnip, and hyssop, are reported to repel cabbage butterflies (McKillip 1973, Hunter and Hines 1971). Studies investigating these claims have yielded contradictory results. Some report no difference in number of eggs laid on hosts growing in a monoculture and those intercropped with various other species (Latheef and Irwin 1979b, Cranshaw 1984, Root and Kareiva 1984). Others report a significant increase in the number of eggs laid on intercropped hosts (Maguire 1984b, Mathews et a1. 1983, Latheef and Ortiz 1983a, 1983b). This has led to the suggestion that, instead of being repellent, some non-host 31 species are attractive to g; rapae females, presumably via their odor (Latheef and Ortiz 1983b). Interpretation of these results is, however, compli- cated by the confounding factor of patch size. Contrary to the predictions of Root's (1973) resource concentration hypothesis, the number of g; rapae eggs laid per plant increases as patch size decreases (Maguire 1983, 1984a, Root and Kareiva 1984, Cromartie 1975). Intercropping other species with hosts, in effect, results in the creation of many small host patches. Therefore, the increase in the number of eggs laid per plant on intercropped hosts may be due to a reduction in patch size rather than to an alleged attractiveness of non-hosts (Root and Kareiva 1984). Some results, however, cannot be reconciled with this explanation. For example, Latheef and Irwin (1983a) found that more eggs were laid on collard plants surrounded by various herb species than on non-boardered controls. In addition, significantly different numbers of eggs were laid on collard surrounded by different species of herbs. For example, significantly more eggs were laid on collard bordered by wormwood than on collards bordered by santolina. The possibility still exists, therefore, that companion plants do affect the behavior of ovipositing P; rapae. 32 MATERIALS AND METHODS Establishment and Maintenance 9; the Butterfly Culture The P; rapae butterflies used as experimental animals in this study were confined to a 1.8 m x 1.8 m x 1.8 m field cage set up over wild second-year vegetation at the Collins Road Entomology Field Research Center MSU, E. Lansing, MI. The field cage served both to house butterflies and as a confined area in which to conduct experiments. Flowers and potted host plants inside of the cage provided food and ovipositional sites, respectively, for the caged butterflies. Butterflies were placed in the field cage as adults and were left until they died (or, on rare occasions, escaped). To maintain the relatively high density of butterflies necessary for experiments, butterflies were added almost daily. A total of 1,119 adult butterflies was placed in the cage during the course of the study (163 were captured from wild populations, 956 were reared from eggs indoors). An average of nine individuals per day was added. Approximately 76% of the individuals added to the cage were female. Shortly after the cage population was established, I observed that males persistently flew at and attempted to copulate with unreceptive females, thus interfering with their ovipositional behavior. Maintaining a 76:24 male to female ratio reduced male "harassment," while still ensuring fertilization of virgin females. 33 The cage was initially stocked with wild butterflies captured from the local area. On June 2, 3, and 4, 17 wild butterflies from the first (overwintering) generation were captured and placed in the cage. These butterflies laid eggs readily on host plants (reserve hosts) placed in the cage for that purpose. These butterflies were among the last individuals to emerge from the overwintering generation. Shortly after this capture, wild g; rapae could not be located. The initial cage population died out. The egg-laden hosts were removed from the cage and taken indoors for rearing. Wild, second—generation adults were first observed flying approximately June 20. Individuals were captured and used to restock the cage. Like their predecessors, these individuals laid readily on reserve hosts. A total of 147 wild second-generation butterflies was captured and placed in the cage from June 20 to July 14. During this period adults emerged indoors from the eggs laid by butterflies of the overwintering generation and were added to the cage. By July 15, with the emergence of adults from eggs laid by the second-generation adults, a continuous culture had been established. After this date only adults reared from larvae indoors were added to the cage. Butterflies added to the cage at the end of the study were two to four generations removed from their wild, captured ancestors. 34 To maintain the culture, egg-laden reserve hosts were removed from the cage every 3 to 5 days, replaced with clean plants, and taken indoors for rearing. Larvae were reared in two different indoor rearing areas. The first was a controlled environment rearing room. The temperature in the controlled rearing room was maintained at approximately 22 degrees C, with a 16L:8D light cycle. The second rearing area was a room where light and temperature were not controlled. Each group of reserve hosts consisted of six plants, four plants of which were taken to the controlled environment rearing room and two of which were taken to the non-controlled rearing area. Caterpillars were reared in two different environments for two reasons. The first was the threat of disease, which if it were to occur could destroy a culture. If a disease outbreak had occurred, experiments could have been delayed for three to four weeks (the time it would take to establish a new continuous culture). Raising caterpillars in two separate locations minimized the risk of losing the entire culture to disease. The second reason for rearing caterpillars in two different environments was to increase genetic variability in the culture and reduce inbreeding. Eggs laid on reserve hosts which belonged to the same group (i.e. were present in the cage during the same time interval) were the progeny of a single group of parents. There was a greater chance that two butterflies reared from one group of reserve hosts were siblings than for two 35 individuals raised from eggs laid on reserve hosts present in the cage at different times. Because eggs from one group of reserve hosts were all laid during the same three to five day interval, the chances were good that larvae would pupate and emerge at about the same time if they were exposed to the same environmental conditions. Raising the eggs from one group of reserve hosts under different temperatures resulted in increased variation in developmental time. During periods of relatively high external temperatures, caterpillars raised in the non- controlled area pupated two to five days earlier than those raised in the controlled rearing room. Likewise, during periods of cool temperatures, pupation occurred two to seven days later than in the controlled rearing room. The net effect was a greater "mixing" of individuals raised from eggs laid on different groups of reserve hosts. Such a mixing is desirable in that it may minimize the effects of inbreeding common to laboratory cultures. Plants fed to larvae were mostly from reserve host stock grown in the greenhouse. All available species and varieties of hosts (see p 38) were used. NO attempt was made to keep the food plant species or variety constant during the larval period. Thus, an individual caterpillar may have fed on all varieties of reserve host during its larval life. There were two reasons for providing caterpillars with different varieties of food plants. First, although the 36 idea is not well accepted, there is some evidence that the food plant species fed to larvae influences adult oviposition (Yamamoto et a1. 1969). Hovanitz and Chang (1963) developed different ”strains" of g; rapae, that showed a marked "preference" for the host species on which they had been raised as larvae. Whether these strains were produced by unintentional artificial selection or whether larval experience influenced ovipositional "preference" is irrelevant. The fact remains that differences in larval food resulted in strains of butterflies differing in ovipositional "preference." Such artificially induced "strains" could have biased the results of this experiment. A second reason for feeding larvae a variety of host plants concerns variation in developmental time. Chew (1975) found that raising Pieris napi larvae on a variety of host plant species increased the variance of the time to pupation. If the same phenomenon exists in 21.32222: then the degree of "genetic mixing" explained earlier would be increased by feeding larvae different host varieties. Most of the time, whole, rooted, living plants were used to feed larvae. Eggs laid on reserve hosts were allowed to hatch and the larvae allowed to feed until they had consumed a major portion of the reserve host. At this point, larvae were transferred to new plants by cutting pieces of the original plant and placing them directly on the new plants. Young larvae were not handled directly because of potential injury (Stone and Midwinter 1975). 37 Food plants were changed in this manner as often as necessary. When caterpillars were nearly mature (2.5 to 4 cm long) they were transferred to individual pupation chambers. These were constructed as follows: pieces of host plant leaves were cut into small pieces and placed in the bottom or top halves of petri dishes (9 cm). A caterpillar was then lifted off its food plant with forceps and carefully placed on top of the cut leaves. Finally, a 0.24 1 plastic cup was inverted over the leaves and larvae. Caterpillars fed on the leaves on the bottom of the dish until ready to pupate, at which time they climbed the sides of the inverted cup and pupated. After pupation, the original petri dish was replaced by a clean one. Emergence generally occurred from 7 to 14 days after pupation. Pupation chambers were kept in dimly lit areas, because under low-light conditions the adults did not attempt to fly. For transfer of adults to the field cage, pupation chambers containing emerged adults were placed in the bottom of a plant tray which was covered with a second inverted plant tray. This arrangement kept the trays dark and thus prevented adult injury. The Plants With the exception of soybean, all potted plants used .in this study were grown from seed. Standard commerCial potting soil was used as the growing medium. Seeds were 38 sown in flats and allowed to germinate. Following germination, plants were placed under fluorescent lights. After reaching the one to three leaf stage, the seedlings were transplanted into 2.5 cm square plastic “six-pack" containers and transferred to a greenhouse. The temperature, humidity, and light conditions in the greenhouse varied with external weather conditions. No supplemental lighting was provided. Plants were transplanted two more times before being used in the field cage: the first time to 8 cm round plastic pots, the second time to 15 cm round clay pots. Transplanting occurred when the plants were judged to have outgrown their containers. Plants were fertilized with 10-30-10 soluble fertilizer at each transplanting, and every two to three weeks between transplantings. Both host and non-host plants were grown in this manner. Host plants were used as food for caterpillars, to collect eggs for the maintenance of the butterfly culture (reserve hosts), and as experimental plants (target plants). The following varieties of host were used: cabbage (tastie), broccoli (italian green sprouting), mustard (tendergreen), cauliflower (snowball), brussel sprouts (long island improved), and collards (georgia). To avoid large differences in size and age between host plants used at the beginning and end of the study, three separate plantings were made. Seeds were first sown on April 16, and these plants were used as target plants and reserve hosts from 39 June 29 to August 15. Plants were 74 to 120 days old during this interval. The second planting occurred on May 14. These plants were used in the field cage from August 15 to September 6. Plants were 93 to 114 days old during this interval. The third planting was made on June 11. These plants were used in the field cage between September 6 and October 17. The plants were between 87 and 128 days old during this interval. Non-hosts were used as experimental target plants. The non-host species used included tansy, sage, romaine lettuce and soybean. Tansy (Tanacetum vulgare) seeds were sown on March 23, 30 and April 2 and used in the experiments as target plants between July 12 to October 17 (age: 111 to 208 days). It was necessary, especially later in the season, to modify the size of the tansy plants. Initially, the tansy plants were pruned by pinching Off the tops. Later in the season cuttings of the original tansy plants were used, in lieu of whole plants. Sage (Salvia Officinalis) seeds were sown on March 30 and were used in experiments from July 21 to Aug 21. Plants were between 113 to 115 days old during this interval. Romaine Lettuce (Lactuca sativa) was sown on June 11 and was used in experiments from August 14 to September 22. The plants were between 64 and 103 days old during this time. Soybean (Glycine max) plants were obtained from a field of plants sown approximately July 1. Plants were dug out of 40 the ground, placed, along with potting soil, into 15 cm clay pots, and used in experiments within one to three days after transplanting. Soybean plants were used in experiments from August 30 to September 22. Field Cage The field cage was intially set up on June 1, 1984 and remained up for the duration of the study, being taken down on November 4 1984. As mentioned previously, the field cage was set up over a field of wild, second year vegetation. The dominant vegetation in this field included: yellow rocket (Barbarea vulgaris), aster (Aster spp), cinquefoil (Potentilla spp), thistle (Circium spp), and various grasses. Although the vegetation surrounding the cage was mowed once in late July, the wild vegetation inside the cage was, for the most part, left to grow naturally. To provide a more uniform vegetational coverage, white dutch clover (Trifolium repens) and annual rye grass (Lolium persicum) were sown in bare areas of the cage on July 7, 1984. During extended dry periods the wild vegetation inside of the cage was watered, both to prevent dessication of the wild plants and to increase the humidity of the cage, humidity being an important factor in survival of butterflies in outdoor cages (Crane and Fleming 1953). Consequently, the cage interior provided a semi-natural habitat for the butterflies. 41 The percent coverage for each wild plant species inside the cage was determined at the end of the study season (Table 1). The method used to determine percent coverage was a modification of the line-intercept sampling method (Brower and Zar 1977). Eight line transects were established at vegetation height. Two were made by connecting string to diagonal corners of the cage. The remaining six were made by connecting string from each reserve host position to the target plant. The transect length of each plant or clump of the same species was recorded. The sum of the transect lengths for a single plant species divided by the total transect lengths covering vegetation was taken as the percentage coverage for that species. This measure provides an estimate of the percentage of the total vegetation area composed of each species. Six host plants (reserve hosts) were present in the cage amid the wild vegetation at all times. These hosts were kept in holes dug in the ground for that purpose (Figure 3). The positions of the 6 reserve hosts (i.e., the holes dug in the ground) are shown in Figure 3. The position of each reserve host was 12 cm from the side of the cage, 24.5 cm from the nearest cage corner, and 70 cm from the center of the target area (target plant). In addition, all reserve host positions were 59 cm from the neighboring positions (with the exception of positions 1 and 6 which were 93 cm apart). 42 Table 1. Percent coverage of wild plants growing in the field cage. Plant Percent Coverage Grasses and Clovers 53.6 Dandelion (Taraxacum Officinale) 20.5 Cut flowers in bottles 10.6 cinquefoil (Potentilla spp) 7.5 Dock (511933 spp) 3 . 8 Prickly lettuce (Lactuca serriola) 1.2 White cocklebur (Lychnis alba) 1.2 Alfalfa (Medicago sativa) 1.0 Yellow Toadflax (Linaria vulgaris) 0.7 43 F RHSO F Observer Area F Figure 3. Map of the field cage interior. T=target plant, P=plant area, S=soil area, RH=reserve host, F=cut flowers. be he fl 44 Plants used as reserve hosts were randomly selected from greenhouse plants and randomly placed into one of the six positions. The day following their initial placement in the field cage, reserve hosts were removed, the number of eggs laid on each plant was counted, and the plants were placed into new randomly selected positions. Reserve hosts were left in the cage for three to five days before being removed and replaced by clean plants. Three different sets of plants were used as reserve hosts. From June 29 to August 15, the six reserve hosts consisted of one plant each of mustard, broccoli, brussel sprouts, cabbage, cauliflower, and collards. Ten replicates of this set were used. During this time the plants were from 74 to 120 days old. The second set of reserve hosts included one plant each of mustard, turnip, broccoli, brussel sprouts, cauliflower, and collards. Five replicates of this set were used from August 15 to September 6. Plants were from 93 to 114 days old. The third set, which was comprised of two 87 to 127 day old plants each of collards, cauliflower, and broccoli, served as reserve hosts for the remainder of the study period--from September 6 to October 17. In addition to reserve hosts, flowers and artificial nectar sources were placed amid the wild vegetation. Cut flowers were collected from nearby weedy fields and placed in plastic half liter soda pop bottles which were kept in hc th VP. US: mo: art lc flc wi: arc wlt 9:0 fad day "f1. Car. 0086 a)- 45 holes in the ground. Eleven buried bottles were located amid the wild vegetation (Figure 3). These were kept full of water and the flowers changed every three to five days. The flower species used during the study are given in Table 2. Pieris rapae feeds on a great diversity of flower species (Opler and Krizek 1984). The species composition of cut flowers changed during the study, reflecting differences in blooming times. Early in the summer the most commonly used species were alfalfa and red clover. Goldenrod was the most commonly used flower during late summer and early fall. To support the relatively high butterfly population, artifical nectar sources were provided in addition to cut flowers. Cubes of yellow cellulose sponge (approximately 3 cm3) were placed on the ends of green wire (artificial flower stems). To prevent the sponge from slipping down the wire, and to delay dessication, aluminum foil was wrapped around the base of the sponge. The sponge was saturated with a 20% honey water solution and the wires placed in the ground among the wild vegetation. Extra honey water solution was added daily. When the sponges became soiled or faded, they were replaced (approximately every week to ten days). Butterflies were observed to feed readily on these "flowers." The wild vegetation was cleared from the center of the cage to provide an area in which butterfly behavior could be observed and recorded during experiments. This area (target area) consisted of a cylinder Of space 68 cm in diameter and Ta} Wi] Cic Ci; Yel Tre 46 Table 2. Cut flowers used to feed Pieris rapae in the field cage. Common Name Scientific Name Red Clover Alfalfa Bull thistle Sow thistle Goldenrod Aster Wild mustard Cichory cinquefoil Yellowrocket Trefoil Trifolium pratense Medicago sativa Cirsium vulgare Sonchus oleraceus Solidago nemoralis Aster pilosus Brassica spp Cichorium intybus Potentilla recta Barbarea vulgaris Lotus corniculatus 30 cm area 1 point buriec by 6 l grounc I the ta with d OCCUp} termed I CYlind (Plant buried ClOSes Pots t Space Space. 8011 8] string Tc r9590“ (‘5 target during gr0wing 47 30 cm in height (Figure 3). The outer boundary of the target area was 92 cm from the cage corners, 54 cm, at the nearest point, from the sides, and was marked with a partially buried brown string. The height of the cylinder was marked by 6 green wires (artificial flower stems) placed in the ground extending upward 30 cm. A 15 cm clay pot, buried in the center of the floor of the target area, was removed during experiments and replaced with different test species of potted plants. The plant occupying this central position during experiments was termed the target plant. The target area was divided into two concentric cylinders: an outer (soil) space (16 cm wide) and an inner (plant) space (36 cm in diameter). Four 15 cm pots were buried in the ground at the base of the soil space, one pot closest to each Of the four cage corners. The buried flower pots touched both the inner and outer boundaries of the soil space and served only to mark the boundaries of the soil space and the plant space. The inner boundary (between the soil space and the plant space) was also marked by a brown string partially buried in the ground. To prevent butterflies from entering the target area in response to objects other than the target plant, the entire target area was maintained free of vegetation and debris during the experiments. In addition, any plant parts growing into the target area from the surrounding wild vegetation were removed. 48 The area inside the cage in which I sat and recorded behavioral sequences is termed the observer area. The Observer area was located in the middle of the east side of the cage, next to the zipper (Figure 3, p 43). This area was also used to gain access to other parts of the cage, e.g., reserve hosts and flowers. To limit the damage done to wild vegetation, walking was confined to the observer area and the target area. As a result of being repeatedly walked and sat upon, the vegetation in the observer area died. Sequence Description During the interval of June 21 to July 11, preliminary observations of g; rapae behavior in the field cage were made. Because the target area inside the cage was not constructed until June 26, some of these observations were of butterflies flying around the wild vegetation. Particular notice was taken of behaviors exhibited in the vicinity of host plants. ' Table 3 gives a list and description of the behavior patterns identified during the preliminary observation period. Behavior patterns were identified using the generalized criteria of Blurton-Jones (1968), i.e., they were fairly common, readily recognizable, and distinguishable. An individual sequence was defined as consisting of the temporal order of specific behavior patterns performed by an P] M Vi Fl CC La: nu. Com Table 3. Description of Pieris rapae behavior patterns recorded in the target area. Name Given to Behavior Symbol Description Soil s Flying into the soil space. Plant p Flying into the plant space. Air-to-Plant Ap Flying into the plant space directly, by descending from above. Visit V An observable response to the target plant. Flutter f Slowing of flight around an Object. Contact c Making tarsal contact with an object which lasts less than .5 seconds. The wings remain moving. Land 1 Making tarsal contact with an object which lasts longer than .5 seconds. Land-rest 1r A land lasting more than five seconds during which no curling, drumming or wing movement occurs. Curl c Curling the abdomen downward so that the abdomen tip touches the substrate. Ending Behaviors: Flutter-host fh Fluttering around a host. Flutter-vegetation fv Fluttering around a non- host. Flutter-flower ff Fluttering around a flower. Contact-host ch Contacting a host. Soc Air Be}: are Flu C01: Lam Beh are 50 Table 3 cont. Name Given to Behavior Symbol Description Contact-vegetation cv Contacting a non-host. Contact-flower cf Contacting a flower. Land-host lh Landing on a host. Land-vegetation lv Landing on a non-host. Land-flower 1f Landing on a flower. Land-host-oviposit lh-ov Landing on a host and exhibiting behavior associated with oviposition, such as drumming or curling. Social 5 Fly toward and into another butterfly. Air A Fly upward in cage. Behaviors Excluded from Hypothetical Sequence because they are Irrelevent to Oviposition: Flutter-soil fs Fluttering above bare ground. Contact-soil cs Contacting bare ground. Land-soil ls Landing on bare ground. Behaviors Excluded from Hypothetical Sequence because they are Difficult to Observe Objectively: Drumming dr Scraping the foretarsi across the plant with a rapid, alternating motion. Wings w Extended wing movement after landing. Behaviors Excluded from the Hypothetical Sequence Because they are Rare During Oviposition: Proboscis Extension prob Extending the proboscis. inc‘ du: rec the (51 CO: pe: th. re; See 0V Se. 0b. an Se: 51 individual butterfly while in the target area. Because the duration of the individual behaviors was shorter than the recording capabilities of any timing device available to me, the recorded sequences lack a time base. A hypothetical sequence was constructed for analysis (see Figure 4). This consisted of a series of steps, each corresponding to an individual behavior pattern. The hypothetical sequence included all possible transitions from one type of behavior to another (certain transitions were forbidden by definition). Most of the data analysis involved comparing the transitional frequencies from a given behavior to all possible succeeding behaviors for sequences performed in response to different plant species. Several identified behavior patterns were excluded from the hypothetical sequence (these are listed, along with the reason for their exclusion, at the end of Table 3). Three types of behaviors were eliminated from the hypothetical sequence: 1) behaviors judged to be irrelevant to oviposition, 2) behaviors not equally observable during all sequences; the position of the butterfly relative to the observer determined whether the behavior could be observed, and 3) behaviors rare during the ovipositional sequence. The following is an explanation of the hypothetical sequence, including a list of all possible transitions, and a description of the behaviors involved. The hypothetical sequence is presented in Figure 4. 52 Target Plant V / , / , .. l / C faend \ K i Caend OJ, «\ leend \ 80“ Plant Space Space Target Plant Figure 4A. Spatial representation Of hypothetical sequence of behaviors of Pieris rapae in target area. S=SOII space, p=plant Space, A=enter target area from above, V=vnsnt, =flutter, l=land, c=contact, cu=curl, end=leave target area. Flgu 53 | Fly into 55 I Flutter lib") Figure 4B. Diagramatical representation of hypothetical sequence of behavior Of Pieris rapae in target area. S= soil space, p=plant space, A=enter target area from above, V=visit, f=flutter, l=land, c=contact, cu=curl, end=leave target area. ' n) t1". en ti: WO‘ ta: be} 54 Observation of a sequence began when a butterfly flew into the target area and ended when the butterfly either left the target area or curled (see below). Since the target area was divided into a soil space (s) and a plant space (p), butterflies could begin a sequence by flying more or less horizontally from the surrounding vegetation into the soil space (s) or descending more or less vertically from above (A) into the plant space directly (Ap). Sequences initiated by butterflies which flew downwards and cut across the outer edge of the soil space while flying toward the wild vegetation were not recorded. Once a butterfly entered the target area, it could end a sequence by flying out of the target area or by curling (see below). Flying out of the target area is designated as "end" and could occur at any step in the sequence. The following is an explanation of the sequence steps, and the transitions between steps, that were possible after sequence initiation: 1. Soil space t2 plant space (s-->p). A butterfly flying into the soil space would ultimately either fly into the plant space (s-->p) or leave the target area (s-->end). 2. Plant space 39 Visit (p-->V). Once a butterfly had entered the plant space (p), either by flying into it from the soil space (s-->p) or by entering it directly (Ap), it would either leave the target area (p-->end) or visit the target plant (p-->V). A "visit" was a category of behaviors, not a single act. Included in the "visit" cc la re SE 0C du la; €01 55 category were the behaviors "flutter," "contact," and "land.” These behaviors were lumped into one category because they constituted an observable response to the target plant. Behaviors included in the visit category are described below. 3. Flutter (f): This behavior consisted of an Observed slowing of flight, and/or stopping in the vicinity of the target plant, i.e., the butterfly appeared to hover around the plant. A flutter could last between approx- imately .5 and 10 seconds. No tarsal contact occurred during fluttering, although butterflies sometimes appeared to make contact with the plant with their wings and/or antennae. Fluttering butterflies were often observed to fly into the underside of target plant leaves. 4. Contact (c): The butterfly made brief tarsal contact with the plant (usually with the leaves). Contact lasted for less than .5 seconds. A butterfly's wings remained moving during a contact. 5. Land (1): The butterfly landed on the plant surface (usually the leaves). Tarsal contact lasted for more than .5 seconds. Typically, some wing movement occurred after landing, but it usually did not last for the duration of the land. The first behavior performed during a visit could be a land, a contact, or a flutter. Thereafter, a butterfly could end the visit by leaving the target area, or continue the visit by performing additional behaviors. For bu de; tr the It wa: 00! dog Suv- fi tav- \ 00::- was lint. 10m. aCCI 56 butterflies continuing the visit, the type of act performed depended, to some extent, on the preceding act. Some transitions were forbidden by definition. Possible transitions included: 1) a fluttering butterfly could leave the target area (f-->end), land (f-->l), or contact (f-->c). It could not immediately repeat the flutter since a flutter was defined as continuing until another behavior was performed, 2) contacting butterflies could leave the target area (c-->end), flutter (c-->f), land (c-->l), or repeat the contact(c-->c), and 3) Landing butterflies could leave the target area (l-->end), flutter (l-->f), contact (l-->c), or re-land (1-->l). In addition, landing butterflies could also rest (1r) or curl (lcu). A rest (lr) was defined as a land on a plant which lasted longer than five seconds. The butterfly performed no obvious movement such as wing fluttering or drumming during a rest. During a curl (cu) a butterfly curled her abdomen downward so that the ovipositor tip pointed toward the leaf surface. After a butterfly had curled, it could either leave the target area or continue the visit (by performing a land, contact, or flutter). When recording sequences, an attempt was made to record the behavior of individual butterflies until they left the target area. However, during especially long sequences involving numerous curls, it was difficult to accurately record data. Therefore, a sequence was con: sax; spe sari Ban er, rep 57 considered complete when a butterfly performed the first curl. Sampling Procedure A day during which data were recorded is termed a sampling date. On each sampling date, behavioral sequences in response to from three to six different target plant species were recorded. The period of time during which the sequences toward one target plant were recorded is termed a sampling interval. No more than one sampling interval was performed for any target plant species on any one sampling date. The procedure for recording data was as follows: the empty pot in the center of the target area was removed and replaced with a target plant randomly selected from the species to be sampled on that day. I sat in the observer area (Figure 3, p 43) of the cage and recorded a pre- determined number Of sequences, as outlined below. After the required number of sequence Observations had been taken, the target plant was removed and taken out of the cage. If the plant had received eggs during the sampling interval, these were counted, recorded, and gently brushed off of the plant. After at least three minutes, another target plant was placed in the target area. This procedure was repeated until observations had been made on all the target plant species to be tested on that particular sampling date. ta St W; 58 In addition to recording sequences and eggs, the times of day (at both the beginning and end of the sampling interval) and the total length of the sampling interval were recorded. Data were replicated by repeating sampling intervals on different days. The sampling procedure was a combination of focal animal and sequence sampling (Lehner 1979). All behaviors performed by an individual butterfly between the initiation and termination of the sequence were recorded in the order of performance. In addition, any behaviors performed immediately after an individual left the target area were also recorded. While Observation and recording of an individual sequence was occurring, all other individuals entering the target area were ignored. Sampling of sequences was randomized in the sense that the first individual to enter the target area after the previously sampled individual had left became the next subject. An equal number of sequences was recorded for each target plant used on the same sampling date. The number of sequences sampled per target plant varied from day to day, mostly ranging from 30 to 39 sequences per day. This number was decided on prior to initiation of sampling. The criteria for deciding the number of sequences to be sampled were: the number of target plants to be used, the weather conditions, and an intuitive assessment of the responsive- ness of the butterflies. seqae influ inter eliml 5 gas saapi thes: reco: sequ eqaa com; Dre: 0f . trp aCc 4) Th. ac re 59 In cases where the behavior of a butterfly during a sequence was ambiguous or where other factors appeared to influence its behavior (for example, a male seeming to interfere with a female's behavior) it was necessary to eliminate recorded sequences from the data. When problem sequences were identified while recording data, additional samples were taken to compensate. Occasionally, however, these problem sequences were identified only after data recording was complete. As a result, the total number of sequences taken on each plant in a comparison is not always equal (less than 1% of the sequences fOr any plant in any comparison were eliminated). Experiments A major objective of this study was to establish the presence or absence of discrimination occurring at each step of the ovipositional sequence between the following plant types: 1) hosts vs non-hosts, 2) hosts of different acceptabilities, 3) different species of non-hosts, and 4) repellent and non-repellent plants. To that end, the following plants were selected as target plants: 1) Broccoli. As a reserve host, broccoli received a greater mean number of eggs than other host species or variety. Therefore, as a target plant, broccoli was used as a highly acceptable host. 2) Mustard. As a reserve host, mustard received a lower mean number of eggs than broccoli. In addition, mustard, being a different species, is morphc was us in out repute studie ovipos 1983b] tansy larvae rapeli “repei (and 1 these Plant Dotti a non Addit host repel Soybe 60 morphologically different from broccoli. Therefore, mustard was used both as a less acceptable host and a host differing in outward appearance from broccoli. 3) Tangy. Tansy is a reputed "repellent" plant which has been used in several studies on the effects of companion plants on g; rapae oviposition (Mathews et al. 1984, Latheef and Ortiz 1983a, 1983b). In addition, Brewer and Ball (1981) found that tansy extract acts as a feeding deterrent to g; rapae larvae. Therefore, tansy was used in this study as a repellent plant. 4) Sage. Sage is another purported "repellent" plant (McKillip 1973) which is morphologically (and presumably, chemically) quite different from tansy. In these experiments it was used as as a second "repellent" plant species. 5) 293. A 15 cm clay pot filled with potting soil was used as a control. 6) Lettuce. Lettuce is a non-host species not reported to be repellent to g; rapae. Additionally, lettuce is, to humans, visually similar to the host species, mustard. Lettuce was used here as a non- repellent, "host-like" non-host plant. 7) Soybean. Soybean is also a non-host species not reported to be repellent to g; rapae. Unlike lettuce, it is not visually similar to hosts. Therefore, in this study soybean was used (as a non-repellent, non-host unlike a host in appearance. 8) Host-Repellent Combination: Broccoli-tansy . This was a smatted broccoli plant with a tansy cutting placed in the Scail next to it. Broccoli-tansy presumably possessed both hcast and repellent properties, and was used in the study to ESE COI sat pl us Da $8 61 assess the butterflies' responses toward a host-repellent combination. In all, five experiments were done, each involving the same proceedure but using a different combination of target plants. The particular combination of target plant species used in each experiment is referred to as a comparison. Data were taken on the plants within a comparison on the same set of sampling dates. The reason for conducting five separate experiments, rather than using all the target plants in a single experiment is as follows: First, there was wide day-to-day variation in the behavior of the butterflies, mostly in their level of activity and in their responsiveness toward target plants. Therefore, data for different target plant species could only be compared if it were taken on the same dates. Second, the activity and responsivness of the butterflies also varied during the course of a single day. Butterflies were usually most active between the hours of 11:00 and 15:00. Outside of this period, there was minimal activity in the cage. The length of the period of maximum responsiveness varied from day to day and changed over the course of the study. To minimize the possible effects Of this natural "waxing and waning" of response on between treatment comparisons, only 3 to 6 different target plants could be used each day. The plant comparisons are designated as Comparison I - Comparison V, and are given, along with the number of sampi Tabll a no: obse: The l ovip: corp; poin‘ diff: the 4 disc] 0Vip< whick diVic‘ 62 sampling intervals and the purpose of each comparison, in Table 4. Methods 2; Analysis The first objective of this study, that of determining the sequence of behaviors leading to g; rapae oviposition on a normal host, was accomplished during preliminary observations and has been described previously (see p 48). The second objective of determining where in the ovipositional sequence discrimination occurs, involved comparing the behavior exhibited by butterflies at the same point of ovipositional sequences performed in response to different plant types. If butterflies, at a given point in the ovipositional sequence, behaved significantly differently toward different plant types, then discrim- ination was said to exist at that point. For example, if the proportions of butterflies curling and leaving the target area (ending) after having landed on the target plant differed significantly between hosts and non-hosts, then discrimination between hosts and non-hosts was said to occur during the land. An implicit assumption of this study was that discrimination could be shown at several points in the ovipositional sequence. The points in the sequence during which discrimination was to be assessed were determined by dividing the total sequence, first into a series of con- secutive stages, and then dividing each stage into a series oIUCUEHDOQXO m 058 CM UUIQQOII IE3 COMUICwEuHUUfiU £0M£3 C003U0£ EUQKU UCQNQ UEFIIICOEflLEQEOU UCENK «? UNOQF 63 .AucOu mo bummmo osu Ou vousnmwu run on canoe mecca hunch Imaoooown vocwbaoo IMHOOooum onu can maoooopn .hmcoy cooauon noocowommmn x x «H .maooooum > coonhom oosuuoq x x as sauna >H ensues: .oosuqu .hmcce x x x x ma «Hooooum HHH .ucmaa «nu ob uncommon cw voEMOMHOO meow>onon mo newuucmEuouoc you :uom: cmaoaao Houucoo u owum .hucoa no :uoa: Mo on: ugh x x x ma .maooooun HH hocma x Ne .msouuoum H mazmzxoo mez<4m Hmozlzoz meHHAHm< mymomlzoz mma memo: ozHAm= quaqummm HZMfiMhmHa azummmmun ho mumzaz waz<4m azOZ< ~Q>h~a£~ WC-QEET £62 Ufl>m0001 tififi L0 LUEEZZ 2332 N 4¥>E4 \ 65 .Avcmfiou can .Huso ocv uneducoo I ucoo .oooam acm~d u a .oooam amen I m .o>ono Boom noun uowuou ocuco I d< .Luso I so .uobus~u I u .uoaucoo I o .vcw~ I H .aouo uoquu o>mo~ I ecu .umam> I > .oocosaom I cum co>mw om cow: boob acumuomuuum .AoocoMum> mo mmwmaocc I <>oz< .uaou o I UV xon mo uocuoo ucwmu peso“ em .BOOB oco mucououaou Ho>o~ a mo cOmmm>mcnsu zoom .A>H cwsouzu Hy u~o>oq o>mmmooosm co coumfioco am when .muasuou orb mo some gamma em amax~ucu ecu mo c0maoumcowuo .n ouswmm commune o> «Looooub Loco m ocuum memcco mOMHLLOBBsn mo cOMBuoaoum ”mic >H penance Am>w4 can .zmcou .MHOOOOMB HOL mucosvom 5mmm> 0:5 mo cOmuamuomoa “~13 o u o <>oz< ammo: cco s new a can a muomucoo Baum“ m> mccofl ob cCOdmou new o>moouoa amm~w m> uCOo * m> m> Baum“ mo comuuoaoum "mrmm tumuusb cums3 um oocmumma “mi ucoo * BCOO * m> vco e m> m> U U o > a we a> > a< e + e + s + e zuuco Lo cosmos m> m0w~pruusn memummm> mo c0muuoaoum "¢I Icoz m> m> o o 0 20000.; Iooum “sew w a» o a» com a s can H: wcmcco can ~§ m> \Vvo % m> a4 % .MV 6 a Am>mq loamucoo .waHusu mOMHmuoBBSB w § > a Sony a mamucwmfim wo acmuuoaoum "NIUM ammuusfim cOMOMmcmua .< Sosa woman 0 umumm umow> cobumumcm ucu~a ob memumou Doc mcco umumw e m> mamaofifiow ummm> a Ca ob woman moocosvoo woman ~mOn mamumou mono umumm a u0m>ocom you umumm Osman mo cOmuuoa Boom acme "asset Lo couauoaoua ”1-0m “Nina "Liam "m-oz “Lue V73...» 3 + or— > Au>ua acmummcoue camumocoua coflummcmua Huzo Ou uswm~< "UN ucwm~< cu uncommom “mm uncommom Ou ucQEo>oz u<~ «x m.coEp6mum mumooem ucufim bounce coco uOu a <>oz< ~o>u0ucH mcm~a5om hum vo>mooom mwwm mo nonasz coo: Am>mq ovipl stag ente This ovip visi the beha were alig butt ov1; Cons was beer inc] butt the 66 These stages appear to be natural subdivisions of the ovipositional sequence. Prior to the response (visit) stage, an Observer could not be sure if butterflies were entering the target area in response to the target plant. This stage necessarily included many sequences irrelevant to oviposition. Once a butterfly exhibited a behavior in the visit category, an observer could be fairly confident that the butterfly was responding to the target plant. The behaviors in a visit, up to and including the first alight, were elicited by non-contact plant stimuli. After the first alight, contact stimuli also potentially also influenced the butterflies' subsequent response. The abdominal curl was correlated highly enough with oviposition to be considered synonymous with the consummatory act of the ovipositional sequence. However, it was not possible to record consistently whether an egg had been deposited after curling, so egg deposition was not included as a behavior in the sequence. After curling, butterflies would either leave the target area or continue the visit. In order to lay an egg, a butterfly had to progress from one stage to the next. However, a butterfly could also terminate the sequence after reaching any stage by leaving the target area (end). The analysis of data in this section involved comparing the proportion of butterflies at a given stage that advanced to the next stage, vs the proportion that ended the sequence, for sequences performed in response 67 to different plant types. The purpose of the analysis of Level II was to establish and compare the amount of discrimination occurring between different plant types at each sequence stage. The stages identified in Level II, above, were then subdivided into their behavioral components. Each discrete behavior performed during the sequence is referred to as a step. Most of the analysis of data at this level (Level III) involved comparing the proportions of butterflies, that had performed a given behavior, that subsequently performed the possible following behaviors, for sequences performed in response to different plant types. The purpose of the analysis of Level III was to determine exactly which of the behaviors in the sequence involved discrimination. Finally, the visit sequences to 3 different target plant species--broccoli, mustard and tansy--were analyzed as a Markov chain in Level IV. A summary of the tests performed on each of the 4 levels is shown in Figure 5. It was hypothesized that discrimination would occur at the various sequence stages and steps because the nature of the plant stimuli perceived by the butterflies during any given stage or step would affect butterfly response. Specifically, it was hypothesized that: 1. Host plants possess host-specific stimuli that cause butterflies to respond to the plant. Therefore, it was predicted that a higher proportion of butterflies at any given stage or step of the ovipositional sequence would 68 respond to the plant (and thus continue the sequence) for sequences performed in response to hosts than for sequences performed in response to non-hosts. 2. Host plants differ quantitatively and/or quali- tatively in host-specific stimuli perceived by butterflies. These differences in stimuli result in differences in the degree of response exhibited by butterflies which, in turn, result in differences in acceptability. Therefore, it was predicted that a higher proportion of butterflies at any given step or stage in the ovipositional sequence would respond to a host Of higher acceptability (broccoli) than to a host of lesser acceptability (mustard). 3. General plant stimuli (i.e., stimuli that are not host-specific) affect the responses of butterflies. Lettuce is somewhat visually similar (to humans) to the host mustard. Visually perceived aspects of plants influence the degree of response exhibited toward plants. Therefore, it was predicted that the proportion of butterflies at any given step or stage in the sequence responding to the plant would be higher for sequences performed in response to lettuce than for sequences performed in response to other non-host species. 4. Tansy and sage generate repellent plant stimuli that butterflies are able to perceive and which deter response to the plant. Therefore, it was predicted that a lower proportion of butterflies at any given step or stage of the sequence would respond to tansy than to other 69 non-host species. In addition, it was predicted that, as the result Of the presence of tansy, a lower proportion of butterflies would respond to broccoli-tansy than to broccoli alone. I Hypotheses and predictions concerning transitions between specific stages and steps are given in Table 5. The final Objective of this study was to compare the relative degree of discrimination occurring: 1) between hosts and non-hosts during different steps and stages of the ovipositional sequence, and 2) among different plant types during the same sequence step or stage. Assessing the relative degree of discrimination between hosts and non- hosts at different stages and steps was accomplished by calculating and comparing the normalized transmission for each stage and step of the ovipositional sequence. To determine the relative degree of discrimination occurring among different plant types during the same sequence stage or step, an index, the percent relative discrimination, was developed. The percent relative discrimination was calculated and compared for: hosts vs non-hosts, hosts of different acceptabilities, and different species of non- hosts, for each of the stages and steps in the ovipositonal sequence. {IL-nckc C0fiUICHE¢hUId1 00503700 #ECCWUNICQW>0 USU Cw Uhfi£3 NCwaEhUUUC Cw UQQQUU ECOmUUNUULQ 3C8 QEQQtUCQKt an 0N£§H 70 .mwooooua coca hmcmulmaoooowm .m .oomoooo uooalcoc nozuo cosu own» one Anew? .n Adena uomuwu orb uuoousoo no so mama Ham: o>ooH no ucoHo nowauuouusp mcmumom> mo coauuonouo woaofi < osu uoobcoo no .uomooen uuosisoc uoauo sang ousuuod .o co coca Hafiz .kuuosa coca «Hooooum .m .wc«u£wwao up vcoeoou Ocean uowuou .nuoo:1:os song mono: .< Hams hosu wonuopa ocm5houov orb wowumom> «couscoo no no page flaw: woman HHMS oowawuouusn wcmuwom> oomamuouusmv luouusa wcmuwom> mo acmuuonoun karma; < up co>moouoe «Hoamuo ucoam OCO\oI~AII > mu .maooooun canu hocoulmaooooun .u .oomooea uooslcos uosuo coco omen can %oco9 .n Amouo uomuou “umow> flaws noun uowuou ozu osu o>oo~ no memu0u=o oomfimuouusn mo cowuuoaoue wosofi < booed bowuou onu .cuouosa euro «Hooooum .o .ucoae any umom> flaw: umom> Haws .oomooeo uoonlcoc uosuo coca oosuuog .m hose uosuoca ocmahouoc HHwa mono uomuou .ouoosicos coca wonuuu «boom .< ucoao 6 Boom noocouomv uuoco any wsmwOOCO "umom> Haws mono bowuuu ecu wcmuOBCO one hora oamg3 mowawuouusn oomamuouuomv nemamuouusb mo acmuuooouc uoswm; < up co>moouon masamuo ocean c:0\>ATI com Hmmmoobm zmmshmm monHHmzmu .uusooo c0muusm8muuomv oocosvon HucOmumuoom>o osu om opus? wcmcmshouov cm counou o:0muomvoua use monocuonhm «n Ounce IUCQU W ”Nan-NIH 71 .maooooua coco Aocwulmaooooum .M .oomoono uoosicoc uosuo coco omen veg honey .9 ”ob uncommon cm ocean Ocean onu ousw haw haucosvounSu Hams ounce Hmou ozu Ousm mswham nowamuuuusn mo c0muuoeoun woaoa < .oomooao uoo;1coc uosuo cusu oosuuon .o .cuouosfi can» «Hoooowm .m .ouooslcoc coca mono: .4 no» uncommon cw oooco Osman Ono oucm ham hHuOOSVOuAsO Hflms woman Hmoo onu Ousw wcmhfim oomHmuouusb mo camuuomoua honwms < .ouoObisoc mo oomooeo econommmc Macao meson on Ham: vofiuso unnu OOMHMHOOBsn wsmucwMHo mo comuuoeoua onu cm oocouommmc oz .wcwxomoueco hp vsoeoou ~Hm3 oomamuouusn uoxuogz coco roams“ nucoee Boom noocouomc boos» comma an own newawwouusn dams} co>moowoe .musamuo ocean Amouo uowwou onu o>uo~ no momma Hmoo orb wcmwobco nowamuouusmv eco\a..u a HuHmmmoubm ZNMBHMQ monHHmzmA noose uomuou .oococoou Huso onu MO was o>oo~ .vwouosa coca «Hooooum .m owooooHow onu one wcmucwmao no Huso Hams .ouoonicoc coca mono: .¢ woumo osmamuouusp an co>moo oomawuouusn "so Huso Hams nomamuouusn upon Aofioomsoso uOOuCOOV wcmucwmadv wcmumam> mo sowuuoaowe nonwm: < masamuo ommmoocoruoom vOO\=oal| Ora UN monHoHaumm mummmaomwm onsHmz cm can uonwm neocu no nouucou no nouusam Hana oomaunouusb we coaunonona nonwws < .oOMOOOo uoonlcoc nosuo cosu oosuuoq .o .cnouosa cozy “Hoooonm .m .ouooslsoc coco oboe: .< «on ouwon> an uouusoo no nouusau Haw: samunocone noaoa hawcwvcoeoonnoo u can umom> a on you uunnu neonu on come Ham: nomaunouusb mo acmunonone nocmms < .waoooonb coca nocoulnaoooonn .w .o0mooeu noonlcos nonuo coca omen can hoses .n “unomn Hens ouueo ocean Ono cw monamn0uu=n mo sowunoaone nosoa < .oomooco uaonleoc nonuo can» uosuuoa .o .vnouosa coco «Hoooonm .n .uuoonicoc coca manor .< «umom> Haws cocoa usuae on» :« nowaunOuusp mo sowunoeona nonwmn 4 .umo«> m cm uou uanmm man oocouscm Hana .ucoae u Mansowoneeo one honu cams} nomamnouusn up co>moonoo .masamuu ucuflm .usoan ecu umom> new: oomamnouusa nonuopa oocosHmcm saw: Masamuo usuam humom> a on you uonwm nmozu on coca no .uoousoo .nouusam new: . usage a mewumom> nowamnouusmv :0: Ant > 7mm Amonu uounoO onu o>oo~ no usuae osu umow> afims ounce usuae as» wcmnouco nowamnouusmv 98?... a «rem mZOHHoHnumm mHmmeOmwm zonnnmzeen zoneomu /'| .ocoo n «seen 73 .ouoO£|soc mo oomooeo uconommwv Macao venom on Ham: .vco no .oscmu Icoo .anso .uzwmao co>mm new so .uocu nowaw Inouusn mo somunooone o3» cm oosonouwmc oz .hocoulmaoooonp song «Hoooonm .0 .vnouosa sonu “Hoooonm .m .ouoonico: corn ouuom .4 «so ucwmao co>mm new noumo oscmusoo no anso Ham? oomaunouusn mo coMunooona nonmms < .maoooonp coca honouIMHoooonm .M .oomooeo uuoslsoc nonuo conu owoo coo nosoe .o ”no uswwao haucoavoonso Ham: umonn o On you gonna nwosu no nouusfiu unnu oomfiunouusn mo =0munoeonm nosoa < .oowooeo nooslcoc nonuo coca oosuuoq .o .cnouosa cosy «Hoooonm .m .ouoosicoc coca ouoom .< "co Auooucoo no vcoHV u£MMHo hausosuoonso Hama unow> o am now uonmm nmosu on nouusfim guru oomamnouuan mo coMunoeone nonwm; < .unom> onu ouosmanou Ou hamnouusn wcmuswmao so ooOsECM «Heamuo nooucoo omwmooaoluoos mo oocoono one .oocoeoon anso onu mo onooooHon nnoamna onu ono mcmunwwfio noumo co>woonoe wfisawuo nooucoo ommmooeoluoom .ucoHe o co u:MMHo Hams hHm Inouusb o nosuoga oocosausm Ham: .ooMHmnouusA wcmnouuofiu hp coemoonon .masamuo ocean Amono uownou ozu o>ooH no .ouOOucoo no ovcoa naoononeeo menses no unan> orb oosmucoo use anso no: .anso noSumo Hons arouse co>mw new noumo .oonamnouusmv coo \.ucOO\=OAII org Hlom Amono uownou onu o>oo~ no .uooucoo .ccoH Hews umom> o on non nonmm nmosu no womnouusflm nonfiwnouusmv oeo\O\HArr m «Inn monhoHnumm mHmmmhom>m car: was: 28.8% /I .ucoo n «noun 74 RESULTS AND DISCUSSION Establishing where Discrimination Occurs Level I: Acceptability of Plants for Oviposition Results: Broccoli Eggs laid on broccoli were counted in 37 of the 42 sampling dates during which broccoli served as the target plant. A total of 364 eggs was laid. The mean number of eggs laid per sampling interval was 9.73 (+/- 1.59 S.E.). The distribution of the number of eggs laid on broccoli per sampling interval is shown in Figure 6. This distribution fits a negative binomial distribution with a k of .836 (Chi-Square = .326, df=1). Since this is a distribution of eggs laid per sampling interval, it represents a distribution over time. A negative binomial distribution has also been shown to fit spatial distributions of the number of 2;.EEEEE eggs laid per plant in the wild (Harcourt 1961) and in a field cage (Kobayashi 1965, 1966). Bliss (1958) points out that there are two reasons for obtaining a contagious distribution such as a negative binomial: contagious interactions between individuals, or unequal exposure of the units to infestation (most likely due to environmental heterogeneity). Over time, many factors may lead to heterogeneity: weather, the number of butterflies in the cage, the age and sex composition of the population, changes in the wild 75 _ _ _ _ _ _ 6 5 4 3 2 l mum-u mo mum—.5: cubs-mama uz_>_muu~_ «#24: hug—E. no mum—.5: ache. EGGS LAID PER SAMPLING INTERVAL Figure 6. Distribution of the number of eggs laid per sampling interval on the broccoli target plant (comparison I). 76 vegetation in the cage over time, etc. Of these, weather, age (Gossard and Jones 1977), and parental density (Kobayashi 1965) are known to affect egg laying behavior and/or the spatial distribution of eggs by E; rapae. Kobayashi (1966) found that a negative binomial spatial distribution of eggs on cabbage plants in a field cage was produced by a random distribution of butterflies visiting plants compounded by a logarithmic distribution of the number of eggs laid per visit. Considering this, it seems likely that the negative binomial temporal distribution of eggs was produced by heterogeneity factors rather than an interaction of butterflies visiting plants, such as a visiting butterfly inducing other butterflies to visit. Still, it intuitively appeared to this Observer that butterflies might be inducing each other to visit plants. To check for this possibility, a runs test (Schefler 1980) was performed using the broccoli data. Separate runs tests were performed for each sampling interval. Each butterfly entering the plant space was classified as (V) visiting or (N) not visiting the target plant. The runs test checks whether the number of runs obtained (a run, in this example, is defined as a list of consecutive V's or N's) is significantly different from that expected due to random occurrences of independent events. If the visits were contagiously distributed (i.e., butterflies inducing each other to visit) significantly fewer runs than expected would result. 77 In none of the sampling intervals were the number of runs significantly different from that expected due to random fluctuations. Thus, the idea that butterflies are inducing each other to visit is not supported. Results: Comparisons The mean number of eggs per sampling interval received by each target plant species is shown in Table 6. It is evident that P; rapae exhibits a great deal of discrimination between hosts and non-hosts. None of the non-host species received eggs. Non-hosts are clearly unacceptable for ovipositing g; rapae. In comparison III, the host broccoli received a mean of 11.91 eggs per sampling interval (+/- S.E.=2.64) while mustard received a mean of 5.21 eggs per sampling interval (+/-S.E.=1.18). These means are significantly different (F=6.01, P=.0247, df=1). Therefore, broccoli is more acceptable as a target plant than mustard and some degree of discrimination occurs during the ovipositional sequence between these two host species. In comparison V the broccoli-tansy combination target received a mean of 11.91 eggs per sampling interval (+/- S.E.=3.60) while the broccoli target received a mean of 12.07 eggs per sampling interval (+/- S.E.=2.28). Performance of a Friedman's Chi-Square test for randomized blocks resulted in a Chi-Square value of 1.23 which is not significantly different at the .05 level (df=l, P=.267). Thus, from these data, the broccoli-tansy combination target 78 Table 6. Mean number of eggs received per sampling interval by each plant species in the five comparisons. Comparison Plant Species Mean Number of Eggs I Broccoli Tansy OOH COO ONO“) COOP ooo II Broccoli Tansy Pot Sage III Broccoli Tansy Mustard Lettuce H IV Tansy Lettuce Soybean V Broccoli ' Tansy Broccoli-Tansy 3.; HON OOO OUIOH 000‘! om p 1. Data significantly different (P=.025) by a 2-way ANOVA (randomized block design with days as blocks). The log (x + 1) transform was used to stabilize the variance (Harcourt 1961). 2. Data not significantly different (P=.267) by a Friedman's Chi-Square test for randomized blocks (Sokal and Rholf 1969) with days as blocks. 79 does not appear to be less acceptable to ovipositing 2; rapae than the broccoli target alone. Level II: Transitions between Successive Stages of Oviposition As discussed previously, on Level II the total ovipositional sequence was divided into four consecutive stages: movement, response, alight, and curl. In order to lay an egg a butterfly necessarily had to advance through each of these stages. However, a butterfly could also end the sequence at any stage by leaving the target area (end). Analysis of data on this level consisted of comparing the proportion of butterflies at each stage that advanced to the next stage vs the proportion that ended the sequence by leaving the target area, for sequences performed in response to different plant species. The number of butterflies entering each stage of the ovipositional sequence and the transitional probabilities (the probability of entering the next stage given that the butterfly is in the previous stage) for the plants in each comparison are given in Appendix B. For the movement to response transition (Seq-->V), a small proportion of butterflies flying into the target area visited the plant (Table 7). Percentages ranged from 25.5% (broccoli, comparison V) to 2.6% (pot, comparison II). Thus, it can be assumed that most butterflies entering the target area did so for reasons other than response to the target plant. 80 Table 7. Number of butterflies entering the target area ending (seq-->end) and visiting seq-->V) for different plant species. Plant Number Number Percent Comparison Species Ending Visiting Visiting I Broccoli 1137 260 18.6 a Tansy 1345 49 3.5 b ***G=175.56, df=1, P <.001 II Broccoli 518 92 15.1 a Tansy 582 30 4.9 b Pot 594 16 2.6 b Sage 577 34 5.6 b ***G=76.93, df=3, P <.001 III Broccoli 505 118 18.9 a Tansy 608 13 2.1 c Mustard 544 79 12.7 c Lettuce 599 23 3.7 c ***G=145.59, df=3, P <.001 IV Tansy 324 7 2.1 a Lettuce 318 15 4.5 a Soybean 330 9 2.3 a G=3.28, df=2, P=.194 V Broccoli 345 118 25.5 a Tansy 447 13 2.8 b Broccoli-Tansy 350 110 23.9 a ***c=129.45, df=2, P <.001 a,b,c, a Data were analyzed by a G-test (Sokal and Rholf 1969). Percentages followed by the same letter are not significantly different at the .05 level by an STP test. * = P§.05 ** = P§.01 *** = P5.001 81 There are several reasons for butterflies to fly into the target area without visiting the plant: 1) a butterfly may respond to an object outside of the target area (i.e., a flower or another butterfly), and fly through the target area in approaching it, 2) a butterfly may not respond to any particular object, but fly through the target area as a result of random (non-directed) movement, 3) a butterfly may respond to the target plant by approaching, but not make an observable response, or make a response so brief and so subtle that the observer fails to notice. Because many butterflies presumably entered the target area for reasons other than oviposition, it is a slight misnomer to state that all butterflies entering the target area are in the "movement" stage of oviposition. However, it was not possible to determine the motivation of butterflies prior to the response stage. The "pot" target in comparison II received fewer visits than any plant species in that comparison. The proportion of butterflies visiting the pot was significantly different from the proportion of butterflies visiting the broccoli, tansy, and sage targets. This result supports the assumption presented earlier: that visit behaviors represent a behavioral response to the plant. In addition, the number of butterflies visiting the pot target is overinflated due to the wish to be conservative. Butterflies often "fluttered" above the bare soil in the target area and occasionally landed on the soil Of the pot target. When an 82 actual plant served as the target it was relatively easy to distinguish these behaviors from the behaviors involved in a visit response (e.g., those directed toward the plant), and they were not recorded as a "visit". However, when the target was the pot, the plant itself was lacking as a point of reference. To correct for any bias toward Obtaining expected results, I was intentionally biased in the opposite direction: All flutters above the soil in the plant space and all landings on the soil in the pot were recorded as visits. Thus, the actual unbiased number of visits to the pot target was most likely considerably lower. The proportion of butterflies visiting hosts was significantly higher than the proportion visiting non-hosts in all comparisons involving both hosts and non-hosts. Host plants received between 3 and 9 times as many visits as non- hosts. Therefore, discrimination between hosts and non- hosts occurs during the movement stage Of oviposition. Broccoli (a highly acceptable host) received a significantly higher proportion of visits than did mustard (a host of lower acceptability) (comparison III). Thus, discrimination also occurs between hosts of different acceptabilities during the movement stage. The proportion of butterflies visiting target non-hosts did not differ significantly for any of the non-host species in comparisons involving two or more non-hosts (comparisons II, III and IV). Although the tread obtained in comparison number IV is consistent with the hypothesis (repellent tansy 83 with the lowest proportion of visits, lettuce with the highest proportion), these proportions are not very different. Thus, these data do not support the idea that discrimination between non-hosts occurs during the movement stage of oviposition. The presence of tansy near the broccoli host did not have the hypothesized "repellent" effect during the movement stage. The proportion of butterflies visiting the broccoli- tansy combination target, although lower, was not significantly different from the proportion visiting broccoli alone. In summary, discrimination was observed between hosts and non-hosts and between hosts of different acceptabilities during the movement stage of oviposition. Thus, the following two ideas are supported by these data: 1) host- specific stimuli are responsible for eliciting visit behaviors, and 2) quantitative and qualitative variations in host-specific stimuli result in hosts of higher acceptability being recognized sooner and/or by a higher proportion of butterflies than hosts of lower acceptability. Analysis of the response to alight stage (V-->l-c) revealed that, in most cases, a large proportion of the butterflies visiting a target plant ultimately landed on and/or contacted that plant (Table 8). The percentage Of responding butterflies alighting ranged from 100% (lettuce, comparison III) to 33% (tansy, comparison II). 84 Table 8. Number of visiting butterflies alighting (V-->l-c) and ending (V-->end) for different plant species. Plant Number‘ Number Percent Comparison Species Ending Alighting Alighting I Broccoli 38 222 85.4 a Tansy 27 22 44.9 b ***c=31.52, df=1, P <.001 II Broccoli 19 73 79.4 a Tansy 20 10 33.3 b Sage 18 16 47.1 b ***G=25.89, df=3, P <.001 III Broccoli 13 105 89.0 a Tansy 7 6 46.0 b Mustard 10 69 87.3 a Lettuce 0 23 100.0 a ***G=19.l3, df=3, P <.001 IV Tansy 2 5 71.4 a Lettuce 0 15 100.0 b Soybean 3 6 66.7 a *G=7.56, df=2, P=.0228 V Broccoli 13 105 90.0 a Tansy 5 8 61.5 b Broccoli-Tansy 21 89 80.9 a *G=6.93, df=2, P=.0312 a,b,c a Data were analyzed by a G test (Sokal and Rholf 1969). Percentages followed by the same letter are not significantly different at the .05 level by an STP test. * - P<.05 ** - R§.01 *** = P5.001 85 In most comparisons, a much higher proportion of butterflies visiting hosts alighted than butterflies visiting non-hosts. For example: in comparison I, over 85% of the butterflies visiting broccoli alighted while less than 45% of visiting butterflies alighted on tansy. The sole exception to this trend involved visits to the lettuce target plant (comparison III). A higher proportion of visiting butterflies ultimately alighted on the non-host lettuce than on either host (broccoli or mustard) in that comparison (100%, 87% and 89% respectively). Thus, some degree of discrimination between hosts and non-hosts occurs during the response stage of the ovipositional sequence, although there are exceptions. A slightly lower proportion of visiting butterflies alighted on mustard (a less acceptable host) than on broccoli (a more acceptable host). This difference though, was not statistically significant (P=.989). Therefore, the idea that discrimination between hosts of different acceptabilities occurs during the response stage of oviposition is not supported by these data. The only significant difference among non-hosts in the proportion of visiting butterflies alighting is the aforementioned case of the lettuce target. In comparison IV, the proportion of visiting butterflies that alighted on lettuce was 100%. The proportions for tansy and soybean were 71% and 67% respectively. Thus, some degree of 86 discrimination between different non-host species appears to occur during the response stage of oviposition. The data do not support the hypothesis that tansy has a repellent effect on visiting butterflies. In comparison IV, the proportion of visiting butterflies alighting on the tansy target was only slightly lower than the proportion alighting on the soybean target. In comparison V, the proportion of visiting butterflies alighting on the broccoli-tansy combination target was also only slightly lower than the proportion alighting on the broccoli target. In both cases, this difference was not statistically significant. In summary, the analysis of the response to alight transition shows that discrimination occurs between hosts and non-hosts and among different species of non-host during the alight stage of the ovipositional sequence. Thus, the ideas that host-specific stimuli elicit alighting and that visual stimuli influence the nature of the visit are supported by these data. The analysis of the alight to curl transition (l-c-->cu) showed that, in all comparisons, a large proportion of butterflies alighting on host plants curled their abdomens (Table 9). With a single exception, none of the butterflies alighting on non-hosts did so. Thus, strong discrimination occurs between hosts and non-hosts during the alight stage of the ovipositional sequence. 87 Table 9. Number of alighting butterflies curling (l-c-->cu) and ending (l-c-—>end) for different plant species. Plant Number Number Percent Comparison Species Ending Curling Curling I Broccoli 31 191 86.0 a Tansy 22 0 0.0 b ***G=66.ll, df=1, P <.001 II Broccoli 13 60 82.2 a Tansy 10 0 0.0 b Sage 16 0 0.0 b ***G=64.36, df=3, P <.001 III Broccoli 12 93 89.0 a Tansy 6 0 0.0 b Mustard 16 53 76.8 a Lettuce 22 1 4.4 b ***G=8l.54, df=3, P <.001 IV Tansy 5 0 0.0 a Lettuce 15 0 0.0 a Soybean 6 0 0.0 a V Broccoli 12 93 88.6 a Tansy 8 0 0.0 b Broccoli-Tansy 18 71 79.8 a ***G=31.07, df=2, P <.001 a,b,c = Data were analyzed by a G test (Sokal and Rholf 1969). Percentages followed by the same letter are not significantly different at the .05 level by an STP test. * = P<.05 ** = RE.01 *** = P5.001 88 In comparison III, 89% of the butterflies alighting on the broccoli target curled their abdomen while 76.8% of the butterflies alighting on mustard curled. Although the results conformed to the prediction (a lower proportion of alighting butterflies curling on a less acceptable host), the difference is not statistically significant (P=.244). As expected, the proportion of alighting butterflies that curled did not differ for different species of target non-hosts. In almost all cases, butterflies alighting on non-hosts did not curl. The only exception involved lettuce (comparison III). One of the 23 individuals alighting on lettuce curled her abdomen (4%), but did not lay an egg. Therefore, the data support the idea that discrimination does not occur between different non-host species during the alight stage. As expected, a lower proportion of butterflies alighting on the combined broccoli-tansy target curled their abdomen than those alighting on broccoli alone (80%, broccoli-tansy: 89%, broccoli). The difference, however, was not statistically significant. Therefore, these data do not support the idea that the presence of tansy has a significant repellent effect on alighting butterflies. In summary, these results show that discrimination occurs only between hosts and non-hosts during the alight stage of oviposition. Thus, the idea that host-specific contact chemicals release the curl response is supported by these data. 89 Level III: Transitions between Successive Steps of Oviposition On this level, the component behaviors involved in each of the stages examined in Level II are analyzed. Whereas the analysis of Level II established the presence or absence of discrimination at each stage, the present analysis determines precisely how this discrimination is manifested behaviorally. IIIA: Movement t2 Response Transition In the analysis of Level IIA (movement to response transition), discrimination was assessed by comparing the proportion of butterflies entering the target area that exhibited an observable response (visit) toward the target plant for different target plant species. In order to exhibit a response (visit) toward the target plant a butterfly flying into the target area had to either: 1) fly into the plant space directly via the air (i.e., descend from above) and visit the plant (Ap-->V), or 2) fly into the soil space, fly into the plant space, and then visit the plant (s-->p-->V). In this section these individual steps of the movement stage are examined. The numbers of butterflies making each specific behavioral transition, and the transitional probabilities between behaviors are given in Appendix C. Approximately half of all butterflies flying into the soil space also flew into the plant space (Table 10). Percentages ranged from 90 Table 10. Number of butterflies entering the soil space that entered the plant space (s-->p) and that ended (s-->p) for different plant species. Number Percent Entering Entering Plant Number Plant Plant Comparison Species Ending Space Space I Broccoli 680 682 51.1 a Tansy 837 539 39.2 b ***G=32.99, df=1, P <.001 II Broccoli 324 274 45.8 a Tansy 393 212 35.0 b Pot 369 239 39.3 ab Sage 382 218 36.3 b ***G=l7.34, df=3, P <.001 III Broccoli 298 315 51.4 a Tansy 362 252 40.1 b Mustard 323 289 47.2 a Lettuce 378 238 38.6 b ***G=25.18, df=3, p <.001 IV Tansy 186 144 43.6 a Lettuce 193 136 41.3 a Soybean 178 148 45.4 a G=l.ll, df=2, P=.574 V Broccoli 183 264 59.1 a Tansy 229 225 49.6 b Broccoli-Tansy 181 265 59.4 a **G=ll.42, df=2, P=.0033 a,b,c = Data were analyzed by a G test (Sokal and Rholf 1969). Percentages followed by the same letter are not significantly different at the .05 level by an STP test. * - 25.05 ** = 25.01 *** = 25.001 91 59.4% (broccoli-tansy, comparison V) to 35.0 % (tansy, comparison II). Without exception, the percentage of butterflies flying into the plant space was higher for target hosts than for target non-hosts. In most cases, these differences were significant. Therefore, these results support the idea that host-specific stimuli stimulate the approach to plants. . As expected, a lower proportion of butterflies flew from the soil space to the plant space when the less acceptable host, mustard, was the target plant rather than the more acceptable host, broccoli (comparison III). This difference, however, was not statistically significant (P=.546). Contrary to expectation, the proportion of butterflies flying from the soil space to the plant space was lower for lettuce than for other non-host species in comparisons III and IV. Therefore, these data do not support the idea that general plant appearance influences the approach to the plant. The proportion of butterflies flying into the plant space from the soil space was lower for the tansy and sage targets than for the pot target in comparison II. Initially, this may appear to confirm the repellent effect of tansy and sage. However, butterflies may have simply perceived tansy and sage as a obstacle in their flight path to be avoided. The target plant took up considerable area within the plant space. This fact alone may explain the 92 lower proportion of butterflies flying into the plant space with the tansy and sage targets as compared with the pot target. There were no non-repellent non-hosts in comparison II to test this idea. In agreement with the idea that the lower proportion of butterflies flying from the soil space to the plant space is not due to a repellent effect of tansy or sage, the proportion was not lower for the tansy target than for other non-host targets in comparisons III and IV. In fact, this proportion was higher for tansy than for lettuce. In addition, the proportion of butterflies flying from the soil space to the plant space was not different for the broccoli- tansy combination target than for the broccoli target alone (59.1% and 59.4% respectively). The proportion of butterflies in the plant space that visited the plant (p-->V) ranged from 42.1% (broccoli, comparison V) to 7% (pot, comparison II) (Table 11). The trends are the same as those discussed in the analysis of the response to alight transition (Level II). The advantage to the present analysis is that certain differences (in the proportion of butterflies visiting) between plant species within a comparison may become significant due to the elimination of superfluous data (i.e., many of the butterflies entering the target area for reasons other than response to the target plant left the target area without entering the plant space and are, thus, not considered in this analysis). 93 Table 11. Number of butterflies entering the plant space visiting (p-->V) and ending (p-->end) for different plant species. Plant Number Number Percent Comparison Species Ending Visiting Visiting I Broccoli 457 260 36.3 a Tansy 508 49 8.8 b ***G=l40.66, df=l, P <.001 II Broccoli 194 92 32.2 c Tansy 189 30 13.7 ab Pot 225 16 7.3 a Sage 195 34 14.8 b ***G=l40.66, df=3, P <.001 III Broccoli 207 118 36.3 a Tansy 246 13 5.0 b Mustard 221 79 26.3 a Lettuce 221 23 9.4 b ***G=121.83 df=3, P <.001 IV Tansy 133 7 4.8 a Lettuce 125 15 10.7 a Soybean 143 9 5.9 a G=4.03, df=2, P=.258 V Brocco1i 162 118 42.1 a Tansy 218 13 5.6 b Broccoli-Tansy 169 110 39.4 a ***G=218.58, df=2, P <.001 a,b,c = Data were analyzed by a G test (Sokal and Rholf 1969). Percentages followed by the same letter are not significantly different at the .05 level by an STP test. * = P<.05 ** - P§.01 *** = P§.001 94 In comparison II the proportion of butterflies entering the plant space that visited the plant is significantly lower for the "pot" than for the other targets. This supports the idea presented earlier: that visit behaviors represent a response to the plant itself. However, in comparison III the difference in the proportion of butterflies visiting broccoli and mustard falls short of significance at the .05 level (G=7.23, df=3, P=.065). In summary, analysis of the soil space to plant space (s-->p) and plant space to visit (p-->V) transitions support the idea that butterflies are able to perceive host-specific stimuli while at least short distances from the plant and respond to these stimuli by approaching and by visiting. The data do not support the idea that visually perceived characteristics that lettuces shares with the host mustard play a role in the approach or visit response. Neither do the data support the idea that tansy and sage possess repellent stimuli that deter the approach and visit response. Proportion 9f butterflies entering the target area from 32913 gs the proportion entering via the soil space (Ap,y§ _). As explained previously, there were two ways for butterflies to enter the target area and, thus, initiate sequences: by flying into the soil space (5) or by flying into the plant space directly (Ap). It was predicted that the proportion of butterflies entering the target area via 95 the air would be higher for host targets than for non-host targets. This prediction rests on two ideas: 1) butterflies entering the target area from above (Ap) were less likely to be entering for reasons other than response to the target plant, than those entering via the soil space (see next section), and 2) butterflies are able to perceive and respond to plant stimuli from outside of the target area. If these ideas are true, then the effect of a host in the target area would be to draw in responsive butterflies from above. The presence of a host in the target area would also draw responsive butterflies into the soil space. However, since the number entering the target area via the soil space for reasons other than oviposition would not change and because the number of sequences observed per day was fixed, a higher proportion of butterflies would be entering the target area from above (Ap increases more than 5 because a higher proportion of the latter group of butterflies are responding to the plant). It was found that a very small proportion of butterflies entering the target area entered the plant space directly (Ap) (Table 12). Percentages ranged from .3% (tansy, comparison IV: pot, comparison II) to 3.5% (broccoli, comparison V). In all comparisons, the proportion of butterflies entering the target space directly from above (Ap) was higher for hosts than for non-hosts. However, in only two of the comparisons (comparisons I and II) were these proportions shown to be significantly 96 G=5.27, df=2, P=.0717 Table 12. Number of butterflies entering the target space via the soil space (s) and from above (Ap) for different plant species. Number Number Percent Plant Via Soil From From Comparison Species Space Above Above I Broccoli 1362 35 2.5 a Tansy 1376 18 1.3 b *G=5.62, df=1, P=.0178 II Broccoli 598 12 2.0 a Tansy 605 7 1.1 ab Pot 608 2 0.3 b Sage 600 11 1.8 ab *G=9.42, df=3, P=.0242 III Broccoli 613 10 1.6 a Tansy 614 7 1.1 a Mustard 612 11 1.8 a Lettuce 616 6 1.0 a G=2.03, df=3, P=.566 IV Tansy 330 1 0.3 a Lettuce 329 4 1.2 a Soybean 326 4 1.2 a G=2.42, df=2, P=.298 V Broccoli 447 16 3.5 a Tansy 454 6 1.3 a Broccoli-Tansy 446 14 3.0 a a,b,c = Data were analyzed by a G test (Sokal and Rholf 1969). Percentages followed by the same letter are not significantly different at the .05 level by an STP test. * - Pi-OS ** = P5.01 *** = P5.001 97 different. This may be, in part, due to the low numbers of butterflies entering the plant area directly (Ap). As expected, the proportion of butterflies entering the target area from above was less for the "pot" target than for any target plant species in comparison II. This agrees with the idea that a certain proportion of butterflies entered the target area in response to the target plant. The proportion entering from above was significantly different between pot and broccoli. There is no significant difference among the proportions entering from above between pot, sage, and tansy. Therefore, these data support the ideas that: 1) host-specific stimuli elicit the approach to the plant, 2) butterflies are able to perceive these host- specific stimuli from outside the target area, and 3) butterflies entering the target area from above are less likely to be entering for reasons other than in response to the target plant than butterflies entering the target area via the soil space. Proportion g§_visiting butterflies y§ method 9; entry. The idea that a higher proportion of butterflies entering the plant space from above (Ap) do so in response to the target plant than butterflies entering the target area via the outer (soil) space (5) was further tested by comparing the proportion of butterflies visiting the plant for these two methods of entry. It was predicted that a higher proportion of those butterflies entering the plant space from above would visit the plant (Ap-->V) than those 98 entering the plant space via the soil space (s-->p-—>V). A significantly higher proportion of butterflies entering the target area from above visited the broccoli target (comparison I) (Figure 7). Although the proportion of butterflies entering from above that visited was higher for both the tansy and sage targets (comparison II), the differences were not significant. It is, perhaps, not surprising that the proportion of butterflies visiting from above was not significantly greater for the non-host plant species. As discussed previously, there is good evidence that butterflies perceive host-specific plant stimuli and respond preferentially to host plants while they are at least short distances from plants. A non-host plant, lacking host-specific releasers, would not have the effect, expected with hosts, i.e., of "drawing" in responsive butterflies from above. The surprise result was the mustard target. A higher proportion of butterflies visited mustard from the soil space than from above (35% from soil space, 18% from above). The number of butterflies entering the plant space from above was low, (11 sequences), and this may in part account for the unexpected result. However, it may be that mustard, being a less acceptable host, is not recognized from the same distances as is broccoli. In summary, the data support the idea that butterflies entering the plant space from above are more likely to visit than those entering from the soil space for at least highly 99 BUTTERFLIES THAT: ENTERED PLANT AREA FROM ABOVE (AP) 70— I ENTERED PLANT AREA VIA SOIL SPACE (S-->P) 60‘ 50‘ 40" 30" PERCENTAGE OF BUTTERFLIES B ROCCOLI TA "5? SAGE HUSTA RD Figure 7. Percentage of butterflies entering the plant area from above (Ap) and via the soil space (s-->p) that visited broccoli, tansg, sage and mustard. Percentages shown under the same letter are not significantly different at the .05 level. 100 acceptable target plants (broccoli). This result suggests that butterflies are able to recognize and respond to some host species from short distances. Distance from which butterflies perceive and respond tg pests. Analyses in previous sections have suggested that butterflies are capable of perceiving host-specific plant stimuli while some distance from hosts and that they respond to these stimuli by approaching the plant. With these data there is no way of determining the exact distance from which host-specific stimuli are perceived. However, it is possible to determine indirectly whether most perception and response takes place at distances shorter than, or longer than, the soil space/plant space boundary. Butterflies entering the soil space can be divided into three groups according to their ultimate fate: 1) those that leave the soil area without entering the plant space (s-->end), 2) those that enter the plant space and leave the target area without visiting the plant (s-->p-->end), and 3) those that visit the target plant (s-->p-—>V). Since the total number of sequences observed each day was equal for each target plant species, if one assumes that a greater proportion of butterflies respond to (visit) broccoli than tansy (established previously), then the proportion of butterflies in group 3 (s—->p-->V) was greater for broccoli than for tansy, and the proportion of group 1 and/or group 2 MUST be less for broccoli than for tansy. 101 If the majority of butterflies first perceive and respond to the plant at distances greater than the soil space/plant space border, then the proportion of group 1 (s-->end) will be lower for broccoli than for tansy. If butterflies perceive and respond to plants only after entering the inner space, then the proportion of group 2 (s-->p-->end) will be lower for broccoli than for tansy. If perception and response occurs at both distances, then the proportions in both groups 1 and 2 will be lower for broccoli than for tansy. Results discussed previously (see p 94) suggest that most recognition of hosts occurs prior to butterflies crossing the soil space/plant space border. This hypothesis was tested in this section. Two 2-way ANOVA's were performed on the percentage of butterflies entering the target area each day that left the target area without entering the inner plant space (s-->end), and that entered the inner plant space, but left the target area without visiting the target plant (s-->p-->end), for broccoli and tansy in comparison I (randomized block design, days as blocks). The mean percentage of butterflies that entered the soil space, but left the target area without entering the plant space (s-->end) was 49.3% for broccoli and 60.0% for tansy (Figure 8). The analysis of variance yielded a highly significant F value (F=2.97, df=4l, P=.0003) 70‘ 60" 50" 40" 30" 20" PERCENTAGE OF BUTTERFLIES 102 [ BROCCOLI I TANSY I LEFT THE TARGET AREA LEFT THE TARGET AREA WITHOUT FLYING INTO AFTER FLYING INTO THE THE PLANT SPACE I PLANT SPACE, BUT DID NOT (S-->END) VISIT THE TARGET PLANT (S-->P—-)END) Figure 8. Percentage of butterflies entering the soil space that left the target area without visiting the plant for the broccoli and tansg target plants (comparison I). 103 The mean percentage of butterflies flying into the soil area that subsequently flew into the plant area, but left without visiting the plant (s-->p-->end) was 32.1% for broccoli and 36.2% for tansy. These means are significantly different at the .052 level (F=1.67, df=4l, P=.0518). Thus, the data support the hypothesis that most visiting butterflies recognize host plants prior to crossing the soil space/plant space boundary. The analysis also suggests that a smaller proportion of visiting butterflies are induced to visit hosts only after crossing the soil space/plant space boundary. The exact distances at which butterflies are able to perceive host-specific stimuli cannot, however, be determined from these data. Illéi Response 39 Alight Transition In the analysis in Level IIB (response to alight transition) discrimination was assessed by comparing the proportion of responding (visiting) butterflies that ultimately performed at least one land or contact on the target plant (as Opposed to the proportion that fluttered and then left the target area) for the target plant species in each comparison. There were two ways for visiting butterflies to land on or contact target plants. First, a butterfly could perform a land or contact as the first act in a visit. Second, a butterfly could perform a flutter followed by a land or a contact (f-->l-c). Analysis of data in the present section consisted of 1) comparing the Proportion of visiting butterflies that performed each of 104 the three possible acts (flutter, contact, or land) as the initial act in a visit for different target plant species, 2) comparing the transitional frequencies between an initial flutter and the three possible succeeding acts (land, contact, or end) for different target plant species, and 3) comparing the relative proportions of first contacts and first lands. Because very few visits were made to the target plants in comparison IV, this comparison was not included in the analyses. It was assumed that flutters and contacts were more exploratory behaviors than lands. Therefore, a land was likely to be performed only when the plant stimuli the butterfly perceived indicated with a fair degree of certainty that the plant was a host. The numbers of butterflies performing each behavior during the response stage to alight stage transition, and the transitional probabilities between behaviors are given in Appendix D. The highest proportion of butterflies fluttered as their first act in a visit (Table 13). This was true for all target plant species in all comparisons. The proportions fluttering ranged from 90% (tansy, comparison II) to 48% (mustard, comparison III) . A relatively low proportion of butterflies performed a contact as their first act in a visit (ranging from 17%, lettuce, comparison III; to 4%, tansy, comparison II). The proportion of butterflies initially landing ranged from 39% (mustard, comparison III) to 7% (tansy, comparison II). 105 Table 13. Number of butterflies fluttering, contacting, and landing as the first act in a visit to different plant species. Plant Number Number Number Comparison Species Fluttering Contacting Landing I Broccoli 162 24 74 a Tansy 42 4 3 b ***c=14.49, df=1, P <.001 II Broccoli 56 11 25 a Tansy 27 1 2 b Sage 27 3 4 ab *G=12.37, df=3, P=.015 III Broccoli 66 ll 41 a Tansy 11 1 1 b Mustard 38 10 31 ab Lettuce l6 4 3 a *G=12.89, df=3, P=.045 V ' Broccoli 73 10 35 a Tansy ll 1 1 a Broccoli-tansy 80 6 24 a G=5.85, df=2, P=.211 a,b,c = Data were analyzed by a G test (Sokal and Rholf 1969). Percentages followed by the same letter are not significantly different at the .05 level by an STP test. * = P<.05 ** = P§.01 *** = P5.001 106 In most cases, the first behavior performed in visits to hosts differed significantly from the first behavior performed in visits to non-hosts. In comparison I the behavior of butterflies visiting broccoli differed significantly from the behavior of butterflies visiting tansy. A higher proportion of the butterflies visiting broccoli immediately landed (29%) than butterflies visiting tansy (6%). Butterflies initiallty fluttering constituted 86% of the butterflies visiting tansy and 62% of the butterflies visiting broccoli. In comparison II the proportions of butterflies performing each of the three types of initial acts differed significantly between broccoli and tansy. There was no significant difference between broccoli and sage. Even so, the trends were consistent with the predictions: a higher proportion of visiting butterflies immediately landed on broccoli (27%) than on tansy or sage (7% and 12% respectively). Correspondingly, the proportion of fluttering butterflies was higher for sage (79%) and tansy (90%) than for broccoli (61%). In comparison III initial behavior toward hosts was not significantly different from that toward the lettuce target. Additionally, the initial behavior toward mustard did not differ significantly from that toward tansy (although the behavior toward broccoli and tansy did differ significantly). Even so, a much higher proportion of butterflies visiting hosts immediately landed (35% broccoli 107 and 39% mustard) than those visiting non-hosts (8% tansy, 13% romaine lettuce). There was no major difference between the initial behavior of butterflies visiting a highly acceptable host (broccoli) and a less acceptable host (mustard). In fact, 'the proportion of visiting butterflies that landed was slightly higher for the less acceptable host, mustard, (39%) than for the highly acceptable host, broccoli (35%). More butterflies visiting broccoli initially fluttered (56%) than those visiting mustard (48%). The proportion of contacts was slightly higher for mustard than for broccoli (13% and 9% respectively). The proportion of the butterflies visiting lettuce (comparison III) that immediately landed (13%) was not significantly different than the proportion visiting tansy (8%). Also, the initial behavior of butterflies visiting broccoli-tansy was not significantly different than the initial behaviors of butterflies visiting broccoli (comparison V). It is my contention that the three visit behaviors have different functions. Flutters and contacts are more tentative, exploratory behaviors. A butterfly may perform a flutter or a contact if the signal it receives from the plant (i.e., the stimuli it perceives) during the movement stage of the ovipositional sequence conveys ambiguous information about the nature of the plant. During a flutter or a contact, a butterfly examines the plant further to 108 obtain more precise information. Even though additional information is gathered after landing it is probable that a land is performed by butterflies already quite "know- ledgeable" about the nature of the plant. Analysis of the first behavior performed in visits to different target plant species supports the idea that the signal perceived by butterflies (i.e., plant stimuli) before they visit the plant is fundamentally different between hosts and non-hosts. The data do not support the ideas that the signal perceived is different between two hosts of different acceptabilities or between different species Of non-host. The idea that broccoli-tansy transmits a more ambiguous (i.e., less "host-like) signal than broccoli alone is also not supported by these data. The proportion of butterflies landing, contacting, and ending following an initial flutter around the target plant was strongly influenced by the target plant species (Table 14). Behavior following an initial flutter around a host was significantly different from behavior following an initial flutter around non-hosts for all plant species in all comparisons. With one exception (to be discussed later), the proportion of butterflies that left the target area following an initial flutter (f-->end) was higher for non-hosts than for hosts. Percentages of butterflies leaving the target area after an initial flutter ranged from 34% (broccoli, comparison II) to 18% (broccoli, comparison V) for hosts and from 74% (tansy, comparison 11) 109 Table 14. Number of butterflies contacting, landing, and ending following an initial flutter above different plant species. Plant Number Number Number Comparison Species Contacting Landing Ending I Broccoli 32 93 37 a Tansy 9 6 27 b ***G=31.87, df=1, P <.001 II Broccoli 12 25 19 a Tansy 4 3 20 b Sage 7 2 18 b ***G=22.15, df=3, P <.001 III Broccoli 13 40 13 a Tansy 3 1 7 b Mustard 10 18 10 a Lettuce 14 2 0 c ***G=40.77, df=3, P <.001 V ' Broccoli 11 49 13 a Tansy 4 2 5 c Broccoli-Tansy 27 32 21 b **G=17.67, df=2, P=.00143 a,b,c = Data were analyzed by a G test (Sokal and Rholf 1969). Percentages followed by the same letter are not significantly different at the .05 level by an STP test. * = P<.05 ** = P§.01 *** = P5.001 110 to 45% (tansy, comparison V) for non-hosts (excluding the exception to be discussed later). There was no significant difference between the proportion of fluttering butterflies that landed, contacted, and ended between the more acceptable host, broccoli, and the less acceptable host, mustard (G=1.7l, df=3, P=.0635). None of the butterflies fluttering around lettuce left the target area (comparison III). The percentages of butterflies continuing (landing or contacting) are actually higher for lettuce than for the two hosts in this comparison. The difference is statistically significant at the .05 level both between lettuce and the two hosts, and between lettuce and the other non-host species, tansy. The percentage of fluttering butterflies that left the target area was significantly higher for broccoli-tansy (26%) than for broccoli (18%). Therefore, tansy does appear to have a repellent effect on the behavior of fluttering butterflies, although this effect is not great. In summary, the ideas that host-specific stimuli induce fluttering butterflies to land, and that repellent plant stimuli may inhibit alighting by fluttering butterflies, are supported by these data. In addition, the data support the idea that visually-perceived general plant stimuli may influence whether fluttering butterflies will alight. The idea that variation in host-specific stimuli influences alightment of fluttering butterflies is not supported. 111 Proportions pf lands yg contacts. Previously, it was hypothesized that the act "contact" represents a more exploratory behavior than the act "land." If this were true, then the relative proportions of alighting butterflies that contacted or landed would differ among different target plant species. To examine this idea, the percentages of alighting butterflies that landed or contacted on their first alight were calculated. To obtain these percentages the number of individuals landing (or contacting) as the first act in a visit was added to the number of individuals landing (or contacting) after fluttering first. The sum was then divided by the total number of butterflies alighting. In general the results support the hypothesis that the act "contact" represents a more tentative, examining type of behavior than the act land. In all cases, the percentage of landing butterflies was higher for hosts than for non-hosts (Table 15). In most cases, this difference was significant. The percentage of alighting butterflies which landed was not significantly different between the more acceptable host and the less acceptable host mustard (G=.820, df=3, P=.845). These data also support the contention that placing tansy next to a broccoli plant results in a more ambiguous signal. A significantly higher percentage of alighting butterflies landed on broccoli alone (80%) than on the broccoli-tansy combination (63%). 112 Table 15. Number of butterflies contacting and landing on their initial alight on different plant species. Plant Number Number Percent Comparison Species Contacting Landing Landing I Broccoli 56 167 74.9 a Tansy 13 9 40.9 b **G=10.18, df=l, P=.00142 II Broccoli 23 50 68.5 a Tansy 5 5 50.0 a Sage 10 6 37.5 a G=5.84, df=2, P=.0539 III Broccoli 24 81 77.1 a Tansy 4 2 33.3 b Mustard 20 49 71.0 a Lettuce 18 5 21.7 b ***G=28.36, df=3, P <.001 V . Broccoli 21 84 80.0 a Tansy 5 3 37.5 b Broccoli-Tansy 33 56 62.9 b **G=10.97, df=2, P=.0041 a,b = Data were analyzed by a G test (Sokal and Rholf 1969). Percentages followed by the same letter are not significantly different at the .05 level by an STP test. * 8 P<.05 ** = P§.01 *** = P5.001 113 IIIC: Alight tg Curl Transition In the analysis of level 2C, discrimination was assessed by comparing the proportion of alighting butterflies that ultimately curled their abdomens at least once (as opposed to the proportion that left the target area without curling) for different target plant species. Although the ultimate act of an alighting butterfly was either to curl or not to curl, one to several lands and/or contacts were made before that ultimate act was performed. In this section each successive alight (land or contact) is considered separately. Considered in this way, three different events were possible after alighting: l) curl: The butterfly curled its abdomen, 2) end: The butterfly did not curl its abdomen and left the target area without making additional lands or contacts, and 3) continue: the butterfly did not curl its abdomen, but continued the visit by making additional lands or contacts. In this section discrimination between target plants is assessed by comparing the transitional frequencies between any given alight and the three possible succeeding acts (curl, end, or continue) for different target plant species. In addition, this section addresses whether the probabilities of performing these three acts change as the visit progresses. Results: Broccoli. Alighting butterflies made between one and eight lands and/or contacts on broccoli before either curling or leaving the target area (Figure 9). Six 114 ALICHTS 22 I T nasr ALIGHTS , 2 :5 _/ T f T— 1.: END FIRST courmue 20 81 REST ONLY'S G ‘ PIRST CURL I 14 SECOND ALIGHTS SECOND CONTINUE 85 THIRD ALIGHT SECOND CURL - 44 —— . THIRD CURL 10 3"! END TOTAL CURLS 101 TOTAL EN” ALIGHT as FOURTH OR MORE CURL 18 Figure 9. Number Of butterflies performing each behavior in the alight stage of oviposition for the broccoli target (comparison l). The width of the boxes is proportional to the number of butterflies performing the designated behavior (this number is given in the box). The width of the arrows is proportional to the transitional probabilities (Shown next to the arrow). The total area of the boxes is insignificant. 115 (2.7%) alighting butterflies ”rested" on the plant on their first alight, then left the target area without performing additional visit behaviors. These sequences were eliminated from further analysis. Figure 10 shows the proportion of alighting butterflies that: 1) curled their abdomens (curl), 2) did not curl but continued the visit by making additional lands and/or contacts (continue), and 3) left the target area without making additional lands and/or contacts (end) for each successive alight. For example, 53% of the butterflies curled on their first land or contact, 37.7% continued the visit, and 9.3% left the target area. I The proportion of butterflies performing each of the three possible events after landing can be used to estimate the probability that an individual butterfly will perform these events after a land or a contact. The probability of a butterfly performing any given event varies slightly between successive alights. For example, the probability of curling steadily decreases from 0.530 on the first alight to 0.438 on the fourth alight. The probability of continuing increases from 0.337 on the first alight to 0.562 on the fourth alight. A G test was performed to test if these probabilities ‘were independent of the number of alights performed. This was not found to be significant (G=5.77, df=4, P=.217). The distribution of the first curl performed by a butterfly as a function of the number of alightings made PERCENTAGE OF BUI IERFLIES 116 BEHAVIOR AFTER ALIGHTING: CURL I CONTINUE [IIIIIII END 60" 50" 40" 30‘ 20" IO" FIRST SECOND THIRD FOURTH ALIGI'IT ALIGHT ALIGHT ALIGHT Figure 10. Percentage of butterflies curling, continuing and ending on successive alights on broccoli (comparison I). 117 before curling is shown in Figure 11. The curve produced is similar to an exponential decay curve. Therefore, it is possible that the probability of curling remains constant on successive alights (provided the butterfly has not curled previously in the visit). This is in agreement with the results discussed previously. Results: Comparisons. What an observer records as a visit to a plant may, in reality, be a behavior performed in any one of three differents contexts: 1) oviposition: a butterfly is responding to a plant as an ovipositional substrate, 2) social: a butterfly is responding to another butterfly on or near the plant, and 3) resting: a butterfly is responding to the plant as a surface to rest on. Butterflies responding to the plant in a social context could easily be identified and eliminated from the data. Those responding in a resting context could only be identified after the rest had been performed. Up to this point, rests have been included as a visit behavior because rests constituted a very small portion of total visits. Because the analysis in this section involves very fine dissection of the post-alight sequence, it is important to identify rests and eliminate them from the data. Before doing this, the proportion of visiting butterflies resting was examined. It was hypothesized that butterflies do not respond differentially to different plant . species when resting. Therefore, it was predicted that the PERCENTAGE OF BUTTERFL I ES MB 70" 60" 50" 40" 30"“ 20‘ 10‘ FIRST HIRD ALIGHT ALIGHT ALIGHT ALIGHT ALIGHT Figure l 1. Percentage of butterflies performing their first curl on the first through fifth alight on broccoli (comparison I). 119 proportion of butterflies resting only (i.e. not exhibiting other visit behaviors) would be independent of target plant species. The proportion of butterflies "resting only" did not differ significantly for the plants in comparisons I and II (Figure 12). In comparison III a significantly higher proportion of butterflies rested on mustard (93%) than on either broccoli (33%) or lettuce (0%). In comparison IV the proportion of butterflies resting on soybean (83%) was significantly higher than the proportion resting on lettuce (0%). A Fisher's exact test was used to analyze the latter set of data due to the low number of visits. In comparison V the proportion of butterflies resting on broccoli (40%) was not significantly different (by a Fisher's exact test) from the proportion resting on broccoli-tansy (P=.l70). Therefore, it can be concluded that the probability of a butterfly resting is apt independent of target plant species. Some species of plant (i.e., soybean and mustard) elicit proportionally more rests than others. The stimuli affecting resting behavior are not known, but they are not the same stimuli as those responsible for ovipositional behavior, since there were no consistent differences in the proportion of rests on hosts and non—hosts. The numbers of butterflies ending, continuing, and curling on each land or contact for broccoli and tansy (comparison I) are shown in Figure 13. Percentages are 120 .mcEsooo mews Lo Lopez: 32 o£ 2 one cognac so: Soon. soc—31.0805" ._.m .cooc>omu>m .oo::o_u.. confine“: 693$ 595$. .__oooobumm .65. mo. o5 Lo anoEo zEmoEch Lo: mam coco. oEmm o5 coca: Esocm momoEooaon. .mo_ooam Ema E9916 :0 25 some 85 8:525. 95:93 Le ommbcooaoa .9 929.1 mum—.52 zen—acute“- > >_ _: = _ ION Fov Foo Tos loo— SBI'HEIEIIIIIQ :IO 39V1N3333d FIRST ALIGHT SECOND ALIGHT THIRD ALIGHT END 20 (9%) END 2 (2%) END 3 (93) IIRCDCBCEDILI] \ CURL I I4 (533) \/ CONTINUE BI (38%) CURL 44 (543) \/ CONTINUE 35 (A31) \ CURL 16 (463) \/ CONTINUE 16 (462) 12 l TAMI? /‘ CURL 0 \/ CONTINUE 3 (I82) CURL \/ CONTINUE I (3.3.3) / CURL \/ CONTINUE I END 14 (822) END (67%) END Figure 13. Number of butterflies curling, continuing, and ending after the first three alights on broccoli and tansg (comparison I). The number in parentheses is the percentage of butterflies making that particular alighting that performed the designated behavior. 122 based on the proportion of butterflies making the particular land/contact that performed each event. The most obvious difference between the two plants lies in the percentage of butterflies curling. Approximately half of the butterflies curled after each alight on broccoli while none of the butterflies curled on tansy. The second most striking difference is in the percentage ending. A higher proportion of butterflies landing on or contacting tansy ended than those alighting on broccoli. The effect is so striking that only 3 of the 17 butterflies alighting on tansy made more than one land or contact. These results are consistent with the idea that host-specific stimuli, perceived by the butterfly after alighting, are crucial to eliciting post-alight ovipositional behavior (curl or continue). The proportion of butterflies curling, continuing, and ending after each succesive land and/or contact for broccoli and for mustard (combination III) remains fairly constant for broccoli during the first three alights (Figure 14). The proportion curling on mustard increases with successive alights. The proportion ending after alighting on mustard increases and the proportion continuing decreases. To test whether the data for mustard were stationary (i.e., whether the frequency of butterflies performing each of the events differed on successive alights) a G-test was performed. The G-statistic was not significant at the .05 level. This indicates that, despite the apparent trend in 123 5.19.2 o~==._. . 339:: can 288.5 cc 3593 3...: EC 2: Leta 9:28 uca $52550 .9250 mmztmfian .6 3359.3 5.12.? ozcumm FIG—Ad Ema—n. Max:290 e: .250 max-hzau can .250 wait-cu can 925:: I 388% E .250 lo— TON Fom lov Flow [00 low .3 950: SEII'IdEISllOQ :IO 39V1N3383d 124 the data, the probability that a butterfly will continue, end, or curl is not significantly influenced by the number of times it has previously alighted on the plant. Because data for successive alightings were stationary, they were lumped together to test if the probability of ending, continuing, or curling was independent of the target plant species. The G-statistic obtained was 3.88, which is not significant at the .05 level (P=.l437, df=2). Therefore, the hypothesis that post-alighting ovipositional behavior is influenced by variation in host-specific stimuli between these two hosts is not supported by these data. The numbers of butterflies ending vs continuing after alighting on non-hosts are too low to be analyzed statistically. However, proportions appear not to differ greatly among different non-host species (Table 16). Very few butterflies (a total of eight of the sixty-four butterflies alighting on all non-host species combined) made more than one land or contact. This supports the idea that the absence of host-specific plant stimuli results in termination of the ovipositional sequence. These data also suggest that a single land or contact is sufficient for a butterfly to detect the absence of host-specific chemicals. Figure 15 gives the proportions of butterflies that curled, continued, or ended after alighting on broccoli and broccoli-tansy (comparison V). Again, these proportions are fairly constant for the first three alightings on broccoli. 125 Table 16. Number of butterflies ending and continuing after their first alight on different non-host species. Plant Number Number Percent Comparison Species Continuing Ending Continuing ,II Tansy 0 6 0 Sage l 10 9 III Tansy 1 5 lo Lettuce 2 20 20 IV Tansy l 4 20 Lettuce 3 10 23 Soybean 0 l 0 126 $ch “3.: 9.: Late 9%.; use .mcSczcoo .9226 853:3 Co $35.88; ._.:m:._< Dm__._._. 5207:8005 ucc 288.5 co 359:0 ._._._m:._< ozoum—m P1246. Fwd—u max—#29. aw: .EDU — _ mum—2:20“. azu 43:0 m:z_hzou 92m Ana—UH >mzfi-:ouuo% I 388% a lom loo ION .m— 9.50: SEII'HEIEILLOQ :IO 39V1N3333d 127 The proportions for broccoli-tansy vary with the number of alightings made. One notable difference between the data for broccoli and broccoli-tansy is in behavior of butterflies after the first land or contact. A higher proportion of butterflies curled after making their first alight on broccoli and a correspondingly lower proportion ended than those alighting on broccoli-tansy. The question arising is whether the increase in the proportion of butterflies ending after alighting on broccoli-tansy results from individual butterflies which made their first alighting on tansy foliage instead of on broccoli foliage. Figure 16 shows the proportion of butterflies ending, curling, and continuing after making their first alightment on the broccoli leaves of the broccoli-tansy combination vs those alighting on the tansy leaves. Eleven butterflies made their first alightment on tansy. Of these, 1 (9%) curled, 6 (55%) ended, and 4 (36%) continued. These proportions are quite different from the same proportions of butterflies making their first alightment either on the broccoli target alone, on the tansy plant alone, or on broccoli leaves in the broccoli-tansy combination. Two points merit emphasis. First, after alighting on tansy leaves in the broccoli-tansy combination a butterfly behaves differently than after alighting on a host. Therefore, it can be concluded that contact cues are important determinants of butterfly behavior. Secondly, the behavior 128 I I BROCCOLI LEAVES TANSY LEAVES 60" 50“ 40" 30" 20" 10" PERCENTAGE OF BUTTERFLIES CURL END CONTINUE Figure 16. Percentage of butterflies curling, ending, and continuing after making their first alight on the broccoli leaves and the tansg leaves of the broccoli-tansy target. 129 of butterflies initially alighting on tansy leaves in the broccoli-tansy combination was different from the behavior of butterflies alighting on the tansy target alone. During the entire study, not one alighting butterfly curled on tansy. Yet one of the eleven butterflies making its initial alightment on the tansy leaves of the broccoli-tansy combination curled. The proportion of butterflies ending after the first alight on tansy (comparison V) was 88%. Only 55% of butterflies making their initial alight on the tansy leaves of the broccoli-tansy combination ended. Thus, even though the results suggest that contact stimuli strongly influence post-alighting behavior, they apparently are not the sole determinants. Non-contact plant stimuli perceived either prior to or after alighting seem to also influence post-alighting response. In summary, analysis of the behaviors involved in the alight stage to curl stage transition support the idea that host-specific contact stimuli are releasers of the curl response. Differences in host-specific contact stimuli between broccoli and mustard do not appear to influence post-alighting behavior. The absence of these host-specific contact stimuli induces a butterfly to terminate the visit. However, data from the broccoli-tansy target indicates that other, non-contact stimuli also influence post-alighting behavior. 130 Level IV: The Visit Sequence The purpose of this analysis is to describe visits to a target plant as a sequence of behaviors. Three visit sequence descriptions are produced: one to each of the target plant species broccoli, tansy, and mustard. The visit sequences are analyzed and described as Markov chains. As Cane (1978) states, "The advantage of describing a sequence of observations on the behavior of an animal in terms of a Markov chain is that a summary of the behavior is produced that enables comparisons to be made between different animals and between different conditions for the same animal." Thus, producing descriptions of visit sequences to three different target plant species as Markov chains allows comparisons among sequences and may give some insight into causation of the behaviors. Two or more behavior patterns are performed by an animal close together in time either because the performance of one act in some way facilitates the performance of the other and/or because the acts have causal factors in common (Slater 1973). Causal factors potentially include the internal state of the animal and/or environmental stimuli (Slater 1973). Sequence analysis enables determination of the degree to which one behavior influences another and, by default, the degree to which the two are determined by common causal factors. It does 29E allow for separation of internal from external causal factors. 131 The average internal motivation was assumed to be the same for groups of butterflies visiting different target plant species. This seems a reasonable assumption, although it can neither be proved nor disproved here. The average environmental conditions were also assumed to be roughly equivalent, with the exception of the target plant species. Under these assumptions, differences in the visit sequence result from differences in target plant species. The description of the visit sequence includes all possible acts and transitional probabilities between acts. A transitional probability is the probability that, given that the butterfly has performed a specified behavior, the next behavior performed will be a given act. This probability is potentially affected by a number of factors. To meet the requirements for a Markov chain, transitional probabilities must be constant over time and be affected only by the type of acts immediately preceding the transition (i.e., not be affected by the butterflies' previous histories). The number of immediately preceeding acts known to affect a transitional probability is the order of the Markov chain. The requirement for transitional probabilities to remain constant over time and not be influenced by history is termed stationarity. It is frequently recommended that data be initially tested for stationarity (Chatfield and Lemon 1970, Slater 1973). While it is never possible to prove stationarity in the absolute sense, the more obvious 132 factors can be checked for influences on the transitional probabilities. The sequence performed by butterflies visiting the broccoli target plant (comparison I) was tested for stationarity. It was not possible to test whether events occurring prior to the entry of the butterflies into the target area influenced transitional probabilities, but it was assumed that any such influence would tend to average out between plant types. In any sequence of behaviors, the transitional probabilities between acts may change during the course of a sequence due to changing levels of motivation and/or due to information gained during the sequence (for examples see Lemon and Chatfield 1971, Oden 1977, Fraser and Nelson 1984). Broccoli data were analyzed to answer the following questions: 1) Do the transitional probabilities between two given acts change as the sequence progresses, and 2) Does the occurrence of certain crucial events in the sequence (i.e., landing and curling) affect transitional probabilities between acts? Effect 2; Step Number .A sequence step was defined as corresponding to the performance of one behavior pattern. Therefore, the number of steps in the sequence equals the number of behaviors performed, and the nth step in the sequence corresponds to the nth behavior performed (Oden 1977). An easy way to check for non-stationarity during the sequence is to compare frequencies of various acts at 133 different steps of the sequence (Oden 1977). Table 17 gives these frequencies for visits to broccoli. Because the relative proportions of various acts appear to change as the visit progresses, initial examination of this table leads to the conclusion that data are not stationary. However, this apparent non-stationarity could be due to the definition of the various acts rather than to a true trend: of the nine transitions, one, flutter to flutter, was forbidden as defined. By definition, a flutter was considered to continue until the butterfly performed a different type of act (land, contact, or end). The reduction in the relative proportion of butterflies fluttering at step two could simply be due to the high number fluttering at step one which were, by definition, prevented from fluttering on the second step. To check data for stationarity, and to correct for the influenceof the forbidden flutter-to-flutter transition, three transition matrices were constructed: one each for transitions from a flutter, a contact, and a land, as a function of step number (data includes behaviors only up to the first curl). These matrices are shown in Table 18. Each matrix was tested for independence of transitional frequency from step number. The G-statistics obtained were not significant (flutter transitions, G=5.73, df=6, P=.454; contact transitions, G=4.0l, df=8, P=.856; land transition, G=7.68, df=8, P=.465). Thus, until the first curl (or through step 4) the transitional probabilities remain 134 Table 17. Number of butterflies performing a flutter, a contact, a land, a land-curl, or an end on steps 1 through 4 of visits to the broccoli target (comparison I). Step Land- Number Flutter Contact Land Curl End 1 158 24 22 49 __ 2 7 41 26 84 46 3 16 17 7 26 7 4 2 10 6 l6 7 Behavior is included only up to the first curl. 135 Table 18. Number of butterflies performing specified transitions to all possible behaviors following a flutter, a contact, and a land as a function of step number. SECOND BEHAVIOR Step Land- Number Flutter Contact Land Curl End FROM A FLUTTER AS THE FIRST BEHAVIOR: 1 to 2 * 33 22 65 38 2 to 3 * 3 1 3 O 3 to 4 * 3 1 7 5 G =5.73, df =6, P =.454 FROM A CONTACT AS THE FIRST BEHAVIOR: 1 to 2 4 5 2 9 4 2 to 3 7 11 2 16 5 3 to 4 ' 1 6 2 7 1 :3 =4.01, df =8, P =.856 FROM A LAND AS THE FIRST BEHAVIOR: 1 to 2 3 3 2 10 4 2 to 3 9 3 4 7 2 3 to 4 l 1 3 2 1 G =7.68, df =8, P =.465 * The flutter-to-flutter transition is forbidden by definition. 136 constant. This indicates that the data up to the first curl are stationary with respect to step number. Effect 9; Key Events It was hypothesized that the occurrence of two "key" events--landing and curling--could change the transitional probabilities between acts. Landing had the potential to affect transitional probabilities in that butterflies potentially received additional information about the plant after landing. Curling could affect transitional probabilities by changing the internal state (motivation) of a butterfly. The transitional frequencies between acts were visually inspected to detect differences between behavior prior to and following the first land. There were only minimal differences in transitional probabilities. Thus, the event of landing does not change transitional probabilities. To test whether the transitional probabilities change after the performance of a curl, four matrices were constructed. Each matrix included the transitional frequencies from one of four behaviors-~a flutter, contact, land and land-curl--as a function of "experience." "Experience" categories included: 1) behaviors prior to the first curl, 2) behaviors from the first curl up to the second curl, and 3) behaviors from the second to the third curl. These matrices are shown in Table 19. A G-test was performed to test whether the transitional frequencies were independent of experience. The transitional frequencies 137 Table 19. Number of butterflies performing specified transitions to all possible behaviors following a flutter, a contact, a land, and a land-curl as a function of "experience”. SECOND BEHAVIOR Experience Land- Cateogory Flutter Contact Land Curl FROM A FLUTTER AS THE FIRST BEHAVIOR: Before first curl * 39 24 75 After first curl * 6 4 28 After second curl * 4 8 17 6 =8.15, df =4, P =.oa7 FROM A CONTACT AS THE FIRST BEHAVIOR: Before first curl 12 22 6 32 After first curl 2 3 O 5 After second curl 2 5 l 5 G =2.oz, df =6, P =.918 FROM A LAND AS THE FIRST BEHAVIOR: Before first curl 13 7 9 19 After first curl 2 O 1 10 After second curl 4 l 3 4 6 =8.79, df =6, P =.186 FROM.A LAND-CURL as THE FIRST BEHAVIOR: After first curl 45 2 8 24 After second curl 31 6 l 4 **G =15.24, df =3, P =.002 * The flutter-to-flutter transition’is forbidden by definition. 138 from a flutter, a contact, and a land without a curl were independent of experience (flutter transitions: G=8.14, df=4, p=.087; contact transitions: G=2.02, df=6, p=.918; land-no curl transitions: G=8.79, df=6, P=.186). However, the transitional frequencies from a land with a curl were not independent of experience (G=15.25, df=3, P=.002). A butterfly was more likely to flutter or contact, and less likely to land or land-curl following the second curl as compared to after the first curl. Figure 17 shows the relative proportions of four possible acts (flutter, contact, land-no curl, and land- curl) at four key points in the visit sequence: 1) as the first behavior in a visit, 2) following the first land or contact in a visit, 3) following the first curl, and 4) following the second curl. The frequencies of these acts are not independent of the "key points" in the sequence listed above (G=87.91, df=9, P=4.3xlO-15). Thus, although the sequence of behaviors up to the first curl is stationary, the performance of a curl changes the transitional probabilities of at least the first act following a curl. As Figure 17 reveals, the probability of fluttering decreases following the first land or contact (as compared to the probability of fluttering as the first behavior in a visit), but increases again after a curl has been performed. The probability of contacting or landing without curling increases after the first land or contact has been 139 ._So 6:83 9: 956:8 a. new i8 as 9: 6522.2 8 figs as 9: 6532.2 a .=m_> m :_ S ”8328 E: 9: mm .50 a 5:5 9.652 new .6598. 6589.50 65.35: 3:553 Lo mmmEmEma .5 9:9... .250 a: Juan—lazca $2.. buchzcu “HP—b.7— 0— ON on CW om oo SEII'IJEIEIILI'IQ :IO 39V1N33213d :53. Ga; op .38 028mm “152.3348 M ME 6533...: I .58 5%... at 02.336... E cm; 4 2. WM 8 “moSaEm 5a.... 140 made (compared to the probability of landing or curling as the first behavior in a visit), and, likewise, decreases after a curl has been performed. Collectively, this indicates the existence of a discrete behavioral unit--the "curl sequence"--which can be defined as the series of acts leading up to a curl. The transitional probabilities of the various acts are similar at the beginning of the visit and after a curl has been performed. The curl sequence is most likely to begin with a flutter. The probability of curling increases if at least one prior behavior has been performed in the particular curl sequence. This implies that the butterflies do not curl in bouts, but rather, after curling, start the curl sequence over again. Determination g; the Order 9; the Chain As the previous analysis indicates, the visit sequence is not stationary. The transitional probabilities between acts are affected by the event of curling. Therefore, in this section data only up to the first curl are included in calculations. The next step in the analysis is to determine the number of preceding behaviors that influence the probability of performing an act, and thus establish the order of the Markov chain. Initially the data are tested for independence of acts immediately following one another. If succeeding acts are independent of preceding acts, then the conditional probability of act j occurring given that act 1 has just occurred is equal to the unconditional probability 141 of act j occurring (Chatfield and Lemon 1970). A sequence of independent events is called a "zeroth-order" Markov chain (Slater 1973). If data do not fit the independence model, then they are tested for fit to a first order Markov chain, where the probability of performing a given act depends only on the immediately preceding act and not on earlier ones. If this model is also rejected, the next step is to test for fit to a second order Markov chain, where the probability of performing a given act depends only on the two preceding acts. This proceedure is repeated until the order of the Markov chain is determined (Fagen and Young 1978). The usual test for independence is to construct a matrix listing the transitional frequencies from preceding behaviors (rows) to following behaviors (columns), and to conduct a Chi-Square or G-test of independence. With these data, however, the forbidden flutter-to-flutter transition could result in erroneous conclusions with this method (Slater 1973). The method used with these data was a test for quasi-independence given by Goodman (1968). The entire table is collapsed four ways to yield four subtables (see Table 20). The subtable containing the impossible transition is eliminated from conSideration. Chi-square tests are performed on the remaining subtables, and their sum is used as the test statistic. The degrees of freedom for this statistic is the sum of the degrees of freedom of the three individual subtables, or the degrees of freedom of 142 Table 20. Chi-Square test for quasi-independence of preceding behavior from following behavior on a table including an impossible transition (flutter-to-flutter). ORIGINAL TABLE:* Following Behavior Preceeding Land- Behavior Flutter Contact Land Curl End Flutter 39 24 75 43 Contact 12 22 6 32 10 Land 13 7 9 l9 7 SUBTABLE A: Following Behavior Preceeding Land- Behavior Contact Land Curl End Contact 22 6 32 10 Land 7 9 19 7 Chi-Square = 52, df=3 SUBTABLE B: Following Behavior Preceeding Behavior Flutter Not Flutter Contact 12 70 Land 13 42 Chi-Square = 79, df=1 SUBTABLE C: Following Behavior Preceeding Land- Behavior Contact Land Curl End Flutter 39 24 75 43 Not Flutter 29 15 51 17 Chi-Square = 31, df=3 SUBTABLE D:** Following Behavior Preceeding Behavior Flutter Not Flutter Flutter 181 Not Flutter 13 112 * The original table is broken into four subtables and a Chi- Square test is performed on each subtable. ** The subtable containing the impossible transition is eliminated. 143 the original table minus one for the missing cell. The results are shown in Table 20. The overall Chi- Square is 8.84, which is not significant at the .05 level (P=.264, df=7). This implies that the acts performed in an individual "curl sequence" (i.e., before curling) are independent. The sequence approximates a "zeroth-order" Markov chain. Therefore, the sequence is not structured internally, but rather represents the random performance of acts (random in the sense of independent, not in the sense of being equally probable). The probability of a butterfly performing a given act is, therefore, affected only by the internal and external conditions. Assuming the internal motivation of butterflies visiting different target plants is equal, the probability distribution for the various acts is altered only by environmental conditions; specifically, the target plant species. This allows for direct comparisons among sequences performed in response to different target plant species. Visits £9 Different Target Plant Species Figures 18, 19, and 20 show descriptions of the visit sequence to three different target plants: broccoli, tansy, and mustard. The acts in these sequences are quasi- independent in that they depend on the preceding act only because of the forbidden flutter-to-flutter transition. Because of this artificial dependence, the transitional probabilities from a flutter and from a land or contact are depicted separately. The probability of performing a given 144 FLUTTER GIVEN FLUTTER .222 CONTACT (msz LAND on CONTACT '13 LAND u41 - LAND-CURL .24 .13 END 4%; Figure 18. The visit sequence to the broccoli target (comparison I). The size of the boxes is proportional to the relative frequency Of the behavior. The width of the arrows is proportional to the transitional probability between behaviors (given next to the arrow). The relative proportion Of the behavior as the first act in a visit is flown by the arrow in the upper right corner of each x. 14S I .91 FLUTTER .04 GIVEN 09 FLUTTER , 4-=——— .22 .18 CONTACT V), response to alight (V-->l-c), and alight to curl (l-c-->cu)--were calculated for comparisons I and III. The normalized transmissions were multiplied by 100% and presented as the percent communication from plant to butterfly (Figure 22). The percent communication in comparisons I and III was very similar at equivalent stages. About 9% of the unpredictability in the butterflies' behavior was eliminated by knowing the target plant species at the movement to response transition. This compares with an approximately 10% reduction in unpredictability at the response to alight 157 TRANSITIONS BETWEEN STAGES: I SEQUENCE-->VISIT YlSlT-->LAND-CONTACT LAND—CONTACT-->CURL 40— 30‘ 20" 10" PERCENT COMMUNICHTION . . .‘ >._ ........... M10003 COMPHBISON I COMPHBISON III Figure 22. Percent communication from plant to butterfly during different stages of the ovipositional sequence. 158 transition and a 25% to 35% reduction at the alight to curl transition. Two points are worthy of note. First, as the sequence progresses, the degree of communication increases. This is the trend one would logically expect if butterflies are assessing the plant at all stages of the sequence. The amount of information gathered about the plant should increase as the sequence progresses, and therefore, so should the degree to which the butterfly's behavior is influenced by the plant. Secondly, the highest degree of discrimination occurs after alighting. This is in keeping with the idea of the primacy of perception of host-specific contact chemicals in host plant identification. However, it is clear that contact chemicals are not solely responsible for the total amount of discrimination shown between plants, as a good deal of discrimination takes place prior to alighting. Level III: Relative Discrimination at Different Steps 9; the OvipositiOnal Sequence To compare the degree of discrimination occurring at different steps of the ovipositional sequence, the total sequence was divided into a series of six steps representing the following five transitions: 1) movement from the soil space to the plant space (s-->p), 2) movement from the plant space to visit (p-->V), 3) first act exhibited in a visit (flutter, contact or land), 4) behavior after a first 159 flutter (f--> land, contact or end), and 5) behavior after a first alight (l-c --> cu, end or continue). The normalized transmissions for each of these transitions for plant comparisons I and III were calculated and are presented as the percent communication from plant to butterfly (normalized transmission x 100%) (Figure 23). As with the percent communication at different stages (see above), the percent communication at different steps is very similar for comparisons I and III. Percent communication is relatively low at the soil space step (only .9% of the uncertainty in the butterflies' behavior at this stage is eliminated by knowing the target plant species for comparison I and only.7% for comparison III). This low degree of communication probably results from the large number of butterflies entering the target area for reasons other than response to the target plant. During the plant space step about 10% of the uncertainty in the butterflies' behavior is explained by the target plant species. The percent communication occurring at this step is probably a much better estimate of communication occurring in the movement stage, as many irrelevant sequences have been eliminated by butterflies leaving the soil space without entering the plant space (s-->end). Knowing the target plant species reduced the uncertainty about the butterflies' first act in a visit about 3%. Thus, the target plant species determines whether PERCENT COMMUNICATION 160 TRANSITIONS BETWEEN STEPS: K ’ x’x’ SOIL SPACE-->PLANT SPACE PLANT SPACE-->VISIT FIRST BEHAVIOR IN A VISIT FLUTTER-->LAND, CONTACT 0R END LAND-CONTACT-->CONTINUE 0R END 2 6 '- 24— 22“ 2 0 ’— I 6 ‘— I 6 T" I 4 I 2 — ‘ ~ . ,. '\’\ I 0 :\$\ iii-95:13:32 ;\$\ — \ \ . \ \ 6 I I // /\/\ ,\,\ — \ x x x 6 «4 «4 — ’\’\ ’\’\ 4 «A «I. 2 "'_ 3“ §.:;.;5;;. ;\$\ . \ \: - ....... I. \ : COMPARISON I COMPARISON III Figure 23. Percent communication from plant to butterfly during different steps in the ovipositional sequence. 161 butterflies will respond to the plant (visit) more than it determines the first behavior in a response. In contrast with the 3% reduction in uncertainty about the butterfly's first behavior, knowing the target plant species reduces the uncertainty of behavior after a flutter approximately 8 to 15% and about 11 to 25% after a first land or contact. This indicates that a butterfly gains information concerning the nature of the plant during the course of a visit. Slightly more information is gained by alighting than by fluttering, but significant information is gained during the act of fluttering. A butterfly gains more information while fluttering than while approaching the plant. As with the previous section, the amount of information gained by the butterfly increases as the sequence progresses. Degree of Discrimination 'Occurring between Different Plant Types Discrimination between plant species has been shown to exist at all stages and several steps of the ovipositional sequence. In addition, discrimination between different types of plants occurs at single stages and steps of the sequence. At almost all transitions between steps in the sequence the transitional probabilities of hosts differed significantly from those of non-hosts, thus establishing the existance of discrimination between these two plant types. At some steps discrimination between hosts of different 162 acceptabilities occurred. Occasionally, discrimination between different species of non-hosts also occurred. The purpose of the present section is to compare the degree of discrimination occurring at the same stage or step of the ovipositional sequence between the following plant types: hosts vrs non-hosts, hosts of differing acceptabilities and different species of non-host. Because plant comparison III included two species of host (broccoli and mustard) and two species of non-host (lettuce and tansy), data from this comparison were used. The index used to determine relative discrimination between different plant types was percent relative discrimination. This is the percentage of the maximum difference in the butterflies' response that is accounted for by the difference in response toward the plant types in question. It is calculated by the formula: % Relative Discrimination = R1 - R2 x 100% M.D. where R1 = the response to one plant type, R2 = the response to the second plant type, and M.D. = the maximum difference (defined below). The response in the above formula is determined by dividing the number of butterflies advancing to the next stage or step of the ovipositional sequence for the plant type in question by the number of butterflies entering the previous stage or step in the ovipositional sequence for that plant type. 163 The maximum difference is calculated by subtracting the lowest response from the highest response in the comparison for the particular transition in question. The percent relative discrimination between hosts and non-hosts is: R(hosts) -MRénon-hosts) x 100% The percent relative discrimination between hosts is: R(broccoli) - R(mustard) x 100% M.D. And, the percent relative discrimination between non-hosts is: R(lettuce) - R(tansy) x 100% M.D. Level II: Stages 9; Oviposition The percent relative discrimination for hosts vs non- hosts, hosts of differing acceptabilities and different species of non-hosts, for each stage of the ovipositional sequence is shown in Figure 24. During the movement and the alighting stage, relative discrimination is highest between hosts and non-hosts, less between different species of host, and minimal between non-hosts. This indicates that host- specific stimuli are the most important determinants of response during these stages. That host specific stimuli vary in quantity and/or quality (as perceived by the butterflies) is shown by the moderate degree of discrimination occurring between hosts of differing acceptabilities. Little discrimination occurred between PERCENT RELATIVE DISCRIMINATION I00“ 60‘ 60— 40- 20‘ 164 HOSTS vs NON-HOSTS - HOSTS OF DIFFERING ACCEPTABILITIES E DIFFERENT SPECIES or NON-HOSTS [A SEO-->V V-->L-C L-C-->+ (3V) (SL-C) (3+) TRANSITIONS BETWEEN STAGES Figure 24. Percent relative discrimination between hosts and non-hosts, hosts of differing acceptabilities, and different species of non-hosts during different stages of the ovipositional sequence. 165 lettuce and tansy in these stages despite their different morphologies, and the reputed repellent effect of tansy. During the response stage, discrimination between the two non-host species is very high during the response stage, higher than that between hosts and non-hosts. Discrim- ination between hosts is minimal at this stage. This indicates that host specific stimuli is relatively less important in eliciting alighting than are general plant characterisitics, such as visually perceived stimuli (e.g., the lettuce target) and/or the presence or absence of a repellent compound (e.g., the tansy target). Level III: Steps 9; Oviposition The percent relative discrimination for hosts vs non- hosts, hosts of differing acceptabilities, and different species of non-host at each step of the ovipositional sequence is shown in Figure 25. Both steps in the movement stage (s-=>p and p-->v) show the same pattern. The relative degree of discrimination is highest for hosts vs non-hosts, lower between hosts, and lower still between the two non- hosts. As discussed previously, this indicates that host- specific stimuli are the most important factors eliciting an approach to the plant. Differences in the quality and/or quantity of host plant stimuli are relatively less important, and differences in general plant stimuli are of relatively minor importance. The step corresponding to the first behavior in a visit is presented twice, once representing the difference in the I66 .8533 68.5.8626 65 .6 33m 238:6 9:56 98.75: so 33QO E332. 65665233386 octets .3 $8: .38... use: use 38: $253 5:658:86 2,528 23th .mw 959“. CR» :Iv Aaufiv AaZHMv Aaflv Anamflv uo_>¢:um merrczmm a>Rv Anny :UAIIUIA QZUAIIUIA Jail..— azwail... Chm...— hma: >allm mallm 381-202 do 39% Emmmta @ . 3.553.584 53:15 .5 $8: I 281-202 m> 8.8: a ON 0? on 00— NOIIVNIHIIIOSIO JAIIV'IJII INJOUI-Id 167 proportions of first flutters, the second time representing the difference in the proportions of first landings. In both cases, maximal discrimination occurs between hosts and non-hosts. This indicates that host-specific stimuli are more important in determining the first behavior of visiting butterflies than are general plant stimuli or differences in host-specific stimuli. The degree of discrimination among non-hosts is higher than that between hosts for the first behavior in a visit. This difference is large for the proportion of first flutters, and minimal for the proportion of first lands. This indicates that differences in general plant stimuli influence the probability of fluttering more than do quantitative and/or qualitative differences between hosts, but that these two types of stimuli affect the probability of landing about equally. The behavior following an initial flutter is also presented twice, once representing the proportions of flutters that were followed by a butterfly leaving the target area (f-->end); the second time representing the proportions of flutters resulting in a land (f-->l). Host-specific stimuli are very important in determining landing after a flutter, as indicated by the high degree of discrimination occurring between hosts and non-hosts at this point. Host-specific stimuli are much less important in determining the proportion of fluttering butterflies that left the target area. For the flutter to end transition 168 (f-->end), the degree of discrimination is highest among non-hosts. This indicates that differences in general plant characteristics, such as morphological differences and/or the presence or absence or repellent characteristics, are important in determining the proportion of fluttering butterflies that leave the target area. Behavior after the first land or contact is depicted twice, once representing the proportion of alighting butterflies that left the target area (l-c-->end), the second time representing the proportion of alighting butterflies that curled (l-c-->cu). The degree of discrimination, for both proportions, is greatest between hosts and non-hosts, indicating that host-specific stimuli are the most important determinanats of behavior after alighting. Quantitative and/or qualitative differences in host-specific stimuli are moderately important in determining the proportion of alighting butterflies that curled, as indicated by the moderate degree of discrimination occurring between hosts at this point. Such differences apparently most influence the proportion curling. Differences between non-host species apparently have minimal influence on post-alighting behavior. Miscellaneous Results and Discussion Correlation of Number of Visits to Number of Eggs Because behavioral sequences of only a fraction of the butterflies entering the target area during a sampling 169 interval were recorded, a check on the accuracy of the sampling method is needed. Two independent measures of butterfly responsiveness during a sampling interval were recorded: the number of visits to the target plant (using the sequence sampling method) and the total number of eggs laid on the target plant. Correlations between these two independent measures were performed for the broccoli target (comparison I), the mustard target (comparison III), and the broccoli-tansy target (comparison IV). Significant correlation coefficients were obtained for all three target plants (Figures 26, 27, 28). The correlation coefficient was .758 (P (r=0)=.006) for broccoli, 0.856 (P(r=0)=.001) for mustard, and .866 (P(r=0)=.037) for broccoli-tansy. This indicates that the sampling method accurately measured differences in the day- to-day responsiveness of the butterflies. Therefore, it can be assumed that this method also accurately measured differences in g; gapge responsiveness toward different species of target plant. Increase in Butterfly Responsiveness over the Season Because the broccoli and tansy targets were included in most plant comparisons, data were taken on these plants during the entire study period. An examination of the results in Tables 7 through 15 reveal that some transitional frequencies for the broccoli and tansy targets progressively increase from one comparison to the next. For example, the EGGS LAID ON TARGET PLANT PER SAMPLING INTERVAL 170 40“ ' 35" 30- , 25" 20" ° 15- O 0 0 10- ° ° 5‘ ’ . —l. .0 0 IIIIIIIII 013456789101112 VISITS TO TARGET PLANT PER SAMPLING INTERVAL 4i N-TOO Figure 26. Eggs laid per sampling interval vs visits per sampling interval for broccoli (comparison I). r;758. 171 —I c >- or 14.1 p— ! _ . c516 2 :i o. I: 3 “12- ° ° U o a '— o z ¢ 0 —I 9.8- I: . (a G o 4: p. o z 0 c: 4- 0 a '2 -I o m 0 to Q A A “" 0 l I I l l 1 I I I 012 34 5 6 7 8 9 VISITS TO TARGET PLANT PER SAMPLING INTERVAL Figure 27. Eggs laid per sampling interval vs visits per sampling interval for mustard (comparison lll)> r:856 172 25' ° 20“ PER SAMPLING INTERVAL '6 51‘ I I O O EGGS LAID ON TARGET PLANT UT I O OIITIIIIIIIIIFFII 012345678910lll213l4l516 VISITS TO TARGET PLANT PER SAMPLING INTERVAL Figure 28. Eggs laid per sampling interval vs visits per sampling interval for broccoli-tansg (comparison V). r=.866. 173 proportion of butterflies flying into the soil space that flew into the plant space (s-->p) was 0.458 for broccoli in comparison II, 0.514 in comparison III, and 0.591 in comparison V. The same proportions for the tansy target were 0.350 in comparison II, 0.401 in comparison III, 0.436 in comparison IV and 0.496 in comparison V. Because data were taken on successively numbered comparisons progressively later in the season the trend noted above may indicate an increase in the measured responsiveness of butterflies as the study season progressed. To check this, the percentage of butterflies advancing from one step to the next (for the transitions s-->p, p-->v, f-->l and l-->cu) for broccoli in comparison II were compared to the percentages for the same steps for broccoli in comparison V. Comparisons II and V were used because no overlap occurred in sampling dates, comparison II being done between July 21 and August 21 and comparison V being done between September 6 and October 17. This proceedure was repeated for tansy. Percentages for the same step, and the same target plant, but in different comparisons, were tested for significant differences using a test for equality of two percentages (Sokal and Rholf 1969). With the exception of the percentage p-->V for tansy, the proportion of butterflies advancing from any one step to the succeeding one was greater in comparison V than in comparison II at every step for both the broccoli and the tansy target (Table 21). The percentage of butterflies 174 Table 21. Percentage of butterflies making the same transition between behaviors for the same plant (broccoli or tansy) in different comparisons (different sampling dates). COMPARISON COMPARISON TRANSITIONS II V t* P** BROCCOLI: Soil Space to Plant Space 45.8 59.1 -4.27 <.005 (s-->p) Plant Space to Visit 32.2 42.1 -2.44 (.02 (P-->V) Flutter to Land or Contact 66.1 82.2 -2.08 <.05 (f-->l-c) Land or Contact to Curl 82.2 88.6 -1.20 >.05 (l-c-->+) TANSY: Soil Space to Plant Space 35.0 49.6 -4.78 <.005 (Sm->13) Plant Space to Visit 13.7 05.6 2.97 <.Ol (P-->V) Flutter to Land or Contact 25.9 36.4 -6.36 >.05 (f-->l-c) * t B t-statistic, ** P = probability. 175 flying from the outer (soil) space to the inner (plant) space (s-->p) was significantly higher in comparison V than in comparison II for both tansy and broccoli. The percentage of butterflies entering the plant space that visited the plant (p-->V) was significantly higher in comparison V for broccoli and significantly lower in comparison V for tansy. The proportion of fluttering butterflies that alighted was significantly higher in comparison V for the broccoli target, but not for the tansy target. There was no significant difference in the percentage of alighting butterflies that curled on broccoli. Therefore, the idea that the measured responsiveness of butterflies increased as the study season progressed is partially supported by these data. There are two alternate reasons for the measured responsiveness to increase during the course of the study: either a change in the judgments made by the observer or an actual change in the behavior of butterflies. There is evidence supporting the latter explanation. The mean number of eggs laid per sampling interval on broccoli is higher for those comparisons done later in the season (see Table 6, p 78). The mean number of eggs laid per sampling interval on broccoli is 7.13 in comparison II, 11.94 in comparison III, and 12.07 in comparison V. The number of eggs laid on a target plant during a sampling interval was measured independently of the number of visits to the plant, 176 indicating that the observed increase in responsivness was a real phenomenon and not a result of observer bias. There are many possible explanations for the increase in butterfly responsiveness as the season progressed. First, there was a change in the wild vegetation in the cage. During the season the wild plants continued to grow and the vegetation as a whole became more dense. This change may have increased the probability of ovipositing butterflies entering the target area. A second possibile explanation for the increase in responsiveness is unintentional artificial selection. Since the eggs used to continue the 2;.23222 culture were laid by butterflies in the cage, the females most responsive to hosts in the cage left the most progeny. The culture may have become increasingly more adapted to cage life. The butterflies used at the end of the study had been ovipositing in the cage for two to four generations. Whatever the explanation, the difference in the proportion of butterflies advancing from any given step to the next among different target plant species remains relatively constant from one comparison to the next. For example, the difference between broccoli and tansy in the percentage of butterflies flying from the soil space to the plant space was 10.8% for comparison I and 9.5% for comparison V. The difference between broccoli and tansy in the percentage of fluttering butterflies that alighted (f-->l-c) was 40.2% for comparison I and 45.8% for 177 comparison V. Therefore, this increase in responsiveness probably had little effect on results. Ending Behavior Butterflies entered the target area for a variety of reasons, including response to the target plant. While it is virtually impossible to evaluate the motivation of butterflies entering the target area, by observing what they did after leaving the target area it was possible to infer, to some extent, their motivation upon exiting. The general hypothesis tested in this section was that the motivation of butterflies would be affected by their experience with the target plant, and therefore, the ending behavior (i.e., what a butterfly did after leaving the target area) of butterflies that had progressed to different steps of the sequence before exiting would differ. Ending behavior was recorded whenever possible for observed sequences. Descriptions of the ending behaviors can be found at the bottom of Table 3 (p 50). Ending behaviors were grouped into three categories: oviposition, feeding, and "other." Feeding behaviors included behaviors directed toward real or artificial flowers: fluttering around a flower (ff), contacting a flower (cf), and landing on a flower (1f). Ovipositional behaviors included fluttering around a reserve host (fh), contacting a reserve host (ch), contacting wild vegetation (cv), and landing on a host or non-host and exhibiting behaviors associated with 178 oviposition, such as drumming or curling (lh-ov, lv-ov). The ovipositional category is further subdivided into behaviors directed toward hosts (fh, ch, lh-ov) and those directed toward non-hosts (cv, lv-ov). Ending behaviors classified in the "other" category include: flying toward another butterfly (S), flying upward into the air (A), fluttering around non-hosts (fv), and landing on hosts (lh) or on non-hosts (1v), but not exhibiting behaviors associated with oviposition. The total number of ending behaviors in each category was calculated for the broccoli target (comparison I). The specific hypotheses that were tested include: 1. It was hypothesized that the closer a butterfly that was motivated to oviposit was to the target host, the more likely it was to recognize the host and direct ovipositional behavior toward it. Therefore, it was predicted that a lower proportion of those butterflies leaving the target area after entering the plant space (p-->end) would exhibit ending behaviors in the ovipositional category than butterflies leaving the target area after entering the soil space (s-->end). 2. It was hypothesized both that most of the butterflies visiting hosts were motivated to oviposit and that the act curl was a consummatory act, i.e., an act, which, when completed, reduces the internal motivation to perform that act. Therefore, it was predicted that a higher proportion of the butterflies visiting the plant but not 179 curling (V-->end; includes the transitions f-->end and l-c-->end) would exhibit ending behavior in the ovipositional category than either butterflies that left the target area without visiting (s-->end and p-->end) or butterflies that curled. 3. It was hypothesized that the experience of tarsal contact with hosts would increase the ability of butterflies to distinguish between hosts and non-hosts. Therefore it was predicted that, of the butterflies exhibiting ending behavior in the ovipositional category, a larger proportion of those alighting on the target host (l-c-->end and curl) would direct their ending behavior toward hosts, than butterflies that had not alighted. Figure 29 shows the proportion of butterflies exhibiting ending behavior in each of the three ending categories as a function of how far they progressed through the ovipositional sequence. As expected, the proportion of butterflies exhibiting ending behavior in the ovipositional category generally increased as the sequence progressed. The proportion exhibiting ending behavior in the feeding category decreased and the proportion in the "other" category remained relatively constant. A total of 7.4% of the butterflies ending after entering the soil space (s-->end) exhibited ending behavior in the ovipositional category, compared to 3.6% for butterflies ending after entering the inner plant space (p-->end). These two percentages are significantly PERCENTAGE OF BUTTERFLIES 60- 50" 40‘ 30" 20" 10“ 180 LAST TRANSITION IN SEQUENCE: - SOIL SPACE-->END FLUTTER-->END PLANT SPACE-->END LAND-CONTACT-->END - CURL :30 .0}:- 12'0 :39 :2. :2. :1. ” :10 :39 :1. :2. hp ‘ FEEDING OVIPOSITION "OTHER" ENDING ENDING ENDING BEHAVIOR BEHAVIOR BEHAVIOR Figure 29. Percentage of butterflies performing ending behaviors in the feeding, oviposition, and ending categories as a function of how far they progressed through the sequence 181 different (using a test for the equality of two percentages, Sokal and Rholf 1969, t= 2.13, p<.025). This indicates that butterflies closer to hosts are more likely to recognize and respond to them. The proportion of butterflies exhibiting ovipositional ending behavior is higher for butterflies that visit the plant, but do not curl than for butterflies not visiting the plant. However, contrary to expectation, the proportion of butterflies exhibiting ovipositional ending behavior does not decrease significantly after curling has occurred. Therefore, the idea that the curl is a consummatory act, decreasing internal motivation, is not supported by these data. Figure 30 shows the proportion of butterflies exhibiting ovipositional ending behavior directed toward hosts (rather than toward non-hosts) as a function of how far they progressed in the sequence. Although the proportions appear to differ, the samples are small and no two percentages differ significantly. Therefore, the idea that contact with host foliage increases the butterflies' ability to discriminate between hosts and non-hosts is not supported by these data. Total Time As mentioned previously, the sampling method used in this study was to record an approximately equal number of sequences for each target species used on any given sampling 182 soILSPAcE -->END FLUTTER -->END I PLANTSPAcE -->END I LAND-CONTACT-->END E com 70- CD a 60" —I LI. 0: E 50" B u. 40'" D LII < 30 I- E u 20" a: LIJ Q. 10- LAST TRANSITION PERFORMED IN SEQUENCE Figure 30. Percentage of butterflies exhibiting ending behavior classed as ”ovipositional“ that directed their ending behavior toward hosts rather than toward non-hosts as a function of how far they progressed in the ovipositional sequence. 183 date. As a result, the length of the sampling intervals varied. If the length of sampling intervals was, on the average, different for different species of target plant, then a bias might occur that could affect results. To check for this possibility, the sampling interval lengths for the different target plant species in a single comparison were analyzed for significant differences using a two-way ANOVA (randomized block design with days as blocks). While the length of sampling intervals varied significantly among different sampling dates (as would be expected from differences in weather conditions) it did not differ significantly among different target plant species (Table 22). Upon initial consideration, it may seem that sampling intervals should be shorter for hosts than for non- hosts because butterflies would be stimulated to approach hosts and, therefore, more would fly into the target area. However, since a higher proportion of butterflies visiting hosts progressed farther in the sequence, the sequence length was longer and took longer to record. Presumably, these two effects cancelled each other out, resulting in no difference in sampling interval length between hosts and non-hosts. 184 Table 22. Mean length of time of sampling intervals (time taken to record sequences) for different target plant species. Plant Mean Length of Time for Comparison Species Sampling Interval (Minutes) I Broccoli 18.8 Tansy 18.0 F=.901, df=1, P=.350 II Broccoli 19.5 Tansy 18.0 Pot 19.3 Sage 19.5 F=.319, df=3, P=.8ll III Broccoli 16.6 Tansy 16.5 Mustard 15.7 Lettuce 15.2 F=.479, df=3, P=.699 IV Tansy 14.3 Lettuce 14.7 Soybean 16.3 F=.585, df=2, P=.572 V Broccoli 14.3 Tansy 13.8 Broccoli-Tansy 15.5 F=.608, df=2, P=.556 185 GENERAL DISCUSSION am The methods used in this study are well-suited to discrimination studies for several reasons. First, an ideal way to study a behaviorally-based phenomenon such as discrimination, is to study the behavior directly. Results based on indirect data, such as the number of eggs laid, or damage done, while useful for many purposes, may lead to erroneous conclusions about behavior. Jones (1977) gives the following example. Kobayashi (1965), observing the egg distribution of Pieris rapae in Japan, found that egg distribution changed with egg density, tending toward a uniform distribution at high densities and a clumped distribution at lower densities. He suggested the butterflies changed their behavior at high densities. Jones (1977) studied the movement behavior of 2;.EEEEE and found that no such change in behavior either occurs or is necessary to account for the change in egg distribution. Rather, the observed change in distribution with density was a statistical result of two parameters--the probability of laying an egg during a visit and the probability of a repeat visit to a host--which jointly determined the number of eggs a butterfly laid on a host. Secondly, there are distinct advantages to studying behavior as a sequence. Most complex behaviors such as oviposition consist of a temporally-ordered series of 186 separate responses, each elicited by its own set of sensory stimuli. Studying discrimination behavior as a sequence allows for determination of the amount of discrimination occuring during different responses. Since the influence of a specific stimulus on response may vary throughout the sequence, this determination may be important to purposes of the study. For example, Kennedy (1965) points out that attempting to control the spread of a virus spread by an aphid vector by masking an odor stimuli would be futile because the odor works only as an arrestant when the aphid is close to the plant, not as a long range attractant. Only by examining the aphids' behavior as a sequence of separate stages could this determination be made. Thirdly, dividing a sequence of behaviors several different ways may be advantageous. Making multiple divisions of a sequence of behavior is justified in that behavior is, in reality, a continuum. "The stream of behavior may be sliced up many ways" (Drummond 1981). Because 3px division of the sequence is artificial, the validity of the divisions used can only be judged by the understanding they promote. Making multiple divisions is advantageous in that, while finer divisions result in a greater understanding of the behavioral components of the sequence (e.g. the actual acts), coarser division allows discrimination to be judged within the context of the entire sequence (for example, it allows for the assessment of the 187 respective roles of pre-alighting vs post-alighting discrimination). The choice of behavioral units to record is dictated by the desired divisions of the sequence. In this study, the smallest reliably observable behavioral units were chosen. This allowed for a more detailed study of behavior than is possible in most field studies of free-flying butterflies. In addition, the smallest units of behavior were recorded because it is always possible to consolidate behavioral units into higher categories and thus study a more coarsly- divided sequence, but it is not possible to subdivide behaviors already recorded (Lehner 1979). Fourth, the use of butterflies confined to a field cage is ideal for discrimination studies.because it maxmizes the advantages and minimizes the disadvantages of using either butterflies confined to small laboratory cages or wild, free-flying butterflies. While the behavior of butterflies confined to small indoor cages is relatively easy to observe and between treatment variation can be kept to a minimum, problems arise when generalizing results obtained with these severely confined butterflies to the behavior of natural populations. This is, perhaps, especially true when the subject of investigation is discrimination. The degree to which discrimination patterns observed in severely confined colonies of butterflies can be considered natural is questionable. First, the conditions of confinement may prevent the expression of certain behavior 188 patterns. For example, butterflies kept in small cages cannot be expected to exhibit normal host-finding behavior (Papaj and Rausher 1983). Secondly, the habitat of such confined animals is often a gross oversimplification of normal conditions. Thus, such animals are usually deprived of stimulus situations which release normal behaviors. In addition, such caged animals are often intentionally deprived of essential stimuli in order to increase their responsiveness during experiments. A common result of such deprivation, both intentional and unintentional, is a lowering of the threshold of stimuli required to release a given response (Singer 1982, Lorenz 1981). Consequently, deprived animals often exhibit responses toward objects that would not normally release such responses. The net result may be abberent patterns of discrimination. "If insects in small cages cannot exhibit normal host location, neither are they likely to display normal host postalighting behavior. Thus confined, the insects' inability to search normally will lead to irregular rates of encounter with the host plants. This, in turn, will lead to irregular levels of discrimination and this will cause irregular patterns of postalight selectivity". Papaj and Rausher 1983. The field cage used in this study was much larger than small laboratory cages often used for studies of ovipositional behavior and contained a "natural habitat". Thus, the complications discussed above were minimized. Other problems arise when using free-flying butterflies in behavioral studies. The first concerns the observability of behavior. That which cannot be observed cannot be 189 described and recorded. It is often difficult to observe the behavior of insects because of their small size. Detailed observations would require a close proximity to free-flying butterflies which is sometimes difficult to maintain with free-flying butterflies. A second problem lies with the heterogeneity of habitats that a free-flying butterfly is likely to encounter. It has been shown that differences in habitat can affect the behavior of phytophagous insects in general and g; £2222 in particular (Stanton 1983, Root and Kareiva 1984, Cromartie 1975, Smith 1976). Since discrimination in this study was assessed on the basis of differences in behavior toward various plant species, it was crucial that such differences be due to differences among the plants themselves and not due to differences in local habitat. The use of a field cage in this studied allowed the observer to be close to the butterflies and also ensured a heterogenous habitat. In sum, four different aspects of the methods used in the present study are particularly well-suited to discrimination studies in general. These include: direct observation of behavior, observation of ovipositional behavior as a sequence, analysis of the sequence divided several different ways, and the use of butterflies in a field cage. 190 Hypotheses and Results The following discussion summarizes the major hypotheses, predictions and results in the first section of the results as they concern discrimination between: 1) hosts vs non-hosts, 2) hosts of differing acceptabilities, 3) different species of non-hosts, and 4) repellent vs non- repellent plants. Hosts vs Non-Hosts Cabbage butterflies exhibit significant discrimination between hosts and non-hosts during all stages and most steps of the ovipositional sequence (Table 23). In virtually every plant comparison (the sole exception to be discussed later), a higher proportion of butterflies exhibited a response toward hosts than toward non-hosts during each sequence step. This higher degree of responsiveness toward hosts occurred even in comparison III, which included two host species differing greatly in external morphology. These results imply that host-specific stimuli exists toward which cabbage butterflies preferentially respond. These host specific stimuli are perceived at all stages and steps of oviposition, including those steps prior to tarsal contact with a plant. The results support the idea that cabbage butterflies are able to perceive host-specific plant stimuli and, as a result, discriminate between hosts and non-hosts: 1) while at least short distances from hosts plants, 2) during the 190A Table 23: Summary of the hypotheses, predictions, and results from the first section of the results for hosts vs non-hosts. Key for Results: * Significant difference between response toward hosts and toward non-hosts for the indicated comparison. The difference between response toward hosts and toward non-hosts is as predicted from the hypothesis, but the difference is not significant at the .05 level. The differenCe between response toward hosts and toward non-hosts is not as predicted from the hypothesis. 191 .auuonIsoc so coca ammo: so AuOm>oson umumm pawns adv cams ssuouuauaes Haas amusmumouas memuwmm> mo :0muuoaoue nonmms < .mumonicoc ssxu sumo: uouucou no so pace Ham: mowamuouusn wsmummm> mo somuuoeoua nonwmn < .mumoslsos gowns» scan sumo: umwumu umum> flaw: cacao Osman osu Ousm wsfihau ODMHMAOuuDA mo acmuuoaoue scrum: < .umonIsoc a ssnu nozucu anon N am ucoaa Dowuuu 0:3 mm woman ucmae Dru OOCM ham Hams women HMOO can mswuouso OOHHMPOuuDA mo s0muuoaoua scrum: < .ouuonIcoc smcu umo: ummm> Haws scum uowusu osu mamuouso nomamuouusn mo s0muuoeoue nonwm: < HHH HH H OCOumuuaaou mCOmuumvoum HImm .mOmamuouusn wauwmw> no unmromouaaw an mu po>woouoa ops Managua ummmooemluaom .N mI use mewnuoouaao Sn upcoamou hamuouusn < .uscHa Dru scum «consummv uuosm uuuoH as aw hamuouusn Dru Dams? <~ vo>moouoe who «Haamua Omumooantumom .H nsofluoom mononuoazm nmnhaus<. mufismom .mumo: pumaou Saawmusouomoue wsmvsoemou mowfimuouusn Cw muHSmou masawua omonu mo COMOOOOAOA .m>moouoa momamuouusb somsa masamum Omumooealuno: Ousuocom Canaan umox us0mue5566< fioposou .nN CHAOH 192 .auuos ADM Cora uuoonlsos DOM Amuwouw £028 on Hafiz uocusoo no mama oso has so mampco nomamuouusp mo acmuuoeoue may .mumOSIsos Pom saga sumo; Pom Amucouw 50:5 on flaws uncucou no puma oco use so maHusu mom~uuouu=a mo samuuoeoua one .oumontso: so cmnu sumo: so Husu a~w3 mowamuouusn mamunwwfiw «Inn .umum> Dru Ono Cu «Ion hfiwuouusb mamunwwao cm mOOOpCm mfioomsonu uowusoo Ommmooequmo: mo oocombu one .oocoeaou ausu Dru mo mummwofiou ops .owusmes Momma vo>woopoe .Amacomamsu « a s s «o acmuuoeoue scrum: nose < om unsusouv masamum umumuoeulumom .m .numosIcoc so such sumo: co OCOH so uouusoo HAM: Oswamuouusb s s s s mewuouusfiu mo COMOAOOODO poswm: 4 «Inn. > HHH Hg H anamuomvoum ac0muuom mononuoaum uCOnmuseEoo ummafics< uuasaom .8800 AN asses 193 approach and the visit, and 3) after alighting on a plant. In addition, the results on relative discrimination also indicate that the host specific stimuli responsible for eliciting a given response differ, however subtly, for different responses, as the degree of discrimination changes as the sequence progress (Figures 22, p 157 & Figure 23, p 1606). One possible reason for the change in the degree of discrimination is that host-specific contact chemicals, which have been shown to be important to egg deposition (Hovanitz and Chang 1964, Renwick and Radke 1983, Traynier 1984), are perceived only after landing. As supporting evidence, relative discrimination between hosts and non- hosts was greatest during the alighting stage. But contact Chemicals cannot be the total explanation for the change in the degree of discrimination during the sequence as the relative discrimination differs even among pre-alight sequence steps. Discrimination between hosts and non-hosts is incomplete only during the visit stage, specifically during the flutter to land-contact transition. However, butterflies exhibit significant discrimination between hosts and non-hosts during all other transitions in the visit stage. Thus, although host-specific stimuli may not be of prime importance in determining the response of a fluttering butterfly, the evidence indicates that host-specific stimuli are primarily responsible for eliciting all other responses, including other responses in a visit. 194 Hosts of Differing Acceptabilities Despite the fact that broccoli received a significantly higher number of eggs than did mustard (Table 6, p 78), only during the movement stage of the ovipositional sequence was discrimination between mustard and broccoli significant (Table 24). Nevertheless, even during responses where the difference in the proportion of butterflies responding to broccoli and mustard was not significant, the results were generally as predicted from the hypotheses: a lower proportion of butterflies at any one particular step responded to mustard than to broccoli. This was true for all three transitions between successive stages of oviposition and for five out of the seven transitions between steps that were tested for differences in response. In light of this, it is probable that discrimination between the highly acceptable broccoli and the less acceptable mustard occurred throughout the ovipositional sequence, but that the difference in response for any one step was minimal. Minimal differences in the proportion of butterflies responding to broccoli and mustard at several steps in the overall sequence can lead to significant differences in the number of eggs laid on the two plants. The probability that a butterfly, entering the target area, will lay an egg on the plant is the product of the probabilities that the butterfly will continue the sequence at each transition between responses in the sequence. 194A Table 24: Summary of the hypotheses, predictions, and results from the first section of the results for hosts of differing acceptabilities. Key for Results: * Significant difference between response toward broccoli and toward mustard, for the indicated comparison. + The differences between the response toward broccoli and toward mustard is as predicted from the hypothesis, but the difference is not significant at the .05 level. 0 The difference between the response toward broccoli and toward mustard is not as predicted from the hypothesis. 195 .pusuusa so sssu ummmP s sm uos unpwm was as «Hooooun so pssa Hams nomawumuusn o wsmummm> mo sOHuuosoua scrum; < glam .vpsumsa saga «Hooooub unsusoO no so pssa Hama ammampouusn + . wsmummm> mo s0muuosous noswmn < mu .uowumu vusumsa was sung uowusu Mfiooooub Dru uwaw> Ham: mouse usuae Dru wswuouso + nOmaupOuusn mo s0muuosous nocwms < mI one no oususs osu Ousosamsm HHH3 hamumuusn wsmpsoamop a an vo>mouuos «HsEmum ommmoosmlumo; sm moosouomwma .N «Hooooun numa Donna mowusu was Ousw Sam Hams Donna Hmoo as» wswuouso + somamuouusb mo s0muuosous scrum; < Hl Hams sous bowusu onu wswnouso posmwn mo sumo: sw uHsmou mHsEmum % nomamuouusn mo samuuosoue ponwm: < <~ Omumoosqumos sm noosououmma .H HHH saumussaoo usOmuompoum OCOwuoom momosuosmm ODHSOOM mmnhfiss< .>um~wnmusooos sw nousouommmp sw uasaop susu sw soars .momamuouusn an vouwbwnxo oasoauou mo moumop Dru sw uneconommmp sm uHsmou «Haawuu sw noosouommmp 060:9 .OOMHMPOuusn hp po>mouuoa as wasamun ummmomemluaos sm hao>musumassu uo\pss >Ho>muuumusssu nomump mussfis umom «sOHusEsmm< Hmposou .ON manna 196 .maouuoun so ssnu pusunsa so uswmas so>mw ass scams Dssmusoo no ps0 o OOMHMAOuusn mo sowuuosous uoswm: < «Ion .pusumsa so smcu .uow>mson msmuswwaslumos wfiouooun so Huso Haws nomamuouusn moomms flaws MHsEmum uosusoo + wsmustHu mo s0muuosous uoswm; 4 ON ommmoosqumos sm moosoummmmn .m .pswuosa susu «Hooooun unsusoo no so psmH flaws ammamuOuusb + wsmuouusfim mo sOMuuosous scrum; < «Inn HHH soOwussEoo usoMuompoum as0muoom ammonuossm oufisaom ammSHss< .ucoo an essay 197 Consequently, small differences in the probability of responding at these various transitions can result in significant differences in the probability of completing the sequence. For example, small differences in the proportion of butterflies responding to broccoli and to mustard at all sequence steps resulted in a differences in the proportion of butterflies entering the movement stage that progressed through the sequence and ultimately curled (Seq-->cu) of 14.9% vs 8.5% for broccoli and mustard respectively. This percentage is significantly different (using a test for equality of two percentages, Sokal and Rholf 1969, ts = 3.54, P<.05). Therefore, in general, these results support the idea that hosts differ in the quantity or quality of host specific stimuli as perceived by butterflies and that these differences result in differences in the degree of response, which, in turn, result in differences in acceptability. The results suggest that differences in host-specific stimuli definitely affect the responses of the movement stage and probably also affect responses in the visit and alight stages, but the evidence for the latter is less clear cut. One last result requires discussion. Mustard proved to be much more acceptable for resting than broccoli. As yet, little attention has been paid to factors eliciting resting behavior in butterflies, although it has been recognized as a behavior (Vaidya 1969, Root and Kareiva 1984, Stanton 1984). It is interesting that, contrary to the findings of 198 Vaidya (1969) for the lemon citrus butterfly, Papilio demoleus, the optimal stimulus situation for resting and oviposition are not the same in g; rapae. Different Species of Non-Host During most stages and steps of the ovipositional sequence, cabbage butterflies exhibit little, if any, discrimination among various non-host species (Table 25). With one exception, the proportion of butterflies responding at any one step or stage was virtually identical for different non-host species. The exception was behavior toward target non-hosts in comparisons III and IV during the response stage. Significantly more visiting butterflies alighted on lettuce than on any other non-host species (Table 8, p 84). The higher proportion of butterflies alighting on lettuce resulted from differences in the behavior of fluttering butterflies. Significantly more of those butterflies fluttering around lettuce landed or contacted than those fluttering around other non-hosts species (Table 14 pg 109). However, all other responses in the response stage were similar. Although to humans lettuce appears visually similar to the host mustard, the butterflies' response toward the lettuce target was, with the above exception, more similar to that toward other non-host species (tansy and soybean) and quite different from that toward the host mustard. Thus, the idea that lettuce is perceived by butterflies as 198A Table 25: Summary of the hypotheses, predictions, and results from the first section of the results for different non-host species. Key for Results: * Significant difference between response toward lettuce and toward other non-host species, as predicted from the hypothesis, for the indicated comparison. The difference between the response toward lettuce and toward other non-host species is as predicted from the hypothesis, but the difference is not significant at the .05 level. The difference between the response toward lettuce and toward other non-host species is not as predicted from the hypothesis. 199 .uomoosn uaosisos uosuo so sosu .uwnm> s s« you uuumm Dru no .oosuuoH so vssfi HHms numamuouusn + wsmumum> mo somuuoeous poswm: < HImm .mOmoosm umOSIsos Porno sons oosuuoa uuousoo no so vssa Haws momHMPOuusn s s msmumam> mo sOmunosous nonww: < mm .mowoosm uoosIsos nosuo sons oosuuofi uammP Ham: moses usmfis onu wsmuouso + + numamuouusn mo COHuuosous Panam: < mI Hams sous uuwuou osu wswuobso .umm«> Dru mo unsuss Dru ousosamsw hamuouusn wswuwmm> no wswnuoOPOOs so an poPMOoqu «enemas assmm> .N .omsosmou ummw> use sumoussm Dru oososamsm musofis + + nowamuouusn mo samuuosoua scrum: < .H >H HHH ms0muompoum asOMuoom mowosuoshm usOmmussaoo Omahass< unannom .omsosmou u.%~muouusn s oososHmsm massage vo>moouoa Saassnw> .mowamuouusn an oneness anon Dru cu unawamm haaosmm> ms po>moosos mm ousuuoq .onsosmou .aomfimuoDusn 0:6 uoommw AommmoosqumonIsosv masamuo Osman Hauosoo ”usoMumasumd asuosou .mN manna 200 .mOmuosm umonlsos nauoouuma sou saunas no: 3333 newssu so>mw ass scams wswvso use .wswssmusoo os .wsmanso ammfiuuouusp mo sOwuuosous one mIom assessmmwv .aomuosm uaonIsos .uowoosm umoslsos ussomumswwm sooauon nommmu uos flaws wsmausu woman .wswapso mo mumssoHou was who os Ipouusa wswuswmfis mo s0muuosoue 059 UN masumaoso uosusoo ummmuosmlumom .m .mowoosa umosIsos nosuo sass ousuuoH momusoo no so pssH AHA: nomawuouusn « wsmuouusfim mo s0muuosous nonwm: < «Inn >H HHH msoMuOmpoum us0wuoom ammonuoahm osoumuseaoo Omahass< uufisuom .bcoo mu asses 201 similar to the host mustard and that visual stimuli are important in eliciting an approach or a visit is not supported by the results. This is not to say that these results suggest that visual stimuli are unimportant during the movement stage of oviposition, only that visual characteristics that we perceive lettuce as sharing with mustard do not affect butterfly behavior. This suggests that visual characteristics possessed in common by lettuce and hosts are important only in inducing a fluttering butterfly to alight. Lettuce is similar to both mustard and broccoli in having broad, rather flat leaves. Whatever this common characteristic is, other responses in the response stage are not affected. As was expected, no discrimination between non-hosts was shown by butterflies during the alight stage. This is consistent with the idea that host-specific contact chemicals are releasers of post-alighting ovipositional behavior, in particular the curl response (Renwick and Radke 1983, Traynier 1984). Of all alights on non-hosts, only one resulted in a curl and it is interesting to note that this was an alight on the lettuce target. The visit sequence of the individual that curled on lettuce was: lcu c c c c, which indicates the butterfly immediately landed and curled on the lettuce target and followed this behavior by four contacts. The initial act performed in this visit (lcu) was something to be observed only in visits to hosts. In addition, the four contacts subsequently performed were 202 unusual in that the probability of continuing after alighting on non-hosts is low. However, because this was an isolated incident few, if any, conclusions can be drawn. Repellent vs Non-Repellent Plants The results obtained from the two "repellent" target plants, tansy and sage, suggest that these plants, traditionally considered to repel §;_gapae, have no such effect (Table 26). With the exception of the previously discussed difference in behavior of butterflies fluttering in response to lettuce and tansy, no significant discrimination occurs between tansy and other non-host species at any stage or step of the ovipositional sequence. Neither does significant discrimination occur at any step of the sequence between the two "repellent" plants, tansy and sage: such a difference would be expected if these two plants differed in repellent properties. AlthOugh these results do not support the idea that tansy and sage are repellent to P; rapae, neither do they support the idea, suggested by Latheef and Ortiz (1983b), that they are attractive. Rather, the behavior of butterflies toward these alleged repellent plants is similar to their behavior toward other non-host species. Even though the results suggest that tansy, by itself, is not repellent to cabbage butterflies, it is still possible that the presence of tansy near a host plant would reduce the butterflies' degree of responsiveness toward the 203 .oson wHouooun ssnu mussuIMHooooun ummmP Haws Dumas usage 0:6 wsmuouso sumamuouusb mo sOMDLOOOHO nosoa 4 .mowoosm umonIsos nonuo ssnu mussu uwsm> HHms woman Osman was wswuouso OOMHMHOuusn mo s0muuoaoua nozoa < .osoHs umwusquaoououn onu sum: smnu Downs» xmsmulwfiouooun was sums woman usage onu Ousw may flaws woman Hmom onu wsmuouso o ammawuouusn mo s0muuomoua nozofi < .muowusu amonIsos Honuo Laws sssu quumu xmsou onu nuw3 woman uowusu onu Ousm haw Hams moose Hmom Dru wsmuouso mmwamuouusb mo soMuuoaoua unsoa < .oson «Hooooun smnu hassuIMHoOOoub uwmw> Haws sous mowusu or» wswuouso + mmmawumuusn mo COMDAOOOHO ADSOH < .mOmoosm umoslsos nosuo smnu hmsmu mamm> Haws scum mowumu Osu wsmuouso womamuouusb mo sOMOHOQODQ nosoa < Ml pss suoouass osu Ouwbwnsm Manama» Ononu mo soduaouuom .uoosuuump muons unsoH us Scum po>mouuos >H HHH ms0mmuosEoo muasmom msOMOOMpmum asOMuuom mononuosh: mmm%ass< .ussas was pusaou omsoamou .mowawuobusn osu nouop sown; vsm o>moouoa ob Dabs mum unmawuouusb Lums3 «Hoamum usmas usofifioaou moumuosow Sussh us0mumasmw< Houosmo .ON maan 203A Table 26: Summary of the hypotheses, predictions, and results from the first section of the results for repellent vs non-repellent plants. ' Key for Results: * Significant difference between response toward: 1) tansy and other non-host species, or 2) broccoli-tansy and broccoli alone, as predicted from the hypothesis for the indicated comparison. + The difference between the response toward: 1) tansy and other non-host species, or 2) broccoli-tansy and broccoli alone, is as predicted from the hypothesis, but the difference is not significant at the .05 level. 0 The difference between the response toward: 1) tansy and other non-host species, or 2) broccoli—tansy and broccoli alone, is not as predicted from the hypothesis. 204 .osoHs «Hooooss so sass hmsoquHoooosn so unsusoo so used Haws nowHMAOUusb wsmsouusHm mo soMuuosous AOSOH < .momooeu OOOSIsos uosuo so sass hoses so uosusou so pssH aama numamuouusn wsmsouusfim mo sowuuosous uosofi < .osoHs «Hoououn so sass SassuImHoooosn so .umom> s sm sum unswm 0:6 as .psoH Ham3 OOHHMAOuusn mo samuuosous noses < .mowoosa unOSIsos uosuo so suns hassu so .umawP s s« uoo unsww was as .psoH Hama nomamsouusp mo sOmuuosous sosofi < .osoHs «HoOooun suns hmsmuIMHoooosb uosusoo so so ussH Hams mDmampOuusn wsmuwum> mo sOMuuososs soaoa < .mOmooso unosisos sosuo suns mossu uosusoo so so pssa Haws OOMHMPOuusn wsmummm> mo sOmusososa soaoH < mlmm mImn Htmn filmm mm mm .Omsoeoos uswwas osu mo sowumbmssm sw wswuasmos .uwum> Dru mo unsuss osu oososamsm sausouusp wsmummfl> so wsmsosOAAQC so an poPmoosoa .mHsEmuo usuas usoHHosoM .N > >H HHH nsoOmusssoo auasuom usOmuowposm asOMuoom ummxfiss< mouosuoszm .ucou SN asses 205 host simply by interfering with the host signal (stimuli) perceived by the butterflies. The results from the broccoli-tansy combination target are contradictory. In the first place, the acceptability of the broccoli-tansy combination did not differ significantly from that of broccoli. This would suggest that tansy has no inhibitory effect on the ovipositional behavior of butterflies toward broccoli. This is in part confirmed by the fact that in all stages and most of the steps of the ovipositional sequence (the sole exception is discussed below), the proportion of butterflies responding to broccoli-tansy does not differ significantly from the proportion responding to broccoli. However, with the exception of the soil space to plant space transition (s-->p), the results are as predicted from the hypotheses; a lower proportion of butterflies at any given step or stage responded to broccoli-tansy than to broccoli alone. For example, a lower proportion of butterflies visited, alighted on, and curled on broccoli-tansy than on broccoli alone. The first behavior in visits to broccoli- tansy was more like that toward non-hosts (more flutters, less lands) and butterflies were more likely to contact (rather than land) on broccoli-tansy than on broccoli. Even so, the minor differences in the proportion of butterflies responding have little cumulative effect on the overall proportion of butterflies completing the sequence. The percentage of all butterflies entering the target area that curled on the target plant (Seq-->V) was 20.1% for 206 broccoli and 15.4% for broccoli-tansy. These two percentages are not significantly different (by a test for equality of two percentages, Sokal and Rholf 1969, ts- 1.87, P>.05). The only exception to the above discussion was the flutter to land-contact transition (f-->l-c). A significantly lower proportion of fluttering butterflies alighted on broccoli-tansy than on broccoli (Table 14, p 109). This result suggests that tansy perhaps does have a minor deterrent effect on fluttering butterflies, possibly due to an odor. However, the overall effect on the total sequence is negligible. The Sequence Although it has been generally accepted that ovipositional behavior consists of a sequence of responses (Dethier 1953, Beck 1963, 1965, Kennedy 1965), it is not always studied as such. Investigations of ovipositional behavior may focus only on a single response or on the number of eggs laid on the plant. In such cases too much emphasis may be placed on a single stimulus as a determinant of ovipositional behavior. In this study discrimination was assessed within the context of the entire sequence and differences in the degree of response between hosts and non-hosts were found to occur at every sequence step. Discrimination resulting from differences in the degree of response toward two objects at 207 multiple sequence steps is probably quite common. For example, conspecific eggs, which deter oviposition by the large white butterfly (Pieris brassicae), are perceived at several steps of the ovipositional sequence and exert a deterrent effect via a number of different stimuli (Rothschild and Schoonhoven 1977, Behan and Schoonhoven 1978, Klijnstra 1982). During approach and fluttering further response is deterred by visual perception of the yellow eggs on the plant and by olfactory perception of a volatile component of the oviposition deterrent chemical deposited with eggs. After alighting, deterrent chemicals perceived via the tarsi appear to inhibit further response. Finally, the oviposition deterrent pheromone is also perceived through abdomenal chemoreceptors, resulting in a reduction of the number of eggs laid by an ovipositing female. In addition, although not explicitly tested, the results imply that different sensory stimuli are involved in different sequence responses. This is consistent with several experimental results (Saxena and Goyal 1978, Saxena and Saxena 1975, Copp and Davenport 1978). Discrimination patterns involving differential responses at multiple points of the ovipositional sequence and involving information gained from multiple stimulus sources are probably advantageous evolutionarily. Because a sequence involves several responses, an object must possess the right stimulus configuration for each response in order 208 for the sequence to be completed and this makes mistakes in oviposition much less likely. In addition, because discrimination is shown in several of the sequece responses, a butterfly's investment (in terms of time, energy, and vulnerability to predators) will be proportional to the degree to which the object responded to possesses the characteristics of a host. Plants that are obviously or grossly different from hosts will be eliminated early in the sequence. On the average, only those plants that most closely resemble hosts (and are, therefore, most likely to be hosts) will receive a major investment in time and energy. The relative discrimination occurring at each stage or step in the sequence increases as the sequence progresses, indicating that the longer butterflies interact with the plant, the more precise the information they gain about the plant's identity. If the stimuli influencing each response differs somewhat (Kennedy 1965), then the butterflies' behavior is being influenced by an ever increasing number of different stimuli. As Miller and Strickler (1984) have pointed out, the more kinds of stimuli affecting an insects host-finding behavior, the higher the chance that the plant the insect finds will be a suitable host. Pre-Alighting Discrimination The results of this study have shown that ovipositing cabbage butterflies do not land indiscriminately on plants. 209 This is in contrast to the observations of Traynier (1979) whose caged butterflies flew with equal frequency to pieces of cabbage leaves, lettuce leaves and green and yellow index cards. The explanation for this difference in results may be a difference in methodology. Traynier used naive females confined to small cages devoid of anything green (before experiments). Lettuce leaves, cabbage leaves and index cards were presented to these females separately, so their simultaneously expressed, differential responsiveness toward the objects cannot be assessed. Insects in a no-choice situation may appear to exhibit equivalent responsiveness toward different objects when, in reality, their responses would be quite different if the objects were offered simultaneously. For example, Yamamoto et al. (1969) observed that, in the absence of hosts, tobacco hornworm moths (Manduca sexta) approached and oviposited on artificial leaves sprayed with host extract but in the presence of a normal host their response toward the artificial leaves was lessened. In contrast to Traynier's (1979) study, the present study was, in reality, a choice situation in that a wide variety of hosts and non-hosts was always available in the field cage. A second major difference in methodology between Traynier's (1979) study and the present study is that his butterflies were deprived of all plant or plant-like stimuli until the experiment. As discussed previously, deprived animals often experience the phenomenon of threshold 210 lowering resulting in less specificity in the stimuli required to elicit a given response. "It is a general observation, however, that as animals become increasingly deprived, they become decreasingly finicky" (Dethier 1982). Because the butterflies used in this study had vegetation (hosts and non-hosts) available at all times, they were not deprived and therefore were probably more specific in their responses than were Traynier's butterflies. Other experiments using deprived butterflies have led to different conclusions. For example, the observations of David and Gardiner (1962) that caged g; brassicae females "fluttered eagerly" and apparently with equal intensity toward hosts and non-hosts outside of the cage led to the conclusion that olfaction played no role in host-plant finding in this insect. Since then, evidence has been uncovered indicating that host-specific volatiles do indeed influence ovipositional behavior of g; brassicae (Rothschild and Schoonhoven 1977, Behan and Schoonhoven 1978, Mitchell 1978). Discrimination exhibited during the pre-alighting stages of the sequence contributes significantly to the overall discrimination among plants. Although the results on relative discrimination show that the degree of discrimination occuring at the alighting stage was higher than at any other stage of the sequence (Figure 22, p 157: Figure 23, p 160), alighting occurs in the latter part of the sequence and a significant number of butterflies have 211 been eliminated from sequences toward non-hosts (as compared to sequences toward hosts) prior to this. For example, only 22 butterflies alighted on tansy compared to 222 alighting on broccoli (Table 8, p 84). Approximately equal numbers of butterflies initiated sequences toward these two plants (1397 and 1394 for broccoli and tansy, respectively). Pre- alighting discrimination, although seemingly minimal, resulted in a ten-fold reduction in the number of butterflies alighting on tansy. The results of this study have shown that pre-alighting discrimination was greater than post-alightinging discrimination between 2 hosts of different acceptabilities (broccoli and mustard). This agrees with results obtained by Wolfson (1980) who found that the number of §;_£§pae eggs laid on black mustard plants (Brassica nigra) was not correlated with the total sinigrin content of the plant. If chemicals are perceived only after alighting, then chemical differences between plants are irrelvent to pre-alighting behavior during which the most significant discrimination between hosts of different acceptabilities occurs. Traynier (1984) has shown that E; £3222 females do discriminate during pre-alighting behavior between varieties of cabbage differing in outward appearance. A Renwick and Radke (1983) found that host extract concentration affected the number of 2;.23222 eggs received by artificial ovipositional substrates. Initially, these results may appear to be in conflict with those of Wolfson 212 (1980). However, because the artificial ovipositional substrates were presumably equal in appearance, pre- alighting discrimination would not occur in this situation. This leaves only post-alighting discrimination to affect differences in the number of eggs laid. In addition, the fact that Renwick and Radke's experiments were conducted in small laboratory cages, suggests that differences in concentration of host-specific contact chemicals may be more relevent to the distribution of eggs on a single host (which is the result of behavioral responses occuring after initial pre-alight discrimination) rather than to differences in the number of eggs laid on the two plants. Although the purpose of this study was not to investigate the sensory basis of discrimination, the results provide at least a basis for speculation and further study. The results support the ideas that host-specific stimuli are primarily responsible for eliciting both the visit and the approach response (Table 7, p 80 & Table 8, p 84). It is difficult to speculate on what this set of host-specific characterisitics are. Because it is likely to be host- specific, odor is the logical candidate . Although a number of studies have led to the conclusion that odor plays no role in Pieris rapae ovipositional behavior (Hovanitz and Chang 1964, Renwick and Radke 1983), there are problems in methodology that jeopardize that conclusion, namely the use of deprived butterflies, the use of butterflies confined to small laboratory cages, and the failure to record actual 213 ovipositional behaviors. Although this means that an influence of host-specific volatiles on the approach and 'visit response can not be ruled out, it is still unlikely that host odor is entirely responsible for the higher degree of response shown toward hosts, as odor gives little information as to exact location. It is possible that visual stimuli are solely responsible for eliciting the approach and visit. Vision almost certainly plays some role in this behavior, especially the visually perceived aspect of color (Traynier 1984, Miyakawa 1976, Hovanitz and Chang 1964), but to totally account for the higher proportion of approaches and visits to hosts, visual stimuli would have to be fairly host specific, and color is not (Prokopy and Owens 1983). The fact that lettuce and mustard are seen by humans as more similar than either are to broccoli, but that the responses of butterflies to mustard and broccoli are more similar than either are to lettuce may seem, at first, to exclude visual stimuli from being a major factor in eliciting response. However, it is possible that distinguishing characteristics go unnoticed by humans and that mustard and lettuce are not seen as similar by butterflies. Wide variation in appearance among hosts also leads to the inference that visually perceived stimuli are not the major factors eliciting the approach and visit. Broccoli and mustard are distinct in overall shape, leaf shape and color. However, broccoli and mustard may share visual 214 characteristics easily overlooked by humans, but easily perceived by butterflies. Alternatively, individual ngrapag females may vary in their response, some individuals responding preferentially toward broccoli, others responding preferentially toward mustard. Individual variation in the degree of response exhibited toward hosts differing in visually perceived characterisitics has been shown to exist in some butterflies (Rausher 1978, Tabashnik et. a1. 1981, Rausher and Papaj 1983). Individual differences in ovipositional behavior were impossible to detect in this study, as the individuals were not marked. If individual variation in the degree of response toward different hosts occurs in g; gapae, it could be due either to innate differences or to an effect of experience. The host-finding behavior of some other butterfly species is affected by the type of plants it has previously landed on (Stanton 1984, Rausher 1978), thus suggesting a role of experience. Traynier (1979) found that tarsal contact with host foliage increased the tendency of g; gapag to land and oviposit. Host chemicals have an excitatory influence on g; rapae and, in such circumstances, learning may occur. In a later study, Traynier (1984) discovered that g; Eppgg does learn to associate the appearance of an ovipositional substrate with the contact chemical stimulation it provides. Partial discrimination between hosts and non-hosts occurs during the response stage of the ovipositional sequence. The results suggest that host-specific stimuli 215 are important in determining the first act in a visit, but not in determining behavior after an initial flutter (Table 13, p 105: Table 14, p 109). Of the specific stimuli potentially affecting the behavior of fluttering butterflies, odor is one candidate, especially repellent odors. Behan and Schoonhoven (1978) report that the aversion pheromone associated with conspecific eggs in Pieris brassicae is detected while the butterflies flutter over the leaves. In the present study, the significantly lower proportion of fluttering butterflies alighting on the broccoli-tansy target as compared with the broccoli target (Table 14, p 109), may have been due to a repellent odor of tansy. A second possible influence on the behavior of fluttering butterflies is leaf shape. Broccoli, mustard and lettuce all have rather large, wide and simple (undivided) leaves while the leaves of the other non-hosts (soybean, tansy and sage) are not of this type. Experimental findings have suggested that leaf shape influences pre-alight discrimination in several butterfly species (Rausher 1978, Stanton 1982). Observations of Feeny et a1. (1983) suggest that Papilio polyxenes selectively responds to plants possessing finely dissected leaves like those of their umbelliferous hosts. If g; rapae, during pre-alighting ovipositional behavior, responded preferentially toward plants with undivided leaves, a large number of non-host species would be eliminated from consideration. Personal 216 observation of free-flying g; papae suggests that this may be the case. Very few plants possessing compound leaves receive attention. In contrast, butterflies often flutter around non-hosts with broad, simple leaves, such as dandelion or dock. Whatever these general plant characteristics affecting the behavior of fluttering butterflies are, they appear to be important in the final acceptance of a host plant by a butterfly. This is shown in that, once initiated, a visit to a host is more likely to be terminated following a flutter than following a contact or a land (Figure 18, p 144: Figure 20, p 145).. Post-Alighting Discrimination Most of the emphasis has been placed on host-specific contact chemicals as determinants of g; gape; ovipositional behavior. The results of this study suggest that host- specific chemicals have a major influence on post-alighting ovipositional behavior, but that other stimuli are also involved. As results from Traynier (1984) indicate, host- specific contact chemicals are probably the major releasers of curling. In this study, from 77% to 89% of all butterflies alighting on hosts curled, while only one butterfly alighting on a non-host curled. However, the data from butterflies alighting on broccoli-tansy suggest that stimuli other than host-specific contact chemicals also influence post-alighting ovipositional behavior. 217 Traynier (1979) found that contact with host foliage resulted in an increase in the probability of landing and laying. Extrapolating from this, the probability of curling and continuing might be expected to increase as the number of alights increased. However, results from this study do not support this (Figure 30, p 182). Therefore, although initial contact with host foliage has an excitatory effect on g; £2222: the results of this study suggest that after the first alight the internal motivation of the butterflies' remains constant. This indicates that the factors determining the alight on which a butterfly will first curl are extrinsic, belonging to the plant. The probability that a butterfly will alight on a site possessing adequate stimuli to elicit a curl response remains constant throughout a visit. Since internal motivation does not change, the probability that a butterfly will curl on any given alight remains constant. This view is consistent with the finding of Renwick and Radke (1983) that butterflies are able to discriminate between two sides of an index card, one containing cabbage extract, the other lacking that extract. Since butterflies normally do not walk on leaves to any great extent, they apparently find suitable ovipositional sites by repeated alightings. Further Questions Several questions needing research are suggested by the results of this study. One concerns the distances from 218 which cabbage butterflies recognize hosts. These results have shown that cabbage butterflies are stimulated to approach hosts preferentially from at least short distances. The actual distances involved may be considerably greater, but this cannot be ascertained from the data. A number of factors potentially affect the distances from which hosts are recognized. Plant size has already been shown to influence the probability of g; ggpge landing on a plant (Jones 1977). Size probably exerts this effect by increasing the distance over which hosts are seen by g; rapae, as has been shown for the pierid butterfly, Anthocharis cardamines (Courtney 1982). A second factor potentially influencing the distance from which hosts are recognized is neighboring vegetation. A host surrounded by non-hosts may not be as easily detected by Pieris rapae as isolated hosts would be. This has already been shown to occur in the butterfly Battus philenor and its Aristolochia hosts (Rausher 1981). Another question awaiting further research is whether the results of this study are generally applicable to free-flying g; Egpgg. Although the butterflies used in this study were provided with a relatively large and complex habitat, they were, nonetheless, confined. Under natural conditions, 2; gapag females cover approximately 700 meters on daily flights, ending up approximatly 2 km from their starting point during the course of their lifetime (Jones et al. 1980). While butterflies confined to small cages might 219 be expected to exhibit lower degrees of discrimination due to the phenomenon of threshold lowering, those confined to large field cages may exhibit higher degrees of discrimination than free-flying butterflies. This would be likely to occur if butterflies are capable of learning to associate the visual appearance of a plant with contact chemicals perceived after alighting, as has been shown already for g; ggpge (Traynier 1984). Because butterflies confined to a field cage are exposed to limited vegetational diversity, they may learn faster and better than free-flying butterflies. Perhaps the most striking issue remaining to be answered is--precisely which stimuli elicit the responses in the ovipositional sequence, especially responses in the pre- alighting portion of the sequence? The fact that discrimination occurs at every step of the sequence indicates that the optimal stimulus configuration eliciting eagg respOnse is host-specific. A number of different stimuli, potentially influencing ovipositional behavior, but whose role has not yet been investigated have already been mentioned: odor, plant shape, leaf shape, etc. It is some as yet unknown set of perceived stimuli to which butterflies respond. The experimental approach used in this study is ideal for investigating the perceptual basis of discrimination. the advantages have already been stressed: Animals are not deprived and, therefore, are less likely to exhibit abnormal 220 discrimination patterns due to threshold lowering and the effect of a stimulus pattern on acceptability can be traced to its influence on one or more responses in the sequence. In addition, the experimental approach used in this study would be ideal for testing directly the idea that the response-influencing stimuli change over the course of the sequence (Kennedy 1965). In the present method, each response is recorded separately. Thus, it is possible to determine if the influence of a given stimulus changes at different responses, and whether the major stimuli influencing response differ for different responses. The fact that the amount of communication from plant to butterfly increases as the sequence progresses suggests that the number of different stimuli affecting a response also increases as the sequence progresses. Because each response is recorded separately, this could also be tested using the present method. The methods of analysis used in this study, especially Markov chain analysis and Information analysis are useful tools for investigating such questions. The chief advantage of the Markov chain analysis is that it allows comparisons of sequences that involve multiple transition possibilities at each step. For the P; gapae ovipositional sequence described in this study, the number of possible transitions involved are limited up until the visit. This portion of the sequence is somewhat deterministic in that the behaviors possible at any given step depend, in part, on the 221 preceeding behavior. For example, butterflies may visit the plant only if they have entered the plant area (s-->p) and not if they have left the target area (s-->end). The semi- deterministic nature of the sequence up to this point makes it possible to compare sequence steps among different plant types by use of a Chi-Square or G test. At the visit step of the sequence, however, the number of possible behaviors both preceeding and proceeding transitions at any given step increase greatly. In order to compare sequences during this phase of oviposition, one can either focus on a key outcome (such as was done on Levels II and III), or use Markov chain analysis to produce exact descriptions of sequences. Comparisons of Markov chain descriptions of ovipositional sequences in g; gapgg would be valuable to determine: 1) the influence of specific stimuli, or sets of stimuli, on ovipositional behavior, 2) whether inter-individual variation in ovipositional behavior occurs, and 3) whether intra-individual differences in ovipositional behavior occur (e.g., whether learning takes place). In addition, the Markov model of oviposition, once perfected, should predict the number of eggs a plant will receive. Predictive models of g; Eapae_ovipositional behavior which focus mostly on movement behavior have already been created (Jones 1977, Root and Kareiva 1984). These models explain mostly the effects of plant density, distribution, and variety on the abundance and distribution of eggs in a field. The value of the Markov model of the visit sequence 222 proposed above is that it could predict the number of eggs individual plants would receive on the basis of stimulus characteristics possessed by those plants. The use of information analysis, in particular the normalized transmission index, is another potentially valuable tool for investigations of g; £2222 ovipositional behavior specifically, and plant-insect interactions in general. Two aspects of the normalized transmission index make it especially useful in such investigations. First, normalized transmission is a measure of communication. It has been used as a measure of communication between two interacting animals (Dingle 1969, Conanat 1974). As far as I know, it has not been used to measure the influence of a non-animal, such as a plant, on an animal's behavior. However, this measure seems ideally suited to this purpose. The second aspect of the normalized transmission index which makes it a valuble tool for studying insect-plant interactions is that it is a relative measure. This makes it possible to compare directly the influence of different stimuli on the same sequence step, or the influence of the same stimulus on different sequence steps. In summary, this study raises far more questions about the cause of g; £2222 ovipositional behavior than it answers. The chief value of this study, therefore, is that it gives a basis for further research by: 1) providing a description of ovipositional behavior, 2) analyzing how 223 ovipositional behavior varies when directed toward different plant species, and 3) suggesting specific questions and methods of research to answer these questions. Control Implications The results of this study have implications for the control of g; gapae larvae on economically important crops. First, the determination of where discrimination occurs in the ovipositional sequence is important to control. The responses showing maximal discrimination are the "weak links" of the chain for control purposes. Factors contributing to plant resistance via non-preference work by making a plant less acceptable to an insect. This implies that these factors reduce the probability of response for one or more behaviors in the sequence. Discrimination is measured by differences in the degree of response. Where discrimination is minimal, there is little potential to alter the probability of response by manipulating stimuli. Conversely, the presence of maximal discrimination indicates that the probability of response can be lowered by changing the stimuli eliciting that response. The degree to which the probability of response can potentially be lowered is proportional to the degree of discrimination observed. Secondly, the finding that g; gapae discriminates between hosts and non-hosts prior to alighting has important control implications. Plant resistance to g; ggpgg infestation results from non-preference by ovipositing 224 females (Radcliffe and Chapman 1966a, 1966b, Dickson and Eckenrode 1975). Any plant characteristic affecting the probability that a female will lay an egg also affects resistance. While host-specific thioglucosides undoubtedly have a major influence on the behavior of g; gapae females after alightment, increasing resistance by reducing thioglucoside content is not desireable since these chemicals are very important constituents of humanly-desired ‘crucifer flavor (MacLeod 1976). Non-chemical characteristics affecting pre-alighting ovipositional behavior hold more promise for manipulation by breeders. For example, the resistance shown by some strains of red cabbage may be due to a reduced probability of alighting, although this has not been investigated. Discovering and manipulating factors affecting the probability of alighting may hold promise for producing resistant varieties of cruciferous crops. CONCLUSIONS The results of this study indicate that: 1. Virtually all g; gapae ovipositional behaviors are elicited by host-specific stimuli. 2. Quantitative and/or qualitative variation in host- specific stimuli between 2 different host species 225 (broccoli and mustard) primarily affect the movement stage of oviposition. Variation in general plant stimuli (non-host-specific) influences only the behavior of fluttering butterflies. Tansy, a reputed "repellent" plant deters butterfly response only during the flutter step, and then only minimally. Butterflies become increasingly discriminating as the ovipositional sequence progresses. For most stages and steps of the ovipositional sequence, the most important stimuli affecting response are host- specific. General plant stimuli are relatively more important than host-specific stimuli only in determining the behavior of fluttering butterflies. LIST OF REFERENCES LIST OF REFERENCES Beck, S.D. 1963. Behavioral aspects of host plant specificity. Proc. N. Cent. Br. ESA 18:26-27. 1965. Resistance of plants to insects. Annu. Rev. Entomol. 10:207-232. Behan, M; and L.M. Schoonhoven. 1978. Chemoreception of an oviposition deterrent associated with eggs in Pieris brassicae. Ent exp & appl. 24:163-179. Bjorkman, R. 1976. Properties and function of plant myrosinases. In: J.G. Vaughan, A.J. MacLeod and B.M.G. Jones, Eds. The Biology and Chemistry pf the Cruciferae. Academic Press. London. pp 191-205. Bliss, C.I. 1958. The analysis of insect counts as negative binomial distributions. Proc 10th Intnl. Cong. Entomol. 2:1015-1031. Blurton-Jones, N.G. 1968. Observations and experiments on the causation of threat displays of the great tit (Parus major). Anim. Behav. Monogr. 1:75-158. Brewer, G.J. and H.J. Ball 1981. A feeding deterrent effect of a water extract of tansy (Tanacetum vulgare L., Composite) on three lepidopterous larvae. J. Kansas Entomol. Soc. 54:733-736. Brower, J.E. and J.H. Zar 1977. Field and Laboratory Methods for General Ecology. Wm. Brown Co, Dubuque, Iowa. 194pp. Brues, C.T. 1920. The selection of food plants by insects, with special reference to lepidopterous larvae. Am. Nat. 54:313-332. Cane, V.R. 1978. On fitting low-order Markov chains to behaviour sequences. Anim. Behav. 26:332-338. Chatfield, C. and R.E. Lemon 1970. Analysing sequences of behavioural events. J. theor. Biol. 29:427-445. 226 227 Chew, F.S. 1975. Coevolution of pierid butterflies and. their cruciferous foodplants. I. The relative quality of available resources. Oecologia (Berl.) 20:117-127. 1977. Coevolution of pierid butterflies and their cruciferous foodplants. II. The distribution of eggs on potential foodplants. Evolution 31:568-579. Chun, M.W. and L.M. Schoonhoven. 1973. Tarsal contact chemosensory hairs of the large white butterfly Pieris brassicae and their possible role in oviposition behavior. Ent exp & appl. 16:343-357. Copp, N.H. and D. Davenport 1978. Agraulis and Passiflora II. Behavior and sensory modalities. Biol. Bull. 155:113-124. Courtney, S.P. 1982. Coevolution of pierid butterflies and their cruciferous foodplants. IV. Crucifer apparency and Anthocharis cardamines (L.) oviposition. Oecologia (Berl.) 52:258-265. Crane, J. and H. Fleming 1953. Construction and operation of butterfly insectaries in the tropics. Zoologica 38:161-172. . Cranshaw, W.S. 1984. Effect of a hay mulch and of a companionate planting on cabbage pest populations. Great Lakes Entomol. 17:9-12. Cromartie, W.J. 1975. Effect of stand size and vegetational background on the colonization of cruciferous plants by herbivorous insects. J. Appl. Ecol. 12:517-533. David, W.A.L. and B.O.C. Gardiner 1962. Oviposition and the hatching of the eggs of Pieris brassicae (L.) in a laboratory culture. Bull. Entomol. Res. 53:91-109. 1966. Mustard oil glucosides as feeding stimulants for Pieris brassicae larvae in a semi-synthetic diet. Ent. exp & appl. 9:247-255. Dethier, V.G. 1953. Host plant perception in phytophagous insects. Trans 9th Int. Congr. Entomol. 2:81-89. 1969. Whose real world? Am. Zool. 9:241-249. 1976. The importance of stimulus patterns for host- plant recognition and acceptance. In: T. Jermy Ed. The Host Plant i3 Relation £9 Insect Behavior and Reproduction. Akademiai, Budapest. pp 67-70. 228 1982. Mechanism of host-plant recognition. Ent. exp & appl. 31:49-56. Dickson, M.H. and G.J. Eckenrode. 1975. Variation in Brassica oleracea resistance to cabbage looper and imported cabbage worm in the greenhouse and field. J. Econ. Entomol. 68: 757-760. Dingle, H. 1969. A statistical and information analysis of aggressive communication in the mantis shrimp Gonodactylus bredini Manning. Anim. Behav. 17:561-575. Drummond, H. 1981. The nature and description of behavior patterns. In: P.P.G. Klopfer and P.H. Bateson, Eds. Perspectives 13 Ethology. Vol 4. pp 1-31. Ehrlich, P.R. and P.H. Raven. 1964. Butterflies and plants: A study in coevolution. Evolution 18:586-608. Erickson, J.M. and P. Feeny 1974. sinigrin: A chemical barrier to the black swallowtail butterfly, Papilio polyxenes. Ecology 55:103-111. Esau, K. 1977. Anatomy 9; Seed Plants, 2nd Ed. John Wiley & Sons. New York. 550pp. Estabrook, G.F. 1967. An information theory model for character analysis. Taxon 16:86-97. Fagen, R.M. and D.Y. Young. 1978. Temporal patterns of behavior: durations, intervals, latencies and sequences. In: P.W. Colgan, Ed. Quantitative Ethology. John Wiley and Sons, New York. 364pp Feeny, P. 1977. Defensive Ecology of the Cruciferae. Feeny, P., K.L. Paauwe, and N.J. Demong 1970. Flea beetles and mustard oils: Host plant specificity of Phyllotreta cruciferae and P; striolata adults (Coleoptera: Chrysomelidae). Ann. Entomol. Soc. Am. 63:832-841. Feeny, P., L. Rosenberry, and M. Carter 1983. Chemical aspects of oviposition behavior in butterflies. In: S. Ahmad Ed. Herbivorous Insects: Host-seekipg Behavior and Mechanisms. Academic Press. New York. pp 27-76. Finch, S. 1978. Volatile plant chemicals and their effect on host plant finding by the cabbage root fly (Delia brassicae). Ent. exp & appl. 24:150-159. 229 Fox, R.M. 1966. Forelegs of butterflies I. Introduction: Chemoreception. J. Res. Lepid. 5:1-12. Fraser, J. and M.C. Nelson. 1984. Communication in the courtship of the madagascan hissing cockroach I. Normal courship. Anim. Behav. 32:194-203. Gangwere, S.K. 1965. Food selection in the Oedipodinae grasshopper Arphia sulphurea (Fabricius). Amer. Mid. Naturl. 74:67-75. Goodman, L.A. 1968. The analysis of cross-classified data: Independence, quasi-independence, and interactions in contingency tables with or without missing entries. J. Amer. Stat. Ass. 63:1091-1131. Gossard, T.W. and R.E. Jones 1977. The effects of age and weather on egg-laying in Pieris rapae L. J. Appl. Ecol. 14:65-71. Gupta, P.D. and A.J. Thorsteinson. 1960. Food plant relationships of the diamond-back moth (Plutella ,maculipennis (Curt.)). II. Sensory regulation of oviposition of the adult female. Ent. exp & appl. 3:305-314. Harcourt, D.G. 1961. Spatial pattern of the imported cabbageworm, Pieris rapae (L.) (Lepidoptera: Pieridae), on cultivated Cruciferae. Can. Entomol. 98:945-952. 1963. Biology of cabbage caterpillars in eastern Ontario. Proc. Entomol. Soc. Ont. 93:61-75. 1966. Sequential sampling for the imported cabbageworm, Pieris rapae (L.). Can. Entomol. 98:741-746. Harrison, P.K. and R.W. Brubaker 1943. The relative abundance of cabbage caterpillars on cole crops grown under similar conditions. J. Econ. Entomol. 36:589-592. Hemingway, J.S., H.J. Schofield and J.G. Vaughan 1969. Volatile mustard oils of brassica juncea seeds. Nature 192:993. Hovanitz, W. and V.C.S. Chang 1963. Comparison of the selective effect of two mustard oils and their gluco- sides to Pieris larvae. J. Res. Lepid. 2:281-288. 1964. Adult oviposition responses in Pieris rapae. J. Res. Lepid. 3:159-172. 230 Hsiao, T.H. and G. Fraenkel 1968. Selection and specificity of the colorado potato beetle for solanaceous and nonsolanaceous plants. Ann. Entomol. Soc. Amer. 61:493-503. Hunter, B.T. and B. Hines 1971. Gardening without Poisons. Houghton Mifflin, Boston. 318pp. Ilse, D. 1956. Behaviour of butterflies before oviposition. J. Bombay Nat. Hist. Soc. 53:486-488. Ives, P.M. 1978. How discriminating are cabbage butterflies? Aust. J. Ecol. 3:261-276. Jermy, T. 1966. Feeding inhibitors and food preferences in chewing phytophagous insects. Ent. exp. & appl. 9:1-12. Jones, R.E. 1977. Movement patterns and egg distribution in cabbage butterflies. J. Anim. Ecol. 46:195-212. Jones, R.E., N. Gilbert, M. Guppy and V. Nealis 1980. Long distance movement of Pieris rapae. J. Anim. Ecol. 49:629-642. Josefsson, E. 1967. Distribution of thioglucosides in different parts of Brassica plants. Phytochemistry (Oxf.) 6:1617-1627. Kennedy, J.S. 1965. Mechanisms of host plant selection. Ann. appl. biol. 56: 317-322. Kjaer, A. 1963. The distribution of sulphur compounds. In: T. Swain Ed. Chemical Plant Taxonomy. Academic Press. London. pp 453-473. 1976. Glucosinolates in the Cruciferae. In: J.G. Vaughan, A.J. MacLeod and B.M.G. Jones, Eds. The Biology and Chemistry of the Cruciferae. Academic Press. London. pp 207-219. Klijnstra, J.W. 1982. Perception of the oviposition deterrent pheromone in Pieris brassicae. Proc. 5th Int. Symp. Insect-plant relationships, Wageningen 1982. Pudoc, Wageningen. pp 145-151. Kobayashi, S. 1965. Influence of parental density on the distribution pattern of eggs in the common cabbage butterfly, Pieris rapae crucivora. Res. Popul. Ecol. 7:109-117. _____l966. Process generating the distribution pattern of eggs of the common cabbage butterfly, Pieris rapae crucivora. Res. Popul. Ecol. 8:51-61. 231 Latheef, M.A. and R.D. Irwin 1979a. Factors affecting oviposition of Pieris rapae on cabbage. Environ. Entomol. 8:606-609. 1979b. The effects of companionate planting on lepidopteran pests of cabbage. Can. Entomol. 111:863-864. Latheef, M.A. and J.H. Ortiz 1983a. Influence of companion plants on oviposition of imported cabbageworm, Pieris rapae (Lepidoptera: Pieridae), and cabbage looper, Trichoplusia p; (Lepidoptera: Noctuidae), on collard plants. Can. Entomol. 115:1529-1531. 1983b. The influence of companion herbs on egg distribution of the imported cabbageworm, Pieris rapae (Lepidoptera: Pieridae), on collard plants. Can. Entomol. 115:1031-1038. Lawrence, G.H.M. 1951. Taxonomy 9g Vascular Plants. Macmillan Co. New York. 823pp. Lehner, P.N. 1979. Handbook 2; Ethological Methods. Garland STPM Press. New York. 403pp. Lemon, R.E. and C. Chatfield. 1971. Organization of song in cardinals. Anim. Behav. 19:1-17. Lorenz, K.Z. 1981. The Foundations 9; Ethology. Simon and Schuster. New York. 380pp. Losey, G.S. Jr. 1978. Information theory and communication. In: P.W. Colgan Ed. Quantitative Ethology. John Wiley and Sons. New York. pp 44-78. Lundgren, L. 1975. Natural plant chemicals acting as oviposition deterrents on cabbage butterflies (Pieris brassicae (L.), g; rapae (L.), and g; napi (L.)). Zool. Scr. 4:253-258. MacLeod, J.A. 1976. Volatile flavour compounds of the Cruciferae. In: J.G. Vaughan, A.J. MacLeod and B.M.G. Jones Eds. The Biology and Chemistry 9; eye Cruciferae. Academic Press. London. pp 307-330. Maguire, L.A. 1983. Influence of collard patch size on population densities of lepidopteran pests (Lepidoptera: Pieridae, Plutellidae). Environ. Entomol. 12:1415-1419. 1984a. Choosing host plant patch sizes for experiments on Pieris rapae (L.) populations (Lepidoptera: Pieridae). Environ. Entomol. 13: 459—463. 232 1984b. Influence of surrounding plants on densities of Pieris rapae (L.) eggs and larvae (Lepidoptera: Pieridae) on collards. Environ. Entomol. 13:464-468. Mathews, D.L., P.S. Michalak and R.J. MacRae 1983. Effect of traditional insect-repellent plants on insect numbers in a mixed planting system. Proc 4th Intern. Federation of Organic Agriculture Movements (IFOAM). Boston Mass. 1981. pp 117-177. McKillip, B.B. (Ed.) 1973. Getting the Bugs out 9; Organic Gardening. Rodale Press. Emmaus Pennsylvania. 115pp. Miller, J.R. and K.L. Strickler 1984. Finding and accepting host plants. In: W.J. Bell and R.J. Carde, Eds. Chemical Ecology e; Insects. Chapman Hall Ltd. pp 127-157. Mitchell, N.D. 1978. Differential host selection by Pieris brassicae L. (the large white butterfly) of Brassica oleraceae L. subs. oleraceae (the wild cabbage). Ent exp.& appl. 22:208-219. Miyakawa, M. 1976. Flower-visiting behavior of small white butterfly, Pieris rapae crucivora. Ann. Zool. Jap. 4:261-273. Moss, J.E. 1933. The natural control of the cabbage caterpillars, Pieris spp.. J. Anim. Ecol. 2:210-231. Munakata, K. 1977. Insect feeding deterrents in plants. In: H.H. Shores and J.J. McKelvey Jr. Eds. Chemical Control 9; Insect Behavior: Theory and AppliCation. John Wiley & Sons. New York. pp 93-102 Nair, K.S.S., F.L. McEwen and V. Snieckus 1976. The relationship between glucosinolate content of cruciferous plants and oviposition preferences of Hylema brassicae (Diptera: Anthomyiidae). Can. Entomol. 108:1031-1036. Nieuwhof, M. 1969. Cole Crops. Leonard Hill Books. London. 353pp. Obara, Y. 1970. Studies on the mating behavior of the white cabbage butterfly, Pieris rapae crucivora Boisduval. III. Near-ultra-violet reflection as the signal of intraspecific communication. 2. vgl. Physiol. 69:99-116. 233 Obara, Y. and T. Hidaka 1968. Recognition of the female by the male, on the basis of ultra-violet reflection, in the white cabbage butterfly, Pieris rapae crucivora Boisduval. Proc. Jpn. Acad. 44:829-832. Oden, N. 1977. Partitioning dependence in non-stationary behavior sequences. In: B.A. Hazlett Ed. Quantitative Methods ie the Study 9; Animal Behavior. Academic Press. New York. pp203-220. Opler, P.A. and 6.0. Krizek 1984. Butterflies East 23 the Great Plains. Ag Illustrated Natural History. John Hopkins Univ. Press. Baltimore. 294 pp. Papaj, D.R. and M.D. Rausher 1983. Individual variation in host location by phytophagous insects. In: S. Ahmad Ed. Herbivorous Insects: Host-seeking Behavior and Mechanisms. Academic Press. New York. pp 77-124.- Prokopy, R.J. and E. D. Owens 1983. Visual detection of plants by herbivorous insects. Annu. Rev. Entomol. 28:337-364. Radcliffe, E.B. and R.K. Chapman 1965. The relative resistance to insect attack of three cabbage varieties at different stages of plant maturity. Annl. Entomol. Soc. Amer. 58:897-902. 1966a. Varietal resistance to insect attack in various cruciferous crops. J. Econ. Entomol. 59:120-125. 1966b. Plant resistance to insect attack in commercial cabbage varieties. J. Econ. Entomol. 59:116-120. Rahman, M; 1968. Oviposition by Pieris rapae L. in the insectory. Pak. J. Sci. 20:278-280. Rausher, M.D. 1978. Search image for leaf shape in a butterfly. Science 200:1071-1073. 1981. The effect of native vegetation on the susceptibility of Aristolochia reticulata (Aristolochiaceae) to herbivore attack. Ecology 62:1187-1195. Rausher, M.D. and D.R. Papaj 1983. Host plant selection by Battus philenor butterflies: Evidence for individual differences in foraging behaviour. Anim. Behav. 31:341-347. Renwick, J.A.A. and C.D. Radke 1983. Chemical recognition of host plants for oviposition by the cabbage butterfly, Pieris rapae (Lepidoptera: Pieridae). Environ. Ent. 12: 446-450. 234 Richards, O.W. 1940. The biology of the small white butterfly (Pieris rapae), with special reference to the factors controlling its abundance. J. Anim. Ecol. 9:243-288. Rodman, J.E. and F.S. Chew 1980. Phytochemical correlates of herbivory in a community of native and naturalized Cruciferae. Biochem. System. Ecol. 8:43-50. Root, R.B. 1973. Organization of a plant-arthropod association in simple and diverse habitats: The fauna of collards (Brassica oleracea). Ecol. Monographs 43:95-120. Root, R.B. and P.M. Kareiva 1984. The search for resources by cabbage butterflies (Pieris rapae): Ecological consequences and adaptive significance of Markovian movements in a patchy environment. Ecology 65:147-165. Rothschild, M. and L.M. Schoonhoven 1977. Assessment of egg load by Pieris brassicae (Lepidoptera: Pieridae). Nature 266:352-355. Rothschild, M. and J.W. Fairbairn 1980. Ovipositing butterfly (Pieris brassicae) distinguishes between aqueous extracts of two strains of Cannabis sativa L. and THC and CBI. Nature 286: 56-59. Rumsey, W.E. and F.E. Brooks 1909. Cabbageworms and suggestions for destroying them. W. Virginia Agri. Exp St Bulletin 120. Saxena, K.N. and S. Goyal 1978. Host-plant relations of the citrus butterfly Papilio demoleus L: Orientational and ovipositional responses. Ent. exp. & appl. 24:1-10. Saxena, K.N. and R.C. Saxena 1975. Patterns of relationships between certain leafhoppers and plants. Part IV. Sequence of stimuli determining arrival on a plant. Ent exp & appl. 18:207-212. Scheffler, W.C. 1980. Statistics for the Biological Sciences. Addison-Wesley Co. Mass. 276 pp. Schoonhoven, L.M. 1977. Insect chemosensory responses to plant and animal hosts. In: H.H. Shorey and J.J. . McKelvey Jr. Eds. Chemical Control 2; Insect BehaVIOr: Theory and Application. John Wiley and Sons, New York. pp7-l4. 235 Scudder, S.H. 1887. The introduction and spread of Pieris rapae in North America, 1860-1886. Mem. Boston Soc. Shannon, C.E. and W. Weaver. 1969. The Mathematical Theory 9; Communication. Univ. of Illinois Press, Urbana. Shapiro, A.M. 1974. Host-plant utilization by Pieris napi populations in California (Lepidoptera: Pieridae). Psyche 81:361-366. 1975. Ecological and behavioral aspects of coexistence in six crucifer-feeding pierid butterflies in the Central Sierra Nevadas. Amer. Midl. Natur. 93:424-433. Singer, M.C. 1982. Quantification of host preference by manipulation of oviposition behavior in the butterfly Euphydryas editha. Oecologia (Berl.) 52:224-229. Slater, P.J.B. 1973. Describing sequences of behavior. In: P.P.G. Bateson and P.H. Klopfer Eds. Perspectives lg Ethology. Plenum Press. New York. pp131-152. Smith, J.G. 1976. Influence of crop background on aphids and other phytophagous insects on brussels sprouts. Ann. appl. Biol. 83:1-13. Sokal, R.R. and F.J. Rholf 1969. Biometry W.H. Freeman and Co., San Francisco. 776pp. Solereder, H. 1908. Systematic Anatomy 2; the Dicotyledons, vol l. Clarendon Press, Oxford. 644pp. Stanton, M.L. 1982. Searching in a patchy environment: Foodplant selection by Colias p; eriphyle butterflies. Ecology 63:839-853. 1983. Spatial patterns in the plant community and their effects on insect search. In: S. Ahmad Ed. Herbivorous Insects: Host-seeking Behavior and Mechanisms. Academic Press. New York. pp 125-157. 1984. Short-term learning and the searching accuracy of egg-laying butterflies. Anim. Behav. 32:33-40. Steinberg, J.B. 1977. Information theory as an ethological tool. In: B.A. Hazlett Ed. Quantitative Methods ip the Study 9; Animal Behavior. Academic Press. New York. pp47-74. Steinberg, J.B. and R.C. Conant 1974. An informational analysis of the inter-male behaviour of the grasshopper Chortophaga viridifasciata. Anim. Behav. 22:617-627. 236 Stevens, W.C. 1924. Plant Anatomy. P. Blakestons and Sons. Philadelphia. 398pp. Stone, J.L.S. and H.J. Midwinter 1975. Butterfly Culture: 5 Guide Ee Breeding Butterfliesy Moths and Other Insects. Blanford Press. Dorset. 104pp. Tabashnik, B.B., H. Wheelock, J.D. Rainbolt, and W.B. Watt 1981. Individual variation in oviposition preference in the butterfly, Colias eurytheme. Oecologia (Berl.) 50: 225-230. Thorsteinson, A.J. 1952. The chemotactic responses that determine host specificity in an oligophagous insect (Plutella maculipennis (Curt.) Lepidoptera). Can. J. Zool. 31:52-72. 1958. Acceptability of plants for phytophagous insects. Proc. 10th Intnl. Congr. Entomol. 2 1956. pp 599-601. 1960. Host selection in phytopagous insects. Annu. Rev. Entomol. 5:193-218. Traynier, R.M.M. 1979. Long-term changes in the oviposition behaviour of the cabbage butterfly, Pieris rapae, induced by contact with plants. Physiol. Entomol. 4:87-96. 1984. Associative learning in the ovipositional behaviour of the cabbage butterfly, Pieris rapae. Physiol. Entomol. 9:465—472. Twinn, C.R. 1924. Studies in the life history, bionomics and control of the cabbage worm in Ontario. 54th Ann. Rep. Entomol. Soc. Ont. 1923:82-86. van Emden, H.F. 1972. Aphids as phytochemists. In: J.B. Harborne, Ed. Phytochemical Ecology. Academic Press, London. pp 25-43. van Emden, H.F. and M.A. Bashford. 1969. A comparison of the reproduction of Brevicoryne brassicae and Myzus persicae in relation to soluble nitrogen concentration and leaf age (leaf position) in the brussels sprout plant. Ent. exp. & appl. 12:351-364. von Uexkull, J. 1957. A stroll though the world of animals and men: A picture book of invisible worlds. In: C.H. Schiller, Ed. Instinctive Behavior. International Universities Press, Inc. 328pp. 237 Vaidya, V.G. 1969. Investigations on the role of visual stimuli in the egg-laying and resting behaviour of Papilio demoleus L. (Papilionidae, Lepidoptera). Anim. Behav. 17:350-355. Verschaffelt, E. 1911. The cause determining the selection of food in some herbivorous insects. Proc. Acad. Sci. Amsterdam 13:536-542. Wensler, R.J.D. 1962. Mode of host selection by an aphid. Nature 195:830-831. Wolfson, J.L. 1980. Oviposition response of Pieris rapae to environmentally induced variation in Brassica nigra. Ent. exp. & appl. 27:223-232. Yamamoto, R.T., R.Y. Jenkins, and R,K. McClusky 1969. Factors determing the selection of plants for oviposition by the tobacco hornworm Maduca sexta. Ent. exp. & appl. 12:504-508. APPENDICES APPENDIX A Sampling dates during which different plant comparisons were tested. 238 Appendix A Sampling dates during which different plant comparisons were tested. Plant Comparison Number Date I II III IV July 12 July 13 July 19 July 21 July 22 July 23 July 24 July 25 July 28 July 30 July 31 August 1 August 2 August 5 August 9 August 10 August 13 August 14 August 15 August 16 August 17 August 20 August 21 August 25’ August 26 August 28 August 30 September 2 September 6 September 8 September 10 September 11 September 17 September 18 September 19 September 20 September 21 September 22 October 2 October 3 October 12 October 17 NXXXNXX NNNNNXXXNX >4 XXXNNXXNNXXXXXNNXXXXXMNNXXXNNXNNNNXNNNXXNK XXX N xxxxxxxxxxxxxxx XXX >4 XNNNNN XNXXNNNXNXNNXN APPENDIX B Number of butterflies entering each stage of the ovipositional sequence for the plant species in each comparison. The width of each box is proportional to the number of butterflies entering the stage (given in or near the approprate box). The width of each arrow is proportional to the transitional probability (given next to each arrow). The total area of each box is not significant. 239 Appendix B SEQUENCES 1394 1.96 .04 VISIT 49 .45 END 1394 '°°° ALIGHT 22 I COMPARISON 1 : BROCCOLI SEQUENCES 1397 I... END I206 COMPARISON I: TANSY 240 Appendix B (cont) i «masses 1... ENC) 550 .15 .2: vuuT ""‘ oz .19 .18 ALIGHT ' To .52 CURL so COMPARISON II :BROCCOLI SEOUENCES 612 I... END 012 COMPARISON II : .os vunT so .33 ALIGHT Io TANSY sEDUENCEs 01: ('L" .pa .5: VHHT 34 .47 END '90 AUGHT 011 To COMPARISON II : SAGE Appendix B (cont) SEOUENCES 621 END 821 COMPARISON III : TANSY SEOUENCES 823 .87 END 570 COMPARISON III : MUSTARD .02 VISIT 13 .46 1.00 ALIGHT .13 V48” 79 .87 AUGHT 69 J7 CURL 53 241 SEOUENCES 823 ('LBI .Is 141‘ vwsn’ - 118 .as END 11 ALIGHT 530 105 .89 CURL 93 COMPARISON III 2 BROCCOLI SEOUENCES 622 .'L96 .04 quT 23 L00 .96 END ALégHT 621 ".04 CURL 1 COMPARISON III : LETTUCE 242 Appendix B (cont) SEOUENCES 331 .98 .02 VISIT 7 .71 END ALISHT 331 COMPARISON IV :TANSY SEQUENCE 333 =4 ,'I95 .05 Visn 15 L00 END 1.0 AUGHT 333 15 COMPARISON IV :LETTUCE SEOUENCES 330 .'I98 .02 .33 VHHT 9 .67 END 1.0 ALIGHT 330 a COMPARISON IV : SOYBEAN Appendix B (cont) 243 SEQUENCES 403 1." END 370 .11 VISIT 118 ALIGH'I’ 105 CURL 93 COMPARISON V :BROCCOLI sEOUENces no [or .03 .aa vssn 13 .02 END ‘-°° ALIGHT 460 a COMPARISON V : TANSY SEQUENCES 480 '.76 .24 .10 VISIT .8! END .20 ALIGHT 389 89 1.00 CURL 71 COMPARISON V : BROCCOLI-TANSY APPENDIX C Number of butterflies performing the behaviors in the movement stage of oviposition for the plant species in each comparison. The width of each box is proportional to the number of butterflies performing the behavior (given in or near the appropriate box). The width of each arrow is proportional to the transitional probability (given next to each arrow). Appendix C 244 Enter Soil Space 1362 A9 35 .511 .025 Enter Plant Space 717 .489 .637 .363 Visit End 1137 260 COMPARISON I : BROCCOLI Enter Soil Space 1376 Ap 18 .392 .013 Enter Plant Space 557 .608 .912 .088 End 1345 Visit 49 COMPARISON l : TANSY Appendix C (cont) 245 Enter Soil Space 508 Ap 12 .458 .02 Enter Plant Space 288 .542 Visit End 518 92 COMPARISON II : BROCCOLI Enter Sell Space 808 An 2 .398 .003 Enter Plant Space 241 End 594 COMPARISON II : POT .07 VIsit 18 Enter Sell Space 605 .01 Enter Plant Space 218 End 582 Vlait SO COMPARISON II : TANSY Enter Sell Space eoo .08 Enter Plant Space 229 .837 . 1 5 End 577 Visit 34 COMPARISON II : SAGE Appendix C (cont) Enter Sell Space 818 .480 Enter Plant Space 325 End 505 Vlalt 110 COMPARISON III : BROCCOLI Enter Sol Space 812 .528 Ap 11 .472 .010 Enter Plant Space 300 End 544 COMPARISON III : MUSTARD 246 Enter Soil Space 814 Enter Plant Space 258 .598 End 808 COMPARISON III : TANSY Enter Soil Space 818 Enter Plant Space 244 .814 End 590 COMPARISON III : LETTUCE ‘ 247 Appendix C (cont) Enter Soil Space 330 .003 Enter Plant Space 145 .584 End 324 COMPARISON IV : TANSY .05 Visit 7 Enter Sell Space 829 .413 .012 Enter Plant Space 140 .587 End 318 Visit 15 COMPARISON IV :LETTUCE Enter Soil Space 328 .454 .012 Enter Plant Space 152 .546 .88 .06 End 321 Visit 9 COMPARISON IV : SOYBEAN Appendix C (cont) 5 24? Enter Sell Space 447 Enter Plant Space 280 .408 .578 .421 1 r Vialt End 345 "a COMPARISON V : BROCCOLI Enter Soil Space 454 Ap8 .488 .013 Enter Plant Space 281 Lu. End 447 COMPARISON V : TANSY Enter Soil Space 448 An 14 05“ Enter Plant Space 279 .408 .808 .394 i r Visit End 350 110 COMPARISON V : BROCCOLI-TANSY APPENDIX D Number of butterflies fluttering, contacting, landing, and ending as the first two behaviors in visits to the plant species in each comparison. The width of each box is proportional to the number of butterflies performing the behavior (given in or near the appropriate box). The width of each arrow is proportional to the transitional probability (given next to each arrow). Appendix D 249 Total Visit 260 1- First Flutters 162 .23 t... and 37 1.20 '.09 t_.. 9 33 l l 13L contact .56 24 .28 lst. f-' 91 land 74 _COMPARISON I:BROCCOLI Total Visits 49 '— First Flutters 42 .21 .08 v t... c 9 .4 ‘ 'T lst c 4 t... and 27 14 ' t...| 6 .J lst. l COMPARISON IzTANSY .06 3 250 Appendix D (cont) Total Vlaita 92 1... First Flutters 25 .21 .12 V I- 131 c 12c 11 .34 v t... .27 end 19 J, .45 "" 1at125 25 COMPARISON II : BROCCOLI Total Vinita 30 .04 .80 First Fluttera 27 .15 7 1.. 4 I -7‘ lat c l T-. .nd 20 .07 .n .. —'l 3 “Hi "' ""d 2 [ Total Visits 34 COMPARISON II : TANSY 1: First Flutters 27 .28 .09 I—.c '57 lat c 3 t—end 18 r. .07 1" I12 l—l 2 COMPARISON II : SAGE 251 Appendix D (cont). Total VIaIta 1 18 Total Vlaita 13 1.. 1... [First Flutter'a 88 Plrat Fluttera 11 l .27 .01 t— 8 3 I -“ 1at c 1 1— 0nd 7 if“ f l roe ' n ma ‘0 1atlanda a .09- l—t. 1v— 1atl 1 COMPARISON m : BROCCOLI COMPARISON Ill : TANSY Total Visits 19 "I“ "'"' ’3 l.48 .70 Firet Flutters 38 First “an.“ ‘0 ' .88 .17 .20 $.13 n lat _. let on 1 10 _ t " c 14 act 4 .20 ‘ P I— end '0 .47 :39 $.13 LT. 11ldS1 .‘2' r; 23 a an _ I... t- I 2 COMPARISON III : MUSTARD COMPARISON III: LETTUCE Appendix D (cont) 252 Total Visits 118 1.. First Flutters 73 ,15 .03 1...“: C 10 1 1 .18 I "d .87 ,, U.30 t_.l 49 'lst land 35 Total Visits 13 COMPARISON V : BROCCOLI .85 First Flutters 11 1.38 .08 v 1.. c 4 l "5 151 c 1 t_. and 5 $.08 18 t—l 2 ‘1 COMPARISON V : TANSY Total Visits 110 .73 First Flutters 80 1.30 .05 t_.c 27 26 18108 21 .nd .40 .22 I 32 land 24 COMPARISON V : BROCCOLI-TANSY 'lICHIGAN smrE UNIV, LIBRARIES ilHlilmlmIII!“Iililiiliilllilllllililillliliililllliiiil 31293006759934