FORAGING ECOLOGY OF A NEOTROPICAL FOLIVORE, LAMPONIUS PORTORICENSIS REHN (PHASMATODEA:PHASMATIDAE) by ELIZABETH SANDLIN SMITH, B.S. A THESIS IN BIOLOGY

Submitted to the Graduate Faculty of Texas Tech University in Partial Fulfillment of the Requirements for the Degree of MASTER OF SCIENCE

Approved

Accepted

December, 1989 T3 I tffCj }J4' jlj../, c"/' :J

ACKNOWLEOOEMENTS

My gratitude extends in many directions and to many special friends. Foremost, I

thank Dr. Michael R. Willig for his support, guidance, patience, and nurturing hand

throughout my graduate training. I also thank Dr. John C. Zak and Dr. M. Kent Rylander

for their encouragement and input in the development and implementation of my research.

Next, I wish to thank the Department of Energy and Oak Ridge Associated Universities for

awarding me a summer Research Fellowship, and I am grateful to Dr. Robert B. Waide of

the Center for Energy and Environment Research in Puerto Rico for his financial assistance

and provision for use of equipment and facilities at El Verde. Additionally, I wish to

acknowledge and thank Sr. Alejo Estrada for his assistance in establishing my laboratory

and work area and also for lending his exhaustive knowledge of the flora and fauna within

the Tabonuco Rainforest. My deep thanks for friendship, advice, ideas, and assistance in

the field and elsewhere go to Deborah J. Kyrouac, Michael R. Gannon, Gerardo R.

Camilo, David R. Ficklen, Robert Huber, Moira J. van Staaden, and Javier Alvarez. I am

indebted to my parents for their tireless support and their avid interest in my education.

Finally, and most importantly, I wish to express my heartfelt gratitude to my husband,

Hunt, who unselfishly and patiently provided the support I needed to see this project to

fruition.

11 TABLE OF CONTENTS

ACKNOWLEDGEMENTS...... n

ABSTRACT...... v

LIST OFTABLES ...... vii

LIST OF FIGURES ...... viii

CHAPTER I. INTRODUCTION AND PRESENTATION OF THE PROBLEMS...... 1 Optimal Foraging Theory ...... 1 Modifications of Classical OFT...... 2 Herbivores...... 2

Factors Affecting Food Choice...... 3

Toxins and Nutrients...... 3 Position...... 4 Prior Experience...... 4

Age ...... 4 Phasmatids ...... 5 The Problems ...... 6

CHAPTER II. MATERIALS AND METHODS...... 9

Study Site...... 9

Collection and Maintenance ...... 9

Experimental Plants ...... 10

Leaf Selection ...... 11

111 Feeding Trials ...... 11

Age- and Sex-Specific Variation ...... 12

Variation Due to Preexposure...... 13

Intraspecific Food Variation ...... 13

Data Analyses ...... 14

CHAPTER III. RESULTS ...... 18

Effects of Age- and Sex-Specific Variation ...... 18

Effects of Preexposure to One Food...... 19

Effects of Intraspecific Variation in Food Quality ...... 21

CHAPTER IV. DISCUSSION...... 31

Herbivory...... 31

Lamponius portoricensis ...... 32

Special Problems Faced By Females ...... 33 Implications of Observed Patterns of Consumption ...... 34

Influence of Age and Sex on Diet Composition ...... 34

Influence of Previous Experience on Diet Composition ...... 35

Food Quality as it Relates to Succession ...... 35

Effects of Food Variation Within a Forage Species ...... 37

Monophagy, Polyphagy, and Plant Successional Status...... 38

For the Future...... 40

LITERATURE CITED ...... 42

APPENDIX ...... 47

IV ABSTRACT

Recent attention in ecology has focused on responses of consumers to a variety of factors which influence foraging behavior. Herein, I evaluate the responses to different choices of food plants exhibited by an herbivore which is abundant within the Tabonuco

Rainforest of Puerto Rico. Previous work indicates that the walking stick, Lamponius

portoricensis. appears to forage on a limited array of plant species and selects habitats

which contain high densities of Piper treleaseanum. However, in food choice

experiments,£,. treleaseanum is its least preferred food. In an attempt to delineate factors

contributing to this apparent dilemma, I designed three separate experiments to evaluate

( 1) if walking sticks of different ages or of different sex have different food preferences,

(2) if previous exposure to only one food type affects subsequent diet composition, and

(3) if walking sticks distinguish between leaves of varying quality from the same plant .

Four plants known to be forage for this insect (Dendropanax arboreus, Piper hispidum, £,.

treleaseanum, and~ baccifera) were used in food choice experiments. Multivariate

analyses reveal that, at different ages, males and females exhibit different patterns of consumption. Likewise, preexposure to only one food influences subsequent diet differently depending upon preexposure regime and sex. In addition, preferences are shown for different qualities of leaves within single forage species. In particular, lower leaves of P. treleaseanum are preferred, while leaves of D. arboreus and .U.. baccifera are eaten indiscriminately. These results corroborate an earlier suggestion that walking sticks

v choose their diets to reflect nutritional constraints. Additionally, walking sticks distinguish between plant species, and varying leaf quality within a single species, and also modify their diet to reflect past experience. Thus, they may have an impact upon successional processes and, ultimately, the composition of plant communities within forest light gaps.

Vl LIST OF TABLES

1. Annotated results for averaged univariate and multivariate repeated measures analysis of variance of percent consumption data from an experiment showing age-specific variation in food choice ...... 22

2. Annotated results for averaged univariate and multivariate repeated measures analysis of variance of percent consumption data from an experiment showing effects of preexposure on food choice ...... 27

3. Annotated results for averaged univariate and multivariate repeated measures analysis of variance of percent consumption data from an experiment showing variation in intraspecific food choice ...... 28

Vll LIST OF FIGURES

1. Schematic representations of the physiognomy of plant taxa used during feeding trials...... 17

" Histograms showing average percent consumption by L. portoricensis of four foods offered during feeding trials for each age class and sex ...... 24

3. Histograms showing average percent consumption by L. portoricensis of four foods offered during feeding trials for each preexposure treatment and sex of adult...... 26

4. Histograms showing average percent consumption by adult L. portoricensis of three leaf types offered during feeding trials for each sex ...... 30

Vlll CHAPTER I

INTRODUCTION AND PRESENTATION OF THE

PROBLEMS

Optimal Foraging Theory

Much research has been devoted to the particulars of foraging behavior, especially as it relates to optimization (see review by Pyke et al. 1977). The central theme of classical optimal foraging theory (OFT) is that natural selection favors the maximization of energy intake per unit time. The basic tenets of OFT are that an organism should eat only those foods which confer the greatest benefit, relative to cost, and that the choice to consume particular foods is based solely upon the relative abundances of foods of greater value. The theory assumes that a forager can rank its food according to net benefit derived from that food and that the forager should stop eating less beneficial foods as better ones become available, eliminating less valuable foods in reverse rank order (e. g., eliminate the "worst" food first). Finally, OFT predicts that all food types included in the diet should be consumed in proportion to their encounter rates, and that there should be no partial inclusion of lower-ranked foods within the diet. If the highest-ranking food is in sufficient abundance, a forager should specialize on that food alone.

There is a continuing debate regarding the merits of studies conducted to test these predictions (Otte 1975, Fox 1981, Schluter 1981, Taghon 1981, Lacheret al. 1982, Willig

1983 unpubl. MS, Vitousek 1985, Stephens and Krebs 1986). Schluter (1981) states that the energetic basis for optimal foraging theory may be erroneous and contends that it is unrealistic to assume that a forager is able to assess relative abundances of foods or may

1 2 not be capable of making an optimal decision based upon food ranking. Belovsky (1984) considers the strict energetic optimization approach naive. Predictions made by OFf have been falsified by empirical studies in natural (Belovsky 1984 and references therein), seminatural (Lacher et al. 1982), and artificial conditions (Kaufman and Collier 1981,

Willig 1983 unpubl. MS). In general, results from studies testing various aspects of OFf provide equivocal evidence to support of this classical and simplistic approach.

Modifications of Classical OFf

In their book outlining various models of foraging theory, Stephens and Krebs ( 1986) present certain constraints that are placed upon foragers which could cause them to behave contrary to classical predictions. Those constraints include (1) incomplete information regarding food abundance and distribution, (2) certain "rules of thumb" which dictate fixed behavior(s) in a fluctuating environment, (3) sampling behavior for assessment of suitability of various foods, ( 4) risk aversion, (5) predator avoidance tactics, and (6) nutritional requirements which can be satisfied only with a broad diet or the inclusion of food otherwise considered low in value. The most commonly studied of these concerns nutrient constraints; dozens of projects have been concerned with relative nutritional contributions of items in a forager's diet (e. g., Milton 1979, Cates 1980, Taghon 1981,

Willig 1983 unpubl. MS, Strong et al. 1984, Vitousek 1985, Forno and Semple 1987).

Herbivores

Foragers often face conflicting demands regarding food choice, and herbivores must assess more variables to determine food selection than do other consumers (Stephens and

Krebs 1986). Most herbivore diets comprise many foods because overall food quality of any one type is relatively low (when compared to food quality for carnivores) and that one food type rarely will provide the forager with all essential nutrients for survival (Pulliam 3 1975, Joem 1979, Milton 1979, Cates 1980, Moran 1980, Taghon 1981, Lacher et al.

1982, Willig 1983 unpubl. MS, Belovsky 1984, Vitousek 1985, Stephens and Krebs

1986, Lamberti et al. 1987, Quiring and McNei11987). The consensus is that herbivores forage to acquire a balanced diet by consuming food from a variety of plant species.

Factors Affecting Food Choice

An herbivore's decision to accept or reject a certain food may be based upon extrinsic characteristics of that food. For example, the food of highest value may occur in a dangerous locality or may be least suitable for fmding a mate or as a nesting site (Stephens and Krebs 1986). Strong et al. ( 1984) state that the presence of a phytophage on a host may be dependent upon factors such as soil nutrient variation, individual differences in host quality, presence of parasitoids, degree of previous insect damage, degree of protection from predators afforded by plant architecture, and species density. Cates (1980) cites plant apparency (its "visibility" to the forager), rather than overall abundance, as one of the most important factors affecting food selection.

Toxins and Nutrients

Current research shows that food selection is affected by numerous intrinsic properties of a food. Much work has been directed toward plant defense compounds and their effects on herbivory (see Special Feature in Ecology 69(4) 1988 for a current perspective on this topic). The presence of toxic compounds is of undoubtable importance; however, other chemical components of a plant such as energy content, organic nitrogen content, concentration of essential amino acids, and texture (Taghon 1981 ), as well as lignin, fiber, protein content, and masticability (Milton 1979) affect a forager's decision to consume a particular food. 4 Position

Differences in quality of various parts of the same plant are mentioned as determinants of food choice (Milton 1979), and herbivores may choose to concentrate their activities in only one region on a plant. For instance, leaf miners feed and oviposit more often on apical leaflets than on older (lower) leaves on unexploited alfalfa hosts (Quiring and McNeil

1987). Milton ( 1979) proposes that, because of greater protein content and lower toxin levels, young leaves may be preferred over older leaves because they generally contain higher fiber content and greater lignin concentrations than young leaves, or they may become less palatable as a result of prior herbivore damage. However, her overall conclusion is that many factors influence food selection and different plant species store defense compounds in either young or mature leaves.

Prior Experience

Several studies focus upon the contention that "an individual's prior feeding experience may strongly influence its subsequent ability to utilize different host plants"

(Karowe 1989). Some degree of induced preference exists as a consequence of selection of habitat or oviposition sites (Stanton 1982, Quiring and McNeil 1987, Papaj and Prokopy

1988) or selection of optimal food types based upon larval experience or physiological specialization (Otte 1975, Redfearn and Pimm 1988, Karowe 1989).

Variation in diet composition may exist among conspecific herbivores (Cates 1980,

Willig 1983 unpubl. MS, Karowe 1989) or change through time as a function of age, sex, or morphology (Gustafsson 1988). Factors which cause a change in diet composition may result in an "ontogenetic niche shift" within an individual. Different nutrients or proportions of the same nutrients may be required by individuals during different stages in 5 their life cycle. For example, insects have complex life cycles, including metamorphosis and niche shifts, which sometimes allow parents and offspring to coexist without competing for resources (Price 1984). If different-aged individuals have different nutrient requirements, or if developmental stages are related to age, then a difference in diet composition may exist between young and older individuals.

Phasmatids

The order Phasmatodea (walking sticks) contains more than 2,500 externally-chewing folivorous species (Strong et al. 1984) which occur worldwide in both temperate and tropical regions. Most are nocturnal and often will mimic other organisms (namely, scorpions, Bedford 1978) or various plant parts by their activities. Their fundamental survival strategy is crypsis, although they exhibit a variety of predator-avoidance mechanisms such as rocking movements, dropping from a plant followed by catalepsy or death feigning, displays with front limbs which often show aposematic (flash) coloration, rapid leg movements, and production of sounds. Some species actively deter predators by biting or flinging themselves at a potential predator or by discharging defensive secretions

(Bedford 1978). Many species are polyphagous and show distinct feeding preferences

(Bedford 1978). However, little basic research has been done concerning the biology and ecology of the Phasmatodea, and very little is known about most species in this group.

The nocturnal walking stick, Lamponius ponoricensis Rehn, is commonly found within the Luquillo Experimental Forest (LEF) of Puerto Rico. This insect is thought to be an important species in this ecosystem, due to its abundance and longevity (Willig et al.

1986). Phasmatids can reach sufficient numbers to severely damage forests in other regions (Bedford 1978), and herbivores, in general, are considered to be key agents in maintaining forest dynamics and ecosystem stability (Lowman 1984, Brokaw 1985,

Collins et al. 1985, Schowalter 1985). 6 Previous work with L. ponoricensis has revealed that individuals exhibit a variety of food preferences, that considerable daily variation exists in the composition of an

individual's diet, and that there is variation in diet composition depending on sex (Willig

1983 unpubl. MS, Willig et al. 1986, Willig 1987 unpubl. data, Sandlin 1988 unpubl. data). A recent study regarding habitat selection (Sandlin et al. in prep.) has shown that

these walking sticks differentially prefer habitats which contain high densities of Piper

treleaseanum, a common light gap shrub. However, in all food choice experiments, L.

portoricensis ranks this plant as its least preferred food, usually by at least one order of

magnitude (Willig 1983 unpubl. MS, Willig 1987 unpubl. data, Sandlin 1988 unpubl. data).

The Problems

The results of these previous experiments may represent maladaptive behavior. OFf

predicts that a forager should actively seek its preferred food. However, L. portoricensis

prefers to be in areas which contain high densities of its least preferred food. Since this

insect seems to forage in a manner consistent with many hypotheses predicting the

incorporation of nutrient constraints into foraging strategies (Pulliam 1975, Joem 1979,

Lacher et al. 1982, Willig 1983 unpubl. MS, Belovsky 1984, Stephens and Krebs 1986), it is instructive to examine the foraging behavior of L... portoricensis in more detail. I was interested in learning if (1) age or sex, (2) previous exposure to only one food, or (3) intraspecific food variation influence food selection by L. portoricensis.

Although phasmatids do not exhibit metamorphosis (they are hemimetabolous; they have no pupal stage), they undergo several molts; upon reaching adulthood, they have the added constraint of reproductive output (especially females). Thus, a dietary niche shift may occur between nymph and adult walking sticks. Therefore, I first asked the question:

Does L. portoricensis exhibit variation in diet composition as a function of age or sex? 7

L. ponoricensis is wingless, does not travel far within a patch (Willig et al. 1986), and may be constrained by incomplete information regarding available resources within a light gap (patch). Thus, individuals may not forage as they would if they were more familiar with the food composition of a patch (Stephens and Krebs 1986) and could be influenced by previous exposure to forage species within limited homeranges (Willig et al. 1986). For example, individual walking sticks might exhibit partial preference in the laboratory because they are sampling an unfamiliar patch (Pyke et al. 1977, Krebs 1978, Stephens and Krebs 1986) which might present foods they were previously unaware were available.

Additionally, if walking sticks have strict nutrient budgets, they may switch food types after obtaining sufficient nutrient(s) unique to a particular food type and crucial to their fitness. If walking sticks were offered only one food type for an extended period, they might subsequently avoid that food in favor of other foods if they had no immediate nutritional need for the previous food type. If this is true, then after extended exposure they would be expected to switch foods in order to meet nutritional needs which could only be satisfied by other foods. Deviations from preferences shown in the past (Willig 1983 unpubl. MS, Willig 1987 unpubl. data) might be exhibited if walking sticks were exposed to only one food for extended periods of time before given a choice of different food types.

I raised a second question: Does an adult walking stick's previous experience with only one food item influence subsequent food choices?

Partial preferences for different forage species may reflect avoidance of low quality or toxic leaves (Stephens and Krebs 1986). It is known that plants usually concentrate secondary defense compounds (e.g., tannins, phenols, alkaloids) within their topmost

(newest) shoots and leaves which protect them from herbivore damage (Strong et al. 1984,

Salisbury and Ross 1985). Thus, within an individual plant species, some leaves may be unsuitable as food for walking sticks. If distinct preferences for leaves of a particular age

(position) are shown, then earlier results may have been confounded by indiscriminate leaf 8 choice (by researchers), and earlier foraging studies should be repeated including only leaf types known to be forage substrate for L. portoricensis. Thus, I asked a third question:

Do walking sticks perceive differences in the quality of leaves representing various positions (age) on a food plant? CHAPTER IT

MATERIALS AND METIIODS

Study Site

The Luquillo Experimental Forest (LEF) of Puerto Rico has been extensively surveyed for at least thirty years (Odum and Pigeon 1970, Crow 1980, Gines et al. 1984 ). El Verde

Field Station (18° 19' N, 650 45' W) is maintained by the University of Puerto Rico and the Terrestrial Ecology Division of the Center for Energy and Environment Research. It is located within the Lower Montane Rain Forest (dominated by Daczyodes excelsa,

Tabonuco) Life Zone (Ewel and Whitmore 1973). All field work was conducted during the summer of 1988 within the forest immediately surrounding El Verde Field Station, the

Tabonuco Rainforest. This area is a pastiche of patches, many of which are newly­ generated light gaps or gaps in various stages of secondary succession. Within these gaps,

Lamponius portoricensis is found in high densities. Other walking sticks occur in this forest, but their numbers are considerably lower than this species (Van Den Bussche et al.

1989). L. portoricensis population density for the area around El Verde Field Station is between 35.2 and 56.5 individuals per 100m2 plot as estimated by the Bailey daily population estimation method (Willig et al. 1986). The population size of L. portoricensis fluctuates on an approximate three-year cycle (J. Lodge, personal communication), but it is not known what factors influence this cycling or what impact these changes have within the forest.

9 10

Collection and Maintenance

Specimens were captured at night from various light gaps within the Tabonuco

Rainforest near El Verde Field Station. Each individual was marked with a unique symbol and weighed to the nearest 0.001 g. Total length from the tip of the head to the end of the abdomen was measured to the nearest mm. Walking sticks were acclimated to laboratory conditions in glass terraria before they were subjected to feeding trials. During acclimation periods, many walking sticks occupied the same terrarium. All terraria contained water­ filled vials for the maintenance of humidity and had twigs upon which walking sticks could climb and rest. Food was suspended from wire mesh covers and many walking sticks preferred to hang from the mesh. A single walking stick was placed into a 5 x 12 x 13 inch compartment within a terrarium during feeding trials and given a suite of food choices on each of three days.

Experimental Plants All plants used in the food choice experiments occur naturally in light gap areas of the forest and were collected from the same vicinity as were walking sticks. Previous work

(Willig 1983 unpubl. MS., Willig et al. unpubl. data) showed that walking sticks commonly are found foraging on four plant species in the forest: a mid-successional canopy tree, Dendropanax arboreus (DA) (Araliaceae); two species of shrub, Piper hispidum (PH) and ,e. treleaseanum (PT) (Piperaceae); and a woody shrub which grows from prostrate stems,~ baccifera (UB) (Urticaceae). Schematic representations of these plants are shown in Fig. 1. Both species of Piper have similar physiognomic structures. Dendropanax arboreus, when it occurs in light gaps, is a sapling. Urera baccifera usually grows as a tall branch from a ground-running stem in rocky outcrop pings. Leaves of U. baccifera tend to grow in a tight spiral arrangement. 11 Leaf Selection

Considerable intra- and inter-individual variation in leaf quality may exist within each of the four experimental food types. Leaves with extensive insect damage were not used and no more than two leaves were taken from the same plant on any given night in order to ensure a wide representation of available forage substrate in the laboratory experiments.

Whenever possible, leaves were collected from different patches.

Feedin~ Trials

For each feeding trial, walking sticks were given single 500 mg portions of leaves placed atop wire mesh platforms early in the night. This amount of food was chosen because it is in excess of average nightly consumption for a single insect (Willig 1983 unpubl. MS., 1985 unpubl. data, 1987 unpubl. data). Leaves were measured using a

Licor LI-3000 Portable Area Meter in conjunction with a Licor LI-3050A Transparent Belt

Conveyor Assembly (resolution: 1 mm2; accuracy: plus or minus 1% for 10 cm2 samples) before each trial. Remaining leaf area was measured the following morning.

The difference in area measurements for a particular leaf piece represents consumption by the walking stick as well as area loss because of desiccation. Leaves in any terrarium which were not partially consumed during the night, as well as five additional sets of leaves not available to walking sticks, were used to determine the average per cent area loss due to desiccation for each leaf type for each day. A separate desiccation correction term was calculated for each day of feeding trials for each type of leaf offered. This desiccation correction factor (DCF) was used to adjust leaf area measurements to reflect the actual leaf consumption by a walking stick. For a more detailed explanation of the DCF and its use, see the Appendix. In this way, a series of three feeding trials was conducted for each walking stick. Average percent consumption of each food type for the three nights was used in all subsequent statistical analyses. 12

A~e- and Sex-Specific Variation

Male and female walking sticks may have different growth rates (Bedford 1978), and young individuals may have differently-composed diets that reflect either lack of experience

(Price 1984 ), differences in morphology (Gustaffson 1988), or different nutritional requirements. Consequently, age- or sex-related differences in diet may be expected.

Adults and nymphs were fed all four foods for at least three nights previous to their use in food choice experiments. Based upon head-to-abdomen length, nymphs were classified into three age categories (S1, 10-37 mm, N = 24; S2, 38-60 mm, N = 24; S3, 61-75 mm,

N = 24 ). Because of their small size, the sex of S 1 nymphs was difficult to determine based upon the morphology of the genitalia and terminal abdominal segments. The sex of seven of these smallest nymphs was determined by morphological comparison to similarly­ sized nymphs of known sex. Classification of adults (N = 24) was based upon total length measurements and evidence of copulatory behavior. In general, a male was considered an adult if its total length was a minimum of 79 mm, whereas a female was considered an adult if it was a minimum of 80 mm (Willig et al. 1986).

Feeding trials consisted of presenting individuals with equal masses (500 mg) of D. arboreus, E. hispidum, E. treleaseanum, and !l.. baccifera leaves on each of three consecutive nights. Because of constraints on the number of walking sticks which could be subjected to feeding trials simultaneously (only 32 individuals on any given night), six separate series of feeding trials (16 individuals on 7-9 July, 10-12 July, 17-19 July, 22-24

July, 31 July-5 August, and 32 individuals on 13-15 July) were conducted during this experiment. Individuals which did not eat during two consecutive days or which molted during a feeding trial were not used in the food choice experiments. 13 Variation Due to Preexposure

Food choice may be affected by previous exposure to a food type. The effects of limited preexposure to one of the four food types on subsequent food choice were evaluated separately for each food in a series of experiments. A control experiment in which individuals were exposed to all four food types was conducted as well. Control individuals were maintained in acclimation terraria and fed all four foods for at least ten nights after capture and before the feeding trials. Individuals for each preexposure treatment (i.e., DA alone, PH alone, PT alone, or UB alone) were fed exclusively one of the four foods for ten consecutive nights. For each preexposure treatment, eight adult males and eight adult females constituted the experimental population. One female died during the DA preexposure. As a result, only seven females were used in this experiment.

Feeding trials consisted of presenting individuals with equal amounts (500 mg) of D. arboreus, £. hispidum, £. treleaseanum, and !l. baccifera leaves for three consecutive nights. Three separate sets of feeding trials were conducted, and all members of a particular preexposure group were measured simultaneously. All experiments were conducted between 6 and 19 August. In order to maintain equal sample sizes for all treatments, average consumption values for each plant species exhibited by the seven females in the pre-DA group were substituted as these values for the dead female.

Intraspecific Food Variation

Leaves from the same plant may differ in their forage quality, and this variation may be detected by a forager. Adults were fed all four foods for at least three nights before feeding trials. Feeding trials consisted of presenting individuals with equal amounts (500 mg) of top, middle, and bottom leaves of one of the four foods (DA, PH, PT, or UB) for three consecutive nights. Two series of feeding trials composed the experiment because only 32 individuals on any given night could be subjected to feeding trials simultaneously. TheDA 14 and PT trials were conducted from 6-11 July; the PH and UB trials were conducted from

24-30 July.

The selection of each type of leaf was made according to specified criteria. Again,

leaves with extensive insect damage were not chosen, and no more than two leaves were

taken from the same plant on any given night. Top leaves were considered to be the first

fully-expanded leaves on any of the four plant species. Because these plants exhibit

different growth forms, different criteria were required to define a middle leaf (see Fig. 1).

Bottom leaves which were relatively undamaged were chosen as the lowest (or oldest)

leaves.

Data Analyses

All data were adjusted for area loss due to desiccation according to unique DCFs for

each leaf type for each night of a feeding trial (see Appendix). The nature of these

corrections is such that some consumption values were corrected to small negative values.

Since these values represent minute quantities of leaves eaten in the feeding trials, these

negative values were set equal to zero in subsequent analyses. Consumption data for each

leaf type offered over the three days were converted to a three-day percent consumption for

each insect. In this way, variation in consumption because of size differences among

individuals was eliminated, with also the data standardized to weight daily consumption of each food by total daily consumption of all foods. Individuals occasionally did not forage; this standardization circumvents the problem of treating zero consumption of four foods as indicative of equal preference. Also, this method does not require mean substitutions

(which might misrepresent actual consumption of individuals) to maintain equal sample sizes.

As indicated above, a multitude of factors may determine foraging decisions, dictating that statistical approaches to examination of most ecological data should be multivariate. In 15 the foraging experiments, the multivariate data contain repeated (and thus non-independent) measures for a given individual (e. g., sex, age, food type, and leaf position). Hence, repeated measures analysis of variance was performed using SPSS-X statistical software package (SPSS-X, Inc. 1988). Tests for sphericity (to detect repeated measures heteroscedasticity) were performed using the Mauchly sphericity test. Both averaged univariate and multivariate tests were performed by SPSS-X procedure MANOV A.

Multivariate tests of significance employed were Pillai's, Hotelling's, and Wilks' tests.

Unfortunately, .a posteriori tests in a doubly multivariate setting have not been adequately developed. As a result, comparisons of particular treatment groups must be based upon trends alone rather than more rigorous statistical analyses. 16 RG. 1. Schematic representations of the physiognomy of plant taxa used during feeding trials. Representative top, middle, and bottom leaves are indicated for (A) Piper hispidum and£. treleaseanum, (B) Dendropanax arboreus, and (C) Urera baccifera. For the sake of clarity, only a few leaves of D. arboreus are shown. 17

A

BOTTOM TOP c

BOTTOM MIDDLE CHAPTER III

RESULTS

Averaged univariate approaches to repeated measures analyses are sensitive to deviations from underlying assumptions. I therefore evaluated sphericity via the Mauchly sphericity test with alpha= 0.01 in order to have greater power at detecting deviations which could confound interpretations of subsequent analyses. This test revealed significant heteroscedasticity in the data for each of the three experiments (Chi-square approximation of Mauchly sphericity test W: p = 0.000 for age-specific variation and preexposure experiments, p = 0.075 for the intraspecific variation experiment). In this context, only the multivariate analyses are meaningful, and they form the basis for subsequent discussions.

Nonetheless, averaged univariate results are in accord with multivariate results in all but one case.

Effects of Age- and Sex-Specific Variation

Multivariate tests showed a significant (p = 0.016) three-way interaction (sex by age by food), indicating that the sexes respond differently to the four food types during different stages of their life cycle (Table 1). Regardless of the age or sex of walking sticks,

DA, PH, and UB each constituted at least 10% of the diet, whereas PT consistently represented a minor proportion of leaf material consumed except for S3 males (Fig. 2).

Examination of patterns of food consumption exhibited by males and females, as well as by different age groups, reveals that relationships are complex among age groups and are different depending upon sex. For example, the proportional representation of PH in the

18 19 diet decreased with age for males but remained relatively constant for females. Similarly,

DA and UB are shown as major dietary components for all age and sex groups except for

S 1 males. In this case, PH dominated, representing over 50% of the diet.

Specific trends are easier to delineate by considering one food at a time. DA consumption patterns differ markedly between males and females, as there is an increase in males' percent consumption, levelling at adulthood, whereas females, regardless of age, consume relatively the same amounts of this food. Males decrease consumption of PH with age, but females eat the same amounts or, if anything, increase the proportion of PH in their diets as they mature. PT patterns are comparable for all individuals (except S3 males), with PT being the least preferred food which is consistently avoided; it does not contribute to the diet of adult males and figures only minimally in the diet of females. UB consumption remains fairly constant for both sexes; however, the proportion of UB in the diet is generally greater for females than for males. Consumption of UB predominates in the diet, especially for males, and is more consistently favored than DA. These results

unequivocally demonstrate that food selection in L. portoricensis is affected by developmental stage and that males and females differ in the manner in which food

selection changes with age.

Effects of Preexposure to One Food

The manner in which males and females modify their diet in response to different preexposure regimes is complex (Fig. 3 offers a visual portrayal of these intricate patterns) as indicated by two significant two-way interactions (Table 2). Preexposure to a particular food type affects subsequent patterns of consumption in the same fashion for males and females, as indicated by the significant interaction between food and preexposure.

Regardless of preexposure regime, the sexes differed in their proportional consumption of the four food types, as indicated by the significant sex by food interaction (Table 2). 20 Combining information from Table 2 and Fig. 3 renders a straightforward

interpretation impossible; relationships are intricate, with direct causalities buried within

the significant two-way interactions for both sex by food and preexposure by food.

Because the pattern of consumption is almost identical in male and female control groups

and the pattern of consumption within a preexposure group or sex was never identical to

the control groups, comparison of control walking sticks with preexposure groups and sex

groups is not further complicated. Effects of preexposure, regardless of sex, are such that

the patterns for each preexposure regime are generally comparable but differ according to

treatment. As an example, comparison of the pattern of UB histograms between males and

females shows the same general pattern but this pattern is not the same as that exhibited by

other preexposure groups.

While consumption patterns between males and females within a preexposure group

are directly comparable in a broad sense, the significant sex by food interaction indicates

that differences are manifested in the extremes to which foods are preferred between the

sexes within any preexposure group and when compared to the control groups. For

instance, males preexposed to DA included proportionately more DA in subsequent diet

than did DA-preexposed females or the control groups. In the same fashion, females in the

PH preexposure group consumed proportionately more of this food than did males or the

control groups in later feeding trials. Moreover, females ate relatively the same proportion

of DA and UB in all preexposure groups, whereas males increased their intake of the

preexposure food when it was DA or UB.

Overall, the consumption of UB and DA (as compared to control consumption) is enhanced, especially when UB and DA are the preexposure foods. In all cases, PT was

not included in the diet of males, whereas females ate small amounts of PT, although it constituted at best only a minute proportion of the diet. In keeping with the statistical results, these patterns were neither paralleled among the various treatment groups within a 21 sex, nor were they paralleled by the proportional representation of each food within control diets. However, the observed consumption patterns clearly demonstrate that previous experience with only one food item influences subsequent food choices exhibited by L. portoricensis.

Effects of Intraspecific Variation in Food Quality

The significant species by position interaction in the repeated measures analyses (Table

3) suggests that the degree to which intraspecific differences in plant quality affect consumption depends upon the particular forage plant and that these effects are consistent in both males and females. The primary cause of this significance appears to be related to preferential consumption of bottom leaves and avoidance of top leaves of PT (Fig. 4). In general, males and females consumed top, middle, and bottom leaves of DA, PH, and UB indiscriminately, but strongly preferred bottom leaves of PT. Males seem to be more fine­ tuned for detection of differences in leaf quality (position= quality; see Milton 1979, Cates

1980) when given DA or UB leaves. They avoided bottom leaves in favor of either top or middle leaves of these plants. Females seem to be more likely to eat DA and UB leaves indiscriminately, with some possible preferences shown for middle UB leaves. Results from this experiment show that L. portoricensis perceives and responds to differences (or their absence) in the quality of leaves representing different positions (age) on forage plants. 22

TABLE 1. Annotated results for averaged univariate and multivariate repeated measures analysis of variance of percent consumption data from an experiment showing age-specific variation in food choice. In the analysis, food is the within-subjects factor, sex and age are the between-subjects factors.

Source of Variation AverageD Univariate Approach Multivariate Approach df ss MS F p Wilks Lambda P

Food 3 8.46 2.82 39.31 0.000 0.080 0.000

Sex X Food 3 0.32 0.11 1.51 0.213 0.948 0.200

Age X Food 9 1.09 0.12 1.69 0.092 0.839 0.081

Sex X Age X Food 9 2.00 0.22 3.09 0.002 0.793 0.016

Within 264 18.93 0.07 23 FIG. 2. Histograms showing average percent consumption by L. portoricensis of four foods offered during feeding trials for each age class and sex. Lines above each histogram bar represent standard errors. DA = Dendropanax arboreus; PH = Piper hispidum; PT =f. treleaseanum; UB = ~ baccifera; S 1 =smallest size class of nymphs (:-.ee text for exact length for size classes); S2 =intermediate-sized nymphs; S3 =largest nymphs; Adult= adult walking sticks. 24

1.0 MALES z • DA 0 ~PH ~ Q. 0.8 lS1 PT :2: 0 UB :> (f) z 0.6 u0 1- 0.4 z LUu c: 0.2 LU Q. 0.0 51 52 53 ADULT AGE CLASS

FEMALES z 1.0 • DA 0 ~PH ~ Q. 0.8 ~ PT :2: 0 UB :> ~ 0.6 0 u 1- 0.4 z LU u c: 0.2 LU Q. 0.0 51 52 53 ADULT AGE CLASS 25 FIG. 3. Histograms showing average percent consumption by adult L. portoricensis of four foods offered during feeding trials for each preexposure treatment and sex of adult. Lines above each histogram bar represent standard errors. ALL = preexposure to all four foods (control); DA = Dendropanax arboreus; PH= Piper hispidum; PT = P. treleaseanum; VB=~ baccifera. 26

MALES z 1.0 • DA 0 j: ~PH Q. 0.8 ISl PT ::E 0 UB ::::> (f) z 0.6 0 u 1-z 0.4 LLJ u a: 0.2 LLJ Q. 0.0 ALL DA PH PT UB PREEXPOSURETREATMENT

FEMALES z 1.0 • DA 0 ~PH j: Q. 0.8 ~ PT ::E 0 UB ::::> ~ 0.6 0 u 1- 0.4 z LLJ u a: 0.2 LLJ Q. 0.0 ALL DA PH PT UB PREEXPOSURETREATMENT 27

TABLE 2. Annotated results for averaged univariate and multivariate repeated measures analysis of variance of percent consumption data from an experiment showing effects of preexposure on food choice. In the analysis, food is the within-subjects factor; sex and preexposure are the between-subjects factors.

Source of Variation Averaged Univariate Approach Multivariate Approach df ss MS F p Wilks Lambda P

Food 3 11.55 3.85 115.77 0.000 0.003 0.000

Sex X Food 3 0.43 0.14 4.35 0.005 0.750 0.000

Preexposure X Food 12 2.21 0.18 5.53 0.000 0.584 0.000

Sex X Preexposure X Food 12 0.77 0.06 1.93 0.032 0.795 0.190

Within 210 6.99 0.03 28

TABLE 3. Annotated results for averaged univariate and multivariate repeated measures analysis of variance of percent consumption data from an experiment showing variation in intraspecific food choice. In the analysis, leaf position is the within-subjects factor, sex and plant species are the between-subjects factors.

Source of Variation Averaged Univariate Awroach Multivariate Approach

df ss MS F p Wilks Lambda P

Position 2 0.16 0.08 2.00 0.141 0.919 0.099

Sex X Position 2 0.07 0.03 0.85 0.432 0.959 0.320

Species X Position 6 3.79 0.63 15.78 0.000 0.299 0.000

Sex X Species X Position 6 0.32 0.05 1.33 0.251 0.856 0.193

Within 112 4.49 0.04 29 FIG. 4. Histograms showing average percent consumption by adult L. portoricensis of three leaf types offered during feeding trials for each sex. Individual walking sticks were offered leaves from one of four plant species. Lines above each histogram bar represent standard errors. DA = Dendropanax arboreus; PH= Piper hispidum; PT = £. treleaseanum; UB = ~ baccifera. 30

MALES z 1.0 • BOTTOM 0 i= 0 MIDDLE c. 0.8 0 TOP ~ ::::> (f) z 0.6 0 u .... 0.4 zw u c: 0.2 w c. 0.0 DA PH PT UB PLANT SPECIES

FEMALES z 1.0 • BOTTOM 0 0 MIDDLE i= c. 0.8 ..,.... 0 TOP ~ ::::> (f) z 0.6 0u .... 0.4 z w u c: 0.2 w c. 0.0 DA PH PT UB PLANT SPECIES CHAPTER IV

DISCUSSION

Herbivory

Stephens and Krebs (1986) delineate special features of herbivores that make them unusual predators, noting that they face problems regarding foraging strategies that other foragers do not. Phytophages spend relatively little time searching for food and most of their time ingesting or digesting food. Foods rarely "occur in neatly packaged prey items," and there may only be the choice of which part of the plant to eat, rather than which plant species to choose as a host. Generalist herbivores have relatively complex diets that must meet the forager's nutritional requirements (Belovsky 1984) while avoiding plant defense mechanisms. As a result, most herbivore diets are characterized by partial consumption of a few or many plant species even when relative abundances of foods favor specialization

(as predicted by classical OFf; see Lacher et al. 1982), and tend to more closely fit models proposed for optimal foraging within nutrient constraints (Otte 1975, Pulliam 1975, Pyke et al. 1977, Fox 1981).

The size and mobility of an herbivore may be the sole determinant of its foraging behavior (Cates and Orians 1975); in particular, a relatively small forager is more likely to perceive its environment as either coarse-grained or homogeneous (Levins 1968). The patch-use model of Stephens and Krebs (1986) predicts that, if a forager must travel great distances to search for food, it may choose lower quality foods in order to decrease search and travel times. For insects, the probability of finding a proper host plant may be no greater than that due to chance alone (Jermy et al. 1988). The choice to feed on a plant

31 32 often is made only after sensory contact is made with the plant (Quiring and McNeil

1987, Blaney and Simmonds 1988). If patches are large relative to the size of a forager, then that forager may be forced to act according to certain "rules of thumb" (Stephens and

Krebs 1986), accepting trade-offs which only ensure its survival, rather than maximizing fitness. Thus, small insect foragers must contend with constraints that may cause them to consume foods of low value.

Diet composition for herbivores is influenced by a seemingly endless array of biotic and abiotic factors. Attributes of plants other than nutritional content, concentration of chemical defenses, abundance, and distribution affect an herbivore's foraging decisions.

Niemela and Tuomi ( 1987) report that plants employ certain antiherbivore tactics, including color variegation and leaves which mimic previous herbivore damage or oviposition, to deter leaf-chewing predators. They contend that plants such as these may deter herbivores by attracting predators which recognize the results of herbivore activity or, to dissuade insects, by attracting parasitoids which ultimately destroy the plant's adversaries. Some insects choose plants as forage because they are able to sequester and reconstitute those defense compounds into their own defense chemistry or use them for pheromones (Schowalter 1985). Anti-parasitoid toxins (Strong et al. 1984) or defensive secretions to deter vertebrate predators (Bedford 1978, Strong et al. 1984) may be the plant's major contribution to its predator. Thus, indirect factors may determine whether a forager includes a particular food item within its diet (Cates and Orians 1975), and these factors may be of overriding importance to the forager.

Lamponius portoricensis

This study was conducted to determine if particular factors or combinations of factors influence the food choice of L. portoricensis. In particular, I addressed three questions:

( 1) Does L. portoricensis exhibit variation in diet composition as a function of age or sex, 33 (2) Does previous experience with only one food item influence subsequent food choices

by adult walking sticks, and (3) Do walking sticks perceive differences in the quality of

leaves representing various positions (age) on a food plant? In consideration of the

previous discussion, factors which influence the observed foraging behavior of L.

portoricensis may be intricately related and may explain its behavior only when considered

in concert. In fact, statistical analyses of the data suggest that variables act in synergistic

ways to affect the overall outcome of feeding trials. In an earlier study of food preferences,

Willig ( 1987 unpubl. data) found similar results, with sex and plant species both influencing diet composition.

Special Problems Faced By Females

Adult female walking sticks are consistently larger than adult males (Willig et al.

1986), a pattern found commonly in phasmatids which is explained in many species by an

extra moult by females (Bedford 1978). Bedford contends that this extra moult might

account for differences in abundance between females and males (in all studies with L.

portoricensis, field capture rate of females was lower than for males--see Willig et al. 1986)

and for higher female mortality. However, females outlive males in many species. These

differences in life history between the sexes could coincide with different foraging

strategies such as those elucidated by this study. For example, females consistently

included small amounts of PT in their diet, regardless of their previous treatment

(preexposure or simple acclimation), whereas males usually excluded this food from their diet. One likely explanation for this difference in PT inclusion is that females are more nutrient-limited than males because of their relatively high reproductive investment in comparison to males and may obtain a portion of their nutritional requirement from PT.

Most of the females used in this experiment were laying eggs continually and, consequently, were potentially functioning with nutritional constraints. Support for this 34 contention is evidenced, as females included a greater variety of foods, or exhibited a more

equal representation of each food, within their diet.

Implications of Observed Patterns of Consumption

Influence of Age and Sex on Diet Composition

The experiment assessing whether age and sex are determinants of food choice reveals

that there is not an abrupt change in the pattern of food consumption between one age class

and the next. However, the manner in which the diet changes with age is different for

males than for females. The overall trend in females is for the diet to become more broadly

based as individuals mature. Conversely, in males, PH is gradually eliminated from the

diet, while DA and UB become established as dominant food types. Otte (1975) found that

first and late instar grasshopper nymphs of three species of Schistocerca exhibited

differences in food preference; Blaney and Simmonds (1988) demonstrated that cues for

food selection, and thus, food selection itself, changed between young and adult

butterflies. Similarly, Gustafsson (1988) found that juvenile coal tits, fill:us ~. foraged

within a more varied niche, with the young using food and foraging sites rarely used by

adults. As these same juveniles matured, they switched to adult feeding habits. Likewise,

Cassidy (1978) reports that older individuals of the Indian walking stick, Carausius

morosus, are more selective (ate fewer types of foods) than younger insects.

Adult L. portoricensis are more mobile than nymphs (Willig et al. 1986). If adults

move more than nymphs, they may be better able to select only their preferred foods and

consume a narrower diet. If nutritional requirements for walking sticks change as they

mature and vary between males and females, and if different nutrients are obtained from

different plants (Otte 1975, Belovsky 1984, Stephens and Krebs 1986), then the observed

patterns of food consumption are expected. Overall, the youngest (S 1) age class shows the

broadest acceptance of all four experimental foods. Adult males constitute the only group 35 which specialized by eliminating PT from the diet, although S3 females show food choices comparable to those of adult males. In general, PT is the least-consumed food for any age or sex class and can be considered the least-preferred food of L. ponoricensis.

Influence of Previous Experience on Diet Composition

Preexposure for only a short time affects subsequent food consumption patterns in adult 6 portoricensis. These results corroborate other contentions that variation in feeding habits of adults reflects previous events (Papaj and Prokopy 1988, Redfearn and Pimm

1988, Karowe 1989), but these studies concentrated upon induced preferences and physiological specialization throughout an individual's lifetime. I did not address previous exposure from earliest instar to adult; the results from my experiment probably are not due to physiological specialization during the preexposure time. However, an induced preference is exhibited by walking sticks preexposed to both DA and UB.

Consumption patterns reveal that DA and UB are distinctly preferred over the other two foods and that neither was consistently preferred over the other. DA and UB may be substitutible currencies--foods which confer equivalent benefits (Stephens and Krebs

1986). Similarly, Willig (1983 unpubl. MS) found no statistically distinguishable differences between consumption of these two foods by L. portoricensis. Thus, it is possible that DA and UB may be foods of equal value and may be perceived as indistinguishable by this consumer. Or, they may both offer unique dietary components required in roughly equivalent amounts.

Food Quality as it Relates to Succession

The production of defense compounds is related to the successional stage of a plant.

Early successional plants concentrate energetic efforts on rapid vegetative growth and early reproduction and devote less energy to chemical protection (Cates and Orians 1975, Otte 36 197 5). Indeed, leaf construction cost for short-lived (light gap) species is sufficiently high

(Williams et al. 1989) that these species may become preferred foods for herbivores because they contain reduced amounts of defense compounds in comparison to longer-lived species (Baldwin and Schultz 1988). Some plant species with relatively persistent leaves have significantly higher concentrations of immobile chemical defenses (lignins and tannins), indicating that growth rate will determine the level of defense within the plant

(Coley 1988). Additionally, mid- and late-successional trees, when in light gaps, may be characterized by rapid growth and thus be more palatable as saplings because of lower concentrations of defense compounds (Baldwin and Schultz 1988). Preexposure feeding trials with L. ponoricensis showed clearly that DA and UB (mid-successional plants, Perez

1988) were included disproportionately often in subsequent diets. Conversely, preexposure to PT and PH (early-successional plants, Brokaw 1985) did not predispose walking sticks to later include more of these plants in their diets. Thus, L. portoricensis forages according to these predictions by preferring young mid-successional plants but does not behave in a manner which supports the prediction that early successional plants will be more palatable. One explanation for this demonstrated avoidance of PH and PT is that these plants have long-lived leaves which may contain high levels of toxins (but see

below). Otte (1975) tested grasshoppers reared on one of eight plants against a control group reared on a mixed diet and found that subsequent diets showed increased preferences for certain species (in this case, late successional plants). If the amount of defense compounds produced is related to successional position of a plant (Otte 1975, Cates 1980, Coley 1988,

Kearsley and Whitham 1989), then Piper spp. would presumably contain fewer toxins than

DA and UB (Fleming 1985). Nonetheless, DA and UB are overwhelmingly preferred in

this and the other experiments. It is likely that walking stick responses to PH and PT are

based upon the nutritional composition and battery of defense compounds they contain. 37 Effects of Food Variation Within a Fora~e Species

Undoubtedly, results from the intraspecific food variation experiment are at least partly explainable in terms of nutritional and defensive compound composition as perceived by L. portoricensis. Females show a lack of discrimination between leaves of DA and UB ' whereas males show avoidance of bottom leaves of these species. However, a distinct preference is shown for lower leaves of PT, the most commonly-encountered gap plant.

Although a model has been proposed which predicts that a food should become more rare within an organism's diet as a consequence of being more abundant in a patch (Engen and

Stenseth 1984), this model assumed only two foods to be available. Clearly, explanation for the observed abundance-preference discrepancy is beyond the scope of this paper.

The PT consumption pattern is worthy of discussion because it is markedly different from patterns for the other three foods. Since the group of walking sticks given only choices of PT leaves demonstrated relatively extreme preferences, they may have chosen the least disagreeable of an unpreferred food in avoidance of starvation, or, for some reason, bottom PT leaves are dietary components regardless of the array of foods available.

In light of previous habitat selection work, speculation about why PT might be preferred has many facets. There may be a trade-off between food quality and some aspects of survivorship. For example, PT may provide the best camouflage available within gaps.

Or, the choice to remain in patches of PT may be a result of lack of knowledge about other potential host plants at or near the periphery of these patches (see Stephens and Krebs 1986 for a discussion and model for "rules of thumb" resulting from lack of experience).

Further, fidelity to this plant may be a result of preferential egg deposition by females (as a result of L. portoricensis' differential selection of patches containing high densities of PT) or of feeding by nymphs on the host resulting in physiological specialization (Stanton

1982, Karowe 1989). 38 In contrast to PT consumption, the pattern of consumption for PH leaves does not differ between the sexes or for any leaf type, indicating that leaves of this species do not vary in quality or that any variation is insufficient to be distinguished by L. portoricensis.

This is a widely-distributed plant species which has been studied in neotropical regions including Panama (Brokaw 1985), Costa Rica (Fleming 1985, Denslow et al. 1987,

Baldwin and Schultz 1988), and Mexico (Chazdon and Field 1987, Williams et al. 1989).

It has been described as a habitat generalist (Chazdon and Field 1987, Williams et al. 1989) which is able to tolerate low-light situations but probably only establishes within light gaps

(Brokaw 1985). Thus, PH leaves may have very low quantities of secondary defense compounds and, perhaps, relatively high nutritional value (Denslow et al. 1987, Baldwin and Schultz 1988, Williams et al. 1989). If this is true for PH, then the observed indiscriminate foraging behavior of L. portoricensis is to be expected. Results from this experiment unequivocally demonstrate that L. portoricensis does not respond to any intraspecific variability in leaf quality for PH, but that there is a demonstrated response to variability in leaves of PT , and to a lesser degree, to variability within leaves of DA and

UB.

Monophagy. Polyphagy. and Plant Successional Status

The differences in leaf quality on a single plant species may be nutrient-related or may reflect changes resulting from prior leaf damage (Forno and Semple 1987), with older leaves likely to show signs of increased damage and lowered nutritional content. Although young leaves may be more nutritious, Cates (1980) points out that young leaves of many species may have the highest concentrations of defense compounds and that leaves of early­ successional plants harbor fewer toxins than comparable leaves of later successional plants.

In addition, he proposes that individuals of mono- and oligophagous species should be able to respond better to these toxins than should polyphagous species. Thus, polyphagous 39 insects should prefer to eat those leaves which have the lowest levels of defense compounds--specifically, older leaves on any plant, and leaves of early-successional plant species. The converse should be true for mono- and oligophagous species.

Cates ( 1980) found that larvae of mono- and oligophagous species preferred young leaves and leaves of early-successional plant species, whereas polyphagous herbivores preferred mature leaf tissue and leaves of mid- and late-successional plants. L. portoricensis is, according to Cates' definition, a polyphagous species because it consumes foods from three or more plant families. My results, then, corroborate those of Cates: L. portoricensis distinctly prefers mature leaves of an early-successional plant, PT, and does not discriminate between leaves of later-successional species such as DA and UB. In contrast to his study, however, L. portoricensis shows an overall preference for the later­ successional species, DA and UB. Reasons for this contrast are probably related to properties of the four plants which may include masticability and fiber, nutrient, or toxin content. Both PH and PT have relatively thick leaves, and these may be less suitable for chewing or may simply take too much time to consume. As OFf predicts, foragers should maximize intake of some currency per unit time, and these two plants may be too costly to consume. I attempted to measure the relative hardness of different leaves but found that there was too much variability in my measurements for reliable estimates to be made. The possibility that there is considerable variation in the nutritional quality of these plants seems likely. Future studies of relative nitrogen, phosphorus, and carbon contents of these plants and assessments of their relative toxicity will certainly serve to elucidate these discrepancies.

Foraging studies with L. ponoricensis demonstrate that this organism clearly does not fit classical models of OFf because least preferred foods are included in its diet, even when food abundances are high (Willig 1983 unpubl. MS), and that energy is probably not an important currency for this walking stick to maximize. My preference results corroborate 40 Willig's earlier experiments (1983 unpubl. MS, 1987 unpubl. data) and suggest that a

multivariate explanation for the causes of L. portoricensis' foraging behavior is appropriate.

For the Future

There is evidence that learning is involved in decisions regarding food and habitat

selection (Otte 1975, Milton 1979, Stanton 1982, Papaj and Prokopy 1988). Variation in

learning ability among individuals is "raw material for natural selection" (Gotceitas and

Colgan 1988) and may account for observations of individual specialization contained

within a polyphagous population (Karowe 1989). The idea that insects base present decisions upon previous experience corroborates ideas regarding age-related decisions, but

it provides other interesting avenues for thought as well. For example, learning to forage

on a particular host may involve density-dependent (and, thus, dynamic) experiences with

both the forager's host and its potential predators. Predicted preferences of L.

portoricensis could be incorporated into future research assessing the impact of this insect

on the success (or failure) of gap species or later successional saplings as gap colonizers.

Stick insects in other regions (e.g., Australia) are important agents of major forest

disturbance because they may defoliate forested areas when they reach outbreak population densities (Bedford 1978). Because L. por1Qricensis is locally abundant and its population

density fluctuates, my results compliment studies of reforestation and general forest

response to disturbance within the Tabonuco Rainforest. Outbreaks and other disturbances

(most notably, gap formation) have been purported to be major regulators of forest stability

(Pickett and White, 1985). Since L. portoricensis is found in high densities in light gaps

(Willig et al. 1986), this herbivore may be a key agent in fostering not only forest stability but also successional change. If herbivores, including L. portoricensis, are considered major determinants of plant community structure (Lowman 1984, Strong et al. 1984, 41

Collins et al. 1985, Schowalter 1985, Simms and Rausher 1989), then knowledge of their foraging behavior may have direct and dramatic effects on the understanding of entire ecosystems. LITERATURE CITED

Baldwin, I. T. and J. C. Schultz. 1988. Phylogeny and the patterns ofleafphenolics in gap- and forest-adapted Piper and Miconia understory shrubs. Oecologia 75:105- 109.

Bedford, G. 0. 1978. Biology and ecology of the Phasmatodea. Annual Review of Entomology 23:125-149.

Belovsky, G. E. 1984. Herbivore optimal foraging: a comparative test of three models. American Naturalist 124:97-115.

Blaney, W. M. and M.S. J. Simmonds. 1988. Food selection in adults and larvae of three species of Lepidoptera: a behavioural and electrophysiological study. Entomologica Experimentale et Applicata 49:111-121.

Brokaw, N. V. 1985. Treefalls, regrowth, and community structure in tropical forests. Pages 53-69 in S. T. A. Pickett and P. S. White, editors. The ecology of natural disturbance and patch dynamics. Academic Press, New York, New York, USA.

Cassidy, M.D. 1978. Development of induced food plant preference in the Indian stick insect, Carausius morosus. Entomologica Experimentale et Applicata 24:87-93.

Cates, R. G. 1980. Feeding patterns of monophagous, oligophagous, and polyphagous insect herbivores: the effect of resource abundance and plant chemistry. Oecologia 46:22-31.

----- and G. H. Orians. 1975. Successional status and the palatability of plants to generalized herbivores. Ecology 56:410-418.

Chazdon, R. L. and C. B. Field. 1987. Determinants of photosynthetic capacity in six rainforest Piper species. Oecologia 73:222-230.

Coley, P. D. 1988. Effects of plant growth rate and leaf lifetime on the amount anf type of anti-herbivore defense. Oecologia 74:531-536.

Collins, B.S., K. P. Dunne, and S. T. A. Pickett. 1985. Responses of forest herbs to canopy gaps. Pages 218-234 in S. T. A. Picke~t and P. S. \Yhite, editors. The ecology of natural disturbance and patch dynamics. Academic Press, New York, New York, USA.

42 43 Crow, T. R. 1980. A rainforest chronicle: a 30-year record of change in structure and composition at El Verde, Puerto Rico. Biotropica 12:42-55.

Denslow, J. S., P.M. Vitousek, and J. C. Schultz. 1987. Bioassays of nutrient limitation in a tropical rain forest soil. Oecologia 74:370-376.

Engen, S. and N. ~- Sten~eth. 1984. An ecologic~} paradox: a food type may become more rare m the dtet as a consequence of bemg more abundant. American Naturalist 124:352-359.

Ewel, J. J. and J. L. Whitmore. 1973. The ecological1ife zones of Puerto Rico and the United States Virgin Islands. Forestry Service Research Paper ITF-18. Available from the Institute of Tropical Forestry, Rio Piedras, Puerto Rico.

Fleming, T. H. 1985. Coexistence of five sympatric Piper (Piperaceae) species in a tropical dry forest. Ecology 66:688-700.

Forno, I. W. and J. L. Semple. 1987. Response of Salvinia molesta to insect damage: changes in nitrogen, phosphorus and potassium content. Oeco1ogia 73:71-74.

Fox, L. 1981. Defense and dynamics in plant-herbivore systems. American Zoologist 21:853-864.

Gines, C., N. L. Ortega, and R. B. Waide. 1984. Bibliography of research at El Verde, Puerto Rico. CEER-T -193 Available from: Center for Energy and Environment Research, PO Box 3682, Rio Piedras, PR. 00936.

Gotceitas, V. and P. Colgan. 1988. Individual variation in learning by foraging juvenile bluegill sunfish (Lepomis macrochirus). Journal of Comparative Psychology 102:294-299.

Gustafsson, L. 1988. Foraging behaviour of individual coal tits, Parus ater, in relation to their age, sex and morphology. Animal Behaviour 36:696-704.

Jermy, T., A. Szentesi, and J. Horvath. 1988. Host plant finding. in phytophagous insects: the case of the Colorado potato beetle. Entomologtca Expenmentale et Applicata 49:83-98. Joern, A. 1979. Feeding patterns in grasshoppers (Orthoptera: Acrididae): factors influencing diet specialization. Oecologia 38:325-347.

Karowe D. N. 1989. Facultative monophagy as a consequence of prior feeding ~xperience: behavioral and physiological specialization in Coli as philodice larvae. Oecologia 78:106-111. Kaufman, L, W. and G. Collier. 1981. The economics of seed handling. American Naturalist 118:46-60.

Kearsley, M. J. C. and T. G. Whitham. 1989. Developme~tal changes in resistance to herbivory: implications for individuals and populations. Ecology 70:422-434. 44 Krebs, J. R. 1978. Optimal foraging: decision rules for predators. in J. R. Krebs and N. B. Davies, editors. Behavioural ecology: an evolutionary approach. Blackwell Scientific Publications, Oxford, England.

Lacher, T. E., Jr., M. R. Willig, and M.A. Mares. 1982. Food preference as a function of ~esource a~undance with mu~tiple prey types: an experimental analysis of optrmal foragtng theory. Amencan Naturalist 120:297-316.

Lamberti, G. A., J. W. Feminella, and V. H. Resh. 1987. Herbivory and intraspecific competition in a stream caddisfly population. Oecologia 73:75-81.

Levins, R. 1968. Evolution in changing environments. Princeton University Press, Princeton, New Jersey, USA.

Lowman, M. D. 1984. An assessment of techniques for measuring herbivory: is rainforest defoliation more intense than we thought? Biotropica 16:264-268.

Milton, K. 1979. Factors influencing leaf choice by howler monkeys: a test of some hypotheses of food selection by generalist herbivores. American Naturalist 114:362-378.

Moran, V. C. 1980. Interactions between phytophagous insects and their Opuntia hosts. Ecological Entomology 5:153-164.

Niemela, P. and J. Tuomi. 1987. Does the leaf morphology of some plants mimic caterpillar damage? Oikos 50:256-257.

Odum, H. T. and R. F. Pigeon, editors. 1970. A tropical rainforest: a study of irradiation and ecology at El Verde, Puerto Rico. U.S. Atomic Energy Commission. NTIS, Springfield, Virginia, USA.

Otte, D. 1975. Plant preference and plant succession; a consideration of evolution of plant preference in Schistocerca. Oecologia 18:129-144.

Papaj, D. R. and R. J. Prokopy. 1988. The effect of prior adult experience on components of habitat preference in the apple maggot fly (Rhagoletis pomonella). Oecologia 76:538-543.

Perez, I. E. 1988. Field guide to the trees ofBisley Watersheds, Luquillo Experimental Forest, Puerto Rico. an Institute of Tropical Forestry Publication, Southern Forest Experiment Station, USDA Forest Service, in collaboration with the University of Puerto Rico.

Pickett, S. T. A. and P. S. White, editors. 1985. The ecology of natural disturbance and patch dynamics. Academic Press, Inc., New York, New York, USA.

Price, P. W. 1984. Insect ecology. John Wiley and Sons, Inc., New York, New York, USA.

Pulliam, H. R. 1975. Diet optimization with nutrient constraints. American Naturalist 109:765-768. 45 Pyke, G.~-· H. R. Pulliam, and E. L. Charnov. 1977. Optimal foraging: a selective review of theory and tests. Quarterly Review of Biology 52:137-154.

Quiring, D. T. and J. N. McNeil. 1987. Foraging behavior of a dipteran leaf miner on exploited and unexploited hosts. Oecologia 73:7-15.

Redfearn, A. and S. L. Pimm. 1988. Population variability and polyphagy in herbivorous insect communities. Ecological Monographs 58:39-55.

Salisbury, F. B. and C. W. Ross. 1985. Plant physiology, 3rd edition. Wadsworth, Inc., Belmont, California, USA.

Schluter, D. 1981. Does the theory of optimal diets apply in complex environments? American Naturalist 118: 139-147.

Schowalter, T. D. 1985. Adaptations of insects to disturbance. Pages 235-252 in S. T. A. Pickett and P. S. White, editors. The ecology of natural disturbance and patch dynamics. Academic Press, New York, New York, USA.

Simms, E. L. and M. D. Rausher. 1989. The evolution of resistance to herbivory in Ipomoea purpurea. II. Natural selection by insects and costs of resistance. Evolution 43:573-585.

SPSS Inc., 1988. SPSS-X user's guide, 3rd edition. Chicago, Illinois, USA.

Stanton, M. L. 1982. Searching in a patchy environment: foodplant selection by Colias p. eriphyle butterflies. Ecology 63:839-853.

Stephens, D. W. and J. R. Krebs. 1986. Foraging Theory. Princeton University Press, Princeton, New Jersey, USA.

Strong, D. R., J. H. Lawton, and Sir Richard Southwood. 1984. Insects on plants. Blackwell Scientific Publications. Oxford, England.

Taghon, G. L. 1981. Beyond selection: optimal ingestion rage as a function of food value. American Naturalist 118:202-214.

Van Den Bussche, R. A., M. R. Willig, R. K. Chesser, and R. B. Waide. 1989. Genetic variation and systematics of four taxa of neotropical walking sticks (Phasmatodea: Phasmatidae). Proceedings of the Entomological Society of Washington 90:422- 427.

Vitousek, P. M. 1985. Community turnover and ecosystem nutrient dynamics. in S. T. A. Pickett and P. S. White, editors. The ecology of natural disturbance and patch dynamics. Academic Press, ONew York, New York, USA.

Williams, K., C. B. Field, and H. A. Mooney. 1989. Relationships among leaf construction cost, leaf longevity, and light environment in rain-forest plants of the genus Piper. American Naturalist 133:198-211.

Willig, M. R. 1983. Optimality theory: individual variability and the determinants of food preference in a tropical folivore. Unpublished Manuscript. 46 Willig, M. R., R. W. Garrison, and A. J. Bauman. 1986. Population dynamics and natural history of a neotropical walking stick, Lamponius portoricensis Rehn (Phasmatodea: Phasmatidae). Texas Journal of Science 38:121-137. APPENDIX

The manner in which raw data were manipulated for use with the SPSS-X program

MANOVA is described herein. The original data represent the area of a 500-mg portion of leaf before ill) and after (A) a single night of feeding by a single walking stick. However, quantity A reflects area lost because of desiccation overnight.in addition to area lost as a result of leaf consumption. Since leaf area lost because of desiccation can be substantial

(up to 28% in PH), A was corrected to account for the confounding effects this loss creates. Five additional 500-mg sets of leaf portions (for each type of leaf used in the experiment) were cut each night, placed in wire mesh containers, and measured the next morning with those pieces which were offered to walking sticks. The average percent of area lost from these five sets of leaf portions and any 500-mg portions left undisturbed by walking sticks represents the average leaf area lost by desiccation alone. This value, unique to each type of leaf for each night of a feeding trial, was subtracted from 1.00 to obtain a value representing the average percent of leaf remaining after desiccation. This number is the Desiccation Correction Factor (DCF) and is given by

Y = {B- [A (100%/DCF)]}B -1 (500mg).

Thus, the quantity in brackets, the corrected value for A, represents leaf area consumed by a walking stick. The quantity Y is the total mg of leaf (for each type of leaf in the experiment) eaten by a walking stick on one night of a feeding trial. Values for Y were calculated using unique DCFs for all data collected. 47 48 Three-day average percent consumption values for each leaf type used in an experiment were calculated in two steps. First, values of Y for each day of an experiment

were converted into daily percent consumption values for each leaf type. For example,

percent consumption of top leaf portions for day 1 (%Top 1) was calculated by dividing the

amount of top leaf eaten by the sum of top, middle, and bottom leaf portions consumed for

day 1. In this way, values for daily percent consumption by an individual walking stick

were calculated for each leaf type. In the example, a total of nine values (three leaf types

for three days) was calculated. Second, three-day average percent consumption values

were calculated for each leaf type in the experiment. For example, %TOP was calculated

by first adding each daily percent value for top leaves (%Topl + %Top2 + %Top3) and

dividing this sum by the sum of daily percent consumption of all leaf types (%Topl +

%Top2 + %Top3 +%Middle! + ... %Bottom3). The resulting number represents the

average percent of the diet composed of that particular leaf type over three days of feeding

by a single individual. These data for all individuals were used by MANOVA to answer

questions outlined in Chapter II.