Citation
Population studies of the citrus snow scale Unaspis citri (Comstock)

Material Information

Title:
Population studies of the citrus snow scale Unaspis citri (Comstock)
Creator:
Arias Reveron, Julio M., 1960-
Publication Date:
Language:
English
Physical Description:
xi, 118 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Female animals ( jstor )
Instars ( jstor )
Internet search systems ( jstor )
Life tables ( jstor )
Mortality ( jstor )
Natural enemies ( jstor )
Parasitoids ( jstor )
Snow ( jstor )
Species ( jstor )
Temperature scales ( jstor )
Citrus -- Diseases and pests ( lcsh )
Citrus snow scale ( lcsh )
Dissertations, Academic -- Entomology and Nematology -- UF
Entomology and Nematology thesis, Ph. D
Scale insects ( lcsh )
City of Gainesville ( local )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1995.
Bibliography:
Includes bibliographical references (leaves 106-117).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Julio M. Arias Reveron.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
022272898 ( ALEPH )
34467124 ( OCLC )

Downloads

This item has the following downloads:


Full Text








POPULATION STUDIES OF THE CITRUS SNOW SCALE
UNASPIS CITRI (COMSTOCK)










By

JULIO M. ARIAS REVERON














A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1995

UNIVERSITy OF FLORIDA LIBRARIES













ACKNOWLEDGMENTS


This research would have not been possible without the guidance, critical evaluation, encouragement and financing by the committee chair, Dr. Harold W. Browning. I appreciated especially his immense patience. Very valuable input was received from the other members of my advisory committee. Dr. Jon C. Allen helped me understand the intricacies of computer modeling, Dr. Carmine A. Lanciani advised on life table analysis during the planning phase of this project, Dr. Clay W. McCoy taught me about insect pathology and the complexities of the researcher's occupation. Dr. Fred Bennett was instructive about biological control and quarantine procedures, and helpful while he was part of the committee, until the day of his retirement.

This research was possible also by the kindness of Mr. Maurice Patrick who gave me the liberty to do and undo at the citrus plots of his property. During the work for this research I got invaluable help from Mrs. Pamela Russ, Mr. Ian Jackson and Mr. Mark Bryan and had the opportunity to share their friendship. Many people not directly involved in the development of this research need to be thanked, because they greatly contributed to my well being during my stays in Gainesville and Lake Alfred: these include Jackie J. Belwood, Vinnod Kutty, the Nielsen family, Faith and David Oi, Eliane Quintela, Devesh Singh, Hugh Smith and Laurie Wilkins:


ii








without their friendship I could not have succeded. I am perpetually indebted to Ms. Lois Wood, for her love and friendship and her invaluable editorial help.

Finally, I need to thank my father Julio, my mother Isabel and my son Gabriel for their love, support and patience. They never gave up on asking "When are you going to finish?", but they never doubted that I would finish it.


































iii













TABLE OF CONTENTS


pAge

ACKNOWLEDGMENTS .................................... ii

LIST OF TABLES ......................................... vii

LIST O F FIG URES ........ ............................... ix

ABSTRACT ............................................. x

CHAPTERS

1- INTRODUCTION ..................................... 1

2 LITERATURE REVIEW .................................. 4

Taxonomy of the Genus Unaspis ......................... 4
Economic Importance ................................. 6
Biology ......................................... 6
D ispersal .. ............. ........................... 10
Phenology .......... ............................ 10
Trophic Relationships ................................ 11
Host Plants ................................... 11
Natural Enemies ............................... 12
Management ....................................... 15
Sampling and Forecast .......................... 15
Chemical control ............................... 16
Biological control ............................... 16

3 DEVELOPMENT AND MORTALITY
UNDER CONSTANT TEMPERATURE
AND RELATIVE HUMIDITY ............................ 18

Introduction ................................. ...... 18
Materials and Methods ................................ 20
Results ....................................... 24


iv








D iscussion .......................................... 30

4 STUDY OF MORTALITY OF THE CITRUS SNOW SCALE
UNDER FIELD CONDITIONS, BASED ON THE ANALYSIS
OF PARTIAL LIFETABLES ............................ 35

Introduction ........--.................................35
Materials and Methods ................................ 36
Data Recording with Photographs .................. 36
Photographic Analysis ........................... 37
Statistical Analysis .............................. 41
Results ..........................................42
Discussion .......................................47

5 COMPARATIVE STRUCTURE OF CITRUS SNOW SCALE
POPULATIONS ON THE CANOPY OF CITRUS TREES
AND THROUGH TIME ................................ 54

Introduction .......................................54
Materials and Methods ................................55
Results ............. ............................58
Discussion ......................................65

6 THE ROLE OF WIND IN CRAWLER DISPERSION ............ 70

Introduction .......................................70
Methods ........................................70
R esults .. .............. ........................... 72
Mylar traps .................................. 72
Infestation Scoring Maps .........................76
Discussion .......................................76

7 MODELING CITRUS SNOW SCALE POPULATIONS ........... 82

Introduction .................... ... ................ 82
Materials and Methods ................................ 83
M odel Structure ................................ 83
Development .................................85
Mortality ....................................85
Reproduction and Migration ....................... 86
Environmental Variables ......................... 86
Model Implementation .......................... 87
Model Calibration .............................. 87
Model Validation ...............................87
Results ................ ..........................88

V








Discussion ..................... ..................... 93

8- CONCLUSIONS ....................................... 98

APPENDIX: SIMULINK BLOCK DIAGRAMS
FOR THE SNOW SCALE MODEL ...................... 102

REFERENCES CITED .................................. 106

BIOGRAPHICAL SKETCH ................................ 118







































vi













LIST OF TABLES


Table page

3.1. Temperature and relative humidity means sd
conditions used in the study of citrus snow scale
development ....................................... ..21
3.2. Duration of development for citrus snow scale at
different constant temperatures and relative
humidities ...................... .............. 25
3.3. Parameters calculated to fit citrus snow scale
developmental data to the modified Logan
developmental model (Lactin et al. 1995). ................. 27
3.4. Effect of constant temperature across constant relative
humidity on mortality of citrus snow scale. ................. 28
3.5. Effect of relative humidity across constant temperatures
on mortality of citrus snow scale. ....................... 28
3.6. Parameters calculated to fit citrus snow scale mortality
data to a gamma distribution function. ................. 29
3.7. Polynomial regressions between temperature and the
parameters of a gamma distribution (a, 13) fitted to the
probability density function of the survival curves ............ 30
4.1. Significance of log-rank and Wilcoxon homogeneity
test, likelihood ratio test for comparison of survival of
citrus snow scale in patches within trees ................... 44
4.2. Significance of log-rank and Wilcoxon homogeneity
test, likelihood ratio test for comparison of survival of
citrus snow scale between trees by dates. ............... 45
4.3. Outcome observed for citrus snow scale individuals
under field conditions......................................... 47
4.4. Total percentage of mortality (adx) and percent mortality
by unknown causes against parasitism and detectable predation in multiple decrement life tables, calculated
by tree. ................. ....................... 48
4.5. Total percent mortality (all causes) adx presented by
stage ............................................. 49
6.1. Frequency and speed of wind on the four quadrants in
Lake Alfred, CREC weather station ......... .......... 75

vii








7.1 Parameters used for the simulation of development for
the citrus snow scale ................................. 88
7.2 Average duration and percentage of survival from each
stage at each environmental condition studied ............. 89
7.3 Parameters for the distributed delay model, calculated
from data at constant temperature and RH ............... 90
7.4 Linear regression between attrition rate (AR) and
temperature (T). ..................................... 91
7.5 Parameter values, standard error and confidence
intervals at 95% for development parameters. ............ 92
7.6 Parameter values, standard error and confidence
intervals at 95% for mortality parameters ................... 93


































viii













LIST OF FIGURES


Figure page

4.1 Survival and hazard function of snow scale populations
on selected trees .................................... 46
5.1 Age structure of citrus snow scale populations on citrus
leaves and twigs ..................................... 59
5.2. Ratio of first instar per gravid female of snow scale on
samples of citrus twigs and leaves ....................... 60
5.3 Relative proportion of sexes of snow scale on samples
of citrus leaves and twigs ............................... 62
5.4. Incidence of natural enemies of snow scale on samples
of leaves and twigs.................................. 63
5.5 Relative importance of natural enemies of snow scale
females on samples of leaves and twigs...... ........... 64
6.1. Wind speed vs direction and catches of snow scale
crawlers during trial 1 and 3 ............................ 73
6.2. Wind speed vs direction and catches of snow scale
crawlers during trial 5 and 6 ........................... 74
6.3. Maps of trees infested with citrus snow scale in a citrus
plot (CREC, N-40, block 16)........................... 77
7.1. Duration of development of citrus snow scale
individuals under constant temperature. ................... 94
7.2. Duration of development of citrus snow scale
individuals under greenhouse conditions ................... 95













ix













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

POPULATION STUDIES ON THE CITRUS SNOW SCALE UNASPIS CITRI (COMSTOCK) By

Julio M. Arias Rever6n

December, 1995

Chairman: Harold W. Browning Major Department: Entomology and Nematology

The goal of this research was to expand knowledge of the biology and ecology of the citrus snow scale Unaspis citti (Comstock) (Homoptera: Diaspididae), including duration of development on a suitable substrate, causes of mortality under field conditions, life table analyses, dispersal mechanisms, and computer simulation of developmental dynamics.

Development of snow scale is temperature dependent. The optimal temperature for development fell in a range of 25 to 38*C, near 290C for most stages. Development was faster on grapefruit leaves than previously reported on lemon fruits. The effect of relative humidity was only significant for first instar males, but the range of relative humidity was too narrow (10%) to observe more differences.



x








Field studies were accomplished with the use of macrophotography as a sampling tool to study snow scale colonies on limbs and trunk, and destructive leaf and twig samples to study snow scale populations within the canopy. Little evidence of Aphytis lingnanensis Compere (Hymemoptera: Aphelinidae) causing mortality on citrus snow scale on trunk and limbs was observed. Encarsia spp. were observed often, attacking both immature males and second instar females. Predators were evident via disrupted scale covers of some stages. Attack by Encarsia ranged between 0 and 38%, while observable attacks by predators ranged between 0 and 49%. Analysis of data from leaf and twig samples suggests that different conditions within trunk and canopy may affect mortality factors. Aphytis lingnanensis and fungi were present in most of the samples taken during the 2-year study. These played a more important role on snow scale populations found on leaves and twigs. Mortality caused by Aphytis lingnanensis was higher on snow scale in the canopy, which suggests that although this aphelinid has an effect on canopy populations, the main scale population on trunk and limbs is unaffected. Crawlers were found to disperse, taking advantage of air currents, as has been reported for other scale insect species.

A simulation model of inmature development of citrus snow scale was derived from experimental data obtained under constant temperature conditions. Since the model describes only part of the life cycle, further research and model development is needed to better describe snow scale population dynamics.





xi













CHAPTER 1
INTRODUCTION


Few animal species are able to manage their environment to maximize the collection of resources needed for their survival, other than gathering those resources readily available. Agriculture is the process of managing ecosystems, the manipulation of vegetal and animal species to maximize the development and reproduction of a useful species. Efficiency in management techniques has been achieved through natural selection within fungus-culturing ants, but for humans, it has been attained by other means of selection: trial and error, the transfer of traditional knowledge and research.

It is difficult to manage a system for which we have little understanding. Solutions chosen without knowledge will very likely lead to waste of resources or to unexpected secondary effects, as happens with development of pesticidal resistance and pollution produced by the indiscriminate use of pesticides. Research leading to an increase in our understanding of a natural or artificial ecosystem is an essential component of the design of management strategies.

Another characteristic of agriculture is that successful crops are usually planted in regions far away from their area of origin (Kloppenburg and Kleimman 1987). This separates the crop plants from organisms that may have coevolved with the plant and use their tissues as food resources. When one or more of these


1







2

organisms make their way to a new area where the host plant is being cultivated as a crop, they usually reach the area without their own natural enemies, and become free to colonize and exploit the resource plant without constraints of predation, parasitism or diseases. These organisms then become pests.

The citrus ecosystem in Florida is dominated by an exotic plant species brought to a geographical area with a benign environment, and where many of its pest organisms have been arriving one at a time. Management of these citrus pests has relied upon use of pesticides, but in the recent times, more pest species have been successfully managed by a technique of ecosystem reconstruction, biological control. Biological control consists of bringing to the new ecosystem some of the missing components from the original ecosystem, particularly the natural enemies that feed or develop on the insect pests. Biological control is a management technique that requires a very good understanding of the relationships between the plant, the pest and its natural enemies.

The citrus snow scale Unaspis citri (Comstock) is today an important pest of citrus in Florida. It was a minor pest in the past, but its relative importance has increased over the years, with geographical spread throughout the state and the successful biological control of other, once more important pests (Browning 1994). Knowledge about its biology is meager and attempts to control it with pesticides or with natural enemies have been unsuccessful. The present work is aimed at understanding details of the citrus snow scale life history and relationships with its surroundings that could contribute towards developing successful biological control. I was interested in expanding the knowledge of its biology, measuring the duration








3

of development on suitable substrata, elucidating causes of mortality that affect citrus snow scale in the field, and looking into the results of previous attempts of biological control and the potential reasons why those may have been unsuccessful.

The present work is organized into chapters along major areas of the research. The next chapter presents background information about the genus Unaspis, the species citri and its close relatives. It also introduce the techniques that were used in the research reported in later chapters. The third chapter describes the development of snow scale under constant temperature and constant relative humidity conditions, on a suitable substratum of a common Florida citrus variety. The fourth chapter discusses mortality factors affecting snow scale under field conditions on the bark of citrus trees. Using photographic techniques and survival table analysis, the role of natural enemies is characterized. The fifth chapter complements the fourth chapter with observations on mortality causes, age structure and sex ratios in the canopy of citrus trees. The sixth chapter looks into the potential of dispersion of snow scale via wind. The seventh chapter summarizes the snow scale knowledge in a simulation model that evaluates the quality of the new information and identifies gaps requiring additional research. Since each chapter is designed to stand alone as a single research paper, a certain amount of redundancy can be found between Chapter 2 (Literature review) and the introductory sections of each subsequent chapter.













CHAPTER 2
LITERATURE REVIEW


Published accounts of research on the citrus snow scale (Unaspis citri Comstock) are not extensive, probably because this insect has been considered a minor pest. Knowledge of this pest has not been a prime consideration in citrus pest management. The genus Unaspis includes eight described species and only three of these are of economic importance. The accumulated knowledge on the citrus snow scale and the other economically important species in the genus is reviewed here in an effort to link the knowledge about snow scale with literature accounts of other Unaspis species. Not surprising is the fact that very little information has been published on the other species of Unaspis that are not of economical importance.



Taxonomy of the Genus Unaspis.

The diaspidid genus Unaspis probably originated in Asia, in the continental region between southeastern India and eastern China, where the distributions of the economically important species converge (UK 1962, 1970, 1988). This genus was described in 1921 by A.D. MacGillivray. J. H. Comstock originally described Unaspis citri in 1883 from Citrus in Louisiana and placed it in the genus Chionaspis (Ferris 1937). Other synonyms for U. citri are Dinaspis annae Malenotti 1917,



4








5

Prontaspis citri (Comstock) MacGillivray 1921 and Dinaspis veitchi Green & Laing 1923. Ferris (1937) assigned those synonyms to Unaspis citri.

Unaspis citri is recorded from Citrus and related genera in the family Rutaceae and occurs in China, Indochina, Australia, Central Africa, South America and the Caribbean (UK 1962). The same source reports this species from the citrus areas of North America, but more recent information restricts it to Florida and Louisiana (Browning 1994). In addition to U. citri, the two other Unaspis species of economic importance are the arrowhead scale, U. yanonensis and the euonymus scale, U. euonymi. Unaspis yanonensis infests Citrus in Japan, China and southern France, but does not occur in the Americas or Australia (UK 1988). Unaspis euonymi attacks ornamental plants in the genus Euonymus (Celastraceae), Prunus (Pomaceae) and Hibiscus (Malvaceae). It occurs in Asia (Japan, China), Europe and North America (UK 1970). The remaining species, Unaspis acuminata, U. atricolor, U. flava and U. permutans were described from material collected from various host plants in Sri Lanka and the south and east of India, while U. turpiniae was described from material collected in the Philippines (Rao 1949).

The diagnostic characteristic that separates U. citri from the other species of the genus is the reduction or absence of perivulvar glands in mature females (Rao 1949). Perivulvar glands are associated with ovoviviparity. Although some degree of ovoviviparity occurs in most diaspidid scale insects, the bounds of this phenomenon are not well delimited. Prevailing theory indicates that part or all of the embryo's development could occur internally in the mother of an ovoviviparous species. Conversely, the presence of an egg shell or chorion is considered to be








6

an indication of oviparity. The perivulvar glandular system may be associated with the presence of the egg shell and thus its reduction in some species is thought to parallel progressing viviparity (Koteja 1990a).



Economic Importance

Unaspis citri is considered an important pest of citrus in the Americas and Australia, based on the stress that this species inflicts on the general health of the tree and on the cost of its control. Bark and scaffold limbs are the primary sites of attack on citrus trees. However, in heavy infestations, it also attacks leaves and to a lesser extent fruits. Bark splitting, twig dieback and tree death are also associated with severe infestations (Beattie & Gellatley 1983, Smith & Papacek 1985, Browning 1994).

Unaspis yanonensis is an important pest of citrus in Japan, where the increase of populations in the absence of effective natural enemies causes damage to citrus fruits (Ohkubo 1981). Unaspis euonymi is very important in landscape horticulture in North America and Europe, producing chlorosis, reduced photosynthesis, leaf abscission and stunting in Euonymous plants (Vinis 1977, Gill et al. 1982, Brewer & Oliver 1987, Cockfield & Potter 1987).



Biology

The biology of citrus snow scale, as studied by Dickens (1968), typifies the life cycle of many other diaspidid scales. Females have two molts, or two nymphal stages and the imago or adult stage. Males have four molts leading to two nymphs,







7

two pupal stages and the imago. The first instar (crawler) and the adult male are the only mobile stages. The crawlers search for a location to settle and feed, where they will remain throughout their development. Second instar females secrete a proteinaceous shield, while males produce a white waxy tricarinated shield. After the second molt, females reach adulthood and produce a dark shell-like secretion that continues to enlarge until oviposition begins. Conversely, males complete feeding in the second instar and molt to the first pupal stage. Subsequently, they progress through the third molt to a second pupal stage and then after the fourth molt a winged male emerges (Dickens 1968, Koteja 1990b).

Female citrus snow scales reach reproductive maturity and deposit their eggs beneath the secreted shield. The incubation period is very short, probably between 30-60 minutes, suggesting that most egg development occurs inside the body of the female. Ovoviviparity is presumed to be an ancestral condition associated with tropical environmental conditions, while retarded embryonic development and oviparity, often associated with winter diapause, is considered a secondary adaptation to temperate regions (Koteja 1990a). Unaspis citti embryos may be laid enclosed in membranes instead of egg shells (Koteja 1990a), since there is no trace of a chorion after the crawler ecloses.

The duration of development of poikilotherm organisms is dependent upon temperature, and the relationship between development and temperature is a fundamental feature of an insect life history (Taylor 1981). To characterize the developmental rate, it is necessary to study the duration of development under a range of enviromental conditions and to describe a relationship with temperature








8

and other relevant factors. The relationship between temperature and development has been described in many ways, including linear relationships (Baskerville & Emin 1969, Sevacherian et al. 1977), and curves and asymmetrical functions (Logan et al. 1976, Sharpe & DeMichele 1977, Taylor 1981, Wagner et al. 1984). The developmental rate increases with an increase of temperature. Development stops below a lower and above an upper temperature threshold.

Dickens (1968) studied duration of development of the citrus snow scale on 'Valencia' orange fruits, at a constant temperature of 26C. He found that first instar development may require around 13 days, while the duration of the second instar is about 11 days for males and 18 days for females. Females started producing eggs about 60 days after they were born. Casares (1974) studied citrus snow scale development on lemon fruits under several constant temperatures, and confirmed that developmental rate increased with an increase in temperature. These efforts to describe citrus snow scale development were made on citrus fruits, which are not the preferred substrata under field conditions.

Development of U. yanonensis has been studied extensively, including development under constant temperatures (Okudai et al. 1971, 1974, Huang et al. 1983) and fluctuating temperatures (Korenaga et al. 1976). The optimal temperature for development was calculated to be about 270C and the lower developmental threshold was near 10C (Okudai et al. 1971, 1974). Studies also addressed the relationship between temperature and ovarial development, forecasting the appearance of the first instar (Nishino & Furuhashi 1971a, b, Okudai et al. 1975). Nishino (1974) has summarized most of this research. Furuhashi








9

(1974) documented survival and developmental differences induced by host plant variety. Gill et al. (1982) reported observations on U. euonymi obtained from field sites, but not from constant-temperature studies.

Fecundity of armored scale insects is not easy to study since births occur concealed from the observer beneath the scale cover. Nonetheless, fecundity of snow scale was described in Australia by Waterhouse & Norris (1987). They reported a maximum of 169 crawlers produced during 5 months. Dickens (1968) observed a mean of 108 crawlers with a maximum of 250-256, in Florida and under laboratory conditions. Longevity of gravid females was reported by Dickens (1968) to average 125 days (4.2 months).

Embryogenesis and fecundity of the arrowhead scale has been described by Nishino & Furuhashi (1971a) and Adachi & Korenaga (1991), with Adachi & Korenaga (1991) reporting a bimodal fertility curve. This was explained by the fact that eggs remain in the ovary until embryonic development is almost completed, and as well-developed eggs fill the ovary, egg production ceases. The production of new eggs resumes when mature eggs are deposited.

The sex ratio of U. citn is strongly biased toward the males, with ranges of 36 males per female reported by Dickens (1968) and Casares (1974). The sex ratio of U. euonymi was described as 7 males per 3 females (Benassy & Pinet 1972), but Gill et al. (1982) and Cockfield & Potter (1987, 1990) report variable sex ratios depending upon the plant organ infested.








10

Dispersal

Wind is presumed to be the main means of dispersion of scale insects between plants, as has been shown for other species (Greathead 1990). It also has been proposed that scale insect crawlers disperse on workers' clothing and equipment (Simanton 1973) or passing animals (Greathead 1990). Xinnian & Browning (1991) observed that citrus snow scale crawlers were positively phototactic and that female crawlers walked farther than males. This behavior could facilitate the expansion of colonies, decrease intraspecific competition between females, which feed for longer periods than males, and facilitate the movement of first instars upwards on the tree, where they might be more easily dislodged and carried away by wind. Similarly, behavior of U. yanonensis crawlers has been studied by Korenaga (1983) and Wang & Chen (1989).



Phenology

Citrus snow scale does not exhibit diapause in Florida, and thus it reproduces throughout the year and has overlapping generations. At least two population peaks have been described in Florida (Dickens 1968, Casares 1977) and in Australia (Summerville 1935, Anon. 1954, Beattie & Gellatley 1983). Arrowhead scale and euonymus scale show synchronized generations, with 2 or 3 being reported per year, depending on temperatures. Arrowhead scales can overwinter in several life stages (Nishino 1974), while euonymus scales overwinter as adult females (Williams et al. 1977, Gill et al. 1982).









Trophic Relationships

Host Plants. Citrus species are the primary hosts plant for snow scale, although the scale has been reported from hosts in related genera of Rutaceae such as Murraya paniculata, Severina buxifolia, and Fortunella sp. (Rutaceae) (Casares 1974, Williams & Watson 1988). Reports of infestations of citrus snow scale on plants in other families (Palmae, Celastraceacea, Oleacea, Bromeliacea) may be inaccurate (Casares 1974, 1977) and represent misidentification of other armored scale species with similar appearance.

Mechanisms of feeding in diaspidid scales are not well understood. It has been proposed that the stylets penetrate woody tissues to find active sieve tubes close to the cambium. Another suggested mechanism is that the scales feed on the contents of any cell invaded by the stylets. The third proposal is that both forms of feeding may occur. The first mechanism does not explain why diaspidid scales do not produce large amounts of honeydew secretions. It is hypothesized that unused materials may be returned to the plant by the insect pumping them with its powerful salivary glands (Banks 1990). Studies on U. euonymi support the second hypothesis, intracellular cell feeding on palisade parenchyma in leaves, and feeding on xylem tissue in limbs (Sadof & Neal 1993). Citrus snow scale damage and feeding mechanisms were studied by Albrigo & Brooks (1977). They showed that stylets penetrate the plant tissue intracellularly, damaging cells on the way to locating the phloem vessels, but they assumed phloem feeding without confirming it.








12

It has been shown that differential preferences exist for the substrate chosen by each sex of U. euonymi. Therefore, sex ratio was observed to be different on different organs of the plant (Benassy & Pinet 1972). Males prefer to settle on leaves, while females prefer to settle on woody organs. This could be related to intraspecific competition and risks of mortality associated with leaf abscission. Since females need to feed for a longer period than males, it may be argued that females prefer to settle on twigs, which are a more permanent substrate than leaves. Males on the other hand, tend to settle on leaves to avoid competition with females and since they only feed for the first two instars, they may survive even if the leaf falls from the plant (Cockfield & Potter 1987). Citrus trees have perennial foliage and leaves remain on the tree for long periods, so the ratios or settling preferences could be different for snow scale.

Natural Enemies. Natural enemies of snow scale have been reported in a variety of publications and have been reviewed by Waterhouse & Norris (1987). A discussion of the more important species follows.

Encarsia (=Aspidiotiphagus) citrina (Craw.) and E. lounsburyi (Berlese and Paoli)(Hymenoptera: Aphelinidae) are the most commonly found and probably most widely distributed citrus snow scale parasitoids (Hely 1944, Casares 1977, Selhime & Brooks 1977, Terdn et al. 1985, Castihreiras & Obreg6n 1986, Fernandez Argudin 1987). They are polyphagous endoparasitoids that have been closely studied in relation with other species of host scales (Benassy & Pinet 1972, Kajita 1972, 1976, 1977a, b, Murakami et al. 1972, Kanda & Kajita 1977, Gill et al. 1982). Encarsia








13

citrina and Encarsia longsburyi prefer to attack young second instars of both sexes of citrus snow scale.

Aphytis lingnanensis Compere (Hymenoptera: Aphelinidae) preferentially attacks late second instar or young adult (prereproductive) females of citrus snow scale. It is also a polyphagous parasitoid originally from Asia; and this species develops as an arrhenotokous ectoparasitoid. Several related forms or strains have been identified, morphologically indistinguishable from A. lingnanensis but with varying degrees of reproductive isolation (DeBach & Rao 1969). These strains have been widely used in attempts at biological control of several pests, including the introduction of the Hong Kong "HK-1" strain of A. lingnanensis into Florida for the control of snow scale and the introduction of the Hong Kong "HK-J" strain into Japan for control of the arrowhead scale (Rosen & DeBach 1979, Tanaka 1981).

Aphytis gordoni (Hymenoptera: Aphelinidae) is another ectoparasitoid, half the size of A. lingnanensis, discovered during exploration for snow scale parasitoids in Hong Kong. All of the material was obtained from U. citri hosts (Rosen & DeBach 1979), which could be an indication of host specificity and make this species worthy of more research as a potential biological control agent for snow scale.

Chilocorus circumdatus (Coleoptera: Coccinellidae) was recently reported feeding on citrus snow scale in Australia where it was originally introduced from China for the control of Aonidiella aurantii (Houston 1991). This coccinellid is a diaspidid scale feeder and was introduced during biological control projects for other species of armored scales in the U.S. (Rosen & DeBach 1977).








14

Fungal diseases have been reputed to play an important role in citrus snow scale population regulation, especially because crowded citrus snow scale colonies provide an ideal situation for the spread of disease. Several fungal genera have been reported in association with snow scale (Sphaerostilbe, Podonectria) (Dickens 1968). Most of those reports are of species now recognized within the genus Nectria (anamorph Fusarium, Samson et al. 1988) Some debate and contradictory results of experimentation emerged from the period between the end of last century to the late 1940s regarding the effectiveness of these fungi in the control of scale insect populations. Nectria species were confirmed as developing saprophytically on the bodies and covers of dead scales, but it is not clear whether these fungi also function as true pathogens (Ziegler 1949, Fisher et al. 1949, Fisher 1950).

Natural enemies are an important means of population regulation. The role of a given species can be studied by laboratory experimentation or by field observations and manipulations. Unfortunately, observations in the field could be influenced by many uncontrollable factors. One way to quantify mortality and its causes is by using life table analysis. Life tables consist of a systematic accounting of the number of individuals alive and dead at each age or stage and, when possible, the causes of death. Life table analysis is a technique borrowed from the insurance business that was applied to animal populations (Hutchinson 1978). Applications of life table analysis to insects are reviewed in Harcourt (1969), Varley et al. (1973), Southwood (1978) and Carey (1993).








15

Management

Sampling and Forecasting. The snow scale presents special problems for sampling, since it occurs mainly on citrus trunks and scaffold limbs. These substrata cannot be easily harvested and processed in the laboratory, as leaf samples are. Sampling methods for citrus snow scale include the use of rating systems for the appearance of scale colonies and the clearing of rectangular patches within active colonies on the trunk. The patches are colonized by scales, and then scales are counted in situ with the help of a hand lens (Casares 1977). The rating methods tend to overestimate the active scale populations since they rely on the appearance of male covers which can accumulate on the trunk for extended periods, and thus represent both live scales and those that have already emerged. Female citrus snow scales are generally overlooked in rating methods since they are difficult to see on citrus bark.

Extensive efforts have been invested in creating models that predict population densities of the arrowhead scale, U. yanonensis (Takezawa & Uchida 1969, Nishino & Furuhashi 1971a, b, Korenaga et al. 1974, 1976, Nishino 1974, Korenaga & Sakagami 1981, Sakagami & Korenaga 1982, Adachi & Korenaga 1992). These models also predicted the results of interactions between biological and chemical control strategies (Adachi & Korenaga 1992). Simulation models have been used to predict and manage other species of scale insects as well (Pfeiffer 1985, McClain et al. 1990a, b).








16

Chemical control. Since snow scale has overlapping generations in Florida citrus, it is extremely difficult to effectively target pesticidal applications to the most susceptible stages, as is done for the arrowhead scale (Adachi & Korenaga 1992). The recommendations for materials and equipment for chemical control are given by Knapp (1995). Recommendations include applying pesticides with hand-held equipment in localized areas where the scale is a problem and trying to achieve the maximum coverage possible.



Biological control. Biological control of citrus snow scale in Florida began in the 1970's. The first recorded introductions of E. citrina and A. lingnanensis occurred in 1969. A. lingnanensis "HK-1" was introduced into Florida in 1972, and was reported as being established and effectively controlling U. citri (Riehl et al. 1980). During the period 1974-1985, approximately 6.5 million parasitoids were released, and savings early in the program were estimated at 8-10 million dollars in pesticide applications (Mead 1976). The program of mass release of HK-1 was discontinued because of the effectiveness of the parasitoid was not demonstrated, and new efforts to identify alternative solutions have begun (Browning 1994). Several other strains of A. lingnanensis and two other species, A. khunti, and A. yanonensis, have been introduced into Florida, but establishment has not been confirmed (Browning 1994).

Unaspis yanonensis became a very serious problem in Japan after it became established there without natural enemies (Ohkubo 1981, Takagi 1981). Aphytis yanonensis and Physcus fulvus (Hymenoptera: Aphelinidae) were introduced from








17

China in 1980 and showed good results (Furuhashi & Nishino 1983). A coccinellid, Chilocorus kuwanae, and a nitidulid, Cybocephalus sp., have been introduced from Korea into the U.S.A. for the control of Unaspis euonymi and field released in 1984 (Drea & Carlson 1987, 1988).













CHAPTER 3
DEVELOPMENT AND MORTALITY
UNDER CONSTANT TEMPERATURE
AND RELATIVE HUMIDITY


Introduction

Citrus snow scale, Unaspis citri (Comstock), is regarded as a pest of citrus in Florida and in many other tropical and subtropical citrus production areas (Dickens 1968, Smith and Papacek 1985, Rose 1990). The relative importance of the snow scale in Florida increased over the period 1963-1971 (Simanton 1973) because of improved management of other important pests and the spread of infested nursery stock throughout the state (Browning 1994).

The life cycle of the snow scale described by Dickens (1968) is typical of most armored scales. The sexes are dimorphic; females proceed through 2 nymphal stages before reaching maturity, whereas males have 2 nymphal stages similar to females, and then 2 non-feeding stages, called the prepupa and pupa. Development under constant temperatures was studied on citrus fruits by Dickens (1968) and Casares (1974). However snow scale is seldom found on fruit in the field, living mostly on trunks, branches, twigs and, to a lesser extent, foliage. Casares (1977) conducted ecological studies on the phenology of the scale and concluded that rainfall, variety and tree age are important factors in determining snow scale population levels.

18








19

Diverse approaches have been taken to simulate development of poikilothermic organisms, from simple assumptions of linear dependency between temperature and development to more complicated nonlinear equation models (Curry and Feldman 1987). Developmental processes depend on temperaturedependent chemical reactions and are bounded by upper and lower thresholds. Below lower temperature thresholds, developmental processes are slow or halted; likewise, above upper temperature thresholds development is retarded, proteins are denatured and subsequent death of the organism occurs (Sharpe et al. 1977). Lower developmental thresholds are easily calculated when the developmental rate is assumed to be linear. However, when the rates are assumed to be nonlinear, the developmental functions that have been proposed approach zero asymptotically, making it impossible to deduce a lower threshold from the mathematical relationship estimated (Logan et al. 1976, Wagner et al. 1984). Development at low temperature may proceed so slowly that it may not be perceptible to the observer, and most models cannot account for this phenomenon. Lactin et al. (1995) proposed a modification of Logan et al. (1976) that estimates lower developmental thresholds. This modification adds a parameter that changes the scale of the curve, allowing it to intercept with the x-axis (zero development).

In an effort to achieve effective control of citrus snow scale, researchers are currently striving to better understand its biology and ecology. The current study relates development and mortality to temperature and relative humidity under laboratory conditions, and adds new information by using a substrate similar to that inhabited by the scale under natural conditions.








20

Materials and Methods

Snow scale development and mortality were studied on leaves of 'Duncan' grapefruit seedlings, Citrus paradisi, under 9 constant temperature and relative humidity conditions. The seedlings were planted in 150-cm3 plastic conical containers and maintained using standard nursery practices. These seedlings were infested with snow scale by attaching them to the trunk of infested trees in the field (Casares 1974). Infested seedlings were carefully inspected to eliminate possible contaminants (mealybugs, mites, whiteflies, other species of scales, and possible predators of the established 1st instar), and held in cages.

Male scales and immature stages were removed after F, crawler activity was evident, leaving only reproductive females on the seedlings. Seven to 10 seedlings bearing gravid females were then placed in each environmental chamber. The gravid females were removed after 48 h, and the newly settled crawlers were located and mapped. These cohorts were examined under a dissecting microscope at 48 to 72 h intervals, depending on the temperature of the chamber. The observations continued until insects reached the adult stage.

Eight computer-controlled Florida Reach-In chambers (Walker et al. 1993) and another chamber (Lunaire, Lunaire Environmental, Williamsport, PA) were used in the studies. The 8 Reach-In chambers were set at combinations of 4 temperatures and 2 relative humidities, with a constant photoperiod of 12:12 h for each treatment (Table 3.1). The additional incubator (chamber 45, Table 3.1) was set at 300C and relative humidity was maintained near 60% with the use of saturated NaCI solutions (Carr and Harris 1949, Greenspan 1977, Marcandier and








21

Table 3.1. Temperature and relative humidity means + sd conditions used in the
study of citrus snow scale development.

Chamber
Temp. OC RH % no.

17 15.95 1.00 60.11 4.11 18 20.95 1.06 59.86 3.06 19 16.10 0.98 71.84 5.72 20 20.96 0.94 69.85 3.15 41 24.00 0.10 60.07 1.29 42 27.99 0.18 60.00 0.46 43 24.06 0.64 69.70 3.55 44 27.98 0.39 69.59 1.83 45 29.77 1.14 56.28 7.55
Conditions in chambers 17-44 (Florida Reach-In) were logged in by the controlling computer every 10 min, conditions in chamber 45 (Lunaire) monitored with hygrothermograph and the statistics calculated from 2-h intervals.



Khachatourians 1987). Conditions inside the Florida Reach-In chambers were monitored at 10-min intervals with built-in sensors, downloaded to a file and plotted weekly. Conditions in the Lunaire chamber were monitored using a hygrothermograph (model H-302, WeatherMeasure, Sacramento CA).

Developmental time and survival for each stage and sex, from settled crawler to adult, were recorded at each observation date. Molts were easily recognized under stereomicroscopy. Females showed darkening and separation of the exuviae; males expelled the exuviae from under the armor at the distal end. Female development from the last molt to the reproductive stage and egg development were not assessed because these periods could not be observed without lifting the armor and disrupting the insect.







22

The modification of the Logan et al. (1976) nonlinear model by Lactin et al. (1995) was chosen to fit the data. This model was preferred because it is descriptive, the parameters may be interpreted biologically, it is more realistic than linear models or symmetrical nonlinear models (Lamb et al. 1984), it is simpler to fit to the data than models with more parameters (Wagner et al. 1984), and it includes a developmental threshold. The modified Logan model has 4 parameters to describe the effect of temperature on the development of poikilotherm organisms:



p ( Tm T
d(T) eP T -e PT A A (3.1)





where T is temperature, p is the rate of increase to optimum temperature, Tm is the maximum lethal temperature, AT is the difference between the lethal temperature and the optimal temperature of development, and A is a parameter that makes the curve intercept the x-axis, allowing the estimation of a developmental threshold. In the Logan model, temperature values in Celsius are transformed to the base temperature, the values for duration of development are inverted to calculate rates of development, and the means of log transformed rates are used to calculate the model parameters (Logan et al. 1976). In this research, as in the work of Lactin et al. (1995), all those manipulations were avoided, minimizing the errors that may arise from computation (Kramer et al. 1991). The inverse equation of the model was used to fit the original mean duration of each stage (in days), instead of using the








23

rates. The mean durations were weighted by the sample size. Temperature was not transformed to the base temperature because this substraction is simply a change in scale and should not produce any effect on values of the parameters when fitting the curve. The curves were fit using the Marquart method (PROC NLIN, SAS Institute 1989).

Mortality was analyzed through life table methods, using insect age as the time variable. Survival curves for the 1st and 2nd nymphs were calculated by pooling data for both sexes. For male prepupal and pupal stages and virgin and gravid females, stages were pooled and the survival curves calculated for each sex. This was done for practical purposes because sexes are difficult to separate for 1 st and 2nd instars and stages are difficult to separate for later instars. Differences between the curves were tested using the Wilcoxon test, which is more sensitive to differences at early survival times and the Savage (log-rank) test, which is more sensitive to differences at later survival times (PROC LIFETEST, SAS Institute 1989).

The rate at which the survival curve decays with respect to time corresponds to the 1st derivative of the function (the slope of the tangent to the curve), or the probability density function of the survival curve (SAS Institute 1989). This function was calculated for each stage and sex and fitted, using the DUD method (PROC NLIN, SAS Institute 1989), to a gamma distribution model (Curry and Feldman 1987):








24


M (t) = 13a ta- 1 e-Pt (3.2) r (a)





where t is age and a and 13 are parameters that govern the shape of the distribution. The parameters obtained were related to temperature using regression of polynomial equations (PROC REG, SAS Institute 1989).



Results

Table 3.2 presents average duration of development in days, with standard deviation and sample size for each sex and stage, at each combination of temperature and humidity.

The effect of temperature on developmental time was statistically significant, except for the pupal stage at 60% RH (Table 3.2). The length of each stadium decreased as temperature increased until an optimal temperature for development was reached. Duration of development then increased at temperatures higher than optimal. The effect of relative humidity was significant only for 1st nymphal males, where 70% RH expedited development by 1-2 d. Interaction effects were significant for 1st nymphs of both sexes and 2nd nymphal males (Table 3.2).

No significant differences were found between sexes in development of 1st stages under either relative humidity condition (F = 0.45, df = 1, 341; P = 0.51 for 60% RH, F = 0.80, df = 1, 247; P = 0.38 for 70% RH, covariance analysis using temperature as covariates, Steel and Torrie 1980). Gender differences were














Table 3.2. Duration of development for citrus snow scale at different constant temperatures and relative humidities
(mean sd in days). (F values from 2-way ANOVA for nymph I and II, 1-way ANOVA for prepupa and pupa.)


RH Temp n 1st stage 2nd stage n 1st stage 2nd stage n Prepupa Pupa % (oC) females females males males male male

16 17 11.8 3.6 16 51.4 3.6 60 10.8 3.5 29 61.8 4.6 3 5.0 1.7 0
21 8 7.0 0.0 6 22.5 3.7 39 7.6 1.4 12 34.5 6.6 5 6.0 2.1 6 5.0 1.6
60 24 25 5.1 1.4 22 18.8 2.1 55 5.2 1.7 12 23.0 2.5 9 3.3 2.6 8 4.8 1.8
28 22 5.7 2.4 22 14.5 1.8 49 5.7 1.7 26 18.4 2.1 12 4.3 1.9 9 4.7 2.0
30 16 8.1 5.4 11 18.9 6.4 53 5.8 2.1 7 19.1 2.1 3 2.0 0.0 2 5.0 1.4

16 18 10.8 5.0 17 52.8 6.3 58 12.5 4.7 14 58.7 4.5 0 0 21 21 9.1 5.3 21 26.1 7.2 54 8.1 2.6 34 32.6 3.7 13 4.8 2.3 10 7.8 2.1
24 14 7.0 0.0 14 17.9 1.8 42 7.1 0.3 1 28.0 1 2.0 5 3.2 2.7
28 7 6.4 1.1 5 15.2 2.3 36 6.2 1.0 1 39.0 0 1 4.0


Effects of temperature were significant for each stage (nymph I females: F = 13.969; nymph II females: F = 172.633; nymph I males: F = 79.826; nymph II males: F = 536.401; prepupa: F = 3.750 @60%; pupa: F = 11.458 @70%) except for pupa at 60% RH. Effects of RH were significant for nymph I males (F = 25.988). Interactions were significant for nymph I females: (F = 3.003) nymph I males (F=4.879) and nymph II males (F=17.020).







26

significant for the 2nd stage at both humidities (F = 44.31, df = 1, 160 for 60% RH, F = 30.66, df = 1, 104 for 70% RH, both P < 0.05), with females showing a shorter 2nd instar (Table 2). No significant differences in sex ratio were found at any of the environmental conditions (F = 0.33, df = 8, 67; P = 0.95). The average sex ratio ranged between 2.43 and 4.97 males per female, with a mean of 3.38.

Table 3.3 presents the parameters and R2 values for the fit to the modified Logan model to the developmental data. It also shows the optimal temperatures calculated for development of each stage and the estimated developmental threshold. Sample sizes of later instars were reduced by mortality in previous stages, yielding less precise estimation of the fitted curves. The optimal temperatures for development were between 25 and 380C for most stages, but the thresholds varied widely from 0.08 to 18.20C, with the extreme values being quite high and clearly unrealistic estimates.

Tables 3.4 and 3.5 present the effects of temperature and relative humidity respectively, on survival of snow scale. There were significant differences between the survival curves at different temperatures for all of the stages except adult females (Table 3.4), and relative humidity affected 1st instars at temperatures >16*C, 2nd instars only at 16*C and males at 24 and 280C.

Table 3.6 presents the parameters for the fit of a gamma distribution to the probability density function of the survival curves. Data were pooled for older stages (males and females). Table 3.7 presents the regression of the parameters a and 13 of the gamma distribution versus temperature using polynomial equations.









Table 3.3. Parameters calculated to fit citrus snow scale developmental data to the modified Logan developmental model
(Lactin et al. 1995).

Parameters Temp. (C)
Sex Stage RH n R2 Thresp Tm, AT A Optimal hold Female 1st stage 60 85 0.99 0.01 35.49 2.61 -1.09 32.8 8.0 70 60 1.00 0.11 37.07 9.01 0.00 28.1 --2"d stage 60 76 0.99 0.11 35.10 9.00 -0.10 26.1 13.0 70 58 0.98 0.11 34.67 9.00 -0.10 25.6 13.5 Male 1st stage 60 251 1.00 0.15 34.01 6.63 0.00 27.4 --70 191 0.99 0.01 37.00 3.00 -1.10 34.0 9.5 2"d stage 60 87 0.98 0.11 34.80 9.00 -0.10 25.8 13.2 70 51 0.95 0.11 34.01 9.00 -0.10 25.0 13.6 Prepupa 60 32 0.93 0.11 40.00 9.00 0.00 31.0 --70 14 1.00 0.10 34.32 7.46 -2.76 26.8 19.4 Pupa 60 25 1.00 0.01 44.26 6.03 -1.00 38.2 0.1 70 16 0.99 0.11 35.10 8.50 -0.90 26.5 18.2 Tm is the maximum lethal temperature, AT is the difference between the lethal temperature and the optimal temperature of development, and A is a parameter that makes the curve intercept the x-axis, allowing the estimation of a developmental threshold.








28

Table 3.4. Effect of constant temperature across constant relative humidity on
mortality of citrus snow scale. X2 values are for log rank test (above) and
Wilcoxon tests (below) (SAS Institute Inc., 1989).


Nymph I Nymph II Females Males
RH
S n X2 n X2 n X2 n X2
(%)

159.63*** 173.31*** 11.60* 83.86*** 60 1530 889 198 245 241.18*** 127.43*** 7.94 ns 45.39*** 4.47 ns 208.29*** 4.61 ns 26.56*** 70 926 681 143 182 15.72** 141.44** 4.52 ns 13.98**

= P < 0.001, **= P < 0.01, *= P < 0.05, ns = not significant.



Table 3.5. Effect of relative humidity across constant temperatures on mortality of
citrus snow scale. X values are for log rank test (above) and Wilcoxon tests
(below) (SAS Institute Inc., 1989).


Nymph I Nymph II Females Males
(C) n X2 X2 n X2 n X
(OC)

1.35 ns 12.87*** 0.00 ns --1.93 ns 12.11*** 0.00 ns
1.30 ns 2.44 ns 0.91 ns 0.02 ns 11.42 *** 0.09 ns 0.91 ns 0.47 ns
86.95*** 0.02 ns 0.66 ns 0.75*
24 548 380 102 79
105.68** 5.50 ns 0.66 ns 0.47* 52.01** 0.07 ns 0.20 ns 5.23*
28 467 285 76 98
74.93** 0.23 ns 0.18 ns 1.99 ns

*= P < 0.001, **= P < 0.01, *= P < 0.05, ns = not significant.








29

Table 3.6. Parameters calculated to fit citrus snow scale mortality data to a gamma
distribution function.

Parameters
Stage RH Tem R C

1"t stage 60 16 0.99 42.22 4.62 21 0.62 36.54 5.19 24 0.98 8.18 1.27 28 0.99 7.19 1.14 30 1.00 6.16 0.80 70 16 0.94 34.96 3.41 21 0.95 4.92 0.60 24 0.90 20.75 2.55 28 0.79 35.57 4.90 2nd stage 60 16 0.93 4.08 0.09 21 1.00 8.17 0.30 24 0.74 4.08 0.17 28 0.93 13.81 0.68 30 0.99 16.11 0.68
70 16 0.95 4.13 0.06 21 0.78 3.93 0.12 24 0.80 35.16 1.54 28 0.92 35.50 1.79
Female 60 21 0.81 80.00 1.03 28 0.88 6.94 0.06 30 0.16 2.40 0.02 70 21 0.96 21.94 0.25 28 0.73 80.00 0.82 Pooled 16 0.99 20.13 0.17
RH 24 0.98 19.52 0.21
Males 60 21 0.90 80.00 1.10 24 0.18 2.03 0.06 28 1.00 51.33 1.20 30 0.93 56.49 1.25 70 21 0.89 11.17 0.13 24 0.28 4.76 0.04 28 0.85 80.00 1.43








30

Table 3.7. Polynomial regressions between temperature and the parameters of a
gamma distribution (a, 3) fitted to the probability density function of the
survival curves


Stage RH R2 Equation

1" stage 60 0.84 a = 131.70 6.79 T+ 0.084 T2
0.74 P = 7.51 0.07 T- 0.0056 P 70 0.86 a = 338.20 30.10 T+ 0.691 P
0.91 3= 37.11 3.41 T + 0.081 P 2"d stage 60 0.82 a = 30.24 2.89 T+ 0.081 PZ
0.84 3 = 0.79 0.09 T+ 0.029 T2 70 0.74 a = -10.14 0.58 T+ 0.083 P
0.82 1 = 0.31 0.13 T+ 0.007 P Females Pool 0.13 a = -242.50 + 25.49 T- 0.56 P
0.19 3 = -3.48 + 0.36 T- 7.89 x 10 P Males Pool 0.33 a =1085.20 87.71 T+ 1.81 P
0.55 3 = 17.10 1.43 T+ 0.03 T2





Discussion

Optimal temperature for development and upper developmental thresholds

(T,) can be estimated from the experimental data and its inclusion in the Logan model. Optimal temperatures ranged between 25 and 380C for all the stages and both sexes, with values falling in the vicinity of 29*C for 1st instars and 260C for 2nd instars. Upper developmental thresholds ranged between 34 and 440C with most values falling in the vicinity of 350C (Table 3.3). The optimal temperatures for 2nd








31

stages are lower than the typical temperatures occurring during the summer in central Florida citrus groves (29*C average of daily mean temperatures from 20 June to 22 September 1992, 370C mean of maximum temperatures during the same period), which supports the observation that snow scale populations do not do well during the hottest months of the year. One is also led to speculate that the geographical areas where snow scale may have originated and thus may be best adapted (i.e. higher latitudes or higher altitudes), experience lower mean temperatures than Florida citrus regions.

Casares (1974) cultured citrus snow scales on fruits in a range of temperatures between 12.5 and 29.40C, and concluded that the lowest temperature at which the snow scale could develop was 18.30C. Insects did not develop at 12.50C, but survived and could resume development when higher temperature conditions occurred. In our research, scale development was completed at 160C. Thus, the lower threshold for development on grapefruit leaves must occur below this temperature. Thresholds estimated from the modified Logan model yield a wide range of results, including some > 16*C (Table 3.3), but many of them approach 120C as observed by Casares (1974). In our studies, the extreme threshold estimates were derived from prepupal and pupal stages (males), which were based on reduced sample sizes caused by mortality in early instars. The estimates also were limited in precision by the length of the stadium approximating the interval between observations. To obtain a more precise estimate of the developmental threshold for these stages, one must generate more data at the low end of the temperature range, with more frequent observations.







32

Casares (1977) discussed the possible effects of humidity on citrus snow scale populations. Casares cited observations by W.A.T Summerville in Australia that heavier infestations of citrus snow scale occurred during dry periods than during wet periods, concluding that humidity was an important factor in snow scale development. In our work, the effect of relative humidity on development was not consistently significant, but the interaction with temperature was, indicating that combined factors have an effect (Table 3.2). Only 2 conditions of humidity, differing by 10%, were used, because this was not a major objective of the developmental studies. These humidities were chosen to represent average conditions occurring in the field (range between 50 and 90% RH during 1992 and 1993). Differences in developmental parameters would be expected from studies using relative humidities representing a wider range and thus incorporating more marginal conditions.

The modification to the Logan nonlinear model was suitable to describe the data presented (Table 3.3). However, the reliability of the parameter estimates was not satisfactory for development of prepupal and pupal males because of the imprecision previously mentioned.

It has been shown that humidity may affect the rate of development for other species (Aonidiella aurantii, McClure [1990b]) but it may have a more important effect on survival than on development (Atkinson 1983). This effect may be direct in the field, altering the survival, rate of development and reproduction of the insect (McClure 1990a) or may be indirect, altering the action of its natural enemies.

In the current studies, temperature most notably affected mortality associated with 1st and 2nd instars and males (Table 3.4). Relative humidity effects were







33

shown to be important for survival of 1st instars and for 2nd instars at low temperatures (Table 3.5). The diaspidid armor is an effective protection against environmental hazards (Foldi 1990). Thus 1st instars are the most likely stage to die as a result of unfavorable environmental conditions given that they lack this protection.

Gamma distribution parameters calculated to fit the rate of change of the survival curves produced figures that varied widely with temperature and relative humidity, sometimes yielding a poor fit (e.g. females at 300C, Table 3.6). In characterizing the relationship between the shape of the mortality rate curve and temperature (Table 3.7), the fit obtained for 2nd-order polynomial equations was good for 1st and 2nd stages, but poor for older stages. The number of points used for these regressions was small (5 temperatures for 60% RH, 4 for 70% RH, and only 3 points for males because none survived at 160C) (Table 3.2). More extensive experimentation would be required to more accurately describe relationships between mortality through time and the temperatures to which the insects are exposed.

Host-plant substrate has been well documented as being responsible for morphological differences in diaspidid scales (Stoetzel 1976, Miller and Kostarab 1979, Cooper and Oetting 1986). Other biological characteristics such as development, mortality and fecundity are likewise affected (McClure 1990b). The developmental rate of citrus snow scale maintained on Duncan grapefruit leaves in this study was much higher than that observed by Casares (1974) for scales maintained on lemon fruits. We avoid drawing conclusions regarding these effects







34

of substrate because different citrus species are involved and effects of host species on citrus snow scale has been documented (Reed et al. 1967). However, the low frequency of fruit naturally infested by citrus snow scale in the field could indicate that the substrate also has an effect on settling preference, developmental rate, or survival of the scales.













CHAPTER 4
STUDY OF MORTALITY OF THE CITRUS SNOW SCALE UNDER FIELD CONDITIONS, BASED ON THE ANALYSIS OF PARTIAL LIFE TABLES


Introduction

Attempts to manage citrus snow scale by classical biological control began during the 1970's, but the results of these efforts were never clearly documented (Browning 1994). Snow scale is still present in citrus groves to the extent that the use of pesticidal suppression is sometimes warranted (Knapp 1995). There is a need to study the relationships between the citrus snow scale, its natural enemies and other biotic and abiotic mortality factors to clarify their value as population control factors and clarify the success or failure of previous attempts at biological control.

Life table analysis is an important tool in the study of populations (Harcourt 1969, Varley & Gradwell 1971, Southwood 1978, Carey 1993), providing a detailed account of the mortality affecting a cohort. A special type of table, the multiple decrement life table, allows the separation of different factors of mortality, simplifying the determination of their relative importance (Varley et al. 1973, Carey 1993).

This chapter reports on the mortality of the citrus snow scale under field conditions in Florida, adapting photographic techniques for sampling of the




35








36

populations, and using life table analysis to separate and weight the importance of the different mortality factors.



Materials and Methods

Data Recording with Photographs. Given the small size of the snow scale and the presence of an armor, direct observations in the field are difficult. Sampling of scale insects usually involves taking a sample of substrate to the laboratory and dissecting it under magnification. This destructive sampling does not permit the study of the fate of those individuals alive at the moment of sampling, that may die later of any cause. Collection and analysis of the remains of the insects are unreliable because it is difficult to establish the time of death, since the remains of the insect stay in place on the substrate for variable periods. Also, some causes of death do not leave tangible host remains. Direct counting in the field is difficult and inaccurate.

To overcome these difficulties, I developed a method for direct observation, recording on photographs the events occurring in a snow scale colony under field conditions. This method was modified from Summy et al. (1984) and Terry & Edwards (1989), and chosen because is more accurate than direct counts using hand lenses. It allows for nondestructive sampling and the following of specific individuals during their development.

Data were collected over a 2-year period between December 1991 and December 1993. Active colonies were located on the trunk or main branches of 'Valencia' orange trees in a commercial grove in Lake Alfred, Polk County, Florida,








37

where no insecticides were applied during the study. Between two and four observation areas (4x4 cm2) were selected on the bark of each tree. To obtain a cohort of newly settled crawlers of known age, I dislodged all individuals in the area (hereafter referred to as patches) using a hard brush (Casares 1977). A map pin was positioned in the upper left corner of the patch as a reference marker. Photographs of patches were taken weekly over a period that varied from eight to 20 weeks. New patches were established periodically (approximately every two weeks) on different trees. Eighty-two patches were studied on 36 trees during the observation period, but only 47 patches on 21 trees were colonized and therefore were used for the life table analysis.

A 35-mm camera (Canon AE-1, Canon Inc. Tokyo, Japan) equipped with a macro lens (Canon FD 50 mm f/3.5 Macro) mounted on bellows allowed image enlargements of 2X. The lighting consisted of two flashes, one mounted on a bracket beside the camera level with the lens and the other hand-held on the opposite side of the lens and equipped with a photosensitive trigger. The camera and flash were mounted on a sturdy tripod. High resolution color transparency film (Kodak Ektachrome 64 and Fujichrome 50) was exposed at 1/60 sec. and f/22 aperture and was processed through commercial laboratories.

Photographic Analysis. The sequences of insect development were studied while viewed on a slide projector with a built-in 9x9-inch screen (Kodak Ektagraphic AudioViewer/Projector model 260, Eastman Kodak Co., Rochester, NY 14650). The total area covered by the field of the camera was 18x12 mm2 of trunk surface. The area chosen for analysis was equivalent to 14x9 mm2 in the center of the field.








38

Individual insects were located and marked on the screen with the help of a clear plastic grid (1xl cm2 divisions ). Approximately 20 newly settled individuals were located and marked from the first scene of each sequence, and followed through the set of slides until their death or to the end of the sequence.

The causes of death were classified in the following manner:

1- Competition: When individuals were dislodged by the growth of a neighbor scale cover; included in this category were also individuals that were covered

by algae, this was observed in only one instance.

2- Parasitism: Individuals that showed emergence holes. Parasitized male scales exhibited these holes from the time they started to secrete the white armor until attaining their full length. Holes on males were attributed to parasitism by Encarsia spp. (Hymenoptera: Aphelinidae). Females were attacked by Encarsia spp. or by Aphytis lingnanensis. Encarsia parasitism was evident early in the second instar, and emergence holes appeared before the second molt. Emergence of Aphytis, on the other hand, occurred on young third instar females, with the emergence holes being larger and more rounded than those produced by Encarsia spp. All of these parasites have been observed emerging from non-preferred host stages and sexes (Aphytis from males and gravid females, Encarsia spp from adult females [Browning 1994]) Those atypical occurrences were not observed during the

present study.

3- Visible predation: Assigned to those cases in which the armor was clearly damaged but some remains persisted on the site. This damage was








39

attributed to predators that could chew or break through the armor, such as coccinellids. It is impossible to measure total predation. Predation can be underestimated, particularly when the scale armor or body of early nymphs is totally removed. On the other hand, overestimation of predation may occur when the disturbed armor did not contain live prey, or if the damage inflicted to the cover is due to scavengers. Encarsia-parasitized female scales attacked by predators before the parasite emerged also were observed. However, these were classified as death by parasitism since they would not survive. It is possible that the same phenomenon occurred with males, but it was impossible to observe.

4- Fungi: White mycelial growth was observed in several instances growing on individuals that had died. Unfortunately, it was impossible to determine if the insect was killed by the fungus or was colonized after dying by another cause. The most common fungi in snow scale colonies are from the genus Nectria (teleomorph of Fusarium) but their role is unclear. There is confusion over whether those fungi are acting as pathogens or as saprophytes (Ziegler 1949). Kuno & Col6n-Ferrer (1973) showed increased crawler mortality but saprophytic behavior on older stages. As with predation, death by diseases is being underestimated. Deaths that occurred under the scale cover and did not show mycelium or fructiferous bodies within the observation period were unaccounted. Given the impossibility to clear the issue of pathogenicity, we assumed any fungal appearance to be pathogenic and causing death to the scale with which it was associated.








40

5- Unknown causes: Mortality causes that could not be separated or were dubious or not clearly attributed to a given factor were grouped under this category. Several outcomes are described here:

a) Collapsed: A few male armors were observed to shrink or become deformed. This collapse could be attributed to the action of predators other than coccinellids, e.g., thrips, neuropteran, mites, that can feed

on the scale insect without disturbing the armor.

b) Failed development: Assigned when an individual scale stopped its development at an immature stage, with no outward signs of disturbance. This failure of development could be attributed to physiological or pathological causes, mainly for first instars, in which a change in coloration occurred as they dried out; to predation by organisms that act without disturbing the armor, such as predatory mites, thrips or lacewing larvae; or host-feeding by aphelinid

parasitoids.

c) Lost: Assigned when an individual scale disappeared without leaving a clue about its fate. Occurred frequently with first instars or with developing males. Could be the action of predators, or

mechanical factors such as rain or wind.

7- Indefinite outcome, covered by others: Used in the cases when the patch was overcrowded and other scales grew over the observed individuals. These were excluded from the analysis.








41

8- Completed life cycle: Individuals that reached the adult size and appearance and were not victims of any of the previous mortality causes were classified as completing their life cycle. The female reproductive period was not isolated for analysis since it occurs beneath the armor, making it

impossible to observe.

Statistical Analysis. Life table analyses were used to estimate mortality trends in populations of citrus snow scale on chosen patches. Single decrement life tables were generated (PROC LIFETEST, SAS Institute Inc. 1989). Comparisons were made between patches in each tree and between trees in the whole sample site. Multiple decrement life tables were built using the mortality categories listed and following the procedures and notation explained in Carey (1993):

K.= Number of individuals beginning stage x

Dix= Number of individuals dying of cause i at stage x

Dx= Total deaths in stage x

aq,= DJKx Fraction of deaths from cause i in stage x in the presence of all other causes, given that the individual is alive at beginning of stage x

aqx= D)/Kx Fraction of deaths from all causes in stage x, given that the individual is alive at beginning of stage x (_aq,) alx= alx-1(1 -aqx) Fraction of survivors at age x out of the original cohort, which is assigned al,=1

adm= al.(aqi) Fraction of deaths in stage x from cause i among alx living at stage x








42

adx= alx-alx.1 Fraction of deaths in stage x from all causes ('ad)



Since it is not possible to separate sexes for the first stage of development and since no attempt to build fecundity tables was made, the life tables include both sexes. The stages considered in the construction of the mortality tables were:

1- First instar, from crawler settling to the onset of first molt.

2- Second instar, before production of secretion, which is from the first molt until the cover secretion begins. Sexes are not distinguishable to this point.

3- Secretion, period during the 2nd instar during which the cover is built; separate calculations were made for each sex. Males pass through two more stages before adult (two molts) but these were unrecognizable because

they occurred beneath the armor. Females molt only once more.

4- Third instar (adult) females, started when females passed the second molt

and began depositing the last section of the armor.

The periods of insect development beyond this point to reproduction, and between 1st instar eclosion and settlement were excluded from the study, thus the mortality tables are based on only the part of the life cycle of the insect. A more detailed description of the life cycle was presented in Chapter 2.



Results

Single decrement life tables compared within trees by the log rank test, Wilcoxon test of homogeneity and likelihood ratio test showed significant differences for the patches in six of 17 trees. Four trees were not tested since each had a







43

single patch. The tree location, date in which the patches were initiated, date until they were observed and significance are shown in table 4.1. It is important to note that the sets of patches that showed significant differences were all initiated during spring and photographed into summer of both years.

Despite the differences observed between patches within some trees, the data from all of the patches in each tree were pooled and used to compare between trees and starting dates. The same statistical tests showed significant differences (Log-Rank X2 =91.40, Wilcoxon X2 =87.65, P< 0.0001 for both tests, 20 d.f., tree as strata, PROC LIFETEST, SAS Institute Inc. 1989) suggesting differential survival at different times of the year or different tree locations. A comparison between trees at each date on which a set of patches was started was possible for 6 dates (Table 4.2). Four of the 6 dates showed significant differences, suggesting that there is a site (tree) component contributing to the variation. The survival curves drawn from these data were diagonal or slightly convex, with the slope becoming steeper during the late spring and summer months, periods when U. citri development was also shortened by increased temperatures. Figure 4.1 shows survival and hazard functions (age-specific probability of death) from selected trees in winter, spring, summer and fall periods. Mortality was initially low and increased slowly and steadily through the scale lifetime but variation in mortality patterns was high. A summary of the fate of the 960 individual scales used in the analysis from a total of 1082 observed is presented in table 4.3. Only 13% of the observed individuals appeared to complete their life cycle. The overall observed sex ratio was 4.7 males/female.








44


Table 4.1. Significance of log-rank test, Wilcoxon homogeneity test, likelihood ratio
test for comparison of survival of citrus snow scale in patches within trees.

# of
Tree Patches Date started Date ended Significance 36-2 3 17/Dec/1 991 27/Apr/1 992 ns 36-7 2 17/Dec/1 991 27/Apr/1 992 ns 35-15 2 23/Jan/1 992 19/May/1 992 ns 35-18 2 06/Mar/1 992 14/Jul/1992 ns 34-21 5 14/Mar/1 992 26/May/1 992 ns 35-17 4 27/Apr/il 992 18/Aug/1 992 All tests 34-19 3 19/May/1 992 24/Sept/1 992 All tests 32-15 2 17/Jun/1i 992 16/Sept/1 992 All tests 34-11 1 29/Jul/1 992 24/Jan/1 993 26-9 2 24/Aug/1 992 15/Jul/1 993 ns 28-7 2 24/Aug/1 992 23/Jun/1993 ns 26-16 2 12/Oct/1 992 14/May/1 993 ns 27-14 2 26/Oct/1 992 30/Apr/1 993 ns 28-14 1 26/Oct/1 992 8/Apr/1 993 -24-2 2 30/Apr/i 993 15/Jul/1993 All tests 23-13 3 24/May/1993 1/Dec/1993 Log rank P <0.025 24-12 2 24/May/1993 1/Oct/1993 Log rank P <0.031 22-9 3 17/Aug/1 993 16/Nov/1i 993 ns 25-6 3 16/Sept/1993 14/Apr/1993 ns 21-12 1 6/Oct/1993 1/Dec/1993 22-14 1 6/Oct/1 993 1/Dec/1 993 --








45

Table 4.2. Significance of log-rank test, Wilcoxon homogeneity test, likelihood ratio
test for comparison of survival of citrus snow scale between trees by dates. Date started # of trees Significance 17/Dec/1991 2 Wilcoxon P < 0.047 06/Mar/1992 2 Log-Rank P < 0.040 24/Aug/1992 2 Log-Rank P < 0.008 26/Oct/1992 2 ns 24/May/1 993 2 ns 6/Oct/1993 2 Log-Rank P < 0.024 Wilcoxon P < 0.016


Survival ranged between 0 (Jun., Aug., and Oct. 1992) and 30.5% (Aug. 1992) in the 21 mortality tables constructed. The main mortality was classified in the "unknown causes" category, but predation and parasitism provided an important contribution to the total (Table 4.4). Between 0 (Jan 1992, Oct. 1993) and 48.8% (Sept 1993) of the initial cohort died by the action of predators, between 0 (Jun., Jul. 1992, Apr., May, Aug., Oct. 1993) and 38.3% (Aug. 1992) died by the action of parasitoids. It is important to mention that most of the parasitism observed was attributed to Encarsia, only one female out of 13 observed parasitized was killed by Aphytis. Unknown causes accounted for 29.3% (Sept 1993) to 90% (Oct. 1993).

Table 4.5 shows a summary of the total mortality (all the causes) presented by stage. The highest mortality usually occurred on armor-secreting males. However, in three cases, mortality was the highest in the second stage prior to scale secretion (Dec. 1991, Jun. 1992, May 1993); in one case mortality was greatest on








Tree 36-2, 17/Dec/1991 Tree 35-17, 271April992 Tree 32-15, 171Jun/1992 Tree 26-16, 121/Oct11992
1.0
0.8
S 0.6 I
0.4 0.2
0.0
0.3 I !
0.2 I
0.1 I
0.0
0 20 40 60 80 100 100 20 40 60 80 100 120 20 40 60 80 100 120 20 40 60 80 100 120
Tree 23-13, 24/May/1993 Tree 22-9, 171Aug1993 Tree 25-6, 161Sept/1993
1.0 -.
0.8
First molt 0 I
S 0.6 Secretion 1 0.4
0.2 I Female 2nd molt 0.0
0.3 I Full size male 0.2 -I
0.1
Full size female 0.1 I
0.0
0 20 40 60 80 100 190 20 40 60 80 100 100 20 40 60 80 100 120 Age (days)
Figure 4.1 Survival and hazard functions of snow scale population on selected trees.
Vertical lines represent time of molts. c








47

Table 4.3. Outcome observed for citrus snow scale individuals under field
conditions. Summary, data pooled for all patches and dates.


1"t 2nd Secreting 3rd Outcome instar Total instar instar Male Female female Completed N/A N/A 118 N/A 10 128 Detectable predation 0 1 119 6 16 142 Parasitism 0 0 95 8 5 108 Competition 1 7 0 0 0 8 Fungi 0 3 1 1 1 6 Unknown causes 103 244 158 35 24 568
Lost 51 78 37 6 5 177 Failed to develop 52 166 125 29 19 391
Total 104 255 495 50 56 960


adult (third instar) females (Jul. 1993), and finally, in two cases, first instars suffered the highest levels of mortality (Aug. and Oct. 1993).



Discussion

Survival curves of citrus snow scale populations showed significant differences between patches on six trees. These patches shared a feature in being initiated between May and mid-June of each year, when temperatures were increasing. The temperature increase is responsible for a shortened life cycle and may increase activity of natural enemies that could have affected variability within these six trees.








48


Table 4.4. Total percentage of mortality (adx) and percent mortality by unknown
causes against parasitism and detectable predation in multiple decrement life
tables, calculated by tree.

Detectable Unknown
Tree Date adx Detectable Parasitism Unknown predation causes
36-2 17-Dec-91 0.7939 0.0153 0.0763 0.6718 36-7 17-Dec-91 0.8387 0.0968 0.1452 0.5968 35-15 23-Jan-92 0.8333 0.0000 0.2708 0.5000 35-18 06-Mar-92 0.9355 0.2903 0.1935 0.4516 34-21 14-Mar-92 0.9775 0.1910 0.2022 0.5730 35-17 27-Apr-92 0.8106 0.0303 0.1212 0.6591 34-19 19-May-92 0.9464 0.2857 0.0179 0.6429 32-15 17-Jun-92 1.0000 0.1707 0.0000 0.8293 34-11 29-Jul-92 0.8000 0.3000 0.0000 0.5000 26-9 24-Aug-92 1.0000 0.2647 0.3824 0.3529 28-7 24-Aug-92 0.7250 0.2250 0.0500 0.4500 26-16 12-Oct-92 1.0000 0.0645 0.2903 0.6452 27-14 26-Oct-92 0.8333 0.1250 0.0833 0.6250 28-14 26-Oct-92 0.9412 0.2353 0.1765 0.5294 24-2 30-Apr-93 0.6970 0.3333 0.0000 0.3636 23-13 24-May-93 0.8868 0.2453 0.0000 0.6415 24-12 24-May-93 0.7576 0.0909 0.0000 0.5152 22-9 17-Aug-93 0.9231 0.1154 0.0000 0.7692 25-6 16-Sep-93 0.9024 0.4878 0.1220 0.2927 21-12 06-Oct-93 0.9500 0.0500 0.0000 0.9000 22-14 06-Oct-93 0.7500 0.0000 0.1250 0.6250








49

Table 4.5. Total percent mortality (all causes) adx presented by stage.

Date Secreting 3" instar
Tree a 1" instar 2'd instar
Started Male Female Female
36-2 17-Dec-91 0.1374 0.4656 0.1756 0.0076 0.0076 36-7 17-Dec-91 0.0806 0.2742 0.4194 0.0645 0.0000 35-15 23-Jan-92 0.1667 0.3125 0.3542 0.0000 0.0000 35-18 06-Mar-92 0.0323 0.1290 0.7097 0.0323 0.0323 34-21 14-Mar-92 0.0225 0.2360 0.5955 0.1124 0.0112 35-17 27-Apr-92 0.1818 0.2045 0.3485 0.0379 0.0379 34-19 19-May-92 0.0357 0.3214 0.5357 0.0179 0.0357 32-15 17-Jun-92 0.1220 0.4146 0.3659 0.0732 0.0244 34-11 29-Jul-92 0.0000 0.1000 0.2000 0.1000 0.4000 26-9 24-Aug-92 0.0588 0.2353 0.5000 0.0588 0.1471 28-7 24-Aug-92 0.0500 0.2000 0.2250 0.0500 0.2000 26-16 12-Oct-92 0.0968 0.3871 0.5161 0.0000 0.0000 27-14 26-Oct-92 0.0417 0.2500 0.4583 0.0833 0.0000 28-14 26-Oct-92 0.1176 0.1765 0.5882 0.0000 0.0588 24-2 30-Apr-93 0.0303 0.2727 0.3636 0.0303 0.0000 23-13 24-May-93 0.0755 0.1509 0.4340 0.0943 0.1321 24-12 24-May-93 0.0909 0.3333 0.1818 0.0303 0.1212 22-9 17-Aug-93 0.4231 0.0769 0.3077 0.1154 0.0000 25-6 16-Sep-93 0.0488 0.0732 0.512 0.1220 0.1463 21-12 06-Oct-93 0.4000 0.2000 0.3500 0.0000 0.0000 22-14 06-Oct-93 0.0000 0.0000 0.3750 0.3750 0.0000



Significant differences were detected between some patches in different trees, and between trees where patches were set at the same dates. This could be due to the different times of the year in which the patches were initiated (Table 4.1), and also could be due to different locations of the trees inside the grove (Table 4.2), or a combination of both. Given that the patches were set on different dates through








50

the year, the insects were exposed to different weather conditions. Since development depends on temperature (Chapter 3), scale insects have shorter life cycles during the warmest months (Figure 4.1). It is also expected that extreme hot temperatures increase mortality, but this is not obvious from the data presented in Figure 4.1. It is also assumed that natural enemies have shortened life cycles at higher temperatures.

Overall, within all of the cohorts studied, (Table 4.2) 24% of males (118 out of 495) and 18% of females (10 out of 56) survived to the end of the observation periods. Most of the mortality observed could not be assigned to a specific cause. This unknown mortality affected 59% of the individuals in the cohorts (568 out of 960) and accounted for 68% of the total mortality. Natural enemies were an important component; detectable predators killed 15% of the individuals, while parasitoids killed 11%, representing 17% and 13% of the total causes of mortality, respectively.

The extent of variability in the data can be observed in table 4.3. Despite this variability, total mortality (adx) never fell below 70%, and varied up to 100%. Detectable predation appeared more important than parasitism (0-49% against 038%) and males demonstrated higher probabilities of being attacked by both factors than females (Table 4.2). Secreting males and second instars of both sexes were the stages with highest mortality (Table 4.5), perhaps because the higher proportion of males over females means higher availability of male prey or hosts. Males could also be more susceptible given their softer shield. Males and females are available for attack by parasitoids for a similar period. Even though males have a shorter life








51

cycle, females are not available for parasitism after they reach reproductive status (Browning 1994). On the other hand, reproductive females will remain targets for predators longer than males. Undoubtedly, predation is underestimated, and likely an important fraction of the observed unknown mortality is predation. In the present research, attempts to trap predators failed but chrysopid larvae (Neuroptera) were observed, collected, and reared from citrus snow scale colonies.

Parasitism occurred through most of the study period, but was lacking from our patches in June and July, 1992 and April to August, 1993. Aphytis lingnanensis has been present in Florida for some time. A race of this species collected in Hong Kong and named HK-1 was released in Florida in the early 1970's (Browning 1994). This species was found in leaves and twig samples in the grove where this study was performed, but the low frequency in which parasitism by Aphytis was observed in the photographic analysis suggests that Aphytis is not a major factor in the regulation of citrus snow scale populations on trunks and therefore it is not as successful a biological control agent as has been reported (Fisher 1985). The contributions of parasitism and detectable predation appeared not to be related as they seemed to fluctuate independently between patches (Table 4.4). High incidence of one factor did not accompany low incidence of the other. Apparently, they are not complementary in this sense. In fact, predators may be a competing factor with the survival of parasitoids that attack early stages of citrus snow scale.

Contrary to what was expected, diseases were observed in a very small proportion of scales examined, despite the fact that reproductive bodies of the fungus were observed on snow scale colonies on the same trees where the patches








52

were photographed. It was anticipated that the crowded conditions in citrus snow scale colonies would favor the spread of diseases. It is very likely that a portion of the unknown mortality is composed of individuals killed by diseases. Potential explanations for not observing fungi in our study are that the period between infection and the development of evident fructiferous bodies could be longer than the period of our observations (8-20 weeks), or that the disturbance created when cleaning the patches affected somehow the incidence of fungi, reducing the crowdedness of the insect colony or the age of abandoned shields, or that the fungi are nothing more than saprophytes and would not develop on live insects. Nectria spp. have been mentioned as important mortality factors for scales in citrus (Fisher et al. 1949) and are found frequently in snow scale colonies, but they as well could be saprophytes that thrive on the abandoned covers of deceased individuals or emerged males (Ziegler 1949; Fisher 1950). Kuno & Col6n-Ferrer (1973) studied the pathogenicity of two species of Fusarium over scale insects. They found that crawler mortality could be increased by the fungi, but there was not effect on older instars, and the fungi showed saprophytic growth.

In general, mortality of snow scale in the observed patches was very high (between 70% and 100%, Table 4.4). A substantial percentage of mortality measured in the various patches was due to unknown causes. Very likely, a part of it can be attributed to natural enemies (parasitoids, predators and diseases), but there is no way to separate it using the current data. The observed effects of natural enemies were considerable and could account for about a quarter of the total mortality. However, this mortality was insufficient to reduce the densities of snow








53

scale to rare occurrence in the site studied. The white covers of the snow scale males are very evident against the dark background of the citrus bark, and the covers abandoned by males that already emerged remain on site for an undetermined but presumably long period. Despite the observed high mortalities, the accumulation of male scale covers may give the impression that densities of live citrus snow scales are greater than what they really are. This suggests that the snow scale problem is in part a problem of perception by the grower. It also identifies the need for careful observation to determine the status of live scales in the field when management options are considered.













CHAPTER 5
COMPARATIVE STRUCTURE OF CITRUS SNOW SCALE
POPULATIONS ON THE CANOPY OF CITRUS TREES AND THROUGH TIME.


Introduction

A good knowledge of the biology and ecology of a pest is necessary to design optimal pest control programs. Periodic sampling of the population produces useful information concerning age structure, sex ratios, and mortality factors. Difficulties arise when measurement of absolute density of an organism is needed, since it requires an estimation of the area sampled. Alternatively, population intensity (number of individuals per unit of habitat, e.g. leaf) and relative estimates are available (Southwood 1978). The percentage of a population subjected to parasitism is a traditional measurement of the effect of these natural enemies, but this measurement has been criticized based on difficulties in obtaining accurate information on the real impact of those parasitoids (van Driesche 1983).

Species that colonize different organs of a plant have been shown to exhibit different traits, such as differences in developmental rate, sex ratio, and even morphology (Chapter 2). In the same way, the association with different host plant organs exposes the insects to different environmental influences. Variable host nutrient composition may occur between leaves, fruits and trunks. Natural enemies




54







55

may also react to different substrata in different ways, such as exhibiting preference for searching on some surfaces over others (Carroll 1979, Murdoch et al. 1989).

The phenology of the citrus snow scale has been previously investigated. Methods included visual counts on trunk, leaves and twig samples that produced population intensity estimates (Casares 1977). Attempts at biological control against this scale were made in 1974 through introduction of parasitoids to Florida. The aphelinid Aphytis lingnanensis "HK-1" was introduced and released. Successful biological control was reported but citrus snow scale persists as a problem in some citrus groves in Florida (Browning 1994). The study by Casares (1977) preceded the introduction of A. lingnanensis "HK-1" and aimed to evaluate the impact and importance of previously established parasitoids. However, the results were not presented in a comparative form and failed to consider an important component of the parasitoid fauna that subsists on nymphal males of the citrus snow scale.

In the present work, population samples of the snow scale were studied with the use of relative population estimates, which are easier to contrast. The objectives were to assess the effects and importance of natural enemies, especially parasitoids, and how populations of snow scale change on the different substrata and through time.



Materials and Methods

Samples of twigs and leaves were collected monthly from three to five snow scale-infested citrus trees in a commercial grove in Lake Alfred, Polk County, Florida. Between five and 12 snow scale infested twigs with leaves were chosen








56

from the interior of the canopy. These samples were processed in the laboratory and up to 100 live individual snow scales were counted from leaves and 100 more from twigs. The information recorded included sex, stage, parasitism, presence of fungi, and presence of mites. Two types of parasitoids were observed, the endoparasitoids, Encarsia species (Hymenoptera: Aphelinidae) and the ectoparasitoid Aphytis lingnanensis (Hym.: Aphelinidae). Encarsia spp. are endoparasitoids of both sexes, ovipositing into the second stage. Parasitism of male scales is evident because the dark parasitoid pupa is visible when male scale covers are lifted. In females, parasitism by Encarsia becomes visible through changes of body coloration of second instar female scales while the parasitoid is in its larval stages. Late parasitoid larvae and pupae are easily visible through the scale cover and body. Two species are likely to occur in the study area, E. citrina (Craw.) and E. longsburyi (Berlese & Paoli) (Browning 1994). On the other hand, Aphytis lingnanensis Compere is an external parasitoid that attacks young third stage females (pre-reproductive adults). Parasitoid larvae and pupae are very evident on their hosts, but eggs are more difficult to detect.

The fungi observed developing on citrus snow scale species were Nectria species (anamorph Fusarium), exhibiting white mycelia and orange fructiferous bodies. An unidentified species of mite also was observed feeding on the bodies of adult females. These mites were observed inserting their mouthparts into the scale integument. Citrus snow scale deaths by predators were not quantified, since there was no way to determine time of death or how long a dead individual would







57

have remained on the substratum. Counting them would have overestimated the effect of predation.

The sampling took place between May 1992 and December 1993, comprising 14 sampling dates. The period between sampling dates was sufficient to allow the emergence of parasitoids in immature stages on the previous sampling date and also to avoid duplicate sampling of the same snow scale cohort.

The proportions of individuals of each age (stage) group were calculated, adding the numbers of healthy individuals and the numbers of parasitized individuals that may have belonged to the same cohort. Internally-parasitized females were classified as second instars, while externally parasitized females were classified with gravid females, because they were attacked as prereproductive females and they are part of the gravid-female cohort. Sex ratio was calculated by adding the number of healthy and parasitized individuals, i.e., total healthy males plus males bearing internal parasitoids, healthy second instars females plus prereproductive adult females plus females bearing any parasitoids. Reproducing females were not added because the older males of equivalent age had emerged and thus were not accounted for in sex ratio calculations. The incidence of parasitoids and fungi was calculated as a percentage of parasitism based on the total individuals of a given sex and stage.

The statistical analysis involved distribution comparison of the percentage age structure between substrata using the Kolmogorov-Smirnov two sample test and Kruskall-Wallis non parametric multiple comparisons for detecting differences between sampling dates (Siegel 1956, SAS Institute Inc. 1989).







58

Results

Figure 5.1 shows the age structure of the population sampled on each date according to the substratum. First instars of undetermined sex and females form the age structure pyramid, whereas males are shown as a bar above the pyramid. The total number of scales in the sample is included. A Kolmogorov-Smirnov two sample distribution test did not show significant differences between substrata, and the Kruskall-Wallis test did not show significant differences between dates. However, the population showed a tendency toward younger stages on leaf samples from May 1992 to June 1993 and September 1993, but reversed on August and October through December of 1993. Age distribution was more homogeneous on twig samples, showing higher proportions of prereproductive females in August and September 1992 and again in August 1993. Higher proportions of first instars occurred during May, June and September 1993. The total number of individuals counted on each sampling date is an indication of the abundance of snow scale at the time of sampling. Low populations were found in November of 1992, and in August, September, November and December of 1993. No live individuals were found on the twig samples from November 1992. These trends should not be taken as conclusive results, given the lack of statistical significance already mentioned.

The ratios of newly born individuals (1st stage nymphs) per reproductive individual (gravid females) are presented in Figure 5.2. These values were consistently higher on leaves, except on the last sampling date in December 1993. High ratios were recorded in June, 1992 and March, May, and September, 1993.








59
May 1992 June 1992 Aug 1992 males 6
462 538 496 1067 54 800 Gravid females
Virgin females
Nymph II fem.
Nymph I
1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0
Sept 1992 Nov 1992 Jan 1993 males 1 E
137 424 171 215 267 Gravid females
Virgin females
Nymph II fem.
Nymph I
1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0
Mar 1993 May 1993 June 1993 males 152 363 397 129 236 436

Gravid females
Virgin females
Nymph II fem. i Nymph I
1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0
Aug 1993 Sept 1993 Oct 1993 males E EU
148 30 75 89 400 52 Gravid females
Virgin females
Nymph II fem.
Nymph I I
1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0
Nov 1993 Dec 1993 males -ILI
77 11 96 103 Sample substratum
77 11 96 103
Gravid females Leaves Virgin females Twigs Nymph II fem.
Nymph I A I
1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0
Figure 5.1 Age structure of citrus snow scale populations on citrus leaves and
twigs. Numbers on the side of each pyramid represent the total number
of individuals on that sample.







60








160- -0-- Leaves >-- Twigs
15030 e 20S10
=It








Figure 5.2. Ratio of snow scale first instars per gravid females on samples
of citrus twigs and leaves.








61

The missing points correspond to samples where no gravid females were found (except Nov., 1992 where no live scales were was found on twigs).

The sex ratios on twigs and leaves are presented in Figure 5.3. In contrast to previous measurements reported in the literature (Casares 1974), observations made in lab cultures (Chapter 3) and on tree trunks in the field (Chapter 4), the sex ratio was close to 1.8 females per male on leaves (average 36% males), with variation ranging from 14-49% males. The proportion of females on twigs was greater (2.2 females:1 male, average 0.31 males), with values ranging 0-53% males. A greater proportion of males was observed in the late spring-early summer months on twigs and leaves in the first year and earlier on leaves during the second year (1993).

The incidence of parasitoids and fungi followed similar patterns on leaves and twigs, especially those associated with female scales (Figure 5.4). Fungi and mites occurred only sporadically. High incidence of internal parasitism occurred in October 1992 and late in 1993 (October-November). Encarsia spp. were abundant on males over most of the observation period, with drops in July 1992 and May to August 1993. The incidence of ectoparasitoids (Aphytis) was usually higher on leaves with peaks in April 1992 and September to December 1993 but some of this increase represents low numbers of host stages encountered.

The incidence of hymenopterous parasitoids, fungi and mites are presented in Figure 5.5, as relative proportions calculated from the totals of attacked individuals observed at each sampling date. Internal parasitoids are predominant except on August 1993 and on October 1993 (twigs). No internal parasitoid attacks









62



Leaves
Dec-93
Nov-93 I I
Oct-93 Females
Sep-93
Aug-93 Males
Jun-93
May-93 Mar-93 Dec-92 Nov-92 Sep-92 Aug-92
Jun-92 May-92

0% 20% 40% 60% 80% 100%



Twigs Dec-93
Nov-93 gg
Oct-93 iFemales
Sep-93 i
Aug-93 Males
Jun-93 May-93 Mar-93 Dec-92
Nov-92 Sep-92 Aug-92
Jun-92 May-92

0% 20% 40% 60% 80% 100%





Figure 5.3. Proportion of males and females on samples of citrus leaves and
twigs.







63


80 A Internal parasitism B Internal parasitism
in males in females
60

40

20


a 80 C External parasitism
o in females
S60
.5 Sample subtratum o -0- Leaves
e 40 - wigs

20
0

80 D Fungi on female E Mites on females
70



1


0





Figure 5.4. Incidence of natural enemies of snow scale on samples of leaves
and twigs. Percentage are calculated over the total of susceptible stages
present in the population.







64
May 1992 Jun 1992 Aug 1992 Sep 1992
6 3
Leaves ( 94 97 100 2..

Twigs as


Nov 1992 Jan 1993 Mar 1993 May 1993 Leaves 0
80 100 86 100


wgs75 100 100 Twigs :


Jun 1993 Aug 1993 Sep 1993 Oct 1993 Leaves 8 1 a 4



Twigs 62


Nov 1993 Dec 1993

Leaves Internal parasites \ External parasites igs Fungi Twigs 56 9Mites

Figure 5.5. Relative importance of natural enemies of snow scale females on
samples of leaves and twigs. Percentages are calculated over the total
attacked females.







65

occurred in September 1993 on twigs. Fungi were predominant in August (leaves, 13 of 24 individuals) and October 1993 (twigs, 7 of 11 individuals). Mites were observed in September, 1992 and January, June and October 1993.

Mites were found living under the bodies of live adult reproductive females with their mouthparts inserted into the insect body, apparently sucking fluids. The scale continued to oviposit. Unlike the other natural enemies, these mites were not preventing reproduction, but very likely they were producing some deleterious effect.



Discussion

The predominance of early instars in the age structure of a population suggests high natality in the population, or high mortality of older stages (Odum 1971). In the case of scale insects, high immigration of crawlers is also possible, since the first stage is the only one able to move. A high proportion of young instars were observed on leaves over twigs during most of 1992 and 1993, while predominance of older stages occurred in October to December 1993. This pattern roughly follows the changes in air temperature for those times of the year. As shown in chapter 3, development of citrus snow scale is dependent on temperature.

The high ratio of newly settled first instars to reproductive females (Figure 5.2) could be the result of higher natality on leaves, but more probably it is explained by movement of crawlers from the twigs and limbs onto the leaves, or a combination of both. Carroll (1979) observed differences in substrate preference by California red scale Aonidiella aurantii, a species that can colonize most of the organs of a citrus tree. He explained the substrate preference based on phenological and







66

topological characteristics and on differential mortality. Since it is known that different substrata influence development and survival (see chapter 3 and Miller & Kostarab 1979, Cooper & Oetting 1986), the possibility of increased citrus snow scale natality cannot be dismissed.

Mortality factors also may be operating differently between the various substrata. Murdoch et al. (1989) demonstrated that rates of parasitism by Aphytis melinus and Encarsia sp. on California red scale (Aonidiella aurantii) were higher on the periphery of the canopy than on the trunk of citrus trees. They suspected that lower attractiveness of the bark surface to parasitoids may play a role in this phenomenon. Carroll (1979) also observed this phenomenon with Comperiella on A. aurantii. The citrus snow scale system, while sharing many features, is not the same as the one observed by Murdoch et al. Citrus snow scale samples in the present studies came from the interior of the tree, near the trunk and main scaffold limbs, and thus our comparison is between the interior leaves and supporting twigs. However, the principle demonstrated by Murdoch et al. may apply equally well in the sense that different substrata may have an effect on the searching and dispersal capabilities of natural enemies.

Snow scale sex ratios have been previously reported as male biased (Casares 1974). Our studies showed a male-based sex ratio in laboratory cultures (Chapter 3) and in colonies on citrus trunks in the field (Chapter 4, Table 4.3). However, the data from twigs and leaves in this study indicate a female biased sex ratio in the canopy. The factors discussed earlier that could influence natality,







67
mortality and movement of citrus snow scale populations on twigs and leaves could have contributed to the observed reverse trend in sex ratio on the canopy.

The important parasitoids affecting mortality of snow scale in this study were Aphytis lingnanensis and Encarsia species. While Encarsia species attacked both sexes, A. lingnanensis attacked females exclusively. Parasitism by Encarsia was higher on females (Figure 5.3A, 5.3B), favoring a male biased sex ratio. Since females are subject to increased pressure from these two sources of mortality, parasitism likely would push the sex ratio bias toward males. This was not the situation observed on leaves and twigs. Differences in dispersal behavior between sexes have been reported for other species of scales (Arias-Rever6n 1988) and for snow scale (Xinnian & Browning 1991). Female crawlers generally followed a straighter course than males and were more active, reaching farther during this important dispersal phase. Since females feed for a longer period (Chapter 3) and likely require more nutrients for development and reproduction, differential dispersal behavior may be an adaptation to reduce intraspecific competition and to expand colonization into more remote locations. Sexual dimorphism in the adult form in scale insects provides a mechanism for winged males to disperse in search of female receptive to mating, countering the limitations on male crawler dispersal and reducing the impacts of inbreeding.

In a study of citrus snow scale prior to the introduction of A. lingnanensis "HK1", the incidence of parasitism by Encarsia longsburyi was presented as number of parasitized hosts per unit of habitat (leaf) and per unit of area (cm? of twig) (Casares 1977). Relative proportions of the population impacted by Encarsia parasitoids are







68

not apparent through these data, and comparison between substrata is impossible. Inspection of Casares's (1977) figures indicated estimates of parasitism in females might be 25% to 55% per leaf and very low (6-10%) per unit of twig surface. Unfortunately, the figures pooled unparasitized 2nd and 3rd female stages, further obscuring the incidence of Encarsia parasitism. Nevertheless, Casares concluded that in addition to its lack of host specificity, Encarsia was not able to control snow scale because its searches for host on leaves rather than twigs.

In the present investigation, comparable percentages of internal parasitism by Encarsia occurred between leaves and twigs (Figure 5.3A, 5.3B). However differences were visible in ectoparasitism by A. lingnanensis, with a predominance of parasitism on leaves (Figure 5.3C). Results previously reported in Chapter 4 point to the minor contribution of A. lingnanensis to the mortality of citrus snow scale in colonies resident on trunks. These observations support the notion that individuals on leaves are more heavily attacked and suffer higher mortality, and agree with the observations of Carroll (1979) and Murdoch et al. (1989) on California red scale.

Since this increased mortality is occurring on females, which is also the most abundant sex, it would be likely that in the absence of A. lingnanensis the populations of citrus snow scale might increase on leaves. Percent parasitism measurements should be taken with caution since they can be calculated only from the observed dead and live individuals, excluding individuals killed by other causes that otherwise would have been part of the live or parasitized fractions (Carey 1993). It should be kept in mind that we may be overlooking other potentially important








69

components of mortality in predation and in host feeding by aphelinid parasitoids (Encarsia and Aphytis).

The relative importance of mortality factors observed is highly variable through the year and on the different substrata (Figure 5.5). Endoparasitism was predominant most of the time and only on two occasions was overshadowed by other entities, namely incidence of fungi. In one case (August, 1993 on twigs) ectoparasitoids were more frequent than endoparasitoids. A potential source of bias is small samples and aggregation of susceptible hosts that may have facilitated parasitoid oviposition on several hosts in the same sampled twig.. I have no way to determine if the observations of fungi represent the onset of a disease or the indication of a saprophyte, but in any case, we should not disregard the fungi potential to inflict mortality on the scales without solid proof.

It appears that A. lingnanensis and Encarsia spp. may be important factors in regulating populations of citrus snow scale in the canopy of citrus trees. However, given the fact that the majority of the scale population occurs on the trunk and scaffold limbs, and that low incidence of A. lingnanensis was observed on these substrata (Chapter 4), it could be concluded that these parasitoids are contributing little as biological control agents of citrus snow scale at the studied location.













CHAPTER 6
THE ROLE OF WIND IN CRAWLER DISPERSION Introduction

Diaspidid scale insects are sessile organisms with only two very limited motile stages, adult males and newborn first instars (crawlers). Adult males can fly, but by themselves they cannot establish new colonies. Crawlers are very fragile and of limited mobility; even this stage may be limited in colonizing new hosts. It has been proposed that the main form of dispersal for the snow scale is through crawler phoresy on the clothing and equipment of grove crews in the citrus groves (Simanton 1973), but little attention has been paid to the role of air movement as a dispersal agent. Wind have been demonstrated to aid dispersal between plants for several scale insect species (Quayle 1916, Brown 1958, Greathead 1990). This chapter explores the role of wind as a mechanism of dispersion for the citrus snow scale, Unaspis citri.



Methods

Heavily infested, crawler-producing trees were identified in two citrus groves in Lake Alfred, Polk Co. Florida. These 'Valencia' orange trees were about 1.5-2 m tall in a block planted at 5x6.5 m spacing, and adjacent trees held little or no snow scale. Sticky cards fastened to 1.2-m tall poles were set around the source tree at 70








71

a distance of about 1.5 m from the canopy edge. The traps were oriented North, East, South, and West of the source tree. Each pole held two 10x10 cm2 Mylar cards coated with Tanglefoot. The initial experiment was established on May 31, 1993 and the traps were retrieved seven days later. This experiment was repeated five times with small variations in the same and in a new location (CREC, N-40, block 16, Lake Alfred, Florida) in 1993-1994 (July 19, July 28, 1993, February 21, April 6, October 14, 1994).

Crawlers were counted on the entire card, since the number of crawlers trapped was always low. The numbers of snow scale crawlers, male scales and natural enemies (Aphytis and Encarsia) were characterized. Wind speed and direction was not monitored at the site, but wind data were recorded at the nearby Citrus Research and Education Center (approximately 2.4 miles linear distance from the initial site and approximately 2.5 miles from the second site).

Several experimental trials failed due to interference by grove vehicles. In the first experiment (May 31-Jun 3,1993) established around a single tree, traps were set at the four cardinal directions and two distances from the tree. Three of the poles at the distance closest to the tree were knocked down, and thus only data from the farthest distance were considered for analysis. In the second set (July 1922, 1993), two trees in the same grove were used with poles all at the same distance from the source tree. All the poles in the East-West orientation were lost to traffic, making the data unusable. The third trial (July 28-August 2, 1993) was a repetition of the second and went undisturbed. The fourth trial (Feb. 21-28 1994) was established in another grove (N-40 block 16, CREC) on two trees with traps at








72

two distances from the source. This was disturbed by a spray operation, and the data were again discarded. The fifth trial (April 6-14, 1994) was a repetition of the fourth and failed in one tree when at least three traps from the East orientation were disturbed. Data from the second tree were complete. The sixth trial (October 14-21, 1994) was set up around single trees in both sites and these traps were not disturbed.

In addition to collection of trap data, snow scale spread was monitored from an artificial infestation in one grove at the CREC (block 16, N-40) during the last two years. This grove contained 'Valencia' orange and 'Duncan' grapefruit trees. Four trees in the plot were infested with snow scale in August 28, 1992, by attaching snow scale infested seedlings to the trunk or main limbs. Monitoring involved grading of snow scale colonies based on their appearance once every month. A score from 5 (heavy infestation) to 0 (no infestation) was given to each tree in the plot, based on a brief inspection for the presence of snow scale male covers on the trunk and main branches (Casares 1977). Wind direction data for this extended period were available only as daily means and were not used.

Numbers of crawlers trapped at each direction were compared by KruskallWallis non-parametric test (Siegel 1956). Visual comparisons were made between wind patterns and number of catches.



Results

Mylar traps. Figures 6.1 and 6.2 present the number of crawlers caught in trials 1, 3, 5 (data from a single tree without repetition) and 6. The hourly averages







73

Direction (degrees)

00 900 1800 2700
S6 16
A o000 14
12
- 4 Oo o10 E3 0 00 0 8
0 o 6 o 2 o o


z0 0<
North East South West


00 900 1800 2700
-8 I I 16 7 B 14
6 0 12
0

0 5 0 10

s-
, 4 dOo 8

$2- 4


North East South West


Figure 6.1. Wind speed vs direction and catches of snow scale crawlers
A Trial 1, May 31 to June 3 1993. B Trial 3, July 28 to August 2 1993.







74

Direction (degrees)

00 900 1800 2700
6 i 16
A (so 14 E
0 12 4 0o 0 0 00 10 0* o 10 C
0 0 o o S3 0 0 oo0o 8 ( o. o o- 6 02 00 6





0o 90* 180* 270
0 4 12
It:











io 0 0 o 10,
0 0 900 180 2700
0 C g o0 0 06



8 Q 0 o
Swo 1o o 2 2 Z 0 o 0





NW N NE E SE S SW W


Figure 6.2. Wind speed vs direction and catches of snow scale crawlers.
A Trial 5, April 6 to 14,1994. B Trial 6, October14 to 21, 1994







75

Table 6.1. Frequency and speed of wind on the four quadrants in Lake Alfred, CREC weather station. Data from hourly log in a seven-days period after the beginning date.
Average Variance
Trial # Orientation Frequency (Km/h) Variance

North 10 4.732 12.08
1 East 31 5.368 13.51 31/May/93 South 10 4.176 5.49 West 45 5.901 7.09

North 6 5.000 2.56
3 East 39 2.613 2.84
28/Jul/93 South 48 3.865 6.13
West 46 4.998 8.77

North 19 5.528 7.94
5 East 125 7.356 14.19
6/Apr/94 South 45 4.989 8.17
West 27 6.462 8.42

North 58 4.409 5.59
6 East 56 3.478 4.44
14/Oct/94 South 11 2.464 1.90
West 38 3.097 4.81





of wind speed against wind direction (azimuth) are also plotted for each experiment.

Additionally, Table 6.1 presents a summary of the wind conditions. Reference lines

in each wind plot represent quadrants of the compass: 315-450 for North, 45-1350

for East, 135-2250 for South, and 225-3150 for West. West wind means that air

movement occurred from that direction and toward the opposite direction, East. The

only significant differences between the number of crawlers trapped at each cardinal

direction were observed in trial 3 (July 28-August 2, 1993, Kruskall-Wallis test

H=9.4235 P<0.024). However, it is possible to observe a trend relating the speed








76

and frequency of air movement and the number of crawlers trapped in a given direction. In trial 1(May 31 June 3, 1993) the most frequent winds came from the East and West, with the strongest being from the East. The highest catch was observed in the West traps. Trial 3 provides another interesting case, wherein the frequency of wind was equivalent between East and West, but the strongest winds were from the West. The highest catches in this experiment were on the East traps. The wind came predominantly from the East in trial 5, and no catches were obtained from the trap on that side: the few crawlers caught were observed in North and West traps. The dominant wind direction in trial 6 was from North and East, but with slower speed than in previous tests. The highest catches were obtained in East, South, and West traps.

Infestation Scoring Maps. Figure 6.5 show the scores of citrus snow scale infestation on the citrus plot studied. Each bar represents the relative location and score of each tree in the plot. The original infestation was started in August 1992 but it was not evident until March 1993. Through 1993 the spread of snow scale showed a tendency toward the soutwest. This trend became more clear during 1994, but started to become confused by April 1994, when some northern trees started showing infestation.



Discussion

Despite the lack of statistical significance, a role of wind on the dispersion of the citrus snow scale can be inferred from Figures 6.1 and 6.2. The more frequent and strongest winds originated from the direction opposite to the traps that showed








77

August 1992 March 1993
15 15 A
N
10- 10

5- 5

I I I I I I I
4 8 12 16 4 8 12 16
April 1993 May 1993
15- 15

10- 10 .0 5-.
o
5- 5


4 8 12 16 4 8 12 16
June 1993 July 1993
15 15

10- 10

5 5

I I I I I I I
4 8 12 16 4 8 12 16

X Coordinate

Artificial infestations Score 3 o No infestation 0 Score 4 o Score 1 Score 5
Score 2

Figure 6.3 Maps of trees infested with citrus snow scale in a citrus plot (CREC,
N-40 block 16). Each circle represents the position of a tree in the plot.
Circle's shade indicates the intensity of snow scale infestation measured
as scores from 0 to 5.








78
August 1993 September 1993
15 00 15 A0 00 A
N
10- 10 *

5 54 8 12 16 4 8 12 16
October 1993 November 1993
15- 15

10- 105- 5

,I I I I II
14 8 12 16 4 8 12 16
o
0 December 1993 January 1994
15 15

10- 105- 5


4 8 12 16 4 8 12 16
February 1994 March 1994
15- 0 0 15 00

10- 10

5 5I I III
4 8 12 16 4 8 12 16 Figure 6.3 -- Continuation. X Coordinate








79
April 1994 May 1994
15- n a 15 A

10- 10- 5 5

II IIII
4 8 12 16 4 8 12 16
July 1994 August 1994
15 15- 10 10
0
O 5- 5 5- =


4 8 12 16 4 8 12 16
September 1994
15

10

5



4 8 12 16

X Coordinate









Figure 6.3 -- Continuation.








80
the highest crawler catches. The maps of progress of scale dispersal (Figure 6.3) do not provide compelling evidence for the relationship between wind direction and scale spread, since we lack the wind data, but indicate a general movement toward the southwest.

Quayle (1916) studied dispersal of black scale (Saissetia oleae) in citrus groves in a manner similar to that described here. He trapped black scale crawlers 450 feet (137 m) from the source, the farthest traps that he had set. He also trapped Aonidiella aurantii at 30 to 150 feet (9 to 45 m) from a source of crawlers.

The distance that an airborne object can travel depends on several factors such as wind speed, the height from which the object starts its descent, and the speed at which its falls (Margalef 1977). An additional consideration for the snow scale is the fact that the more dense scale colonies occur in the interior part of the tree (trunk, scaffold branches) where air movement is somewhat restricted and the canopy becomes a barrier for dispersion. Crawlers lifted by the wind from the trunk must overcome that barrier. It is expected that young trees with heavy infestations and sparse canopy will be better sources for short distance colonization, since winds close to the ground are slower and the height of the dispersing crawlers is limited in short trees. On the other hand, larger trees with a dense canopy should not be a good source of crawlers, but those few crawlers that are potentially lifted by wind from a high branch will have the opportunity to travel further.

Xinnian & Browning (1991) studied the movements of snow scale crawlers and concluded that they were positively phototactic and suspected but did not provide conclusive evidence for the effects of gravity. They also observed higher








81

activity and greater distance traveled by female crawlers. It is hypothesized that snow scale infestations may start low on the trunk and progress upwards. This hypothesis is supported by positive phototaxis or negative geotaxis coupled with higher mobility of females crawlers, and could be an adaptation for reaching higher levels on the tree. The possible advantages of this adaptation could be that it will increase the chances for dispersion of crawlers by wind. These conditions should enhance citrus snow scale potential for successfully colonizing neighboring trees.

The importance of phoresy on humans and other animals has been addressed for other species of scale insects. Stephens & Aylor (1978) collected and examined the feathers, beak and feet of birds for crawlers of Matsucoccus resinosae and failed to find them. On the other hand, Washburn & Frankie (1981) exposed humans and dogs to iceplants infested with Pulvinariella mesembryanthemi and found crawlers and egg sacs on clothing and fur. It appears unlikely that citrus snow scale is dispersed to any extent by humans given the location of citrus snow scale on the interior wood and the thorny nature of the citrus canopy. Both factors should limit contact between human clothing and the snow scale colonies on bark.













CHAPTER 7
MODELING CITRUS SNOW SCALE POPULATIONS Introduction

Models are hypotheses of how a system functions (McKinion 1992). A model can be designed to graphically represent relationships or to mathematically represent a system. Mathematical models can vary in complexity, creating a compromise between simplicity and accuracy. Simulation models tend to be complicated, and produce outputs that can be used to forecast the state of a system in the future, be it weather conditions, the populations of an organism, or the dynamics of a market. Models has been used extensively in forecasting and management of biological populations (Ruesink 1975, 1976, Getz & Gutierrez 1982, Curry & Feldman 1987).

Overall, models help to understand the processes and provide means to organize information about complex systems (Wagner & Willers 1992). However, in order to create an accurate representation, thorough knowledge of the system is required. Models also can be used to test the quality of the information available regarding a system. The model then becomes a tool to check the integrity of our estimation of its parameters.

Models are built through a series of steps. Ruesink (1976) describes four such steps: definition of the system and the objects within it; creation of

82








83

mathematical equations for each object defined; interconnection of the defined objects into a working model; and finally, testing the performance of the model against the real world. Once a model is constructed, two types of tests can be performed. Validation is testing the goodness of fit of the model against real data, while sensitivity analysis is the analysis of the model's dynamics when the values of the parameter estimates are changed.

The present chapter presents the development of a model for citrus snow scale dynamics, tests the estimates obtained from studies presented in previous chapters, establishes a framework for future expansion of studies, and identifies gaps in knowledge on biology and ecology of the citrus snow scale.



Materials and Methods

Model Structure. A distributed delay model (Manetsch 1976) was created to simulate the population dynamics of citrus snow scale. Distributed delay is especially useful since it models different stages (age) and the variation in the transit time of the individuals through a given stage (Berry & Stinner 1992). In this model, both sexes were simulated since we are interested in development and not in reproduction. In addition, males are an important numerical component of the population and the most obvious to the observer. The life cycle was composed of first nymphal stages, which develop into two sexes with sex ratio 8, a male stage, a second female stage, and adult female (third) stage. Eggs and crawlers were not included in the model, nor were the events occurring after sexual maturation of








84

females, since the information about those stages is incomplete (see discussion in Chapter 3). The model for males and females was structured as follows:

dN (t)
= S1 N (t) pi N(t)
dt

dN (t)
dN2'ft = S1 N (t) S2 N2,f() P2 N2,f(t) (Females )


dN 3, f (t) (7.1)
dt = S, N2,f(t) 3 3f
dt

dN2, m(t)
S = (1-) S N (t) S2 N2(t)- P2(t) N2,m(t) (Males)

dN3,m(t)
dt = S2 N2,mM(t) P3 N3, m(t)




Where N represents the number of individuals in the stage i; S, represents the developmental rate from stage i to stage i +1 (1/ time), p, represents the rate of mortality at stage i (instantaneous rate of death); and 6 represents the sex ratio as females per male. The parameters for the distributed delay model with attrition were calculated according to Vansickle (1977) as:

-2

S2

(7.2)
DEL = k

AR= 1 DEL I T DEL








85

Where and s2 are the average and variance of the time spent by the individuals in a stage, k is the number of delay stages (subsystems) that constitute each modeled stage, DEL is the mean delay that an individual spends in the modeled stage, e is the proportion that survive at the end of the stage, and AR is the instantaneous attrition rate, equivalent to the mortality rate p, previously mentioned.

Development. Development was modeled after Lactin et al.'s (1995) modification of Logan et al.'s (1976) poikilotherm model, using the data presented in chapter 3. Lactin's modification to the model of development is:

p T- T m T (7.3) rate = eT- e m T +A



where T is temperature, p is the rate of increase to optimum temperature, Tm is the maximum lethal temperature, AT is the difference between the lethal temperature and the optimum temperature and A is a scaling factor that allows for incorporating developmental thresholds (Lactin et al. 1995).

Mortality. Innate mortality was modeled using survival estimates for snow scale cultures at constant temperature (Chapter 3) and was expressed as the percentage of survival at the end of the stage (% of censored observations) (Procedure LIFETEST, SAS Institute Inc. 1989).

Innate mortality was modeled as dependent upon temperature; it could not be modeled as dependent upon RH because there were not sufficient data or evidence of this dependency (see Chapter 3). The fraction dying is a proportion of dead individuals in the previous stage (p,N,.,) and is subtracted from the rate of








86

change between stages (Equation 7.1) (Vansickle 1977, Curry & Feldman 1987). Linear regression was fitted to express the relationship between the proportion dying and temperature for each stage simulated.

Reproduction and Migration. Reproduction and migration are not considered in this model because they have not been adequately measured. Since reproduction occurs beneath the mother's secreted cover, it is not possible to obtain direct observations. Also, the patterns of movement of crawlers is such that newly emerged crawlers walk away from their mothers (Xinnian & Browning 1991). Reproduction could be irrelevant if a small area, such as the 18x12 mm2 patches of bark surface used in chapter 5, are considered for the simulation. Most of the newly emerged crawlers move beyond this area. On the other hand, if the simulation considers a larger scale, such as a branch or a whole tree, then reproduction becomes relevant. Inversely, migration will be the main component of recruitment on a small scale, but probably quite irrelevant at large scale (Chapter 6). Recruitment by migration was studied from field observations (Chapter 5) and the analysis of photo slides, but was not used to estimate the recruitment of new scales.

Environmental Variables. Temperature was the only environmental variable incorporated into the model. It was initially set constant for development of the model and for comparison to the constant temperature developmental data presented in Chapter 2. Temperature then was modeled as a compound cosine function after Brewster & Allen (1991):








87


T (t) = a cos 2 T7 Pea ))





where a was the amplitude of the departure from the average temperature T (maximum and minimum) and () is a fraction that synchronizes the period of the curve. The daily average temperature was in turn simulated in the same equation but using the annual measurement of average temperature and amplitude.

Model Implementation. The model was developed in a PC graphic environment using MATLAB 4.0 for Windows with Simulink v1.2c. (The MathWorks, Inc., 24 Prime Park Way, Natick, MA 01760). Runge-Kutta fifth order method with fourth order step-size control was used to integrate the system of differential equations (Press et al. 1992). The minimum step size used was 1x10' days. The Simulink block diagrams for the model are included in the Appendix.

Model Calibration. Data from the study of development at constant temperatures (Chapter 2) were used to calibrate the model. These data were recorded from nine cohorts of snow scale cultured on 'Duncan' grapefruit seedlings and maintained at five temperatures and two relative humidities. The goodness of fit of the model trajectories was tested against the data by X2 and KolmogorovSmirnov tests (Siegel 1956, Sokal & Rohlf 1981).

Model Validation. The study of development was repeated with a set of eight 'Duncan' seedlings infested with citrus snow scales under variable temperature (greenhouse conditions). The methods used to infest and record development of








88

scales were those described in chapter 2. The seedlings were maintained inside a cage in a greenhouse and continuous readings of temperature were taken with an hygrothermograph. The data were compared against the model by X2 and Kolmogorov-Smirnov tests.

Results

Table 7.1 presents the parameters calculated for the simulation of development of the snow scale from the observations reported in chapter 3. The values chosen were for curves at 60% RH which represents the data set with the most temperature observations.

Table 7.2 shows the mean, variance and the percentage of survival at the end of the stage for cultures of snow scale maintained under constant temperature and relative humidities. The parameters for the distributed delay model, calculated according to the methods described by Vansickle (1977) are shown in Table 7.3. The relationship between the fraction dying at the end of the stage and the temperature are expressed as the regression equations presented in Table 7.4. The sex ratio 6 was set at 2 males per female or 0.667 males/female (Chapter 3). Table 7.1 Parameters used for the simulation of development for the citrus snow
scale.

Stage p Tm AT A

1" stage 9 0.0109 35.4890 2.6085 -1.0910 i" stage a" 0.1503 34.0149 6.6330 -1.0x105
2d stage 9 0.1100 35.1010 9.0010 -0.1001
Males 0.1100 34.8024 9.0017 -0.0996









89




Table 7.2 Average duration and percentage of survival from each stage at each
environmental condition studied (Chapter 3).


Observed days in stage
Stage T RH % survival Average Variance
16 60 5.39 31.52 0.596 70 5.26 23.25 0.721 21 60 3.75 17.08 0.485 70 3.69 21.81 0.750 1" stage 24 60 1.98 6.69 0.697 70 3.80 17.56 0.535 28 60 2.73 13.08 0.615 70 3.19 12.74 0.548 30 60 2.76 7.82 0.512 16 60 32.54 241.10 0.531 70 34.26 251.44 0.680 21 60 15.78 46.50 0.410
2nd 70 1923 102.05 0.693
stage 24 60 13.29 37.22 0.618
females 70 12.31 41.79 0.373
28 60 11.64 24.74 0.682 70 11.65 37.22 0.333 30 60 13.98 39.18 0.547 16 60 25.67 311.15 1.000 70 33.97 358.29 1.000 21 60 33.61 229.41 0.857 70 17.18 161.22 0.976 Males* 24 60 15.61 66.70 1.000 70 17.29 81.34 1.000 28 60 17.91 121.49 0.973 70 17.29 65.34 0.938 30 60 3.72 81.12 0.850 16 60 64.09 30.13 1.000 70 56.37 229.27 0.947 21 60 44.10 65.64 0.857
Adult 70 60.27 97.38 0.970
females 24 60 44.23 121.00 1.000 70 29.14 21.92 1.000 28 60 34.74 91.58 0.976 70 50.62 331.22 1.000 30 60 20.81 331.93 0.727
* Includes secreting male 2nd stages, "prepupa" and "pupa".




Full Text
LIST OF FIGURES
Figure page
4.1 Survival and hazard functian gf snow scale populations
on selected trees 46
5.1 Age structure of citrus snow scale populations on citrus
leaves and twigs 59
5.2.Ratio of first instar per gravid female of snow scale on
samples of citrus twigs and leaves 60
5.3 Relative proportion of sexes of snow scale on samples
of citrus leaves and twigs 62
5.4. Incidence of natural enemies of snow scale on samples
of leaves and twigs 63
5.5 Relative importance of natural enemies of snow scale
females on samples of leaves and twigs 64
6.1. Wind speed vs direction and catches of snow scale
crawlers during trial 1 and 3 73
6.2. Wind speed vs direction and catches of snow scale
crawlers during trial 5 and 6 74
6.3. Maps of trees infested with citrus snow scale in a citrus
plot (CREC, N-40, block 16) 77
7.1. Duration of development of citrus snow scale
individuals under constant temperature 94
7.2. Duration of development of citrus snow scale
individuals under greenhouse conditions 95
IX


66
topological characteristics and on differential mortality. Since it is known that
different substrata influence development and survival (see chapter 3 and Miller &
Kostarab 1979, Cooper & Oetting 1986), the possibility of increased citrus snow
scale natality cannot be dismissed.
Mortality factors also may be operating differently between the various
substrata. Murdoch et al. (1989) demonstrated that rates of parasitism by Aphytis
melinus and Encarsia sp. on California red scale (Aonidiella aurantii) were higher on
the periphery of the canopy than on the trunk of citrus trees. They suspected that
lower attractiveness of the bark surface to parasitoids may play a role in this
phenomenon. Carroll (1979) also observed this phenomenon with Comperiella on
A. aurantii. The citrus snow scale system, while sharing many features, is not the
same as the one observed by Murdoch et al. Citrus snow scale samples in the
present studies came from the interior of the tree, near the trunk and main scaffold
limbs, and thus our comparison is between the interior leaves and supporting twigs.
However, the principle demonstrated by Murdoch et al. may apply equally well in the
sense that different substrata may have an effect on the searching and dispersal
capabilities of natural enemies.
Snow scale sex ratios have been previously reported as male biased
(Casares 1974). Our studies showed a male-based sex ratio in laboratory cultures
(Chapter 3) and in colonies on citrus trunks in the field (Chapter 4, Table 4.3).
However, the data from twigs and leaves in this study indicate a female biased sex
ratio in the canopy. The factors discussed earlier that could influence natality,


88
scales were those described in chapter 2. The seedlings were maintained inside a
cage in a greenhouse and continuous readings of temperature were taken with an
hygrothermograph. The data were compared against the model by x2 and
Kolmogorov-Smirnov tests.
Results
Table 7.1 presents the parameters calculated for the simulation of
development of the snow scale from the observations reported in chapter 3. The
values chosen were for curves at 60% RH which represents the data set with the
most temperature observations.
Table 7.2 shows the mean, variance and the percentage of survival at the
end of the stage for cultures of snow scale maintained under constant temperature
and relative humidities. The parameters for the distributed delay model, calculated
according to the methods described by Vansickle (1977) are shown in Table 7.3.
The relationship between the fraction dying at the end of the stage and the
temperature are expressed as the regression equations presented in Table 7.4. The
sex ratio 0 was set at 2 males per female or 0.667 males/female (Chapter 3).
Table 7.1 Parameters used for the simulation of development for the citrus snow
scale.
Stage
P
Tm
AT
A
1sr stage 9
0.0109
35.4890
2.6085
-1.0910
1st stage o"
0.1503
34.0149
6.6330
-1.0x10*5
2nd stage 9
0.1100
35.1010
9.0010
-0.1001
Males
0.1100
34.8024
9.0017
-0.0996


113
Okudai, S., R. Korenaga & Y. Sakagami. 1971. The effect of temperature on the
development of the arrowhead scale, Unaspis yanonensis. I. development
of the first generation under the constant temperatures. Bull. Hortic. Res. Stn.
Ser. B (Okitsu). 11: 193-201.
. 1974. Studies on the autumnal occurrence of larvae of the arrowhead
scale, Unaspis yanonensis Kuwana. II. On the relationship between
temperature in the early stage of the autumn adults and the autumnal
appearance of larvae. Bull. Hortic. Res. Stn. Ser. B (Okitsu). 1: 101-113.
. 1975. Effect of temperature on the ovarian development of the arrowhead
scale, Unaspis yanonensis Kuwana, in the late hibernating period. Bull. Fruit
Tree Res. Stn. Ser. B (Okitsu). 2: 97-106.
Pfeiffer, D. G. 1985. Pheromone trapping of males and prediction of crawler
emergence for San Jose scale (Homoptera: Diaspididae) in Virginia apple
orchards. J. Entomol. Sci. 20: 351-353.
Press, W. H., S. A. Teukolsky, W. T. Vetterling & B. P. Flannery. 1992. Numerical
recipes in FORTRAN: the art of scientific computing, 2nd. ed. Cambridge
University Press, Cambridge.
Quayle, H. J. 1916. Dispersion of scale insects by the wind. J. Econ. Entomol. 9:
486-493.
Rao, V. P. 1949. The genus Unaspis MacGillivray. Microentomology. 14: 59-72.
Reed, D. K., A. G. Selhime & C. R. Crittenden. 1967. Occurrence of citrus snow
scale, Unaspis citri, on several varieties of citrus in Florida. J. Econ. Entomol.
60: 300-301.
Riehl, L. A., R. F. Brooks, C. W. McCoy, T. W. Fisher & H. A. Dean. 1980.
Accomplishments toward improving integrated pest management for citrus,
pp. 319-363. In C. B. Huffaker [ed.], New technology of pest control. Willey-
Interscience, New York.
Rose, M. 1990. Diaspidid pest problems and control in crops: Citrus, pp. 535-541.
In D. Rosen [ed.], Armored scale insects: their biology, natural enemies and
control. Elsevier, Amsterdam.
Rosen, D. & P. DeBach. 1977. Diaspididae, pp. 78-128. In C. P. Clausen [ed.],
Introduced parasites and predators of arthropod pests and weeds: a world
review. ARS USDA, Washington, D C.


30
Table 3.7. Polynomial regressions between temperature and the parameters of a
gamma distribution (a, P) fitted to the probability density function of the
survival curves
Stage
RH
R2
Equation
1st stage
60
0.84
0.74
a = 131.70-6.79 7+0.084 T2
3 = 7.51 -0.07 7-0.0056 72
70
0.86
0.91
a = 338.20-30.10 7+0.691 72
3 = 37.11 -3.41 7+0.081 72
2nd stage
60
0.82
0.84
a = 30.24-2.89 7+0.081 72
3 = 0.79-0.09 7+0.029 72
70
0.74
0.82
a = -10.14-0.58 7+0.083 72
3 = 0.31 -0.13 7+0.007 72
Females
Pool
0.13
0.19
a = -242.50 + 25.49 7-0.56 72
3 = -3.48 + 0.36 7- 7.89 x 10'3 72
Males
Pool
0.33
0.55
a =1085.20-87.71 7+1.81 72
3 = 17.10-1.43 7+0.03 72
Discussion
Optimal temperature for development and upper developmental thresholds
(Tm) can be estimated from the experimental data and its inclusion in the Logan
model. Optimal temperatures ranged between 25 and 38C for all the stages and
both sexes, with values falling in the vicinity of 29C for 1st instars and 26C for 2nd
instars. Upper developmental thresholds ranged between 34 and 44C with most
values falling in the vicinity of 35C (Table 3.3). The optimal temperatures for 2nd


110
distribution of the hosts on citrus leaves and searching behavior of the
parasite. Sci. Bull. Fac. Agrie. Kyushu Univ. 31: 113-117.
. 1977a. Searching and ovipositing behavior of Aspidiotiphagus citrinus
(Craw), a parasite of the arrowhead scale, Unaspis yanonensis (Kuwana). 4.
effect of host density on the time spent by the parasite on a citrus leaf and
the number of hosts parasitized. Sci. Bull. Fac. Agrie. Kyushu Univ. 31: 151-
157.
. 1977b. Searching and ovipositing behaviour of Aspidiotiphagus citrinus
(Craw), a parasite of the arrowhead scale, Unaspis yanonensis (Kuwana). 3.
influence of the structure of scales on the ovipositing behavior of the parasite.
Sci. Bull. Fac. Agrie. Kyushu Univ. 31: 125-131.
Kanda, K. & FI. Kajita. 1977. Searching and ovipositing behavior of Aspidiotiphagus
citrinus (Craw), a parasite of the arrowhead scale, Unaspis yanonensis
(Kuwana). 2. searching behavior for host females on the citrus leaf. Sci. Bull.
Fac. Agrie. Kyushu Univ. 31: 119-124.
Kloppenburg, J., Jr. & D. L. Kleimman. 1987. The plant germplasm controversy.
BioScience. 37: 190-198.
Knapp, J. L. (ed.). 1995. 1995 Florida citrus pest management guide. Florida
Cooperative Extension Service/IFAS Univ. of Florida, Gainesville.
Korenaga, R. 1983. The time of hatching and crawling speed of U. yanonensis
(Flomoptera: Coccidae). Jap. J. Appl. Entomol. Zool. 27: 308-310.
Korenaga, R. & Y. Sakagami. 1981. Prediction of the occurrence of the arrowhead
scale Unaspis yanonensis (Kuwana), pp. 700-704. In K. Matsumoto [ed.],
Proceedings of the International Society of Citriculture. Organizing
committee, Tokyo, Japan.
Korenaga, R., S. Okudai & Y. Sakagami. 1974. Studies on forecasting the
occurrence of the arrowhead scale, Unaspis yanonensis Kuwana. I.
Forecasting occurrence date of the first generation first instar larvae from the
development of male larvae and of ovaries of mature female in late
hibernating period. Bull. Hortic. Res. Stn. Ser. B (Okitsu). 1: 85-99.
Korenaga, R., Y. Sakagami & S. Okudai. 1976. The effect of temperature on the
development of the arrowhead scale, Unaspis yanonensis Kuwana. II.
development of the first generation under fluctuating temperature. Bull. Fruit
Tree Res. Stn. Ser. B (Okitsu). 3: 47-56.


57
have remained on the substratum. Counting them would have overestimated the
effect of predation.
The sampling took place between May 1992 and December 1993, comprising
14 sampling dates. The period between sampling dates was sufficient to allow the
emergence of parasitoids in immature stages on the previous sampling date and
also to avoid duplicate sampling of the same snow scale cohort.
The proportions of individuals of each age (stage) group were calculated,
adding the numbers of healthy individuals and the numbers of parasitized individuals
that may have belonged to the same cohort. Internally-parasitized females were
classified as second instars, while externally parasitized females were classified with
gravid females, because they were attacked as prereproductive females and they
are part of the gravid-female cohort. Sex ratio was calculated by adding the number
of healthy and parasitized individuals, i.e., total healthy males plus males bearing
internal parasitoids, healthy second instars females plus prereproductive adult
females plus females bearing any parasitoids. Reproducing females were not
added because the older males of equivalent age had emerged and thus were not
accounted for in sex ratio calculations. The incidence of parasitoids and fungi was
calculated as a percentage of parasitism based on the total individuals of a given
sex and stage.
The statistical analysis involved distribution comparison of the percentage
age structure between substrata using the Kolmogorov-Smirnov two sample test
and Kruskall-Wallis non parametric multiple comparisons for detecting differences
between sampling dates (Siegel 1956, SAS Institute Inc. 1989).


17
China in 1980 and showed good results (Furuhashi & Nishino 1983). A coccinellid,
Chilocorus kuwanae, and a nitidulid, Cybocephalus sp., have been introduced from
Korea into the U.S.A. for the control of Unaspis euonymi and field released in 1984
(Drea & Carlson 1987, 1988).


90
Table 7.3 Parameters for the distributed delay model, calculated from data at
constant temperature and RH (Chapter 3).
Stage
T
RH
k
DEL
AR*
16
60
0.9
9.449
0.0735
70
1.2
6.924
0.0544
21
60
0.8
9.029
0.1284
70
0.6
5.850
0.0624
1st stage
24
60
0.6
3.665
0.1360
70
0.8
8.141
0.1154
28
60
0.6
6.418
0.1198
70
0.8
6.776
0.1325
30
60
1.0
5.486
0.1754
16
60
4.4
37.582
0.0181
70
4.7
37.215
0.0108
21
60
5.4
18.646
0.0521
70
3.6
21.279
0.0181
2nd stage
24
60
4.7
14.707
0.0344
females
70
3.6
16.164
0.0702
28
60
5.5
12.487
0.0318
70
3.6
15.746
0.0814
30
60
5.0
15.777
0.0406
16
60
2.1
25.669
0.0000
70
3.2
33.973
0.0000
21
60
4.9
34.683
0.0045
70
1.8
17.404
0.0014
Males**
24
60
3.7
15.611
0.0000
70
3.7
17.295
0.0000
28
60
2.6
18.095
0.0015
70
4.6
17.532
0.0037
30
60
0.2
9.637
0.0282
16
60
136.3
64.091
0.0000
70
13.9
56.589
0.0010
21
60
29.6
44.334
0.0035
70
37.3
60.322
0.0005
MUUIl
24
60
16.2
44.231
0.0000
females
70
38.7
29.143
0.0000
28
60
13.2
34.804
0.0007
70
7.7
50.619
0.0000
30
60
1.3
26.566
0.0136
* k is the number of subsystems (delay equations) that constitute each stage; DEL is the delay that
an individual spend in each stage; AR is the attrition rate.
** Includes secreting 2nd stage male, "prepupa" and "pupa".


67
mortality and movement of citrus snow scale populations on twigs and leaves could
have contributed to the observed reverse trend in sex ratio on the canopy.
The important parasitoids affecting mortality of snow scale in this study were
Aphytis lingnanensis and Encarsia species. While Encarsia species attacked both
sexes, A. lingnanensis attacked females exclusively. Parasitism by Encarsia was
higher on females (Figure 5.3A, 5.3B), favoring a male biased sex ratio. Since
females are subject to increased pressure from these two sources of mortality,
parasitism likely would push the sex ratio bias toward males. This was not the
situation observed on leaves and twigs. Differences in dispersal behavior between
sexes have been reported for other species of scales (Arias-Revern 1988) and for
snow scale (Xinnian & Browning 1991). Female crawlers generally followed a
straighter course than males and were more active, reaching farther during this
important dispersal phase. Since females feed for a longer period (Chapter 3) and
likely require more nutrients for development and reproduction, differential dispersal
behavior may be an adaptation to reduce intraspecific competition and to expand
colonization into more remote locations. Sexual dimorphism in the adult form in
scale insects provides a mechanism for winged males to disperse in search of
female receptive to mating, countering the limitations on male crawler dispersal and
reducing the impacts of inbreeding.
In a study of citrus snow scale prior to the introduction of A. lingnanensis "HK-
1", the incidence of parasitism by Encarsia longsburyi was presented as number of
parasitized hosts per unit of habitat (leaf) and per unit of area (cm2 of twig) (Casares
1977). Relative proportions of the population impacted by Encarsia parasitoids are


3
of development on suitable substrata, elucidating causes of mortality that affect
citrus snow scale in the field, and looking into the results of previous attempts of
biological control and the potential reasons why those may have been unsuccessful.
The present work is organized into chapters along major areas of the
research. The next chapter presents background information about the genus
Unaspis, the species citri and its close relatives. It also introduce the techniques
that were used in the research reported in later chapters. The third chapter
describes the development of snow scale under constant temperature and constant
relative humidity conditions, on a suitable substratum of a common Florida citrus
variety. The fourth chapter discusses mortality factors affecting snow scale under
field conditions on the bark of citrus trees. Using photographic techniques and
survival table analysis, the role of natural enemies is characterized. The fifth
chapter complements the fourth chapter with observations on mortality causes, age
structure and sex ratios in the canopy of citrus trees. The sixth chapter looks into
the potential of dispersion of snow scale via wind. The seventh chapter summarizes
the snow scale knowledge in a simulation model that evaluates the quality of the
new information and identifies gaps requiring additional research. Since each
chapter is designed to stand alone as a single research paper, a certain amount of
redundancy can be found between Chapter 2 (Literature review) and the
introductory sections of each subsequent chapter.


65
occurred in September 1993 on twigs. Fungi were predominant in August (leaves,
13 of 24 individuals) and October 1993 (twigs, 7 of 11 individuals). Mites were
observed in September, 1992 and January, June and October 1993.
Mites were found living under the bodies of live adult reproductive females
with their mouthparts inserted into the insect body, apparently sucking fluids. The
scale continued to oviposit. Unlike the other natural enemies, these mites were not
preventing reproduction, but very likely they were producing some deleterious effect.
Discussion
The predominance of early instars in the age structure of a population
suggests high natality in the population, or high mortality of older stages (Odum
1971). In the case of scale insects, high immigration of crawlers is also possible,
since the first stage is the only one able to move. A high proportion of young instars
were observed on leaves over twigs during most of 1992 and 1993, while
predominance of older stages occurred in October to December 1993. This pattern
roughly follows the changes in air temperature for those times of the year. As
shown in chapter 3, development of citrus snow scale is dependent on temperature.
The high ratio of newly settled first instars to reproductive females (Figure
5.2) could be the result of higher natality on leaves, but more probably it is explained
by movement of crawlers from the twigs and limbs onto the leaves, or a combination
of both. Carroll (1979) observed differences in substrate preference by California
red scale Aonidiella aurantii, a species that can colonize most of the organs of a
citrus tree. He explained the substrate preference based on phenological and


50
the year, the insects were exposed to different weather conditions. Since
development depends on temperature (Chapter 3), scale insects have shorter life
cycles during the warmest months (Figure 4.1). It is also expected that extreme hot
temperatures increase mortality, but this is not obvious from the data presented in
Figure 4.1. It is also assumed that natural enemies have shortened life cycles at
higher temperatures.
Overall, within all of the cohorts studied, (Table 4.2) 24% of males (118 out
of 495) and 18% of females (10 out of 56) survived to the end of the observation
periods. Most of the mortality observed could not be assigned to a specific cause.
This unknown mortality affected 59% of the individuals in the cohorts (568 out of
960) and accounted for 68% of the total mortality. Natural enemies were an
important component; detectable predators killed 15% of the individuals, while
parasitoids killed 11%, representing 17% and 13% of the total causes of mortality,
respectively.
The extent of variability in the data can be observed in table 4.3. Despite this
variability, total mortality (adx) never fell below 70%, and varied up to 100%.
Detectable predation appeared more important than parasitism (0-49% against 0-
38%) and males demonstrated higher probabilities of being attacked by both factors
than females (Table 4.2). Secreting males and second instars of both sexes were
the stages with highest mortality (Table 4.5), perhaps because the higher proportion
of males over females means higher availability of male prey or hosts. Males could
also be more susceptible given their softer shield. Males and females are available
for attack by parasitoids for a similar period. Even though males have a shorter life


45
Table 4.2. Significance of log-rank test, Wilcoxon homogeneity test, likelihood ratio
test for comparison of survival of citrus snow scale between trees by dates.
Date started
# of trees
Significance
17/Dec/1991
2
Wilcoxon P < 0.047
06/Mar/1992
2
Log-Rank P < 0.040
24/Aug/1992
2
Log-Rank P < 0.008
26/Oct/1992
2
ns
24/May/1993
2
ns
6/Oct/1993
2
Log-Rank P < 0.024
Wilcoxon P < 0.016
Survival ranged between 0 (Jun., Aug., and Oct. 1992) and 30.5% (Aug.
1992) in the 21 mortality tables constructed. The main mortality was classified in the
"unknown causes" category, but predation and parasitism provided an important
contribution to the total (Table 4.4). Between 0 (Jan 1992, Oct. 1993) and 48.8%
(Sept 1993) of the initial cohort died by the action of predators, between 0 (Jun., Jul.
1992, Apr., May, Aug., Oct. 1993) and 38.3% (Aug. 1992) died by the action of
parasitoids. It is important to mention that most of the parasitism observed was
attributed to Encarsia, only one female out of 13 observed parasitized was killed by
Aphytis. Unknown causes accounted for 29.3% (Sept 1993) to 90% (Oct. 1993).
Table 4.5 shows a summary of the total mortality (all the causes) presented
by stage. The highest mortality usually occurred on armor-secreting males.
However, in three cases, mortality was the highest in the second stage prior to scale
secretion (Dec. 1991, Jun. 1992, May 1993); in one case mortality was greatest on


TABLE OF CONTENTS
page
ACKNOWLEDGMENTS ii
LIST OF TABLES vii
LIST OF FIGURES ix
ABSTRACT x
CHAPTERS
1 INTRODUCTION 1
2 LITERATURE REVIEW 4
Taxonomy of the Genus Unaspis 4
Economic Importance 6
Biology 6
Dispersal 10
Phenology 10
Trophic Relationships 11
Host Plants 11
Natural Enemies 12
Management 15
Sampling and Forecast 15
Chemical control 16
Biological control 16
3 DEVELOPMENT AND MORTALITY
UNDER CONSTANT TEMPERATURE
AND RELATIVE HUMIDITY 18
Introduction 18
Materials and Methods 20
Results 24
iv


87
T (t) = a cos
1 1 t peak N N
2 n
period
T
7.4
where a was the amplitude of the departure from the average temperature T
(maximum and minimum) and 0 is a fraction that synchronizes the period of the
curve. The daily average temperature was in turn simulated in the same equation
but using the annual measurement of average temperature and amplitude.
Model Implementation. The model was developed in a PC graphic
environment using MATLAB 4.0 for Windows with Simulink v1,2c. (The MathWorks,
Inc., 24 Prime Park Way, Natick, MA 01760). Runge-Kutta fifth order method with
fourth order step-size control was used to integrate the system of differential
equations (Press et al. 1992). The minimum step size used was 1x1 O'3 days. The
Simulink block diagrams for the model are included in the Appendix.
Model Calibration. Data from the study of development at constant
temperatures (Chapter 2) were used to calibrate the model. These data were
recorded from nine cohorts of snow scale cultured on 'Duncan' grapefruit seedlings
and maintained at five temperatures and two relative humidities. The goodness of
fit of the model trajectories was tested against the data by x2 and Kolmogorov-
Smirnov tests (Siegel 1956, Sokal & Rohlf 1981).
Model Validation. The study of development was repeated with a set of eight
'Duncan' seedlings infested with citrus snow scales under variable temperature
(greenhouse conditions). The methods used to infest and record development of


97
predators on the citrus snow scale and how changes in these elements will modify
the dynamics of its populations.


POPULATION STUDIES OF THE CITRUS SNOW SCALE
UNASPIS CITRI (COMSTOCK)
By
JULIO M. ARIAS REVERON
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1995
UNIVERSITY OF aORIDA LIBRAR,ES


Hazard Survival
Tree 36-2, 17/Dec/1991 Tree 35-17, 27/Apr/1992 Tree 32-15,17/Jun/1992 Tree 26-16,12/Oct/1992
Tree 23-13, 24/May/1993 Tree 22-9, 17/Aug/1993 Tree 25-6, 16/Sept/1993
First molt
Secretion
Female 2nd molt
Full size male
Full size female
0 20 40 60 80 100 120 20 40 60 80 100 120 20 40 60 80 100 120
Age (days)
Figure 4.1 Survival and hazard functions of snow scale population on selected trees.
Vertical lines represent time of molts.


14
Fungal diseases have been reputed to play an important role in citrus snow
scale population regulation, especially because crowded citrus snow scale colonies
provide an ideal situation for the spread of disease. Several fungal genera have
been reported in association with snow scale (Sphaerostilbe, Podonectria) (Dickens
1968). Most of those reports are of species now recognized within the genus
Nectria (anamorph Fusarium, Samson et al. 1988) Some debate and contradictory
results of experimentation emerged from the period between the end of last century
to the late 1940s regarding the effectiveness of these fungi in the control of scale
insect populations. Nectria species were confirmed as developing saprophytically
on the bodies and covers of dead scales, but it is not clear whether these fungi also
function as true pathogens (Ziegler 1949, Fisher et al. 1949, Fisher 1950).
Natural enemies are an important means of population regulation. The role
of a given species can be studied by laboratory experimentation or by field
observations and manipulations. Unfortunately, observations in the field could be
influenced by many uncontrollable factors. One way to quantify mortality and its
causes is by using life table analysis. Life tables consist of a systematic accounting
of the number of individuals alive and dead at each age or stage and, when
possible, the causes of death. Life table analysis is a technique borrowed from the
insurance business that was applied to animal populations (Hutchinson 1978).
Applications of life table analysis to insects are reviewed in Harcourt (1969), Varley
et al. (1973), Southwood (1978) and Carey (1993).


12
It has been shown that differential preferences exist for the substrate chosen
by each sex of U. euonymi. Therefore, sex ratio was observed to be different on
different organs of the plant (Benassy & Pinet 1972). Males prefer to settle on
leaves, while females prefer to settle on woody organs. This could be related to
intraspecific competition and risks of mortality associated with leaf abscission. Since
females need to feed for a longer period than males, it may be argued that females
prefer to settle on twigs, which are a more permanent substrate than leaves. Males
on the other hand, tend to settle on leaves to avoid competition with females and
since they only feed for the first two instars, they may survive even if the leaf falls
from the plant (Cockfield & Potter 1987). Citrus trees have perennial foliage and
leaves remain on the tree for long periods, so the ratios or settling preferences could
be different for snow scale.
Natural Enemies. Natural enemies of snow scale have been reported in a
variety of publications and have been reviewed by Waterhouse & Norris (1987). A
discussion of the more important species follows.
Encarsia (=Aspidiotiphagus) citrina (Craw.) and E. lounsburyi (Berlese and
Paoli)(Hymenoptera: Aphelinidae) are the most commonly found and probably most
widely distributed citrus snow scale parasitoids (Hely 1944, Casares 1977, Selhime
& Brooks 1977, Tern et al. 1985, Castieiras & Obregn 1986, Fernandez Argudin
1987). They are polyphagous endoparasitoids that have been closely studied in
relation with other species of host scales (Benassy & Pinet 1972, Kajita 1972, 1976,
1977a, b, Murakami et al. 1972, Kanda & Kajita 1977, Gill et al. 1982). Encarsia


64
Leaves
Twigs
Leaves
Twigs
May 1992 Jun 1992 Aug 1992 Sep 1992

8
Nov 1992 Jan 1993 Mar 1993 May 1993
O
Leaves
Twigs
Jun 1993
Aug 1993
6_
Sep 1993
0
Oct 1993
Leaves
Twigs
Nov 1993 Dec 1993
Figure 5.5. Relative importance of natural enemies of snow scale females on
samples of leaves and twigs. Percentages are calculated over the total
attacked females.


15
Management
Sampling and Forecasting. The snow scale presents special problems for
sampling, since it occurs mainly on citrus trunks and scaffold limbs. These
substrata cannot be easily harvested and processed in the laboratory, as leaf
samples are. Sampling methods for citrus snow scale include the use of rating
systems for the appearance of scale colonies and the clearing of rectangular
patches within active colonies on the trunk. The patches are colonized by scales,
and then scales are counted in situ with the help of a hand lens (Casares 1977).
The rating methods tend to overestimate the active scale populations since they rely
on the appearance of male covers which can accumulate on the trunk for extended
periods, and thus represent both live scales and those that have already emerged.
Female citrus snow scales are generally overlooked in rating methods since they are
difficult to see on citrus bark.
Extensive efforts have been invested in creating models that predict
population densities of the arrowhead scale, U. yanonensis (Takezawa & Uchida
1969, Nishino & Furuhashi 1971a, b, Korenaga et al. 1974, 1976, Nishino 1974,
Korenaga & Sakagami 1981, Sakagami & Korenaga 1982, Adachi & Korenaga
1992). These models also predicted the results of interactions between biological
and chemical control strategies (Adachi & Korenaga 1992). Simulation models have
been used to predict and manage other species of scale insects as well (Pfeiffer
1985, McClain et al. 1990a, b).


100
Snow scale crawlers can disperse by taking advantage of air currents as has
been demonstrated for many other scale species. The importance of this means of
dispersion needs to be investigated further. Questions arise regarding how far
crawlers can be carried by wind, and how long they remain viable after leaving the
site of their birth.
A distributed delay model of the citrus snow scale built around data obtained
on studies of development (Chapter 3) mimics the dynamics of the cohort
development but it is not precise, especially for late instars (adults) and under low
temperature conditions. Further research and development are needed to refine the
model for citrus snow scale population dynamics.
Systems in which measurement of stage-specific population densities can
serve to establish timing for management intervention, are good candidates for the
implementation of simulation models to aid in the decision making process (e.g.,
Pfeiffer 1985, McClain et al. 1990). Simulation models do not yet seem promising
as management aids in the case of the citrus snow scale, because knowledge about
this insect is inadequate. However, the model was useful as a tool to understand
the system and identify areas for additional research. Expansion of the model
developed in the studies reported here to a more precise, comprehensive stage can
be a powerful tool to study the interactions between citrus snow scale and its natural
enemies, helping to predict the outcome of biological control experiments.
Although our comprehension of citrus snow scale biology has increased, it
is necessary to expand our understanding of the impact of the citrus snow scale on
the biology of citrus trees. This understanding could lead to enhanced efforts to


49
Table 4.5. Total percent mortality (all causes) adx presented by stage.
Tree
Date
Started
1st instar
2nd instar
Secreting
3rd instar
Female
Male
Female
36-2
17-Dec-91
0.1374
0.4656
0.1756
0.0076
0.0076
36-7
17-Dec-91
0.0806
0.2742
0.4194
0.0645
0.0000
35-15
23-Jan-92
0.1667
0.3125
0.3542
0.0000
0.0000
35-18
06-Mar-92
0.0323
0.1290
0.7097
0.0323
0.0323
34-21
14-Mar-92
0.0225
0.2360
0.5955
0.1124
0.0112
35-17
27-Apr-92
0.1818
0.2045
0.3485
0.0379
0.0379
34-19
19-May-92
0.0357
0.3214
0.5357
0.0179
0.0357
32-15
17-Jun-92
0.1220
0.4146
0.3659
0.0732
0.0244
34-11
29-Jul-92
0.0000
0.1000
0.2000
0.1000
0.4000
26-9
24-Aug-92
0.0588
0.2353
0.5000
0.0588
0.1471
28-7
24-Aug-92
0.0500
0.2000
0.2250
0.0500
0.2000
26-16
12-Oct-92
0.0968
0.3871
0.5161
0.0000
0.0000
27-14
26-Oct-92
0.0417
0.2500
0.4583
0.0833
0.0000
28-14
26-Oct-92
0.1176
0.1765
0.5882
0.0000
0.0588
24-2
30-Apr-93
0.0303
0.2727
0.3636
0.0303
0.0000
23-13
24-May-93
0.0755
0.1509
0.4340
0.0943
0.1321
24-12
24-May-93
0.0909
0.3333
0.1818
0.0303
0.1212
22-9
17-Aug-93
0.4231
0.0769
0.3077
0.1154
0.0000
25-6
16-Sep-93
0.0488
0.0732
0.512
0.1220
0.1463
21-12
06-Oct-93
0.4000
0.2000
0.3500
0.0000
0.0000
22-14
06-Oct-93
0.0000
0.0000
0.3750
0.3750
0.0000
Significant differences were detected between some patches in different
trees, and between trees where patches were set at the same dates. This could be
due to the different times of the year in which the patches were initiated (Table 4.1),
and also could be due to different locations of the trees inside the grove (Table 4.2),
or a combination of both. Given that the patches were set on different dates through


CHAPTER 6
THE ROLE OF WIND IN CRAWLER DISPERSION
Introduction
Diaspidid scale insects are sessile organisms with only two very limited motile
stages, adult males and newborn first instars (crawlers). Adult males can fly, but by
themselves they cannot establish new colonies. Crawlers are very fragile and of
limited mobility; even this stage may be limited in colonizing new hosts. It has been
proposed that the main form of dispersal for the snow scale is through crawler
phoresy on the clothing and equipment of grove crews in the citrus groves
(Simanton 1973), but little attention has been paid to the role of air movement as a
dispersal agent. Wind have been demonstrated to aid dispersal between plants for
several scale insect species (Quayle 1916, Brown 1958, Greathead 1990). This
chapter explores the role of wind as a mechanism of dispersion for the citrus snow
scale, Unaspis citri.
Methods
Heavily infested, crawler-producing trees were identified in two citrus groves
in Lake Alfred, Polk Co. Florida. These 'Valencia' orange trees were about 1.5-2 m
tall in a block planted at 5x6.5 m spacing, and adjacent trees held little or no snow
scale. Sticky cards fastened to 1,2-m tall poles were set around the source tree at
70


32
Casares (1977) discussed the possible effects of humidity on citrus snow
scale populations. Casares cited observations by W.A.T Summerville in Australia
that heavier infestations of citrus snow scale occurred during dry periods than during
wet periods, concluding that humidity was an important factor in snow scale
development. In our work, the effect of relative humidity on development was not
consistently significant, but the interaction with temperature was, indicating that
combined factors have an effect (Table 3.2). Only 2 conditions of humidity, differing
by 10%, were used, because this was not a major objective of the developmental
studies. These humidities were chosen to represent average conditions occurring
in the field (range between 50 and 90% RH during 1992 and 1993). Differences in
developmental parameters would be expected from studies using relative humidities
representing a wider range and thus incorporating more marginal conditions.
The modification to the Logan nonlinear model was suitable to describe the
data presented (Table 3.3). However, the reliability of the parameter estimates was
not satisfactory for development of prepupal and pupal males because of the
imprecision previously mentioned.
It has been shown that humidity may affect the rate of development for other
species (Aonidiella aurantii, McClure [1990b]) but it may have a more important
effect on survival than on development (Atkinson 1983). This effect may be direct
in the field, altering the survival, rate of development and reproduction of the insect
(McClure 1990a) or may be indirect, altering the action of its natural enemies.
In the current studies, temperature most notably affected mortality associated
with 1st and 2nd instars and males (Table 3.4). Relative humidity effects were


11
Trophic Relationships
Host Plants. Citrus species are the primary hosts plant for snow scale,
although the scale has been reported from hosts in related genera of Rutaceae such
as Murraya paniculata, Severina buxifolia, and Fortunella sp. (Rutaceae) (Casares
1974, Williams & Watson 1988). Reports of infestations of citrus snow scale on
plants in other families (Palmae, Celastraceacea, Oleacea, Bromeliacea) may be
inaccurate (Casares 1974, 1977) and represent misidentification of other armored
scale species with similar appearance.
Mechanisms of feeding in diaspidid scales are not well understood. It has
been proposed that the stylets penetrate woody tissues to find active sieve tubes
close to the cambium. Another suggested mechanism is that the scales feed on the
contents of any cell invaded by the stylets. The third proposal is that both forms of
feeding may occur. The first mechanism does not explain why diaspidid scales do
not produce large amounts of honeydew secretions. It is hypothesized that unused
materials may be returned to the plant by the insect pumping them with its powerful
salivary glands (Banks 1990). Studies on U. euonymi support the second
hypothesis, intracellular cell feeding on palisade parenchyma in leaves, and feeding
on xylem tissue in limbs (Sadof & Neal 1993). Citrus snow scale damage and
feeding mechanisms were studied by Albrigo & Brooks (1977). They showed that
stylets penetrate the plant tissue intracellularly, damaging cells on the way to
locating the phloem vessels, but they assumed phloem feeding without confirming
it.


22
The modification of the Logan et al. (1976) nonlinear model by Lactin et al.
(1995) was chosen to fit the data. This model was preferred because it is
descriptive, the parameters may be interpreted biologically, it is more realistic than
linear models or symmetrical nonlinear models (Lamb et al. 1984), it is simpler to fit
to the data than models with more parameters (Wagner et al. 1984), and it includes
a developmental threshold. The modified Logan model has 4 parameters to
describe the effect of temperature on the development of poikilotherm organisms:
(3.1)
where 7is temperature, pis the rate of increase to optimum temperature, Tm is the
maximum lethal temperature, AT is the difference between the lethal temperature
and the optimal temperature of development, and A is a parameter that makes the
curve intercept the x-axis, allowing the estimation of a developmental threshold. In
the Logan model, temperature values in Celsius are transformed to the base
temperature, the values for duration of development are inverted to calculate rates
of development, and the means of log transformed rates are used to calculate the
model parameters (Logan et al. 1976). In this research, as in the work of Lactin et
al. (1995), all those manipulations were avoided, minimizing the errors that may
arise from computation (Kramer et al. 1991). The inverse equation of the model was
used to fit the original mean duration of each stage (in days), instead of using the


107
Berry, I. L & R. E. Stinner. 1992. Computer development of insect population
models, pp. 54-73. In J. L. Goodenough and J. M. McKinion [eds.], Basic of
Insect Modeling. American Society of Agricultural Engineers.
Brewer, B. S. & A. D. Oliver. 1987. Euonymus scale, Unaspis euonymi (Comstock)
(Homoptera: Diaspididae): effects of host cultivar age, and location on
infestation levels. J. Entomol. Sci. 22: 119-122.
Brewster, C. C. & J. C. Allen. 1991. Simulation of plant resistant in a celery-leaf
miner-parasitoid model. Florida Entomol. 74: 24-41.
Brown, C. E. 1958. Dispersal of the pine needle scale, Phenacaspis pinifoliae
(Fitch), (Diaspididae: Homoptera). Can. Entomol. 90: 685-690.
Browning, H. W. 1994. Biological control of the citrus snow scale Unaspis citri in
Florida: evaluation of Aphytis and other natural enemy species, pp. 119-142.
In D. Rosen [ed.], Advances in the study of Aphytis (Hymenoptera:
Aphelinidae). Intercept Press, Andovers, UK.
Carey, J. R. 1993. Applied demography for biologists, with special emphasis on
insects. Oxford Univ. Press, New York.
Carr, D. S. & B. L. Harris. 1949. Solutions for maintaining constant relative humidity.
Ind. Eng. Chem. 41: 2014-2015.
Carroll, D. P. 1979. Within-tree distribution and host substrate influences on
California red scale, Aonidiella aurantii (Mask.); density, survival,
reproduction and parasitation. Ph D. Thesis University of California,
Riverside.
Casares, R. 1974. Effects of temperature on the development and fecundity of citrus
snow scale, Unaspis citri (Comstock) (Homoptera: Diaspididae). M.Sc.
Thesis University of Florida, Gainesville.
. 1977. Population dynamics of the citrus snow scale Unaspis citri
(Comstock) (Homoptera: Diaspididae) at Paradise Island, Osceola county,
Florida. Ph.D. Thesis University of Florida, Gainesville.
Castieiras, A. & O. Obregn. 1986. Toxicity of six pesticides used on citrus crops
against Aspidiotiphagus lounsburyi. Ciencia Tec. Agrie., Proteccin de
Plantas 9: 73-79.
Cockfield, S. D. & D. A. Potter. 1987. Distribution, development, and feeding impact
of Eunonymus scales (Homoptera: Diaspididae) on Eunonymus fortunei
under greenhouse conditions. Environ. Entomol. 16: 917-921.


I certify that I have read this study and that in my opinion it conforms
to acceptable standards of scholarly presentation and is fully adequate, in
scope and quality, as a dissertation for the degree of Doctor of Philosophy.
Harold W. Browning, Cha/r
Associate Professor of Entomology
and Nematology
I certify that I have read this study and that in my opinion it conforms
to acceptable standards of scholarly presentation and is fully adequate, in
scope and quality, as a dissertation for the degree of Doctor of Philosophy.
P. Allen
ssor of Entomology and
Nematology
I certify that I have read this study and that in my opinion it conforms
to acceptable standards of scholarly presentation and is fully adequate, in
scope and quality, as a dissertation for the degree of Doctor of Philosophy.
Q/jftr
Carmine A. Lanciani
Professor of Zoology
I certify that I have read this study and that in my opinion it conforms
to acceptable standards of scholarly presentation and^i fully adequate, in,
scope and quality, as a dissertation for theopOodpr laytor/W. McCoy Jr. /
Professor of Entomology and
Nematology


2
organisms make their way to a new area where the host plant is being cultivated as
a crop, they usually reach the area without their own natural enemies, and become
free to colonize and exploit the resource plant without constraints of predation,
parasitism or diseases. These organisms then become pests.
The citrus ecosystem in Florida is dominated by an exotic plant species
brought to a geographical area with a benign environment, and where many of its
pest organisms have been arriving one at a time. Management of these citrus pests
has relied upon use of pesticides, but in the recent times, more pest species have
been successfully managed by a technique of ecosystem reconstruction, biological
control. Biological control consists of bringing to the new ecosystem some of the
missing components from the original ecosystem, particularly the natural enemies
that feed or develop on the insect pests. Biological control is a management
technique that requires a very good understanding of the relationships between the
plant, the pest and its natural enemies.
The citrus snow scale Unaspis citri (Comstock) is today an important pest of
citrus in Florida. It was a minor pest in the past, but its relative importance has
increased over the years, with geographical spread throughout the state and the
successful biological control of other, once more important pests (Browning 1994).
Knowledge about its biology is meager and attempts to control it with pesticides or
with natural enemies have been unsuccessful. The present work is aimed at
understanding details of the citrus snow scale life history and relationships with its
surroundings that could contribute towards developing successful biological control.
I was interested in expanding the knowledge of its biology, measuring the duration


69
components of mortality in predation and in host feeding by aphelinid parasitoids
(Encarsia and Aphytis).
The relative importance of mortality factors observed is highly variable
through the year and on the different substrata (Figure 5.5). Endoparasitism was
predominant most of the time and only on two occasions was overshadowed by
other entities, namely incidence of fungi. In one case (August, 1993 on twigs)
ectoparasitoids were more frequent than endoparasitoids. A potential source of bias
is small samples and aggregation of susceptible hosts that may have facilitated
parasitoid oviposition on several hosts in the same sampled twig.. I have no way
to determine if the observations of fungi represent the onset of a disease or the
indication of a saprophyte, but in any case, we should not disregard the fungi
potential to inflict mortality on the scales without solid proof.
It appears that A. lingnanensis and Encarsia spp. may be important factors
in regulating populations of citrus snow scale in the canopy of citrus trees. However,
given the fact that the majority of the scale population occurs on the trunk and
scaffold limbs, and that low incidence of A. lingnanensis was observed on these
substrata (Chapter 4), it could be concluded that these parasitoids are contributing
little as biological control agents of citrus snow scale at the studied location.


71
a distance of about 1.5 m from the canopy edge. The traps were oriented North,
East, South, and West of the source tree. Each pole held two 10x10 cm2 Mylar
cards coated with Tanglefoot. The initial experiment was established on May 31,
1993 and the traps were retrieved seven days later. This experiment was repeated
five times with small variations in the same and in a new location (CREC, N-40,
block 16, Lake Alfred, Florida) in 1993-1994 (July 19, July 28, 1993, February 21,
April 6, October 14, 1994).
Crawlers were counted on the entire card, since the number of crawlers
trapped was always low. The numbers of snow scale crawlers, male scales and
natural enemies (Aphytis and Encarsia) were characterized. Wind speed and
direction was not monitored at the site, but wind data were recorded at the nearby
Citrus Research and Education Center (approximately 2.4 miles linear distance from
the initial site and approximately 2.5 miles from the second site).
Several experimental trials failed due to interference by grove vehicles. In
the first experiment (May 31-Jun 3,1993) established around a single tree, traps
were set at the four cardinal directions and two distances from the tree. Three of the
poles at the distance closest to the tree were knocked down, and thus only data
from the farthest distance were considered for analysis. In the second set (July 19-
22, 1993), two trees in the same grove were used with poles all at the same
distance from the source tree. All the poles in the East-West orientation were lost
to traffic, making the data unusable. The third trial (July 28-August 2, 1993) was a
repetition of the second and went undisturbed. The fourth trial (Feb. 21-28 1994)
was established in another grove (N-40 block 16, CREC) on two trees with traps at


83
mathematical equations for each object defined; interconnection of the defined
objects into a working model; and finally, testing the performance of the model
against the real world. Once a model is constructed, two types of tests can be
performed. Validation is testing the goodness of fit of the model against real data,
while sensitivity analysis is the analysis of the model's dynamics when the values
of the parameter estimates are changed.
The present chapter presents the development of a model for citrus snow
scale dynamics, tests the estimates obtained from studies presented in previous
chapters, establishes a framework for future expansion of studies, and identifies
gaps in knowledge on biology and ecology of the citrus snow scale.
Materials and Methods
Model Structure. A distributed delay model (Manetsch 1976) was created to
simulate the population dynamics of citrus snow scale. Distributed delay is
especially useful since it models different stages (age) and the variation in the transit
time of the individuals through a given stage (Berry & Stinner 1992). In this model,
both sexes were simulated since we are interested in development and not in
reproduction. In addition, males are an important numerical component of the
population and the most obvious to the observer. The life cycle was composed of
first nymphal stages, which develop into two sexes with sex ratio 0, a male stage,
a second female stage, and adult female (third) stage. Eggs and crawlers were not
included in the model, nor were the events occurring after sexual maturation of


APPENDIX
SIMULINK BLOCK DIAGRAM
FOR THE SNOW SCALE MODEL
Citrus snow scale model
Only intrinsic mortality included
102


111
Koteja, J. 1990a. Embryonic development; ovipary and vivipary, pp. 233-242. In D.
Rosen [ed.], Armored scale insects, their biology, natural enemies and
control. Elsevier, Amsterdam.
. 1990b. Life history, pp. 243-254. In D. Rosen [ed.], Armored scale insects,
their biology, natural enemies and control. Elsevier, Amsterdam.
Kramer, D. A., R. E. Stinner & F. P. Hain. 1991. Time Versus rate in Parameter
Estimation of Nonlinear Temperature-Dependent Development Models.
Environ. Entomol. 20: 484-488.
Kuno, G., M.A. Coln-Ferrer. 1973. Pathogenicity of two Fusarium fungi to an
armored scale insect, Selenaspidus articulatus. J. Invertebr. Pathol.
22(3):473-474.
Lactin, D. J., N. J. Holliday, D. L. Johnson & R. Craigen. 1995. Improved rate model
of temperature-dependent development by arthropods. Environ. Entomol. 24:
68-75.
Lamb, R. J., G. H. Gerber & G. F. Atkinson. 1984. Comparison of developmental
rate curves applied to egg hatching data of Entomoscelis americana Brown
(Coleptera: Chrysomelidae). Environ. Entomol. 13: 868-872.
Logan, J. A., D. J. Wolkind, S. C. Hoyt & L. K. Tanigoshi. 1976. An analitical Model
for Description of Temperature Dependent Rate Phenomena in Arthropods.
Environ. Entomol. 5: 1133-1140.
Manetsch, T. J. 1976. Time-varying distributed delays and their use in aggregative
models of large systems. IEEE Transactions on Systems, Man, and
Cybernetics SMC-6: 547-553.
Marcandier, S. & G. G. Khachatourians. 1987. Evolution of relative humidity and
temperature within a closed chamber used for entomological studies. Can.
Entomol. 119: 893-900.
Margalef, R. 1977. Ecologa. 2nd. ed. Omega, Barcelona. (In Spanish).
McClain, D. C., G. C. Rock & R. E. Stinner. 1990a. San Jose Scale (Homoptera:
Diaspididae): Simulation of seasonal phenology in North Carolina orchards.
Environ. Entomol. 19: 916-925.
1990b. Thermal requirements for development and simulation of the
seasonal phenology of Encarsia perniciosi (Hymenoptera: Aphelinidae), a
parasitoid of the San Jose scale (Homoptera: Diaspididae) in North Carolina
orchards. Environ. Entomol. 19: 1396-1402.


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
POPULATION STUDIES ON THE CITRUS SNOW SCALE
UNASPIS CITRI (COMSTOCK)
By
Julio M. Arias Revern
December, 1995
Chairman: Harold W. Browning
Major Department: Entomology and Nematology
The goal of this research was to expand knowledge of the biology and
ecology of the citrus snow scale Unaspis citri (Comstock) (Homoptera: Diaspididae),
including duration of development on a suitable substrate, causes of mortality under
field conditions, life table analyses, dispersal mechanisms, and computer simulation
of developmental dynamics.
Development of snow scale is temperature dependent. The optimal
temperature for development fell in a range of 25 to 38C, near 29C for most
stages. Development was faster on grapefruit leaves than previously reported on
lemon fruits. The effect of relative humidity was only significant for first instar males,
but the range of relative humidity was too narrow (10%) to observe more
differences.
x


44
Table 4.1. Significance of log-rank test, Wilcoxon homogeneity test, likelihood ratio
test for comparison of survival of citrus snow scale in patches within trees.
Tree
# of
Patches
Date started
Date ended
Significance
36-2
3
17/Dec/1991
27/Apr/1992
ns
36-7
2
17/Dec/1991
27/Apr/1992
ns
35-15
2
23/Jan/1992
19/May/1992
ns
35-18
2
06/Mar/1992
14/Jul/l 992
ns
34-21
5
14/Mar/1992
26/May/1992
ns
35-17
4
27/Apr/1992
18/Aug/1992
All tests
34-19
3
19/May/1992
24/Sept/1992
All tests
32-15
2
17/Jun/1992
16/Sept/1992
All tests
34-11
1
29/Jul/1992
24/Jan/1993

26-9
2
24/Aug/1992
15/Jul/1993
ns
28-7
2
24/Aug/1992
23/Jun/1993
ns
26-16
2
12/Oct/1992
14/May/1993
ns
27-14
2
26/Oct/1992
30/Apr/1993
ns
28-14
1
26/Oct/1992
8/Apr/1993

24-2
2
30/Apr/1993
15/Jul/l 993
All tests
23-13
3
24/May/1993
1/Dec/1993
Log rank P <0.025
24-12
2
24/May/1993
1/Oct/1993
Log rank P <0.031
22-9
3
17/Aug/1993
16/NOV/1993
ns
25-6
3
16/Sept/1993
14/Apr/1993
ns
21-12
1
6/Oct/1993
1/Dec/1993

22-14
1
6/Oct/1993
1/Dec/1993



Number of crawlers caught Number of crawlers caught
Direction (degrees)
74
North East South West
NW N NE E SE S SW W
Figure 6.2. Wind speed vs direction and catches of snow scale crawlers.
A-Trial 5, April 6 to 14, 1994. B Trial 6, October 14 to 21, 1994
Average wind speed Average wind speed (Km/h)


Field studies were accomplished with the use of macrophotography as a
sampling tool to study snow scale colonies on limbs and trunk, and destructive leaf
and twig samples to study snow scale populations within the canopy. Little evidence
of Aphytis lingnanensis Compere (Hymemoptera: Aphelinidae) causing mortality on
citrus snow scale on trunk and limbs was observed. Encarsia spp. were observed
often, attacking both immature males and second instar females. Predators were
evident via disrupted scale covers of some stages. Attack by Encarsia ranged
between 0 and 38%, while observable attacks by predators ranged between 0 and
49%. Analysis of data from leaf and twig samples suggests that different conditions
within trunk and canopy may affect mortality factors. Aphytis lingnanensis and fungi
were present in most of the samples taken during the 2-year study. These played
a more important role on snow scale populations found on leaves and twigs.
Mortality caused by Aphytis lingnanensis was higher on snow scale in the canopy,
which suggests that although this aphelinid has an effect on canopy populations, the
main scale population on trunk and limbs is unaffected. Crawlers were found to
disperse, taking advantage of air currents, as has been reported for other scale
insect species.
A simulation model of inmature development of citrus snow scale was derived
from experimental data obtained under constant temperature conditions. Since the
model describes only part of the life cycle, further research and model development
is needed to better describe snow scale population dynamics.
XI


10
Dispersal
Wind is presumed to be the main means of dispersion of scale insects
between plants, as has been shown for other species (Greathead 1990). It also has
been proposed that scale insect crawlers disperse on workers' clothing and
equipment (Simanton 1973) or passing animals (Greathead 1990). Xinnian &
Browning (1991) observed that citrus snow scale crawlers were positively
phototactic and that female crawlers walked farther than males. This behavior could
facilitate the expansion of colonies, decrease intraspecific competition between
females, which feed for longer periods than males, and facilitate the movement of
first instars upwards on the tree, where they might be more easily dislodged and
carried away by wind. Similarly, behavior of U. yanonensis crawlers has been
studied by Korenaga (1983) and Wang & Chen (1989).
Phenology
Citrus snow scale does not exhibit diapause in Florida, and thus it reproduces
throughout the year and has overlapping generations. At least two population peaks
have been described in Florida (Dickens 1968, Casares 1977) and in Australia
(Summerville 1935, Anon. 1954, Beattie & Gellatley 1983). Arrowhead scale and
euonymus scale show synchronized generations, with 2 or 3 being reported per
year, depending on temperatures. Arrowhead scales can overwinter in several life
stages (Nishino 1974), while euonymus scales overwinter as adult females
(Williams et al. 1977, Gill et al. 1982).


Discussion 93
8-CONCLUSIONS 98
APPENDIX: SIMULINK BLOCK DIAGRAMS
FOR THE SNOW SCALE MODEL 102
REFERENCES CITED 106
BIOGRAPHICAL SKETCH 118
VI


33
shown to be important for survival of 1st instars and for 2nd instars at low
temperatures (Table 3.5). The diaspidid armor is an effective protection against
environmental hazards (Foldi 1990). Thus 1st instars are the most likely stage to
die as a result of unfavorable environmental conditions given that they lack this
protection.
Gamma distribution parameters calculated to fit the rate of change of the
survival curves produced figures that varied widely with temperature and relative
humidity, sometimes yielding a poor fit (e.g. females at 30C, Table 3.6). In
characterizing the relationship between the shape of the mortality rate curve and
temperature (Table 3.7), the fit obtained for 2nd-order polynomial equations was
good for 1 st and 2nd stages, but poor for older stages. The number of points used
for these regressions was small (5 temperatures for 60% RH, 4 for 70% RH, and
only 3 points for males because none survived at 16C) (Table 3.2). More extensive
experimentation would be required to more accurately describe relationships
between mortality through time and the temperatures to which the insects are
exposed.
Host-plant substrate has been well documented as being responsible for
morphological differences in diaspidid scales (Stoetzel 1976, Miller and Kostarab
1979, Cooper and Oetting 1986). Other biological characteristics such as
development, mortality and fecundity are likewise affected (McClure 1990b). The
developmental rate of citrus snow scale maintained on Duncan grapefruit leaves in
this study was much higher than that observed by Casares (1974) for scales
maintained on lemon fruits. We avoid drawing conclusions regarding these effects


BIOGRAPHICAL SKETCH
Julio M. Arias Revern was born in Caracas, Venezuela, in September
29 1960, the son of a Costarrican office clerk and a Venezuelan school
teacher. In 1977 Julio moved to his father's homeland to go to college,
where he also adopted his fathers nationality. He obtained his B.Sc. degree
in Biology in 1982, and his M.Sc. in Biology in 1988 at the University of Costa
Rica. Julio married in 1981 and his son Gabriel was born in August 1984.
Julio later divorced in 1987.
After little less than a year working at an iguana research project in
Panama and Costa Rica, Julio initiated his Ph.D. program at the Department
of Entomology of the University of Florida in January 1989.
118


109
Furuhashi, K. 1974. Developmental difference of arrowhead scale (Unaspis
yanonensis Kuwana) after infest on Satsuma mandarin and Natsudaidai. Bull.
Shizuoka Prefect. Citrus Exp. Sta. 11: 68-73.
Furuhashi, K. & M. Nishino. 1983. Biological control of arrowhead scale, Unaspis
yanonensis, by parasitic wasps introduced from the People's Republic of
China. Entomophaga 28: 277-286.
Getz, W. M. & A. P. Gutierrez. 1982. A perspective on system analysis in crop
production and insect pest management. Ann. Rev. Entomol. 27: 447-466.
Gill, S. A., D. R. Miller & J. A. Davidson. 1982. Bionomics and taxonomy of the
euonymus scale Unaspis euonymi (Comstock), and detailed biological
information on the scale in Maryland (Homoptera: Diaspididae). (MP 969)
Maryland Agricultural Experiment Station, College Park.
Greathead, D. J. 1990. Crawler behaviour and dispersal, pp. 305-308. In D. Rosen
[ed.], Armored scale insects, their biology, natural enemies and control.
Elsevier, Amsterdam.
Greenspan, L. 1977. Humidity fixed points of binary saturated aqueous solutions.
J. Res. Nat. Bur. Stand. Sect. A 81 A: 89-95.
Harcourt, D. G. 1969. The development and use of life tables in the study of natural
insect populations. Ann. Rev. Entomol. 14: 175-196.
Hely, P. C. 1944. The white louse scale (Protaspis cith). A pest of coastal citrus
trees. Agrie. Gazette 55: 283-285.
Houston, K. J. 1991. Chilocorus circumdatus Gyllenhal newly established in
Australia and additional records for Coccinella undecimpunctata L.
(Coleptera: Coccinellidae). J. Aust. Entomol. Soc. 30: 341-342.
Huang, L. L, D. W. Wang, Q. B. Zhang & W. S. Zhu. 1983. A study on the biology
and control of the arrowhead scale (Unaspis yanonensis Kuwana). Acta
Phytophylacica Sin. 10: 19-24.
Hutchinson, G. E. 1978. An introduction to population ecology. Yale Univ. Press,
New Haven.
Kajita, H. 1972. The age effect on the number of ovarian and deposited eggs of
Aspidiotiphagus citrinus Craw. Jap. J. Appl. Entomol. Zool. 16: 202-204.
. 1976. Searching and oviposition behavior of Aspidiotiphagus citrinus
(Craw), a parasite of the arrowhead scale, Unaspis yanonensis (Kuwana). 1.


CHAPTER 1
INTRODUCTION
Few animal species are able to manage their environment to maximize the
collection of resources needed for their survival, other than gathering those
resources readily available. Agriculture is the process of managing ecosystems, the
manipulation of vegetal and animal species to maximize the development and
reproduction of a useful species. Efficiency in management techniques has been
achieved through natural selection within fungus-culturing ants, but for humans, it
has been attained by other means of selection: trial and error, the transfer of
traditional knowledge and research.
It is difficult to manage a system for which we have little understanding.
Solutions chosen without knowledge will very likely lead to waste of resources or to
unexpected secondary effects, as happens with development of pesticidal
resistance and pollution produced by the indiscriminate use of pesticides. Research
leading to an increase in our understanding of a natural or artificial ecosystem is an
essential component of the design of management strategies.
Another characteristic of agriculture is that successful crops are usually
planted in regions far away from their area of origin (Kloppenburg and Kleimman
1987). This separates the crop plants from organisms that may have coevolved with
the plant and use their tissues as food resources. When one or more of these
1


84
females, since the information about those stages is incomplete (see discussion in
Chapter 3). The model for males and females was structured as follows:
dW1
(0
dt
dN 2 j
(0
dt
dN 3,f
(0
dt
^2,m
(0
dt
dN o m
o, m
(0
dt
- S,N,(t) p1 N1 (0
= e S1 A/1 (/) S2/V2f(0 M2A/2 f(0
- S2N2 f(t) M3/V3,f(0
- (1-6) S,N,(t) S2N2(t) V2(t)N2m(t)
M3^3,m (0
(Females )
(Males )
(7.1)
Where /^ represents the number of individuals in the stage /; S, represents
the developmental rate from stage / to stage / +1 (1/ time), p, represents the rate of
mortality at stage / (instantaneous rate of death); and 0 represents the sex ratio as
females per male. The parameters for the distributed delay model with attrition were
calculated according to Vansickle (1977) as:
k =
DEL = xs k
(7.2)
1 _L
x DEL
AR >


38
Individual insects were located and marked on the screen with the help of a clear
plastic grid (1x1 cm2 divisions ). Approximately 20 newly settled individuals were
located and marked from the first scene of each sequence, and followed through the
set of slides until their death or to the end of the sequence.
The causes of death were classified in the following manner:
1- Competition: When individuals were dislodged by the growth of a neighbor
scale cover; included in this category were also individuals that were covered
by algae, this was observed in only one instance.
2- Parasitism: Individuals that showed emergence holes. Parasitized male
scales exhibited these holes from the time they started to secrete the white
armor until attaining their full length. Holes on males were attributed to
parasitism by Encarsia spp. (Hymenoptera: Aphelinidae). Females were
attacked by Encarsia spp. or by Aphytis lingnanensis. Encarsia parasitism
was evident early in the second instar, and emergence holes appeared
before the second molt. Emergence of Aphytis, on the other hand, occurred
on young third instar females, with the emergence holes being larger and
more rounded than those produced by Encarsia spp. All of these parasites
have been observed emerging from non-preferred host stages and sexes
(Aphytis from males and gravid females, Encarsia spp from adult females
[Browning 1994]) Those atypical occurrences were not observed during the
present study.
3- Visible predation: Assigned to those cases in which the armor was clearly
damaged but some remains persisted on the site. This damage was


7
two pupal stages and the imago. The first instar (crawler) and the adult male are the
only mobile stages. The crawlers search for a location to settle and feed, where
they will remain throughout their development. Second instar females secrete a
proteinaceous shield, while males produce a white waxy tricarinated shield. After
the second molt, females reach adulthood and produce a dark shell-like secretion
that continues to enlarge until oviposition begins. Conversely, males complete
feeding in the second instar and molt to the first pupal stage. Subsequently, they
progress through the third molt to a second pupal stage and then after the fourth
molt a winged male emerges (Dickens 1968, Koteja 1990b).
Female citrus snow scales reach reproductive maturity and deposit their eggs
beneath the secreted shield. The incubation period is very short, probably between
30-60 minutes, suggesting that most egg development occurs inside the body of the
female. Ovoviviparity is presumed to be an ancestral condition associated with
tropical environmental conditions, while retarded embryonic development and
oviparity, often associated with winter diapause, is considered a secondary
adaptation to temperate regions (Koteja 1990a). Unaspis citri embryos may be laid
enclosed in membranes instead of egg shells (Koteja 1990a), since there is no trace
of a chorion after the crawler ecloses.
The duration of development of poikilotherm organisms is dependent upon
temperature, and the relationship between development and temperature is a
fundamental feature of an insect life history (Taylor 1981). To characterize the
developmental rate, it is necessary to study the duration of development under a
range of enviromental conditions and to describe a relationship with temperature


Y Coordinate
4 8 12 16 4
X Coordinate
8 12 16
Figure 6.3 -- Continuation.


20
Materials and Methods
Snow scale development and mortality were studied on leaves of 'Duncan'
grapefruit seedlings, Citrus paradisi, under 9 constant temperature and relative
humidity conditions. The seedlings were planted in 150-cm3 plastic conical
containers and maintained using standard nursery practices. These seedlings were
infested with snow scale by attaching them to the trunk of infested trees in the field
(Casares 1974). Infested seedlings were carefully inspected to eliminate possible
contaminants (mealybugs, mites, whiteflies, other species of scales, and possible
predators of the established 1st instar), and held in cages.
Male scales and immature stages were removed after F1 crawler activity was
evident, leaving only reproductive females on the seedlings. Seven to 10 seedlings
bearing gravid females were then placed in each environmental chamber. The
gravid females were removed after 48 h, and the newly settled crawlers were
located and mapped. These cohorts were examined under a dissecting microscope
at 48 to 72 h intervals, depending on the temperature of the chamber. The
observations continued until insects reached the adult stage.
Eight computer-controlled Florida Reach-In chambers (Walker et al. 1993)
and another chamber (Lunaire, Lunaire Environmental, Williamsport, PA) were used
in the studies. The 8 Reach-In chambers were set at combinations of 4
temperatures and 2 relative humidities, with a constant photoperiod of 12:12 h for
each treatment (Table 3.1). The additional incubator (chamber 45, Table 3.1) was
set at 30C and relative humidity was maintained near 60% with the use of
saturated NaCI solutions (Carr and Harris 1949, Greenspan 1977, Marcandier and


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EEYCDNDQN_WKSYE6 INGEST_TIME 2015-03-31T17:47:47Z PACKAGE AA00029832_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


75
Table 6.1. Frequency and speed of wind on the four quadrants in Lake Alfred,
CREC weather station. Data from hourly log in a seven-days period after the
beginning date.
Trial #
Orientation
Frequency
Average
(Km/h)
Variance
North
10
4.732
12.08
1
East
31
5.368
13.51
31/May/93
South
10
4.176
5.49
West
45
5.901
7.09
North
6
5.000
2.56
3
East
39
2.613
2.84
28/Jul/93
South
48
3.865
6.13
West
46
4.998
8.77
North
19
5.528
7.94
5
East
125
7.356
14.19
6/Apr/94
South
45
4.989
8.17
West
27
6.462
8.42
North
58
4.409
5.59
6
East
56
3.478
4.44
14/Oct/94
South
11
2.464
1.90
West
38
3.097
4.81
of wind speed against wind direction (azimuth) are also plotted for each experiment.
Additionally, Table 6.1 presents a summary of the wind conditions. Reference lines
in each wind plot represent quadrants of the compass: 315-45 for North, 45-135
for East, 135-225 for South, and 225-315 for West. West wind means that air
movement occurred from that direction and toward the opposite direction, East. The
only significant differences between the number of crawlers trapped at each cardinal
direction were observed in trial 3 (July 28-August 2, 1993, Kruskall-Wallis test
H=9.4235 P<0.024). However, it is possible to observe a trend relating the speed


42
adx= alx-alx+1 Fraction of deaths in stage x from all causes (£adix)
Since it is not possible to separate sexes for the first stage of development
and since no attempt to build fecundity tables was made, the life tables include both
sexes. The stages considered in the construction of the mortality tables were:
1- First instar, from crawler settling to the onset of first molt.
2- Second instar, before production of secretion, which is from the first molt
until the cover secretion begins. Sexes are not distinguishable to this point.
3- Secretion, period during the 2nd instar during which the cover is built;
separate calculations were made for each sex. Males pass through two
more stages before adult (two molts) but these were unrecognizable because
they occurred beneath the armor. Females molt only once more.
4- Third instar (adult) females, started when females passed the second molt
and began depositing the last section of the armor.
The periods of insect development beyond this point to reproduction, and
between 1st instar eclosin and settlement were excluded from the study, thus the
mortality tables are based on only the part of the life cycle of the insect. A more
detailed description of the life cycle was presented in Chapter 2.
Results
Single decrement life tables compared within trees by the log rank test,
Wilcoxon test of homogeneity and likelihood ratio test showed significant differences
for the patches in six of 17 trees. Four trees were not tested since each had a


76
and frequency of air movement and the number of crawlers trapped in a given
direction. In trial 1(May 31 June 3, 1993) the most frequent winds came from the
East and West, with the strongest being from the East. The highest catch was
observed in the West traps. Trial 3 provides another interesting case, wherein the
frequency of wind was equivalent between East and West, but the strongest winds
were from the West. The highest catches in this experiment were on the East traps.
The wind came predominantly from the East in trial 5, and no catches were obtained
from the trap on that side: the few crawlers caught were observed in North and West
traps. The dominant wind direction in trial 6 was from North and East, but with
slower speed than in previous tests. The highest catches were obtained in East,
South, and West traps.
Infestation Scoring Maps. Figure 6.5 show the scores of citrus snow scale
infestation on the citrus plot studied. Each bar represents the relative location and
score of each tree in the plot. The original infestation was started in August 1992
but it was not evident until March 1993. Through 1993 the spread of snow scale
showed a tendency toward the soutwest. This trend became more clear during
1994, but started to become confused by April 1994, when some northern trees
started showing infestation.
Discussion
Despite the lack of statistical significance, a role of wind on the dispersion of
the citrus snow scale can be inferred from Figures 6.1 and 6.2. The more frequent
and strongest winds originated from the direction opposite to the traps that showed


58
Results
Figure 5.1 shows the age structure of the population sampled on each date
according to the substratum. First instars of undetermined sex and females form the
age structure pyramid, whereas males are shown as a bar above the pyramid. The
total number of scales in the sample is included. A Kolmogorov-Smirnov two
sample distribution test did not show significant differences between substrata, and
the Kruskall-Wallis test did not show significant differences between dates.
However, the population showed a tendency toward younger stages on leaf samples
from May 1992 to June 1993 and September 1993, but reversed on August and
October through December of 1993. Age distribution was more homogeneous on
twig samples, showing higher proportions of prereproductive females in August and
September 1992 and again in August 1993. Higher proportions of first instars
occurred during May, June and September 1993. The total number of individuals
counted on each sampling date is an indication of the abundance of snow scale at
the time of sampling. Low populations were found in November of 1992, and in
August, September, November and December of 1993. No live individuals were
found on the twig samples from November 1992. These trends should not be taken
as conclusive results, given the lack of statistical significance already mentioned.
The ratios of newly bom individuals (1st stage nymphs) per reproductive
individual (gravid females) are presented in Figure 5.2. These values were
consistently higher on leaves, except on the last sampling date in December 1993.
High ratios were recorded in June, 1992 and March, May, and September, 1993.


CHAPTER 8
CONCLUSIONS
With this work, we have contributed new knowledge to the biology of the
citrus snow scale and the population dynamics of this species in the field.
Development of snow scale is dependent on temperature. Optimal
temperature for development falls in a range of 25 to 38C with most frequent values
falling near 29C. The effect of relative humidity was found to be significant only for
first instar males but the range of relative humidity tested was too narrow (10%).
Testing of a wider humidity range would be expected to identify citrus snow scale
tolerance limits and optimal humidity conditions.
In light of these new data, it can be predicted that snow scale will fare poorly
during the hot Florida summer months when temperatures easily surpass the
calculated optimum. Development was faster on grapefruit leaves than was
previously reported for citrus snow scale on lemon fruits, but it is still not known if
snow scale development on bark is different from rates already characterized on
leaves or fruits. In any case, grapefruit leaves represent a more suitable substratum
than fruits. It appears that host plant variety and/or plant organ infested, can affect
the rate of citrus snow scale development.
Field studies by photographic techniques, life table analysis, and destructive
canopy samples showed Encarsia spp. present on snow scale infesting twigs and
98


101
manage populations of citrus snow scale within commercial citrus planting through
applied biological control and other integrated pest management strategies.


males -
Gravid females -
Virgin females -
Nymph II fern. -
Nymph I ^
1.0
males -
Gravid females -
Virgin females -
Nymph II fern. -
Nymph I -
1.0 0.5
0.0 0.5 1.0 0.5
0.0 0.5 1.0 0.5
0.0 0.5 1.0
Mar 1993
May 1993
June 1993
males
ru
;
- 152
363 397
129 236
436
Gravid females -
Virgin females
Nymph II fern. -
Nymph I -
A
A
A.
1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0
males -
Gravid females -
Virgin females -
Nymph II fern. -
Nymph I -
1.0
males -
Gravid females -
Virgin females -
Nymph II fern. -
Nymph I
1.0
0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0
Nov 1993 Dec 1993
0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0
Figure 5.1 Age structure of citrus snow scale populations on citrus leaves and
twigs. Numbers on the side of each pyramid represent the total number
of individuals on that sample.


81
activity and greater distance traveled by female crawlers. It is hypothesized that
snow scale infestations may start low on the trunk and progress upwards. This
hypothesis is supported by positive phototaxis or negative geotaxis coupled with
higher mobility of females crawlers, and could be an adaptation for reaching higher
levels on the tree. The possible advantages of this adaptation could be that it will
increase the chances for dispersion of crawlers by wind. These conditions should
enhance citrus snow scale potential for successfully colonizing neighboring trees.
The importance of phoresy on humans and other animals has been
addressed for other species of scale insects. Stephens & Aylor (1978) collected and
examined the feathers, beak and feet of birds for crawlers of Matsucoccus resinosae
and failed to find them. On the other hand, Washburn & Frankie (1981) exposed
humans and dogs to iceplants infested with Pulvinariella mesembryanthemi and
found crawlers and egg sacs on clothing and fur. It appears unlikely that citrus snow
scale is dispersed to any extent by humans given the location of citrus snow scale
on the interior wood and the thorny nature of the citrus canopy. Both factors should
limit contact between human clothing and the snow scale colonies on bark.


115
Simanton, W. A. 1973. Changes in Florida citrus pest populations during 16 years,
pp. 491-496. In 0. Carpena [ed.], Proceedings I International Congress of
Citriculture, Murcia, Spain.
Smith, D. & D. F. Papacek. 1985. Integrated pest management in Queensland
citrus. Queensland Agr. J. 111: 249-259.
Sokal, R. R. & F. J. Rohlf. 1981. Biometry, 2nd. ed. Freeman, New York.
Southwood, T. R. E. 1978. Ecological methods, with particular references to the
study of insect populations, 2nd. ed. Chapman & Hall, New York.
Steel, R. G. D. & J. H. Torrie. 1980. Principles and procedures of statistics: a
biometrial approach, 2nd. ed. McGraw-Hill, New York.
Stephens, G. R. & D. E. Aylor. 1978. Aerial dispersal of red pine scale,
Matsucoccus resinosae (Homoptera: Margarodidae). Environ. Entomol. 7:
556-563.
Stoetzel, M. B. 1976. Scale-cover formation in the diaspididae (Homoptera:
Coccoidea). Proc. Entomol. Soc. Wash. 78: 323-332.
Summerville, W. A. T. 1935. White louse of citrus. Advisory leaflet. Department of
Agriculture and Stock, Queensland.
Summy, K. R., M. R. Davis, W. G. Hart & F. E. Gilstrap. 1984. Use of close-up
photography in nondestructive monitoring of citrus blackfly cohorts. J. Rio
Grande Val. Hortic. Soc. 37: 55-60.
Takagi, K. 1981. Evaluation of parasitoids of latent pests with sticky suction trap, pp.
627-630. In K. Matsumoto [ed.], Proceedings of the International Society of
Citriculture. Organizing committee, Tokyo, Japan.
Takezawa, H. & M. Uchida. 1969. Relationship between the ovarial development
and the appearance of nymphs in arrowhead scale (Unaspis yanonensis)
with special reference to its application to forecasting. Jap. J. Appl. Entomol.
Zool. 13: 31-39.
Tanaka, M. 1981. Biological control of arrowhead scale, Unaspis yanonensis
(Kuwana), in the implementation of IPM programs of citrus orchards in
Japan, pp. 636-640. In K. Matsumoto [ed.], Proceedings of the International
Society of Citriculture. Organizing committee, Tokyo, Japan.
Taylor, F. 1981. Ecology and evolution of physiological time in insects. Am. Nat
117: 1-23.


28
Table 3.4. Effect of constant temperature across constant relative humidity on
mortality of citrus snow scale, x2 values are for log rank test (above) and
Wilcoxon tests (below) (SAS Institute Inc., 1989).
Nymph I
Nymph II
Females
Males
RH
(%)
n
X2
n
x2
n x2
n
X2
60
1530
159.63***
889
173.31***
es ivr
7.94 ns
245
83.86***
241.18***
127.43***
45.39***
70
926
4.47 ns
15.72***
681
208.29***
141.44***
143 2 "S
4.52 ns
182
26.56***
13.98***
***= P < 0.001, **= P < 0.01, *:
= P < 0.05, ns = not significant.
Table 3.5. Effect of relative humidity across constant temperatures on mortality of
citrus snow scale.
X2 values are for log rank test (above) and Wilcoxon tests
(below) (SAS Institute Inc., 1989).
Nymph I
Nymph II
Females
Males
T
(C)
n
X2
n
X2
n x2
n
X2
16
554
1.35 ns
1.93 ns
420
12.87***
12.11***
yi- 0.00 ns
0.00 ns
99
21
435
I.30 ns
II.42 ***
296
2.44 ns
0.09 ns
cc 0.91 ns
66 0.91 ns
161
0.02 ns
0.47 ns
24
548
86.95***
105.68***
380
0.02 ns
5.50 ns
in9 0.66 ns
0.66 ns
79
0.75*
0.47*
28
467
52.01***
74.93***
285
0.07 ns
0.23 ns
7fi 0.20 ns
0.18 ns
98
5.23*
1.99 ns
***= P < 0.001, **= P < 0.01, *= P < 0.05, ns = not significant.


47
Table 4.3. Outcome observed for citrus snow scale individuals under field
conditions. Summary, data pooled for all patches and dates.
Outcome
1st
instar
2nd
instar
Secreting
3rd
instar
female
Total
Male
Female
Completed
N/A
N/A
118
N/A
10
128
Detectable predation
0
1
119
6
16
142
Parasitism
0
0
95
8
5
108
Competition
1
7
0
0
0
8
Fungi
0
3
1
1
1
6
Unknown causes
103
244
158
35
24
568
Lost
51
78
37
6
5
177
Failed to develop
52
166
125
29
19
391
Total
104
255
495
50
56
960
adult (third instar) females (Jul. 1993), and finally, in two cases, first instars suffered
the highest levels of mortality (Aug. and Oct. 1993).
Discussion
Survival curves of citrus snow scale populations showed significant
differences between patches on six trees. These patches shared a feature in being
initiated between May and mid-June of each year, when temperatures were
increasing. The temperature increase is responsible for a shortened life cycle and
may increase activity of natural enemies that could have affected variability within
these six trees.


Nymph I
103
Female nymph II
AR


# of 1rst nymphs per gravid female
60
Figure 5.2. Ratio of snow scale first instars per gravid females on samples
of citrus twigs and leaves.


116
Tern, A. L, M. L. Collado de Manes, S. Glenross, R. Alvarez & H. Lazaro. 1985.
Primary and secundary Hymenopteran parasitoids of Diaspidids in Tucuman
(Argentina) citrus, except Aonidiella aurantii. Rev. Invest. CIRPON 3: 25-33.
(In Spanish).
Terry, L. I. & G. J. Edwards. 1989. Efficacy of densitometric and multispectral
techniques for monitoring infestations of citrus snow scale on citrus bark.
Photogram. Eng. Remote Sensing 55: 1471-1475.
UK, CABI. Institute of Entomology. 1962. Distribution maps of pests. Pest: Unaspis
citri (Comst.). Commonwealth Agriculture Bureaux. Series A (Agricultural).
Map 149.
. 1970. Distribution maps of pests. Pest: Unaspis eunoymi (Comst.).
Commonwealth Agriculture Bureaux. Series A (Agricultural). Map 269.
. 1988. Distribution maps of pests. Pest: Unaspis yanonensis (Kuwana).
Commonwealth Agriculture Bureaux. Series A (Agricultural). Map 503.
van Driesche, R. G. 1983. Meaning of "percent parasitism" in studies of insect
parasitoids. Environ. Entomol. 12: 1611-1622.
Vansickle, J. 1977. Attrition in distributed delay models. IEEE Trans. Syst. Man
Cybern. SMC-7: 635-638.
Varley, G. C. & G. R. Gradwell. 1971. The use of models and life tables in
assessing the role of natural enemies, pp. 93-112. In C. B. Huffaker [ed.],
Biological control. Plenum/Rosetta.
Varley, G. C., G. R. Gradwell & M. P. Hassell. 1973. Insect population ecology: an
analitical approach. Blackwell Sci. Publ., Oxford.
Vinis, G. 1977. The spindle-berry scale, Unaspis euonymi Comst. in Hungary.
Novenyvedelem. 13: 5-10.
Wagner, T. L. & J. L. Willers. 1992. The role and relationship of the database with
insect population models, pp. 9-22. In J. L. Goodenough and J. M. McKinion
[eds.], Basics of insect modeling. American Society of Agricultural Engineers,
St. Joseph.
Wagner, T. L, Hsin-I. Wu, P. J. H. Sharpe, R. M. Schoolfield & R. N. Coulson. 1984.
Modeling insect development rates: a literature review and application of
biophysical model. Ann. Entomol. Soc. Am. 77: 208-225.


51
cycle, females are not available for parasitism after they reach reproductive status
(Browning 1994). On the other hand, reproductive females will remain targets for
predators longer than males. Undoubtedly, predation is underestimated, and likely
an important fraction of the observed unknown mortality is predation. In the present
research, attempts to trap predators failed but chrysopid larvae (Neuroptera) were
observed, collected, and reared from citrus snow scale colonies.
Parasitism occurred through most of the study period, but was lacking from
our patches in June and July, 1992 and April to August, 1993. Aphytis lingnanensis
has been present in Florida for some time. A race of this species collected in Hong
Kong and named HK-1 was released in Florida in the early 1970's (Browning 1994).
This species was found in leaves and twig samples in the grove where this study
was performed, but the low frequency in which parasitism by Aphytis was observed
in the photographic analysis suggests that Aphytis is not a major factor in the
regulation of citrus snow scale populations on trunks and therefore it is not as
successful a biological control agent as has been reported (Fisher 1985). The
contributions of parasitism and detectable predation appeared not to be related as
they seemed to fluctuate independently between patches (Table 4.4). High
incidence of one factor did not accompany low incidence of the other. Apparently,
they are not complementary in this sense. In fact, predators may be a competing
factor with the survival of parasitoids that attack early stages of citrus snow scale.
Contrary to what was expected, diseases were observed in a very small
proportion of scales examined, despite the fact that reproductive bodies of the
fungus were observed on snow scale colonies on the same trees where the patches


Discussion
30
4 STUDY OF MORTALITY OF THE CITRUS SNOW SCALE
UNDER FIELD CONDITIONS, BASED ON THE ANALYSIS
OF PARTIAL LIFE TABLES 35
Introduction 35
Materials and Methods 36
Data Recording with Photographs 36
Photographic Analysis 37
Statistical Analysis 41
Results 42
Discussion 47
5 COMPARATIVE STRUCTURE OF CITRUS SNOW SCALE
POPULATIONS ON THE CANOPY OF CITRUS TREES
AND THROUGH TIME 54
Introduction 54
Materials and Methods 55
Results 58
Discussion 65
6 THE ROLE OF WIND IN CRAWLER DISPERSION 70
Introduction 70
Methods 70
Results 72
Mylar traps 72
Infestation Scoring Maps 76
Discussion 76
7 MODELING CITRUS SNOW SCALE POPULATIONS 82
Introduction 82
Materials and Methods 83
Model Structure 83
Development 85
Mortality 85
Reproduction and Migration 86
Environmental Variables 86
Model Implementation 87
Model Calibration 87
Model Validation 87
Results 88
v


89
Table 7.2 Average duration and percentage of survival from each stage at each
environmental condition studied (Chapter 3).
Stage
T
RH
Observed days in stage
/o
survival
Average Variance
16
60
5.39
31.52
0.596
70
5.26
23.25
0.721
21
60
3.75
17.08
0.485
70
3.69
21.81
0.750
1st stage
24
60
1.98
6.69
0.697
70
3.80
17.56
0.535
28
60
2.73
13.08
0.615
70
3.19
12.74
0.548
30
60
2.76
7.82
0.512
16
60
32.54
241.10
0.531
70
34.26
251.44
0.680
21
60
15.78
46.50
0.410
?nc* 70
19.23
102.05
0.693
females
24
60
13.29
37.22
0.618
70
12.31
41.79
0.373
28
60
11.64
24.74
0.682
70
11.65
37.22
0.333
30
60
13.98
39.18
0.547
16
60
25.67
311.15
1.000
70
33.97
358.29
1.000
21
60
33.61
229.41
0.857
70
17.18
161.22
0.976
Males*
24
60
15.61
66.70
1.000
70
17.29
81.34
1.000
28
60
17.91
121.49
0.973
70
17.29
65.34
0.938
30
60
3.72
81.12
0.850
16
60
64.09
30.13
1.000
70
56.37
229.27
0.947
21
60
44.10
65.64
0.857
Adult
70
60.27
97.38
0.970
females
24
60
44.23
121.00
1.000
70
29.14
21.92
1.000
28
60
34.74
91.58
0.976
70
50.62
331.22
1.000
*
30
60
20.81
331.93
0.727
* Includes secreting male 2nd stages, "prepupa" and "pupa".


6
an indication of oviparity. The perivulvar glandular system may be associated with
the presence of the egg shell and thus its reduction in some species is thought to
parallel progressing viviparity (Koteja 1990a).
Economic Importance
Unaspis citri is considered an important pest of citrus in the Americas and
Australia, based on the stress that this species inflicts on the general health of the
tree and on the cost of its control. Bark and scaffold limbs are the primary sites of
attack on citrus trees. However, in heavy infestations, it also attacks leaves and to
a lesser extent fruits. Bark splitting, twig dieback and tree death are also associated
with severe infestations (Beattie & Gellatley 1983, Smith & Papacek 1985, Browning
1994).
Unaspis yanonensis is an important pest of citrus in Japan, where the
increase of populations in the absence of effective natural enemies causes damage
to citrus fruits (Ohkubo 1981). Unaspis euonymi is very important in landscape
horticulture in North America and Europe, producing chlorosis, reduced
photosynthesis, leaf abscission and stunting in Euonymous plants (Vinis 1977, Gill
et al. 1982, Brewer & Oliver 1987, Cockfield & Potter 1987).
Biology
The biology of citrus snow scale, as studied by Dickens (1968), typifies the
life cycle of many other diaspidid scales. Females have two molts, or two nymphal
stages and the imago or adult stage. Males have four molts leading to two nymphs,


40
5- Unknown causes: Mortality causes that could not be separated or were
dubious or not clearly attributed to a given factor were grouped under this
category. Several outcomes are described here:
a) Collapsed: A few male armors were observed to shrink or become
deformed. This collapse could be attributed to the action of predators
other than coccinellids, e.g., thrips, neuropteran, mites, that can feed
on the scale insect without disturbing the armor.
b) Failed development: Assigned when an individual scale stopped its
development at an immature stage, with no outward signs of
disturbance. This failure of development could be attributed to
physiological or pathological causes, mainly for first instars, in which
a change in coloration occurred as they dried out; to predation by
organisms that act without disturbing the armor, such as predatory
mites, thrips or lacewing larvae; or host-feeding by aphelinid
parasitoids.
c) Lost: Assigned when an individual scale disappeared without
leaving a clue about its fate. Occurred frequently with first instars or
with developing males. Could be the action of predators, or
mechanical factors such as rain or wind.
7- Indefinite outcome, covered by others: Used in the cases when the patch
was overcrowded and other scales grew over the observed individuals.
These were excluded from the analysis.


This dissertation was submitted to the Graduate Faculty of the College
of Agriculture and to the Graduate School and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy.
December, 1995
Os ciitture
Dean, College of Agri
Dean, Graduate School


85
Where x and s2 are the average and variance of the time spent by the
individuals in a stage, k is the number of delay stages (subsystems) that constitute
each modeled stage, DEL is the mean delay that an individual spends in the
modeled stage, e is the proportion that survive at the end of the stage, and AR is the
instantaneous attrition rate, equivalent to the mortality rate p, previously mentioned.
Development. Development was modeled after Lactin et al.'s (1995)
modification of Logan et al.'s (1976) poikilotherm model, using the data presented
in chapter 3. Lactin's modification to the model of development is:
rate
epT
A
(7.3)
where T is temperature, p is the rate of increase to optimum temperature, Tm is the
maximum lethal temperature, AT is the difference between the lethal temperature
and the optimum temperature and A is a scaling factor that allows for incorporating
developmental thresholds (Lactin et al. 1995).
Mortality. Innate mortality was modeled using survival estimates for snow
scale cultures at constant temperature (Chapter 3) and was expressed as the
percentage of survival at the end of the stage (% of censored observations)
(Procedure LIFETEST, SAS Institute Inc. 1989).
Innate mortality was modeled as dependent upon temperature; it could not
be modeled as dependent upon RH because there were not sufficient data or
evidence of this dependency (see Chapter 3). The fraction dying is a proportion of
dead individuals in the previous stage (p,/VM) and is subtracted from the rate of


CHAPTER 3
DEVELOPMENT AND MORTALITY
UNDER CONSTANT TEMPERATURE
AND RELATIVE HUMIDITY
Introduction
Citrus snow scale, Unaspis citri (Comstock), is regarded as a pest of citrus
in Florida and in many other tropical and subtropical citrus production areas
(Dickens 1968, Smith and Papacek 1985, Rose 1990). The relative importance of
the snow scale in Florida increased over the period 1963-1971 (Simanton 1973)
because of improved management of other important pests and the spread of
infested nursery stock throughout the state (Browning 1994).
The life cycle of the snow scale described by Dickens (1968) is typical of
most armored scales. The sexes are dimorphic; females proceed through 2
nymphal stages before reaching maturity, whereas males have 2 nymphal stages
similar to females, and then 2 non-feeding stages, called the prepupa and pupa.
Development under constant temperatures was studied on citrus fruits by Dickens
(1968) and Casares (1974). However snow scale is seldom found on fruit in the
field, living mostly on trunks, branches, twigs and, to a lesser extent, foliage.
Casares (1977) conducted ecological studies on the phenology of the scale and
concluded that rainfall, variety and tree age are important factors in determining
snow scale population levels.
18


21
Table 3.1. Temperature and relative humidity means sd conditions used in the
study of citrus snow scale development.
Chamber
no.
Temp. C
RH %
17
15.95 1.00
60.11 4.11
18
20.95 1.06
59.86 3.06
19
16.10 0.98
71.84 5.72
20
20.96 0.94
69.85 3.15
41
24.00 0.10
60.07 1.29
42
27.99 0.18
60.00 0.46
43
24.06 0.64
69.70 3.55
44
27.98 0.39
69.59 1.83
45
29.77 1.14
56.28 7.55
Conditions in chambers 17-44 (Florida Reach-In) were logged in by the controlling
computer every 10 min, conditions in chamber 45 (Lunaire) monitored with
hygrothermograph and the statistics calculated from 2-h intervals.
Khachatourians 1987). Conditions inside the Florida Reach-In chambers were
monitored at 10-min intervals with built-in sensors, downloaded to a file and plotted
weekly. Conditions in the Lunaire chamber were monitored using a hygro
thermograph (model H-302, WeatherMeasure, Sacramento CA).
Developmental time and survival for each stage and sex, from settled crawler
to adult, were recorded at each observation date. Molts were easily recognized
under stereomicroscopy. Females showed darkening and separation of the exuviae;
males expelled the exuviae from under the armor at the distal end. Female
development from the last molt to the reproductive stage and egg development were
not assessed because these periods could not be observed without lifting the armor
and disrupting the insect.


without their friendship I could not have succeded. I am perpetually indebted to Ms.
Lois Wood, for her love and friendship and her invaluable editorial help.
Finally, I need to thank my father Julio, my mother Isabel and my son Gabriel
for their love, support and patience. They never gave up on asking "When are you
going to finish?", but they never doubted that I would finish it.
in


Y Coordinate
77
August 1992
March 1993
A
N
4 8 12 16
4
8 12
X Coordinate
Artificial infestations
Score 3
No infestation
Score 4
Score 1
@ Score 2

Score 5
Figure 6.3 Maps of trees infested with citrus snow scale in a citrus plot (CREC,
N-40 block 16). Each circle represents the position of a tree in the plot.
Circle's shade indicates the intensity of snow scale infestation measured
as scores from 0 to 5.


24
M (t)
pa t 1
e pt
r (a)
(3.2)
where t is age and a and 3 are parameters that govern the shape of the distribution.
The parameters obtained were related to temperature using regression of
polynomial equations (PROC REG, SAS Institute 1989).
Results
Table 3.2 presents average duration of development in days, with standard
deviation and sample size for each sex and stage, at each combination of
temperature and humidity.
The effect of temperature on developmental time was statistically significant,
except for the pupal stage at 60% RH (Table 3.2). The length of each stadium
decreased as temperature increased until an optimal temperature for development
was reached. Duration of development then increased at temperatures higher than
optimal. The effect of relative humidity was significant only for 1 st nymphal males,
where 70% RH expedited development by 1-2 d. Interaction effects were significant
for 1st nymphs of both sexes and 2nd nymphal males (Table 3.2).
No significant differences were found between sexes in development of 1st
stages under either relative humidity condition (F = 0.45, df = 1, 341; P = 0.51 for
60% RH, F = 0.80, df = 1, 247; P = 0.38 for 70% RH, covariance analysis using
temperature as covariates, Steel and Torrie 1980). Gender differences were


16
Chemical control. Since snow scale has overlapping generations in Florida
citrus, it is extremely difficult to effectively target pesticidal applications to the most
susceptible stages, as is done for the arrowhead scale (Adachi & Korenaga 1992).
The recommendations for materials and equipment for chemical control are given
by Knapp (1995). Recommendations include applying pesticides with hand-held
equipment in localized areas where the scale is a problem and trying to achieve the
maximum coverage possible.
Biological control. Biological control of citrus snow scale in Florida began in
the 1970's. The first recorded introductions of E. citrina and A. lingnanensis
occurred in 1969. A. lingnanensis "HK-1" was introduced into Florida in 1972, and
was reported as being established and effectively controlling U. citri (Riehl et al.
1980). During the period 1974-1985, approximately 6.5 million parasitoids were
released, and savings early in the program were estimated at 8-10 million dollars in
pesticide applications (Mead 1976). The program of mass release of HK-1 was
discontinued because of the effectiveness of the parasitoid was not demonstrated,
and new efforts to identify alternative solutions have begun (Browning 1994).
Several other strains of A. lingnanensis and two other species, A. khunti, and A.
yanonensis, have been introduced into Florida, but establishment has not been
confirmed (Browning 1994).
Unaspis yanonensis became a very serious problem in Japan after it became
established there without natural enemies (Ohkubo 1981, Takagi 1981). Aphytis
yanonensis and Physcus fulvus (Hymenoptera: Aphelinidae) were introduced from


CHAPTER 5
COMPARATIVE STRUCTURE OF CITRUS SNOW SCALE
POPULATIONS ON THE CANOPY OF CITRUS TREES
AND THROUGH TIME.
Introduction
A good knowledge of the biology and ecology of a pest is necessary to design
optimal pest control programs. Periodic sampling of the population produces useful
information concerning age structure, sex ratios, and mortality factors. Difficulties
arise when measurement of absolute density of an organism is needed, since it
requires an estimation of the area sampled. Alternatively, population intensity
(number of individuals per unit of habitat, e.g. leaf) and relative estimates are
available (Southwood 1978). The percentage of a population subjected to
parasitism is a traditional measurement of the effect of these natural enemies, but
this measurement has been criticized based on difficulties in obtaining accurate
information on the real impact of those parasitoids (van Driesche 1983).
Species that colonize different organs of a plant have been shown to exhibit
different traits, such as differences in developmental rate, sex ratio, and even
morphology (Chapter 2). In the same way, the association with different host plant
organs exposes the insects to different environmental influences. Variable host
nutrient composition may occur between leaves, fruits and trunks. Natural enemies
54


96
Among faults in the conceptualization of the model could be the exclusion of
relevant factors, perhaps relative humidity, conditions of the host plant or other
unknown factors. Programming errors are unlikely, since the use of a simulation
graphical tool (MATLAB's Simulink) greatly simplifies this task. The most likely
problem remaining then, is inaccuracy in the estimation of parameters. Table 7.5
presents confidence intervals for the parameters used in simulating development
and table 7.6 shows the same information for mortality. Confidence intervals for the
rates of development of the first instar (both sexes) are very wide, which may
account for the problems with the first stage curve. The parameter estimation used
thus may not accurately represent the real values. At the same time, regression of
the instantaneous mortality rate (AR) against temperature showed low coefficients
of determination (R2, Table 7.4), which implies poor approximation of the observed
mortality. Further development of methods to simulate mortality may be necessary.
The simulation of this factor might be improved using a nonlinear or a polynomial
model. Furthermore, other factors besides temperature (eg. interaction of
temperature with humidity) likely affect mortality.
Further improvement of this model requires filling gaps in knowledge of the
snow scale system. It is important to determine reproductives rates, the longevity
of the adult reproductive females, the rate of dispersion (crawling) of new individuals
on a substrate, and the effects of substrate (plant organ and plant variety) on
development and survival. These data will give us a more precise idea of the
potential for pest outbreak of this species. On the other hand, other important
factors that must be incorporated into the model are the effects of parasitoids and


19
Diverse approaches have been taken to simulate development of
poikilothermic organisms, from simple assumptions of linear dependency between
temperature and development to more complicated nonlinear equation models
(Curry and Feldman 1987). Developmental processes depend on temperature-
dependent chemical reactions and are bounded by upper and lower thresholds.
Below lower temperature thresholds, developmental processes are slow or halted;
likewise, above upper temperature thresholds development is retarded, proteins are
denatured and subsequent death of the organism occurs (Sharpe et al. 1977).
Lower developmental thresholds are easily calculated when the developmental rate
is assumed to be linear. However, when the rates are assumed to be nonlinear, the
developmental functions that have been proposed approach zero asymptotically,
making it impossible to deduce a lower threshold from the mathematical relationship
estimated (Logan et al. 1976, Wagner et al. 1984). Development at low temperature
may proceed so slowly that it may not be perceptible to the observer, and most
models cannot account for this phenomenon. Lactin et al. (1995) proposed a
modification of Logan et al. (1976) that estimates lower developmental thresholds.
This modification adds a parameter that changes the scale of the curve, allowing it
to intercept with the x-axis (zero development).
In an effort to achieve effective control of citrus snow scale, researchers are
currently striving to better understand its biology and ecology. The current study
relates development and mortality to temperature and relative humidity under
laboratory conditions, and adds new information by using a substrate similar to that
inhabited by the scale under natural conditions.


CHAPTER 7
MODELING CITRUS SNOW SCALE POPULATIONS
Introduction
Models are hypotheses of how a system functions (McKinion 1992). A model
can be designed to graphically represent relationships or to mathematically
represent a system. Mathematical models can vary in complexity, creating a
compromise between simplicity and accuracy. Simulation models tend to be
complicated, and produce outputs that can be used to forecast the state of a system
in the future, be it weather conditions, the populations of an organism, or the
dynamics of a market. Models has been used extensively in forecasting and
management of biological populations (Ruesink 1975, 1976, Getz & Gutierrez 1982,
Curry & Feldman 1987).
Overall, models help to understand the processes and provide means to
organize information about complex systems (Wagner & Willers 1992). However,
in order to create an accurate representation, thorough knowledge of the system is
required. Models also can be used to test the quality of the information available
regarding a system. The model then becomes a tool to check the integrity of our
estimation of its parameters.
Models are built through a series of steps. Ruesink (1976) describes four
such steps: definition of the system and the objects within it; creation of
82


56
from the interior of the canopy. These samples were processed in the laboratory
and up to 100 live individual snow scales were counted from leaves and 100 more
from twigs. The information recorded included sex, stage, parasitism, presence of
fungi, and presence of mites. Two types of parasitoids were observed, the
endoparasitoids, Encarsia species (Hymenoptera: Aphelinidae) and the
ectoparasitoid Aphytis lingnanensis (Hym.: Aphelinidae). Encarsia spp. are
endoparasitoids of both sexes, ovipositing into the second stage. Parasitism of male
scales is evident because the dark parasitoid pupa is visible when male scale covers
are lifted. In females, parasitism by Encarsia becomes visible through changes of
body coloration of second instar female scales while the parasitoid is in its larval
stages. Late parasitoid larvae and pupae are easily visible through the scale cover
and body. Two species are likely to occur in the study area, E. citrina (Craw.) and
E. longsburyi (Berlese & Paoli) (Browning 1994). On the other hand, Aphytis
lingnanensis Compere is an external parasitoid that attacks young third stage
females (pre-reproductive adults). Parasitoid larvae and pupae are very evident on
their hosts, but eggs are more difficult to detect.
The fungi observed developing on citrus snow scale species were Nectria
species (anamorph Fusarium), exhibiting white mycelia and orange fructiferous
bodies. An unidentified species of mite also was observed feeding on the bodies
of adult females. These mites were observed inserting their mouthparts into the
scale integument. Citrus snow scale deaths by predators were not quantified, since
there was no way to determine time of death or how long a dead individual would


53
scale to rare occurrence in the site studied. The white covers of the snow scale
males are very evident against the dark background of the citrus bark, and the
covers abandoned by males that already emerged remain on site for an
undetermined but presumably long period. Despite the observed high mortalities,
the accumulation of male scale covers may give the impression that densities of live
citrus snow scales are greater than what they really are. This suggests that the
snow scale problem is in part a problem of perception by the grower. It also
identifies the need for careful observation to determine the status of live scales in
the field when management options are considered.


23
rates. The mean durations were weighted by the sample size. Temperature was
not transformed to the base temperature because this substraction is simply a
change in scale and should not produce any effect on values of the parameters
when fitting the curve. The curves were fit using the Marquart method (PROC NUN,
SAS Institute 1989).
Mortality was analyzed through life table methods, using insect age as the
time variable. Survival curves for the 1st and 2nd nymphs were calculated by
pooling data for both sexes. For male prepupal and pupal stages and virgin and
gravid females, stages were pooled and the survival curves calculated for each sex.
This was done for practical purposes because sexes are difficult to separate for 1st
and 2nd instars and stages are difficult to separate for later instars. Differences
between the curves were tested using the Wilcoxon test, which is more sensitive to
differences at early survival times and the Savage (log-rank) test, which is more
sensitive to differences at later survival times (PROC LIFETEST, SAS Institute
1989).
The rate at which the survival curve decays with respect to time corresponds
to the 1st derivative of the function (the slope of the tangent to the curve), or the
probability density function of the survival curve (SAS Institute 1989). This function
was calculated for each stage and sex and fitted, using the DUD method (PROC
NUN, SAS Institute 1989), to a gamma distribution model (Curry and Feldman
1987):


99
leaves as well as trunk and limbs. Attack by Encarsia ranged between zero and
38% on citrus trunks at different sites over time, while observable attack by
predators ranged between zero and 49%. An important percentage of the mortality
could not be identified (Chapter 4), which must be composed of predation, diseases,
parasitoids host feeding and abiotic causes. Since the effects of predation and
diseases must be greatly underestimated by not counting the proportion included
in this percentage, the observed incidences should be taken as only an indication
of trends. Definitely, the effect of predation must be greater than what was
observed. In the case of diseases, is more difficult to assign a value as population
regulator since the question about pathogenicity is not yet resolved. Nonetheless,
the species of natural enemies observed are considerable mortality factors for snow
scale, and therefore their conservation should be considered when pest
management actions need to be taken.
Studies in the field found little evidence of Aphytis lingnanensis contributing
to mortality on populations of snow scale infesting trunks and main limbs. Analysis
of samples from leaves and twigs indicated that the conditions on bark and leaves
may be different with regard to mortality factors. Aphytis lingnanensis and fungi
were present in nearly all of the samples taken during the two year study. These
appeared to play a more important role on marginal portions of snow scale
populations that inhabit the canopy. Mortality invoked by Aphytis lingnanensis is
higher on the canopy, which suggests that although this aphelinid has an effect on
snow scale populations on the canopy, the main component of the population that
resides on trunk and main limb bark escapes parasitism by Aphytis.


13
citrina and Encarsia longsburyi prefer to attack young second instars of both sexes
of citrus snow scale.
Aphytis lingnanensis Compere (Hymenoptera: Aphelinidae) preferentially
attacks late second instar or young adult (prereproductive) females of citrus snow
scale. It is also a polyphagous parasitoid originally from Asia; and this species
develops as an arrhenotokous ectoparasitoid. Several related forms or strains have
been identified, morphologically indistinguishable from A. lingnanensis but with
varying degrees of reproductive isolation (DeBach & Rao 1969). These strains have
been widely used in attempts at biological control of several pests, including the
introduction of the Hong Kong "HK-1" strain of A. lingnanensis into Florida for the
control of snow scale and the introduction of the Hong Kong "HK-J" strain into Japan
for control of the arrowhead scale (Rosen & DeBach 1979, Tanaka 1981).
Aphytis gordoni (Hymenoptera: Aphelinidae) is another ectoparasitoid, half
the size of A. lingnanensis, discovered during exploration for snow scale parasitoids
in Hong Kong. All of the material was obtained from U. citri hosts (Rosen & DeBach
1979), which could be an indication of host specificity and make this species worthy
of more research as a potential biological control agent for snow scale.
Chilocorus circumdatus (Coleptera: Coccinellidae) was recently reported
feeding on citrus snow scale in Australia where it was originally introduced from
China for the control of Aonidiella aurantii (Houston 1991). This coccinellid is a
diaspidid scale feeder and was introduced during biological control projects for other
species of armored scales in the U.S. (Rosen & DeBach 1977).


37
where no insecticides were applied during the study. Between two and four
observation areas (4x4 cm2) were selected on the bark of each tree. To obtain a
cohort of newly settled crawlers of known age, I dislodged all individuals in the area
(hereafter referred to as patches) using a hard brush (Casares 1977). A map pin
was positioned in the upper left corner of the patch as a reference marker.
Photographs of patches were taken weekly over a period that varied from eight to
20 weeks. New patches were established periodically (approximately every two
weeks) on different trees. Eighty-two patches were studied on 36 trees during the
observation period, but only 47 patches on 21 trees were colonized and therefore
were used for the life table analysis.
A 35-mm camera (Canon AE-1, Canon Inc. Tokyo, Japan) equipped with a
macro lens (Canon FD 50 mm f/3.5 Macro) mounted on bellows allowed image
enlargements of 2X. The lighting consisted of two flashes, one mounted on a
bracket beside the camera level with the lens and the other hand-held on the
opposite side of the lens and equipped with a photosensitive trigger. The camera
and flash were mounted on a sturdy tripod. High resolution color transparency film
(Kodak Ektachrome 64 and Fujichrome 50) was exposed at 1/60 sec. and f/22
aperture and was processed through commercial laboratories.
Photographic Analysis. The sequences of insect development were studied
while viewed on a slide projector with a built-in 9x9-inch screen (Kodak Ektagraphic
AudioViewer/Projector model 260, Eastman Kodak Co., Rochester, NY 14650). The
total area covered by the field of the camera was 18x12 mm2 of trunk surface. The
area chosen for analysis was equivalent to 14x9 mm2 in the center of the field.


LIST OF TABLES
Table page
3.1. Temperature and relative humidity means sd
conditions used in the study of citrus snow scale
development 21
3.2. Duration of development for citrus snow scale at
different constant temperatures and relative
humidities 25
3.3. Parameters calculated to fit citrus snow scale
developmental data to the modified Logan
developmental model (Lactin et al. 1995) 27
3.4. Effect of constant temperature across constant relative
humidity on mortality of citrus snow scale 28
3.5. Effect of relative humidity across constant temperatures
on mortality of citrus snow scale 28
3.6. Parameters calculated to fit citrus snow scale mortality
data to a gamma distribution function 29
3.7. Polynomial regressions between temperature and the
parameters of a gamma distribution (a, (3) fitted to the
probability density function of the survival curves 30
4.1. Significance of log-rank and Wilcoxon homogeneity
test, likelihood ratio test for comparison of survival of
citrus snow scale in patches within trees 44
4.2. Significance of log-rank and Wilcoxon homogeneity
test, likelihood ratio test for comparison of survival of
citrus snow scale between trees by dates 45
4.3. Outcome observed for citrus snow scale individuals
under field conditions 47
4.4. Total percentage of mortality (adx) and percent mortality
by unknown causes against parasitism and detectable
predation in multiple decrement life tables, calculated
by tree 48
4.5. Total percent mortality (all causes) adxpresented by
stage 49
6.1. Frequency and speed of wind on the four quadrants in
Lake Alfred, CREC weather station 75
vii


Table 3.3. Parameters calculated to fit citrus snow scale developmental data to the modified Logan developmental model
(Lactin et al. 1995).
Parameters
Temp. (C)
Sex
Stage
RH
n
R2
P
Tm
AT
A
Optimal
Thres
hold
Female
1st stage
60
85
0.99
0.01
35.49
2.61
-1.09
32.8
8.0
70
60
1.00
0.11
37.07
9.01
0.00
28.1

2nd stage
60
76
0.99
0.11
35.10
9.00
-0.10
26.1
13.0
70
58
0.98
0.11
34.67
9.00
-0.10
25.6
13.5
Male
1st stage
60
251
1.00
0.15
34.01
6.63
0.00
27.4

70
191
0.99
0.01
37.00
3.00
-1.10
34.0
9.5
2nd stage
60
87
0.98
0.11
34.80
9.00
-0.10
25.8
13.2
70
51
0.95
0.11
34.01
9.00
-0.10
25.0
13.6
Prepupa
60
32
0.93
0.11
40.00
9.00
0.00
31.0

70
14
1.00
0.10
34.32
7.46
-2.76
26.8
19.4
Pupa
60
25
1.00
0.01
44.26
6.03
-1.00
38.2
0.1
70
16
0.99
0.11
35.10
8.50
-0.90
26.5
18.2
Tm is the maximum lethal temperature, AT is the difference between the lethal temperature and the optimal temperature of
development, and A is a parameter that makes the curve intercept the x-axis, allowing the estimation of a developmental threshold, w


9
(1974) documented survival and developmental differences induced by host plant
variety. Gill et al. (1982) reported observations on U. euonymi obtained from field
sites, but not from constant-temperature studies.
Fecundity of armored scale insects is not easy to study since births occur
concealed from the observer beneath the scale cover. Nonetheless, fecundity of
snow scale was described in Australia by Waterhouse & Norris (1987). They
reported a maximum of 169 crawlers produced during 5 months. Dickens (1968)
observed a mean of 108 crawlers with a maximum of 250-256, in Florida and under
laboratory conditions. Longevity of gravid females was reported by Dickens (1968)
to average 125 days (4.2 months).
Embryogenesis and fecundity of the arrowhead scale has been described by
Nishino & Furuhashi (1971a) and Adachi & Korenaga (1991), with Adachi &
Korenaga (1991) reporting a bimodal fertility curve. This was explained by the fact
that eggs remain in the ovary until embryonic development is almost completed, and
as well-developed eggs fill the ovary, egg production ceases. The production of new
eggs resumes when mature eggs are deposited.
The sex ratio of U. citri is strongly biased toward the males, with ranges of 3-
6 males per female reported by Dickens (1968) and Casares (1974). The sex ratio
of U. euonymi was described as 7 males per 3 females (Benassy & Pinet 1972), but
Gill et al. (1982) and Cockfield & Potter (1987, 1990) report variable sex ratios
depending upon the plant organ infested.


92
Table 7.5 Parameter values, standard error and confidence intervals at 95% for
development parameters.
Stage
Parameter
Value
Std. error
Asymptotic
Confidence intvl.
P
0.011
0.011
-0.132
0.154
Tm
35.489
9.538
-85.700
156.678
1st stage ?
AT
2.609
5.565
-68.095
73.312
A
-1.091
0.210
-3.754
1.572
P
0.150
0.196
-2.344
2.645
Tm
34.015
8.430
-73.094
141.124
1 stage AT
6.633
8.529
-101.733
114.999
A
-1.0x1 O'5
0.240
-3.055
3.055
P
0.110
3.0x1 O4
0.108
0.112
Tm
35.101
0.769
31.791
38.411
2nd stage ?
AT
9.001
0.000
9.001
9.001
A
-0.100
0.037
-0.258
0.058
P
0.110
7.7x1 O'4
0.107
0.113
Tm
34.802
1.395
28.800
40.805
Males
AT
9.002
0.000
9.007
9.002
A
-0.100
0.741
-0.418
0.219
nymphal females and males were overestimated at low and high temperatures, and
were closer to the observed values at temperatures between. Finally, adult females
were consistently overestimated.
Figure 7.2 shows the simulation plotted over data obtained from a snow scale
culture maintained under variable temperatures in a greenhouse. The simulation
was generated using the mean temperature and amplitude measured in the


95
80 90 100 110 120 130
Time (julian days)
Figure 7.2 Duration of development of citrus snow scale individuals under
greenhouse conditions. Circles: observed data. Lines: simulated values.


68
not apparent through these data, and comparison between substrata is impossible.
Inspection of Casares's (1977) figures indicated estimates of parasitism in females
might be 25% to 55% per leaf and very low (6-10%) per unit of twig surface.
Unfortunately, the figures pooled unparasitized 2nd and 3rd female stages, further
obscuring the incidence of Encarsia parasitism. Nevertheless, Casares concluded
that in addition to its lack of host specificity, Encarsia was not able to control snow
scale because its searches for host on leaves rather than twigs.
In the present investigation, comparable percentages of internal parasitism
by Encarsia occurred between leaves and twigs (Figure 5.3A, 5.3B). However
differences were visible in ectoparasitism by A. lingnanensis, with a predominance
of parasitism on leaves (Figure 5.3C). Results previously reported in Chapter 4
point to the minor contribution of A. lingnanensis to the mortality of citrus snow scale
in colonies resident on trunks. These observations support the notion that
individuals on leaves are more heavily attacked and suffer higher mortality, and
agree with the observations of Carroll (1979) and Murdoch et al. (1989) on California
red scale.
Since this increased mortality is occurring on females, which is also the most
abundant sex, it would be likely that in the absence of A. lingnanensis the
populations of citrus snow scale might increase on leaves. Percent parasitism
measurements should be taken with caution since they can be calculated only from
the observed dead and live individuals, excluding individuals killed by other causes
that otherwise would have been part of the live or parasitized fractions (Carey 1993).
It should be kept in mind that we may be overlooking other potentially important


91
Table 7.4 Linear regression between attrition rate (AR) and temperature (T).
Stage
Model
R2
1st stage
AR =
-0.0402 + 6.54x1 O'3 T
0.7023
2nd stage 9
AR =
1/(124.853-3.798 T)
0.5485
Males
AR =
-0.0303 + 1.494x1 O3 T
0.3124
Adult females
AR =
-0.0103 + 6.12x1 O'4 T
0.3947
The value of k (the number of substages comprising each stage) was chosen
as averages of the values presented in table 7.3 for each stage. It was initially set
to ^ for first nymph, k=5 for female second nymph, k=3 for males and ^ for
adult females. This adult female stage acts as a sink because death at old age is
not being modeled. The resulting model poorly fitted the observed data. The model
was then modified to k=2 for second nymphs and males, which improved the
correspondence between modeled and observed values.
Figure 7.1 shows the correspondence between observed cohorts of citrus
snow scale kept at constant temperatures and the curves of the simulation model.
Visual inspection of the curves suggested a good correspondence of the simulated
curves and the data for the first instar and the second instar female at all the
temperatures. However, the correspondence between simulated and observed data
for adult females was poor, especially at temperatures at the extreme of the range
studied. Tests of goodness of fit (x2 and Kolmogorov-Smirnov) were significant for
every stage in every chamber, indicating poor fitting. For first nymphs there was a
consistent underestimation of the number at the beginning of the simulation that
changed to slight overestimation after the first 5-6 days of simulation. Second


104
Males
Temperature
Adult females
1
in 1
U i*N i
}
1/s
Adults females
dN/dt
afemales
To Workspace


REFERENCES CITED
Adachi, I. & R. Korenaga. 1991. Fertility schedules of Unaspis yanonensis
(Flemiptera: Diaspididae) in relation to daily temperature. Res. Popul. Ecol.
33: 57-68.
. 1992. A simulation model for the arrowhead scale (Hemiptera: Diaspididae)
population dynamics on citrus trees in relation to pest-management
programs. Res. Popul. Ecol. 34: 155-171.
Albrigo, L. G. & R. F. Brooks. 1977. Penetration of citrus cuticules and cells by citrus
snow scale, Unaspis citri (Comst.), pp. 463-467. In W. Grierson [ed.],
Proceedings of the International Society of Citriculture. Orlando, Florida.
Anon. 1954. White louse scale (Unaspis citri). Agrie. Gazette 65: 270-271.
Arias-Revern, J. M. 1988. Aspects of the biology and population ecology of the
purple scale Lepidosaphes beckii (Newmann) and the glover scale L. gloveri
(Packard) (Flomoptera: Diaspididae), citrus pests in Costa Rica. M.Sc. Thesis
Universidad de Costa Rica, San Pedro de Montes de Oca. (In Spanish).
Atkinson, P. R. 1983. Estimates of natural mortality related to environmental factors
in a population of citrus red scale Aonidiella aurantii (Maskell) (Hemiptera:
Diaspididae). Bull. Entomol. Res. 73: 239-258.
Banks, H. J. 1990. Physiology and biochemistry, pp. 267-274. In D. Rosen [ed.],
Armored scale insects. Their biology, natural enemies and control. Elsevier,
Amsterdam.
Baskerville, G. L. & P. Emin. 1969. Rapid estimation of heat accumulation from
maximum an minimum temperatures. Ecology 50: 514-517.
Beattie, G. A. C. & J. G. Gellatley. 1983. Citrus scale insects. (Agfact H2.AE.2)
Department of Agriculture, New South Wales.
Benassy, C. & C. Pinet. 1972. Notes on Unaspis yanonensis Kuw. (Homoptera,
Diaspidinae) in the Maritime Alps. Ann. Zool. Ecol. Anim. 4: 187-212.
106


CHAPTER 2
LITERATURE REVIEW
Published accounts of research on the citrus snow scale (Unaspis citri
Comstock) are not extensive, probably because this insect has been considered a
minor pest. Knowledge of this pest has not been a prime consideration in citrus pest
management. The genus Unaspis includes eight described species and only three
of these are of economic importance. The accumulated knowledge on the citrus
snow scale and the other economically important species in the genus is reviewed
here in an effort to link the knowledge about snow scale with literature accounts of
other Unaspis species. Not surprising is the fact that very little information has been
published on the other species of Unaspis that are not of economical importance.
Taxonomy of the Genus Unaspis.
The diaspidid genus Unaspis probably originated in Asia, in the continental
region between southeastern India and eastern China, where the distributions of the
economically important species converge (UK 1962, 1970, 1988). This genus was
described in 1921 by A.D. MacGillivray. J. H. Comstock originally described
Unaspis citri in 1883 from Citrus in Louisiana and placed it in the genus Chionaspis
(Ferris 1937). Other synonyms for U. citri are Dinaspis annae Malenotti 1917,
4


34
of substrate because different citrus species are involved and effects of host species
on citrus snow scale has been documented (Reed et al. 1967). However, the low
frequency of fruit naturally infested by citrus snow scale in the field could indicate
that the substrate also has an effect on settling preference, developmental rate, or
survival of the scales.


55
may also react to different substrata in different ways, such as exhibiting preference
for searching on some surfaces over others (Carroll 1979, Murdoch et al. 1989).
The phenology of the citrus snow scale has been previously investigated.
Methods included visual counts on trunk, leaves and twig samples that produced
population intensity estimates (Casares 1977). Attempts at biological control against
this scale were made in 1974 through introduction of parasitoids to Florida. The
aphelinid Aphytis lingnanensis "HK-1" was introduced and released. Successful
biological control was reported but citrus snow scale persists as a problem in some
citrus groves in Florida (Browning 1994). The study by Casares (1977) preceded
the introduction of A. lingnanensis "HK-1" and aimed to evaluate the impact and
importance of previously established parasitoids. However, the results were not
presented in a comparative form and failed to consider an important component of
the parasitoid fauna that subsists on nymphal males of the citrus snow scale.
In the present work, population samples of the snow scale were studied with
the use of relative population estimates, which are easier to contrast. The
objectives were to assess the effects and importance of natural enemies, especially
parasitoids, and how populations of snow scale change on the different substrata
and through time.
Materials and Methods
Samples of twigs and leaves were collected monthly from three to five snow
scale-infested citrus trees in a commercial grove in Lake Alfred, Polk County,
Florida. Between five and 12 snow scale infested twigs with leaves were chosen


105
Subsystem
U i*N i


Table 3.2. Duration of development for citrus snow scale at different constant temperatures and relative humidities
(mean sd in days). (F values from 2-way ANOVA for nymph I and II, 1-way ANOVAfor prepupa and pupa.)
RH
%
Temp
(C)
n
1st stage
females
n
2nd stage
females
n
1st stage
males
n
2nd stage
males
n
Prepupa
male
n
Pupa
male
16
17
11.8 3.6
16
51.4 3.6
60
10.8 3.5
29
61.8 4.6
3
5.0 1.7
0
....
21
8
7.0 0.0
6
22.5 3.7
39
7.6 1.4
12
34.5 6.6
5
6.0 2.1
6
5.0 1.6
60
24
25
5.1 1.4
22
18.8 2.1
55
5.2 1.7
12
23.0 2.5
9
3.3 2.6
8
4.8 1.8
28
22
5.7 2.4
22
14.5 1.8
49
5.7 1.7
26
18.4 2.1
12
4.3 1.9
9
4.7 2.0
30
16
8.1 5.4
11
18.9 6.4
53
5.8 2.1
7
19.1 2.1
3
2.0 0.0
2
5.0 1.4
16
18
10.8 5.0
17
52.8 6.3
58
12.5 4.7
14
58.7 4.5
0

0

70
21
21
9.1 5.3
21
26.1 7.2
54
8.1 2.6
34
32.6 3.7
13
4.8 2.3
10
7.8 2.1
24
14
7.0 0.0
14
17.9 1.8
42
7.1 0.3
1
28.0 -
1
2.0 -
5
3.2 2.7
28
7
6.4 1.1
5
15.2 2.3
36
6.2 1.0
1
39.0 -
0

1
4.0 -
Effects of temperature were significant for each stage (nymph I females: F = 13.969; nymph II females: F = 172.633;
nymph I males: F = 79.826; nymph II males: F = 536.401; prepupa: F = 3.750 @60%; pupa: F = 11.458 @70%) except
for pupa at 60% RH. Effects of RH were significant for nymph I males (F = 25.988). Interactions were significant for
nymph I females: (F = 3.003) nymph I males (F=4.879) and nymph II males (F=17.020).
K)
cn


62
Leaves
Twigs
Figure 5.3. Proportion of males and females on samples of citrus leaves and
twigs.


108
. 1990. Euonymus scale (Homoptera: Diaspididae) effects on plant growth
and leaf abscission and implications for differential site selection by male and
female scales. J. Econ. Entomol. 83: 995-1001.
Cooper, R. M. & R. D. Oetting. 1986. Adaptations of scale insects to host variability.
J. Agrie. Entomol. 3: 163-169.
Curry, G. L. & R. M. Feldman. 1987. Mathematical foundations of population
dynamics. The Texas Engineering Experiment Station Monographs Texas
A&M Univ. Press, College Station.
DeBach, P. & S. V. Rao. 1969. Experimental studies on hybridiation and sexual
isolation between some Aphytis species (Hymenoptera: Aphelinidae) II.
Experiments on sexual isolation. Hilgardia 39: 555-567.
Dickens, T. H. 1968. Life history of Citrus Snow Scale Unaspis citri (Comstock).
M.Sc. Thesis University of Florida, Gainesville.
Drea, J. J. & R. W. Carlson. 1987. The establishment of Chilocorus kuwanae
(Coleptera: Coccinellidae) in Eastern United States. Proc. Entomol. Soc.
Wash. 89: 821-824.
. 1988. Establishment of Cybocephalus sp. (Coleptera: Nitidulidae) from
Korea on Unaspis euonymi (Homoptera: Diaspididae) in the eastern United
States. Proc. Entomol. Soc. Wash. 90: 307-309.
Fernandez Argudn, M. 1987. Morfology of Aspidiotiphagus sp. (Hymenoptera:
Aphelinidae). Rev. Prot. Veg. 2: 134-139.
Ferris, G. F. 1937. Atlas of the scale insects of North America. The Diaspididae,
Part I Stanford Univ. Press, Stanford.
Fisher, F. E. 1950. Entomogenous fungi attacking scale insects and rust mites on
Citrus in Florida. J. Econ. Entomol. 43: 306-309.
Fisher, F. E., W. L. Thompson & J. T. Griffiths Jr. 1949. Progress report on the
fungus diseases of scale insects attacking citrus in Florida. Florida Entomol.
32: 1-11.
Fisher, J. 1985. Parasitic wasp prey on snow scale. Citrus Ind. 66: 36,40-42.
Foldi, I. 1990. The scale cover, pp. 43-54. In D. Rosen [ed.], Armored scale insects:
their biology, natural enemies and control. Elsevier, Amsterdam.


Percentage of incidence
63
80
60
40
20
0
80
-C External parasitism
in females
Figure 5.4. Incidence of natural enemies of snow scale on samples of leaves
and twigs. Percentage are calculated over the total of susceptible stages
present in the population.


CHAPTER 4
STUDY OF MORTALITY OF THE CITRUS SNOW SCALE
UNDER FIELD CONDITIONS, BASED ON THE ANALYSIS
OF PARTIAL LIFE TABLES
Introduction
Attempts to manage citrus snow scale by classical biological control began
during the 1970's, but the results of these efforts were never clearly documented
(Browning 1994). Snow scale is still present in citrus groves to the extent that the
use of pesticidal suppression is sometimes warranted (Knapp 1995). There is a
need to study the relationships between the citrus snow scale, its natural enemies
and other biotic and abiotic mortality factors to clarify their value as population
control factors and clarify the success or failure of previous attempts at biological
control.
Life table analysis is an important tool in the study of populations (Harcourt
1969, Varley & Gradwell 1971, Southwood 1978, Carey 1993), providing a detailed
account of the mortality affecting a cohort. A special type of table, the multiple
decrement life table, allows the separation of different factors of mortality, simplifying
the determination of their relative importance (Varley et al. 1973, Carey 1993).
This chapter reports on the mortality of the citrus snow scale under field
conditions in Florida, adapting photographic techniques for sampling of the
35


29
Table 3.6. Parameters calculated to fit citrus snow scale mortality data to a gamma
distribution function.
Stage
RH
Temp
C
Parameters
R2
a
P
1st stage
60
16
0.99
42.22
4.62
21
0.62
36.54
5.19
24
0.98
8.18
1.27
28
0.99
7.19
1.14
30
1.00
6.16
0.80
70
16
0.94
34.96
3.41
21
0.95
4.92
0.60
24
0.90
20.75
2.55
28
0.79
35.57
4.90
2nd stage
60
16
0.93
4.08
0.09
21
1.00
8.17
0.30
24
0.74
4.08
0.17
28
0.93
13.81
0.68
30
0.99
16.11
0.68
70
16
0.95
4.13
0.06
21
0.78
3.93
0.12
24
0.80
35.16
1.54
28
0.92
35.50
1.79
Female
60
21
0.81
80.00
1.03
28
0.88
6.94
0.06
30
0.16
2.40
0.02
70
21
0.96
21.94
0.25
28
0.73
80.00
0.82
Pooled
16
0.99
20.13
0.17
RH
24
0.98
19.52
0.21
Males
60
21
0.90
80.00
1.10
24
0.18
2.03
0.06
28
1.00
51.33
1.20
30
0.93
56.49
1.25
70
21
0.89
11.17
0.13
24
0.28
4.76
0.04
28
0.85
80.00
1.43


72
two distances from the source. This was disturbed by a spray operation, and the
data were again discarded. The fifth trial (April 6-14, 1994) was a repetition of the
fourth and failed in one tree when at least three traps from the East orientation were
disturbed. Data from the second tree were complete. The sixth trial (October 14-21,
1994) was set up around single trees in both sites and these traps were not
disturbed.
In addition to collection of trap data, snow scale spread was monitored from
an artificial infestation in one grove at the CREC (block 16, N-40) during the last two
years. This grove contained 'Valencia' orange and 'Duncan' grapefruit trees. Four
trees in the plot were infested with snow scale in August 28, 1992, by attaching
snow scale infested seedlings to the trunk or main limbs. Monitoring involved
grading of snow scale colonies based on their appearance once every month. A
score from 5 (heavy infestation) to 0 (no infestation) was given to each tree in the
plot, based on a brief inspection for the presence of snow scale male covers on the
trunk and main branches (Casares 1977). Wind direction data for this extended
period were available only as daily means and were not used.
Numbers of crawlers trapped at each direction were compared by Kruskall-
Wallis non-parametric test (Siegel 1956). Visual comparisons were made between
wind patterns and number of catches.
Results
Mylar traps. Figures 6.1 and 6.2 present the number of crawlers caught in
trials 1, 3, 5 (data from a single tree without repetition) and 6. The hourly averages


43
single patch. The tree location, date in which the patches were initiated, date until
they were observed and significance are shown in table 4.1. It is important to note
that the sets of patches that showed significant differences were all initiated during
spring and photographed into summer of both years.
Despite the differences observed between patches within some trees, the
data from all of the patches in each tree were pooled and used to compare between
trees and starting dates. The same statistical tests showed significant differences
(Log-Rank x2 =91.40, Wilcoxon x2 =87.65, P< 0.0001 for both tests, 20 d.f., tree as
strata, PROC LIFETEST, SAS Institute Inc. 1989) suggesting differential survival at
different times of the year or different tree locations. A comparison between trees
at each date on which a set of patches was started was possible for 6 dates (Table
4.2). Four of the 6 dates showed significant differences, suggesting that there is a
site (tree) component contributing to the variation. The survival curves drawn from
these data were diagonal or slightly convex, with the slope becoming steeper during
the late spring and summer months, periods when U. citri development was also
shortened by increased temperatures. Figure 4.1 shows survival and hazard
functions (age-specific probability of death) from selected trees in winter, spring,
summer and fall periods. Mortality was initially low and increased slowly and
steadily through the scale lifetime but variation in mortality patterns was high. A
summary of the fate of the 960 individual scales used in the analysis from a total of
1082 observed is presented in table 4.3. Only 13% of the observed individuals
appeared to complete their life cycle. The overall observed sex ratio was 4.7
males/female.


52
were photographed. It was anticipated that the crowded conditions in citrus snow
scale colonies would favor the spread of diseases. It is very likely that a portion of
the unknown mortality is composed of individuals killed by diseases. Potential
explanations for not observing fungi in our study are that the period between
infection and the development of evident fructiferous bodies could be longer than
the period of our observations (8-20 weeks), or that the disturbance created when
cleaning the patches affected somehow the incidence of fungi, reducing the
crowdedness of the insect colony or the age of abandoned shields, or that the fungi
are nothing more than saprophytes and would not develop on live insects. Nectria
spp. have been mentioned as important mortality factors for scales in citrus (Fisher
et al. 1949) and are found frequently in snow scale colonies, but they as well could
be saprophytes that thrive on the abandoned covers of deceased individuals or
emerged males (Ziegler 1949; Fisher 1950). Kuno & Coln-Ferrer (1973) studied
the pathogenicity of two species of Fusarium over scale insects. They found that
crawler mortality could be increased by the fungi, but there was not effect on older
instars, and the fungi showed saprophytic growth.
In general, mortality of snow scale in the observed patches was very high
(between 70% and 100%, Table 4.4). A substantial percentage of mortality
measured in the various patches was due to unknown causes. Very likely, a part
of it can be attributed to natural enemies (parasitoids, predators and diseases), but
there is no way to separate it using the current data. The observed effects of natural
enemies were considerable and could account for about a quarter of the total
mortality. However, this mortality was insufficient to reduce the densities of snow


POPULATION STUDIES OF THE CITRUS SNOW SCALE
UNASPIS CITRI (COMSTOCK)
By
JULIO M. ARIAS REVERON
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1995
UNIVERSITY OF aORIDA LIBRAR,ES

ACKNOWLEDGMENTS
This research would have not been possible without the guidance, critical
evaluation, encouragement and financing by the committee chair, Dr. Harold W.
Browning. I appreciated especially his immense patience. Very valuable input was
received from the other members of my advisory committee. Dr. Jon C. Allen
helped me understand the intricacies of computer modeling, Dr. Carmine A. Lanciani
advised on life table analysis during the planning phase of this project, Dr. Clay W.
McCoy taught me about insect pathology and the complexities of the researcher's
occupation. Dr. Fred Bennett was instructive about biological control and quarantine
procedures, and helpful while he was part of the committee, until the day of his
retirement.
This research was possible also by the kindness of Mr. Maurice Patrick who
gave me the liberty to do and undo at the citrus plots of his property. During the
work for this research I got invaluable help from Mrs. Pamela Russ, Mr. Ian Jackson
and Mr. Mark Bryan and had the opportunity to share their friendship. Many people
not directly involved in the development of this research need to be thanked,
because they greatly contributed to my well being during my stays in Gainesville and
Lake Alfred: these include Jackie J. Belwood, Vinnod Kutty, the Nielsen family, Faith
and David Oi, Eliane Quntela, Devesh Singh, Hugh Smith and Laurie Wilkins:

without their friendship I could not have succeded. I am perpetually indebted to Ms.
Lois Wood, for her love and friendship and her invaluable editorial help.
Finally, I need to thank my father Julio, my mother Isabel and my son Gabriel
for their love, support and patience. They never gave up on asking "When are you
going to finish?", but they never doubted that I would finish it.
in

TABLE OF CONTENTS
page
ACKNOWLEDGMENTS ii
LIST OF TABLES vii
LIST OF FIGURES ix
ABSTRACT x
CHAPTERS
1 INTRODUCTION 1
2 LITERATURE REVIEW 4
Taxonomy of the Genus Unaspis 4
Economic Importance 6
Biology 6
Dispersal 10
Phenology 10
Trophic Relationships 11
Host Plants 11
Natural Enemies 12
Management 15
Sampling and Forecast 15
Chemical control 16
Biological control 16
3 DEVELOPMENT AND MORTALITY
UNDER CONSTANT TEMPERATURE
AND RELATIVE HUMIDITY 18
Introduction 18
Materials and Methods 20
Results 24
iv

Discussion
30
4 STUDY OF MORTALITY OF THE CITRUS SNOW SCALE
UNDER FIELD CONDITIONS, BASED ON THE ANALYSIS
OF PARTIAL LIFE TABLES 35
Introduction 35
Materials and Methods 36
Data Recording with Photographs 36
Photographic Analysis 37
Statistical Analysis 41
Results 42
Discussion 47
5 COMPARATIVE STRUCTURE OF CITRUS SNOW SCALE
POPULATIONS ON THE CANOPY OF CITRUS TREES
AND THROUGH TIME 54
Introduction 54
Materials and Methods 55
Results 58
Discussion 65
6 THE ROLE OF WIND IN CRAWLER DISPERSION 70
Introduction 70
Methods 70
Results 72
Mylar traps 72
Infestation Scoring Maps 76
Discussion 76
7 MODELING CITRUS SNOW SCALE POPULATIONS 82
Introduction 82
Materials and Methods 83
Model Structure 83
Development 85
Mortality 85
Reproduction and Migration 86
Environmental Variables 86
Model Implementation 87
Model Calibration 87
Model Validation 87
Results 88
v

Discussion 93
8-CONCLUSIONS 98
APPENDIX: SIMULINK BLOCK DIAGRAMS
FOR THE SNOW SCALE MODEL 102
REFERENCES CITED 106
BIOGRAPHICAL SKETCH 118
VI

LIST OF TABLES
Table page
3.1. Temperature and relative humidity means sd
conditions used in the study of citrus snow scale
development 21
3.2. Duration of development for citrus snow scale at
different constant temperatures and relative
humidities 25
3.3. Parameters calculated to fit citrus snow scale
developmental data to the modified Logan
developmental model (Lactin et al. 1995) 27
3.4. Effect of constant temperature across constant relative
humidity on mortality of citrus snow scale 28
3.5. Effect of relative humidity across constant temperatures
on mortality of citrus snow scale 28
3.6. Parameters calculated to fit citrus snow scale mortality
data to a gamma distribution function 29
3.7. Polynomial regressions between temperature and the
parameters of a gamma distribution (a, (3) fitted to the
probability density function of the survival curves 30
4.1. Significance of log-rank and Wilcoxon homogeneity
test, likelihood ratio test for comparison of survival of
citrus snow scale in patches within trees 44
4.2. Significance of log-rank and Wilcoxon homogeneity
test, likelihood ratio test for comparison of survival of
citrus snow scale between trees by dates 45
4.3. Outcome observed for citrus snow scale individuals
under field conditions 47
4.4. Total percentage of mortality (adx) and percent mortality
by unknown causes against parasitism and detectable
predation in multiple decrement life tables, calculated
by tree 48
4.5. Total percent mortality (all causes) adxpresented by
stage 49
6.1. Frequency and speed of wind on the four quadrants in
Lake Alfred, CREC weather station 75
vii

7.1 Parameters used for the simulation of development for
the citrus snow scale 88
7.2 Average duration and percentage of survival from each
stage at each environmental condition studied 89
7.3 Parameters for the distributed delay model, calculated
from data at constant temperature and RH 90
7.4 Linear regression between attrition rate (AR) and
temperature (T) 91
7.5 Parameter values, standard error and confidence
intervals at 95% for development parameters 92
7.6 Parameter values, standard error and confidence
intervals at 95% for mortality parameters 93
viii

LIST OF FIGURES
Figure page
4.1 Survival and hazard functian gf snow scale populations
on selected trees 46
5.1 Age structure of citrus snow scale populations on citrus
leaves and twigs 59
5.2.Ratio of first instar per gravid female of snow scale on
samples of citrus twigs and leaves 60
5.3 Relative proportion of sexes of snow scale on samples
of citrus leaves and twigs 62
5.4. Incidence of natural enemies of snow scale on samples
of leaves and twigs 63
5.5 Relative importance of natural enemies of snow scale
females on samples of leaves and twigs 64
6.1. Wind speed vs direction and catches of snow scale
crawlers during trial 1 and 3 73
6.2. Wind speed vs direction and catches of snow scale
crawlers during trial 5 and 6 74
6.3. Maps of trees infested with citrus snow scale in a citrus
plot (CREC, N-40, block 16) 77
7.1. Duration of development of citrus snow scale
individuals under constant temperature 94
7.2. Duration of development of citrus snow scale
individuals under greenhouse conditions 95
IX

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
POPULATION STUDIES ON THE CITRUS SNOW SCALE
UNASPIS CITRI (COMSTOCK)
By
Julio M. Arias Revern
December, 1995
Chairman: Harold W. Browning
Major Department: Entomology and Nematology
The goal of this research was to expand knowledge of the biology and
ecology of the citrus snow scale Unaspis citri (Comstock) (Homoptera: Diaspididae),
including duration of development on a suitable substrate, causes of mortality under
field conditions, life table analyses, dispersal mechanisms, and computer simulation
of developmental dynamics.
Development of snow scale is temperature dependent. The optimal
temperature for development fell in a range of 25 to 38C, near 29C for most
stages. Development was faster on grapefruit leaves than previously reported on
lemon fruits. The effect of relative humidity was only significant for first instar males,
but the range of relative humidity was too narrow (10%) to observe more
differences.
x

Field studies were accomplished with the use of macrophotography as a
sampling tool to study snow scale colonies on limbs and trunk, and destructive leaf
and twig samples to study snow scale populations within the canopy. Little evidence
of Aphytis lingnanensis Compere (Hymemoptera: Aphelinidae) causing mortality on
citrus snow scale on trunk and limbs was observed. Encarsia spp. were observed
often, attacking both immature males and second instar females. Predators were
evident via disrupted scale covers of some stages. Attack by Encarsia ranged
between 0 and 38%, while observable attacks by predators ranged between 0 and
49%. Analysis of data from leaf and twig samples suggests that different conditions
within trunk and canopy may affect mortality factors. Aphytis lingnanensis and fungi
were present in most of the samples taken during the 2-year study. These played
a more important role on snow scale populations found on leaves and twigs.
Mortality caused by Aphytis lingnanensis was higher on snow scale in the canopy,
which suggests that although this aphelinid has an effect on canopy populations, the
main scale population on trunk and limbs is unaffected. Crawlers were found to
disperse, taking advantage of air currents, as has been reported for other scale
insect species.
A simulation model of inmature development of citrus snow scale was derived
from experimental data obtained under constant temperature conditions. Since the
model describes only part of the life cycle, further research and model development
is needed to better describe snow scale population dynamics.
XI

CHAPTER 1
INTRODUCTION
Few animal species are able to manage their environment to maximize the
collection of resources needed for their survival, other than gathering those
resources readily available. Agriculture is the process of managing ecosystems, the
manipulation of vegetal and animal species to maximize the development and
reproduction of a useful species. Efficiency in management techniques has been
achieved through natural selection within fungus-culturing ants, but for humans, it
has been attained by other means of selection: trial and error, the transfer of
traditional knowledge and research.
It is difficult to manage a system for which we have little understanding.
Solutions chosen without knowledge will very likely lead to waste of resources or to
unexpected secondary effects, as happens with development of pesticidal
resistance and pollution produced by the indiscriminate use of pesticides. Research
leading to an increase in our understanding of a natural or artificial ecosystem is an
essential component of the design of management strategies.
Another characteristic of agriculture is that successful crops are usually
planted in regions far away from their area of origin (Kloppenburg and Kleimman
1987). This separates the crop plants from organisms that may have coevolved with
the plant and use their tissues as food resources. When one or more of these
1

2
organisms make their way to a new area where the host plant is being cultivated as
a crop, they usually reach the area without their own natural enemies, and become
free to colonize and exploit the resource plant without constraints of predation,
parasitism or diseases. These organisms then become pests.
The citrus ecosystem in Florida is dominated by an exotic plant species
brought to a geographical area with a benign environment, and where many of its
pest organisms have been arriving one at a time. Management of these citrus pests
has relied upon use of pesticides, but in the recent times, more pest species have
been successfully managed by a technique of ecosystem reconstruction, biological
control. Biological control consists of bringing to the new ecosystem some of the
missing components from the original ecosystem, particularly the natural enemies
that feed or develop on the insect pests. Biological control is a management
technique that requires a very good understanding of the relationships between the
plant, the pest and its natural enemies.
The citrus snow scale Unaspis citri (Comstock) is today an important pest of
citrus in Florida. It was a minor pest in the past, but its relative importance has
increased over the years, with geographical spread throughout the state and the
successful biological control of other, once more important pests (Browning 1994).
Knowledge about its biology is meager and attempts to control it with pesticides or
with natural enemies have been unsuccessful. The present work is aimed at
understanding details of the citrus snow scale life history and relationships with its
surroundings that could contribute towards developing successful biological control.
I was interested in expanding the knowledge of its biology, measuring the duration

3
of development on suitable substrata, elucidating causes of mortality that affect
citrus snow scale in the field, and looking into the results of previous attempts of
biological control and the potential reasons why those may have been unsuccessful.
The present work is organized into chapters along major areas of the
research. The next chapter presents background information about the genus
Unaspis, the species citri and its close relatives. It also introduce the techniques
that were used in the research reported in later chapters. The third chapter
describes the development of snow scale under constant temperature and constant
relative humidity conditions, on a suitable substratum of a common Florida citrus
variety. The fourth chapter discusses mortality factors affecting snow scale under
field conditions on the bark of citrus trees. Using photographic techniques and
survival table analysis, the role of natural enemies is characterized. The fifth
chapter complements the fourth chapter with observations on mortality causes, age
structure and sex ratios in the canopy of citrus trees. The sixth chapter looks into
the potential of dispersion of snow scale via wind. The seventh chapter summarizes
the snow scale knowledge in a simulation model that evaluates the quality of the
new information and identifies gaps requiring additional research. Since each
chapter is designed to stand alone as a single research paper, a certain amount of
redundancy can be found between Chapter 2 (Literature review) and the
introductory sections of each subsequent chapter.

CHAPTER 2
LITERATURE REVIEW
Published accounts of research on the citrus snow scale (Unaspis citri
Comstock) are not extensive, probably because this insect has been considered a
minor pest. Knowledge of this pest has not been a prime consideration in citrus pest
management. The genus Unaspis includes eight described species and only three
of these are of economic importance. The accumulated knowledge on the citrus
snow scale and the other economically important species in the genus is reviewed
here in an effort to link the knowledge about snow scale with literature accounts of
other Unaspis species. Not surprising is the fact that very little information has been
published on the other species of Unaspis that are not of economical importance.
Taxonomy of the Genus Unaspis.
The diaspidid genus Unaspis probably originated in Asia, in the continental
region between southeastern India and eastern China, where the distributions of the
economically important species converge (UK 1962, 1970, 1988). This genus was
described in 1921 by A.D. MacGillivray. J. H. Comstock originally described
Unaspis citri in 1883 from Citrus in Louisiana and placed it in the genus Chionaspis
(Ferris 1937). Other synonyms for U. citri are Dinaspis annae Malenotti 1917,
4

5
Prontaspis citri (Comstock) MacGillivray 1921 and Dinaspis veitchi Green & Laing
1923. Ferris (1937) assigned those synonyms to Unaspis citri.
Unaspis citri is recorded from Citrus and related genera in the family
Rutaceae and occurs in China, Indochina, Australia, Central Africa, South America
and the Caribbean (UK 1962). The same source reports this species from the citrus
areas of North America, but more recent information restricts it to Florida and
Louisiana (Browning 1994). In addition to U. citri, the two other Unaspis species of
economic importance are the arrowhead scale, U. yanonensis and the euonymus
scale, U. euonymi. Unaspis yanonensis infests Citrus in Japan, China and southern
France, but does not occur in the Americas or Australia (UK 1988). Unaspis
euonymi attacks ornamental plants in the genus Euonymus (Celastraceae), Prunus
(Pomaceae) and Hibiscus (Malvaceae). It occurs in Asia (Japan, China), Europe
and North America (UK 1970). The remaining species, Unaspis acuminata, U.
atricolor, U. flava and U. permutans were described from material collected from
various host plants in Sri Lanka and the south and east of India, while U. turpiniae
was described from material collected in the Philippines (Rao 1949).
The diagnostic characteristic that separates U. citri from the other species of
the genus is the reduction or absence of perivulvar glands in mature females (Rao
1949). Perivulvar glands are associated with ovoviviparity. Although some degree
of ovoviviparity occurs in most diaspidid scale insects, the bounds of this
phenomenon are not well delimited. Prevailing theory indicates that part or all of
the embryo's development could occur internally in the mother of an ovoviviparous
species. Conversely, the presence of an egg shell or chorion is considered to be

6
an indication of oviparity. The perivulvar glandular system may be associated with
the presence of the egg shell and thus its reduction in some species is thought to
parallel progressing viviparity (Koteja 1990a).
Economic Importance
Unaspis citri is considered an important pest of citrus in the Americas and
Australia, based on the stress that this species inflicts on the general health of the
tree and on the cost of its control. Bark and scaffold limbs are the primary sites of
attack on citrus trees. However, in heavy infestations, it also attacks leaves and to
a lesser extent fruits. Bark splitting, twig dieback and tree death are also associated
with severe infestations (Beattie & Gellatley 1983, Smith & Papacek 1985, Browning
1994).
Unaspis yanonensis is an important pest of citrus in Japan, where the
increase of populations in the absence of effective natural enemies causes damage
to citrus fruits (Ohkubo 1981). Unaspis euonymi is very important in landscape
horticulture in North America and Europe, producing chlorosis, reduced
photosynthesis, leaf abscission and stunting in Euonymous plants (Vinis 1977, Gill
et al. 1982, Brewer & Oliver 1987, Cockfield & Potter 1987).
Biology
The biology of citrus snow scale, as studied by Dickens (1968), typifies the
life cycle of many other diaspidid scales. Females have two molts, or two nymphal
stages and the imago or adult stage. Males have four molts leading to two nymphs,

7
two pupal stages and the imago. The first instar (crawler) and the adult male are the
only mobile stages. The crawlers search for a location to settle and feed, where
they will remain throughout their development. Second instar females secrete a
proteinaceous shield, while males produce a white waxy tricarinated shield. After
the second molt, females reach adulthood and produce a dark shell-like secretion
that continues to enlarge until oviposition begins. Conversely, males complete
feeding in the second instar and molt to the first pupal stage. Subsequently, they
progress through the third molt to a second pupal stage and then after the fourth
molt a winged male emerges (Dickens 1968, Koteja 1990b).
Female citrus snow scales reach reproductive maturity and deposit their eggs
beneath the secreted shield. The incubation period is very short, probably between
30-60 minutes, suggesting that most egg development occurs inside the body of the
female. Ovoviviparity is presumed to be an ancestral condition associated with
tropical environmental conditions, while retarded embryonic development and
oviparity, often associated with winter diapause, is considered a secondary
adaptation to temperate regions (Koteja 1990a). Unaspis citri embryos may be laid
enclosed in membranes instead of egg shells (Koteja 1990a), since there is no trace
of a chorion after the crawler ecloses.
The duration of development of poikilotherm organisms is dependent upon
temperature, and the relationship between development and temperature is a
fundamental feature of an insect life history (Taylor 1981). To characterize the
developmental rate, it is necessary to study the duration of development under a
range of enviromental conditions and to describe a relationship with temperature

8
and other relevant factors. The relationship between temperature and development
has been described in many ways, including linear relationships (Baskerville & Emin
1969, Sevacherian et al. 1977), and curves and asymmetrical functions (Logan et
al. 1976, Sharpe & DeMichele 1977, Taylor 1981, Wagner et al. 1984). The
developmental rate increases with an increase of temperature. Development stops
below a lower and above an upper temperature threshold.
Dickens (1968) studied duration of development of the citrus snow scale on
'Valencia' orange fruits, at a constant temperature of 26C. He found that first instar
development may require around 13 days, while the duration of the second instar
is about 11 days for males and 18 days for females. Females started producing
eggs about 60 days after they were born. Casares (1974) studied citrus snow scale
development on lemon fruits under several constant temperatures, and confirmed
that developmental rate increased with an increase in temperature. These efforts
to describe citrus snow scale development were made on citrus fruits, which are not
the preferred substrata under field conditions.
Development of U. yanonensis has been studied extensively, including
development under constant temperatures (Okudai et al. 1971, 1974, Huang et al.
1983) and fluctuating temperatures (Korenaga et al. 1976). The optimal
temperature for development was calculated to be about 27C and the lower
developmental threshold was near 10C (Okudai et al. 1971, 1974). Studies also
addressed the relationship between temperature and ovarial development,
forecasting the appearance of the first instar (Nishino & Furuhashi 1971a, b, Okudai
et al. 1975). Nishino (1974) has summarized most of this research. Furuhashi

9
(1974) documented survival and developmental differences induced by host plant
variety. Gill et al. (1982) reported observations on U. euonymi obtained from field
sites, but not from constant-temperature studies.
Fecundity of armored scale insects is not easy to study since births occur
concealed from the observer beneath the scale cover. Nonetheless, fecundity of
snow scale was described in Australia by Waterhouse & Norris (1987). They
reported a maximum of 169 crawlers produced during 5 months. Dickens (1968)
observed a mean of 108 crawlers with a maximum of 250-256, in Florida and under
laboratory conditions. Longevity of gravid females was reported by Dickens (1968)
to average 125 days (4.2 months).
Embryogenesis and fecundity of the arrowhead scale has been described by
Nishino & Furuhashi (1971a) and Adachi & Korenaga (1991), with Adachi &
Korenaga (1991) reporting a bimodal fertility curve. This was explained by the fact
that eggs remain in the ovary until embryonic development is almost completed, and
as well-developed eggs fill the ovary, egg production ceases. The production of new
eggs resumes when mature eggs are deposited.
The sex ratio of U. citri is strongly biased toward the males, with ranges of 3-
6 males per female reported by Dickens (1968) and Casares (1974). The sex ratio
of U. euonymi was described as 7 males per 3 females (Benassy & Pinet 1972), but
Gill et al. (1982) and Cockfield & Potter (1987, 1990) report variable sex ratios
depending upon the plant organ infested.

10
Dispersal
Wind is presumed to be the main means of dispersion of scale insects
between plants, as has been shown for other species (Greathead 1990). It also has
been proposed that scale insect crawlers disperse on workers' clothing and
equipment (Simanton 1973) or passing animals (Greathead 1990). Xinnian &
Browning (1991) observed that citrus snow scale crawlers were positively
phototactic and that female crawlers walked farther than males. This behavior could
facilitate the expansion of colonies, decrease intraspecific competition between
females, which feed for longer periods than males, and facilitate the movement of
first instars upwards on the tree, where they might be more easily dislodged and
carried away by wind. Similarly, behavior of U. yanonensis crawlers has been
studied by Korenaga (1983) and Wang & Chen (1989).
Phenology
Citrus snow scale does not exhibit diapause in Florida, and thus it reproduces
throughout the year and has overlapping generations. At least two population peaks
have been described in Florida (Dickens 1968, Casares 1977) and in Australia
(Summerville 1935, Anon. 1954, Beattie & Gellatley 1983). Arrowhead scale and
euonymus scale show synchronized generations, with 2 or 3 being reported per
year, depending on temperatures. Arrowhead scales can overwinter in several life
stages (Nishino 1974), while euonymus scales overwinter as adult females
(Williams et al. 1977, Gill et al. 1982).

11
Trophic Relationships
Host Plants. Citrus species are the primary hosts plant for snow scale,
although the scale has been reported from hosts in related genera of Rutaceae such
as Murraya paniculata, Severina buxifolia, and Fortunella sp. (Rutaceae) (Casares
1974, Williams & Watson 1988). Reports of infestations of citrus snow scale on
plants in other families (Palmae, Celastraceacea, Oleacea, Bromeliacea) may be
inaccurate (Casares 1974, 1977) and represent misidentification of other armored
scale species with similar appearance.
Mechanisms of feeding in diaspidid scales are not well understood. It has
been proposed that the stylets penetrate woody tissues to find active sieve tubes
close to the cambium. Another suggested mechanism is that the scales feed on the
contents of any cell invaded by the stylets. The third proposal is that both forms of
feeding may occur. The first mechanism does not explain why diaspidid scales do
not produce large amounts of honeydew secretions. It is hypothesized that unused
materials may be returned to the plant by the insect pumping them with its powerful
salivary glands (Banks 1990). Studies on U. euonymi support the second
hypothesis, intracellular cell feeding on palisade parenchyma in leaves, and feeding
on xylem tissue in limbs (Sadof & Neal 1993). Citrus snow scale damage and
feeding mechanisms were studied by Albrigo & Brooks (1977). They showed that
stylets penetrate the plant tissue intracellularly, damaging cells on the way to
locating the phloem vessels, but they assumed phloem feeding without confirming
it.

12
It has been shown that differential preferences exist for the substrate chosen
by each sex of U. euonymi. Therefore, sex ratio was observed to be different on
different organs of the plant (Benassy & Pinet 1972). Males prefer to settle on
leaves, while females prefer to settle on woody organs. This could be related to
intraspecific competition and risks of mortality associated with leaf abscission. Since
females need to feed for a longer period than males, it may be argued that females
prefer to settle on twigs, which are a more permanent substrate than leaves. Males
on the other hand, tend to settle on leaves to avoid competition with females and
since they only feed for the first two instars, they may survive even if the leaf falls
from the plant (Cockfield & Potter 1987). Citrus trees have perennial foliage and
leaves remain on the tree for long periods, so the ratios or settling preferences could
be different for snow scale.
Natural Enemies. Natural enemies of snow scale have been reported in a
variety of publications and have been reviewed by Waterhouse & Norris (1987). A
discussion of the more important species follows.
Encarsia (=Aspidiotiphagus) citrina (Craw.) and E. lounsburyi (Berlese and
Paoli)(Hymenoptera: Aphelinidae) are the most commonly found and probably most
widely distributed citrus snow scale parasitoids (Hely 1944, Casares 1977, Selhime
& Brooks 1977, Tern et al. 1985, Castieiras & Obregn 1986, Fernandez Argudin
1987). They are polyphagous endoparasitoids that have been closely studied in
relation with other species of host scales (Benassy & Pinet 1972, Kajita 1972, 1976,
1977a, b, Murakami et al. 1972, Kanda & Kajita 1977, Gill et al. 1982). Encarsia

13
citrina and Encarsia longsburyi prefer to attack young second instars of both sexes
of citrus snow scale.
Aphytis lingnanensis Compere (Hymenoptera: Aphelinidae) preferentially
attacks late second instar or young adult (prereproductive) females of citrus snow
scale. It is also a polyphagous parasitoid originally from Asia; and this species
develops as an arrhenotokous ectoparasitoid. Several related forms or strains have
been identified, morphologically indistinguishable from A. lingnanensis but with
varying degrees of reproductive isolation (DeBach & Rao 1969). These strains have
been widely used in attempts at biological control of several pests, including the
introduction of the Hong Kong "HK-1" strain of A. lingnanensis into Florida for the
control of snow scale and the introduction of the Hong Kong "HK-J" strain into Japan
for control of the arrowhead scale (Rosen & DeBach 1979, Tanaka 1981).
Aphytis gordoni (Hymenoptera: Aphelinidae) is another ectoparasitoid, half
the size of A. lingnanensis, discovered during exploration for snow scale parasitoids
in Hong Kong. All of the material was obtained from U. citri hosts (Rosen & DeBach
1979), which could be an indication of host specificity and make this species worthy
of more research as a potential biological control agent for snow scale.
Chilocorus circumdatus (Coleptera: Coccinellidae) was recently reported
feeding on citrus snow scale in Australia where it was originally introduced from
China for the control of Aonidiella aurantii (Houston 1991). This coccinellid is a
diaspidid scale feeder and was introduced during biological control projects for other
species of armored scales in the U.S. (Rosen & DeBach 1977).

14
Fungal diseases have been reputed to play an important role in citrus snow
scale population regulation, especially because crowded citrus snow scale colonies
provide an ideal situation for the spread of disease. Several fungal genera have
been reported in association with snow scale (Sphaerostilbe, Podonectria) (Dickens
1968). Most of those reports are of species now recognized within the genus
Nectria (anamorph Fusarium, Samson et al. 1988) Some debate and contradictory
results of experimentation emerged from the period between the end of last century
to the late 1940s regarding the effectiveness of these fungi in the control of scale
insect populations. Nectria species were confirmed as developing saprophytically
on the bodies and covers of dead scales, but it is not clear whether these fungi also
function as true pathogens (Ziegler 1949, Fisher et al. 1949, Fisher 1950).
Natural enemies are an important means of population regulation. The role
of a given species can be studied by laboratory experimentation or by field
observations and manipulations. Unfortunately, observations in the field could be
influenced by many uncontrollable factors. One way to quantify mortality and its
causes is by using life table analysis. Life tables consist of a systematic accounting
of the number of individuals alive and dead at each age or stage and, when
possible, the causes of death. Life table analysis is a technique borrowed from the
insurance business that was applied to animal populations (Hutchinson 1978).
Applications of life table analysis to insects are reviewed in Harcourt (1969), Varley
et al. (1973), Southwood (1978) and Carey (1993).

15
Management
Sampling and Forecasting. The snow scale presents special problems for
sampling, since it occurs mainly on citrus trunks and scaffold limbs. These
substrata cannot be easily harvested and processed in the laboratory, as leaf
samples are. Sampling methods for citrus snow scale include the use of rating
systems for the appearance of scale colonies and the clearing of rectangular
patches within active colonies on the trunk. The patches are colonized by scales,
and then scales are counted in situ with the help of a hand lens (Casares 1977).
The rating methods tend to overestimate the active scale populations since they rely
on the appearance of male covers which can accumulate on the trunk for extended
periods, and thus represent both live scales and those that have already emerged.
Female citrus snow scales are generally overlooked in rating methods since they are
difficult to see on citrus bark.
Extensive efforts have been invested in creating models that predict
population densities of the arrowhead scale, U. yanonensis (Takezawa & Uchida
1969, Nishino & Furuhashi 1971a, b, Korenaga et al. 1974, 1976, Nishino 1974,
Korenaga & Sakagami 1981, Sakagami & Korenaga 1982, Adachi & Korenaga
1992). These models also predicted the results of interactions between biological
and chemical control strategies (Adachi & Korenaga 1992). Simulation models have
been used to predict and manage other species of scale insects as well (Pfeiffer
1985, McClain et al. 1990a, b).

16
Chemical control. Since snow scale has overlapping generations in Florida
citrus, it is extremely difficult to effectively target pesticidal applications to the most
susceptible stages, as is done for the arrowhead scale (Adachi & Korenaga 1992).
The recommendations for materials and equipment for chemical control are given
by Knapp (1995). Recommendations include applying pesticides with hand-held
equipment in localized areas where the scale is a problem and trying to achieve the
maximum coverage possible.
Biological control. Biological control of citrus snow scale in Florida began in
the 1970's. The first recorded introductions of E. citrina and A. lingnanensis
occurred in 1969. A. lingnanensis "HK-1" was introduced into Florida in 1972, and
was reported as being established and effectively controlling U. citri (Riehl et al.
1980). During the period 1974-1985, approximately 6.5 million parasitoids were
released, and savings early in the program were estimated at 8-10 million dollars in
pesticide applications (Mead 1976). The program of mass release of HK-1 was
discontinued because of the effectiveness of the parasitoid was not demonstrated,
and new efforts to identify alternative solutions have begun (Browning 1994).
Several other strains of A. lingnanensis and two other species, A. khunti, and A.
yanonensis, have been introduced into Florida, but establishment has not been
confirmed (Browning 1994).
Unaspis yanonensis became a very serious problem in Japan after it became
established there without natural enemies (Ohkubo 1981, Takagi 1981). Aphytis
yanonensis and Physcus fulvus (Hymenoptera: Aphelinidae) were introduced from

17
China in 1980 and showed good results (Furuhashi & Nishino 1983). A coccinellid,
Chilocorus kuwanae, and a nitidulid, Cybocephalus sp., have been introduced from
Korea into the U.S.A. for the control of Unaspis euonymi and field released in 1984
(Drea & Carlson 1987, 1988).

CHAPTER 3
DEVELOPMENT AND MORTALITY
UNDER CONSTANT TEMPERATURE
AND RELATIVE HUMIDITY
Introduction
Citrus snow scale, Unaspis citri (Comstock), is regarded as a pest of citrus
in Florida and in many other tropical and subtropical citrus production areas
(Dickens 1968, Smith and Papacek 1985, Rose 1990). The relative importance of
the snow scale in Florida increased over the period 1963-1971 (Simanton 1973)
because of improved management of other important pests and the spread of
infested nursery stock throughout the state (Browning 1994).
The life cycle of the snow scale described by Dickens (1968) is typical of
most armored scales. The sexes are dimorphic; females proceed through 2
nymphal stages before reaching maturity, whereas males have 2 nymphal stages
similar to females, and then 2 non-feeding stages, called the prepupa and pupa.
Development under constant temperatures was studied on citrus fruits by Dickens
(1968) and Casares (1974). However snow scale is seldom found on fruit in the
field, living mostly on trunks, branches, twigs and, to a lesser extent, foliage.
Casares (1977) conducted ecological studies on the phenology of the scale and
concluded that rainfall, variety and tree age are important factors in determining
snow scale population levels.
18

19
Diverse approaches have been taken to simulate development of
poikilothermic organisms, from simple assumptions of linear dependency between
temperature and development to more complicated nonlinear equation models
(Curry and Feldman 1987). Developmental processes depend on temperature-
dependent chemical reactions and are bounded by upper and lower thresholds.
Below lower temperature thresholds, developmental processes are slow or halted;
likewise, above upper temperature thresholds development is retarded, proteins are
denatured and subsequent death of the organism occurs (Sharpe et al. 1977).
Lower developmental thresholds are easily calculated when the developmental rate
is assumed to be linear. However, when the rates are assumed to be nonlinear, the
developmental functions that have been proposed approach zero asymptotically,
making it impossible to deduce a lower threshold from the mathematical relationship
estimated (Logan et al. 1976, Wagner et al. 1984). Development at low temperature
may proceed so slowly that it may not be perceptible to the observer, and most
models cannot account for this phenomenon. Lactin et al. (1995) proposed a
modification of Logan et al. (1976) that estimates lower developmental thresholds.
This modification adds a parameter that changes the scale of the curve, allowing it
to intercept with the x-axis (zero development).
In an effort to achieve effective control of citrus snow scale, researchers are
currently striving to better understand its biology and ecology. The current study
relates development and mortality to temperature and relative humidity under
laboratory conditions, and adds new information by using a substrate similar to that
inhabited by the scale under natural conditions.

20
Materials and Methods
Snow scale development and mortality were studied on leaves of 'Duncan'
grapefruit seedlings, Citrus paradisi, under 9 constant temperature and relative
humidity conditions. The seedlings were planted in 150-cm3 plastic conical
containers and maintained using standard nursery practices. These seedlings were
infested with snow scale by attaching them to the trunk of infested trees in the field
(Casares 1974). Infested seedlings were carefully inspected to eliminate possible
contaminants (mealybugs, mites, whiteflies, other species of scales, and possible
predators of the established 1st instar), and held in cages.
Male scales and immature stages were removed after F1 crawler activity was
evident, leaving only reproductive females on the seedlings. Seven to 10 seedlings
bearing gravid females were then placed in each environmental chamber. The
gravid females were removed after 48 h, and the newly settled crawlers were
located and mapped. These cohorts were examined under a dissecting microscope
at 48 to 72 h intervals, depending on the temperature of the chamber. The
observations continued until insects reached the adult stage.
Eight computer-controlled Florida Reach-In chambers (Walker et al. 1993)
and another chamber (Lunaire, Lunaire Environmental, Williamsport, PA) were used
in the studies. The 8 Reach-In chambers were set at combinations of 4
temperatures and 2 relative humidities, with a constant photoperiod of 12:12 h for
each treatment (Table 3.1). The additional incubator (chamber 45, Table 3.1) was
set at 30C and relative humidity was maintained near 60% with the use of
saturated NaCI solutions (Carr and Harris 1949, Greenspan 1977, Marcandier and

21
Table 3.1. Temperature and relative humidity means sd conditions used in the
study of citrus snow scale development.
Chamber
no.
Temp. C
RH %
17
15.95 1.00
60.11 4.11
18
20.95 1.06
59.86 3.06
19
16.10 0.98
71.84 5.72
20
20.96 0.94
69.85 3.15
41
24.00 0.10
60.07 1.29
42
27.99 0.18
60.00 0.46
43
24.06 0.64
69.70 3.55
44
27.98 0.39
69.59 1.83
45
29.77 1.14
56.28 7.55
Conditions in chambers 17-44 (Florida Reach-In) were logged in by the controlling
computer every 10 min, conditions in chamber 45 (Lunaire) monitored with
hygrothermograph and the statistics calculated from 2-h intervals.
Khachatourians 1987). Conditions inside the Florida Reach-In chambers were
monitored at 10-min intervals with built-in sensors, downloaded to a file and plotted
weekly. Conditions in the Lunaire chamber were monitored using a hygro
thermograph (model H-302, WeatherMeasure, Sacramento CA).
Developmental time and survival for each stage and sex, from settled crawler
to adult, were recorded at each observation date. Molts were easily recognized
under stereomicroscopy. Females showed darkening and separation of the exuviae;
males expelled the exuviae from under the armor at the distal end. Female
development from the last molt to the reproductive stage and egg development were
not assessed because these periods could not be observed without lifting the armor
and disrupting the insect.

22
The modification of the Logan et al. (1976) nonlinear model by Lactin et al.
(1995) was chosen to fit the data. This model was preferred because it is
descriptive, the parameters may be interpreted biologically, it is more realistic than
linear models or symmetrical nonlinear models (Lamb et al. 1984), it is simpler to fit
to the data than models with more parameters (Wagner et al. 1984), and it includes
a developmental threshold. The modified Logan model has 4 parameters to
describe the effect of temperature on the development of poikilotherm organisms:
(3.1)
where 7is temperature, pis the rate of increase to optimum temperature, Tm is the
maximum lethal temperature, AT is the difference between the lethal temperature
and the optimal temperature of development, and A is a parameter that makes the
curve intercept the x-axis, allowing the estimation of a developmental threshold. In
the Logan model, temperature values in Celsius are transformed to the base
temperature, the values for duration of development are inverted to calculate rates
of development, and the means of log transformed rates are used to calculate the
model parameters (Logan et al. 1976). In this research, as in the work of Lactin et
al. (1995), all those manipulations were avoided, minimizing the errors that may
arise from computation (Kramer et al. 1991). The inverse equation of the model was
used to fit the original mean duration of each stage (in days), instead of using the

23
rates. The mean durations were weighted by the sample size. Temperature was
not transformed to the base temperature because this substraction is simply a
change in scale and should not produce any effect on values of the parameters
when fitting the curve. The curves were fit using the Marquart method (PROC NUN,
SAS Institute 1989).
Mortality was analyzed through life table methods, using insect age as the
time variable. Survival curves for the 1st and 2nd nymphs were calculated by
pooling data for both sexes. For male prepupal and pupal stages and virgin and
gravid females, stages were pooled and the survival curves calculated for each sex.
This was done for practical purposes because sexes are difficult to separate for 1st
and 2nd instars and stages are difficult to separate for later instars. Differences
between the curves were tested using the Wilcoxon test, which is more sensitive to
differences at early survival times and the Savage (log-rank) test, which is more
sensitive to differences at later survival times (PROC LIFETEST, SAS Institute
1989).
The rate at which the survival curve decays with respect to time corresponds
to the 1st derivative of the function (the slope of the tangent to the curve), or the
probability density function of the survival curve (SAS Institute 1989). This function
was calculated for each stage and sex and fitted, using the DUD method (PROC
NUN, SAS Institute 1989), to a gamma distribution model (Curry and Feldman
1987):

24
M (t)
pa t 1
e pt
r (a)
(3.2)
where t is age and a and 3 are parameters that govern the shape of the distribution.
The parameters obtained were related to temperature using regression of
polynomial equations (PROC REG, SAS Institute 1989).
Results
Table 3.2 presents average duration of development in days, with standard
deviation and sample size for each sex and stage, at each combination of
temperature and humidity.
The effect of temperature on developmental time was statistically significant,
except for the pupal stage at 60% RH (Table 3.2). The length of each stadium
decreased as temperature increased until an optimal temperature for development
was reached. Duration of development then increased at temperatures higher than
optimal. The effect of relative humidity was significant only for 1 st nymphal males,
where 70% RH expedited development by 1-2 d. Interaction effects were significant
for 1st nymphs of both sexes and 2nd nymphal males (Table 3.2).
No significant differences were found between sexes in development of 1st
stages under either relative humidity condition (F = 0.45, df = 1, 341; P = 0.51 for
60% RH, F = 0.80, df = 1, 247; P = 0.38 for 70% RH, covariance analysis using
temperature as covariates, Steel and Torrie 1980). Gender differences were

Table 3.2. Duration of development for citrus snow scale at different constant temperatures and relative humidities
(mean sd in days). (F values from 2-way ANOVA for nymph I and II, 1-way ANOVAfor prepupa and pupa.)
RH
%
Temp
(C)
n
1st stage
females
n
2nd stage
females
n
1st stage
males
n
2nd stage
males
n
Prepupa
male
n
Pupa
male
16
17
11.8 3.6
16
51.4 3.6
60
10.8 3.5
29
61.8 4.6
3
5.0 1.7
0
....
21
8
7.0 0.0
6
22.5 3.7
39
7.6 1.4
12
34.5 6.6
5
6.0 2.1
6
5.0 1.6
60
24
25
5.1 1.4
22
18.8 2.1
55
5.2 1.7
12
23.0 2.5
9
3.3 2.6
8
4.8 1.8
28
22
5.7 2.4
22
14.5 1.8
49
5.7 1.7
26
18.4 2.1
12
4.3 1.9
9
4.7 2.0
30
16
8.1 5.4
11
18.9 6.4
53
5.8 2.1
7
19.1 2.1
3
2.0 0.0
2
5.0 1.4
16
18
10.8 5.0
17
52.8 6.3
58
12.5 4.7
14
58.7 4.5
0

0

70
21
21
9.1 5.3
21
26.1 7.2
54
8.1 2.6
34
32.6 3.7
13
4.8 2.3
10
7.8 2.1
24
14
7.0 0.0
14
17.9 1.8
42
7.1 0.3
1
28.0 -
1
2.0 -
5
3.2 2.7
28
7
6.4 1.1
5
15.2 2.3
36
6.2 1.0
1
39.0 -
0

1
4.0 -
Effects of temperature were significant for each stage (nymph I females: F = 13.969; nymph II females: F = 172.633;
nymph I males: F = 79.826; nymph II males: F = 536.401; prepupa: F = 3.750 @60%; pupa: F = 11.458 @70%) except
for pupa at 60% RH. Effects of RH were significant for nymph I males (F = 25.988). Interactions were significant for
nymph I females: (F = 3.003) nymph I males (F=4.879) and nymph II males (F=17.020).
K)
cn

26
significant for the 2nd stage at both humidities (F = 44.31, df = 1, 160 for 60% RH,
F = 30.66, df = 1, 104 for 70% RH, both P < 0.05), with females showing a shorter
2nd instar (Table 2). No significant differences in sex ratio were found at any of the
environmental conditions (F = 0.33, df = 8, 67; P = 0.95). The average sex ratio
ranged between 2.43 and 4.97 males per female, with a mean of 3.38.
Table 3.3 presents the parameters and R2 values for the fit to the modified
Logan model to the developmental data. It also shows the optimal temperatures
calculated for development of each stage and the estimated developmental
threshold. Sample sizes of later instars were reduced by mortality in previous
stages, yielding less precise estimation of the fitted curves. The optimal
temperatures for development were between 25 and 38 C for most stages, but the
thresholds varied widely from 0.08 to 18.2C, with the extreme values being quite
high and clearly unrealistic estimates.
Tables 3.4 and 3.5 present the effects of temperature and relative humidity
respectively, on survival of snow scale. There were significant differences between
the survival curves at different temperatures for all of the stages except adult
females (Table 3.4), and relative humidity affected 1st instars at temperatures
>16C, 2nd instars only at 16C and males at 24 and 28C.
Table 3.6 presents the parameters for the fit of a gamma distribution to the
probability density function of the survival curves. Data were pooled for older
stages (males and females). Table 3.7 presents the regression of the parameters
a and p of the gamma distribution versus temperature using polynomial equations.

Table 3.3. Parameters calculated to fit citrus snow scale developmental data to the modified Logan developmental model
(Lactin et al. 1995).
Parameters
Temp. (C)
Sex
Stage
RH
n
R2
P
Tm
AT
A
Optimal
Thres
hold
Female
1st stage
60
85
0.99
0.01
35.49
2.61
-1.09
32.8
8.0
70
60
1.00
0.11
37.07
9.01
0.00
28.1

2nd stage
60
76
0.99
0.11
35.10
9.00
-0.10
26.1
13.0
70
58
0.98
0.11
34.67
9.00
-0.10
25.6
13.5
Male
1st stage
60
251
1.00
0.15
34.01
6.63
0.00
27.4

70
191
0.99
0.01
37.00
3.00
-1.10
34.0
9.5
2nd stage
60
87
0.98
0.11
34.80
9.00
-0.10
25.8
13.2
70
51
0.95
0.11
34.01
9.00
-0.10
25.0
13.6
Prepupa
60
32
0.93
0.11
40.00
9.00
0.00
31.0

70
14
1.00
0.10
34.32
7.46
-2.76
26.8
19.4
Pupa
60
25
1.00
0.01
44.26
6.03
-1.00
38.2
0.1
70
16
0.99
0.11
35.10
8.50
-0.90
26.5
18.2
Tm is the maximum lethal temperature, AT is the difference between the lethal temperature and the optimal temperature of
development, and A is a parameter that makes the curve intercept the x-axis, allowing the estimation of a developmental threshold, w

28
Table 3.4. Effect of constant temperature across constant relative humidity on
mortality of citrus snow scale, x2 values are for log rank test (above) and
Wilcoxon tests (below) (SAS Institute Inc., 1989).
Nymph I
Nymph II
Females
Males
RH
(%)
n
X2
n
x2
n x2
n
X2
60
1530
159.63***
889
173.31***
es ivr
7.94 ns
245
83.86***
241.18***
127.43***
45.39***
70
926
4.47 ns
15.72***
681
208.29***
141.44***
143 2 "S
4.52 ns
182
26.56***
13.98***
***= P < 0.001, **= P < 0.01, *:
= P < 0.05, ns = not significant.
Table 3.5. Effect of relative humidity across constant temperatures on mortality of
citrus snow scale.
X2 values are for log rank test (above) and Wilcoxon tests
(below) (SAS Institute Inc., 1989).
Nymph I
Nymph II
Females
Males
T
(C)
n
X2
n
X2
n x2
n
X2
16
554
1.35 ns
1.93 ns
420
12.87***
12.11***
yi- 0.00 ns
0.00 ns
99
21
435
I.30 ns
II.42 ***
296
2.44 ns
0.09 ns
cc 0.91 ns
66 0.91 ns
161
0.02 ns
0.47 ns
24
548
86.95***
105.68***
380
0.02 ns
5.50 ns
in9 0.66 ns
0.66 ns
79
0.75*
0.47*
28
467
52.01***
74.93***
285
0.07 ns
0.23 ns
7fi 0.20 ns
0.18 ns
98
5.23*
1.99 ns
***= P < 0.001, **= P < 0.01, *= P < 0.05, ns = not significant.

29
Table 3.6. Parameters calculated to fit citrus snow scale mortality data to a gamma
distribution function.
Stage
RH
Temp
C
Parameters
R2
a
P
1st stage
60
16
0.99
42.22
4.62
21
0.62
36.54
5.19
24
0.98
8.18
1.27
28
0.99
7.19
1.14
30
1.00
6.16
0.80
70
16
0.94
34.96
3.41
21
0.95
4.92
0.60
24
0.90
20.75
2.55
28
0.79
35.57
4.90
2nd stage
60
16
0.93
4.08
0.09
21
1.00
8.17
0.30
24
0.74
4.08
0.17
28
0.93
13.81
0.68
30
0.99
16.11
0.68
70
16
0.95
4.13
0.06
21
0.78
3.93
0.12
24
0.80
35.16
1.54
28
0.92
35.50
1.79
Female
60
21
0.81
80.00
1.03
28
0.88
6.94
0.06
30
0.16
2.40
0.02
70
21
0.96
21.94
0.25
28
0.73
80.00
0.82
Pooled
16
0.99
20.13
0.17
RH
24
0.98
19.52
0.21
Males
60
21
0.90
80.00
1.10
24
0.18
2.03
0.06
28
1.00
51.33
1.20
30
0.93
56.49
1.25
70
21
0.89
11.17
0.13
24
0.28
4.76
0.04
28
0.85
80.00
1.43

30
Table 3.7. Polynomial regressions between temperature and the parameters of a
gamma distribution (a, P) fitted to the probability density function of the
survival curves
Stage
RH
R2
Equation
1st stage
60
0.84
0.74
a = 131.70-6.79 7+0.084 T2
3 = 7.51 -0.07 7-0.0056 72
70
0.86
0.91
a = 338.20-30.10 7+0.691 72
3 = 37.11 -3.41 7+0.081 72
2nd stage
60
0.82
0.84
a = 30.24-2.89 7+0.081 72
3 = 0.79-0.09 7+0.029 72
70
0.74
0.82
a = -10.14-0.58 7+0.083 72
3 = 0.31 -0.13 7+0.007 72
Females
Pool
0.13
0.19
a = -242.50 + 25.49 7-0.56 72
3 = -3.48 + 0.36 7- 7.89 x 10'3 72
Males
Pool
0.33
0.55
a =1085.20-87.71 7+1.81 72
3 = 17.10-1.43 7+0.03 72
Discussion
Optimal temperature for development and upper developmental thresholds
(Tm) can be estimated from the experimental data and its inclusion in the Logan
model. Optimal temperatures ranged between 25 and 38C for all the stages and
both sexes, with values falling in the vicinity of 29C for 1st instars and 26C for 2nd
instars. Upper developmental thresholds ranged between 34 and 44C with most
values falling in the vicinity of 35C (Table 3.3). The optimal temperatures for 2nd

31
stages are lower than the typical temperatures occurring during the summer in
central Florida citrus groves (29C average of daily mean temperatures from 20
June to 22 September 1992, 37C mean of maximum temperatures during the same
period), which supports the observation that snow scale populations do not do well
during the hottest months of the year. One is also led to speculate that the
geographical areas where snow scale may have originated and thus may be best
adapted (i.e. higher latitudes or higher altitudes), experience lower mean
temperatures than Florida citrus regions.
Casares (1974) cultured citrus snow scales on fruits in a range of
temperatures between 12.5 and 29.4C, and concluded that the lowest temperature
at which the snow scale could develop was 18.3C. Insects did not develop at
12.5C, but survived and could resume development when higher temperature
conditions occurred. In our research, scale development was completed at 16C.
Thus, the lower threshold for development on grapefruit leaves must occur below
this temperature. Thresholds estimated from the modified Logan model yield a wide
range of results, including some > 16C (Table 3.3), but many of them approach
12C as observed by Casares (1974). In our studies, the extreme threshold
estimates were derived from prepupal and pupal stages (males), which were based
on reduced sample sizes caused by mortality in early instars. The estimates also
were limited in precision by the length of the stadium approximating the interval
between observations. To obtain a more precise estimate of the developmental
threshold for these stages, one must generate more data at the low end of the
temperature range, with more frequent observations.

32
Casares (1977) discussed the possible effects of humidity on citrus snow
scale populations. Casares cited observations by W.A.T Summerville in Australia
that heavier infestations of citrus snow scale occurred during dry periods than during
wet periods, concluding that humidity was an important factor in snow scale
development. In our work, the effect of relative humidity on development was not
consistently significant, but the interaction with temperature was, indicating that
combined factors have an effect (Table 3.2). Only 2 conditions of humidity, differing
by 10%, were used, because this was not a major objective of the developmental
studies. These humidities were chosen to represent average conditions occurring
in the field (range between 50 and 90% RH during 1992 and 1993). Differences in
developmental parameters would be expected from studies using relative humidities
representing a wider range and thus incorporating more marginal conditions.
The modification to the Logan nonlinear model was suitable to describe the
data presented (Table 3.3). However, the reliability of the parameter estimates was
not satisfactory for development of prepupal and pupal males because of the
imprecision previously mentioned.
It has been shown that humidity may affect the rate of development for other
species (Aonidiella aurantii, McClure [1990b]) but it may have a more important
effect on survival than on development (Atkinson 1983). This effect may be direct
in the field, altering the survival, rate of development and reproduction of the insect
(McClure 1990a) or may be indirect, altering the action of its natural enemies.
In the current studies, temperature most notably affected mortality associated
with 1st and 2nd instars and males (Table 3.4). Relative humidity effects were

33
shown to be important for survival of 1st instars and for 2nd instars at low
temperatures (Table 3.5). The diaspidid armor is an effective protection against
environmental hazards (Foldi 1990). Thus 1st instars are the most likely stage to
die as a result of unfavorable environmental conditions given that they lack this
protection.
Gamma distribution parameters calculated to fit the rate of change of the
survival curves produced figures that varied widely with temperature and relative
humidity, sometimes yielding a poor fit (e.g. females at 30C, Table 3.6). In
characterizing the relationship between the shape of the mortality rate curve and
temperature (Table 3.7), the fit obtained for 2nd-order polynomial equations was
good for 1 st and 2nd stages, but poor for older stages. The number of points used
for these regressions was small (5 temperatures for 60% RH, 4 for 70% RH, and
only 3 points for males because none survived at 16C) (Table 3.2). More extensive
experimentation would be required to more accurately describe relationships
between mortality through time and the temperatures to which the insects are
exposed.
Host-plant substrate has been well documented as being responsible for
morphological differences in diaspidid scales (Stoetzel 1976, Miller and Kostarab
1979, Cooper and Oetting 1986). Other biological characteristics such as
development, mortality and fecundity are likewise affected (McClure 1990b). The
developmental rate of citrus snow scale maintained on Duncan grapefruit leaves in
this study was much higher than that observed by Casares (1974) for scales
maintained on lemon fruits. We avoid drawing conclusions regarding these effects

34
of substrate because different citrus species are involved and effects of host species
on citrus snow scale has been documented (Reed et al. 1967). However, the low
frequency of fruit naturally infested by citrus snow scale in the field could indicate
that the substrate also has an effect on settling preference, developmental rate, or
survival of the scales.

CHAPTER 4
STUDY OF MORTALITY OF THE CITRUS SNOW SCALE
UNDER FIELD CONDITIONS, BASED ON THE ANALYSIS
OF PARTIAL LIFE TABLES
Introduction
Attempts to manage citrus snow scale by classical biological control began
during the 1970's, but the results of these efforts were never clearly documented
(Browning 1994). Snow scale is still present in citrus groves to the extent that the
use of pesticidal suppression is sometimes warranted (Knapp 1995). There is a
need to study the relationships between the citrus snow scale, its natural enemies
and other biotic and abiotic mortality factors to clarify their value as population
control factors and clarify the success or failure of previous attempts at biological
control.
Life table analysis is an important tool in the study of populations (Harcourt
1969, Varley & Gradwell 1971, Southwood 1978, Carey 1993), providing a detailed
account of the mortality affecting a cohort. A special type of table, the multiple
decrement life table, allows the separation of different factors of mortality, simplifying
the determination of their relative importance (Varley et al. 1973, Carey 1993).
This chapter reports on the mortality of the citrus snow scale under field
conditions in Florida, adapting photographic techniques for sampling of the
35

36
populations, and using life table analysis to separate and weight the importance of
the different mortality factors.
Materials and Methods
Data Recording with Photographs. Given the small size of the snow scale and
the presence of an armor, direct observations in the field are difficult. Sampling of
scale insects usually involves taking a sample of substrate to the laboratory and
dissecting it under magnification. This destructive sampling does not permit the
study of the fate of those individuals alive at the moment of sampling, that may die
later of any cause. Collection and analysis of the remains of the insects are
unreliable because it is difficult to establish the time of death, since the remains of
the insect stay in place on the substrate for variable periods. Also, some causes of
death do not leave tangible host remains. Direct counting in the field is difficult and
inaccurate.
To overcome these difficulties, I developed a method for direct observation,
recording on photographs the events occurring in a snow scale colony under field
conditions. This method was modified from Summy et al. (1984) and Terry &
Edwards (1989), and chosen because is more accurate than direct counts using
hand lenses. It allows for nondestructive sampling and the following of specific
individuals during their development.
Data were collected over a 2-year period between December 1991 and
December 1993. Active colonies were located on the trunk or main branches of
'Valencia' orange trees in a commercial grove in Lake Alfred, Polk County, Florida,

37
where no insecticides were applied during the study. Between two and four
observation areas (4x4 cm2) were selected on the bark of each tree. To obtain a
cohort of newly settled crawlers of known age, I dislodged all individuals in the area
(hereafter referred to as patches) using a hard brush (Casares 1977). A map pin
was positioned in the upper left corner of the patch as a reference marker.
Photographs of patches were taken weekly over a period that varied from eight to
20 weeks. New patches were established periodically (approximately every two
weeks) on different trees. Eighty-two patches were studied on 36 trees during the
observation period, but only 47 patches on 21 trees were colonized and therefore
were used for the life table analysis.
A 35-mm camera (Canon AE-1, Canon Inc. Tokyo, Japan) equipped with a
macro lens (Canon FD 50 mm f/3.5 Macro) mounted on bellows allowed image
enlargements of 2X. The lighting consisted of two flashes, one mounted on a
bracket beside the camera level with the lens and the other hand-held on the
opposite side of the lens and equipped with a photosensitive trigger. The camera
and flash were mounted on a sturdy tripod. High resolution color transparency film
(Kodak Ektachrome 64 and Fujichrome 50) was exposed at 1/60 sec. and f/22
aperture and was processed through commercial laboratories.
Photographic Analysis. The sequences of insect development were studied
while viewed on a slide projector with a built-in 9x9-inch screen (Kodak Ektagraphic
AudioViewer/Projector model 260, Eastman Kodak Co., Rochester, NY 14650). The
total area covered by the field of the camera was 18x12 mm2 of trunk surface. The
area chosen for analysis was equivalent to 14x9 mm2 in the center of the field.

38
Individual insects were located and marked on the screen with the help of a clear
plastic grid (1x1 cm2 divisions ). Approximately 20 newly settled individuals were
located and marked from the first scene of each sequence, and followed through the
set of slides until their death or to the end of the sequence.
The causes of death were classified in the following manner:
1- Competition: When individuals were dislodged by the growth of a neighbor
scale cover; included in this category were also individuals that were covered
by algae, this was observed in only one instance.
2- Parasitism: Individuals that showed emergence holes. Parasitized male
scales exhibited these holes from the time they started to secrete the white
armor until attaining their full length. Holes on males were attributed to
parasitism by Encarsia spp. (Hymenoptera: Aphelinidae). Females were
attacked by Encarsia spp. or by Aphytis lingnanensis. Encarsia parasitism
was evident early in the second instar, and emergence holes appeared
before the second molt. Emergence of Aphytis, on the other hand, occurred
on young third instar females, with the emergence holes being larger and
more rounded than those produced by Encarsia spp. All of these parasites
have been observed emerging from non-preferred host stages and sexes
(Aphytis from males and gravid females, Encarsia spp from adult females
[Browning 1994]) Those atypical occurrences were not observed during the
present study.
3- Visible predation: Assigned to those cases in which the armor was clearly
damaged but some remains persisted on the site. This damage was

39
attributed to predators that could chew or break through the armor, such as
coccinellids. It is impossible to measure total predation. Predation can be
underestimated, particularly when the scale armor or body of early nymphs
is totally removed. On the other hand, overestimation of predation may occur
when the disturbed armor did not contain live prey, or if the damage inflicted
to the cover is due to scavengers. Encarsia-parasitized female scales
attacked by predators before the parasite emerged also were observed.
However, these were classified as death by parasitism since they would not
survive. It is possible that the same phenomenon occurred with males, but
it was impossible to observe.
4- Fungi: White mycelial growth was observed in several instances growing
on individuals that had died. Unfortunately, it was impossible to determine
if the insect was killed by the fungus or was colonized after dying by another
cause. The most common fungi in snow scale colonies are from the genus
Nectria (teleomorph of Fusarium) but their role is unclear. There is confusion
over whether those fungi are acting as pathogens or as saprophytes (Ziegler
1949). Kuno & Coln-Ferrer (1973) showed increased crawler mortality but
saprophytic behavior on older stages. As with predation, death by diseases
is being underestimated. Deaths that occurred under the scale cover and did
not show mycelium or fructiferous bodies within the observation period were
unaccounted. Given the impossibility to clear the issue of pathogenicity, we
assumed any fungal appearance to be pathogenic and causing death to the
scale with which it was associated.

40
5- Unknown causes: Mortality causes that could not be separated or were
dubious or not clearly attributed to a given factor were grouped under this
category. Several outcomes are described here:
a) Collapsed: A few male armors were observed to shrink or become
deformed. This collapse could be attributed to the action of predators
other than coccinellids, e.g., thrips, neuropteran, mites, that can feed
on the scale insect without disturbing the armor.
b) Failed development: Assigned when an individual scale stopped its
development at an immature stage, with no outward signs of
disturbance. This failure of development could be attributed to
physiological or pathological causes, mainly for first instars, in which
a change in coloration occurred as they dried out; to predation by
organisms that act without disturbing the armor, such as predatory
mites, thrips or lacewing larvae; or host-feeding by aphelinid
parasitoids.
c) Lost: Assigned when an individual scale disappeared without
leaving a clue about its fate. Occurred frequently with first instars or
with developing males. Could be the action of predators, or
mechanical factors such as rain or wind.
7- Indefinite outcome, covered by others: Used in the cases when the patch
was overcrowded and other scales grew over the observed individuals.
These were excluded from the analysis.

41
8- Completed life cycle: Individuals that reached the adult size and
appearance and were not victims of any of the previous mortality causes
were classified as completing their life cycle. The female reproductive period
was not isolated for analysis since it occurs beneath the armor, making it
impossible to observe.
Statistical Analysis. Life table analyses were used to estimate mortality
trends in populations of citrus snow scale on chosen patches. Single decrement life
tables were generated (PROC LIFETEST, SAS Institute Inc. 1989). Comparisons
were made between patches in each tree and between trees in the whole sample
site. Multiple decrement life tables were built using the mortality categories listed
and following the procedures and notation explained in Carey (1993):
K*=
Dix=
D =
aqix= Dfc/K,
aqx= DJK
a\ = alx-1(1 -aqx)
Number of individuals beginning stage x
Number of individuals dying of cause / at stage x
Total deaths in stage x
Fraction of deaths from cause /' in stage x in the
presence of all other causes, given that the individual is
alive at beginning of stage x
Fraction of deaths from all causes in stage x, given that
the individual is alive at beginning of stage x (£aqix)
Fraction of survivors at age x out of the original cohort,
adix= alx(aqix)
which is assigned al^l
Fraction of deaths in stage x from cause / among alx
living at stage x

42
adx= alx-alx+1 Fraction of deaths in stage x from all causes (£adix)
Since it is not possible to separate sexes for the first stage of development
and since no attempt to build fecundity tables was made, the life tables include both
sexes. The stages considered in the construction of the mortality tables were:
1- First instar, from crawler settling to the onset of first molt.
2- Second instar, before production of secretion, which is from the first molt
until the cover secretion begins. Sexes are not distinguishable to this point.
3- Secretion, period during the 2nd instar during which the cover is built;
separate calculations were made for each sex. Males pass through two
more stages before adult (two molts) but these were unrecognizable because
they occurred beneath the armor. Females molt only once more.
4- Third instar (adult) females, started when females passed the second molt
and began depositing the last section of the armor.
The periods of insect development beyond this point to reproduction, and
between 1st instar eclosin and settlement were excluded from the study, thus the
mortality tables are based on only the part of the life cycle of the insect. A more
detailed description of the life cycle was presented in Chapter 2.
Results
Single decrement life tables compared within trees by the log rank test,
Wilcoxon test of homogeneity and likelihood ratio test showed significant differences
for the patches in six of 17 trees. Four trees were not tested since each had a

43
single patch. The tree location, date in which the patches were initiated, date until
they were observed and significance are shown in table 4.1. It is important to note
that the sets of patches that showed significant differences were all initiated during
spring and photographed into summer of both years.
Despite the differences observed between patches within some trees, the
data from all of the patches in each tree were pooled and used to compare between
trees and starting dates. The same statistical tests showed significant differences
(Log-Rank x2 =91.40, Wilcoxon x2 =87.65, P< 0.0001 for both tests, 20 d.f., tree as
strata, PROC LIFETEST, SAS Institute Inc. 1989) suggesting differential survival at
different times of the year or different tree locations. A comparison between trees
at each date on which a set of patches was started was possible for 6 dates (Table
4.2). Four of the 6 dates showed significant differences, suggesting that there is a
site (tree) component contributing to the variation. The survival curves drawn from
these data were diagonal or slightly convex, with the slope becoming steeper during
the late spring and summer months, periods when U. citri development was also
shortened by increased temperatures. Figure 4.1 shows survival and hazard
functions (age-specific probability of death) from selected trees in winter, spring,
summer and fall periods. Mortality was initially low and increased slowly and
steadily through the scale lifetime but variation in mortality patterns was high. A
summary of the fate of the 960 individual scales used in the analysis from a total of
1082 observed is presented in table 4.3. Only 13% of the observed individuals
appeared to complete their life cycle. The overall observed sex ratio was 4.7
males/female.

44
Table 4.1. Significance of log-rank test, Wilcoxon homogeneity test, likelihood ratio
test for comparison of survival of citrus snow scale in patches within trees.
Tree
# of
Patches
Date started
Date ended
Significance
36-2
3
17/Dec/1991
27/Apr/1992
ns
36-7
2
17/Dec/1991
27/Apr/1992
ns
35-15
2
23/Jan/1992
19/May/1992
ns
35-18
2
06/Mar/1992
14/Jul/l 992
ns
34-21
5
14/Mar/1992
26/May/1992
ns
35-17
4
27/Apr/1992
18/Aug/1992
All tests
34-19
3
19/May/1992
24/Sept/1992
All tests
32-15
2
17/Jun/1992
16/Sept/1992
All tests
34-11
1
29/Jul/1992
24/Jan/1993

26-9
2
24/Aug/1992
15/Jul/1993
ns
28-7
2
24/Aug/1992
23/Jun/1993
ns
26-16
2
12/Oct/1992
14/May/1993
ns
27-14
2
26/Oct/1992
30/Apr/1993
ns
28-14
1
26/Oct/1992
8/Apr/1993

24-2
2
30/Apr/1993
15/Jul/l 993
All tests
23-13
3
24/May/1993
1/Dec/1993
Log rank P <0.025
24-12
2
24/May/1993
1/Oct/1993
Log rank P <0.031
22-9
3
17/Aug/1993
16/NOV/1993
ns
25-6
3
16/Sept/1993
14/Apr/1993
ns
21-12
1
6/Oct/1993
1/Dec/1993

22-14
1
6/Oct/1993
1/Dec/1993


45
Table 4.2. Significance of log-rank test, Wilcoxon homogeneity test, likelihood ratio
test for comparison of survival of citrus snow scale between trees by dates.
Date started
# of trees
Significance
17/Dec/1991
2
Wilcoxon P < 0.047
06/Mar/1992
2
Log-Rank P < 0.040
24/Aug/1992
2
Log-Rank P < 0.008
26/Oct/1992
2
ns
24/May/1993
2
ns
6/Oct/1993
2
Log-Rank P < 0.024
Wilcoxon P < 0.016
Survival ranged between 0 (Jun., Aug., and Oct. 1992) and 30.5% (Aug.
1992) in the 21 mortality tables constructed. The main mortality was classified in the
"unknown causes" category, but predation and parasitism provided an important
contribution to the total (Table 4.4). Between 0 (Jan 1992, Oct. 1993) and 48.8%
(Sept 1993) of the initial cohort died by the action of predators, between 0 (Jun., Jul.
1992, Apr., May, Aug., Oct. 1993) and 38.3% (Aug. 1992) died by the action of
parasitoids. It is important to mention that most of the parasitism observed was
attributed to Encarsia, only one female out of 13 observed parasitized was killed by
Aphytis. Unknown causes accounted for 29.3% (Sept 1993) to 90% (Oct. 1993).
Table 4.5 shows a summary of the total mortality (all the causes) presented
by stage. The highest mortality usually occurred on armor-secreting males.
However, in three cases, mortality was the highest in the second stage prior to scale
secretion (Dec. 1991, Jun. 1992, May 1993); in one case mortality was greatest on

Hazard Survival
Tree 36-2, 17/Dec/1991 Tree 35-17, 27/Apr/1992 Tree 32-15,17/Jun/1992 Tree 26-16,12/Oct/1992
Tree 23-13, 24/May/1993 Tree 22-9, 17/Aug/1993 Tree 25-6, 16/Sept/1993
First molt
Secretion
Female 2nd molt
Full size male
Full size female
0 20 40 60 80 100 120 20 40 60 80 100 120 20 40 60 80 100 120
Age (days)
Figure 4.1 Survival and hazard functions of snow scale population on selected trees.
Vertical lines represent time of molts.

47
Table 4.3. Outcome observed for citrus snow scale individuals under field
conditions. Summary, data pooled for all patches and dates.
Outcome
1st
instar
2nd
instar
Secreting
3rd
instar
female
Total
Male
Female
Completed
N/A
N/A
118
N/A
10
128
Detectable predation
0
1
119
6
16
142
Parasitism
0
0
95
8
5
108
Competition
1
7
0
0
0
8
Fungi
0
3
1
1
1
6
Unknown causes
103
244
158
35
24
568
Lost
51
78
37
6
5
177
Failed to develop
52
166
125
29
19
391
Total
104
255
495
50
56
960
adult (third instar) females (Jul. 1993), and finally, in two cases, first instars suffered
the highest levels of mortality (Aug. and Oct. 1993).
Discussion
Survival curves of citrus snow scale populations showed significant
differences between patches on six trees. These patches shared a feature in being
initiated between May and mid-June of each year, when temperatures were
increasing. The temperature increase is responsible for a shortened life cycle and
may increase activity of natural enemies that could have affected variability within
these six trees.

48
Table 4.4. Total percentage of mortality (adx) and percent mortality by unknown
causes against parasitism and detectable predation in multiple decrement life
tables, calculated by tree.
Tree
Date
adx
Detectable
predation
Parasitism
Unknown
causes
36-2
17-Dec-91
0.7939
0.0153
0.0763
0.6718
36-7
17-Dec-91
0.8387
0.0968
0.1452
0.5968
35-15
23-Jan-92
0.8333
0.0000
0.2708
0.5000
35-18
06-Mar-92
0.9355
0.2903
0.1935
0.4516
34-21
14-Mar-92
0.9775
0.1910
0.2022
0.5730
35-17
27-Apr-92
0.8106
0.0303
0.1212
0.6591
34-19
19-May-92
0.9464
0.2857
0.0179
0.6429
32-15
17-Jun-92
1.0000
0.1707
0.0000
0.8293
34-11
29-Jul-92
0.8000
0.3000
0.0000
0.5000
26-9
24-Aug-92
1.0000
0.2647
0.3824
0.3529
28-7
24-Aug-92
0.7250
0.2250
0.0500
0.4500
26-16
12-Oct-92
1.0000
0.0645
0.2903
0.6452
27-14
26-Oct-92
0.8333
0.1250
0.0833
0.6250
28-14
26-Oct-92
0.9412
0.2353
0.1765
0.5294
24-2
30-Apr-93
0.6970
0.3333
0.0000
0.3636
23-13
24-May-93
0.8868
0.2453
0.0000
0.6415
24-12
24-May-93
0.7576
0.0909
0.0000
0.5152
22-9
17-Aug-93
0.9231
0.1154
0.0000
0.7692
25-6
16-Sep-93
0.9024
0.4878
0.1220
0.2927
21-12
06-Oct-93
0.9500
0.0500
0.0000
0.9000
22-14
06-Oct-93
0.7500
0.0000
0.1250
0.6250

49
Table 4.5. Total percent mortality (all causes) adx presented by stage.
Tree
Date
Started
1st instar
2nd instar
Secreting
3rd instar
Female
Male
Female
36-2
17-Dec-91
0.1374
0.4656
0.1756
0.0076
0.0076
36-7
17-Dec-91
0.0806
0.2742
0.4194
0.0645
0.0000
35-15
23-Jan-92
0.1667
0.3125
0.3542
0.0000
0.0000
35-18
06-Mar-92
0.0323
0.1290
0.7097
0.0323
0.0323
34-21
14-Mar-92
0.0225
0.2360
0.5955
0.1124
0.0112
35-17
27-Apr-92
0.1818
0.2045
0.3485
0.0379
0.0379
34-19
19-May-92
0.0357
0.3214
0.5357
0.0179
0.0357
32-15
17-Jun-92
0.1220
0.4146
0.3659
0.0732
0.0244
34-11
29-Jul-92
0.0000
0.1000
0.2000
0.1000
0.4000
26-9
24-Aug-92
0.0588
0.2353
0.5000
0.0588
0.1471
28-7
24-Aug-92
0.0500
0.2000
0.2250
0.0500
0.2000
26-16
12-Oct-92
0.0968
0.3871
0.5161
0.0000
0.0000
27-14
26-Oct-92
0.0417
0.2500
0.4583
0.0833
0.0000
28-14
26-Oct-92
0.1176
0.1765
0.5882
0.0000
0.0588
24-2
30-Apr-93
0.0303
0.2727
0.3636
0.0303
0.0000
23-13
24-May-93
0.0755
0.1509
0.4340
0.0943
0.1321
24-12
24-May-93
0.0909
0.3333
0.1818
0.0303
0.1212
22-9
17-Aug-93
0.4231
0.0769
0.3077
0.1154
0.0000
25-6
16-Sep-93
0.0488
0.0732
0.512
0.1220
0.1463
21-12
06-Oct-93
0.4000
0.2000
0.3500
0.0000
0.0000
22-14
06-Oct-93
0.0000
0.0000
0.3750
0.3750
0.0000
Significant differences were detected between some patches in different
trees, and between trees where patches were set at the same dates. This could be
due to the different times of the year in which the patches were initiated (Table 4.1),
and also could be due to different locations of the trees inside the grove (Table 4.2),
or a combination of both. Given that the patches were set on different dates through

50
the year, the insects were exposed to different weather conditions. Since
development depends on temperature (Chapter 3), scale insects have shorter life
cycles during the warmest months (Figure 4.1). It is also expected that extreme hot
temperatures increase mortality, but this is not obvious from the data presented in
Figure 4.1. It is also assumed that natural enemies have shortened life cycles at
higher temperatures.
Overall, within all of the cohorts studied, (Table 4.2) 24% of males (118 out
of 495) and 18% of females (10 out of 56) survived to the end of the observation
periods. Most of the mortality observed could not be assigned to a specific cause.
This unknown mortality affected 59% of the individuals in the cohorts (568 out of
960) and accounted for 68% of the total mortality. Natural enemies were an
important component; detectable predators killed 15% of the individuals, while
parasitoids killed 11%, representing 17% and 13% of the total causes of mortality,
respectively.
The extent of variability in the data can be observed in table 4.3. Despite this
variability, total mortality (adx) never fell below 70%, and varied up to 100%.
Detectable predation appeared more important than parasitism (0-49% against 0-
38%) and males demonstrated higher probabilities of being attacked by both factors
than females (Table 4.2). Secreting males and second instars of both sexes were
the stages with highest mortality (Table 4.5), perhaps because the higher proportion
of males over females means higher availability of male prey or hosts. Males could
also be more susceptible given their softer shield. Males and females are available
for attack by parasitoids for a similar period. Even though males have a shorter life

51
cycle, females are not available for parasitism after they reach reproductive status
(Browning 1994). On the other hand, reproductive females will remain targets for
predators longer than males. Undoubtedly, predation is underestimated, and likely
an important fraction of the observed unknown mortality is predation. In the present
research, attempts to trap predators failed but chrysopid larvae (Neuroptera) were
observed, collected, and reared from citrus snow scale colonies.
Parasitism occurred through most of the study period, but was lacking from
our patches in June and July, 1992 and April to August, 1993. Aphytis lingnanensis
has been present in Florida for some time. A race of this species collected in Hong
Kong and named HK-1 was released in Florida in the early 1970's (Browning 1994).
This species was found in leaves and twig samples in the grove where this study
was performed, but the low frequency in which parasitism by Aphytis was observed
in the photographic analysis suggests that Aphytis is not a major factor in the
regulation of citrus snow scale populations on trunks and therefore it is not as
successful a biological control agent as has been reported (Fisher 1985). The
contributions of parasitism and detectable predation appeared not to be related as
they seemed to fluctuate independently between patches (Table 4.4). High
incidence of one factor did not accompany low incidence of the other. Apparently,
they are not complementary in this sense. In fact, predators may be a competing
factor with the survival of parasitoids that attack early stages of citrus snow scale.
Contrary to what was expected, diseases were observed in a very small
proportion of scales examined, despite the fact that reproductive bodies of the
fungus were observed on snow scale colonies on the same trees where the patches

52
were photographed. It was anticipated that the crowded conditions in citrus snow
scale colonies would favor the spread of diseases. It is very likely that a portion of
the unknown mortality is composed of individuals killed by diseases. Potential
explanations for not observing fungi in our study are that the period between
infection and the development of evident fructiferous bodies could be longer than
the period of our observations (8-20 weeks), or that the disturbance created when
cleaning the patches affected somehow the incidence of fungi, reducing the
crowdedness of the insect colony or the age of abandoned shields, or that the fungi
are nothing more than saprophytes and would not develop on live insects. Nectria
spp. have been mentioned as important mortality factors for scales in citrus (Fisher
et al. 1949) and are found frequently in snow scale colonies, but they as well could
be saprophytes that thrive on the abandoned covers of deceased individuals or
emerged males (Ziegler 1949; Fisher 1950). Kuno & Coln-Ferrer (1973) studied
the pathogenicity of two species of Fusarium over scale insects. They found that
crawler mortality could be increased by the fungi, but there was not effect on older
instars, and the fungi showed saprophytic growth.
In general, mortality of snow scale in the observed patches was very high
(between 70% and 100%, Table 4.4). A substantial percentage of mortality
measured in the various patches was due to unknown causes. Very likely, a part
of it can be attributed to natural enemies (parasitoids, predators and diseases), but
there is no way to separate it using the current data. The observed effects of natural
enemies were considerable and could account for about a quarter of the total
mortality. However, this mortality was insufficient to reduce the densities of snow

53
scale to rare occurrence in the site studied. The white covers of the snow scale
males are very evident against the dark background of the citrus bark, and the
covers abandoned by males that already emerged remain on site for an
undetermined but presumably long period. Despite the observed high mortalities,
the accumulation of male scale covers may give the impression that densities of live
citrus snow scales are greater than what they really are. This suggests that the
snow scale problem is in part a problem of perception by the grower. It also
identifies the need for careful observation to determine the status of live scales in
the field when management options are considered.

CHAPTER 5
COMPARATIVE STRUCTURE OF CITRUS SNOW SCALE
POPULATIONS ON THE CANOPY OF CITRUS TREES
AND THROUGH TIME.
Introduction
A good knowledge of the biology and ecology of a pest is necessary to design
optimal pest control programs. Periodic sampling of the population produces useful
information concerning age structure, sex ratios, and mortality factors. Difficulties
arise when measurement of absolute density of an organism is needed, since it
requires an estimation of the area sampled. Alternatively, population intensity
(number of individuals per unit of habitat, e.g. leaf) and relative estimates are
available (Southwood 1978). The percentage of a population subjected to
parasitism is a traditional measurement of the effect of these natural enemies, but
this measurement has been criticized based on difficulties in obtaining accurate
information on the real impact of those parasitoids (van Driesche 1983).
Species that colonize different organs of a plant have been shown to exhibit
different traits, such as differences in developmental rate, sex ratio, and even
morphology (Chapter 2). In the same way, the association with different host plant
organs exposes the insects to different environmental influences. Variable host
nutrient composition may occur between leaves, fruits and trunks. Natural enemies
54

55
may also react to different substrata in different ways, such as exhibiting preference
for searching on some surfaces over others (Carroll 1979, Murdoch et al. 1989).
The phenology of the citrus snow scale has been previously investigated.
Methods included visual counts on trunk, leaves and twig samples that produced
population intensity estimates (Casares 1977). Attempts at biological control against
this scale were made in 1974 through introduction of parasitoids to Florida. The
aphelinid Aphytis lingnanensis "HK-1" was introduced and released. Successful
biological control was reported but citrus snow scale persists as a problem in some
citrus groves in Florida (Browning 1994). The study by Casares (1977) preceded
the introduction of A. lingnanensis "HK-1" and aimed to evaluate the impact and
importance of previously established parasitoids. However, the results were not
presented in a comparative form and failed to consider an important component of
the parasitoid fauna that subsists on nymphal males of the citrus snow scale.
In the present work, population samples of the snow scale were studied with
the use of relative population estimates, which are easier to contrast. The
objectives were to assess the effects and importance of natural enemies, especially
parasitoids, and how populations of snow scale change on the different substrata
and through time.
Materials and Methods
Samples of twigs and leaves were collected monthly from three to five snow
scale-infested citrus trees in a commercial grove in Lake Alfred, Polk County,
Florida. Between five and 12 snow scale infested twigs with leaves were chosen

56
from the interior of the canopy. These samples were processed in the laboratory
and up to 100 live individual snow scales were counted from leaves and 100 more
from twigs. The information recorded included sex, stage, parasitism, presence of
fungi, and presence of mites. Two types of parasitoids were observed, the
endoparasitoids, Encarsia species (Hymenoptera: Aphelinidae) and the
ectoparasitoid Aphytis lingnanensis (Hym.: Aphelinidae). Encarsia spp. are
endoparasitoids of both sexes, ovipositing into the second stage. Parasitism of male
scales is evident because the dark parasitoid pupa is visible when male scale covers
are lifted. In females, parasitism by Encarsia becomes visible through changes of
body coloration of second instar female scales while the parasitoid is in its larval
stages. Late parasitoid larvae and pupae are easily visible through the scale cover
and body. Two species are likely to occur in the study area, E. citrina (Craw.) and
E. longsburyi (Berlese & Paoli) (Browning 1994). On the other hand, Aphytis
lingnanensis Compere is an external parasitoid that attacks young third stage
females (pre-reproductive adults). Parasitoid larvae and pupae are very evident on
their hosts, but eggs are more difficult to detect.
The fungi observed developing on citrus snow scale species were Nectria
species (anamorph Fusarium), exhibiting white mycelia and orange fructiferous
bodies. An unidentified species of mite also was observed feeding on the bodies
of adult females. These mites were observed inserting their mouthparts into the
scale integument. Citrus snow scale deaths by predators were not quantified, since
there was no way to determine time of death or how long a dead individual would

57
have remained on the substratum. Counting them would have overestimated the
effect of predation.
The sampling took place between May 1992 and December 1993, comprising
14 sampling dates. The period between sampling dates was sufficient to allow the
emergence of parasitoids in immature stages on the previous sampling date and
also to avoid duplicate sampling of the same snow scale cohort.
The proportions of individuals of each age (stage) group were calculated,
adding the numbers of healthy individuals and the numbers of parasitized individuals
that may have belonged to the same cohort. Internally-parasitized females were
classified as second instars, while externally parasitized females were classified with
gravid females, because they were attacked as prereproductive females and they
are part of the gravid-female cohort. Sex ratio was calculated by adding the number
of healthy and parasitized individuals, i.e., total healthy males plus males bearing
internal parasitoids, healthy second instars females plus prereproductive adult
females plus females bearing any parasitoids. Reproducing females were not
added because the older males of equivalent age had emerged and thus were not
accounted for in sex ratio calculations. The incidence of parasitoids and fungi was
calculated as a percentage of parasitism based on the total individuals of a given
sex and stage.
The statistical analysis involved distribution comparison of the percentage
age structure between substrata using the Kolmogorov-Smirnov two sample test
and Kruskall-Wallis non parametric multiple comparisons for detecting differences
between sampling dates (Siegel 1956, SAS Institute Inc. 1989).

58
Results
Figure 5.1 shows the age structure of the population sampled on each date
according to the substratum. First instars of undetermined sex and females form the
age structure pyramid, whereas males are shown as a bar above the pyramid. The
total number of scales in the sample is included. A Kolmogorov-Smirnov two
sample distribution test did not show significant differences between substrata, and
the Kruskall-Wallis test did not show significant differences between dates.
However, the population showed a tendency toward younger stages on leaf samples
from May 1992 to June 1993 and September 1993, but reversed on August and
October through December of 1993. Age distribution was more homogeneous on
twig samples, showing higher proportions of prereproductive females in August and
September 1992 and again in August 1993. Higher proportions of first instars
occurred during May, June and September 1993. The total number of individuals
counted on each sampling date is an indication of the abundance of snow scale at
the time of sampling. Low populations were found in November of 1992, and in
August, September, November and December of 1993. No live individuals were
found on the twig samples from November 1992. These trends should not be taken
as conclusive results, given the lack of statistical significance already mentioned.
The ratios of newly bom individuals (1st stage nymphs) per reproductive
individual (gravid females) are presented in Figure 5.2. These values were
consistently higher on leaves, except on the last sampling date in December 1993.
High ratios were recorded in June, 1992 and March, May, and September, 1993.

males -
Gravid females -
Virgin females -
Nymph II fern. -
Nymph I ^
1.0
males -
Gravid females -
Virgin females -
Nymph II fern. -
Nymph I -
1.0 0.5
0.0 0.5 1.0 0.5
0.0 0.5 1.0 0.5
0.0 0.5 1.0
Mar 1993
May 1993
June 1993
males
ru
;
- 152
363 397
129 236
436
Gravid females -
Virgin females
Nymph II fern. -
Nymph I -
A
A
A.
1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0
males -
Gravid females -
Virgin females -
Nymph II fern. -
Nymph I -
1.0
males -
Gravid females -
Virgin females -
Nymph II fern. -
Nymph I
1.0
0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0
Nov 1993 Dec 1993
0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0
Figure 5.1 Age structure of citrus snow scale populations on citrus leaves and
twigs. Numbers on the side of each pyramid represent the total number
of individuals on that sample.

# of 1rst nymphs per gravid female
60
Figure 5.2. Ratio of snow scale first instars per gravid females on samples
of citrus twigs and leaves.

61
The missing points correspond to samples where no gravid females were found
(except Nov., 1992 where no live scales were was found on twigs).
The sex ratios on twigs and leaves are presented in Figure 5.3. In contrast
to previous measurements reported in the literature (Casares 1974), observations
made in lab cultures (Chapter 3) and on tree trunks in the field (Chapter 4), the sex
ratio was close to 1.8 females per male on leaves (average 36% males), with
variation ranging from 14-49% males. The proportion of females on twigs was
greater (2.2 females: 1 male, average 0.31 males), with values ranging 0-53%
males. A greater proportion of males was observed in the late spring-early summer
months on twigs and leaves in the first year and earlier on leaves during the second
year (1993).
The incidence of parasitoids and fungi followed similar patterns on leaves and
twigs, especially those associated with female scales (Figure 5.4). Fungi and mites
occurred only sporadically. High incidence of internal parasitism occurred in
October 1992 and late in 1993 (October-November). Encarsia spp. were abundant
on males over most of the observation period, with drops in July 1992 and May to
August 1993. The incidence of ectoparasitoids (Aphytis) was usually higher on
leaves with peaks in April 1992 and September to December 1993 but some of this
increase represents low numbers of host stages encountered.
The incidence of hymenopterous parasitoids, fungi and mites are presented
in Figure 5.5, as relative proportions calculated from the totals of attacked
individuals observed at each sampling date. Internal parasitoids are predominant
except on August 1993 and on October 1993 (twigs). No internal parasitoid attacks

62
Leaves
Twigs
Figure 5.3. Proportion of males and females on samples of citrus leaves and
twigs.

Percentage of incidence
63
80
60
40
20
0
80
-C External parasitism
in females
Figure 5.4. Incidence of natural enemies of snow scale on samples of leaves
and twigs. Percentage are calculated over the total of susceptible stages
present in the population.

64
Leaves
Twigs
Leaves
Twigs
May 1992 Jun 1992 Aug 1992 Sep 1992

8
Nov 1992 Jan 1993 Mar 1993 May 1993
O
Leaves
Twigs
Jun 1993
Aug 1993
6_
Sep 1993
0
Oct 1993
Leaves
Twigs
Nov 1993 Dec 1993
Figure 5.5. Relative importance of natural enemies of snow scale females on
samples of leaves and twigs. Percentages are calculated over the total
attacked females.

65
occurred in September 1993 on twigs. Fungi were predominant in August (leaves,
13 of 24 individuals) and October 1993 (twigs, 7 of 11 individuals). Mites were
observed in September, 1992 and January, June and October 1993.
Mites were found living under the bodies of live adult reproductive females
with their mouthparts inserted into the insect body, apparently sucking fluids. The
scale continued to oviposit. Unlike the other natural enemies, these mites were not
preventing reproduction, but very likely they were producing some deleterious effect.
Discussion
The predominance of early instars in the age structure of a population
suggests high natality in the population, or high mortality of older stages (Odum
1971). In the case of scale insects, high immigration of crawlers is also possible,
since the first stage is the only one able to move. A high proportion of young instars
were observed on leaves over twigs during most of 1992 and 1993, while
predominance of older stages occurred in October to December 1993. This pattern
roughly follows the changes in air temperature for those times of the year. As
shown in chapter 3, development of citrus snow scale is dependent on temperature.
The high ratio of newly settled first instars to reproductive females (Figure
5.2) could be the result of higher natality on leaves, but more probably it is explained
by movement of crawlers from the twigs and limbs onto the leaves, or a combination
of both. Carroll (1979) observed differences in substrate preference by California
red scale Aonidiella aurantii, a species that can colonize most of the organs of a
citrus tree. He explained the substrate preference based on phenological and

66
topological characteristics and on differential mortality. Since it is known that
different substrata influence development and survival (see chapter 3 and Miller &
Kostarab 1979, Cooper & Oetting 1986), the possibility of increased citrus snow
scale natality cannot be dismissed.
Mortality factors also may be operating differently between the various
substrata. Murdoch et al. (1989) demonstrated that rates of parasitism by Aphytis
melinus and Encarsia sp. on California red scale (Aonidiella aurantii) were higher on
the periphery of the canopy than on the trunk of citrus trees. They suspected that
lower attractiveness of the bark surface to parasitoids may play a role in this
phenomenon. Carroll (1979) also observed this phenomenon with Comperiella on
A. aurantii. The citrus snow scale system, while sharing many features, is not the
same as the one observed by Murdoch et al. Citrus snow scale samples in the
present studies came from the interior of the tree, near the trunk and main scaffold
limbs, and thus our comparison is between the interior leaves and supporting twigs.
However, the principle demonstrated by Murdoch et al. may apply equally well in the
sense that different substrata may have an effect on the searching and dispersal
capabilities of natural enemies.
Snow scale sex ratios have been previously reported as male biased
(Casares 1974). Our studies showed a male-based sex ratio in laboratory cultures
(Chapter 3) and in colonies on citrus trunks in the field (Chapter 4, Table 4.3).
However, the data from twigs and leaves in this study indicate a female biased sex
ratio in the canopy. The factors discussed earlier that could influence natality,

67
mortality and movement of citrus snow scale populations on twigs and leaves could
have contributed to the observed reverse trend in sex ratio on the canopy.
The important parasitoids affecting mortality of snow scale in this study were
Aphytis lingnanensis and Encarsia species. While Encarsia species attacked both
sexes, A. lingnanensis attacked females exclusively. Parasitism by Encarsia was
higher on females (Figure 5.3A, 5.3B), favoring a male biased sex ratio. Since
females are subject to increased pressure from these two sources of mortality,
parasitism likely would push the sex ratio bias toward males. This was not the
situation observed on leaves and twigs. Differences in dispersal behavior between
sexes have been reported for other species of scales (Arias-Revern 1988) and for
snow scale (Xinnian & Browning 1991). Female crawlers generally followed a
straighter course than males and were more active, reaching farther during this
important dispersal phase. Since females feed for a longer period (Chapter 3) and
likely require more nutrients for development and reproduction, differential dispersal
behavior may be an adaptation to reduce intraspecific competition and to expand
colonization into more remote locations. Sexual dimorphism in the adult form in
scale insects provides a mechanism for winged males to disperse in search of
female receptive to mating, countering the limitations on male crawler dispersal and
reducing the impacts of inbreeding.
In a study of citrus snow scale prior to the introduction of A. lingnanensis "HK-
1", the incidence of parasitism by Encarsia longsburyi was presented as number of
parasitized hosts per unit of habitat (leaf) and per unit of area (cm2 of twig) (Casares
1977). Relative proportions of the population impacted by Encarsia parasitoids are

68
not apparent through these data, and comparison between substrata is impossible.
Inspection of Casares's (1977) figures indicated estimates of parasitism in females
might be 25% to 55% per leaf and very low (6-10%) per unit of twig surface.
Unfortunately, the figures pooled unparasitized 2nd and 3rd female stages, further
obscuring the incidence of Encarsia parasitism. Nevertheless, Casares concluded
that in addition to its lack of host specificity, Encarsia was not able to control snow
scale because its searches for host on leaves rather than twigs.
In the present investigation, comparable percentages of internal parasitism
by Encarsia occurred between leaves and twigs (Figure 5.3A, 5.3B). However
differences were visible in ectoparasitism by A. lingnanensis, with a predominance
of parasitism on leaves (Figure 5.3C). Results previously reported in Chapter 4
point to the minor contribution of A. lingnanensis to the mortality of citrus snow scale
in colonies resident on trunks. These observations support the notion that
individuals on leaves are more heavily attacked and suffer higher mortality, and
agree with the observations of Carroll (1979) and Murdoch et al. (1989) on California
red scale.
Since this increased mortality is occurring on females, which is also the most
abundant sex, it would be likely that in the absence of A. lingnanensis the
populations of citrus snow scale might increase on leaves. Percent parasitism
measurements should be taken with caution since they can be calculated only from
the observed dead and live individuals, excluding individuals killed by other causes
that otherwise would have been part of the live or parasitized fractions (Carey 1993).
It should be kept in mind that we may be overlooking other potentially important

69
components of mortality in predation and in host feeding by aphelinid parasitoids
(Encarsia and Aphytis).
The relative importance of mortality factors observed is highly variable
through the year and on the different substrata (Figure 5.5). Endoparasitism was
predominant most of the time and only on two occasions was overshadowed by
other entities, namely incidence of fungi. In one case (August, 1993 on twigs)
ectoparasitoids were more frequent than endoparasitoids. A potential source of bias
is small samples and aggregation of susceptible hosts that may have facilitated
parasitoid oviposition on several hosts in the same sampled twig.. I have no way
to determine if the observations of fungi represent the onset of a disease or the
indication of a saprophyte, but in any case, we should not disregard the fungi
potential to inflict mortality on the scales without solid proof.
It appears that A. lingnanensis and Encarsia spp. may be important factors
in regulating populations of citrus snow scale in the canopy of citrus trees. However,
given the fact that the majority of the scale population occurs on the trunk and
scaffold limbs, and that low incidence of A. lingnanensis was observed on these
substrata (Chapter 4), it could be concluded that these parasitoids are contributing
little as biological control agents of citrus snow scale at the studied location.

CHAPTER 6
THE ROLE OF WIND IN CRAWLER DISPERSION
Introduction
Diaspidid scale insects are sessile organisms with only two very limited motile
stages, adult males and newborn first instars (crawlers). Adult males can fly, but by
themselves they cannot establish new colonies. Crawlers are very fragile and of
limited mobility; even this stage may be limited in colonizing new hosts. It has been
proposed that the main form of dispersal for the snow scale is through crawler
phoresy on the clothing and equipment of grove crews in the citrus groves
(Simanton 1973), but little attention has been paid to the role of air movement as a
dispersal agent. Wind have been demonstrated to aid dispersal between plants for
several scale insect species (Quayle 1916, Brown 1958, Greathead 1990). This
chapter explores the role of wind as a mechanism of dispersion for the citrus snow
scale, Unaspis citri.
Methods
Heavily infested, crawler-producing trees were identified in two citrus groves
in Lake Alfred, Polk Co. Florida. These 'Valencia' orange trees were about 1.5-2 m
tall in a block planted at 5x6.5 m spacing, and adjacent trees held little or no snow
scale. Sticky cards fastened to 1,2-m tall poles were set around the source tree at
70

71
a distance of about 1.5 m from the canopy edge. The traps were oriented North,
East, South, and West of the source tree. Each pole held two 10x10 cm2 Mylar
cards coated with Tanglefoot. The initial experiment was established on May 31,
1993 and the traps were retrieved seven days later. This experiment was repeated
five times with small variations in the same and in a new location (CREC, N-40,
block 16, Lake Alfred, Florida) in 1993-1994 (July 19, July 28, 1993, February 21,
April 6, October 14, 1994).
Crawlers were counted on the entire card, since the number of crawlers
trapped was always low. The numbers of snow scale crawlers, male scales and
natural enemies (Aphytis and Encarsia) were characterized. Wind speed and
direction was not monitored at the site, but wind data were recorded at the nearby
Citrus Research and Education Center (approximately 2.4 miles linear distance from
the initial site and approximately 2.5 miles from the second site).
Several experimental trials failed due to interference by grove vehicles. In
the first experiment (May 31-Jun 3,1993) established around a single tree, traps
were set at the four cardinal directions and two distances from the tree. Three of the
poles at the distance closest to the tree were knocked down, and thus only data
from the farthest distance were considered for analysis. In the second set (July 19-
22, 1993), two trees in the same grove were used with poles all at the same
distance from the source tree. All the poles in the East-West orientation were lost
to traffic, making the data unusable. The third trial (July 28-August 2, 1993) was a
repetition of the second and went undisturbed. The fourth trial (Feb. 21-28 1994)
was established in another grove (N-40 block 16, CREC) on two trees with traps at

72
two distances from the source. This was disturbed by a spray operation, and the
data were again discarded. The fifth trial (April 6-14, 1994) was a repetition of the
fourth and failed in one tree when at least three traps from the East orientation were
disturbed. Data from the second tree were complete. The sixth trial (October 14-21,
1994) was set up around single trees in both sites and these traps were not
disturbed.
In addition to collection of trap data, snow scale spread was monitored from
an artificial infestation in one grove at the CREC (block 16, N-40) during the last two
years. This grove contained 'Valencia' orange and 'Duncan' grapefruit trees. Four
trees in the plot were infested with snow scale in August 28, 1992, by attaching
snow scale infested seedlings to the trunk or main limbs. Monitoring involved
grading of snow scale colonies based on their appearance once every month. A
score from 5 (heavy infestation) to 0 (no infestation) was given to each tree in the
plot, based on a brief inspection for the presence of snow scale male covers on the
trunk and main branches (Casares 1977). Wind direction data for this extended
period were available only as daily means and were not used.
Numbers of crawlers trapped at each direction were compared by Kruskall-
Wallis non-parametric test (Siegel 1956). Visual comparisons were made between
wind patterns and number of catches.
Results
Mylar traps. Figures 6.1 and 6.2 present the number of crawlers caught in
trials 1, 3, 5 (data from a single tree without repetition) and 6. The hourly averages

Number of crawlers caught Number of crawlers caught
Direction (degrees)
73
0 90 180 270
16
14
12
10
8
6
4
2
0
North East South West
North East South West
Figure 6.1. Wind speed vs direction and catches of snow scale crawlers
A-Trial 1, May 31 to June 3 1993. B Trial 3, July 28 to August 2 1993.
Average wind speed (Km/h) Average wind speed (Km/h)

Number of crawlers caught Number of crawlers caught
Direction (degrees)
74
North East South West
NW N NE E SE S SW W
Figure 6.2. Wind speed vs direction and catches of snow scale crawlers.
A-Trial 5, April 6 to 14, 1994. B Trial 6, October 14 to 21, 1994
Average wind speed Average wind speed (Km/h)

75
Table 6.1. Frequency and speed of wind on the four quadrants in Lake Alfred,
CREC weather station. Data from hourly log in a seven-days period after the
beginning date.
Trial #
Orientation
Frequency
Average
(Km/h)
Variance
North
10
4.732
12.08
1
East
31
5.368
13.51
31/May/93
South
10
4.176
5.49
West
45
5.901
7.09
North
6
5.000
2.56
3
East
39
2.613
2.84
28/Jul/93
South
48
3.865
6.13
West
46
4.998
8.77
North
19
5.528
7.94
5
East
125
7.356
14.19
6/Apr/94
South
45
4.989
8.17
West
27
6.462
8.42
North
58
4.409
5.59
6
East
56
3.478
4.44
14/Oct/94
South
11
2.464
1.90
West
38
3.097
4.81
of wind speed against wind direction (azimuth) are also plotted for each experiment.
Additionally, Table 6.1 presents a summary of the wind conditions. Reference lines
in each wind plot represent quadrants of the compass: 315-45 for North, 45-135
for East, 135-225 for South, and 225-315 for West. West wind means that air
movement occurred from that direction and toward the opposite direction, East. The
only significant differences between the number of crawlers trapped at each cardinal
direction were observed in trial 3 (July 28-August 2, 1993, Kruskall-Wallis test
H=9.4235 P<0.024). However, it is possible to observe a trend relating the speed

76
and frequency of air movement and the number of crawlers trapped in a given
direction. In trial 1(May 31 June 3, 1993) the most frequent winds came from the
East and West, with the strongest being from the East. The highest catch was
observed in the West traps. Trial 3 provides another interesting case, wherein the
frequency of wind was equivalent between East and West, but the strongest winds
were from the West. The highest catches in this experiment were on the East traps.
The wind came predominantly from the East in trial 5, and no catches were obtained
from the trap on that side: the few crawlers caught were observed in North and West
traps. The dominant wind direction in trial 6 was from North and East, but with
slower speed than in previous tests. The highest catches were obtained in East,
South, and West traps.
Infestation Scoring Maps. Figure 6.5 show the scores of citrus snow scale
infestation on the citrus plot studied. Each bar represents the relative location and
score of each tree in the plot. The original infestation was started in August 1992
but it was not evident until March 1993. Through 1993 the spread of snow scale
showed a tendency toward the soutwest. This trend became more clear during
1994, but started to become confused by April 1994, when some northern trees
started showing infestation.
Discussion
Despite the lack of statistical significance, a role of wind on the dispersion of
the citrus snow scale can be inferred from Figures 6.1 and 6.2. The more frequent
and strongest winds originated from the direction opposite to the traps that showed

Y Coordinate
77
August 1992
March 1993
A
N
4 8 12 16
4
8 12
X Coordinate
Artificial infestations
Score 3
No infestation
Score 4
Score 1
@ Score 2

Score 5
Figure 6.3 Maps of trees infested with citrus snow scale in a citrus plot (CREC,
N-40 block 16). Each circle represents the position of a tree in the plot.
Circle's shade indicates the intensity of snow scale infestation measured
as scores from 0 to 5.

Y Coordinate
4 8 12 16 4
X Coordinate
8 12 16
Figure 6.3 -- Continuation.

Y Coordinate
April 1994 May 1994
4 8 12 16
July 1994
15 -
10 -
5 -
I I I I
4 8 12 16
4 8 12
August 1994
4 8 12
September 1994
4 8 12 16
X Coordinate
Figure 6.3 -- Continuation.

80
the highest crawler catches. The maps of progress of scale dispersal (Figure 6.3)
do not provide compelling evidence for the relationship between wind direction and
scale spread, since we lack the wind data, but indicate a general movement toward
the southwest.
Quayle (1916) studied dispersal of black scale (Saissetia oleae) in citrus
groves in a manner similar to that described here. He trapped black scale crawlers
450 feet (137 m) from the source, the farthest traps that he had set. He also
trapped Aonidiella aurantii at 30 to 150 feet (9 to 45 m) from a source of crawlers.
The distance that an airborne object can travel depends on several factors
such as wind speed, the height from which the object starts its descent, and the
speed at which its falls (Margalef 1977). An additional consideration for the snow
scale is the fact that the more dense scale colonies occur in the interior part of the
tree (trunk, scaffold branches) where air movement is somewhat restricted and the
canopy becomes a barrier for dispersion. Crawlers lifted by the wind from the trunk
must overcome that barrier. It is expected that young trees with heavy infestations
and sparse canopy will be better sources for short distance colonization, since winds
close to the ground are slower and the height of the dispersing crawlers is limited
in short trees. On the other hand, larger trees with a dense canopy should not be
a good source of crawlers, but those few crawlers that are potentially lifted by wind
from a high branch will have the opportunity to travel further.
Xinnian & Browning (1991) studied the movements of snow scale crawlers
and concluded that they were positively phototactic and suspected but did not
provide conclusive evidence for the effects of gravity. They also observed higher

81
activity and greater distance traveled by female crawlers. It is hypothesized that
snow scale infestations may start low on the trunk and progress upwards. This
hypothesis is supported by positive phototaxis or negative geotaxis coupled with
higher mobility of females crawlers, and could be an adaptation for reaching higher
levels on the tree. The possible advantages of this adaptation could be that it will
increase the chances for dispersion of crawlers by wind. These conditions should
enhance citrus snow scale potential for successfully colonizing neighboring trees.
The importance of phoresy on humans and other animals has been
addressed for other species of scale insects. Stephens & Aylor (1978) collected and
examined the feathers, beak and feet of birds for crawlers of Matsucoccus resinosae
and failed to find them. On the other hand, Washburn & Frankie (1981) exposed
humans and dogs to iceplants infested with Pulvinariella mesembryanthemi and
found crawlers and egg sacs on clothing and fur. It appears unlikely that citrus snow
scale is dispersed to any extent by humans given the location of citrus snow scale
on the interior wood and the thorny nature of the citrus canopy. Both factors should
limit contact between human clothing and the snow scale colonies on bark.

CHAPTER 7
MODELING CITRUS SNOW SCALE POPULATIONS
Introduction
Models are hypotheses of how a system functions (McKinion 1992). A model
can be designed to graphically represent relationships or to mathematically
represent a system. Mathematical models can vary in complexity, creating a
compromise between simplicity and accuracy. Simulation models tend to be
complicated, and produce outputs that can be used to forecast the state of a system
in the future, be it weather conditions, the populations of an organism, or the
dynamics of a market. Models has been used extensively in forecasting and
management of biological populations (Ruesink 1975, 1976, Getz & Gutierrez 1982,
Curry & Feldman 1987).
Overall, models help to understand the processes and provide means to
organize information about complex systems (Wagner & Willers 1992). However,
in order to create an accurate representation, thorough knowledge of the system is
required. Models also can be used to test the quality of the information available
regarding a system. The model then becomes a tool to check the integrity of our
estimation of its parameters.
Models are built through a series of steps. Ruesink (1976) describes four
such steps: definition of the system and the objects within it; creation of
82

83
mathematical equations for each object defined; interconnection of the defined
objects into a working model; and finally, testing the performance of the model
against the real world. Once a model is constructed, two types of tests can be
performed. Validation is testing the goodness of fit of the model against real data,
while sensitivity analysis is the analysis of the model's dynamics when the values
of the parameter estimates are changed.
The present chapter presents the development of a model for citrus snow
scale dynamics, tests the estimates obtained from studies presented in previous
chapters, establishes a framework for future expansion of studies, and identifies
gaps in knowledge on biology and ecology of the citrus snow scale.
Materials and Methods
Model Structure. A distributed delay model (Manetsch 1976) was created to
simulate the population dynamics of citrus snow scale. Distributed delay is
especially useful since it models different stages (age) and the variation in the transit
time of the individuals through a given stage (Berry & Stinner 1992). In this model,
both sexes were simulated since we are interested in development and not in
reproduction. In addition, males are an important numerical component of the
population and the most obvious to the observer. The life cycle was composed of
first nymphal stages, which develop into two sexes with sex ratio 0, a male stage,
a second female stage, and adult female (third) stage. Eggs and crawlers were not
included in the model, nor were the events occurring after sexual maturation of

84
females, since the information about those stages is incomplete (see discussion in
Chapter 3). The model for males and females was structured as follows:
dW1
(0
dt
dN 2 j
(0
dt
dN 3,f
(0
dt
^2,m
(0
dt
dN o m
o, m
(0
dt
- S,N,(t) p1 N1 (0
= e S1 A/1 (/) S2/V2f(0 M2A/2 f(0
- S2N2 f(t) M3/V3,f(0
- (1-6) S,N,(t) S2N2(t) V2(t)N2m(t)
M3^3,m (0
(Females )
(Males )
(7.1)
Where /^ represents the number of individuals in the stage /; S, represents
the developmental rate from stage / to stage / +1 (1/ time), p, represents the rate of
mortality at stage / (instantaneous rate of death); and 0 represents the sex ratio as
females per male. The parameters for the distributed delay model with attrition were
calculated according to Vansickle (1977) as:
k =
DEL = xs k
(7.2)
1 _L
x DEL
AR >

85
Where x and s2 are the average and variance of the time spent by the
individuals in a stage, k is the number of delay stages (subsystems) that constitute
each modeled stage, DEL is the mean delay that an individual spends in the
modeled stage, e is the proportion that survive at the end of the stage, and AR is the
instantaneous attrition rate, equivalent to the mortality rate p, previously mentioned.
Development. Development was modeled after Lactin et al.'s (1995)
modification of Logan et al.'s (1976) poikilotherm model, using the data presented
in chapter 3. Lactin's modification to the model of development is:
rate
epT
A
(7.3)
where T is temperature, p is the rate of increase to optimum temperature, Tm is the
maximum lethal temperature, AT is the difference between the lethal temperature
and the optimum temperature and A is a scaling factor that allows for incorporating
developmental thresholds (Lactin et al. 1995).
Mortality. Innate mortality was modeled using survival estimates for snow
scale cultures at constant temperature (Chapter 3) and was expressed as the
percentage of survival at the end of the stage (% of censored observations)
(Procedure LIFETEST, SAS Institute Inc. 1989).
Innate mortality was modeled as dependent upon temperature; it could not
be modeled as dependent upon RH because there were not sufficient data or
evidence of this dependency (see Chapter 3). The fraction dying is a proportion of
dead individuals in the previous stage (p,/VM) and is subtracted from the rate of

86
change between stages (Equation 7.1) (Vansickle 1977, Curry & Feldman 1987).
Linear regression was fitted to express the relationship between the proportion dying
and temperature for each stage simulated.
Reproduction and Migration. Reproduction and migration are not considered
in this model because they have not been adequately measured. Since
reproduction occurs beneath the mothers secreted cover, it is not possible to obtain
direct observations. Also, the patterns of movement of crawlers is such that newly
emerged crawlers walk away from their mothers (Xinnian & Browning 1991).
Reproduction could be irrelevant if a small area, such as the 18x12 mm2 patches of
bark surface used in chapter 5, are considered for the simulation. Most of the newly
emerged crawlers move beyond this area. On the other hand, if the simulation
considers a larger scale, such as a branch or a whole tree, then reproduction
becomes relevant. Inversely, migration will be the main component of recruitment
on a small scale, but probably quite irrelevant at large scale (Chapter 6).
Recruitment by migration was studied from field observations (Chapter 5) and the
analysis of photo slides, but was not used to estimate the recruitment of new scales.
Environmental Variables. Temperature was the only environmental variable
incorporated into the model. It was initially set constant for development of the
model and for comparison to the constant temperature developmental data
presented in Chapter 2. Temperature then was modeled as a compound cosine
function after Brewster & Allen (1991):

87
T (t) = a cos
1 1 t peak N N
2 n
period
T
7.4
where a was the amplitude of the departure from the average temperature T
(maximum and minimum) and 0 is a fraction that synchronizes the period of the
curve. The daily average temperature was in turn simulated in the same equation
but using the annual measurement of average temperature and amplitude.
Model Implementation. The model was developed in a PC graphic
environment using MATLAB 4.0 for Windows with Simulink v1,2c. (The MathWorks,
Inc., 24 Prime Park Way, Natick, MA 01760). Runge-Kutta fifth order method with
fourth order step-size control was used to integrate the system of differential
equations (Press et al. 1992). The minimum step size used was 1x1 O'3 days. The
Simulink block diagrams for the model are included in the Appendix.
Model Calibration. Data from the study of development at constant
temperatures (Chapter 2) were used to calibrate the model. These data were
recorded from nine cohorts of snow scale cultured on 'Duncan' grapefruit seedlings
and maintained at five temperatures and two relative humidities. The goodness of
fit of the model trajectories was tested against the data by x2 and Kolmogorov-
Smirnov tests (Siegel 1956, Sokal & Rohlf 1981).
Model Validation. The study of development was repeated with a set of eight
'Duncan' seedlings infested with citrus snow scales under variable temperature
(greenhouse conditions). The methods used to infest and record development of

88
scales were those described in chapter 2. The seedlings were maintained inside a
cage in a greenhouse and continuous readings of temperature were taken with an
hygrothermograph. The data were compared against the model by x2 and
Kolmogorov-Smirnov tests.
Results
Table 7.1 presents the parameters calculated for the simulation of
development of the snow scale from the observations reported in chapter 3. The
values chosen were for curves at 60% RH which represents the data set with the
most temperature observations.
Table 7.2 shows the mean, variance and the percentage of survival at the
end of the stage for cultures of snow scale maintained under constant temperature
and relative humidities. The parameters for the distributed delay model, calculated
according to the methods described by Vansickle (1977) are shown in Table 7.3.
The relationship between the fraction dying at the end of the stage and the
temperature are expressed as the regression equations presented in Table 7.4. The
sex ratio 0 was set at 2 males per female or 0.667 males/female (Chapter 3).
Table 7.1 Parameters used for the simulation of development for the citrus snow
scale.
Stage
P
Tm
AT
A
1sr stage 9
0.0109
35.4890
2.6085
-1.0910
1st stage o"
0.1503
34.0149
6.6330
-1.0x10*5
2nd stage 9
0.1100
35.1010
9.0010
-0.1001
Males
0.1100
34.8024
9.0017
-0.0996

89
Table 7.2 Average duration and percentage of survival from each stage at each
environmental condition studied (Chapter 3).
Stage
T
RH
Observed days in stage
/o
survival
Average Variance
16
60
5.39
31.52
0.596
70
5.26
23.25
0.721
21
60
3.75
17.08
0.485
70
3.69
21.81
0.750
1st stage
24
60
1.98
6.69
0.697
70
3.80
17.56
0.535
28
60
2.73
13.08
0.615
70
3.19
12.74
0.548
30
60
2.76
7.82
0.512
16
60
32.54
241.10
0.531
70
34.26
251.44
0.680
21
60
15.78
46.50
0.410
?nc* 70
19.23
102.05
0.693
females
24
60
13.29
37.22
0.618
70
12.31
41.79
0.373
28
60
11.64
24.74
0.682
70
11.65
37.22
0.333
30
60
13.98
39.18
0.547
16
60
25.67
311.15
1.000
70
33.97
358.29
1.000
21
60
33.61
229.41
0.857
70
17.18
161.22
0.976
Males*
24
60
15.61
66.70
1.000
70
17.29
81.34
1.000
28
60
17.91
121.49
0.973
70
17.29
65.34
0.938
30
60
3.72
81.12
0.850
16
60
64.09
30.13
1.000
70
56.37
229.27
0.947
21
60
44.10
65.64
0.857
Adult
70
60.27
97.38
0.970
females
24
60
44.23
121.00
1.000
70
29.14
21.92
1.000
28
60
34.74
91.58
0.976
70
50.62
331.22
1.000
*
30
60
20.81
331.93
0.727
* Includes secreting male 2nd stages, "prepupa" and "pupa".

90
Table 7.3 Parameters for the distributed delay model, calculated from data at
constant temperature and RH (Chapter 3).
Stage
T
RH
k
DEL
AR*
16
60
0.9
9.449
0.0735
70
1.2
6.924
0.0544
21
60
0.8
9.029
0.1284
70
0.6
5.850
0.0624
1st stage
24
60
0.6
3.665
0.1360
70
0.8
8.141
0.1154
28
60
0.6
6.418
0.1198
70
0.8
6.776
0.1325
30
60
1.0
5.486
0.1754
16
60
4.4
37.582
0.0181
70
4.7
37.215
0.0108
21
60
5.4
18.646
0.0521
70
3.6
21.279
0.0181
2nd stage
24
60
4.7
14.707
0.0344
females
70
3.6
16.164
0.0702
28
60
5.5
12.487
0.0318
70
3.6
15.746
0.0814
30
60
5.0
15.777
0.0406
16
60
2.1
25.669
0.0000
70
3.2
33.973
0.0000
21
60
4.9
34.683
0.0045
70
1.8
17.404
0.0014
Males**
24
60
3.7
15.611
0.0000
70
3.7
17.295
0.0000
28
60
2.6
18.095
0.0015
70
4.6
17.532
0.0037
30
60
0.2
9.637
0.0282
16
60
136.3
64.091
0.0000
70
13.9
56.589
0.0010
21
60
29.6
44.334
0.0035
70
37.3
60.322
0.0005
MUUIl
24
60
16.2
44.231
0.0000
females
70
38.7
29.143
0.0000
28
60
13.2
34.804
0.0007
70
7.7
50.619
0.0000
30
60
1.3
26.566
0.0136
* k is the number of subsystems (delay equations) that constitute each stage; DEL is the delay that
an individual spend in each stage; AR is the attrition rate.
** Includes secreting 2nd stage male, "prepupa" and "pupa".

91
Table 7.4 Linear regression between attrition rate (AR) and temperature (T).
Stage
Model
R2
1st stage
AR =
-0.0402 + 6.54x1 O'3 T
0.7023
2nd stage 9
AR =
1/(124.853-3.798 T)
0.5485
Males
AR =
-0.0303 + 1.494x1 O3 T
0.3124
Adult females
AR =
-0.0103 + 6.12x1 O'4 T
0.3947
The value of k (the number of substages comprising each stage) was chosen
as averages of the values presented in table 7.3 for each stage. It was initially set
to ^ for first nymph, k=5 for female second nymph, k=3 for males and ^ for
adult females. This adult female stage acts as a sink because death at old age is
not being modeled. The resulting model poorly fitted the observed data. The model
was then modified to k=2 for second nymphs and males, which improved the
correspondence between modeled and observed values.
Figure 7.1 shows the correspondence between observed cohorts of citrus
snow scale kept at constant temperatures and the curves of the simulation model.
Visual inspection of the curves suggested a good correspondence of the simulated
curves and the data for the first instar and the second instar female at all the
temperatures. However, the correspondence between simulated and observed data
for adult females was poor, especially at temperatures at the extreme of the range
studied. Tests of goodness of fit (x2 and Kolmogorov-Smirnov) were significant for
every stage in every chamber, indicating poor fitting. For first nymphs there was a
consistent underestimation of the number at the beginning of the simulation that
changed to slight overestimation after the first 5-6 days of simulation. Second

92
Table 7.5 Parameter values, standard error and confidence intervals at 95% for
development parameters.
Stage
Parameter
Value
Std. error
Asymptotic
Confidence intvl.
P
0.011
0.011
-0.132
0.154
Tm
35.489
9.538
-85.700
156.678
1st stage ?
AT
2.609
5.565
-68.095
73.312
A
-1.091
0.210
-3.754
1.572
P
0.150
0.196
-2.344
2.645
Tm
34.015
8.430
-73.094
141.124
1 stage AT
6.633
8.529
-101.733
114.999
A
-1.0x1 O'5
0.240
-3.055
3.055
P
0.110
3.0x1 O4
0.108
0.112
Tm
35.101
0.769
31.791
38.411
2nd stage ?
AT
9.001
0.000
9.001
9.001
A
-0.100
0.037
-0.258
0.058
P
0.110
7.7x1 O'4
0.107
0.113
Tm
34.802
1.395
28.800
40.805
Males
AT
9.002
0.000
9.007
9.002
A
-0.100
0.741
-0.418
0.219
nymphal females and males were overestimated at low and high temperatures, and
were closer to the observed values at temperatures between. Finally, adult females
were consistently overestimated.
Figure 7.2 shows the simulation plotted over data obtained from a snow scale
culture maintained under variable temperatures in a greenhouse. The simulation
was generated using the mean temperature and amplitude measured in the

93
Table 7.6 Parameter values, standard error and confidence intervals at 95% for
mortality parameters.
Stage
Parameter
Value
Confidence limits
OlU. ciiUi
lower
upper
1st stage
intercept
-0.049
0.038
-0.156
0.057
slope
0.007
0.002
0.003
0.012
2nd stage
intercept
132.611
31.534
45.072
220.150
?
slope
-4.017
1.336
-7.726
-0.309
Males
intercept
-0.030
0.033
-0.122
0.061
slope
0.002
0.001
-0.002
0.005
Adult
intercept
-0.010
0.010
-0.038
0.018
females
slope
0.001
0.000
-0.001
0.002
greenhouse. Correspondence of the simulation to the data was not good. Figure
7.2 shows overestimation in all the stages except second nymphal females.
Discussion
The model for development of citrus snow scale under constant temperatures
approximates the experimental data. However, problems of estimation arose.
Populations were more accurately estimated at intermediate temperatures and early
in the development of the cohort, while under- and over-estimation occurred at high
and low temperatures and in later life stages. The potential reasons for the lack of
correspondence could be faults in the conceptualization of the model, faults in the
programming or inaccuracy in the estimation of the development or mortality
parameters, or inaccuracy in the development data.

150
75
0
135
90
45
0
225
150
75
0
120
80
40
0
225
150
75
0
jre 7
94
16C, 60%RH
21C, 60%RH
24C, 60%RH
i i i
28C, 60%RH
1 T
30C, 60%RH
150
16C, 70%RH
21 C, 70%RH
24C, 70%RH
28C, 70%RH
0

0
O
20 40 60 80
Nynph I
Female Nymph II
Adult females
Males
0 20 40 60 80
Time (days)
Duration of development of citrus snow scale individuals under
>nstant temperature. Circles: observed data. Lines: simulated values.

95
80 90 100 110 120 130
Time (julian days)
Figure 7.2 Duration of development of citrus snow scale individuals under
greenhouse conditions. Circles: observed data. Lines: simulated values.

96
Among faults in the conceptualization of the model could be the exclusion of
relevant factors, perhaps relative humidity, conditions of the host plant or other
unknown factors. Programming errors are unlikely, since the use of a simulation
graphical tool (MATLAB's Simulink) greatly simplifies this task. The most likely
problem remaining then, is inaccuracy in the estimation of parameters. Table 7.5
presents confidence intervals for the parameters used in simulating development
and table 7.6 shows the same information for mortality. Confidence intervals for the
rates of development of the first instar (both sexes) are very wide, which may
account for the problems with the first stage curve. The parameter estimation used
thus may not accurately represent the real values. At the same time, regression of
the instantaneous mortality rate (AR) against temperature showed low coefficients
of determination (R2, Table 7.4), which implies poor approximation of the observed
mortality. Further development of methods to simulate mortality may be necessary.
The simulation of this factor might be improved using a nonlinear or a polynomial
model. Furthermore, other factors besides temperature (eg. interaction of
temperature with humidity) likely affect mortality.
Further improvement of this model requires filling gaps in knowledge of the
snow scale system. It is important to determine reproductives rates, the longevity
of the adult reproductive females, the rate of dispersion (crawling) of new individuals
on a substrate, and the effects of substrate (plant organ and plant variety) on
development and survival. These data will give us a more precise idea of the
potential for pest outbreak of this species. On the other hand, other important
factors that must be incorporated into the model are the effects of parasitoids and

97
predators on the citrus snow scale and how changes in these elements will modify
the dynamics of its populations.

CHAPTER 8
CONCLUSIONS
With this work, we have contributed new knowledge to the biology of the
citrus snow scale and the population dynamics of this species in the field.
Development of snow scale is dependent on temperature. Optimal
temperature for development falls in a range of 25 to 38C with most frequent values
falling near 29C. The effect of relative humidity was found to be significant only for
first instar males but the range of relative humidity tested was too narrow (10%).
Testing of a wider humidity range would be expected to identify citrus snow scale
tolerance limits and optimal humidity conditions.
In light of these new data, it can be predicted that snow scale will fare poorly
during the hot Florida summer months when temperatures easily surpass the
calculated optimum. Development was faster on grapefruit leaves than was
previously reported for citrus snow scale on lemon fruits, but it is still not known if
snow scale development on bark is different from rates already characterized on
leaves or fruits. In any case, grapefruit leaves represent a more suitable substratum
than fruits. It appears that host plant variety and/or plant organ infested, can affect
the rate of citrus snow scale development.
Field studies by photographic techniques, life table analysis, and destructive
canopy samples showed Encarsia spp. present on snow scale infesting twigs and
98

99
leaves as well as trunk and limbs. Attack by Encarsia ranged between zero and
38% on citrus trunks at different sites over time, while observable attack by
predators ranged between zero and 49%. An important percentage of the mortality
could not be identified (Chapter 4), which must be composed of predation, diseases,
parasitoids host feeding and abiotic causes. Since the effects of predation and
diseases must be greatly underestimated by not counting the proportion included
in this percentage, the observed incidences should be taken as only an indication
of trends. Definitely, the effect of predation must be greater than what was
observed. In the case of diseases, is more difficult to assign a value as population
regulator since the question about pathogenicity is not yet resolved. Nonetheless,
the species of natural enemies observed are considerable mortality factors for snow
scale, and therefore their conservation should be considered when pest
management actions need to be taken.
Studies in the field found little evidence of Aphytis lingnanensis contributing
to mortality on populations of snow scale infesting trunks and main limbs. Analysis
of samples from leaves and twigs indicated that the conditions on bark and leaves
may be different with regard to mortality factors. Aphytis lingnanensis and fungi
were present in nearly all of the samples taken during the two year study. These
appeared to play a more important role on marginal portions of snow scale
populations that inhabit the canopy. Mortality invoked by Aphytis lingnanensis is
higher on the canopy, which suggests that although this aphelinid has an effect on
snow scale populations on the canopy, the main component of the population that
resides on trunk and main limb bark escapes parasitism by Aphytis.

100
Snow scale crawlers can disperse by taking advantage of air currents as has
been demonstrated for many other scale species. The importance of this means of
dispersion needs to be investigated further. Questions arise regarding how far
crawlers can be carried by wind, and how long they remain viable after leaving the
site of their birth.
A distributed delay model of the citrus snow scale built around data obtained
on studies of development (Chapter 3) mimics the dynamics of the cohort
development but it is not precise, especially for late instars (adults) and under low
temperature conditions. Further research and development are needed to refine the
model for citrus snow scale population dynamics.
Systems in which measurement of stage-specific population densities can
serve to establish timing for management intervention, are good candidates for the
implementation of simulation models to aid in the decision making process (e.g.,
Pfeiffer 1985, McClain et al. 1990). Simulation models do not yet seem promising
as management aids in the case of the citrus snow scale, because knowledge about
this insect is inadequate. However, the model was useful as a tool to understand
the system and identify areas for additional research. Expansion of the model
developed in the studies reported here to a more precise, comprehensive stage can
be a powerful tool to study the interactions between citrus snow scale and its natural
enemies, helping to predict the outcome of biological control experiments.
Although our comprehension of citrus snow scale biology has increased, it
is necessary to expand our understanding of the impact of the citrus snow scale on
the biology of citrus trees. This understanding could lead to enhanced efforts to

101
manage populations of citrus snow scale within commercial citrus planting through
applied biological control and other integrated pest management strategies.

APPENDIX
SIMULINK BLOCK DIAGRAM
FOR THE SNOW SCALE MODEL
Citrus snow scale model
Only intrinsic mortality included
102

Nymph I
103
Female nymph II
AR

104
Males
Temperature
Adult females
1
in 1
U i*N i
}
1/s
Adults females
dN/dt
afemales
To Workspace

105
Subsystem
U i*N i

REFERENCES CITED
Adachi, I. & R. Korenaga. 1991. Fertility schedules of Unaspis yanonensis
(Flemiptera: Diaspididae) in relation to daily temperature. Res. Popul. Ecol.
33: 57-68.
. 1992. A simulation model for the arrowhead scale (Hemiptera: Diaspididae)
population dynamics on citrus trees in relation to pest-management
programs. Res. Popul. Ecol. 34: 155-171.
Albrigo, L. G. & R. F. Brooks. 1977. Penetration of citrus cuticules and cells by citrus
snow scale, Unaspis citri (Comst.), pp. 463-467. In W. Grierson [ed.],
Proceedings of the International Society of Citriculture. Orlando, Florida.
Anon. 1954. White louse scale (Unaspis citri). Agrie. Gazette 65: 270-271.
Arias-Revern, J. M. 1988. Aspects of the biology and population ecology of the
purple scale Lepidosaphes beckii (Newmann) and the glover scale L. gloveri
(Packard) (Flomoptera: Diaspididae), citrus pests in Costa Rica. M.Sc. Thesis
Universidad de Costa Rica, San Pedro de Montes de Oca. (In Spanish).
Atkinson, P. R. 1983. Estimates of natural mortality related to environmental factors
in a population of citrus red scale Aonidiella aurantii (Maskell) (Hemiptera:
Diaspididae). Bull. Entomol. Res. 73: 239-258.
Banks, H. J. 1990. Physiology and biochemistry, pp. 267-274. In D. Rosen [ed.],
Armored scale insects. Their biology, natural enemies and control. Elsevier,
Amsterdam.
Baskerville, G. L. & P. Emin. 1969. Rapid estimation of heat accumulation from
maximum an minimum temperatures. Ecology 50: 514-517.
Beattie, G. A. C. & J. G. Gellatley. 1983. Citrus scale insects. (Agfact H2.AE.2)
Department of Agriculture, New South Wales.
Benassy, C. & C. Pinet. 1972. Notes on Unaspis yanonensis Kuw. (Homoptera,
Diaspidinae) in the Maritime Alps. Ann. Zool. Ecol. Anim. 4: 187-212.
106

107
Berry, I. L & R. E. Stinner. 1992. Computer development of insect population
models, pp. 54-73. In J. L. Goodenough and J. M. McKinion [eds.], Basic of
Insect Modeling. American Society of Agricultural Engineers.
Brewer, B. S. & A. D. Oliver. 1987. Euonymus scale, Unaspis euonymi (Comstock)
(Homoptera: Diaspididae): effects of host cultivar age, and location on
infestation levels. J. Entomol. Sci. 22: 119-122.
Brewster, C. C. & J. C. Allen. 1991. Simulation of plant resistant in a celery-leaf
miner-parasitoid model. Florida Entomol. 74: 24-41.
Brown, C. E. 1958. Dispersal of the pine needle scale, Phenacaspis pinifoliae
(Fitch), (Diaspididae: Homoptera). Can. Entomol. 90: 685-690.
Browning, H. W. 1994. Biological control of the citrus snow scale Unaspis citri in
Florida: evaluation of Aphytis and other natural enemy species, pp. 119-142.
In D. Rosen [ed.], Advances in the study of Aphytis (Hymenoptera:
Aphelinidae). Intercept Press, Andovers, UK.
Carey, J. R. 1993. Applied demography for biologists, with special emphasis on
insects. Oxford Univ. Press, New York.
Carr, D. S. & B. L. Harris. 1949. Solutions for maintaining constant relative humidity.
Ind. Eng. Chem. 41: 2014-2015.
Carroll, D. P. 1979. Within-tree distribution and host substrate influences on
California red scale, Aonidiella aurantii (Mask.); density, survival,
reproduction and parasitation. Ph D. Thesis University of California,
Riverside.
Casares, R. 1974. Effects of temperature on the development and fecundity of citrus
snow scale, Unaspis citri (Comstock) (Homoptera: Diaspididae). M.Sc.
Thesis University of Florida, Gainesville.
. 1977. Population dynamics of the citrus snow scale Unaspis citri
(Comstock) (Homoptera: Diaspididae) at Paradise Island, Osceola county,
Florida. Ph.D. Thesis University of Florida, Gainesville.
Castieiras, A. & O. Obregn. 1986. Toxicity of six pesticides used on citrus crops
against Aspidiotiphagus lounsburyi. Ciencia Tec. Agrie., Proteccin de
Plantas 9: 73-79.
Cockfield, S. D. & D. A. Potter. 1987. Distribution, development, and feeding impact
of Eunonymus scales (Homoptera: Diaspididae) on Eunonymus fortunei
under greenhouse conditions. Environ. Entomol. 16: 917-921.

108
. 1990. Euonymus scale (Homoptera: Diaspididae) effects on plant growth
and leaf abscission and implications for differential site selection by male and
female scales. J. Econ. Entomol. 83: 995-1001.
Cooper, R. M. & R. D. Oetting. 1986. Adaptations of scale insects to host variability.
J. Agrie. Entomol. 3: 163-169.
Curry, G. L. & R. M. Feldman. 1987. Mathematical foundations of population
dynamics. The Texas Engineering Experiment Station Monographs Texas
A&M Univ. Press, College Station.
DeBach, P. & S. V. Rao. 1969. Experimental studies on hybridiation and sexual
isolation between some Aphytis species (Hymenoptera: Aphelinidae) II.
Experiments on sexual isolation. Hilgardia 39: 555-567.
Dickens, T. H. 1968. Life history of Citrus Snow Scale Unaspis citri (Comstock).
M.Sc. Thesis University of Florida, Gainesville.
Drea, J. J. & R. W. Carlson. 1987. The establishment of Chilocorus kuwanae
(Coleptera: Coccinellidae) in Eastern United States. Proc. Entomol. Soc.
Wash. 89: 821-824.
. 1988. Establishment of Cybocephalus sp. (Coleptera: Nitidulidae) from
Korea on Unaspis euonymi (Homoptera: Diaspididae) in the eastern United
States. Proc. Entomol. Soc. Wash. 90: 307-309.
Fernandez Argudn, M. 1987. Morfology of Aspidiotiphagus sp. (Hymenoptera:
Aphelinidae). Rev. Prot. Veg. 2: 134-139.
Ferris, G. F. 1937. Atlas of the scale insects of North America. The Diaspididae,
Part I Stanford Univ. Press, Stanford.
Fisher, F. E. 1950. Entomogenous fungi attacking scale insects and rust mites on
Citrus in Florida. J. Econ. Entomol. 43: 306-309.
Fisher, F. E., W. L. Thompson & J. T. Griffiths Jr. 1949. Progress report on the
fungus diseases of scale insects attacking citrus in Florida. Florida Entomol.
32: 1-11.
Fisher, J. 1985. Parasitic wasp prey on snow scale. Citrus Ind. 66: 36,40-42.
Foldi, I. 1990. The scale cover, pp. 43-54. In D. Rosen [ed.], Armored scale insects:
their biology, natural enemies and control. Elsevier, Amsterdam.

109
Furuhashi, K. 1974. Developmental difference of arrowhead scale (Unaspis
yanonensis Kuwana) after infest on Satsuma mandarin and Natsudaidai. Bull.
Shizuoka Prefect. Citrus Exp. Sta. 11: 68-73.
Furuhashi, K. & M. Nishino. 1983. Biological control of arrowhead scale, Unaspis
yanonensis, by parasitic wasps introduced from the People's Republic of
China. Entomophaga 28: 277-286.
Getz, W. M. & A. P. Gutierrez. 1982. A perspective on system analysis in crop
production and insect pest management. Ann. Rev. Entomol. 27: 447-466.
Gill, S. A., D. R. Miller & J. A. Davidson. 1982. Bionomics and taxonomy of the
euonymus scale Unaspis euonymi (Comstock), and detailed biological
information on the scale in Maryland (Homoptera: Diaspididae). (MP 969)
Maryland Agricultural Experiment Station, College Park.
Greathead, D. J. 1990. Crawler behaviour and dispersal, pp. 305-308. In D. Rosen
[ed.], Armored scale insects, their biology, natural enemies and control.
Elsevier, Amsterdam.
Greenspan, L. 1977. Humidity fixed points of binary saturated aqueous solutions.
J. Res. Nat. Bur. Stand. Sect. A 81 A: 89-95.
Harcourt, D. G. 1969. The development and use of life tables in the study of natural
insect populations. Ann. Rev. Entomol. 14: 175-196.
Hely, P. C. 1944. The white louse scale (Protaspis cith). A pest of coastal citrus
trees. Agrie. Gazette 55: 283-285.
Houston, K. J. 1991. Chilocorus circumdatus Gyllenhal newly established in
Australia and additional records for Coccinella undecimpunctata L.
(Coleptera: Coccinellidae). J. Aust. Entomol. Soc. 30: 341-342.
Huang, L. L, D. W. Wang, Q. B. Zhang & W. S. Zhu. 1983. A study on the biology
and control of the arrowhead scale (Unaspis yanonensis Kuwana). Acta
Phytophylacica Sin. 10: 19-24.
Hutchinson, G. E. 1978. An introduction to population ecology. Yale Univ. Press,
New Haven.
Kajita, H. 1972. The age effect on the number of ovarian and deposited eggs of
Aspidiotiphagus citrinus Craw. Jap. J. Appl. Entomol. Zool. 16: 202-204.
. 1976. Searching and oviposition behavior of Aspidiotiphagus citrinus
(Craw), a parasite of the arrowhead scale, Unaspis yanonensis (Kuwana). 1.

110
distribution of the hosts on citrus leaves and searching behavior of the
parasite. Sci. Bull. Fac. Agrie. Kyushu Univ. 31: 113-117.
. 1977a. Searching and ovipositing behavior of Aspidiotiphagus citrinus
(Craw), a parasite of the arrowhead scale, Unaspis yanonensis (Kuwana). 4.
effect of host density on the time spent by the parasite on a citrus leaf and
the number of hosts parasitized. Sci. Bull. Fac. Agrie. Kyushu Univ. 31: 151-
157.
. 1977b. Searching and ovipositing behaviour of Aspidiotiphagus citrinus
(Craw), a parasite of the arrowhead scale, Unaspis yanonensis (Kuwana). 3.
influence of the structure of scales on the ovipositing behavior of the parasite.
Sci. Bull. Fac. Agrie. Kyushu Univ. 31: 125-131.
Kanda, K. & FI. Kajita. 1977. Searching and ovipositing behavior of Aspidiotiphagus
citrinus (Craw), a parasite of the arrowhead scale, Unaspis yanonensis
(Kuwana). 2. searching behavior for host females on the citrus leaf. Sci. Bull.
Fac. Agrie. Kyushu Univ. 31: 119-124.
Kloppenburg, J., Jr. & D. L. Kleimman. 1987. The plant germplasm controversy.
BioScience. 37: 190-198.
Knapp, J. L. (ed.). 1995. 1995 Florida citrus pest management guide. Florida
Cooperative Extension Service/IFAS Univ. of Florida, Gainesville.
Korenaga, R. 1983. The time of hatching and crawling speed of U. yanonensis
(Flomoptera: Coccidae). Jap. J. Appl. Entomol. Zool. 27: 308-310.
Korenaga, R. & Y. Sakagami. 1981. Prediction of the occurrence of the arrowhead
scale Unaspis yanonensis (Kuwana), pp. 700-704. In K. Matsumoto [ed.],
Proceedings of the International Society of Citriculture. Organizing
committee, Tokyo, Japan.
Korenaga, R., S. Okudai & Y. Sakagami. 1974. Studies on forecasting the
occurrence of the arrowhead scale, Unaspis yanonensis Kuwana. I.
Forecasting occurrence date of the first generation first instar larvae from the
development of male larvae and of ovaries of mature female in late
hibernating period. Bull. Hortic. Res. Stn. Ser. B (Okitsu). 1: 85-99.
Korenaga, R., Y. Sakagami & S. Okudai. 1976. The effect of temperature on the
development of the arrowhead scale, Unaspis yanonensis Kuwana. II.
development of the first generation under fluctuating temperature. Bull. Fruit
Tree Res. Stn. Ser. B (Okitsu). 3: 47-56.

111
Koteja, J. 1990a. Embryonic development; ovipary and vivipary, pp. 233-242. In D.
Rosen [ed.], Armored scale insects, their biology, natural enemies and
control. Elsevier, Amsterdam.
. 1990b. Life history, pp. 243-254. In D. Rosen [ed.], Armored scale insects,
their biology, natural enemies and control. Elsevier, Amsterdam.
Kramer, D. A., R. E. Stinner & F. P. Hain. 1991. Time Versus rate in Parameter
Estimation of Nonlinear Temperature-Dependent Development Models.
Environ. Entomol. 20: 484-488.
Kuno, G., M.A. Coln-Ferrer. 1973. Pathogenicity of two Fusarium fungi to an
armored scale insect, Selenaspidus articulatus. J. Invertebr. Pathol.
22(3):473-474.
Lactin, D. J., N. J. Holliday, D. L. Johnson & R. Craigen. 1995. Improved rate model
of temperature-dependent development by arthropods. Environ. Entomol. 24:
68-75.
Lamb, R. J., G. H. Gerber & G. F. Atkinson. 1984. Comparison of developmental
rate curves applied to egg hatching data of Entomoscelis americana Brown
(Coleptera: Chrysomelidae). Environ. Entomol. 13: 868-872.
Logan, J. A., D. J. Wolkind, S. C. Hoyt & L. K. Tanigoshi. 1976. An analitical Model
for Description of Temperature Dependent Rate Phenomena in Arthropods.
Environ. Entomol. 5: 1133-1140.
Manetsch, T. J. 1976. Time-varying distributed delays and their use in aggregative
models of large systems. IEEE Transactions on Systems, Man, and
Cybernetics SMC-6: 547-553.
Marcandier, S. & G. G. Khachatourians. 1987. Evolution of relative humidity and
temperature within a closed chamber used for entomological studies. Can.
Entomol. 119: 893-900.
Margalef, R. 1977. Ecologa. 2nd. ed. Omega, Barcelona. (In Spanish).
McClain, D. C., G. C. Rock & R. E. Stinner. 1990a. San Jose Scale (Homoptera:
Diaspididae): Simulation of seasonal phenology in North Carolina orchards.
Environ. Entomol. 19: 916-925.
1990b. Thermal requirements for development and simulation of the
seasonal phenology of Encarsia perniciosi (Hymenoptera: Aphelinidae), a
parasitoid of the San Jose scale (Homoptera: Diaspididae) in North Carolina
orchards. Environ. Entomol. 19: 1396-1402.

112
McClure, M. S. 1990a. Ecology: Influence of environmental factors, pp. 319-329. In
D. Rosen [ed ], Armored scale insects: their biology, natural enemies and
control. Elsevier, Amsterdam.
. 1990b. Life-tables, models and population dynamics, pp. 331-337. In D.
Rosen [ed.], Armored scale insects. Their biology, natural enemies and
control. Elsevier, Amsterdam.
McKinion, J. M. 1992. Getting started: basics of modeling strategies, pp. 1-8. In J.
L. Goodenough and J. M. McKinion [eds.], Basics of insect modeling.
American Asociation of Agricultural Engineers, St. Joseph.
Mead, F. W. 1976. Economic insects in Florida. Florida Department of Agriculture
and Consumer Services, Division of Plant Industries, Gainesville.
Miller, D. R. & M. Kostarab. 1979. Recent advances in the study of scale insects.
Ann. Rev. Entomol. 24: 1-27.
Murakami, Y., H. Uematsu, S. Ohga & H. Kajita. 1972. Parasites of Unaspis
yanonensls in Japan (Hymenoptera, Chalcidoidea). Mushi 46: 45-52.
Murdoch, W. W., R. F. Luck, S. J. Walde, J. D. Reeve & D. S. Yu. 1989. A refuge
for red scale under control by Aphytis: structural aspects. Ecology 70: 1707-
1714.
Nishino, M. 1974. Studies on biology and forecasting of occurrence of the
arrowhead scale Unaspis yanonensis Kuwana. Bull. Shizuoka Prefect. Citrus
Exp. Sta. Special Bulletin No 2: 1-101.
Nishino, M. & K. Furuhashi. 1971a. Studies on forecasting the occurrence of the
arrowhead scale scale, Unaspis yanonensis, Kuwana. IV. Influence of
temperature for development of eggs in ovaries of the adult female. Bull.
Shizuoka Prefect. Citrus Exp. Sta. 9: 115-132.
. 1971b. Studies on forecasting the occurrence of the arrowhead scale,
Unaspis yanonensis Kuwana. III. The period, type and number of emergence
of the 1st instar larvae on the 3rd generation. Bull. Shizuoka Prefect. Citrus
Exp. Sta. 9: 97-114.
Odum, E. P. 1971. Fundamentals of ecology, 3rd. ed. Saunders, Philadelphia.
Ohkubo, N. 1981. Role of petroleum oil sprays in an integrated pest management
system of citrus crops in Japan, pp. 611-614. In K. Matsumoto [ed.],
Proceedings of the International Society of Citriculture. Organizing
committee, Tokyo, Japan.

113
Okudai, S., R. Korenaga & Y. Sakagami. 1971. The effect of temperature on the
development of the arrowhead scale, Unaspis yanonensis. I. development
of the first generation under the constant temperatures. Bull. Hortic. Res. Stn.
Ser. B (Okitsu). 11: 193-201.
. 1974. Studies on the autumnal occurrence of larvae of the arrowhead
scale, Unaspis yanonensis Kuwana. II. On the relationship between
temperature in the early stage of the autumn adults and the autumnal
appearance of larvae. Bull. Hortic. Res. Stn. Ser. B (Okitsu). 1: 101-113.
. 1975. Effect of temperature on the ovarian development of the arrowhead
scale, Unaspis yanonensis Kuwana, in the late hibernating period. Bull. Fruit
Tree Res. Stn. Ser. B (Okitsu). 2: 97-106.
Pfeiffer, D. G. 1985. Pheromone trapping of males and prediction of crawler
emergence for San Jose scale (Homoptera: Diaspididae) in Virginia apple
orchards. J. Entomol. Sci. 20: 351-353.
Press, W. H., S. A. Teukolsky, W. T. Vetterling & B. P. Flannery. 1992. Numerical
recipes in FORTRAN: the art of scientific computing, 2nd. ed. Cambridge
University Press, Cambridge.
Quayle, H. J. 1916. Dispersion of scale insects by the wind. J. Econ. Entomol. 9:
486-493.
Rao, V. P. 1949. The genus Unaspis MacGillivray. Microentomology. 14: 59-72.
Reed, D. K., A. G. Selhime & C. R. Crittenden. 1967. Occurrence of citrus snow
scale, Unaspis citri, on several varieties of citrus in Florida. J. Econ. Entomol.
60: 300-301.
Riehl, L. A., R. F. Brooks, C. W. McCoy, T. W. Fisher & H. A. Dean. 1980.
Accomplishments toward improving integrated pest management for citrus,
pp. 319-363. In C. B. Huffaker [ed.], New technology of pest control. Willey-
Interscience, New York.
Rose, M. 1990. Diaspidid pest problems and control in crops: Citrus, pp. 535-541.
In D. Rosen [ed.], Armored scale insects: their biology, natural enemies and
control. Elsevier, Amsterdam.
Rosen, D. & P. DeBach. 1977. Diaspididae, pp. 78-128. In C. P. Clausen [ed.],
Introduced parasites and predators of arthropod pests and weeds: a world
review. ARS USDA, Washington, D C.

114
. 1979. Species of Aphytis of the World (Hymenoptera: Aphelinidae). Series
entomolgica. Dr. W.Junk BV, The Hague.
Ruesink, W. G. 1975. Analysis and modeling in pest management, pp. 353-376. In
R. L. Metcalf and W. H. Luckmann [eds.], Introduction to insect pest
management. J.Wiley & Sons, New York.
. 1976. Status of the systems approach to pest management. Ann. Rev.
Entomol. 21: 27-44.
Sadof, C. S. & J. J. Neal. 1993. Use of host plant resources by the euonymus scale,
Unaspis euonymi (Homoptera: Diaspididae). Ann. Entomol. Soc. Am. 86:
614-620.
Sakagami, Y. & R. Korenaga. 1982. Studies on forecasting the occurrence of the
arrowhead scale, Unaspis yanonensis Kuwana. II. Forecasting initial
occurrence date of the first instar larvae by multiple regression equation. Bull.
Fruit Tree Res. Stn. Ser. B (Okitsu). 9: 35-51.
Samson, R. A., H. C. Evans & J. P. Latg. 1988. Atlas of entomopathogenic fungi.
Springer-Verlag, Berlin.
SAS Institute Inc. 1989. SAS/STAT User's guide version 6., 4th. ed. 2. SAS Institute
Inc., Cory, NC.
Selhime, A. G. & R. F. Brooks. 1977. Biological control of some armored scale
insects on citrus, pp. 475-478. In W. Grierson [ed.], Proceedings of the
International Society of Citriculture. Organizing committee, Orlando, Florida.
Sevacherian, V., V. M. Stern & A. J. Mueller. 1977. Heat accumulation for timing
Lygus control measures in a Safflower-Cotton complex. J. Econ. Entomol.
70: 399-402.
Sharpe, P. J. H. & D. W. DeMichele. 1977. Reaction kinetics of poikilotherm
development. J. Theor. Biol. 64: 649-670.
Sharpe, P. J. H., G. L. Curry, D. W. DeMichele & C. L. Cole. 1977. Distribution
model of organism development times. J. Theor. Biol. 66: 21-38.
Siegel, S. 1956. Non parametric statistics for the behavioral sciences. McGraw-Hill,
New York.

115
Simanton, W. A. 1973. Changes in Florida citrus pest populations during 16 years,
pp. 491-496. In 0. Carpena [ed.], Proceedings I International Congress of
Citriculture, Murcia, Spain.
Smith, D. & D. F. Papacek. 1985. Integrated pest management in Queensland
citrus. Queensland Agr. J. 111: 249-259.
Sokal, R. R. & F. J. Rohlf. 1981. Biometry, 2nd. ed. Freeman, New York.
Southwood, T. R. E. 1978. Ecological methods, with particular references to the
study of insect populations, 2nd. ed. Chapman & Hall, New York.
Steel, R. G. D. & J. H. Torrie. 1980. Principles and procedures of statistics: a
biometrial approach, 2nd. ed. McGraw-Hill, New York.
Stephens, G. R. & D. E. Aylor. 1978. Aerial dispersal of red pine scale,
Matsucoccus resinosae (Homoptera: Margarodidae). Environ. Entomol. 7:
556-563.
Stoetzel, M. B. 1976. Scale-cover formation in the diaspididae (Homoptera:
Coccoidea). Proc. Entomol. Soc. Wash. 78: 323-332.
Summerville, W. A. T. 1935. White louse of citrus. Advisory leaflet. Department of
Agriculture and Stock, Queensland.
Summy, K. R., M. R. Davis, W. G. Hart & F. E. Gilstrap. 1984. Use of close-up
photography in nondestructive monitoring of citrus blackfly cohorts. J. Rio
Grande Val. Hortic. Soc. 37: 55-60.
Takagi, K. 1981. Evaluation of parasitoids of latent pests with sticky suction trap, pp.
627-630. In K. Matsumoto [ed.], Proceedings of the International Society of
Citriculture. Organizing committee, Tokyo, Japan.
Takezawa, H. & M. Uchida. 1969. Relationship between the ovarial development
and the appearance of nymphs in arrowhead scale (Unaspis yanonensis)
with special reference to its application to forecasting. Jap. J. Appl. Entomol.
Zool. 13: 31-39.
Tanaka, M. 1981. Biological control of arrowhead scale, Unaspis yanonensis
(Kuwana), in the implementation of IPM programs of citrus orchards in
Japan, pp. 636-640. In K. Matsumoto [ed.], Proceedings of the International
Society of Citriculture. Organizing committee, Tokyo, Japan.
Taylor, F. 1981. Ecology and evolution of physiological time in insects. Am. Nat
117: 1-23.

116
Tern, A. L, M. L. Collado de Manes, S. Glenross, R. Alvarez & H. Lazaro. 1985.
Primary and secundary Hymenopteran parasitoids of Diaspidids in Tucuman
(Argentina) citrus, except Aonidiella aurantii. Rev. Invest. CIRPON 3: 25-33.
(In Spanish).
Terry, L. I. & G. J. Edwards. 1989. Efficacy of densitometric and multispectral
techniques for monitoring infestations of citrus snow scale on citrus bark.
Photogram. Eng. Remote Sensing 55: 1471-1475.
UK, CABI. Institute of Entomology. 1962. Distribution maps of pests. Pest: Unaspis
citri (Comst.). Commonwealth Agriculture Bureaux. Series A (Agricultural).
Map 149.
. 1970. Distribution maps of pests. Pest: Unaspis eunoymi (Comst.).
Commonwealth Agriculture Bureaux. Series A (Agricultural). Map 269.
. 1988. Distribution maps of pests. Pest: Unaspis yanonensis (Kuwana).
Commonwealth Agriculture Bureaux. Series A (Agricultural). Map 503.
van Driesche, R. G. 1983. Meaning of "percent parasitism" in studies of insect
parasitoids. Environ. Entomol. 12: 1611-1622.
Vansickle, J. 1977. Attrition in distributed delay models. IEEE Trans. Syst. Man
Cybern. SMC-7: 635-638.
Varley, G. C. & G. R. Gradwell. 1971. The use of models and life tables in
assessing the role of natural enemies, pp. 93-112. In C. B. Huffaker [ed.],
Biological control. Plenum/Rosetta.
Varley, G. C., G. R. Gradwell & M. P. Hassell. 1973. Insect population ecology: an
analitical approach. Blackwell Sci. Publ., Oxford.
Vinis, G. 1977. The spindle-berry scale, Unaspis euonymi Comst. in Hungary.
Novenyvedelem. 13: 5-10.
Wagner, T. L. & J. L. Willers. 1992. The role and relationship of the database with
insect population models, pp. 9-22. In J. L. Goodenough and J. M. McKinion
[eds.], Basics of insect modeling. American Society of Agricultural Engineers,
St. Joseph.
Wagner, T. L, Hsin-I. Wu, P. J. H. Sharpe, R. M. Schoolfield & R. N. Coulson. 1984.
Modeling insect development rates: a literature review and application of
biophysical model. Ann. Entomol. Soc. Am. 77: 208-225.

117
Walker, T. J., J. J. Gaffney, A. W. Kidder & A. B. Ziffer. 1993. Florida Reach-Ins:
environmental chambers for entomological research. Amer. Entomol. 39:
177-182.
Wang, D. W. & B. Q. Chen. 1989. Observations on the behaviour of arrowhead
scale, Unaspis yanonensis Kuwana (Horn.: Diaspididae) crawlers. Chin. J.
Biol. Cont. 5: 76-78.
Washburn, J. O. & G. W. Frankie. 1981. Dispersal of a scale insect, Pulvinariella
mesembryanthemi (Homoptera: Coccoidea) on iceplant in California. Environ.
Entomol. 10: 724-727.
Waterhouse, D. F. & K. R. Norris. 1987. Biological control: Pacific prospects. Inkata
Press, Melbourne.
Williams, D. J. & G. W. Watson. 1988. The scale insects of the tropical South Pacific
region. (Part 1. The armoured scales (Diaspididae)). C.A.B., Wallingford, UK.
Williams, M. L, C. H. Ray & I. E. Daniels. 1977. The euonymus scale in Alabama.
Highlights Agrie. Res. 24: 15.
Xinnian, Z. & H. W. Browning. 1991. Studies on crawler behavior of citrus snow
scale. Acta Agriculturae Jiangxi 3: 132-140.
Ziegler, L. W. 1949. The possible truer status of the red-headed scale-fungus.
Florida Entomol. 32: 151-157.

BIOGRAPHICAL SKETCH
Julio M. Arias Revern was born in Caracas, Venezuela, in September
29 1960, the son of a Costarrican office clerk and a Venezuelan school
teacher. In 1977 Julio moved to his father's homeland to go to college,
where he also adopted his fathers nationality. He obtained his B.Sc. degree
in Biology in 1982, and his M.Sc. in Biology in 1988 at the University of Costa
Rica. Julio married in 1981 and his son Gabriel was born in August 1984.
Julio later divorced in 1987.
After little less than a year working at an iguana research project in
Panama and Costa Rica, Julio initiated his Ph.D. program at the Department
of Entomology of the University of Florida in January 1989.
118

I certify that I have read this study and that in my opinion it conforms
to acceptable standards of scholarly presentation and is fully adequate, in
scope and quality, as a dissertation for the degree of Doctor of Philosophy.
Harold W. Browning, Cha/r
Associate Professor of Entomology
and Nematology
I certify that I have read this study and that in my opinion it conforms
to acceptable standards of scholarly presentation and is fully adequate, in
scope and quality, as a dissertation for the degree of Doctor of Philosophy.
P. Allen
ssor of Entomology and
Nematology
I certify that I have read this study and that in my opinion it conforms
to acceptable standards of scholarly presentation and is fully adequate, in
scope and quality, as a dissertation for the degree of Doctor of Philosophy.
Q/jftr
Carmine A. Lanciani
Professor of Zoology
I certify that I have read this study and that in my opinion it conforms
to acceptable standards of scholarly presentation and^i fully adequate, in,
scope and quality, as a dissertation for theopOodpr laytor/W. McCoy Jr. /
Professor of Entomology and
Nematology

This dissertation was submitted to the Graduate Faculty of the College
of Agriculture and to the Graduate School and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy.
December, 1995
Os ciitture
Dean, College of Agri
Dean, Graduate School



26
significant for the 2nd stage at both humidities (F = 44.31, df = 1, 160 for 60% RH,
F = 30.66, df = 1, 104 for 70% RH, both P < 0.05), with females showing a shorter
2nd instar (Table 2). No significant differences in sex ratio were found at any of the
environmental conditions (F = 0.33, df = 8, 67; P = 0.95). The average sex ratio
ranged between 2.43 and 4.97 males per female, with a mean of 3.38.
Table 3.3 presents the parameters and R2 values for the fit to the modified
Logan model to the developmental data. It also shows the optimal temperatures
calculated for development of each stage and the estimated developmental
threshold. Sample sizes of later instars were reduced by mortality in previous
stages, yielding less precise estimation of the fitted curves. The optimal
temperatures for development were between 25 and 38 C for most stages, but the
thresholds varied widely from 0.08 to 18.2C, with the extreme values being quite
high and clearly unrealistic estimates.
Tables 3.4 and 3.5 present the effects of temperature and relative humidity
respectively, on survival of snow scale. There were significant differences between
the survival curves at different temperatures for all of the stages except adult
females (Table 3.4), and relative humidity affected 1st instars at temperatures
>16C, 2nd instars only at 16C and males at 24 and 28C.
Table 3.6 presents the parameters for the fit of a gamma distribution to the
probability density function of the survival curves. Data were pooled for older
stages (males and females). Table 3.7 presents the regression of the parameters
a and p of the gamma distribution versus temperature using polynomial equations.


114
. 1979. Species of Aphytis of the World (Hymenoptera: Aphelinidae). Series
entomolgica. Dr. W.Junk BV, The Hague.
Ruesink, W. G. 1975. Analysis and modeling in pest management, pp. 353-376. In
R. L. Metcalf and W. H. Luckmann [eds.], Introduction to insect pest
management. J.Wiley & Sons, New York.
. 1976. Status of the systems approach to pest management. Ann. Rev.
Entomol. 21: 27-44.
Sadof, C. S. & J. J. Neal. 1993. Use of host plant resources by the euonymus scale,
Unaspis euonymi (Homoptera: Diaspididae). Ann. Entomol. Soc. Am. 86:
614-620.
Sakagami, Y. & R. Korenaga. 1982. Studies on forecasting the occurrence of the
arrowhead scale, Unaspis yanonensis Kuwana. II. Forecasting initial
occurrence date of the first instar larvae by multiple regression equation. Bull.
Fruit Tree Res. Stn. Ser. B (Okitsu). 9: 35-51.
Samson, R. A., H. C. Evans & J. P. Latg. 1988. Atlas of entomopathogenic fungi.
Springer-Verlag, Berlin.
SAS Institute Inc. 1989. SAS/STAT User's guide version 6., 4th. ed. 2. SAS Institute
Inc., Cory, NC.
Selhime, A. G. & R. F. Brooks. 1977. Biological control of some armored scale
insects on citrus, pp. 475-478. In W. Grierson [ed.], Proceedings of the
International Society of Citriculture. Organizing committee, Orlando, Florida.
Sevacherian, V., V. M. Stern & A. J. Mueller. 1977. Heat accumulation for timing
Lygus control measures in a Safflower-Cotton complex. J. Econ. Entomol.
70: 399-402.
Sharpe, P. J. H. & D. W. DeMichele. 1977. Reaction kinetics of poikilotherm
development. J. Theor. Biol. 64: 649-670.
Sharpe, P. J. H., G. L. Curry, D. W. DeMichele & C. L. Cole. 1977. Distribution
model of organism development times. J. Theor. Biol. 66: 21-38.
Siegel, S. 1956. Non parametric statistics for the behavioral sciences. McGraw-Hill,
New York.


112
McClure, M. S. 1990a. Ecology: Influence of environmental factors, pp. 319-329. In
D. Rosen [ed ], Armored scale insects: their biology, natural enemies and
control. Elsevier, Amsterdam.
. 1990b. Life-tables, models and population dynamics, pp. 331-337. In D.
Rosen [ed.], Armored scale insects. Their biology, natural enemies and
control. Elsevier, Amsterdam.
McKinion, J. M. 1992. Getting started: basics of modeling strategies, pp. 1-8. In J.
L. Goodenough and J. M. McKinion [eds.], Basics of insect modeling.
American Asociation of Agricultural Engineers, St. Joseph.
Mead, F. W. 1976. Economic insects in Florida. Florida Department of Agriculture
and Consumer Services, Division of Plant Industries, Gainesville.
Miller, D. R. & M. Kostarab. 1979. Recent advances in the study of scale insects.
Ann. Rev. Entomol. 24: 1-27.
Murakami, Y., H. Uematsu, S. Ohga & H. Kajita. 1972. Parasites of Unaspis
yanonensls in Japan (Hymenoptera, Chalcidoidea). Mushi 46: 45-52.
Murdoch, W. W., R. F. Luck, S. J. Walde, J. D. Reeve & D. S. Yu. 1989. A refuge
for red scale under control by Aphytis: structural aspects. Ecology 70: 1707-
1714.
Nishino, M. 1974. Studies on biology and forecasting of occurrence of the
arrowhead scale Unaspis yanonensis Kuwana. Bull. Shizuoka Prefect. Citrus
Exp. Sta. Special Bulletin No 2: 1-101.
Nishino, M. & K. Furuhashi. 1971a. Studies on forecasting the occurrence of the
arrowhead scale scale, Unaspis yanonensis, Kuwana. IV. Influence of
temperature for development of eggs in ovaries of the adult female. Bull.
Shizuoka Prefect. Citrus Exp. Sta. 9: 115-132.
. 1971b. Studies on forecasting the occurrence of the arrowhead scale,
Unaspis yanonensis Kuwana. III. The period, type and number of emergence
of the 1st instar larvae on the 3rd generation. Bull. Shizuoka Prefect. Citrus
Exp. Sta. 9: 97-114.
Odum, E. P. 1971. Fundamentals of ecology, 3rd. ed. Saunders, Philadelphia.
Ohkubo, N. 1981. Role of petroleum oil sprays in an integrated pest management
system of citrus crops in Japan, pp. 611-614. In K. Matsumoto [ed.],
Proceedings of the International Society of Citriculture. Organizing
committee, Tokyo, Japan.


31
stages are lower than the typical temperatures occurring during the summer in
central Florida citrus groves (29C average of daily mean temperatures from 20
June to 22 September 1992, 37C mean of maximum temperatures during the same
period), which supports the observation that snow scale populations do not do well
during the hottest months of the year. One is also led to speculate that the
geographical areas where snow scale may have originated and thus may be best
adapted (i.e. higher latitudes or higher altitudes), experience lower mean
temperatures than Florida citrus regions.
Casares (1974) cultured citrus snow scales on fruits in a range of
temperatures between 12.5 and 29.4C, and concluded that the lowest temperature
at which the snow scale could develop was 18.3C. Insects did not develop at
12.5C, but survived and could resume development when higher temperature
conditions occurred. In our research, scale development was completed at 16C.
Thus, the lower threshold for development on grapefruit leaves must occur below
this temperature. Thresholds estimated from the modified Logan model yield a wide
range of results, including some > 16C (Table 3.3), but many of them approach
12C as observed by Casares (1974). In our studies, the extreme threshold
estimates were derived from prepupal and pupal stages (males), which were based
on reduced sample sizes caused by mortality in early instars. The estimates also
were limited in precision by the length of the stadium approximating the interval
between observations. To obtain a more precise estimate of the developmental
threshold for these stages, one must generate more data at the low end of the
temperature range, with more frequent observations.


39
attributed to predators that could chew or break through the armor, such as
coccinellids. It is impossible to measure total predation. Predation can be
underestimated, particularly when the scale armor or body of early nymphs
is totally removed. On the other hand, overestimation of predation may occur
when the disturbed armor did not contain live prey, or if the damage inflicted
to the cover is due to scavengers. Encarsia-parasitized female scales
attacked by predators before the parasite emerged also were observed.
However, these were classified as death by parasitism since they would not
survive. It is possible that the same phenomenon occurred with males, but
it was impossible to observe.
4- Fungi: White mycelial growth was observed in several instances growing
on individuals that had died. Unfortunately, it was impossible to determine
if the insect was killed by the fungus or was colonized after dying by another
cause. The most common fungi in snow scale colonies are from the genus
Nectria (teleomorph of Fusarium) but their role is unclear. There is confusion
over whether those fungi are acting as pathogens or as saprophytes (Ziegler
1949). Kuno & Coln-Ferrer (1973) showed increased crawler mortality but
saprophytic behavior on older stages. As with predation, death by diseases
is being underestimated. Deaths that occurred under the scale cover and did
not show mycelium or fructiferous bodies within the observation period were
unaccounted. Given the impossibility to clear the issue of pathogenicity, we
assumed any fungal appearance to be pathogenic and causing death to the
scale with which it was associated.


36
populations, and using life table analysis to separate and weight the importance of
the different mortality factors.
Materials and Methods
Data Recording with Photographs. Given the small size of the snow scale and
the presence of an armor, direct observations in the field are difficult. Sampling of
scale insects usually involves taking a sample of substrate to the laboratory and
dissecting it under magnification. This destructive sampling does not permit the
study of the fate of those individuals alive at the moment of sampling, that may die
later of any cause. Collection and analysis of the remains of the insects are
unreliable because it is difficult to establish the time of death, since the remains of
the insect stay in place on the substrate for variable periods. Also, some causes of
death do not leave tangible host remains. Direct counting in the field is difficult and
inaccurate.
To overcome these difficulties, I developed a method for direct observation,
recording on photographs the events occurring in a snow scale colony under field
conditions. This method was modified from Summy et al. (1984) and Terry &
Edwards (1989), and chosen because is more accurate than direct counts using
hand lenses. It allows for nondestructive sampling and the following of specific
individuals during their development.
Data were collected over a 2-year period between December 1991 and
December 1993. Active colonies were located on the trunk or main branches of
'Valencia' orange trees in a commercial grove in Lake Alfred, Polk County, Florida,


7.1 Parameters used for the simulation of development for
the citrus snow scale 88
7.2 Average duration and percentage of survival from each
stage at each environmental condition studied 89
7.3 Parameters for the distributed delay model, calculated
from data at constant temperature and RH 90
7.4 Linear regression between attrition rate (AR) and
temperature (T) 91
7.5 Parameter values, standard error and confidence
intervals at 95% for development parameters 92
7.6 Parameter values, standard error and confidence
intervals at 95% for mortality parameters 93
viii


8
and other relevant factors. The relationship between temperature and development
has been described in many ways, including linear relationships (Baskerville & Emin
1969, Sevacherian et al. 1977), and curves and asymmetrical functions (Logan et
al. 1976, Sharpe & DeMichele 1977, Taylor 1981, Wagner et al. 1984). The
developmental rate increases with an increase of temperature. Development stops
below a lower and above an upper temperature threshold.
Dickens (1968) studied duration of development of the citrus snow scale on
'Valencia' orange fruits, at a constant temperature of 26C. He found that first instar
development may require around 13 days, while the duration of the second instar
is about 11 days for males and 18 days for females. Females started producing
eggs about 60 days after they were born. Casares (1974) studied citrus snow scale
development on lemon fruits under several constant temperatures, and confirmed
that developmental rate increased with an increase in temperature. These efforts
to describe citrus snow scale development were made on citrus fruits, which are not
the preferred substrata under field conditions.
Development of U. yanonensis has been studied extensively, including
development under constant temperatures (Okudai et al. 1971, 1974, Huang et al.
1983) and fluctuating temperatures (Korenaga et al. 1976). The optimal
temperature for development was calculated to be about 27C and the lower
developmental threshold was near 10C (Okudai et al. 1971, 1974). Studies also
addressed the relationship between temperature and ovarial development,
forecasting the appearance of the first instar (Nishino & Furuhashi 1971a, b, Okudai
et al. 1975). Nishino (1974) has summarized most of this research. Furuhashi


93
Table 7.6 Parameter values, standard error and confidence intervals at 95% for
mortality parameters.
Stage
Parameter
Value
Confidence limits
OlU. ciiUi
lower
upper
1st stage
intercept
-0.049
0.038
-0.156
0.057
slope
0.007
0.002
0.003
0.012
2nd stage
intercept
132.611
31.534
45.072
220.150
?
slope
-4.017
1.336
-7.726
-0.309
Males
intercept
-0.030
0.033
-0.122
0.061
slope
0.002
0.001
-0.002
0.005
Adult
intercept
-0.010
0.010
-0.038
0.018
females
slope
0.001
0.000
-0.001
0.002
greenhouse. Correspondence of the simulation to the data was not good. Figure
7.2 shows overestimation in all the stages except second nymphal females.
Discussion
The model for development of citrus snow scale under constant temperatures
approximates the experimental data. However, problems of estimation arose.
Populations were more accurately estimated at intermediate temperatures and early
in the development of the cohort, while under- and over-estimation occurred at high
and low temperatures and in later life stages. The potential reasons for the lack of
correspondence could be faults in the conceptualization of the model, faults in the
programming or inaccuracy in the estimation of the development or mortality
parameters, or inaccuracy in the development data.


150
75
0
135
90
45
0
225
150
75
0
120
80
40
0
225
150
75
0
jre 7
94
16C, 60%RH
21C, 60%RH
24C, 60%RH
i i i
28C, 60%RH
1 T
30C, 60%RH
150
16C, 70%RH
21 C, 70%RH
24C, 70%RH
28C, 70%RH
0

0
O
20 40 60 80
Nynph I
Female Nymph II
Adult females
Males
0 20 40 60 80
Time (days)
Duration of development of citrus snow scale individuals under
>nstant temperature. Circles: observed data. Lines: simulated values.


48
Table 4.4. Total percentage of mortality (adx) and percent mortality by unknown
causes against parasitism and detectable predation in multiple decrement life
tables, calculated by tree.
Tree
Date
adx
Detectable
predation
Parasitism
Unknown
causes
36-2
17-Dec-91
0.7939
0.0153
0.0763
0.6718
36-7
17-Dec-91
0.8387
0.0968
0.1452
0.5968
35-15
23-Jan-92
0.8333
0.0000
0.2708
0.5000
35-18
06-Mar-92
0.9355
0.2903
0.1935
0.4516
34-21
14-Mar-92
0.9775
0.1910
0.2022
0.5730
35-17
27-Apr-92
0.8106
0.0303
0.1212
0.6591
34-19
19-May-92
0.9464
0.2857
0.0179
0.6429
32-15
17-Jun-92
1.0000
0.1707
0.0000
0.8293
34-11
29-Jul-92
0.8000
0.3000
0.0000
0.5000
26-9
24-Aug-92
1.0000
0.2647
0.3824
0.3529
28-7
24-Aug-92
0.7250
0.2250
0.0500
0.4500
26-16
12-Oct-92
1.0000
0.0645
0.2903
0.6452
27-14
26-Oct-92
0.8333
0.1250
0.0833
0.6250
28-14
26-Oct-92
0.9412
0.2353
0.1765
0.5294
24-2
30-Apr-93
0.6970
0.3333
0.0000
0.3636
23-13
24-May-93
0.8868
0.2453
0.0000
0.6415
24-12
24-May-93
0.7576
0.0909
0.0000
0.5152
22-9
17-Aug-93
0.9231
0.1154
0.0000
0.7692
25-6
16-Sep-93
0.9024
0.4878
0.1220
0.2927
21-12
06-Oct-93
0.9500
0.0500
0.0000
0.9000
22-14
06-Oct-93
0.7500
0.0000
0.1250
0.6250


80
the highest crawler catches. The maps of progress of scale dispersal (Figure 6.3)
do not provide compelling evidence for the relationship between wind direction and
scale spread, since we lack the wind data, but indicate a general movement toward
the southwest.
Quayle (1916) studied dispersal of black scale (Saissetia oleae) in citrus
groves in a manner similar to that described here. He trapped black scale crawlers
450 feet (137 m) from the source, the farthest traps that he had set. He also
trapped Aonidiella aurantii at 30 to 150 feet (9 to 45 m) from a source of crawlers.
The distance that an airborne object can travel depends on several factors
such as wind speed, the height from which the object starts its descent, and the
speed at which its falls (Margalef 1977). An additional consideration for the snow
scale is the fact that the more dense scale colonies occur in the interior part of the
tree (trunk, scaffold branches) where air movement is somewhat restricted and the
canopy becomes a barrier for dispersion. Crawlers lifted by the wind from the trunk
must overcome that barrier. It is expected that young trees with heavy infestations
and sparse canopy will be better sources for short distance colonization, since winds
close to the ground are slower and the height of the dispersing crawlers is limited
in short trees. On the other hand, larger trees with a dense canopy should not be
a good source of crawlers, but those few crawlers that are potentially lifted by wind
from a high branch will have the opportunity to travel further.
Xinnian & Browning (1991) studied the movements of snow scale crawlers
and concluded that they were positively phototactic and suspected but did not
provide conclusive evidence for the effects of gravity. They also observed higher


Number of crawlers caught Number of crawlers caught
Direction (degrees)
73
0 90 180 270
16
14
12
10
8
6
4
2
0
North East South West
North East South West
Figure 6.1. Wind speed vs direction and catches of snow scale crawlers
A-Trial 1, May 31 to June 3 1993. B Trial 3, July 28 to August 2 1993.
Average wind speed (Km/h) Average wind speed (Km/h)


61
The missing points correspond to samples where no gravid females were found
(except Nov., 1992 where no live scales were was found on twigs).
The sex ratios on twigs and leaves are presented in Figure 5.3. In contrast
to previous measurements reported in the literature (Casares 1974), observations
made in lab cultures (Chapter 3) and on tree trunks in the field (Chapter 4), the sex
ratio was close to 1.8 females per male on leaves (average 36% males), with
variation ranging from 14-49% males. The proportion of females on twigs was
greater (2.2 females: 1 male, average 0.31 males), with values ranging 0-53%
males. A greater proportion of males was observed in the late spring-early summer
months on twigs and leaves in the first year and earlier on leaves during the second
year (1993).
The incidence of parasitoids and fungi followed similar patterns on leaves and
twigs, especially those associated with female scales (Figure 5.4). Fungi and mites
occurred only sporadically. High incidence of internal parasitism occurred in
October 1992 and late in 1993 (October-November). Encarsia spp. were abundant
on males over most of the observation period, with drops in July 1992 and May to
August 1993. The incidence of ectoparasitoids (Aphytis) was usually higher on
leaves with peaks in April 1992 and September to December 1993 but some of this
increase represents low numbers of host stages encountered.
The incidence of hymenopterous parasitoids, fungi and mites are presented
in Figure 5.5, as relative proportions calculated from the totals of attacked
individuals observed at each sampling date. Internal parasitoids are predominant
except on August 1993 and on October 1993 (twigs). No internal parasitoid attacks


5
Prontaspis citri (Comstock) MacGillivray 1921 and Dinaspis veitchi Green & Laing
1923. Ferris (1937) assigned those synonyms to Unaspis citri.
Unaspis citri is recorded from Citrus and related genera in the family
Rutaceae and occurs in China, Indochina, Australia, Central Africa, South America
and the Caribbean (UK 1962). The same source reports this species from the citrus
areas of North America, but more recent information restricts it to Florida and
Louisiana (Browning 1994). In addition to U. citri, the two other Unaspis species of
economic importance are the arrowhead scale, U. yanonensis and the euonymus
scale, U. euonymi. Unaspis yanonensis infests Citrus in Japan, China and southern
France, but does not occur in the Americas or Australia (UK 1988). Unaspis
euonymi attacks ornamental plants in the genus Euonymus (Celastraceae), Prunus
(Pomaceae) and Hibiscus (Malvaceae). It occurs in Asia (Japan, China), Europe
and North America (UK 1970). The remaining species, Unaspis acuminata, U.
atricolor, U. flava and U. permutans were described from material collected from
various host plants in Sri Lanka and the south and east of India, while U. turpiniae
was described from material collected in the Philippines (Rao 1949).
The diagnostic characteristic that separates U. citri from the other species of
the genus is the reduction or absence of perivulvar glands in mature females (Rao
1949). Perivulvar glands are associated with ovoviviparity. Although some degree
of ovoviviparity occurs in most diaspidid scale insects, the bounds of this
phenomenon are not well delimited. Prevailing theory indicates that part or all of
the embryo's development could occur internally in the mother of an ovoviviparous
species. Conversely, the presence of an egg shell or chorion is considered to be


86
change between stages (Equation 7.1) (Vansickle 1977, Curry & Feldman 1987).
Linear regression was fitted to express the relationship between the proportion dying
and temperature for each stage simulated.
Reproduction and Migration. Reproduction and migration are not considered
in this model because they have not been adequately measured. Since
reproduction occurs beneath the mothers secreted cover, it is not possible to obtain
direct observations. Also, the patterns of movement of crawlers is such that newly
emerged crawlers walk away from their mothers (Xinnian & Browning 1991).
Reproduction could be irrelevant if a small area, such as the 18x12 mm2 patches of
bark surface used in chapter 5, are considered for the simulation. Most of the newly
emerged crawlers move beyond this area. On the other hand, if the simulation
considers a larger scale, such as a branch or a whole tree, then reproduction
becomes relevant. Inversely, migration will be the main component of recruitment
on a small scale, but probably quite irrelevant at large scale (Chapter 6).
Recruitment by migration was studied from field observations (Chapter 5) and the
analysis of photo slides, but was not used to estimate the recruitment of new scales.
Environmental Variables. Temperature was the only environmental variable
incorporated into the model. It was initially set constant for development of the
model and for comparison to the constant temperature developmental data
presented in Chapter 2. Temperature then was modeled as a compound cosine
function after Brewster & Allen (1991):


Y Coordinate
April 1994 May 1994
4 8 12 16
July 1994
15 -
10 -
5 -
I I I I
4 8 12 16
4 8 12
August 1994
4 8 12
September 1994
4 8 12 16
X Coordinate
Figure 6.3 -- Continuation.


ACKNOWLEDGMENTS
This research would have not been possible without the guidance, critical
evaluation, encouragement and financing by the committee chair, Dr. Harold W.
Browning. I appreciated especially his immense patience. Very valuable input was
received from the other members of my advisory committee. Dr. Jon C. Allen
helped me understand the intricacies of computer modeling, Dr. Carmine A. Lanciani
advised on life table analysis during the planning phase of this project, Dr. Clay W.
McCoy taught me about insect pathology and the complexities of the researcher's
occupation. Dr. Fred Bennett was instructive about biological control and quarantine
procedures, and helpful while he was part of the committee, until the day of his
retirement.
This research was possible also by the kindness of Mr. Maurice Patrick who
gave me the liberty to do and undo at the citrus plots of his property. During the
work for this research I got invaluable help from Mrs. Pamela Russ, Mr. Ian Jackson
and Mr. Mark Bryan and had the opportunity to share their friendship. Many people
not directly involved in the development of this research need to be thanked,
because they greatly contributed to my well being during my stays in Gainesville and
Lake Alfred: these include Jackie J. Belwood, Vinnod Kutty, the Nielsen family, Faith
and David Oi, Eliane Quntela, Devesh Singh, Hugh Smith and Laurie Wilkins:


41
8- Completed life cycle: Individuals that reached the adult size and
appearance and were not victims of any of the previous mortality causes
were classified as completing their life cycle. The female reproductive period
was not isolated for analysis since it occurs beneath the armor, making it
impossible to observe.
Statistical Analysis. Life table analyses were used to estimate mortality
trends in populations of citrus snow scale on chosen patches. Single decrement life
tables were generated (PROC LIFETEST, SAS Institute Inc. 1989). Comparisons
were made between patches in each tree and between trees in the whole sample
site. Multiple decrement life tables were built using the mortality categories listed
and following the procedures and notation explained in Carey (1993):
K*=
Dix=
D =
aqix= Dfc/K,
aqx= DJK
a\ = alx-1(1 -aqx)
Number of individuals beginning stage x
Number of individuals dying of cause / at stage x
Total deaths in stage x
Fraction of deaths from cause /' in stage x in the
presence of all other causes, given that the individual is
alive at beginning of stage x
Fraction of deaths from all causes in stage x, given that
the individual is alive at beginning of stage x (£aqix)
Fraction of survivors at age x out of the original cohort,
adix= alx(aqix)
which is assigned al^l
Fraction of deaths in stage x from cause / among alx
living at stage x


117
Walker, T. J., J. J. Gaffney, A. W. Kidder & A. B. Ziffer. 1993. Florida Reach-Ins:
environmental chambers for entomological research. Amer. Entomol. 39:
177-182.
Wang, D. W. & B. Q. Chen. 1989. Observations on the behaviour of arrowhead
scale, Unaspis yanonensis Kuwana (Horn.: Diaspididae) crawlers. Chin. J.
Biol. Cont. 5: 76-78.
Washburn, J. O. & G. W. Frankie. 1981. Dispersal of a scale insect, Pulvinariella
mesembryanthemi (Homoptera: Coccoidea) on iceplant in California. Environ.
Entomol. 10: 724-727.
Waterhouse, D. F. & K. R. Norris. 1987. Biological control: Pacific prospects. Inkata
Press, Melbourne.
Williams, D. J. & G. W. Watson. 1988. The scale insects of the tropical South Pacific
region. (Part 1. The armoured scales (Diaspididae)). C.A.B., Wallingford, UK.
Williams, M. L, C. H. Ray & I. E. Daniels. 1977. The euonymus scale in Alabama.
Highlights Agrie. Res. 24: 15.
Xinnian, Z. & H. W. Browning. 1991. Studies on crawler behavior of citrus snow
scale. Acta Agriculturae Jiangxi 3: 132-140.
Ziegler, L. W. 1949. The possible truer status of the red-headed scale-fungus.
Florida Entomol. 32: 151-157.