Biological Control of Insects

1
Biological Control of Insects

• Peter B. McEvoy
• Ent 420/520 Insect Ecology

2
Introduction

• Manipulating natural enemies for pest control by
  introducing, augmenting, or conserving control
  organisms
• Essentially empirical with ample scope for
  improvement through research on the ecology of
  interactions between natural enemies and pests
• Draw theory and practice together, indicating
  where improvements in understanding may
  contribute to improvements in management

3
Early Beginnings

• Beginnings. Began in 1888 with introduction of
  vedaliea beetle (Rodolia cardinalis) into CA from
  AUS for control of cottony cushion scale (Icerya
  purchasi) on citrus (See R.C. Sawyer To Make a
  Spotless Orange)
• Patterns of success. By 1986 (Greathead 1986),
  1162 successful introductions of predators and
  parasitoids
• 25 successfully regulated target pest
• 69 intermittent or partial control (i.e. some
  seasons, generations, cultivars, or climate
  zones)
• 6 failed to provide any control at all

4
Three Approaches to Biological Control

• Three ways to enhance effectiveness of natural
  enemies in pest control
• Classical biological control
• Augmentive biological control
• Conservation of indigenous natural enemies
• Separate subcultures all aimed at the same goal

5
Approaches to Biological ControlConserving and
Augmenting

• Conserving natural enemies by modifying the
  cropping practices
• Increasing vegetation diversity near or in the
  crop (Altieri Letourneau 1984 van Emden 1990)
• Modifying pesticide use (Waage et al 1985 Waage
  1989)
• Augmenting field populations of indigenous
  natural enemies with mass-reared individuals
  released early in season to prevent pest reaching
  damaging levels, or applied in massive and
  repeated applications as biopesticides

6
Approaches to Biological Control Classical
Biological Control

• Exotic pest invades region without their adapted
  natural enemy complex, and, in absence of
  effective natural enemies, reach very high
  population levels
• Control by introducing and establishing effective
  natural enemies from pests area of origin
• Called classical in view of first use in 1800s

7
Compare Biological Control Systems with Natural
Systems

• Many natural arthropod populations are likewise
  regulated by predators, parasites, pathogens
• Classical biological control is simply a special
  case of a general pattern in which populations
  are regulated by density-dependent processes, a
  major class of which involves predator-prey of
  parasitoid-host interactions (DeBach 1974,
  Huffaker and Messenger 1976, Hassell 1978)

8
Challenge for General Theory of Population
Dynamics

• Explanation. Explain when and how natural
  enemies regulate their prey and host populations,
  especially in light of alternative hypotheses
  that deny DD processes are essential (Den Boer,
  1969, Dempster 1983)
• Prediction. Develop methods for selecting and
  evaluating natural enemies that are safe and
  effective

9
Questions for ecology

• Classical biocontrol permits us to compare the
  dynamics of insects with and without (i.e. before
  and after) natural enemies
• Make and test predictions about the role of
  natural enemies in pest population dynamics and
  regulation
• Exotic nature of pest and enemy tends to isolate
  their interaction in new environment,
  facilitating application of simple predator-prey
  models as a way to understand processes and
  generate predictions

10
Traditional Paradigm
11
Traditional Paradigm

• Natural enemy reduces pest density to new, low,
  stable, equilibrium well below that that
  prevailed before introduction (Fig 1)
• Beddington et al. 1978 analyzed 6 projects and
  reached one conclusion (stable pest density lt 2
  of that which prevailed before)
• Murdoch et al. 1985 analyzed same 6 projects plus
  3 others and reached an opposite conclusion (only
  1 of 9 evinced stable equilibrium)
• So where does that leave us? Equilibrium,
  stability, and related concepts depend on scale,
  the perspective of the observer. Possible to
  have success without stability, at least on local
  spatial scale.

12
Steps in a Classical Biocontrol Program

• Evaluate the pest problem in the target region
  for the biocontrol program. Establish taxonomic
  identity of pest and area of origin.
• Foreign exploration for the pest in the area of
  origin. Surveys to assess the complex of natural
  enemies of the pest, their impact and degree of
  specialization.
• Selection of enemies from this complex for
  importation and establishment in the target
  region.
• Quarantine for removing hyperparasitoids, plant
  pathogens and insect pathogens from culture
• Release natural enemies cleared from quarantine
  in the target region.
• If agents establish, monitor change of the
  natural enemy and pest population

13
Patterns of Classical Biological Control Success

• Can historical data can provide guidelines for
  future practice as well as ecological insight?
• Database of introductions of insect natural
  enemies against arthropod pest developed at
  International Institute of Biological Control
  (IIBC), called BIOCAT
• Catalogs gt4,000 separate introductions of insect
  biocontrol agents against insect pests (563
  species of enemies against 292 pest species in
  168 countries)

14
Measure of success

• Homoptera have been both the most targeted and
  most successfully controlled group of exotic
  pests (Greathead 1989)
• Their pest prevalence may reflect their feeding
  on the woody stems of perennial crops this
  enhances potential to be transported around the
  tropics and subtropics on propagative plant
  material
• Their control probably reflects the impact that a
  guild of mobile and effective enemies can have on
  relatively sedentary and exposed pest hosts

15
Decision Making in Classical Biological Control

• Process of selection is not simply reconstruction
  of natural enemy complex of a pest in its exotic
  range
• We must be parsimonious
• Avoid introducing natural enemies with unwanted
  side effects
• Avoid running up costs of the program
• Avoid antagonistic interactions detrimental to
  overall control, e.g. facultative
  hyperparasitoids
• Avoid introducing inferior natural enemy that, if
  introduced first, may make subsequent
  introduction of better species more difficult

16
When Biological Control Gets Out of
Control Parasitic fly Compsilura concinnata
brought from Europe to NA for control of gypsy
moth harms native insects including giant silk
moths Boettner et al. 2000 Conservation Biology
141798-1806
D.L. Wagner/University of Connecticut
The population of the Hyalophora cecropia
(Saturniidae), a moth that can grow to half a
foot across, has been in decline.
The Hyalophora cecropia as a caterpillar.
17
Exploration for Potential Biological Control
Agents

• Coevolution.
• Hokkaen and Pimentel (1984) suggest seek exotic
  natural enemies from species related to target
  pest, so natural enemies are less coevolved and
  therefore more virulent.
• Instances of such new associations have led to
  successful biological control
• Empirical evaluation (Waage Greathead 1989,
  Waage 1990 Hokkanen et al.)
• Theoretical evaluation
• Practical conclusion New associations can be as
  effective as old ones once the natural enemy is
  established. However, establishment of an enemy
  on a species it has not previously encountered is
  difficult. Aside from effectiveness, there are
  safety concerns in using less specific natural
  enemies

18
Maximizing coverage of the natural enemy complex

• Survey Area. How much area should we survey to
  encompass geographic variation in natural enemy
  complex? Some sources of variation
• Spatial variation. Geographic variation appears
  to be small (Askew Shaw 1985), and few
  well-separated locations should suffice.
• Host plant factors may affect enemy complexes and
  success of exploration programs.
• Enemy complex can be reduced on host plant where
  range recently expanded
• Enemy complex can vary with crop
• Host density. Composition of enemy complex can
  vary with host density (Price 1973 Mills 1990).
  Natural enemies collected from outbreaks may not
  be the best for maintaining the pest at low
  densities.

19
Mossy Rose Gall In NA and Europe
Diplolepis rosae (Cynipidae)
Rosa eglanteria (Rosaceae)
20
Diagram of a Food Web
6 Top Predator
Black-capped chickadee Parus atricapillus
5 Secondary parasitoids
Gregarious Tetrastichus
Solitary Torymus bedeguaris
4 Primary parasitoids
Parasitoid Orthopelma mediator
Cynipid gall-wasp Diplolepis rosae
3 Herbivore
Sweetbriar rose Rosa eglanteria
2 Plant
1 Resources
Light, water, nutrients
21
Interpreting the abundance of natural enemies

• Ranking predators. Abundance and perhaps size, a
  correlate of feeding potential, often used to
  rank relative importance of predators
• Ranking parasitoids. Proportion of hosts killed
  is used to rank different species of parasitoids.
  
• Total generation mortality caused by enemy often
  poorly estimated by single point estimates of
  parasitism (van Driesche 1983)
• Collection and analysis of life table information
  in area of origin in two programs, one winter
  moth (Operophtera brumata), other larch
  casebearer (Coleophora laricella)

22
Selection of Agents for Introduction

• Impact. Favor those that appear to have
  substantial impact on host population after
  antagonists removed
• Specificity. Exclude generalists that pose risk
  to non-target species in the area of introduction
• Easy to rear. Practical to rear natural enemy for
  quarantine and release
• Other critical attributes. Opportunity to apply
  holistic and reductionist criteria of what is a
  good agent (Waage 1990)

23
Holistic Criteria

• Related to how enemy fits into ecology of its
  pest and other mortality factors acting on it
• Effective at low host density. Seeking enemy in
  low density rather than high density host
  populations
• Synergistic, additive, antagonistic interactions.
  Investigating structure and dynamics of natural
  enemy complex to see how one enemy might be
  influenced by another
• Single vs. Multiple agents. Introducing more
  than one agent (multiple species introductions)
  in preference to single best agent (single
  species introductions)

24
Single vs Multiple AgentsCan antagonistic
interactions among multiple agents lead to
reduced success?

• Enemies spend more time eating each other than
  eating the target pest.
• Enemy exploitation or interference competition.
  Several agents may compete in such a way that
  single best enemy yields stronger suppression
  than multiple agents
• Godfray and Hassell 1987
• May and Hassell 1981
• Kakehashi et al 1984
• Briggs 1993

25
Multiple introductionsusing natural enemies
occupying different feeding niches on host

• Invulnerable stages. Maximize proportion of pest
  life cycle that is vulnerable to natural enemies
• Antagonistic interactions. Keep in mind dynamic
  natural of enemy complexes and possible strong
  interactions among enemies acting at different
  stages in pest life cycle
• Compensatory responses. Strong density-dependent
  mortalities will tend to have substantial
  negative effects on the impact of natural enemies
  that precede them (May et al. 1981)
• As a practical matter, positive effects of
  multiple introductions outweigh the alternative
  of trying to introduce the best agent the first
  time. Stopping after the first introduction
  just in case it was the best species, and just in
  case the subsequent introduction would reduce the
  efficacy of control, is not justified on field
  experience. (Waage and Mills 1992)

26
Reductionist criteria for agent selection

• Models. Draw on simple predator-prey models for
  clues to factors that affect pest suppression and
  regulation
• Criteria. Searching efficiency (or area of
  discovery), handling time, aggregation, mutual
  interference have all been suggested as criteria
• Behavioral approach. For example, compare
  behavioral responses of candidates to see which
  has highest search efficiency

27
Refuge Theory Now the Outcome of Biocontrol
(Y) Can Be Predicted From a Single, Easily
Measured Parameter (X) (Hold the hype)
Hawkins et al 1994
28
Why are Reductionist Criteria Rarely Used?

• Phenomenological not mechanistic. Some parameters
  hard to estimate or even imagine in field
• Unrealistic estimates of parameters. Estimates
  measured in the lab unlikely to apply in the
  field (e.g. searching efficiency and aggregation)
• Correlations among characters. Attributes are not
  independent but are positively or negatively
  correlated, e.g. fecundity and efficiency of
  parasitoids often traded off against competitive
  ability in a host (Mat et al. 1981)

29
Whole organism, not component parts, forms basis
for predicting success. So what to do

• Establish patterns of correlation in nature
• Incorporate realistic combinations in models
• Emulate models of this kind applied to winter
  moth (Hassell 1980), red scale (Murdoch et al.
  1987), cassava mealybug (Gutierrez et al. 1988),
  and mango mealybug (Godfray Waage 1991)
• Acknowledge mostly retrospective so far, but
  potentially predictive

30
Constraints on testing hypotheses using
biological control systems.Sociology of
Research Who benefits, Who Pays

• Who wants to compare selected and rejected agents
  in different but comparable parts of pests
  exotic range?
• We must consider sensibilities of an affected
  country, where rapid and effective solution to
  serious pest problem is required
• One would be reluctant to risk further losses so
  that investigation could be guinea pig for
  refinement of scientific methods
• Possibilities for adaptive management?

31
Local Case Study

• Biocontrol of European larch casebearer in Oregon
  (Ryan 1990)
• Two Parasitoids
• Agathis pumila (Braconidae)
• Chrysocharis laricinellae (Eulophidae)
• Host Insect
• Coleophora laricella (Lep Coleophoridae)
• Host plant Larix

32
Time Series Regional Means
33
Life Cycle Large Casebearer
Eggs laid singly in Jn-Jl, egg stage last about 1
month L-1 and L-2 mine needle, attack by
Agathis L-3 forms case from hollow
needle Larvae attaches to twig to
overwinter L-4 feeds on new needles in
spring Pupa mid May- mid June Adult May early
July
A
E
Chrysocharis laricinellae
L4
L2-3
Agathis pumila
34
General Approach Key Factor Analysis

• Study life history and develop methods of census
  for each stage
• Construct a life table that is as complete as
  possible, expressing the "killing power" of
  mortality factors as k-values
• Accumulate many life tables
• Plot generation curves and mortalities
• Assess the key-factors which make the biggest
  contribution to change in generation mortality
• Determine the relationship of component
  mortalities to density
• Follow up with intensive studies of key factors
• Make predictions using the model

35
Life Tables with 10 k-values

• k1 Infertility
• k2 Egg predation
• K3 Embryo Death
• K4 Parasitism by Agathis pumila
• K5 Mortality of needle-mining larvae
• K6 Mortality of fall, case-bearing larvae
• K7 Mortality of winter and small, spring
  case-bearing larvae
• K8 Parasite-induced morality of large, spring
  case-bearing larvae and pupae by species other
  than A. pumila
• K9 Sex Ratio
• K10 Adult mortality, reduced fecundity, and
  emigration

36
K4 Parasitism by Agathis is a Key Factor
Variation in k4 closely correlated with variation
in generation mortality
37
Test for Density Dependence

• Strength
• 2. Sign
• 3. Time Delay

1. Strength2. Sign3. Time Delay
38
Graph each k-value against log density of stage
on which it acts
Pattern across three locations
39
Time-Delayed Parasitism by Agathis
40
Approaches differ in insect and weed biocontrol

• Insect biocontrol.
• Mathematical models as metaphors of the
  parasitoid-host interaction to identify processes
  that regulate host populations
• Studies of lab and field populations using life
  table analysis to identify DD processes
  associated with regulating pest populations
• Assumes that effective biocontrol,
  density-dependence and host population regulation
  are linked
• Experimental manipulations (Luck et al 1988)
  championed as alternative to life analysis, but
  experiments and life tables can be combined in
  LTRE

41
Nicholson-Bailey Model Assumptions Hassell 1978

• Host and parasitoid populations have
  non-overlapping generations
• Parasitoid population randomly searches all areas
  containing hosts
• Every host with the population has the same risk
  of attack
• Only one egg matures per host
• Every time a host is encountered, it is
  parasitized, even if it has been previously
  parasitized
• Only the supernumerary eggs die

42
Reconciling behavior of natural and model systems

• Confront the possible with the actual.
  Introduction of a parasitoid can regulate a host
  population, whereas modeled interaction cannot
• Motivates search for stabilizing mechanisms.
  Modifications that might cause individual hosts
  to vary in risk of attack and stabilize model
  interactions
• Aggregation of parasitoid population at denser
  host patches or independent of host density
• Refuges for a portion of the host population
• Decrease in parasitism with increasing host
  density within each generation
• Asynchrony between parasitoid and host
  populations
• Sex ratio variation in the parasitoid population
  via local parental control or with increasing
  parasitoid density
• General class of stabilizing mechanisms.
  Aggregation of risk promotes stability at price
  of higher equilibrium density of the host
  population

43
Criteria for identifying successful biocontrol
(Strong et al. 1984, Huffaker et al 1976, Waage
and Hassell 1982)

• Synchrony or slight asynchrony with the host
  population
• High enemy intrinsic rate of increase relative to
  that of the host
• High enemy searching efficiency
• Interference (interspecific competition) amongst
  the parasitoids
• Aggregation of enemy on host patches
• Significant enemy dispersal ability

44
The Case for Behavioral Studies(Luck 1990)

• How are parasitoid preference and performance
  related to individual host quality?
• Reproductive potential in relation to host size
• Host selection, oviposition, sex ratio in
  relation to host size
• Host quality in relation to host density, plant
  cultivar, location on plant, temperature
  conditions
• If parasitized hosts stop growing (idiobionts).
  Evaluating current resoureces (host density and
  size distribution) allows host population in the
  field to be evaluated from parasitoids
  perspective
• If parasitized hosts continue to grow
  (koinobionts). Requires evaluation of current and
  future resources. Host choice may be related to
  minimizing the risk of intraspecific and
  interspecific competition, e.g. by reducing time
  of offspring within host rather than maximizing
  their size and fecundity
• Value to biocontrol. Helps us assess temporal and
  spatial availability of host resources in the
  field. Links individual behavior with population
  growth of the parasitoid remaining challenge to
  do the same for the host

45
Predictive Modeling in Biological ControlGodfray
and Waage 1991

• Practical situation. Foreign exploration
  invariably yields a number of candidates
• Apply practical criteria. After screening
  candidate by very practical criteria
• easy to rear
• sufficiently host specific
• What then? What criteria can be applied to judge
  potential effectiveness of the remaining
  candidates?

46
Mango Mealy Bug

• Natural History.
• African pest. Mango mealybug Rastrococcous
  invadens (Hemiptera Pseudococcidae) is a pest of
  mango and citrus in West Africa
• Generalist. Pest reported from gt 44 plant species
  in 22 separate plant families. Copious honeydew?
  sooty molds.
• Candidate enemies. Two parasitoids (Encyrtidae)
  collected in India considered as candidates
• Timing of attack. Gyranusoidea tebygi attacks
  young mealy bugs of both sexes while Anagyrus
  mangicola attack older, female insects
• Specific application. Model predicts one
  parasitoid (Gyranusoidea tebygi) will lead to
  greater decrease in host density than other
  parasitoid (Anagyrus mangicola).
• General application. Decision to release G.
  tebygi made before model developed and they take
  not credit for it. Nevertheless, they believe
  models of intermediate complexity offer promise
  for predicting biocontrol

47
Life Cycle Stages Host
Parasitoids
Rastrococcous invadens
Gyranuscoidea. tebygi
Adults (A)
1st stage (F)
Males
Immatures (J)
Males
2nd stage (S)
Adults (A)
Males
Adults (R)
Immatures (J)
Anagyrus mangicola
48
Prospective Modeling of Mango Mealybug

• Easily measured parameters collected in a few
  months of field and lab study
• Easy Parameters include stage of host attacked by
  different parasitoids, age-specific development
  rates for hosts and parasitoids, age-specific
  survivorship of hosts in the field, and adult
  longevities and daily oviposition rates
• Difficult parameters such as search efficiency
  treated as variables
• Predictive Power. Model predicted superiority of
  one of two potential control agents
• Quick and Cheap. While the aim to be accurate,
  such models cannot be the product of many years
  of careful study and must be built quickly and
  cheaply if they are to be useful

49
Pest Suppression
50
Release and Assessment of Selected Agents

• Allocation of effort. Investment in selection
  must be matched by effort in establishment and
  evaluation
• Experimental methods for evaluation of
  effectiveness of natural enemies (Luck et al
  1988)
• Well-quantified assessment of larch casebearer in
  the PNW (Ryan 1990)

51
Why simple analytical models have little if any
real application in biological control

• Non-Independent Characters. Models characterize
  enemy in terms of a number of independent and
  desirable life history traits have little
  practical value. Real organisms constrained by
  pattern of variance and covariance among
  characters.
• Hard to Measure. Key parameters identified by
  analytical models such as searching efficiency,
  handling time, and aggregation are very difficult
  to measure in the field, while their measurement
  in the lab is often unrealistic.
• Strategic value. Simple models are more useful
  for strategic questions e.g. single vs multiple
  control organism species, than detailed
  predictions about the merits of particular
  natural enemies.

52
Arguments against introducing all eligible enemies

• Bad Science. Does little to advance the science
• Waste of Resources. Wastes time and money
• Possibly Counterproductive. Some evidence that
  establishment of one agent makes more difficult
  the establishment of the next, which must occur
  on a reduced pest population (Ehler and Hall
  1982, but see Keller 1984)

53
Models for biological control

• Strategic approach. Analytical models (May and
  Hassell 1988 Hassell 1978) make general
  predictions about equilibrium levels and
  stability
• Tactical approach. Detailed simulation models of
  pest and natural enemy incorporating models of
  crop plant and details of physical conditions
  (Gutierrez et al. models of cassava mealybug)

54
Final Thought

• We believe that a biological control worker
  should take note of model predictions such as
  ours, critically examine the assumptions
  underlyng the model and, if satisfied, combine
  our results with the intuition that has been the
  mainstay of biological control since its
  inception.
• Godfray and Waage 1991

55
Discussion

• What role for ecological theory in biocontrol?
• Tradeoff in generality, precision, realism
• Is every case in BC a special case?
• Distinguish strategic from tactical applications
  of models
• Can success be achieved without stability?
• Combine Inductive vs Deductive approaches
• Why have parasitoids received far more attention
  than predators? Hymenoptera more than Dipetera?
• What hope is there for predicting biocontrol?
• Trial and Error. Collection of poorly documented
  case histories
• Prior experience. Nothing predicts success like
  success how variable are outcomes of biological
  control from time to time and place to place?
• Old fashioned natural history identify
  requirements of enemy, then try to match as many
  requirements as you can