Before introducing a biological control (biocontrol) organism, the benefits and costs should be weighed in both economic and environmental terms.

Problems with Benefit/Cost Analysis:

• Even when the benefits are greater than costs, the magnitude of the costs may be so high as to make the action unacceptable or unfeasible.
• Benefits and costs that are unevenly distributed socially, geographically and generationally can present fairness questions. A process for resolving conflicting interests is needed.
• Excessive uncertainty or questionable valuation techniques may undercut the analysis.

Key attributes of biocontrol agents are their capacity to:

• harm other organisms
• survive
• reproduce
• disperse
• evolve

These attributes can be good or bad depending in part on host specificity. Understanding, prediction, and management of host specificity is the key to the safety issue, but it is not the whole story. A control organism may harm a non-target organism in a multitude of ways - from a direct trophic interaction that arises when the control organism consumes a non-target organism, to direct interference competition, to indirect interactions that can arise when the control organism and the non-target organism interact via intermediate species such as a shared natural enemy or a shared host.

The International Union for the Conservation of Nature (IUCN) position on translocation of living organisms:

• Release of a non-indigenous species (NIS) should be considered only if clear and well-defined benefits to human or natural communities can be foreseen.
• Release of a NIS should be considered only if no indigenous species is suitable for the desired purpose.
• No NIS should be deliberately released into any natural area; releases into semi-natural areas should not occur absent exceptional reasons.
• Planned releases, including those of biological control, entail three critical phases: rigorous assessment of desirability; controlled experimental release; and extensive release accompanied by careful monitoring and pre-arrangement for control or eradication measures, if necessary.
• Special consideration should be given to eradicating existing introductions in ecologically vulnerable areas.

Host specificity testing in biological control:

• predators and pathogens to control weeds
• predators, parasites, and pathogens to control pest arthropods
• microbial biopesticides (using bacteria, fungi, viruses, etc. to control pest arthropods and weeds)
• other types of biological control (fish for control of aquatic weeds, viruses for control of vertebrate pests, and newly evolving technologies based in biotechnology)

Review of safety record of biocontrol agents

Some exotic biocontrol agents were introduced to do a particular job and have ended up doing something different. For example, there has been a problem with agents feeding on non-target species:

1. Chrysolina beetles eating ground cover in northern California
2. Rhinocyllus weevils eating native thistles
3. Tyria caterpillars eating ornamental plants and native wild flowers
4. Cactoblastis caterpillars eating native cacti in Florida

A tingid bug was widely introduced to control a specific weed; it attacked a variety of a crop, sesame. However, a disaster was averted as on-target effects were localized along scales of space, time, and biological organization.

How confident can we be regarding the safety of biocontrol agents?

1. Little monitoring except in agroecosystems
2. Native communities have been neglected
3. Expect time-lags in appearance of problems (it can take 5-10 years to see affects on target species and even longer for effects on non-target species)

Key questions regarding host specificity

1. Test plants: How do we pick the non-targets for host- specificity tests?

   a) dominated by economically useful plants
   b) includes a small fraction of potential wild hosts
   c) inadequate predictive framework

   (Decisions about introductions should be removed from local political decision-making because there are national implications.)

2. Test insects : How do we deal with agent variety?

   a) Specificity is expressed at individual, population, and species levels. b) Host range may change over time given the right combination of genetic variation, selection, and ecological opportunity.

Early Tests
Prior to the introduction of the cinnabar moth into New Zealand from England in 1935, crops and ornamentals were tested to make sure they were not suitable hosts.

More Modern Tests
Instead of testing each plant, an attempt was made to develop a predictive framework. Prior to the release of cinnabar moth in Canada in 1961, several genera of the tribe Senecioneae and representatives of most other tribes of Compositae that occur in Canada were tested. Feeding and development were evaluated; did feeding occur on the host, and did the insect complete development? It was found that restriction for feeding occurs at the sub-family level and that restriction for development occurs at the genus level. This established a new philosophy for host testing.

What is the actual protocol?
It is conventional to assume vulnerability equals suitability for larval development in no-choice feeding test?

• but actual host range (field) is much smaller than potential host range (lab)
• need to distinguish probability and consequences of host use

1. phylogenetic constraints
2. climatic constraints
3. physical and biotic habitat restrictions
4. oviposition requirements
5. plant acceptability for feeding
6. suitability for development

A more productive approach to risk would be to:

1. assess probability of attack; take into account sequence of factors
2. look at consequences of attacks on the non-target host

1. Probability of host utilization: habitat finding, host finding, host recognition, host acceptance, host suitability
2. Amount of potential loss for each level of utilization, taking into account mitigating factors
3. Expectation of loss

The storage effect hypothesis suggested that strong recruitment under favorable periods is stored in the adult population or the soil seed bank and is capable of contributing to reproduction when favorable conditions return. It requires that:

* environmental conditions vary, resulting in fluctuating recruitment rates;
* individuals survive over periods of poor recruitment as adults or buried seed.

There is a need for more evolutionary studies of host shifts:
1- micro-evolutionary requirements for host shift: genetic variability; strong selection; ecological opportunity
2- macro-evolutionary considerations in host shift: distinguish evolution by mutual descent from evolution by colonization

Safety factors in patterns of genetic variation and correlation:

preference x suitability = host utilization
no correlation - independence
positive correlations - host shift may be easier
negative correlations - host shift may be more difficult

Conclusions

Future work should emphasize:

1. Statistical rigor in tests
2. Validation of general approach
3. Post-release monitoring
4. Selected evolutionary studies
5. Attention to both trophic and non-trophic effects
6. Risk analysis weighing costs and benefits in environmental and economic terms (including probability of harm, consequences of harm, etc.)

Discussion

Preventing harmful introductions is the highest priority. We need as good a protocol to protect against exotic invasions as we have to protect against mistakes in biocontrol releases. On the other hand, biocontrol practitioners see themselves as "do-gooders" and exempt themselves from rules. Nearly 1,000 biocontrol agents per year are released. How should we view introducing 60 control organism species to control one target pest species like the Russian wheat aphid? In the past, one agent (a ladybird beetle) was released without an identified target; it simply attacks soft-bodied insects. To cite another area of ignorance, the environmental impacts of agents like microbial pesticides are very poorly known.

There is concern over biocontrol bacteria Bacillus thuringiensis (B.t.) used to eradicate the gypsy moth. It is a broad-spectrum larvacide that killed native lepidoptera. This could have been more harmful environmentally than chemical pesticides. We need to weigh the costs and benefits of losing the lepidoptera in a particular area versus a complete invasion by the gypsy moth.

There are serious questions about using exotics to control "undesirable" native species. The zeal to find such new organisms is frightening. There is a reluctance to attack native plants pests in this way, but not in controlling native insects that are perceived to be pests. One has to guard against philosophical errors on all sides of this issue. We need to look past individual species population impacts to community impacts. We should ask these questions prior to introduction of biocontrol agents, not after.

The current cost/benefit analysis is not very sophisticated; we still hear a lot about potential benefits without any discussion of potential risks and costs, i.e., "this agent could save $60 million for the rangeland" with no consideration of how much the natural system will pay for it.

Biological control of weeds

Dr. McEvoy discussed 26 weed species currently managed with biological control, most of Eurasian origin, and 58 natural enemy species including 32 species of Coleoptera (beetles), as well as Diptera, Lepidoptera, mites, fungi, nematodes, and fish. In South Africa, 36 weed species are controlled by 60 natural enemies, also mainly beetles.

Biological Control of Weeds - a World Catalogue is a good resource; what works in one area has a good chance of working in another. Worldwide databases of biocontrol agents also exist and include information on characteristics of species controlled biologically.

Practitioners of biocontrol were surveyed about what factors they believe are limiting enemy effectiveness. Their responses tallied as follows:

Climate 44%
Predators 22%
Parasitoids 11%
Disease 8%
Incompatibility 33%
Competition 12%

Their recommendations emphasize a need to find better matches between organism and climate and between organism and host, but we also need better understanding of organism interactions within and between different trophic levels.

Basic Questions
What are the patterns in the dynamics of biological control systems?
How are they regulated?
How do they respond to perturbation?

1. show whether current enemies are adequate or whether new enemies are needed
2. create more reliable basis for extrapolation to new areas
3. elaborating and testing principles and predictions about host specificity and predictability

   ---

   Figure 16.1: Two Views on Safety of Biocontrol Agents
   Figure 16.2: Summary Chart of Original Interaction and Evolved Host Switch