Ecological principles that underpin host testing for weed biological control – a case study with the lantana sap-sucking bug, Aconophora compressa Walker (Hemiptera: Membracidae)

ABSTRACT 1. Aconophora compressa Walker (Hemiptera: Membracidae) was introduced into Australia from Mexico in 1995 for the biological control of lantana (Lantana camara L.: Verbenaceae). Pre-release host testing implied minimal non-target effects from this gregarious stem feeding insect. Post-release, A. compressa is sometimes present on several plant species in the field in Australia and is mostly associated with fiddlewood (Citharexylum spinosum), an exotic ornamental tree of Caribbean origin. Field sampling has indicated that lantana harbours A. compressa mainly as a “spillover” from fiddlewood and does not persist on lantana when populations on fiddlewood are low. 2. This thesis investigated the interaction between A. compressa and six host taxa on which it is found in the field in Australia. To this end, adult survival and reproduction, nymphal survival and development and the consequences of the long distance host finding behaviour of A. compressa were quantified experimentally. A model was developed to calculate the number of adults produced on each host taxon, based on glasshouse data similar to that typically collected in biological control host testing. Output from the model was then compared to host use data from the field to examine the consistency between glasshouse and field data. 3. Modelled adult production indicated a discrepancy between the relatively high performance of A. compressa on lantana under glasshouse conditions and its relatively low abundance on lantana in the field. The results indicate that fiddlewood can be considered the primary host plant species of A. compressa in southeast Queensland. In terms of the numbers of eggs laid and nymphal survival and developmental rates, fiddlewood was by far the best host, followed by lantana, Duranta erecta (cultivars geisha girl and Sheena’s gold), jacaranda (Jacaranda mimosifolia) and Myoporum (Myoporum acuminatum). Oviposition and nymphal developmental rates on the last two were low and these plants are used only incidentally in the field by A. compressa. The remaining hosts, including lantana, are effectively secondary hosts because they are evidently used mainly when high densities of A. compressa develop on fiddlewood and then spillover onto the other hosts, where they do not persist as they do on fiddlewood. 4. The effect of nymphal host plants of different quality on adult survival and reproduction were investigated in a reciprocal cross experiment with the primary host (fiddlewood) and a secondary host plant (geisha girl). Time spent feeding on the secondary host during any life stage had a negative affect on the ultimate reproductive capacity of the A. compressa individuals involved, even when insects that had developed on geisha girl as nymphs were moved to fiddlewood at adult eclosion. Modelled adult production indicates that the numbers of A. compressa produced on fiddlewood exceed those produced on geisha girl by a factor of about 10-20. The model output also supports the interpretation derived from field data, indicating that populations of A. compressa on secondary host plants derive mainly from a spillover when densities on fiddlewood are high. 5. The relative colonisation rates across sentinel fiddlewood and lantana plants placed in the field resolved the discrepancy between the ability of A. compressa to survive and reproduce well on lantana under glasshouse conditions, but being found in the field on lantana only at low rates. Aconophora compressa individuals located the sentinel lantana plants to a much lesser extent than they did fiddlewood (20 vs 80%, respectively). Aconophora compressa placed on potted lantana and fiddlewood in the field remained on the plants and laid egg batches at equal rates. Adult production across these hosts was re-modelled to include the rates at which these species are colonised in the field, and output strongly resembled the relative intensities of host use recorded in the field. 6. Surveys in southeast Queensland were conducted over 15 months in a repeated measures design. Data verify that A. compressa was present on secondary and incidental hosts mainly as a spillover from fiddlewood. Field surveys in Mexico revealed that host use by A. compressa was consistent with that in Australia. In Morelos, the state in which A. compressa was most abundant, A. compressa was found on Lantana urticifolia, Duranta erecta and Tecoma stans at similar relatively high rates and on jacaranda and a number of other plant species at relatively low rates. Fiddlewood species were rare and were found only in areas with extremely low densities of A. compressa, so the intensity at which this species are used could not be established at the time. 7. Weed biocontrol host testing methods published over the last 10 years were reviewed, with emphasis on the degree of replication, and which raised questions about the number of replicates required for host testing. There were 489 experiments in 118 papers. Most experiments used relatively few replicates and about 75% of all experiments for which the median number of replicates could be determined (415 cases) had five or fewer replicates (319 cases). About half of all papers (37 of 83 papers with sufficient detail to determine their exact methods) included experiments in which singular replicates were repeated sequentially. About 30% of multiple host choice tests included different combinations of test plants in each cage. In statistical terms, such a design lacks replication. It is hoped that these results will initiate discussion that leads to minimum standards with regard to replication in weed biocontrol host testing experiments. 8. Overall, the experimental data (outlined in points 3 to 6, above) were used to justify the beneficial aspects of determining host plant species status, before release, in terms of their being primary, secondary or incidental hosts and to ensure that the target weed is indeed the primary host. The reason that A. compressa is ineffective against lantana is because lantana is a secondary host and thus falls outside the “sensory target” range of A. compressa. The term, “host specific” is discussed in relation to its use in weed biocontrol; how it implicitly directs the types of experiments conducted and interpretation of host testing results. A framework for identifying host relationships is developed with an eye for continued improvement of the theory and thus the practice of weed biocontrol host testing