- Top of page
- Materials and methods
Uraba lugens Walker (syn Roeselia lugens: Lepidoptera: Noctuidae, Nolinae) is an invasive Australian moth with an exceptionally wide host range centred on Australian Myrtaceae, and a broad geographical distribution (Fig. 1). In its native range, U. lugens has been identified and studied as a pest of Eucalyptus spp. since at least 1899, when notable outbreaks were recorded (Froggatt 1900; Campbell 1962; Farr, Swain & Metcalf 2004). In New Zealand, U. lugens was identified originally on Eucalyptus spp. at the Mount Manganui golf course in 1996, whereupon an intensive eradication campaign in this area resulted in its local eradication. According to the New Zealand Forest Health database, it appears that the population had first been collected at Mount Maunganui in 1992, but the specimen was misidentified originally (J. Bain, Ensis, Forest Health Database, unpublished data). Subsequently, U. lugens was discovered in nearby Auckland in 2001, but the delimitation survey revealed that the population there had dispersed beyond the point where eradication was feasible (Ross 2003). U. lugens is currently found in Auckland, covering more than 40 000 ha, and spreading steadily southward and northward away from the original infestation zone (http://gis.scionresearch.com/maful).
Figure 1. The known distribution of (a) the host plants for Uraba lugens and (b) Uraba lugens in Australia. The records for U. lugens include the pheromone survey sites at which it was trapped during the survey reported here. U. lugens host records were obtained from Australia's Virtual Herbarium, and location records for U. lugens were collated from the Australian National Insect Collection (unpublished data), Brimblecom (1962); Turner (1944), Jane Elek (personal communication), Forestry Tasmania (unpublished data), Charlma Phillips – Forestry South Australia (unpublished data), Chris Burwell – QLD Museum (unpublished data), Campbell (1962), Harris (1974), Andy Gibb HortResearch (unpublished data), Anthony Rice, University of Tasmania (unpublished data), John Ireson; Tasmanian Department of Primary Industries Water and Environment (unpublished data), Kerrie Bacon, State Forests of NSW (unpublished data), Ken Henry, SARDI – Primary Industries Research South Australia, Ross Wylie, Queensland Department of Primary Industries, Vin Patel, University of Tasmania, David de Little, private consultant. In some cases, the Geosciences Australia place names index was used to derive geographical coordinates for locality information.
Download figure to PowerPoint
U. lugens takes its common name, the gum leaf skeletonizer, from the feeding habits of the early larval stadia, which feed on the epidermis of the leaves of gum trees, leaving a skeletal network of veins. It is of concern to New Zealand biosecurity authorities from a number of perspectives. The larvae have hollow urticating hairs or setae that contain an envenomating substance, including a histamine, that can cause severe skin reactions that can last for up to 42 days (Southcott 1978). It also poses an immediate and severe threat to some amenity trees in the Auckland region, with Lophostemon confertus trees being severely defoliated (D. J. Kriticos, personal observation). To a minor extent, some native Myrtaceae that are in the immediate vicinity of suitable Australian Myrtaceae hosts are at risk of minor ‘spillover’ feeding damage when larvae from eucalypt hosts come into contact with the native plants (Kriticos et al. 2005). As the invasion spreads north and south from Auckland, U. lugens will encounter extensive commercial Eucalyptus spp. plantations, posing a significant economic threat (Potter & Stephens 2005; Potter et al. 2005).
The potential distribution and relative abundance of U. lugens in New Zealand is of interest to policymakers and other stakeholders, who wish to understand the potential spatial extent of the invasion so that the impacts can be gauged and appropriate ameliorative measures can be considered prior to the invasion reaching valuable plantation areas.
Climate-based distribution models have been used previously to project the potential distribution of taxa ranging from plants, invertebrates, pathogens and even vertebrates. Numerous systems have been developed for this purpose (Guisan & Zimmermann 2000; Kriticos & Randall 2001). Their strengths lie in being able to identify locations that may be suitable for a taxon. This information is of immense value to conservation managers wishing to identify the range of a rare and endangered species in order to protect its habitat. In the case of an invading organism this information can identify those natural or production assets at risk from the invasion. While each of the habitat models has its own strengths and weaknesses, there are some thematic issues. Models built using native-range only data are likely to be biased conservatively due to the effect of biotic release that distinguishes between the fundamental and realized niches (Davis et al. 1998a; Wharton & Kriticos 2004). Correlative regression-based models such as Bioclim (Busby 1991), grasp (Lehmann et al. 2002) and garp (Stockwell & Peters 1999) are likely to be unreliable when extrapolating into novel environments such as new continents or future climate scenarios. Models based solely on records of invasive species in a newly invaded region are likely to be conservatively biased as the invaded range is likely to be a subset of its suitable range. Finally, many of the models rely solely upon climate variables, ignoring other habitat factors such as disturbance regime and soils. Models such as Bioclim, climate (Pheloung 1996), climex (Sutherst & Maywald 1985; Sutherst et al. 2004) and stash (Sykes et al. 1996) model those areas that are climatically suitable, rather than habitat per se.
The known distribution of U. lugens in Australia was used to develop a climex model to project its potential distribution in New Zealand. climex was chosen for this task because of its popularity as a pest risk assessment tool and its ability to project distributions into novel environments and to provide some insight into the ecoclimatic mechanisms that are limiting a species distribution (Kriticos & Randall 2001; Sutherst 2003). In developing this model, Kriticos et al. (2005) raised some questions regarding the completeness of the known distribution of U. lugens in the Australian state of Tasmania. First, the distribution of U. lugens in Tasmania appeared to be restricted unreasonably to relatively dry areas. Fitting the model to the known distribution would have required applying a suspiciously large degree of wet stress to limit U. lugens to the drier zones, which strongly affected its projected potential distribution in New Zealand. Secondly, the projected potential distribution of U. lugens in New Zealand was also highly sensitive to its precise range limit near Arthurs Lake in the central highlands of Tasmania. The apparent range limit coincided with a steep elevation gradient, terminating on the central plateau, the coldest region in Tasmania. The geographical location of the single record for U. lugens near Arthurs Lake was recorded imprecisely, thereby casting a major uncertainty over the potential distribution of U. lugens in New Zealand. These two sources of uncertainty warranted further investigation because important (and expensive) decisions were being influenced by perceptions of the potential range of this species in New Zealand in relation to the eucalypt forest plantations (M. Ross, Biosecurity New Zealand, personal communicatrion).
Traps baited with pheromone lures have been used extensively to ascertain the presence of a variety of insects – usually in agricultural, horticultural or silvicultural settings where the organism is a known potential pest and the aim is to ascertain if the pest is present through time (Suckling & Karg 2000). Delta traps have been used in Auckland, New Zealand to trap U. lugens to estimate the spatial extent of the invasion through time and to confirm the predicted adult flight phenology (Withers et al. 2003; Suckling et al. 2005).
In order to reduce the uncertainty surrounding the potential distribution of U. lugens in New Zealand, a pheromone trap survey was undertaken to establish the climatic limits of U. lugens in Tasmania. We describe here the development of the climex model, and the application of the pheromone trap survey to the task of parameterizing the climex model.
- Top of page
- Materials and methods
In five of the six primary elevation transects, U. lugens was found in the highest trap site on the transect, demonstrating the suitability of the more extreme climate (and presumably also the more moderate climatic conditions found at lower elevation trap sites that failed to trap moths).
This study indicates that U. lugens is a hardy insect, able to tolerate a wide range of conditions from the warm and dry climate of mainland Australia during summer to the cool, wet subalpine conditions found at high elevation in central and western Tasmania. In so doing, it displays a large degree of plasticity in voltinism.
The apparent discrepancy in central Tasmania between the trapped moth occurrence and the potential distribution (Fig. 3) is due possibly to an inconsistency between the meteorological data set for Tasmania derived using esoclim (Houlder 2004) and the records from the Bureau of Meteorology (http://www.bom.gov.au/). Using values from the Bureau of Meteorology climate average dataset for Liawenee (station 096065) climex indicates that on average there will be 610 °C days above 8 °C available for development of U. lugens, whereas using the esoclim data set, climex indicates that there are only 445 °C days above 8 °C at the same location. The inconsistency is due probably to the paucity of meteorological data from central Tasmania with which to use as input into esoclim. In this study we elected to use the recorded data set as the basis for setting the PDD and cold stress parameters, rather than fit them to an individual derived climatic data set. One implication of this choice is to underline the need for caution in interpreting range limits (in particular the cold range limit) when using other gridded data sets such as that used for New Zealand. That is, the limits should be viewed as indicative, rather than prescriptive.
minimum development heat sum to complete a generation
climex uses the PDD heat sum threshold and the minimum temperature for development (DV0) to estimate the number of generations of the modelled taxa that can be supported at each location. The annual heat sum at Liawenee on the central plateau is only 610 °C days above a base temperature of 8 °C. This is considerably less than the minimum generational heat sum of 1060 °C days predicted by development studies under constant temperature and daylength conditions for a single generation of the bivoltine form (Allen & Keller 1991; D. J. Kriticos, Ensis unpublished data). None the less, moths were caught at each of the trap sites on the Steppes and the Central Plateau where temperatures appear to be insufficient to support adequate development of U. lugens. There are at least five possible explanations for this apparent anomaly. First, the moths could have developed elsewhere under warmer conditions and flown to these high-elevation sites. Given the consistency of trapping success and the geographical spread of sites across the central plateau over which U. lugens was trapped, this seems unlikely. In addition, these moths are known to be poor fliers (Harris 1975; Morgan & Cobbinah 1977). Secondly, the year prior to trapping could have been extremely warm in relation to the long-term average conditions on the central plateau. Thirdly, the univoltine form of U. lugens could have lower thermal requirements for development than the bivoltine form, or there could be a cold-adapted race of U. lugens, although sharing the same pheromone. Fourthly, the usual analytical technique for estimating generation times provides an estimate for the average moth in the population, not the most rapidly developing quantile. At the cool range margins, only the fastest developing moths may survive. To estimate the minimum heat accumulation required to enable any moths to complete their development, it may be more appropriate to sum the thermal development time for each lifestage less two standard deviations. Finally, Allen (1990) notes that in the field, larvae may pupate at the 8th to 13th instars. This suggests that pupation may occur at any point after a developmental threshold has been reached (8th instar) in response to some other external stimuli such as changing daylength or deteriorating food quality (Cobbinah 1985). Thus, the heat requirements to complete a generation may be shortened in the field compared to laboratory studies, where light : dark cycles are held constant and food supply quality and abundance is well maintained.
If the appropriate value for PDD is the minimum heat sum for the most rapidly developing individuals, then there are important implications for interpretation of the Number of Generations variable in climex. It is unclear whether in more moderate climates the average development period dictates the number of generations because of the requirement for synchronized mating within a generation and the consequent effect that the law of central tendency has on selective pressure. Thus, at the cool range margin, the selection trade-off may be skewed toward the minimum development time required for successful reproduction, at the expense of larger adult size and fecundity. At warmer locations the fitness trade-off may swing towards more extended development with larger adults and fecundity, rather than increased voltinism. Without more detailed studies of this problem, predictions of multivoltinism in climex based on multiples of PDD should be interpreted as indicating that the minimum heat sum requirements for multiple generations have been met, but not that the number of generations will be likely to occur. In addition to the possibility that there may be plasticity in the heat sum taken to complete a generation, other factors such as rainfall, day length, soil moisture availability and the coincidence of these factors in the range necessary for population growth may also act to constrain the realized pattern of voltinism in any given location.
These difficulties in defining the minimum heat sum required to complete a generation reinforce the cautions regarding the primacy of geographical distribution data in bioclimatic modelling, and the extreme caution that needs to be applied when comparing biological parameters derived from instantaneous measurements in a laboratory with climatic data. Such comparisons involve a scale change, where the dominant processes may change unpredictably between the two settings.
potential distribution of u. lugens
U. lugens adults are generally thought to fly a maximum of 1 km from their pupation site, though some long-distance dispersal (>12 km) could possibly occur over water (Suckling et al. 2005). This means that the adult male distribution is likely to be indicative of the distribution of the species generally, including the damaging larval stage. This assumption could be unreliable in very windy locations, where the adults could be transported intact for long distances from their natal sites. Outlying data should therefore be treated with caution, and efforts should be made to include data redundancy in survey effort.
A key assumption in developing climate-based distribution models is that the known distribution is limited by climate, and not by factors such as the distribution of suitable hosts. The distribution of U. lugens hosts in Australia is clearly not limiting its distribution throughout most of its range (Fig. 1). It was possible that hosts may have been limiting its range in some cold areas of Tasmania, such as on the Steppes, near Liawenee and at Hartz Mountains, where U. lugens were found at the tree line. However, at Mount Field, despite suitable hosts growing up to 1200 m, U. lugens was not found present above 600 m. This finding is consistent with the climex model that indicates that the annual heat sum would be insufficient for U. lugens above this elevation (Fig. 2), lending further support for the choice of value for the PDD parameter.
The pheromone trap survey extended the known distribution of U. lugens well beyond the known historical distribution limits into much cooler and wetter climates. In New Zealand, the potential distribution has been extended to include the majority of the cool, wet central North Island and a great deal more of the South Island up to the foothills of the Alps (Fig. 5). Central Otago remains climatically marginal to unsuitable for U. lugens due to a combination of dry stress and cold stress (Fig. 5).
The global potential distribution of U. lugens highlights a phytosanitary risk from U. lugens due to exports from Australia or New Zealand to overseas eucalypt growing regions particularly during the pupation period(s). In addition to forestry resources, amenity plantings in places such as California will also be under threat of invasion and attack by U. lugens, with its attendant economic and public health impacts.
The projected potential distribution may be at least partially conservative as it is based upon the realized niche, which includes constraints due to biotic factors such as competition, predation and parasitism (Hutchinson 1957; Davis et al. 1998b). As noted by Brown, Stevens & Kaufman (1996), where biotic factors are constraining ranges, they are most likely to manifest themselves in those regions of the potential range that are relatively warm and wet, and where environmental resources such as heat and moisture are not limiting. Hence, as the warm and wet range limits for U. lugens are not encountered in New Zealand, the model can be considered to be reasonably reliable, and potential climatic range expansion due to biotic release is unlikely. This form of underestimate in potential distribution is likely to occur in subtropical and tropical regions.
pheromone trapping for range delimitation
The pheromone trap system utilized for this study worked well. Despite the traps being set for 10 weeks, the trapped specimens were generally in good condition at the time the traps were retrieved. However, there were a number of traps that contained Lepidopteran carcasses that could have been U. lugens, but were too badly decomposed to be certain of the identity. It is also likely that the invasive wasp, Vespula germanica, and other wasps (many of which were found in traps), had attacked a number of these unidentifiable carcasses. Ten weeks was chosen as the trap duration in order to minimize the survey costs and still cover the expected adult moth flight duration at all surveyed elevations (January–March). The success of the trap system suggests that it is a useful, cost-effective technique that could also be applied to climate change studies to ascertain how insect species distributions change through time.
The lack of suitable data on the distribution of organisms is perhaps the single most common hindrance for bioclimatic modellers who generally have to rely upon museum or herbarium specimen data (Kriticos & Randall 2001; Sutherst 2003). There have been calls for systematically improving the quality and quantity of distribution data for organisms (Sutherst 2003). However, it is unclear how such a system would be paid for, and what factors would motivate relevant local authorities to collect and maintain data on their native or introduced pests when the optimal qualities of the data required for habitat modelling does not coincide with that required for pest or conservation management. In the absence of resolution of these issues and the availability of data in such databases that can support these modelling efforts, models will continue to include potentially large uncertainties.
Species range boundaries can be quite dynamic (Brown, Stevens & Kaufman 1996), while climate data sets are static. While this represents a potential scale mismatch, apart from the case of migratory species, the ramifications of basing a climex model on transect data are generally likely to be trivial. If the transect data extend the range boundary beyond the dynamic equilibrium boundary, the model will project this extra range as being at risk of invasion, albeit usually with a marginal indicated climate suitability. This bias is likely to be preferred by incursion managers.
The extension of the known range of U. lugens using a trapping survey may be an example of a widespread bias in museum and herbaria data sets toward range cores. For cryptic species in particular, it may be more difficult to detect them toward their range margins if their densities decline with habitat suitability. Consequently, many existing projections of climate suitability based on known distribution data may be conservatively biased.
The use of purpose-designed surveys such as that described in this paper may offer one cost-effective means of generating useful, unbiased data of a suitable quality for making more reliable projections. For poorly known species, for relatively little cost, four to five transects running from the edge of known suitable habitat into regions for which the distribution boundaries are unknown should be most informative. A preliminary climex model can be used to inform the choice of location of these transects to directly address the climatic gradients in different parts of the species’ range. If the spatial distribution of hosts or other non-climatic habitat factors are available prior to undertaking the survey, then GIS techniques can be used to further refine the trap system prior to deployment.
- Top of page
- Materials and methods
We are indebted to the following people and organizations that helped us to make this project a success. Dr Marianne Horak provided access to the Australian National Insect Collection. Mr Bevis Jordon, of Gunns Limited, helped in selecting sites and Mr Chris Barnes gave permission to deploy traps in Gunns forest coups. The Tasmanian Department of Primary Industries, Water and Environment, for permission to collect U. lugens throughout Tasmania, and Dr Jane Elek, Forestry Tasmania, for permission to trap in forest coups, and assistance with survey site selection. Ms Anne Drake of Zinifex Rosebery Mine kindly provided vehicle access to Mt Reid for the trapping. Dr John Ireson provided access to laboratory facilities and Dr Peter McQuillan, University of Tasmania, provided expert assistance with insect identifications. Mr Shaun Kolomeitz, University of Queensland, provided valuable help with accessing climate data. Biosecurity New Zealand and the Foundation for Research, Science and Technology (C04X0302) provided financial support for the development of the pheromone and for undertaking the survey. Dr Agathe Leriche helped with the production of some of the figures. Dr Janet Farr (Western Australian Department of Conservation and Land Management), the journal editors and an anonymous referee provided useful suggestions to improve the manuscript.