OUP user menu

Potential Geographical Distribution of the Mediterranean Fruit Fly, Ceratitis capitata (Diptera: Tephritidae), with Emphasis on Argentina and Australia

(CC)
M. Teresa Vera, Rafael Rodriguez, Diego F. Segura, Jorge L. Cladera, Robert W. Sutherst
DOI: http://dx.doi.org/10.1603/0046-225X-31.6.1009 1009-1022 First published online: 1 December 2002

Abstract

The CLIMEX model was used to infer the climatic requirements of the Mediterranean fruit fly, Ceratitis capitata (Wiedemann), from the fruit fly’s observed geographical distribution in the Mediterranean region. The model indicated that the potential distribution was limited by cold to the north in Europe and by dryness in northern Africa and in the south of Spain and Portugal. The model was then used to estimate the potential geographical distribution of the species in Argentina and Australia. The results agreed with the observed distribution in Argentina and much of the historical distribution in Australia, but they did not agree with the present distribution in eastern Australia. In the latter region, another species of fruit fly, Bactrocera (Dacus) tryoni (Froggatt) has been credited with displacing C. capitata. Seasonal and year-to-year variation in climatic suitability was explored at three selected locations in Argentina. The results indicated that some detrimental effects of summer temperatures, or of extremes of precipitation, occurred in particular areas. Some of these limiting factors, especially dry stress, were prolonged enough to restrict the geographical distribution of medfly. However, when irrigation was included in the simulations, the detrimental effect of dryness was removed. Finally, a global risk map for medfly was produced, which highlights the areas at risk from this major quarantine pest.

  • CLIMEX
  • medfly
  • geographical distribution

The Mediterranean fruit fly, Ceratitis capitata (Diptera: Tephritidae), known also as the medfly, is one of the world's most widespread and damaging pests of horticulture. It infests >200 species of fruit and vegetables (Christenson and Foote 1960, Liquido et al. 1991) and is considered to be a major quarantine pest. It originated in Africa (White and Elson-Harris 1992) and spread first to the Mediterranean region during the early 19th century and from there to the rest of world (Headrick and Goeden 1996). It occurs in most tropical and temperate regions with some countries succeeding in eradicating outbreaks of recently introduced populations (Dowell and Penrose 1995, Clark et al. 1996, Penrose 1996) or well-established populations (Hendrichs et al. 1983, Fisher et al. 1985).

In Argentina (Fig. 1), the medfly was first detected in orchards around Buenos Aires city in the early 1900s (Vergani 1952) and in Concordia, Entre Ríos, in 1934. From 1935, several foci were reported in most of the fruit-producing areas north of parallel 36°S (Vergani 1952). By 1952, it was also detected south of this line in the Patagonian region, but the populations were restricted to urban areas and detected only during the favorable season (Aruani et al. 1996, Sanchez et al. 2001). The absence of quarantine barriers between regions and the very active fruit trade supported by road and rail transport probably allowed C. capitata to occupy its potential range. In western areas (Mendoza, San Juan, and La Rioja), it was reported in fruit-producing irrigated valleys and urban areas that are isolated from other fruit fly-infested areas by deserts or mountains with no, or very few, cultivated or wild host plants (CIRPON 1991, Aruani et al. 1996). Presently, on-going eradication and control programs restrict its occurrence in several areas by direct interventions (sterile insect technique and chemical control) and by quarantine barriers (Aruani et al. 1996, De Longo et al. 2000, Frissolo et al. 2001, Sanchez et al. 2001). In other fruit-producing areas, populations reach high infestation levels supported, in some cases, by the presence of wild hosts or abandoned orchards. Although its distribution appears to be restricted ultimately by the severity of the winter, the variety of climates around the country suggests that other climatic factors may limit or at least modulate the species' population dynamics in different areas. Some of the important horticultural areas are located adjacent to the edges of the geographical distribution of medfly, or in marginal areas, making variation in climatic suitability within and between years an important factor to consider in medfly control.

Fig. 1.

The historical geographical distribution of medfly in Argentina. At present, control actions are successfully reducing medfly populations at some marginal areas to the south (Patagonian region) and west (Mendoza, San Juan, and La Rioja). Locations of Cerro Azul, Jachal, and Rivadavia are shown.

The medfly was first introduced into Western Australia from Europe around 1897 (Hooper and Drew 1989, Anonymous 1991). It is now established from Esperance in the south to Derby in the north, with highest populations between Bunbury and Carnarvon (http://www.agric.wa.gov.au/ento/medfly.htm). Late in the 1890s, it spread to the eastern states colonizing the East Coast from north of the Clarence river to south of Sydney with infestations also recorded inland as far as Albury and the Murrumbidgee Irrigation Area, and from Melbourne and Tasmania (Fig. 2, Bateman 1971). A secondary colonization event was reported from the Mediterranean region (Fimiani 1989). Genetic analyses indicate that the Australian population of medfly has low heterogeneity compared with populations from more recent colonization events around the world (Malacrida et al. 1998). C. capitata appears to have been displaced gradually from eastern Australia by the Queensland fruit fly Bactrocera (Dacus) tryoni, as the latter expanded into southern parts of its range early in the 20th century (Allman 1939, Anonymous 1991). There are sporadic outbreaks of medfly reported in Adelaide in South Australia (Maelzer 1990).

Fig. 2.

The historical geographical distribution of medfly in Australia. Medfly is now endemic only in Western Australia.

Several attempts have been made to define the climatic requirements of C. capitata and to estimate its potential to colonize new areas (Gjullin 1931, Messenger and Flitters 1954, Worner 1988; Carey 1990). Climatic requirements were inferred from laboratory studies (Messenger and Flitters 1954, Worner 1988) or from observed distributions (Gjullin 1931). Much of the effort has been focused on the impact of extreme low temperatures with less emphasis on the effect of other potentially limiting conditions. Other studies (McBride 1935, Harris and Lee 1989, Eskafi and Fernandez 1990) evaluated the role of precipitation and soil moisture as detrimental factors, but did not infer any distribution pattern. Papadopoulos et al. (1996) studied medfly overwintering in a marginal location in northern Greece, providing valuable data for fitting and testing models of overwintering ability. Worner (1988) was the first to use a model, CLIMEX (Sutherst and Maywald 1985, Sutherst et al. 1999) to assess the response of C. capitata to climate and to describe its potential distribution in New Zealand.

The CLIMEX approach is based on the assumption that biological organisms are supremely efficient integrators of the effects of climate and other environmental variables. Population abundance, seasonal phenology, and geographical distributions all reflect this process, which is difficult to capture entirely in experiments or process-based simulation models without decades of study. CLIMEX represents a pragmatic approach (Worner 1988) that is most useful when there is limited process-related data available for a species. The aim is to describe the core responses of a species to climate. CLIMEX analyses complement other modeling approaches and have proved illuminating even with species that have been studied or modeled for decades. Sutherst and Maywald (1985) emphasized several caveats related to assumptions about the role of nonclimatic factors and to the available range of parameter values in the geographical areas involved. They also warned about the effect of interspecific competition, resulting in significant restriction of the distribution of each species, and gave a dramatic example involving ticks of the genus Boophilus in Africa. Davis et al. (1998) drew further attention to this phenomenon with regard to laboratory competition between fruit fly species mediated by a parasitoid wasp Leptopilina boulardi Barbotin, Carton and Kelner-Pillaut, but did not support it with data on the distributions of the species involved, namely Drosophila melanogaster Meigen, Drosophila simulans Sturtevant, and Drosophila subobscura Collin.

In this article, the CLIMEX model was used to infer the responses of medfly to climate, and the resulting model was tested against independent observations from Argentina and Australia. A global projection is then made of the areas around the world that are at risk of colonization by medfly, in the absence of blocking by nonclimatic factors.

Materials and Methods

Outline of the CLIMEX Model.

CLIMEX is a dynamic model (Sutherst and Maywald 1985; Sutherst et al. 1995,1999,2000; http://www.ento.csiro.au/climex/climex.html) that integrates the weekly responses of a population to climate into a series of annual indices. A hydrological model is used to calculate weekly soil moisture from rainfall and estimated evaporation. The temperature, moisture, and daylength responses (where appropriate) are combined into a weekly population Growth Index (GIW) for that species. Annual Temperature (TI) and Moisture (MI) indices summarize the species' response to prevailing temperature and moisture conditions, respectively. Responses to extreme conditions are taken into account in a series of "stress indices" that estimate the threat posed to that species by extreme or prolonged cold, hot, dry, or wet weather. In addition, constraints on completion of the life cycle due to prolonged periods of inadequate heat summation or by an inappropriate daylength regimen can be estimated where data are available. Finally, the growth and stress indices are combined into an Ecoclimatic Index (EI), scaled from 0 to 100, to represent the overall suitability of the given geographical location for the propagation and persistence of the species.

As a guide to the potential for establishment of populations, the following categories of EI are suggested: Locations with EI > 25 are very favorable for population growth and persistence; locations with EI = 10-25 are favorable; and those with <10 have low to marginal suitability. An EI = 0 means that the species is not able to persist at the location under average climatic conditions. In these latter areas, it is possible that populations can either persist in years that deviate significantly from average or can increase during the growth season, but are unable to overwinter. In the latter case, immigration in the spring (for example) can result in reinfestation each year. The annual Growth Index, given that the flies immigrate into the area each spring, indicates the potential population growth in such cases. The annual number of degree-days, above the developmental threshold (estimated by the CLIMEX DV0 parameter), for development is calculated so that the number of generations can be estimated.

Variation in the suitability of the climate for population growth over several years can also be assessed with CLIMEX using the Compare Years function. In this case, actual rather than long-term average meteorological data are used, and CLIMEX converts these to monthly averages to smooth the data. It then interpolates the data to weekly values, which make them more comparable with parameters estimated using long-term average data. The Compare Years function provides relative measures of the growth and stress indices that make it possible to compare between seasons and years, although they are not directly comparable with the values derived using the average data because of the different amount of smoothing involved.

The values of the CLIMEX model parameters, which reflect a species' climatic requirements, are inferred from information on the species' known geographical distribution, relative abundance, and seasonal phenology. This procedure, referred to as inverse or inferential modeling, is the reverse of the reductionist approach that is used to build population models based on assembling functions that describe biological processes. In CLIMEX, this inferential process involves first, the development of hypotheses regarding factors that limit the distribution, followed by manually adjusting parameter values until the simulated geographical distribution coincides as closely as possible with the observed distribution.

Meteorological Databases.

For Europe and Australia, meteorological data in the CLIMEX database was used. It consisted of monthly long-term average maximum and minimum temperatures, rainfall, and relative humidity for 285 representative locations in Europe and 676 locations in Australia. In addition, a splined set of global climate averages on a 0.5° grid, referred to as a "climate surface", was acquired from the Climate Research Unit of Norwich University (http://ipcc-ddc.cru.uea.ac.uk/; New et al. 1999) and used to examine the potential distribution of medfly in Europe. In Argentina, data from 152 meteorological stations from the Instituto Nacional de Tecnología Agropecuaria (INTA) and the Servicio Meteorológico Nacional (SMN) meteorological network, with recordings of ≈24 yr in each location, was used to relate to the observed medfly distribution.

Three particular locations of Argentina (Fig. 1) were chosen to assess year-to-year and seasonal variation in climatic suitability for the species. Cerro Azul, located in the northeast of the country in Misiones province, represents extremely humid conditions. At Jachal, located in San Juan province, very dry weather prevails all year round. Rivadavia, located in Salta province, is characterized by extremely hot summers and dry winters. These locations were chosen to represent extreme moisture and temperature conditions other than cold, which was already represented. Daily meteorological data for 30, 11, and 21 yr respectively at the three locations were used to examine the population growth (GIW, TIW, and MIW) and stress indices in detail. Median and quartiles were computed with STATISTICA (StatSoft 2000).

To investigate the effect of climate on horticultural pests, it is necessary to include the effects of irrigation, without which much dry-land horticulture would not be possible. Therefore, when the weekly rainfall was <25 mm, it was topped up with irrigation to equal 25 mm. When rainfall exceeded 25 mm, no irrigation was added.

Results

CLIMEX parameters (Table 1), and particularly those related to low-temperature effects, were estimated for the medfly using the reported geographical distribution in the Mediterranean region (http://www.ecoport.org/EP.exe$PictShow?ID=3052&Subj=), supplemented by specific observations on overwintering survival at Thessaloniki, Greece (Papadopoulos et al. 1996). With the available range of parameter values for the climatic variables in Europe, it was not possible to estimate the limiting effects of either very high temperatures or high summer rainfall on the species. Thus, these limits were undefined and were estimated from the experimental data of Messenger and Flitters (1954). Because this process excludes influences of temporal variation, microclimatic effects, and landscape heterogeneity in the field, the estimates are considered to have lower reliability than the cold and dry stress parameter values. The fit to the distribution, using point-based climatic data from the CLIMEX database and expressed as the EI values, is shown in Fig. 3a. The principal limiting factor was found to be cold stress in the form of inadequate thermal summation to survive the winter. The model was then run using the climate surface to compare the results with those derived using point data. The results (Fig. 3b) revealed some differences between the two projections, in particular, along the Mediterranean coast of France, because of the averaging effect of the grids where there is a heterogeneous topography with mountains abutting the sea. Irrigation increased the area at risk greatly (Fig. 3c). It expanded the suitable areas in the north of Africa and in southern Spain and Portugal, revealing a potential to support larger infestations.

Fig. 3.

(a) The potential geographical distribution of medfly in Europe, as fitted by the CLIMEX Ecocfimatic Index (EI), using point meteorological data, (b) CLIMEX EI using the 50-km2 climate surface, (c) CLIMEX EI with 25 mm of irrigation per week. The EI, and hence inferred suitability of the climate for medfly, is proportional to the size of the circles. Zero values are not shown in (a).

View this table:
Table 1.

Given the distorted information resulting from the use of the climate surface, the other simulations were carried out using point data only. The seasonal potential for growth of medfly populations is shown by the weekly CLIMEX growth index (GI). The seasonal GI values for Thessaloniki with 25 mm of irrigation per week in the warmer months are shown in Fig. 4. The CLIMEX parameters, derived above, were used to compare potential geographical distributions in Argentina (Fig. 5) and Australia (Fig. 6), and the results contrasted sharply. Whereas in Argentina there was a close agreement with the reported distribution (Vergani 1952), in Australia there was good agreement with the distribution in Western Australia, but not in the east.

Fig. 4.

Seasonal CLIMEX Growth Index (GI) values for medfly in Thessaloniki, with 25 mm of irrigation per week in the warmer months. The site was used in fitting the model parameter values.

Fig. 5.

The potential geographical distribution of medfly in Argentina, as fitted by the CLIMEX model. Ecocfimatic Index (EI), using point data. The EI, and hence inferred suitability of the climate for medfly, is proportional to the size of the circles.

Fig. 6.

The potential geographical distribution of medfly in Australia, as fitted by the CLIMEX model. Ecoclimatic Index (EI), using point data. The EI, and hence inferred suitability of the climate for medfly, is proportional to the size of the circles.

The annual and seasonal variation in suitability of the climate for the medfly was explored at Cerro Azul, Jachal, and Rivadavia (Table 2, Figs. 7-9). In Cerro Azul (Fig. 7), located in the wettest area of the country, excessive moisture led to a zero median value in the Moisture Index (MIW), and consequently in the Growth Index (GIW), for periods lasting an average of two consecutive weeks. For particular weeks during late spring, summer, and autumn, the inter-quartile range of MIW was high showing large year-to-year variation (Fig. 7). In the wettest 25th percentile of years analyzed (Fig. 7, first quartile), the MIW reached zero during summer and autumn and lasted a total of 17 wk with a maximum of 9 consecutive weeks. Wet stress (WS) was high and occurred year round, indicating extreme conditions. In those years when conditions were not so wet (third quartile of MIW), no wet stress was revealed and, although there were low values of MIW, they never reached zero.

Fig. 7.

Median (bold) and interquartile range (first and third quartiles) of weekly values for CLIMEX Growth, Temperature, Moisture, and Wet Stress (WSW) indices for Cerro Azul, Argentina.

View this table:
Table 2.

At Jachal (Fig. 8), an extremely dry area, dry stress (DS) was the most important stress factor. It lasted a median of 19 consecutive weeks with a range of 8-40 consecutive weeks (DSW first and third quartiles). Although there was annual variation in the severity of dry stress, the effect of the lack of rainfall was so strong that the MIW reached zero for almost all the year in most of the years. There was also some degree of heat stress in midsummer, but temperatures were mostly suitable from spring to autumn.

Fig. 8.

Median (bold) and interquartile range (first and third quartiles) of weekly values for CLIMEX Growth, Temperature, Moisture, and Dry Stress indices for Jachal, Argentina.

At Rivadavia (Fig. 9), located in the hottest region of Argentina, there were 9 consecutive weeks during which the weekly temperature index, TIw, equaled zero (median value), and hence the GIW also reached zero. However, the model did not indicate any heat stress (HS). There was also a period of dry stress during the winter that lasted a median of 7 wk, and for the 25th percentile of extremely dry years, it lasted up to 16 consecutive weeks (third quartile). MIW and GIW reached zero for 10-24 consecutive weeks (first and third quartiles) with a median value of 19 wk.

Fig. 9.

Median (bold) and interquartile range (first and third quartiles) of weekly values for CLIMEX Growth, Temperature, Moisture, and Dry Stress indices for Rivadavia, Argentina.

Irrigation removed almost all of the effect of the low rainfall on population growth, as shown by comparing the results for Jachal in Fig. 10a with those in Fig. 8. During winter, there was a decline in GIW mainly due to low temperatures (TIW, Fig. 8). At Rivadavia (Fig. 10b), the result was similar with the addition of irrigation during the dry winter completely removing dry stress.

Fig. 10.

Median (bold) and interquartile range (first and third quartiles) of weekly values for CLIMEX Growth Index with irrigation (a) forjachaland (b) for Rivadavia. When the weekly rainfall was less than 25 mm, it was topped up with irrigation to equal 25 mm. When rainfall exceeded 25 mm, no irrigation was added.

Finally, the model was run using a global meteorological data set with ≈3091 locations to define the areas of the world at risk from medfly (Fig. 11a). This pest has a very extensive potential geographical distribution that includes the tropical and subtropical portions of every continent, including those with a Mediterranean climate. Irrigation would extend that distribution substantially into irrigation zones in dry climates, particularly the West Coast of the United States, parts of North Africa and the Middle East, and central Australia (Fig. 11b).

Fig. 11.

The potential geographical distribution of medfly worldwide, as fitted by the CLIMEX model, (a) Ecoclimatic Index (EI), using point data and natural rainfall, (b) CLIMEX EI with 25 mm of irrigation per week in summer to simulate horticulture. The EI, and hence inferred suitability of the climate for medfly, is proportional to the size of the circles.

Discussion

The CLIMEX model was used to explore different hypotheses on the factors that limit the medfly's distribution in Europe, and the resulting model accurately reflected the observed distribution. The results show that a thermal accumulation hypothesis accurately explained the European distribution and the southern and western (due to the Andes) limits of the observed distribution of the fruit fly in Argentina. Low maximum temperatures limit the capacity of the medfly to colonize higher latitudes by denying it adequate thermal energy to sustain development. This contrasts with the common view that lethal minimum temperatures (e.g., freezing) are the usual limiting factor. However, it does not exclude the effect of lethal low temperatures, which will also have an effect, but extreme temperatures appear to be less restrictive to the distribution than the limitation imposed by the need for thermal accumulation in winter. When suitable data are available, it would be desirable to estimate the limiting value for minimum temperature to produce a composition cold stress model for medfly. Finding locations in which there are cold winter nights and warm days could help achieve this.

The results from the Compare Years function at three particular locations in Argentina show that the medfly experiences wet, dry, or hot stress during prolonged periods of the year. This would reduce the potential for survival. Year-to-year variation in MIW was high at Cerro Azul, suggesting that detrimental effects of excessive moisture on population growth are likely to be restricted to particular years. At Jachal, however, dry stress was present almost all the year, and this pattern was constant through the years. Cold stress was present during winter, and the combination of both factors is likely to be responsible for the low abundance of the pest in that location. However, such dry stress should not be overestimated because irrigation removes it, and irrigation is a common practice in horticultural situations. In the case of Rivadavia, the impact of summer temperatures also showed some year-to-year variation, and the high temperatures suppressed population growth but did not appear to be high enough to induce limiting heat stress. This suggests the need for field data to investigate the effects of extremely high temperatures on population growth rates and survival during the summer. Irrigation removed dry stress, suggesting that in the case of this species, dry stress may not act as a key factor in areas under horticulture practice, but may be of importance when assessing large area-wide control or eradication programs. Moreover, in all three locations, MIW showed more variability than did TIW, suggesting that annual fluctuations in the effect of moisture on the medfly's population dynamics should be taken into account when assessing risks to horticulture.

The results also indicated that the potential geographical distribution of the medfly in Australia includes much of the low-altitude habitats in the south of the continent, including parts of Tasmania. Historical infestations in Melbourne and Launceston, Tasmania (Bateman 1971), give confidence in these results. The analysis confirms the feasibility of the "carry-over" hypothesis related to the incidence of medfly in Adelaide in South Australia (Maelzer 1990), which has a substantially milder winter than locations in southern Europe where the medfly is permanently established. The same conclusion can be drawn about the ability of the medfly to overwinter in Los Angeles (Carey 1990) based on the current model (Fig. 9). In Australia, Allman (1939) and Noble (1942) described how the medfly had declined to very low numbers in New South Wales (NSW) from the 1920s, concurrent with B. tryoni spreading south from Queensland around the turn of the century and becoming the dominant fruit fly pest in NSW. The last medfly was seen in Sydney in 1941 (Bateman 1971). The reason for this change of status is not known and has been the subject of much speculation around the competitive role of B. tryoni (Allman 1939; Andrewartha and Birch 1954; Bateman 1971; Fitt 1989). Much of northeastern Australia is highly suitable for B. tryoni (Yonow and Sutherst 1998), and the medfly has been displaced elsewhere by Ceratitis rosa Karch in Mauritius and Bactrocera dorsalis Hendel in low-altitude areas in Hawaii (Fitt 1989). The medfly never became established in Queensland (Tryon 1927), where B. tryoni is most numerous. An unlikely contributing factor could have been the release in the 1930s of nonspecific parasitoids aimed at the control B. tryoni. They were sourced from Hawaii and India in medfly pupae to which they were well adapted (Noble 1942). Although the parasitoids were not recovered from B. tryoni in the field in NSW, it is not clear whether the search included the medfly. Nevertheless, the declining trend of this latter species in NSW was already apparent before the release of the parasitoids.

The CLIMEX projections indicated that the climate of South-Western Australia was also suitable for the persistence of medfly, consistent with the fruit fly's current presence there (Fisher et al. 1985). To confirm the climatic suitability of coastal Queensland for medfly, as indicated by the CLIMEX analysis, a direct climatic comparison was made between the meteorological data of São Paulo (Santos) in Brazil, where medfly is endemic (Suplicy et al. 1987), south to at least the Porto Alegre area (Garcia and Corseuil 1998) and Australian locations. This was done using the "Match Climates" function of CLIMEX. The closest overall climatic match to an Australian meteorological station was at Double Island Point in Queensland, ≈250 Km north of the NSW border; and the closest temperature match was Sandy Cape and Innisfail in north Queensland. The fact that locations in areas of Brazil where the medfly is well-established have a climate very similar to coastal Queensland, provides conclusive evidence that the climate in Queensland per se was not responsible for the failure of medfly to colonize the state. This supports the hypothesis that it was excluded by biotic factors such as competition with B. tryoni in the warmer habitats of Australia.

These results highlight the original caveat that was emphasized by Sutherst and Maywald (1985) that it is necessary to check for nonclimatic limiting factors when estimating the potential geographical distribution of a species. They also showed how an inconsistency in the projections compared with observed distributions could be informative and point to nonclimatic factors operating. Thus, it is important to consider other possible limiting factors in the region from where the parameters are being estimated.

In the present case, the parameters were estimated from the Mediterranean region, and the lack of fit with the current distribution in eastern Australia was explained by the displacement of C. capitata by B. tryoni. If an attempt had been made to estimate the param-eters using the current geographical distribution in Western Australia, the resultant model (which would have had to include irrigation to fit the southern part of the distribution) would have included winter and summer rainfall areas. Hence it would also have flagged the climatic suitability of the east coast of Australia for the fruit fly. Such apparent discrepancies were highlighted by Sutherst and Mayald (1985), who suggested that they are a potentially valuable source of new insights into areas at risk. Consideration of the requirements for irrigation by the medfly would have greatly reduced the likelihood of excluding the east coast of Australia from the area at risk, because irrigation simulates the summer rainfall pattern in the east.

The benefits from validation tests of such models against independent data in other continents are evident from this study, as stated by Sutherst and Maywald (1985). It also highlights the advantages of having a biological perspective when using models, rather than relying on statistical fitting routines to estimate parameter values when pattern-matching meteorological data. Each model constitutes a scientific hypothesis of the factors governing the geographical distribution and so demands biological interpretation of the possible processes involved. The deficiency encountered with attempts to estimate the parameter values from the European distribution, where it was necessary to fall back on experimental data to estimate the medfly's response to high temperature and rainfall, also shows the need to take account of the available amount of variation in any given geographical area when fitting CLIMEX models, as stated by Sutherst and Maywald (1985).

Worner (1988) used biological parameters obtained from reported studies based mainly on laboratory assays to estimate the potential distribution of medfly in New Zealand using CLIMEX. When these parameters were used to estimate the potential geographic distribution in Argentina, the resultant distribution was more restricted than the observed range (data not shown), suggesting that the parameters should be updated in line with the present results. As stated before, it is preferable to use geographical distributions to infer parameter values because laboratory data do not incorporate the effects of environmental heterogeneity, or diurnal ranges of temperature and microclimatic effects that are expressed in the distribution. However, it is not possible with such observations to accurately determine threshold values for development or survival because there is a correlation between the threshold value and the rate of accumulation of stress in linear functions such as those used in CLIMEX (Sutherst et al. 1995). For example, while laboratory studies may indicate a threshold for temperature stress <10°C, the duration of tolerance of different temperatures will vary. Hence a more severe threshold associated with a lower rate of accumulation of stress will give similar answers to a less severe threshold with a higher rate of stress accumulation. In the field, the effect is expressed as a slope in the ramping of populations at the boundaries of geographical distributions (Sutherst et al. 1995). Hence, if good data exist on population densities along a transect across the boundary, it should be possible to confirm the most appropriate combination of threshold and rates. This will rarely happen and so approximations are necessary, supported where possible with laboratory observations.

In conclusion, we have shown that it is possible to infer a species' climatic requirements from its geographical distribution, provided that certain caveats are noted. The present analysis demonstrates an independent validation of the CLIMEX model in Argentina and Australia, but highlights the need for vigilance in identifying factors that override climatic factors in excluding species from certain habitats. The results also demonstrate the extent to which the medfly threatens horticulture around the world and illustrate how observed behavior of local populations, such as those in Argentina, can be interpreted using the CLIMEX model.

Acknowledgments

The authors thank Rick Bottomley for his assistance with the illustrations and H.A.C. Fay for providing a map of the overwintering distribution of medfly in Europe. The global climate surface was provided by the Climate Research Unit of the University of East Anglia, and Gunter Maywald provided support for the CLIMEX model.

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/), which permits non-commercial reuse, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions{at}oup.com

References

View Abstract