Integrated pest management (IPM) is the dominant paradigm that guides most aspects of current research in and implementation of insect pest management. The philosophy and historic origins of the approach are well documented (Perkins 1982; Walter 2003) and, as management systems have developed, various attempts to categorise developmental changes or phases leading to the accepted model have been made (e.g. Kogan 1998; Pedigo 2002). Despite clear, but case-specific, definitions in the numerous theoretical and applied works, which advocate its many virtues, the practical meaning of the ‘I’ in the much used acronym has become a matter of contention. It rarely represents true ‘integration’, usually at best symbolises ‘improved’, and in some circumstances might be somewhat cynically defined as ‘incidental’ to pest management.
Integrated pest management is generally recognised as a reaction to an overuse of insecticides and the management crisis that subsequently ensues (Barfield & Swisher 1994); yet from its earliest days the paradigm has come to encompass two conflicting schools of thought. The first advocates the responsible use of pesticides and has been variously described as ‘the ecologist armed with chemicals’ (Perkins 1982), ‘pesticide management’ (Brunner 1994) and ‘tactical IPM’ (Barfield & Swisher 1994). It is an approach that dates back at least to Michelbacher and Smith (1943) which aspires to rational pesticide use by developing intervention thresholds to be used in conjunction with sampling schedules that determine pest densities, enable the proper timing and, if more than lip service is paid to predators and parasitoids, judicious use of appropriate selective pesticides (see Stern et al. 1959; Binns & Nyrop 1992). This essentially ‘sample, spray and pray’ (SSP) approach is perhaps the dominant form of IPM in most crops. The second school of thought places greater emphasis on a somewhat nebulous, often ill-defined, ‘understanding of the agro-ecosystem’ before appropriate interventions are applied. This has been referred to as ‘strategic IPM’ (Barfield & Swisher 1994) or ‘real IPM’ (Brader 1988) and places far greater emphasis on natural enemies.
Whichever school of thought is subscribed to, the approach was developed in Western agriculture and consequently the individual farmer or manager retains responsibility for decision making in a given management unit, usually represented by a single crop in a particular field. Such fields are managed as essentially independent units often with scant regard for events in adjacent or proximate fields, let alone those considered distal to the management unit. For various reasons, not least the way IPM research and extension tends to be conducted amid various conflicting interest groups (farmers, researchers, chemical company representatives, pest management consultants, environmental groups, local community), implementation and adoption of the technology is almost always less than perfect (Trumble 1998). Nevertheless, in some cropping systems, particularly in high value crops, such as cotton, tactical IPM is at least widely attempted (see Morse & Buhler 1997 and below). Although this widespread implementation has been beneficial and might have rationalised pesticide use to some extent, has it ‘solved’ or even mitigated key pest problems? Are IPM practices ‘stable’ in the sense of Perkins (1982) and/or sustainable, or do we continue to lurch from crisis to crisis in response to uncontrollable incidental events? Such events might include the appearance of new pests detected within a crop or following breaches in quarantine, the evolution of insecticide resistance in endemic pests, outbreaks caused by weather, or increased susceptibility following from new cropping or marketing practices. In such circumstances does the notion of IPM simply enjoy transient reprieves when new selective insecticides or genetically engineered crops reach the market?
The rationalisation of pesticide use is a key mantra of IPM. Consequently widespread implementation of IPM throughout a localised agricultural sector should be expected to result in a long-term reduction in pesticide use in that sector, particularly if the approach replaces scheduled pesticide application regimes. Although reduced pesticide input and increased profit tend to accrue to individual adopters of IPM, these impacts are not necessarily the net outcomes of widespread IPM adoption (Taylor 1980). Furthermore, even rational commercial operations are likely to adopt increased pesticide application in preference to IPM under certain circumstances; e.g. Fenemore and Norton (1985) found that high pesticide use rather than IPM was more likely to be used in high value fruit crops because of reduced risk of producing unacceptable cosmetically damaged fruit, the high costs of monitoring/scouting in order to reduce pesticide use and economies of scale achieved by tank mixing groups of pesticides (e.g. fungicides and insecticides) when no viable alternatives to a single disease problem existed.
In our analysis, we examine some Australian case studies of what are widely considered to be successful IPM programs. We have chosen scale insect management in citrus production in Queensland, management of insect pests of Brassica vegetable crops in the Lockyer Valley and Helicoverpa spp. management in cotton in Queensland and New South Wales. These offer contrasting examples of pest management problems. Citrus is a perennial crop beset by many exotic imported key pests, Brassica crops are cultivated as a series of contiguous short-term (2-month) crops grown sequentially over 9–10 months and are attacked by a complex of mainly exotic lepidopteran host specialists and cotton is a relatively long-lived field crop (5–6 months) attacked by a large complex of pests that is dominated by polyphagous noctuid moths. All three systems have experienced a pest management crisis, or ‘disaster’ phase; essentially when the insecticides available at the time failed to provide economic control. We draw lessons from these case studies, consider whether the IPM paradigm, in whatever guise, can work for certain classes of pest insects and question whether it can be truly effective for any pest insect at the landscape level. We argue it might be time for a reappraisal of the paradigm and suggest how it might need to be amended.
Insecticide use in Australia: just how ‘green’ is our agriculture?
For some cropping systems and pest problems in Australia, the lack of effective alternatives to scheduled insecticide applications still constitutes a major constraint to IPM implementation. The problems posed by fruit-spotting bugs and fruit-piercing moths in certain horticultural crops (Fay 2002) and redlegged and blue oat mites in broad-acre canola crops (Furlong et al. 2008), are examples of co-occurring pests that can confound the development of IPM programs. Nevertheless in some cropping systems, particularly in high value crops, such as cotton and Brassica vegetables, at least tactical IPM is considered to be widely practiced (see below). Indeed most growers of these crops would consider themselves to be implementing IPM if surveyed. Objective measures of the ongoing success and degree of implementation of programs are difficult to obtain. Most attempts to evaluate IPM programs measure the effectiveness and potential sustainability of a given strategy and its economic benefit over short timescales (e.g. Trumble et al. 1997). Long-term adoption is more difficult and costly to measure and not the kind of activity readily supported by funding agencies beyond the lifetime of projects designed to develop or implement IPM (but see Horne et al. 1999; McDougall 2007).
The idea of IPM, even if conflicted, has been promoted in all cropping systems in Australia over the last 30 years. Consequently, it might be expected that insecticide expenditure per ha should have at least stabilised, or maybe even declined, during this period. The data show that the cost of insecticide inputs per ha into Australian agricultural crops increased dramatically in the late1980s because of increasing real costs (Fig. 1). During the 1990s real insecticide costs per ha continued to increase dramatically, despite the price index reaching a plateau early in the decade (Fig. 1); this indicates that insecticide inputs actually increased throughout the period. Real insecticide costs have continued to increase over the last decade with the notable exception of 2001, when inputs were significantly reduced because of devastating drought (Fig. 1). The increased real costs of insecticide inputs during the period can be explained in part by changes in crop composition across the agricultural landscape. Wheat, which has relatively low insecticide input costs, has been gradually replaced by other crops (e.g. various oilseeds, other grains) which have higher insecticide input costs (Fig. 1; ABARE 2007) and are managed using newer, more expensive, insecticides.
Figure 1. The real cost of insecticide input into Australian crop systems, the insecticide price index, the total crop area and the ratio of the area of wheat cropped relative to other crops from 1974/75–2005/06 (ABARE 2007).
Download figure to PowerPoint
Citrus IPM in Queensland
Many of the more than 90 major pests of citrus production in Australia are exotic species (Smith et al. 1997) which arrived in Australia before quarantine procedures were well established, although these barriers continue to be breached (Smith et al. 2005). IPM in Queensland citrus began in 1973 in response to the failure of insecticides following their overuse and the evolution of insecticide resistance in key pest species (Papacek & Smith 1989). The citrus IPM program was founded on successful classical biological control for imported pests, such as white wax scale (Sands et al. 1986), California red scale (Smith 1978a), circular black scale (Smith 1978b), pink wax scale (Smith 1986) and white louse scale (Smith & Papacek 1995), as well as augmentative biological control against the dominant scale pests when required (Smith & Papacek 1993; Smith et al. 1997). In the absence of disruptive insecticides, including the use of bait sprays for fruit fly control (Smith & Papacek 1985) and more recently on an area-wide basis (Lloyd et al. 2007), these biological control agents provide effective control of the major scale pests and reduced insecticide input has also resulted in greater mortality of other minor pests and a reduction in their impact because of the activities of natural enemies (Papacek & Smith 1989). Spider abundance can be much greater in IPM plots than in conventionally managed orchards in Queensland (Green 2005); this presumably contributes to pest mortality although, as in many other agro-ecosystems, this remains to be experimentally tested.
The success of the citrus IPM program hinged in part on its adoption by one of the major growers in the 1978–1979 season. By 1985, approximately 40% of citrus growers were using IPM and, by 1991–1992, this had risen to 75% (Smith & Papacek 1993). Costs of management when using IPM ranged from $237/ha to $421/ha compared with $941/ha to $1784/ha for conventional chemical control (Smith & Papacek 1993), although these comparisons are over time and not via controlled coincident experiments. A 75% reduction in pesticide use has been achieved through a combination of pest and natural enemy monitoring (on which interventions, such as pesticide application or augmentative release of parasitoids and/or predators are based) and system modifications, such as reduced mowing of inter-row grasses to increase the prevalence of predatory mites (Smith & Papacek 1993).
This example of an IPM success story has certain features. Citrus trees are long-lived and the agro-ecosystem is not disrupted by regular ploughing or planting. Many of the key pests are exotic in origin and successful classical biological control programs have resulted in their long-term suppression (see DeBach 1974). It sits squarely in the ‘real’ or ‘strategic’ school of IPM. The program is also effectively area wide in as much as the majority of growers appear to have adopted the practices and benefited from the introduced natural enemies, although large scale or landscape level effects on management within fields are rarely considered (Schellhorn et al. 2008). Other successful IPM programs in Australian fruit crops include the use of pheromones for mating disruption and control of oriental fruit moth, Grapholita molesta Busck and codling moth, Cydia pomonella L. in Victoria (Il'ichev et al. 1998, 2002, 2007; Il'ichev & Williams 2006). In these examples too, an area-wide scale of application appears to be critical.
Brassica IPM in south-east Queensland
Brassica crops in south-east Queensland are attacked by a complex of lepidopterous pest species which includes diamondback moth (Plutella xylostella), cabbage cluster caterpillar (Crocidolomia pavonana), cabbage white butterfly (Pieris rapae), cabbage centre grubs (Hellula hydralis and H. undalis), cluster caterpillar (Spodoptera litura) and bollworms (Helicoverpa sp.). All members of the pest complex rarely occur in unison and relative abundance is influenced by species-specific seasonal phenologies that reflect the origins of each insect. Diamondback moth and cabbage white butterfly, both of temperate origin, are most abundant in spring, winter and autumn months while the indigenous centre grubs and bollworms and the tropical cabbage cluster caterpillar are more abundant in summer and spring. Together these latter species represent what are often referred to as ‘early season pests’.
As elsewhere in the world diamondback moth is the most difficult pest to control with insecticides because of the evolution of resistance to a wide range of chemical groups (Wilcox 1986; Hargreaves 1996; Endersby et al. 2008). Public concern over environmental contamination and food residues combined with insecticide failures, which resulted in large crop losses in the 1980s, prompted the formulation of an insecticide resistance management strategy, the ‘Three Valley Strategy’, in late 1980s (Deuter 1989). Grower groups formed in that period were built on in subsequent initiatives (funded by the horticultural industry and state and federal governments) throughout the 1990s, which aimed to develop alternative pest management strategies to the prophylactic application of broad-spectrum insecticides (Heisswolf et al. 1997). From the mid-1990s, the use of formulations of highly selective Bacillus thuringiensis (Bt) were promoted to preserve endemic natural enemies within crops and available funding strengthened links between grower groups and the state government extension service. The participatory approach to research and extension is credited as a major reason for the success of the program (Heisswolf et al. 1997).
Pest management practices in Brassica crops in the southeast Queensland changed between 1990 and 2002 (Fig. 2). The introduction of a 3-month Brassica production break during the summer months (November to January) was a pillar of the Three Valley Strategy. Designed to minimise the abundance of available host plants of the diamondback moth and, thereby, minimise the locally breeding pest population, the production break was combined with an area-wide policy, which recommended the application of different groups of insecticides in distinct temporal windows in order to mitigate the evolution of insecticide resistance (Deuter 1989). By 1996, over 70% of farmers in the region had used the practice (Fig. 2). However, the practice has declined ever since and there has been a concomitant rise in the proportion of growers using a break of 1 month (Fig. 2). In 2002, although 95% of Brassica growers in the region used a production break, 51% ceased production for only 1 month. Although a production break of 1 month might reduce the pest population on a fine local scale (i.e. a farm), an overall decrease in the duration of the production break on the regional scale will result in a greater availability of host plants in both space and time, a problem that will be exacerbated by lack of coordination of breaks between properties. The use of broad-spectrum insecticides in the region declined markedly between 1990 and 2002 and the intensive extension activity in the mid- to late 1990s saw well over 90% of growers use Bt formulations (Fig. 2). The use of Bt has subsequently declined because of the availability of spinosad (Success®) and indoxacarb (Steward®), two new-generation insecticides which, although selective, can be far more disruptive to key natural enemies of the diamondback moth than applications of Bt (Chen et al. 2008).
Figure 2. Trends in pest management practices targetting the insect pests of Brassica crops in the Lockyer Valley (1990–2002). *, organophosphate and pyrethroid insecticides; **, spinosad and indoxacarb (Heisswolf et al. 1997; L Bilston pers. comm. 2003).
Download figure to PowerPoint
The proportion of Brassica farmers who scout their crops for pests has steadily increased whereas, interestingly, the proportion of farmers who consider that they practice IPM has remained static (60–68%; Fig. 2). Undoubtedly, farmers in the region are now more aware of the agro-ecosystem that they manage and they are conscious of the potential benefits of natural enemies. However, whether or not IPM is widely practiced is debatable: rather, evidence suggests that farmers make judicious use of newer highly effective selective insecticides (Fig. 2). Currently, many growers in the region do not consider diamondback moth to represent the severe problem that it once did and credit their perceived adoption of IPM for the change. However, an alternative explanation is that regional abundance and long-term population dynamics of diamondback moth is a function of climate (Zalucki & Furlong 2008) and has little to do with management practice except at a fine scale. The changes in management practices, which are often equated with the successful widespread adoption of IPM, may in fact be incidental to a decline in abundance, which is due to the less favourable climatic conditions that have prevailed since 1989 (Fig. 3).
Figure 3. The annual growth index for Plutella xylostella at Gatton research station from 1965 to 2003, based on a CLIMEX model (Zalucki & Furlong 2008) and major initiatives to improve P. xylostella pest management since the crises experienced in the 1980s.
Download figure to PowerPoint
The apparent backsliding in IPM practice (Fig. 2) probably also reflects the lack of continued funding which prevents long-term contact between growers and extension officers being maintained. After the initial research and development phase of projects, when contact between farmers and researchers/extension officers is frequent, relationships wane as researchers move onto other funding-driven problems and extension considers the problem solved. Such problems are intensified by the lack of commitment to retain and develop capacity in applied insect ecology within universities and State and Federal agencies. This often results in loss of corporate knowledge, which exacerbates the problems of short-term cycles and perpetuates the inefficient practices of treating pest outbreaks as new phenomena when they inevitably recur.
Helicoverpa spp. IPM
For migratory, polyphagous, outbreak pest species, such as Helicoverpa spp. individual field-based tactical IPM continues to fail (Zalucki et al. 1998). Helicoverpa spp. are ‘key pests’ of many crops in Australia and their management has been the focal point of many IPM programs, most notably in cotton (e.g. Ives et al. 1984). This IPM program is held up as the Holy Grail to which other pest and crop systems should aspire, and a great deal of research and development funding has been invested towards this goal (e.g. the GRDC National Invertebrate Pest Initiative, CSIRO 2008). In light of these investments, we question whether or not the Helicoverpa spp. pest problem has been reduced and whether this is ever likely to happen. We argue that tactical IPM cannot succeed. Indeed, under certain conditions it might exacerbate the evolution of insecticide resistance and contribute to subsequent management crises. Focussing on cotton, we summarise the history of Helicoverpa spp. management in Australia and outline a possible alternative approach which is akin to Perkins' ‘Total Population Management’ paradigm (Perkins 1982), but which does not suggest Helicoverpa eradication. This is essentially an area-wide management approach (sensuKnipling & Stadelbacher 1983) that is commensurate with the spatial and temporal ecology of the pests (see Zalucki et al. 1986).
From the earliest days of cotton production in Australia, Helicoverpa spp. have figured prominently as major pests in all cropping regions, and have been managed or controlled with insecticides. In the early 1970s, resistance to DDT and other insecticides (Fig. 4) spurred the development of various approaches to IPM. Longworth and Rudd (1975) provided the first comprehensive economic analysis of insecticide use and cotton production in Australia and discussed the compounding issues of insecticides, insecticide resistance management and the need to develop ‘pest managers’. The period of 1974–1989 saw the promotion of crop monitoring and the development of sampling plans and intervention thresholds. These incorporated the capacity of the crop to compensate for damage as well as the implicit positive impacts of beneficial insects. Many of these developments were built into the computerised decision-support program SIRATAC and its successor, CottonLOGIC (see Hearn & Bange 2002). The prototype SIRATAC system was used to manage large (10 ha) experimental plots in the mid-1970s (Room 1979) but, by the mid 1980s, it was used to manage 44 613 ha of commercial cotton crops (Brook & Hearn 1990). Although the program initially targeted the management of Helicoverpa spp. its modular structure enabled other components of crop management (e.g. irrigation scheduling, fertiliser application) to be readily incorporated. Although SIRATAC was only used directly by approximately 35% of growers, its principles of pest sampling and the application of action thresholds were readily embraced by private consultants, growers and chemical companies and became the industry standard (Pyke 1985). Despite the widespread use of SIRATAC, pyrethroid insecticides, rather than the expert system, effectively saved the Australian cotton industry when H. armigera became unmanageable with DDT in the 1970s.
Pyrethroids first failed to control H. armigera in January 1983 at Emerald in central Queensland (Gunning et al. 1984). The evolution of pyrethroid resistance caused much consternation and an insecticide resistance management (IRM) strategy was rapidly developed and implemented (see Forrester et al. 1993). In cotton crops, pyrethroid use was restricted to the single H. armigera generation (stage II) in the middle of the crop season (early January to mid-February) and endosulfan was recommended for this and the earlier H. armigera generations (stages I or II), but not the third generation (stage III), for which organophosphate and carbamate insecticides were recommended. Post-season destruction of high-density population of diapausing H. armigera pupae (‘pupae busting’) was also suggested. As H. armigera is polyphagous and moves among crops, all or the majority of growers of susceptible crops need to conform to such a strategy for it to be effective. Despite a high rate of compliance, resistance to pyrethroids and endosulfan continued to increase, building up as each season progressed but declining by the beginning of the next season as the resistant alleles in the population were diluted by subsequent crossing with susceptible insects. Nevertheless, the long-term trend was a decrease in the susceptibility of H. armigera to these insecticides, such that in each year the level of resistance in a given stage was greater than that recorded in the same stage in the previous year (Daly & Fisk 1998). By the early 1990s, a new insecticide resistance crisis loomed and H. armigera exhibited high levels of resistance to all insecticide groups available: synthetic pyrethroids, organophosphates, carbamates and endosulfan (Gunning et al. 1992, 1996, 1997a; Forrester et al. 1993; Murray et al. 2005a,b) The broad-spectrum activity of these insecticides disrupted beneficial insect populations, allowing H. armigera populations to resurge, inducing extra insecticide applications and further selection for resistance (Murray et al. 1998). For cotton, at least, the solution was to use transgenic plants expressing Bt toxins (Fig. 4).
Recent studies have indicated that resistance levels for most conventional insecticides have stabilised or even declined in H. armigera (Rossiter et al. 2007). As in Brassica crops, SSP type IPM in cotton crops (most of which are now transgenic) might have been saved by more expensive selective insecticides, such as indoxacarb, spinosad and formulations of Nuclear Polyhedrosis Virus. These insecticides have been described as having a better ‘IPM fit’ than the broad-spectrum insecticides used in the 1970s and 1980s, as they are less disruptive to predators and parasitoids (Wilson et al. 2007), although the impact of particular insecticides can vary considerably between groups of natural enemies (Scholz et al. 1998); e.g. Indoxacarb has minimal impact on Trichogramma sp. (Scholz & Zalucki 1999) but significantly affects ladybirds (Wilson et al. 2007).
Figure 5. Abundance of both Helicoverpa pest species in September plus October at Narrabri (solid black column) and H. armigera (open columns) caught in the Myall Vale light trap. From 1966 to 1972 the two species were not separated. For the period 1977–1980 the data are only available in Wilson (1983). For the period 1987–2001 they come from Colin Tann and Geoff Baker (CSIRO) and are based on smaller light traps, but the catch has been adjusted to be equivalent, notionally, to larger original light trap based on three years' data when both traps were run simultaneously (WA Rochester, CSIRO, unpubl. 2003). The cross-hatched columns represent catches of H. punctigera in Tasmania (L Hill, Tasmanian Department of Agriculture, unpubl. 2007) in the same months from 1966 to 2005 (original values ×30 for scaling purposes). The Tasmanian data suggest both periodic outbreaks as at Narrabri and a general increase in abundance over time.
Download figure to PowerPoint
Thus, historically, Helicoverpa spp. management in cotton in Australia moved towards tactical IPM of the SSP variety upheld by decision-support systems (Hearn & Bange 2002), culminating in the widespread adoption of transgenic cotton (Fig. 4). Despite widespread adoption of this form of IPM and the rational use of insecticides, resistance to these products has evolved and outbreaks of Helicoverpa spp. still occur and pressurise the transgenic technology (Figs 4 & 5). The apparent lack of field resistance to Bt toxins might have as much to do with the polyphagous nature of the pests and the widespread increase in the areas of non-Bt host crops of Helicoverpa spp. (e.g. various grains), that has occurred recently (Fig. 1) as it does with strategies to manage insecticide resistance.
Clearly, the traditional approach to management of Helicoverpa spp. under the rubric of IPM is failing despite large investments in research. Over the last 10–30 years the pest status of these species has not decreased (Fig. 5). It appears that a fundamental problem is that the approach to IPM, whether it is tactical or real, attempts to manage the problem as an independent local issue at the scale of the farm field. This is a flawed approach for the management of highly migratory, multivoltine, polyphagous pests. The frequency and likelihood of outbreaks in agricultural regions in spring is not necessarily related to local conditions but reflects rainfall and breeding elsewhere (Fig. 5, above). Subsequent dynamics might well depend on local weather and crops but at a scale larger than the farm field on which management decisions are made. Helicoverpa spp. need to be managed on a season-long landscape basis (Brier et al. 2008; Schellhorn et al. 2008). Such an approach recognises crops as host plants for these species which allow populations to increase following early season colonisation and which contribute to the populations that later infest more valuable crops, such as cotton.
Similarly, in mixed cropping areas, sorghum and maize crops can contribute substantially to the local Helicoverpa spp. population size (Maelzer & Zalucki 1999); hence, management of the pests within early season hosts might be advantageous to the cotton farmer later, and management at the end of the season might reduce pest abundance in the subsequent year. Similarly, interventions in low value crops during the season might be needed to reduce subsequent pest populations in crops, which are managed intensively with insecticides. The economics of such interventions need to be viewed in the context of overall farm revenue. These ideas formed the basis of an area-wide management (AWM) strategy, tested (Murray et al. 1998) and then apparently successfully implemented on the Darling Downs in Queensland (Rochester et al. 2002; Murray et al. 2005a). Although the program of area-wide management appears to have worked, there is no mechanism for definitive validation and, as we have argued elsewhere (Zalucki & Furlong 2005), the abundance of Helicoverpa spp. during the period might simply reflect prevailing climatic conditions, which are reflected in the dramatic reduction in the area of cotton production (Fig. 4). The AWM approach requires all growers in an area to co-operate to suppress the population of Helicoverpa spp. and appreciates the futility of commodity-based distinctions by recognising that all crops that are potential Helicoverpa host plants might have contributed to the pest population that needs to be managed. The requisite size of an effective management unit is unknown but it is likely to be a landscape many orders of magnitude greater in area and complexity than an individual field.
Assuming that recent climatic conditions have not been primarily responsible for determining Helicoverpa spp. abundance, the portents for IPM in transgenic cotton are not encouraging when the economics of pest management in cotton are considered (Fig. 4). Even if it is argued that as a result of the introduction of transgenic cotton in 1997, Helicoverpa spp. populations have declined then, in economic terms, the approach has not been a success. The total insect management costs, which includes the $300/ha licence fee for Bollgard II, have not declined consistently when expressed as a percentage of the total production cost and they remained much the same in 2006 as they were in 1996 (Fig. 4). Indeed, Back and Beasley (2007) argue that the success of transgenic cotton in Australia lies in the social and environmental benefits which accrue and that these factors, rather than direct economic gains, have driven adoption. In years when water is scarce and prices are poor (e.g. 2003, Fig. 4) farmers might conserve expenditure on input costs in case of crop failure and such actions must be taken into account when interpreting historical data on management costs. A tangible measure of the success of the introduction of transgenic cotton is the decreased frequency of the detection of insecticide resistance in H. armigera populations (Rossiter et al. 2007), presumably related to the decline in insecticide applications for Helicoverpa in cotton (Brier et al. 2008) and, perhaps, the decline in cotton per se (Fig. 4). To maintain this success, an area-wide approach is essential and the inclination of individuals to plant conventional cotton varieties (and thereby save the fixed cost of the licence fee) when pest Helicoverpa populations are low needs to be discouraged. A critical question for AWM is how to keep the majority of people acting towards the public (i.e. their own) long-term good.