Insect herbivory and ontogeny: How do growth and development influence feeding behaviour, morphology and host use?
Abstract
Herbivorous insects exploit many different plants and plant parts and often adopt different feeding strategies throughout their life cycle. The conceptual framework for investigating insect–plant interactions relies heavily on explanations invoking plant chemistry, neglecting a suite of competing and interacting pressures that may also limit herbivory. In the present paper, the methods by which ontogeny, feeding strategies and morphological characters inhibit herbivory by mandibulate holometabolous insects are examined. The emphasis on mechanical disruption of plant cells in the insect digestive strategy changes the relative importance of plant ‘defences’, increasing the importance of leaf structure in inhibiting herbivory. Coupled with the implications of substantial morphological and behavioural changes in ontogeny, herbivores adopt a range of approaches to herbivory that are independent of plant chemistry alone. Many insect herbivores exhibit substantial ontogenetic character displacement in mandibular morphology. This is tightly correlated with changes in feeding strategy, with changes to the cutting edges of mandibles increasing the efficiency of feeding. The changes in feeding strategy are also characterized by changes in feeding behaviour, with many larvae feeding gregariously in early instars. Non-nutritive hypotheses considering the importance of natural enemies, shelter-building and thermoregulation may also be invoked to explain the ontogenetic consequences of changes to feeding behaviour. There is a need to integrate these factors into a framework considering the gamut of potential explanations of insect herbivory, recognizing that ontogenetic constraints are not only viable explanations but a logical starting point. The interrelations between ontogeny, size, morphology and behaviour highlight the complexity of insect–plant relationships. Given the many methods used by insect herbivores to overcome the challenges of consuming foliage, the need for species-specific and stage- specific research investigating ontogeny and host use by insect herbivores is critical for developing general theories of insect–plant interactions.
INTRODUCTION
Herbivorous insects adopt a range of feeding strategies to successfully consume living plant material, encountering and overcoming a complex suite of plant defences and natural enemies to do so (Bernays 1998; Schoonhoven et al. 1998). Explanations of insect–plant interactions typically address the ‘green world’ phenomenon, reflecting on the inability of insects to successfully consume all plants (Hairston et al. 1960; Murdoch 1966; Bernays & Graham 1988; Abe & Higashi 1991). The merits of general theories attempting to explain insect–plant interactions are often weakened by the sheer diversity of herbivores and food plants and complicated by generalizations based on limited experimental evidence (Peters 1995; Beck 1997). Nevertheless, the explanatory power of general conceptual frameworks supported by extensive experimental evidence necessitates the search for general patterns and the processes underpinning them.
In the present paper, I consider how ontogeny, feeding strategies and morphological characters inhibit herbivory by mandibulate holometabolous insect herbivores. I discuss how factors that limit herbivory may be scale-dependent and describe a framework integrating digestive strategy, functional morphology and feeding behaviour with the properties of food plants. The broader ecological consequences of feeding strategies and non-nutritive explanations of their significance are also described. The focus is on insects successfully consuming living eucalypt foliage, although the implications are relevant to the broader range of potential interactions.
Discussion of insect herbivores in this paper is limited to mandibulate insects with ‘chewing’ mouthparts, of which the Lepidoptera, Coleoptera, Orthoptera and Hymenoptera are most prominent as consumers of living foliage (Schoonhoven et al. 1998). The larvae of holometabolous insects can utilize size changes and developmental flexibility during ontogeny so that a range of feeding strategies, such as leaf mining, skeletonizing and leaf snipping can be used (Gaston et al. 1991). Similarly sized larvae adopting superficially similar feeding strategies may also exhibit considerable morphological variation leading to differences in the way they consume leaf material (Bernays & Janzen 1988). The substantial size changes (often over several orders of magnitude (Reavey 1993)), exhibited by larvae of holometabolous insects constrains feeding behaviours and is the basis for reconsidering feeding behaviour in the context of ontogeny and digestive strategy. Although most previous work within these limits focuses on larval lepidopterans (see Stamp & Casey (1993)), I will follow Reavey (1993) in making sawfly larvae ‘honorary Lepidoptera’, and will also include some larval Coleoptera owing to their similar ecological habits.
The eucalypts (which, for the purpose of this paper, refer to species within Eucalyptus, Angophora and Corymbia) are an abundant food resource for insect herbivores in Australian habitats and are successfully exploited by a range of species adopting a range of feeding strategies (Farrow 1996; Elliott et al. 1998). Utilization of eucalypt foliage by herbivores is limited by its low nutrient concentration as well as the presence of plant secondary metabolites acting as digestibility reducers and toxins (Hume 1999). The acquisition of nutrients is also potentially limited by the high degree of sclerophylly in eucalypts (Morrow 1983; Edwards et al. 2000) as well as a broad spectrum of mechanical defences (Lanyon & Sanson 1986; Hochuli 1996). The use of Eucalyptus foliage by herbivores has been extensively reviewed (e.g. Ohmart & Edwards 1991; Landsberg & Cork 1997), revealing the strategies used by a diverse group of herbivores to overcome the barriers to successfully consuming eucalypts. These reviews also suggest that eucalypts in Australia are characterized by high levels of herbivory, although the extent to which they differ from Northern Hemisphere forest trees is complicated by methodological and philosophical approaches, as well as inherent variation within the taxon (Fox & Morrow 1983; Abbott et al. 1993; Landsberg & Cork 1997). Nevertheless, the diverse group of herbivores on eucalypts and the high levels of herbivory they generally exhibit make them an excellent model system on which general hypotheses explaining herbivory can be tested (Landsberg & Cork 1997). This is also valid in the context of the present paper, as many prominent insect pests on eucalypts exhibit the ontogenetic patterns described subsequently (Common 1990; Elliott et al. 1998). These include the lepidopteran species, Doratifera casta Scott (Limacodidae), Uraba lugens Walker (Noctuidae) and Mnesampela privata (Guenée) (Geometridae) and the coleopterans Chrysophtharta argicola (Chapuis) and Chrysophtharta bimaculata (Olivier) (Common 1990; Farrow 1996; Elliott et al. 1998; Nahrung et al. 2001).
DIGESTIVE STRATEGY AND LIMITS TO HERBIVORY
Digestive strategy and its consequences are central to understanding how insect herbivory may be limited or influenced by plant attributes (Hochuli 1996). Explanations of insect–plant interactions have focused on the role of plant secondary chemicals (e.g. Rosenthal & Berenbaum 1991) to the extent where numerous alternatives considering mechanical attributes (Fernandes 1994), predators and parasitoids (Haukioja 1993; Heinrich 1993) and the interactions among factors (Stamp & Casey 1993) have been identified as under-studied despite their importance. An integrative approach considering interactions among the complex factors influencing herbivory is clearly required to resolve the simplistic explanations usually offered to explain how insects successfully exploit plants (Oedekoven & Joern 2000).
Because of their relative inability to digest and chemically disrupt plant cell walls (Abe & Higashi 1991; Martin 1991; Hochuli & Roberts 1996), most herbivorous insects rely on the mechanical disruption of cell walls to obtain nutrients from cell contents (Hochuli 1996; but see Barbehenn 1992). This emphasis on mechanical disruption in the insect digestive strategy may change the relative importance of different plant ‘defences’, increasing the importance of mechanical attributes inhibiting the initial acquisition of nutrients (Hochuli 1996).
Interpretation of previous work investigating plant mechanical defences has been complicated by methodological problems (Choong 1996; Sanson et al. 2001), particularly with respect to the estimates of the hardness and toughness of leaves. Although measuring these attributes is complex, the consequences for herbivores can be described simply. ‘Toughness’ is the work required to fracture the leaf, making it difficult to break open and requiring enough muscle and effective mandibles to propagate a crack in the leaf. ‘Hardness’ may best be considered with respect to the relative content of very hard components in the diet (e.g. silica content of grasses; Georgiadis & McNaughton 1990) rather than a property that can be attributed to a leaf. A major consequence of consuming diets with high silica content will be mandibular wear induced by mandibles grinding against hard components in leaves and each other during leaf snipping (Raupp 1985; Hochuli 1994). The consequences of consuming leaves with extreme mechanical properties are likely to be reflected in head and mandible morphology, as well as performance on mechanically challenging diets (Raupp 1985; Bernays 1986).
The roles of natural enemies in limiting herbivory may also have been understated (Montlorr & Bernays 1993; Bernays 1998) and are also implicated in understanding feeding strategy. Plant defences affect the ease with which plant material can be ingested and may interact with pressures from predators and parasites of herbivores by forcing them to modify their foraging behaviours (Heinrich 1993; Mueller & Dearing 1994; Oedekoven & Joern 2000). Furthermore, natural enemies may play an important role in enhancing the value of ontogenetically imposed feeding strategies (Hunter 2000; also see the following discussion of gregarious feeding), as well as the development of mutualistic associations often associated with larval communal behaviour (e.g. lycaenid butterflies and ants; Eastwood & Fraser 1999).
Although plant chemistry is clearly a critical factor influencing the success of insect herbivores, the first step in generating a general scheme for describing the potential of leaf attributes to limit herbivory is to evaluate how the herbivore is exposed to plant chemical defences by considering whether barriers to the mechanical disruption of plants exist. In doing this, the link between digestive strategy and feeding must reflect the scale-dependent nature of the mechanical challenges leaves pose for larvae. Being a particular size determines which feeding strategies are available to larvae, which will in turn facilitate the adoption of morphologies and behaviours suitable for particular plants and leaf stages. The interrelationships between larval size, morphology and feeding behaviour and the consequences of these are explored in the following sections.
ONTOGENY AND FEEDING BY HERBIVOROUS INSECTS
Size and ontogeny
The larvae of herbivorous insects usually undergo significant size changes over several orders of magnitude during development, often exhibiting dramatic changes in feeding behaviour and appearance (Greene 1989; Reavey 1993). Reavey (1993) reviewed the role of body size in caterpillar foraging and concluded that it was a key character influencing foraging; he also identified the need for a framework integrating case studies on individual species from a ‘scattered and rather sparse literature’. His review identified many of the characteristics of foraging associated with body size, and generated hypotheses explaining the consequences of challenges faced by caterpillars as they grow. Among the general trends described, modes of nutrient acquisition (i.e. the way leaf material is ingested) and levels of gregariousness were identified as traits that often changed as larvae grew larger.
Larvae operate at different scales at different times in their life cycles, and the different modes of nutrient acquisition adopted may reflect simple mechanical constraints imposed by size (Gaston et al. 1991; Reavey 1993). The continuum of feeding strategies encompasses internal feeders selectively consuming mesophyll tissue through to external feeders consuming entire leaves. Early-instar caterpillars feed differently from late instars, with the most common change for larval microlepidopterans being from leaf mining to external feeding (Gaston et al. 1991). External feeding, as practised by many macrolepidopterans, also encompasses several habits that are likely to be dictated by larval size, such as skeletonizing and snipping (Barbehenn 1992; Hochuli 1994).
Early-instar caterpillars show a greater tendency to feed gregariously, with many species feeding solitarily in later instars (Fitzgerald 1993). The larvae of many species of herbivorous insects, including various families of the Lepidoptera (e.g. Stamp 1981; Damman 1987), Hymenoptera (e.g. Ghent 1960; Macdonald & Ohmart 1993) and Coleoptera (e.g. McCauley & O’Donnell 1984; Breden & Wade 1985), have gregarious larvae. There are numerous potential costs and benefits of gregarious feeding behaviour (Hunter 2000; see following) and their relative importance is species-specific and scale-dependent (Hunter 2000; Zaviezo & Mills 2000). Strongly gregarious behaviour may also be retained in later instars (e.g. pergid sawflies, MacDonald & Ohmart 1993) indicating benefits independent of size.
Feeding behaviour
The major factor influencing the mode of nutrient acquisition is head and mandible morphology, as smaller heads and mouthparts will facilitate selective feeding on leaves in a manner that is impossible in later instars (Reavey 1993). Similarly, the larger heads and mandibles of late-instar larvae facilitate the efficient ingestion of greater quantities of leaf material than is possible for early instars (Bernays 1986). The consequence of this general shift is reduced assimilation efficiency in later instars (Slansky & Scriber 1985), probably as a result of ingesting a higher proportion of low quality food (Slansky 1993). Many early-instar lepidopteran larvae selectively consume portions of the leaf that are low in fibre and easily digestible, but as they develop, the larvae become less discriminating and consume all but the sclerenchymous tissue and veins (e.g. Kogan & Cope 1974; Khalsa et al. 1979).
Gregarious feeding may be an unavoidable consequence of egg-laying behaviour, with several of the benefits of gregariousness in caterpillars being applicable to oviposition strategies. Laying eggs in clustered masses, rather than individually, may reduce the effect of parasitoids and predators and render eggs less vulnerable to dehydration (Stamp 1980). Aggregated larvae may also have an advantage over isolated conspecifics because of increased thermoregulatory ability (Evans 1982; Stamp & Bowers 1990), diluted predation risk (Stamp 1981; Cornell et al. 1987; Stamp & Bowers 1988) and increased foraging efficiency through facilitation (Ghent 1960; Cornell et al. 1987;). However, larvae feeding gregariously may suffer higher intraspecific competition when suitable leaf material is limiting and increase the quality of foraging patches for parasites, which suggests that there may be substantial costs associated with gregarious feeding behaviour.
Solitary animals are more susceptible to parasitoids than their grouped conspecifics (Stamp 1981). The parasitoid-related reduction in fitness of solitary animals can be direct (greater mortality of the parasitized animals) or indirect (decreased growth and development times; Stamp 1981, 1982; Stamp & Bowers 1988). Grouped larvae may display lower mortality than solitary larvae because they are better able to simultaneously respond to predator attacks than their isolated conspecifics, using more effective defensive movements, release greater amounts of defensive chemicals, or build more elaborate and effective shelters (Prop 1960; Myers & Smith 1978; Stamp 1984). Decreased growth may also result from balancing trade-offs between predation risk and resource acquisition, with solitary animals displaying diligent antipredator behaviours more often, which reduce the amount of time available for foraging (Stamp & Bowers 1988).
Solitary early-instar larvae may suffer higher mortality than grouped larvae because of the difficulties of establishing a foraging site (Ghent 1960; Tsubaki 1981; Breden & Wade 1989). Termed ‘establishment mortality’ by Ghent (1960), larvae in groups have an increased chance of overcoming the physical barrier of the leaf surface presented to small larvae and consequently display greater survivorship than isolated conspecifics. Nahrung et al. (2001) manipulated group sizes of larvae of leaf beetle C. argicola and found that larger groups of larvae suffered lower mortality when feeding on mechanically challenging older leaves of Eucalyptus nitens, with neonate establishment being critical to the success of individuals. Neonate larvae of C. bimaculata also suffer difficulty in establishing feeding sites on tough eucalypt leaves (Howlett et al. 2001). These difficulties in establishing foraging sites have been linked to ontogenetically imposed morphological limitations in mandible size (Nahrung et al. 2001). However, as larvae grow and their mandibles and associated musculature develop, the differential between the survival of grouped and solitary larvae decreases, as has been shown for a number of gregarious larvae, including Neodiprion pratti banksianae Roh. (Ghent 1960), Plagiodera versicolora (Laicharting) (Breden & Wade 1989), and D. casta and M. privata (Sutherland 1999)
Mandibular morphology and function
Ontogenetic changes in head and mandible morphology parallel the changes in feeding behaviour, with mandibular morphology changing substantially throughout larval development. Caterpillars replace mandibles at each moult, offering the potential for several morphological strategies to be adopted throughout their feeding life (Hochuli 1994). This is supported by the prevalent association between changes in feeding strategy and mandibular morphology (Weller 1987; Dockter 1993). For instance, the mandibles of most notodontid larvae change from having a toothed cutting edge in the first instar to a smooth edge in later instars; first instars skeletonize the leaves, leaving the epidermis intact, whereas later instars cut through the full leaf thickness (Godfrey et al. 1989). These ontogenetic changes in mandibular morphology are also observed in Persectania ewingii (Westwood) (Noctuidae), where younger larvae scrape and gouge the surface of grasses before snipping leaf material in later instars (Hochuli 1994).
General patterns in morphological changes suggest substantial ontogenetic character displacement, including the development of a retinaculum on the oral surface of mandibles, reduced prominence of points on the cutting edge, and the development of a bladed mandible (Dockter 1993; Hochuli 1994; Sutherland 1999). Sutherland (1999) found that the mandibles of D. casta and M. privata larvae undergo some changes with development. The early-instar larvae of both species have distinct teeth and ridges in early instars, when they skeletonize the leaf surface. Mandibles of D. casta and M. privata late-instar larvae are distinctly different, with older D. casta larvae having distinct blades and a retinaculum, losing most evidence of teeth and ridges. In contrast, M. privata maintain distinct teeth through development, although the shape of the incisal teeth changes, with the definition of the points decreasing, and no retinaculum developing (Sutherland 1999).
The prevalence of pointed teeth on the cutting ridge of mandibles of early-instar larvae may reflect some of the difficulties associated with consuming plant material from the leaf surface (Dockter 1993; Hochuli 1994). The points probably increase the caterpillar’s ability to break through the leaf surface, because forces at the tip of the mandible are concentrated rather than spread along the blade of the cutting edge. Once the leaf surface has been fractured, it is relatively easy for small larvae to feed upon material beneath the leaf epidermis (Lucas et al. 1991). The retinaculum of D. casta may allow leaf tissue to be broken into smaller pieces and thus may allow more nutrients to be extracted from the leaf material, although Barbehenn (1992) found that digestibility was not correlated with bite size and handling time may be more important. Bernays (1991) suggested that tough leaves are efficiently handled by snipping scissor actions, whereas the softer, more flaccid leaves are more efficiently ingested by a tearing, crushing action. Large larvae also have larger heads containing more mandibular muscles, which are capable of producing greater force, which also influences the effectiveness of differing mandibular morphology, and ultimately feeding strategy (Bernays 1986; Dockter 1993).
Non-nutritive consequences of ontogeny
The feeding strategies of herbivores are intertwined with behavioural strategies that increase growth and development through thermoregulation (Casey 1993; Bryant et al. 2000), and the avoidance of natural enemies (Heinrich 1993). Pressures imposed by competition and predation may override those associated with feeding and digestion per se, mooting the significance of leaf properties as limits to herbivory.
For instance, many lepidopteran larvae use leaves to create shelters to lessen the impacts of predation, parasitism and dehydration (Gaston et al. 1991; Fitzgerald 1993). Shelters may be constructed by individuals as well as groups and vary substantially in complexity, ranging from a single rolled leaf, to multiple leaves tied together, through to elaborate tents and leaf shelters formed by gregarious species (Stamp & Wilkens 1993; Fitzgerald 1993). Shelter building is not limited to gregarious larvae, and may involve considerable use of mouthparts, such as in Netrocoryne repanda Felder and Felder (Hesperiidae), which snip leaves and use webbing to create shelters, building new ones with each instar (Braby 2000).
The influence of leaf characteristics on the ability of larvae to form leaf shelters has not been well studied (Mueller & Dearing 1994; Sutherland 1999). Leaf structure may restrict the ability of larvae to form a shelter or greatly increase the time necessary to complete a shelter (Damman 1987; Mueller & Dearing 1994). The properties of ideal leaves for shelter construction and feeding may vary substantially, forcing larvae to compromise and choose poorer quality leaves for one activity. This problem is expected to be more important in small larvae because they are less capable of manipulating inflexible leaves (Reavey 1993) and in small groups of larvae that communally form shelters (Damman 1987). The paw paw caterpillar, Omphalocera munroei Martin constructs a better leaf shelter from the older leaves of its host plant, despite young leaves being higher quality food for larvae (Damman 1987). Group size also influences the ability of these caterpillars to build shelters, with at least 20 first-instar larvae required to construct a shelter from older leaves. Conversely, Mueller and Dearing (1994) found that aquatic larvae of the pyralid moth Parapoynx rugosalis Moschler suffered greater predation when confined to constructing their protective cases from adult leaves of the water lily, Nymphaea ampla (Salisb.) instead of young leaves. In both cases, the physical structure of the leaf limits the ability of larvae to form a shelter and influences the level of predation they suffer.
The thermal balance of solitary and gregarious lepidopteran larvae is likely to differ and may thus have important ecological consequences (Bryant et al. 2000). Butterfly species with either solitary or gregarious larvae have two distinct thermal strategies. Exposed gregarious larvae regulate body temperatures, whereas solitary larvae are largely dependent on ambient temperatures (Bryant et al. 2000). Within-species observations show that group basking facilitates thermoregulation and that the elevated temperatures attained by caterpillars in groups allow them to rapidly exploit host plants (Stamp & Bowers 1990). The wealth of information on larval thermoregulation (synthesized by Heinrich 1996) supports the claim that it is an important non-nutritive factor influencing the ecology of insect herbivores; the aforementioned examples indicate that it may also impose strong indirect effects on foraging behaviour and levels of exposure to natural enemies.
INTEGRATING ONTOGENY INTO THE ECOLOGY OF INSECT HERBIVORY
Foraging behaviour theory is set within a framework of conflicting pressures with an emerging consensus among researchers to look beyond solely nutritionally based explanations of herbivory (Reavey 1993; Bernays 1998). In addressing the range of hypotheses that are generated from the aforementioned observations, it is clear that the complex suite of potential interactions among factors needs to be integrated into tests of how plant characteristics influence herbivory. However, the quest for generality must also recognize the fundamental significance of ontogenetic characters, primarily size, and incorporate a view of herbivory from the caterpillar’s perspective. Autecological nutritional and morphological studies are a common and worthwhile starting point, although the series of questions that needs to be addressed cannot be considered without generating many more regarding interactions with non-nutritive factors.
Is it possible to make generalizations about insect herbivory given the immense complexity of the systems that need to be considered? Starting with even a fundamental question such as ‘Does feeding behaviour change ontogenetically?’ generates multiple questions that must all be addressed. Is the change in feeding behaviour reflected in head and mandible morphology? What morphological characteristics influence the success of individuals? Are these related to the structural characteristics of leaves? How are these factors and changes related? What alternative explanations can be invoked and do these explanations obviate the need for morphological explanations? This search for alternative explanations leads one to consider the broader biotic and abiotic pressures on larvae in their natural environment, introducing additional trophic levels in the process.
The incorporation of non-nutritive explanations into insect herbivory adds many more layers of complexity to potential explanations. Do leaf attributes select for obligate gregariousness or is gregariousness a function of non-nutritive consequences of ontogeny, such as egg clustering, thermoregulation, defensive behaviour to avoid natural enemies or ease of shelter-building? How do ontogenetic characteristics affect interactions with predators and parasites? How are gregariousness and the tendency to solidarity in later instars influenced by competition among larvae?
Landsberg and Cork (1997) point out that it is over-simplistic to consider eucalypts as a single entity, identifying that conjecture of levels of herbivory in eucalypts treated as a single entity has led to ‘unprofitable controversies’. Similarly, it is often not meaningful to ascribe broad generalizations to the nutritional ecology of herbivorous taxa because insects act (functionally) as fundamentally different herbivores at different stages of their life cycle. Many studies are biased toward considering the biology of late-instar larvae when describing host use and choice (Slansky & Scriber 1985; Slansky 1993), a potentially misleading bias given the importance of early-instar mortality (Zalucki et al. 2001). This, and the ambiguity inherent in previous studies of the mechanical properties of plants (Sanson et al. 2001) reflects the difficulty in overcoming methodological and cultural mind-sets in previous studies of insect herbivory.
CONCLUSIONS
The description of patterns in insect–plant ecology too often fails to identify or investigate process. The scale of herbivory may explain many examples of ‘avoidance’, as temporal changes in the use of plants by herbivore assemblages (e.g. oaks; Feeny 1970) may be associated with a number of ontogenetic changes in larvae. Many herbivorous insects seem locked into feeding strategies as a consequence of scale and the morphological constraints that are imposed by scale.
The present paper illustrates several alternative hypotheses to conventional explanations of insect–plant interactions, showing that size and morphology can constrain the way in which caterpillars feed, independently of other factors such as plant chemistry and predation. Although certain plant chemicals clearly function as deterrents in some circumstances, as suggested by temporal differences in host use by insects, their function for these caterpillars may be a moot point because the insect may be unable to process the leaves and get to the chemicals that are within the leaf. Therefore, if an insect does not feed on a particular plant because it cannot adequately process it, it is more appropriate to say that morphological and ontogenetic constraints are restricting their host range rather than plant chemistry. Evidence to suggest otherwise must be provided before factors related to plant chemistry can be invoked.
The interrelations between ontogeny, size, morphology and behaviour highlight the complexity and depth of study required to understand the relationship between insects and their host plants. Given the many ways in which insect herbivores may overcome the challenges of consuming foliage, the need for species-specific and stage-specific research investigating ontogeny and host use by insect herbivores is critical.
Acknowledgements
I thank Martin Steinbauer for the kind invitation to present this paper at the first Symposium on Insect– Eucalypt Interactions held in Canberra. Critical comments from Peter Banks, David Britton and Tish Silberbauer improved early versions of this paper. I am also grateful to the past and present members of the Hochuli Laboratory, especially Michelle Sutherland and Kylie McClelland, for discussion of the ideas presented in this paper.
