Chapter 8 Ant Diversity and Function in Disturbed and Changing Habitats

Stacy M. Philpott, Ivette Perfecto, Inge Armbrecht, and Catherine L. Parr

8.1 Introduction ecosystem engineers (Decae¨ns et al. 2002; Folgarait Habitat transformation and disturbance are signifi- 1998). The impacts of habitat disturbance and trans- cant threats to biodiversity conservation and eco- formation for ants are widespread, yet they vary with system function. Disturbance is generally defined region and ecosystem. Ants can be very sensitive to as any event that removes biomass (Townsend and habitat transformation and disturbance, and for this Hildrew 1994), and is distinguished from habitat reason have been extensively used as indicator spe- transformation or stress, which reduces available cies (Hoffmann and Andersen 2003; see Box 8.1). resources or changes the microclimate or structure Because ants are colonial organisms, removal of in- of the habitat (Andersen 2000; Pickett and White dividuals (mortality) caused by habitat disturbance 1985). Habitat disturbance and transformation or transformation may not translate to extirpation of affect communities in many ways either by altering the colony from the habitat (Andersen 2000). This the balance of competitive interactions, often in may mean the responses of ants to disturbance may effect resetting the process of competitive exclu- differ from other terrestrial animals and plants that sion, or by clearing space for colonization of new may become locally extinct after disturbances. organisms. The degree to which habitat disturbance Disturbance effects on ant communities include loss and transformation affect animal communities in of diversity, changes in species composition, alter- general, and ants in particular, depends largely on ation of interspecific interactions, changes in trophic the frequency and intensity of disturbance, the per- interactions with ant-plants and honeydew-produc- manence with which habitats are transformed, and ing hemipterans, and modification of ant-provided the distance from which propagules travel to recol- ecosystem services such as seed dispersal, predation, onize affected habitats. and soil modification. Virtually all habitats are sub- Ant habitats of all kinds are modified by natural ject to some sort of disturbance, although the distur- disturbances, such as fire, forest gap formation, bance will obviously vary in origin (natural or hurricanes, and flooding, which vary in their extent, human-induced), in scale, and in magnitude. Many magnitude, and frequency. Furthermore, many insights to basic ecology are thus gained by investi- terrestrial ecosystems, especially in tropical regions, gating ecology in disturbed habitats. For example, have been altered by human activities including de- seminal work examining predator-caused distur- forestation, urbanization, agriculture, agricultural bance in intertidal zones has formed the basis for intensification, grazing, and mining. At the same the field of disturbance ecology (e.g. Paine 1996). time, ants themselves are also instigators of habitat Similarly, research in tropical forests affected by hur- modification via their roles as mound builders and ricanes and tree-fall gaps has shaped our knowledge

137 138 ANT ECOLOGY

Box 8.1 Using ants as indicators of ecosystem change Alan N. Andersen

The sensitivity of ant communities to environ- rangeland grazing, with mining having a neg- mental disturbance, combined with their great ligible impact despite its dominant economic functional importance and ease of sampling, contribution. makes them powerful monitoring and assess- ment tools in land management. The use of Reliability ants as indicators of ecosystem change is par- ticularly widespread in Australia, especially in The use of ants as bioindicators is founded the context of mine site rehabilitation, but also on the assumption that the extent of ant for a variety of other land-use situations such as community change reflects broader ecosys- off-site mining impacts, forest management, tem change. How valid is this assumption? and pastoralism (Andersen and Majer 2004). The few relevant studies all suggest that ants Ants have been strongly championed as indi- do indeed reflect broader ecological change, cators in Australia, but could equally be used as rather than providing idiosyncratic responses such in most other parts of the world, wherever that are as uninformative as they are un- they are diverse and abundant. representative. For example, a range of mine site rehabilitation studies show that patterns of ant recolonization reflect those of other What to measure invertebrate groups and of key ecosystem Ant monitoring programmes typically focus on processes such as nutrient cycling (Andersen changes in species composition rather than di- et al. 2004a). However, this is a ripe area for versity, as the latter can remain relatively con- further research, as it is important to un- stant in the face of major compositional change derstand what ecosystem components and and therefore be uninformative (Kaspari and processes ants are representing, and what Majer 2000). Moreover, species diversity can re- they are not. For example, many vertebrate spond in unpredictable ways to disturbance, or groups are likely to respond to different in ways that are highly situation-specific, such habitat variables, and at different spatio- that changes can be difficult to interpret. If spe- temporal scales, than those driving ant (and cies-level responses to disturbance are well un- other invertebrate) community dynamics. derstood, then monitoring can focus on changes in the abundance of individual species. For ex- Feasibility ample, several Australian ant species have been shown to increase or decrease consistently in Concerns are often expressed by land man- abundance in relation to disturbance (Andersen agers that invertebrate monitoring is too dif- et al. 2004a). However, in most cases reliable ficult and too time-consuming to be cost species-level information is unavailable so broad effective. However, a critical evaluation of species compositional change is monitored. different monitoring options suggests that Majer and Beeston (1996) have developed a this is not the case. Majer et al. (2007) have protocol for scaling-up local information on compared the performance of various inver- the effects of disturbance on ant species com- tebrate groups with plants and vertebrates as position to address regional scale effects of indicators of restoration success at Western different land uses. The effects on ant species Australian mine sites. Assemblage composi- composition is multiplied by the proportional tion of a range of invertebrate groups, in- area affected for each land use, and the sum of cluding ants, all reflected trends in the these scores becomes a ‘biodiversity integrity’ composition of other groups to a greater ex- index for the region. On this basis, Majer and tent than did either plants or vertebrates. In- Beeston (1996) concluded that the land use vertebrates were much more efficient than causing most biodiversity loss in Western Aus- vertebrates in terms of information yield per tralia was intensive agriculture, followed by unit time in the field and laboratory. continues ANT BIODIVERSITY AND FUNCTION IN DISTURBED AND CHANGING HABITATS 139

Box 8.2 continued

Moreover, several studies have shown that The future ant sampling and processing can be greatly There is an ongoing need for further research simplified – by recording species presence or on ant responses to different disturbances in absence rather than abundance, or consid- different places, and on how broadly these ering only a subset of species, without losing responses represent general ecological change. indicator effectiveness (Andersen and Majer However, there is already a strong body of 2004). For example, presence–absence data knowledge, and the use of ants as bio-indica- for large species only, gave comparable re- tors in land management is limited more by a sults to comprehensive ant surveys in terms land management tradition of ignoring inver- of detecting off-site mining impacts at tebrates altogether. Land managers could Mount Isa in northwestern Queensland. In- profitably learn from their aquatic colleagues, deed, selected species can actually improve who for decades have been effectively using indicator performance, as has been shown invertebrates as bio-indicators of river health for small subsets of genera in terms of dis- (Hawkins et al. 2000). Ants are acting as criminating land condition in relation to environmental monitors in most terrestrial livestock grazing in western New South habitats – we just need to be asking them what Wales (Andersen and Majer 2004). is going on!

about community assembly (Hubbell et al. 1999; Van- community structure, mechanisms causing biodiver- dermeer et al. 2000). More recently, knowledge about sity loss with habitat transformation, and subsequent relationships between diversity and ecosystem func- implications for trophic interactions and ecosystem tion has advanced by studying predatory effects of services provided by ants in altered habitats (see birds across a range of agricultural management sys- Table 8.1). We also examine the role of ants as eco- tems (Van Bael et al. 2008). Specifically for ants, re- system engineers. In the concluding remarks, we search in agroforests set the groundwork for generalize what is known about the impacts of dis- studying the spatial ecology of ant communities turbance on ant communities. Throughout the chap- and mechanisms underlying pattern formation (e.g. ter, we provide information about management or ant mosaics) (Leston 1973; Majer 1976). Studies in a conservation recommendations (also see Chapter 4) range of agricultural and forest habitats have re- useful or necessary to restore ant communities to vealed the relative importance of competition and states present before drastic human-induced habitat environmental characteristics of habitats in assem- disturbance and transformation. Finally, we will bling ant communities. Furthermore, changes in re- present an agenda for future research that will ad- source availability in disturbed ecosystems have vance our understanding of this important field. provided important insight into factors that are essential in maintaining the diversity of tropical ants. Especially because of the prevalence of habitat 8.2 Agents of habitat disturbance disturbance and transformation in nearly all biomes and transformation and effects on and ecosystems, understanding whether and how ant communities disturbance alters ant behaviour, diversity, composi- 8.2.1 Fire tion, and subsequent changes in ecosystem services is critical. Fire is a frequent and widespread disturbance In this chapter, we summarize the effects of natural in many of the world’s major biomes, in- and anthropogenic disturbance on ant species and cluding savannas, grasslands, boreal forests and Table 8.1 Key findings relating habitat disturbance and effects on ant communities.

Type of disturbance/ transformation Effects on ant community Reference

Fire Resilience and resistance to fire Parr et al. (2004); Parr and Andersen (2008) Shifts in composition Andersen et al. (2006) Negative impact on arboreal and cryptic litter- Arnan et al. (2006) dwelling species Enhanced seed dispersal by ants Parr et al. (2007) Habitat type determines the extent of (burning) Barrow et al. (2007); Farji-Brener et al. effect on ants (2002); Ratchford et al. (2005) Increased abundance for particular functional Hoffmann and Andersen (2003) groups (Opportunists) Flooding and Reduction in species richness Ballinger et al. (2007); Milford (1999) inundation Development of survival and behavioural Ballinger et al. (2007); Klein et al. (1993); mechanisms Lude et al. (1999); Maschwitz and Moog (2000); Nielsen (1997); Nielsen et al. (2006) Swimming or surfing ants Adis (1982); Jaffe´ (1993) Forest tree-fall and No detectable response Feener and Schupp (1998) gap creation Context of gap (primary or secondary forest) affecting impact of herbivory by leaf-cutting Pen˜ aloza and Farji-Brener (2003) ants Deforestation and Decreases in ant species richness Dunn (2004); Majer et al. (1997, and logging references therein) Changes in ant composition Nakamura et al. (2007); Palladini et al. (2007); Vasconcelos (1999a) Transition from stochastic to deterministic Floren et al. (2001) processes of community assembly in disturbed habitats Increased vulnerability to ant invasions Suarez et al. (1998); Vasconcelos et al. (2000) Selective logging favours species richness Azevedo-Ramos et al. (2006) Agricultural Reduction of species richness Perfecto et al. (2007) intensification Favours herbivory by leaf-cutting ants Blanton and Ewel (1985) Increased foraging activity by fire ants Nestel and Dickschen (1990) (Solenopsis geminata) Possible functional disruption on soil food web Amador and Gorres (2007) interactions for maintenance of soil fertility and structure Grazing Relative proportions of functional groups change Bestelmeyer and Wiens (1996); Hoffmann according to grazing practices (2000) No response to intensive pulses of cattle grazing Read and Andersen (2000) Increased arboreal ant species richness with trees Majer and Beeston (1996) Fodder banks increase predatory ant richness Ramı´rez et al. (2007) Mining Increased richness with rehabilitation time Andersen et al. (2003); Majer et al. (1984) Reduced richness with sulphur deposits Hoffmann (2000) Urbanization Richness decline in natural habitats inside urban Lessard and Buddle (2005); Pacheco and areas with respect to rural areas Vasconcelos (2007); Yamaguchi (2004) No change Gibb and Hochuli (2003) Opportunistic or non-native species persisting in Carpintero et al. (2003); Gibb and Hochuli urban sites, compositional changes (2003); Holway and Suarez (2006) ANT BIODIVERSITY AND FUNCTION IN DISTURBED AND CHANGING HABITATS 141 sclerophyllous vegetation (Pyne 1997). Studies on a the effect of burning on biota have been conducted Stress: 0.18 in a correspondingly broad range of habitats. While U-Wet there are several aspects of these fires that can be studied (e.g. season, frequency, intensity, size, B-Wet type), most fire and ant studies have focused on either a comparison of burnt versus unburnt areas, U-Dry or on the effect of applying repeated fires. Here we focus on areas that burn naturally. Overall, ant assemblages exhibit striking resil- B-Dry ience and resistance to burning (Barrow et al. 2007; Parr et al. 2004) with fires generally having little b effect on ant abundance, species richness, assem- Stress: 0.15 blage composition, or structure. Even where ant U-Wet abundance was found to decrease immediately post-fire (ten days post-fire, Andersen and Yen 1985), in the longer term there was little obvious B-Wet negative effect. The degree of response of ant assemblages to U-Dry burning has been linked to habitat type (Barrow et al. 2007; Farji-Brener et al. 2002; Ratchford et al. B-Dry 2005), with variation in resilience linked to the arid- ity of a site (Arnan et al. 2006), and the degree to which the habitat is modified post-fire (Barrow et al. Figure 8.1 Multi-dimensional scaling ordination for burnt 2007; Parr et al. 2004; see Figure 8.1). Exceptions to and unburnt sites for (a) spinifex and (b) sandplain habitat at Purnululu National Park in Western Australia in both this extreme resilience occur either in systems that the wet and dry season sampling periods (U = unburnt, burn extremely infrequently (e.g. low flammability B = burnt, wet = wet season, dry = dry season). Each point vegetation such as rainforest), or those that undergo of the ordination represents a sampling grid in the a major shift in vegetation composition and struc- respective habitats. In the spinifex habitat, which ture post-fire (e.g. in the Mediterranean, Pinus nigra undergoes pronounced structural change with burning, there is a significant difference in ant assemblage forest is converted to shrubland, Rodrigo and Re- composition between burnt and unburnt sites. A more tana 2006). Composition of ant assemblages may limited difference in vegetation structure in the sandplain also sometimes differ in burn/no burn comparisons habitat results in only a slight difference in ant but these instances are usually where repeated, assemblages. (Reproduced with permission, from Barrow rather than once-off fires have been applied (Parr et al. 2007). et al. 2004), or at the other extreme, where there has been long-term fire exclusion in a highly flammable vegetation structure or insolation level affects ant environment. For example, a shift in composition assemblages. toward more rainforest-associated ant species was Although ants in some habitats exhibit remarkable reported where fire was excluded from tropical resilience to fire, there can be striking changes in the savanna in northern Australia (Andersen et al. abundance of different functional groups. For exam- 2006). In terms of habitat modification, clearly the ple, burning increases the abundance of some partic- consumption of dead wood or litter by fire is likely ular functional groups (Hoffmann and Andersen to negatively affect arboreal or cryptic litter-dwell- 2003) (see Box 8.2 for an introduction to functional ing species (Arnan et al. 2006). Although changes in groups). Elsewhere, processes such as myrmecoch- ant assemblages due to fires have primarily been ory (seed dispersal by ants, see Chapter 6) are affect- attributed to changes in vegetation structure, there ed by burning; savanna fires in northern Australia are few studies that directly test how a change in can enhance rates of seed removal and significantly 142 ANT ECOLOGY

Box 8.2 Functional groups in ant community ecology Alan N. Andersen Ecologists often classify species into functional modified and extended for continental and groups as a way of reducing ecological com- intercontinental analyses of biogeographical plexity and allowing for comparative analyses patterns of ant community structure and their of ecological systems with little or no species responses to disturbance (Andersen 1995, overlap. No particular functional group 1997, 2000, 2003; Hoffmann and Andersen scheme can serve all purposes, and groups 2003). based on different ‘functions’ will have differ- The most important functional groups in this ent applications. One approach to functional global scheme are Dominant Dolichoderinae, groups in ant community ecology is to classify Generalized Myrmicinae, and Opportunists, species according to niche dimensions such as because they respectively represent the three diet, nest location, and time of foraging. This is primary ecological types in relation to stress particularly useful for detailed analyses of and disturbance from a global perspective – particular communities. However, such dominant, subdominant, and ruderal (Grime schemes tend to be purely descriptive, and 1979). They strongly parallel the three primary often reveal little insight into fundamental plant life-forms used in vegetation analysis and ecological, biogeographical, and evolutionary classification: trees, shrubs, and grasses (An- processes driving community structure. A dersen 1995). See Figure 8.2.1 for representa- commonly used scheme that aims to redress tives of some functional groups. this is based on global-scale responses of ants From a global perspective, dominant species to environmental stress (factors affecting are those at the top of dominance hierarchies productivity) and disturbance (factors in the most productive environments. For ants, removing biomass), operating at the genus or maximum productivity occurs where the sun’s species-group level (Table 8.2.1). These thermal energy in hot, open and structurally groups originated from P. J. M. Greenslade’s simple environments combine with the meta- pioneering studies in arid Australia bolic energy of carbohydrates from plant exu- (Greenslade 1978), and have since been dates, especially honeydew (Andersen 1995;

Table 8.2.1 Ant functional groups based on global-scale responses to environmental stress and disturbance, and their major constituent taxa.

Functional group Major taxa

Dominant Anonychomyrma, Azteca, Dolichoderus, Dorymyrmex (bicolor group), Forelius, Dolichoderinae Iridomyrmex, Liometopum, Linepithema, Papyrius, Tapinoma (nigerrimum group) Generalized Crematogaster, Monomorium (part), Pheidole Myrmicinae Opportunists Aphaenogaster, Dorymyrmex (insanus group), Ectatomma, Formica (fusca group), Lepisiota, Myrmica, Paratrechina, Rhytidoponera, Tapinoma, Technomyrmex, Tetramorium Subordinate Camponotus, Opisthopsis, Polyrhachis Camponotini Hot-Climate Cataglyphis, Melophorus, Meranoplus, Messor, Monomorium (part), Myrmecocystus, Specialists Ocymyrmex, Pogonomyrmex Cold-Climate Anoplolepis (part), Formica (rufa and exsecta groups), Lasius, Lasiophanes, Temnothorax, Specialists Monomorium (part), Notoncus, Prolasius, Stenamma, Stigmacros Tropical-Climate Many taxa characteristic of tropical rain forest, including Dorylinae, Ecitoninae, and Attini; it Specialists also includes the fire ants (Solenopsis subgenus Solenopsis), and the behaviourally dominant genus Oecophylla Cryptic Species Many genera of small-sized and small-eyed myrmicines and ponerines Specialist Predators Anochetus, Cerapachys, Leptogenys, Myrmecia, Odontomachus (part), Pachycondyla

continues ANT BIODIVERSITY AND FUNCTION IN DISTURBED AND CHANGING HABITATS 143

ab

cd

Figure 8.2.1 (a) Iridomyrmex sp. from Australia is behaviourally dominant and typifies the Dominant Dolichoderinae group; (b) Species of Camponotus belong to the Subordinate Camponotini; (c) Specialist Predators include the genus Pachycondyla;(d) Generalized Myrmicinae, such as species of Monomorium are classified as subdominant according to the Functional Group scheme devised by Andersen (1995). (Photos: Alex Wild) Davidson 1997; Kaspari 2000; Kaspari and temperate regions elsewhere in the world. In Weiser 1999). This coupling of thermal and these regions, behavioural dominance has metabolic energy powers the large colony sizes evolved in Cold-Climate Specialist formicines, and high rates of activity that are characteristic such as Formica (rufa and exsecta groups) of behaviourally dominant species, and is best throughout the Holarctic, and Anoplolepis developed in the canopies of lowland tropical (custodiens group) in southern Africa. No such rainforest (Blu¨ thgen et al. 2000; Davidson et al. behaviourally dominant, cold-adapted formi- 2003; Tobin 1994), and on the ground in arid cine occurs in Australia. Behavioural domi- Australia (Andersen 2003). The behaviourally nance also occurs in the arboreal Tropical- dominant ants in these habitats are typically Climate Specialist Oecophylla, which occurs dolichoderines, notably species of Iridomyrmex throughout the Old World Tropics; it has no in the Australian arid zone (Greenslade 1976), parallel in the New World Tropics. species of Anonychomyrma, Philidris, and Do- Subdominant Generalized Myrmicinae have lichoderus in the Old World tropics (Huxley a much broader distribution in relation to en- 1982; where Philidris is referred to as the cor- vironmental stress and disturbance than do datus group of Iridomyrmex), and species of Dominant Dolichoderinae, and tend to pre- Azteca, Dolichoderus, Linepithema, Liometo- dominate in moderately, rather than highly, pum, Dorymyrmex, and Forelius in the New productive environments for ants (Andersen World (Andersen 1995; Davidson 1997; Tobin 1995). They are often extremely successful at 1994). It should, however, be noted that not all recruiting to and defending food resources, dolichoderines are behaviourally dominant, but compared with Dominant Dolichoderinae with many being behaviourally submissive have lower rates of activity, smaller colony Opportunists (see Table 8.2.1). sizes, and smaller foraging territories, and tend It is noteworthy that although behaviourally to be less aggressive. Generalized Myrmicinae dominant dolichoderines occur in all climatic are often the most abundant ants in warm en- zones of Australia, they are absent in cool- vironments where Dominant Dolichoderinae continues 144 ANT ECOLOGY

Box 8.2 continued

are absent, such as the tropical savannas of ality inevitably comes at the cost of precision. southern Africa (Parr et al. 2004) and Brazil The functional group scheme can provide a (Campos et al., unpublished data), and in leaf useful framework for analysing the broad litter of lowland tropical rainforest throughout structure of particular communities, but it is the world (Ward 2000). Opportunists are un- not designed for studies of community dy- specialized, behaviourally submissive species, namics at local scales that require a detailed often with wide habitat distributions. They understanding of the ecology of individual predominate only at sites where stress or dis- species. The scheme is particularly limited for turbance severely limits ant productivity and local-scale analysis in regions where relatively diversity, and consequently where behavioural few functional groups are represented, such as dominance is low. in cool-temperate regions of the northern The functional group scheme described here hemisphere, or where one functional group is is designed for biogeographical-scale analyses particularly diverse both taxonomically and of ant community structure and dynamics. It is ecologically, such as Tropical-Climate Specia- also useful for other large-scale studies involv- lists in lowland tropical rainforest. A functional ing ants, such as analysing ecological attri- approach is still highly useful in such situa- butes of pest ant species (McGlynn 1999b) and tions, but requires functional groups that are the ant partners of lycaenid butterflies (East- designed for the specific purpose (e.g. Delabie wood and Fraser 1999). However, such gener- et al. 2000).

increase the distance of seed dispersal, more than by the current to dry ground. Where more predict- doubling it for some ant species (Parr et al. 2007). able seasonal flooding occurs, species richness tends to be lower (Majer and Delabie 1994), and soil-, litter-, and shrub-associated species are most 8.2.2 Flooding and inundation negatively affected. Floods cause major habitat changes by scouring Ant species occurring in frequently wet areas flood plains and removing vegetation, soil, and have extraordinary ways of dealing with flood- litter. These catastrophic, large-scale, stochastic ing. For example, mangroves are regularly inun- flood events ‘reset’ areas of flood plain, and dra- dated with sea water. Although most ants in the matically alter habitat availability and quality. For wettest mangrove areas in northern Australia are example, immediately after floodwaters recede, ant arboreal (e.g. Crematogaster), Polyrhachis sokolova species richness and abundance can be reduced nests in the mud and relies on trapping air in (Ballinger et al. 2007; Milford 1999). Over longer nest galleries to survive inundation periods lon- timescales (several years), duration of inundation ger than 3 h (Nielsen 1997). Extreme physiologi- is an important factor. Richness and abundance are cal adaptation enables Camponotus anderseni, lower in areas where inundation duration has been which nests in the twigs of mangrove trees, to longer (Ballinger et al. 2007). Floodplain species survive hours of inundation. Because the head of tend to be opportunists that can recolonize dis- soldiers plugs the nest entrance when the tide turbed areas quickly, with some species specifically comesin,gasexchangeisprevented,andin colonizing young gravel bars devoid of vegetation response, the ants partly switch to anaerobic (e.g. Formica selysi in braided rivers in the Alps, respiration (Nielsen et al. 2006). Important to Lude et al. 1999). Survival mechanisms include point out is that this type of flooding is a consis- evacuating nests to higher ground or in trees tent disturbance that differs from unexpected or (Adis et al. 2001; Ballinger et al. 2007; Lude et al. stochastic events and may elicit physiological 1999), and forming rafts (comprising the queen, adaptations on evolutionary, rather than ecologi- several dozen workers, and brood) that are carried cal timescales. ANT BIODIVERSITY AND FUNCTION IN DISTURBED AND CHANGING HABITATS 145

At a smaller scale, heavy rain can cause flood- Hurricanes may also affect ant assemblages. For ing of nests. Ants exhibit a range of responses to example, Morrison (2002a) examined ant commu- this threat including plugging nest entrances nities on 17 Bahamian islands before and after a with their heads, and removing water that has hurricane that caused significant damage to vegeta- leaked into the nest by ingesting it, and either tion and soils. He found no ant species that went regurgitating (Klein et al.1993)orexcretinga locally extinct as a result of the hurricane, but ob- droplet outside the nest (Maschwitz and Moog served substantial decreases in overall ant abun- 2000). Some ant species are able to continue for- dance and changes in the composition of species aging when their habitat is flooded. For example, visiting baits (Morrison 2002a). Hurricanes may leaf-cutting ants reportedly ‘walk’ on water also alter ant–plant mutualisms over landscape when foraging during flooded periods (Adis scales. During the five years following a hurricane 1982). In the intertidal zone, P. sokolova has in the Nicaragua, proportions of Cecropia spp. trees been described as ‘swimming’ as the tide comes occupied by Azteca spp. were greatly reduced, like- in (see Box 10.1) and even a surfing-like beha- ly leading to high mortality of this ant associate viour has been observed in ants that forage in (Ferguson et al. 1995). the intertidal zone whereby the ants adopt a ‘nymphal’ position enabling them to ride the 8.2.4 Logging wave until the sea water is absorbed into the sand, and then they walk off (Jaffe´ 1993). Logging is a globally important threat to biodiver- sity. However, the degree of tree removal varies from complete extraction (deforestation) to selec- 8.2.3 Forest tree-fall gap creation tive logging, where only certain species of trees and hurricanes are targeted for removal. Ant species richness may In tropical forests, gap creation from falling trees is decrease in logged areas (King et al. 1998), increase an important disturbance shaping ecological and in recently logged stands (Palladini et al. 2007), or evolutionary dynamics. Although gap creation has experience no change with selective logging (e.g. been shown to influence plant and bird dynamics, Kemel et al. 2001; Vasconcelos et al. 2000). Where Feener and Schupp (1998), working in Panama, deforestation occurs, resulting in varying levels of found little to suggest that ant assemblages respond habitat fragmentation, there are usually changes in significantly to tree-fall gap formation. They found ant communities, namely, changes in species com- no differences in species richness, abundance, com- position (e.g. Nakamura et al. 2007; Palladini et al. position, or rates of resource discovery between 2007; Vasconcelos 1999a,b) and demography (e.g. gaps and the surrounding forest. Likewise, Shure Kemel et al. 2001) of the resident species. In boreal and Phillips (1991) found no differences in ant forests, colony abundance of red wood ants (Formi- abundance in recently created forest gaps differing ca rufa group) declined drastically in deforested in size from 0.016–10 ha. Instead, larger-scale pro- areas that were also ploughed, but in areas where cesses may be of greater importance, and seasonali- some trees were left standing, colony abundance ty and habitat difference may reduce the impact of was similar to mature forests (Domisch et al. 2005). gaps on ants. Herbivory caused by leaf-cutting ants Ant assemblage composition in logged forests may (mainly Atta cephalotes) varies with the age of the become similar to that of primary forest but it can surrounding forest matrix, with foliar damage in require several decades, or even centuries, of gaps adjacent to old-growth forest a magnitude natural regeneration (Floren et al. 2001; Palladini higher than that in gaps adjacent to secondary for- et al. 2007). Changes in forest composition affect est (Pen˜aloza and Farji-Brener 2003). Consequently, ant dynamics, possibly causing a transition disturbance from tree-fall and the formation of gaps from stochastic to deterministic – driven processes may play a vital role in providing islands of palat- of community assembly in disturbed habitats (Flo- able resources in an ocean of less palatable forest. ren et al. 2001). Such demographic disruptions 146 ANT ECOLOGY caused by deforestation may increase vulnerability ing tropical, temperate, and boreal forests have re- of native ant communities to ant invasions ported changes in species composition with (Suarez et al. 1998; Vasconcelos et al. 2000). In addi- fragmentation, especially due to edge effects (Car- tion, deforestation may be accompanied by fire, valho and Vasconcelos 1999; Debuse et al. which further severely negatively affects rainforest 2007; Suarez et al. 1998; Vasconcelos et al. 2001). ant species richness within the burned area For example, in central Amazonia, Carvalho and (MacKay et al. 1991) since most species are not Vasconcelos (1999) reported large changes in ant adapted to fire. While deforestation (and concomi- species composition in edges (up to 200 m) and tant conversion to agriculture) usually leads to de- forest fragment interiors, but saw no differences in creases in species richness, selective logging has a species richness. In rainforest areas this edge effect less drastic effect on ant species richness (Dunn has been attributed primarily to increases in leaf 2004). Further, practices such as reduced-impact litter in the forest edge as compared to forest interi- logging may have less of an adverse effect on or, but microclimatic and vegetation changes could ant species richness and composition (Azevedo- also be implicated (Carvalho and Vasconcelos 1999; Ramos et al. 2006), likely because the reduced im- Perfecto and Vandermeer 1996). The diversity of pact practices maintain a forest structure similar ant and hemipteran mutualists also increases in to an unlogged forest. edge habitats, presumably due to higher plant pro- ductivity towards forest edges (Dejean and Giber- nau 2000). 8.2.5 Fragmentation and edge effects One of the most consistently reported effects of It is difficult to distinguish between effects due to fragmentation is the increase of non-native, inva- habitat loss and effects of fragmentation because sive, or aggressive large colony weedy or invasive they often go hand in hand (Debuse et al. 2007). species (Bru¨ hl et al. 2003; Dejean and Gibernau Perhaps the largest and longest-running forest frag- 2000; Lessard and Buddle 2005; Ness 2004; Suarez mentation experiment is the Biological Dynamics et al. 1998). This has detrimental effects on the Forest Fragment Project (BDFFP) initiated in 1979 native ant fauna, decreases fragment habitability in the Amazon basin. Results of 20 years of studies for other ground-dwelling arthropods, and may in this project reveal that fragmentation effects directly or indirectly affect plants and their asso- are diverse and responses of different species and ciated arthropods (see Chapter 15). taxonomic groups are highly individualistic (Laur- One of the most interesting emerging results on ence et al. 2002). Studies of fragmentation have the effects of fragmentation on ant communities is shown variable effects on ants with regard to that the quality of the matrix surrounding forest species diversity and composition. However, most fragments is important, and that matrices that are studies report a decline in species richness and nest more similar to the forest structure will better pro- density within fragments (Bru¨hl et al. 2003; Carvalho mote inter-fragment connectivity (Belshaw and Bol- and Vasconcelos 1999; Vasconcelos 1999b), as well ton 1993; Byrne 1994; Vasconcelos 1999a). For as a higher number of non-native, invasive, or tramp example, immigrants from external source popula- species in fragments as compared to continuous for- tions may help maintain local populations of twig- ests (Suarez et al. 1998). At the landscape level, total nesting ants in forest fragments. Thus, ant popula- abundance of ants tends to increase due to the avail- tions in isolated fragments (i.e. surrounded by pas- ability of young successional areas (Vasconcelos tures) may be more extinction-prone (Byrne 1994). et al. 2001). Carvalho and Vasconcelos (1999) proposed that Studies show no consistent effects of fragment fragmentation effects for ants likely diminish with size on species richness but edges tend to have forest re-growth in pastures, because many ant spe- higher species richness than forest interiors (Dejean cies use these habitats (Belshaw and Bolton 1993). and Gibernau 2000; Majer et al. 1997; Vasconcelos This is also true for coffee agroforestry systems as et al. 2001, but see Golden and Crist 2000). Likewise, forest ant diversity is better maintained in high- many studies across a range of ecosystems includ- quality matrices resembling natural vegetation ANT BIODIVERSITY AND FUNCTION IN DISTURBED AND CHANGING HABITATS 147

(Perfecto and Vandermeer 2002). These results agricultural lands offer habitat for a relatively low link well to disturbance ecology theory in general number of ant species (7), but where the landscape in that distance to sources of propagules to re-colo- includes some meadows, fallow lands, and edge nize disturbed or transformed habitats is critical habitats species richness increases to 19 (70% of (e.g. MacArthur and Wilson 1967). Given that the species known from the area) (Dauber and Wol- there is local species extinction, even in large frag- ters 2004). Thus, maintaining habitat heterogeneity ments, the results from the BDFFP and other stud- in the landscape may be very important for main- ies on forest fragmentation strongly suggest a need taining ant diversity in disturbed agricultural land- to transform highly degraded agricultural matrices scapes. to those of a style that favour migration among In the tropics, differences in ant diversity have patches. been used to assess the consequences of agricultur- al intensification in coffee and cacao crops, specifi- cally – mainly involving different levels and 8.2.6 Agricultural intensification varieties of shade trees (Perfecto et al. 2007). Although some primary forest leaf litter ant species Thousands of hectares of traditionally shaded may survive in agricultural landscapes, such as agroecosystems in the tropics have been trans- cocoa plantations (e.g. Belshaw and Bolton 1993), formed into plantations with little or no shade permanent transformation of forests into agricul- (e.g. sun coffee). Intensification of coffee plantations ture dramatically reduces ant species richness (e.g. significantly reduces the associated biodiversity; Majer et al. 1997), as does agricultural intensifica- for instance, 18 of the 22 studies on ants examined tion. Intensification is generally associated with by Perfecto et al. (2007), showed that ant diversity crop specialization, increasing mechanization, and declined with agricultural intensification. This generalized use of agrochemicals and other exter- trend also applies for most studies including those nal inputs in the crop field. Yet because agricultural on plants, arthropods, and vertebrates. Some of the landscapes form the matrix surrounding forest lost ant fauna may be important biological control fragments, understanding the impacts of agricul- agents (Perfecto et al. 2007) and might positively tural intensification on biodiversity is valuable for affect soil fertility and quality (Amador and Gorres conservation purposes. There is a growing aware- 2007). Because not only the number of ant species ness that agroecosystems should be a priority in the decreases with the removal of shade trees, but also biological conservation agenda because some the abundance, case-specific analyses are needed in agroecosystems are repositories of high levels of order to evaluate the impacts of such changes in ant biodiversity including ants (Perfecto et al. 2007). assemblages. For example, Solenopsis geminata,a Ants are a robust group as ecological indicators, voracious predator of other insects, is extremely and constitute a rare example of the adoption of common in sun coffee plantations of Mexico, but invertebrates as indicators of land management the same species is a seed predator in some open (Andersen and Majer 2004; see Box 8.1). agroecosystems (Nestel and Dickschen 1990). Fur- In temperate regions, ant richness and abun- ther, open agroecosystems may also favour eco- dance are strongly affected by agriculture, and ef- nomically detrimental ants such as Atta cephalotes, fects may vary depending on common agricultural which cut 3.5 times more leaf tissue in a cassava practices and landscape components. For example, monoculture and in plots of non-indigenous plant in a study in Virginia and North Carolina, Peck et al. species than in diverse successional plots (Blanton (1998) found that ant species richness and colony and Ewel 1985). density for most species were lower in more dis- Even though ants have been increasingly used as turbed crop fields than in field margins, in areas the focus group in many studies regarding chang- practicing conservation tillage, and in areas where ing agroecosystems, natural systems, rehabilitation, fewer insecticides were applied. In Germany, active and other land management systems, there is a real 148 ANT ECOLOGY

ant species richness, and particularly the richness of 40 October April soil dwelling ants does not change with intensive 35 pulses of cattle grazing (Read and Andersen 2000). Thus, there are no clear trends about how grazing 30 affects ant richness (see Figure 8.2). Ranching and grazing practices may be improved, from a biodi- 25 versity standpoint, using several techniques. In Bra-

Species richness zil, arboreal ant species richness in isolated trees 20 embedded in tree pastures (and especially in large trees with epiphytes) increased with proximity to 15 forest patches (Majer and Delabie 1999). Converting 0.08 0.5 1.4 310intensive pasture lands to silvopastoral systems Distance from water point (km) (pastures with trees) by planting a diverse selection of trees and shrubs and thereby increasing canopy Figure 8.2 Species richness at varying distances from a waterpoint at Kidman Springs cattle station in northern cover might increase predatory ant richness Australia. The increasing distance from water represents (Ramı´rez et al. 2007). As practiced in Colombia, decreasing grazing pressure. Pitfall sampling was silvopastoral systems include frequent pruning to conducted in October and April. (Reproduced with generate fodder banks for cattle. Such fodder banks permission, from Hoffmann 2000). are extremely labour intensive, but involve very high plant biomass production in short periods of time (a few months). The drastic changes in vegeta- need to establish reliable sampling protocols tion associated with cyclic foliage pruning and re- for using ants in conservation monitoring (see growth alter the physical–physiological conditions Chapter 4). Ants provide invaluable information of the habitat and may hasten colonization or dis- about constantly disturbed habitats such as agroe- placement processes in ant communities (Ramı´rez cosystems in intensification or rehabilitation pro- et al. 2007). Most of what is known about the im- cesses, in a relatively short time and for low cost pacts of heavy grazing on ant communities is from (Underwood and Fisher 2006). tropical systems, although some temperate studies have been done. For example, Dauber et al. (2006a) 8.2.7 Grazing investigated ant richness and composition in regen- erating grasslands of varying size and condition A large fraction of anthropogenically modified used for grazing for centuries. They found that landscapes is designated for cattle pasture. There remnant grassland size did not affect ant richness, is an increasing concern about intense and constant and that smaller patches did not contain subsets of disturbance associated with unsustainable manage- ant species, but that habitat condition, namely tree ment of pasture lands. Ant richness is dramatically and vegetation cover, had strong impacts on differ- higher in tropical (or subtropical) forests compared ences in species composition. with intensively grazed neighbouring grasslands (Quiroz-Robledo and Valenzuela-Gonzalez 1995). 8.2.8 Mining Increases in grazing intensity may also result in declines of ant species richness, especially of litter- Mining represents perhaps the most extreme form inhabiting cryptic species and specialized preda- of habitat disturbance and transformation, result- tors (Bestelmeyer and Wiens 1996), and strong ing essentially in complete habitat loss. Ants have changes in species composition, although the rela- been widely used as an indicator group for mine tive proportions of different functional groups ap- rehabilitation work (see Box 8.1). Studies on mine pear somewhat resilient to grazing pressure site rehabilitation aim to determine how natural (Hoffmann 2000; Rivera and Armbrecht, unpub- undisturbed habitat and its complete associated lished data). However, in arid areas of Australia, ant fauna can be restored. Typically these studies ANT BIODIVERSITY AND FUNCTION IN DISTURBED AND CHANGING HABITATS 149

ab

c d

Figure 8.3 Mine site rehabilitation is a process that takes many years with little guarantee that the habitat will return to its exact former state. These photos from Nhulunbuy, northern Australia, illustrate how sites of different ages since rehabilitation differ in vegetation; this change in habitat strongly influences the ants: (a) un-mined reference site, (b) 2 year old rehabilitated site, (c) 5 year old rehabilitated site, (d) 24 year old rehabilitated site. (Photos: Benjamin D. Hoffmann)

compare a range of rehabilitation sites varying in a monoculture of either indigenous or introduced age with undisturbed reference sites. An important plant species tend to be less successful than those principle to take into account is that proximity to a with mixed vegetation. Importantly, although spe- source of colonizing species influences recovery cies richness increases with rehabilitation age, with from severe disturbances. Species richness is gener- some rehabilitation sites having the same species ally positively associated with time since rehabili- richness as undisturbed reference sites, the compo- tation (Andersen et al. 2003; Majer et al. 1984) and sition of these species can differ significantly. In may also increase with increases in habitat hetero- some systems the recreation of an intact pre-mining geneity and cover of litter and vegetation (Majer assemblage can take a substantial period of time et al. 1984). In this regard, rehabilitation sites with (e.g. >20 years in Mediterranean woodlands; 150 ANT ECOLOGY

Ottonetti et al. 2006; see Figure 8.3), and in cases in urban habitats (Friedrich and Philpott 2009). where the ecological community tends along a dif- Increased soil temperature and decreased soil ferent trajectory (e.g. due to stochastic events), com- moisture may enhance the establishment potential plete re-creation may be impossible. Mining can of some invasive species and reduce the abilities therefore leave a lasting impression on local biota of some native species to persist (Yamaguchi 2004). including ant assemblages. Finally, pollution caused In contrast, other invasive species such as the Ar- by mining can also negatively affect ant commu- gentine ant (Linepithema humile) thrive with higher nities; dry sulphur deposits from mining emissions soil moisture, allowing them to displace native spe- significantly reduce ant richness and abundance, and cies in irrigated or watered urban areas (Holway dramatically alter assemblage composition (Hoff- and Suarez 2006). Dispersal limitation may be im- mann et al. 2000). portant in community assembly in urban areas as founding queens may not arrive in urban centres from source populations (Pacheco and Vasconcelos 8.2.9 Urbanization 2007). Those ants that do persist in urban habitats Urbanization is a driving force behind habitat de- tend to be generalist and opportunistic species, struction, and has dramatic impacts on ant richness competitive dominants, and ants with large, ag- and composition. Ecological studies of urban ants gressive colonies (Carpintero et al. 2003). Addition- generally focus on investigating changes in species ally, factors that cause losses of some native species richness and species composition in different urban may facilitate invasion of non-native tramp species habitat types, urban habitat fragments of different in urban areas (Holway and Suarez 2006; see Chap- size or age, or along urban to rural gradients (Gibb ter 14). Finally, some urban areas are dominated by and Hochuli 2003; Lessard and Buddle 2005; Pacheco smaller-bodied ants (Holway and Suarez 2006) and and Vasconcelos 2007; Yamaguchi 2004). Ant species by phytophagous rather than predatory species richness sometimes declines with reduced size and (Gibb and Hochuli 2003) indicating that urbaniza- increased age of habitat fragments embedded in tion may alter ant species composition thereby urban areas (Yamaguchi 2004), along rural to urban affecting the ecological function of the ant commu- forest gradients (Lessard and Buddle 2005), or from nity in urban habitats. parks at urban edges to inner city squares (Pacheco and Vasconcelos 2007). In contrast, others have 8.3 Mechanisms causing change with found that ant richness does not decline with in- habitat disturbance creases in urban sprawl or with decreasing size of natural habitat fragments in urban areas (Gibb and The effects of habitat disturbance and transforma- Hochuli 2003). Nearly all studies, however, do find tion are brought about through changes in one or clear changes in ant species composition in urban several local-scale factors; these include behaviours habitats compared with nearby natural areas. (e.g. competitive interactions, predator avoidance, There is support for two main groups of factors parasitism, and colonization ability), soil type, and that influence species richness and composition of resource availability (Kaspari et al. 2003). Many of ants in urban areas: habitat and landscape factors the factors influencing ant assemblages are indirect, and competitive interactions. Disappearance of rather than direct, and are linked to habitat alter- necessary nesting resources or food items in ation. For example, disturbances alter habitat struc- urban habitats may affect specialist ant species. ture, which then influences microclimate. For example, due to a decline in the abundance of Removal of vegetation or growth of weedy plants rotting wood resources, generalist ants tend to following disturbance can have a significant effect dominate these nesting resources excluding dead on ant assemblages through changes to the micro- wood specialists from urban areas of Helsinki climate. Although little work has been done on (Vepsa¨la¨inen et al. 2008). Likewise, ants more fre- mechanisms behind disturbance impacts, differ- quently colonize artificial nesting resources most ences in microclimatic conditions including mois- similar to the most limiting cavity nest resources ture gradients, temperature regimes, and exposure ANT BIODIVERSITY AND FUNCTION IN DISTURBED AND CHANGING HABITATS 151 to wind affect ants physiologically and may poten- ity associated with agricultural transformation of tially reduce foraging times and their success. Fur- land or other disturbances (e.g. fire) causes changes thermore, seasonality and vertical stratification in ecological relationships among ant assemblages may also contribute to ant species assembly in Neo- (via parasitism by phorids, competitive interactions, tropical forests (Feener and Schupp 1998). Some refuge presence) and determines the coexistence of types of disturbance are inherent to ecosystem dy- the various ant populations (Wilkinson and Feener namics and constitute mechanisms of biodiversity 2007). When soil-nesting ant assemblages, especially generation and maintenance, while other long- those in temperate regions, reach high densities, they term, large-scale disturbances may challenge spe- may self-thin via competition, drivingtrends towards cies survival. Further, anthropogenic disturbance fewer and larger adult colonies. Likewise, top-down may change the relative balance of bottom-up ver- processes (predation in this case) may partially con- sus top-down population regulation. trol the litter ant community while bottom-up organi- Resources, such as nest sites, food, and refuges, zation (competition, resources) may be controlling among others, are important for ant community as- soil-nesting species in tropical forests (Kaspari 1996). sembly (see Chapter 7) and may become increasingly Dispersal limitation or slow recolonization may limited as natural habitats are disturbed or converted also contribute to changes in ant assemblages. At to productive agroecosystems. At the soil level, local large spatial scales, forest patches embedded in patches of litter naturally suffer continuous distur- disturbed landscapes (e.g. urban areas) may not bance, and evidence shows that plant succession oc- receive necessary colonists to maintain the commu- curring in these differently disturbed patches may nities found in forest patches (Pacheco and Vascon- lead to changes in species composition and may con- celos 2007). But even at small spatial scales, stitute a mechanism of diversity maintenance in trop- colonization and colony presence may highly influ- ical forests (Campos et al. 2007). For instance, leaf ence ant assemblages. Patterns of diversity are litter ant assemblages of several Neotropical forests sensitive to spatial scale, for which careful quantifi- may be partially limited by local resources, such as cation of diversity gradients at different grains patchy food availability (McGlynn 2006; but see is necessary (Kaspari et al. 2003). At the local Torres 1984). Using twig augmentation experiments, level, heterogeneous habitats may provide patchy Kaspari (1996) found some evidence of nest site limi- distributed resources, which may derive from tation on the forest floor of four forests in Panama and small-scale disturbance processes. For instance, ar- Costa Rica (but see Carvalho and Vasconcelos 1999). tificial disturbances at the ~0.1 m2 scale in a Brazi- As natural forests are converted into agroecosystems lian forest showed that the recolonization of leaf and consequently simplified, nest site limitation of litter ant species was limited by the colonizing abil- both litter-dwelling and arboreal ants increases. In ities of ant species and not by the limitation of litter agroecosystems, ant diversity may be lost due to a resources (Campos et al. 2007). Because ant richness shortage of animal (e.g. shells) or plant derived (e.g. is positively correlated with ant colony abundance twigs and seeds) nesting resources (Armbrecht at small scales (1 m2), factors limiting colony abun- and Perfecto 2003; Philpott and Foster 2005; but see dance may ultimately determine ant species pres- Torres 1984). Furthermore, ant nest-site limitation ence, abundance, and richness (Kaspari et al. 2003). may be stronger in more intensively managed sites (Philpott and Foster 2005) or in sites with a lower diversity of twig resources (Armbrecht 8.4 Consequences of habitat disturbance et al. 2004). and transformation for trophic Disturbance can alter competitive interactions and interactions and ecosystem services colony dynamics. For example, in forests, local dis- provided by ants turbances such as caused by army ant (Eciton)raids 8.4.1 Ant–hemipteran–plant interactions may prevent leaf litter or soil ants from reaching densities high enough to saturate nesting and food Disturbance may alter trophic interactions involving resources (Kaspari 1996). The loss of habitat complex- ants, such as predation, symbioses, scavenging, leaf- 152 ANT ECOLOGY cutting activity by attine fungus-growing ants, and 8.4.2 Ants as biological control agents foraging on extrafloral nectaries or hemipteran-se- Ants have been used for biological control of insect cretions. These changes consequently alter ecosys- pests and fungal pathogens in agricultural, agrofor- tem services that ants provide, both in natural and estry, and forestry systems for centuries (Perfecto disturbed ecosystems. Humans often perceive ants and Castin˜eiras 1998; Philpott and Armbrecht 2006; either as dangerous pests that form associations with Way and Khoo 1992; see Box 7.2). However, the sap-sucking insects or as beneficial predators of pests impact of disturbance on the ecosystem function (Philpott and Armbrecht 2006). Studies evaluating of ants as biological control agents has not received how habitat transformation may affect ant–hemip- as much attention (Philpott and Armbrecht 2006). teran interactions are scarce but suggest that interac- The most obvious disturbances that can alter the tions are highly disrupted by human disturbance, biological control activity of ants are those asso- even creating or exacerbating potential pest pro- ciated with agricultural intensification, such as pes- blems. In Indonesia, Ozaki et al. (2000) described ticide application, tillage (de Bruyn 1999), and sharp declines in scale insect populations (Aulacaspis reduction of plant diversity (Armbrecht and Gal- marina) due to predation by Monomorium floricola and lego 2007; Armbrecht and Perfecto 2003). All of Paratrechina sp. in mangroves (Rhizophora mucronata). these factors reduce predatory activity of ants. Remarkably, planted mangroves were severely in- Coffee agroecosystems have received detailed at- fested while in neighbouring natural forest ants con- tention in recent years, especially from the point of trolled the scale insect. Habitat disturbance may also view of understanding how agricultural intensifica- facilitate invasion of exotic species that alter ant– tion affects biological control provided by ants. Sev- hemipteran associations (see Chapters 14 and 15). eral studies demonstrate reductions in ant diversity For example, Paratrechina fulva is a pest in Colombia, with intensification of coffee systems, but only a few because it associates with hemipterans and because of these examine how the predatory activity of ants it depletes native invertebrate fauna (Go´mez et al. is affected (see Philpott and Armbrecht 2006). In a 2002). Spread of this species tends to be highly fa- few studies, ant removal of pests diminishes with voured by anthropogenic disturbances around la- coffee intensification (Armbrecht and Gallego 2007; goons and sugar cane intensive monocultures Armbrecht and Perfecto 2003), but one study did not (Chaco´n et al. 2000). find any change in ant effects along a coffee intensi- Ant–hemipteran–plant interactions may be ex- fication gradient (Philpott et al. 2008a). Armbrecht tremely rich and non-specific, indicating that a high and Gallego (2007) demonstrated that ants have variety of ant responses are possible with distur- stronger predatory effects on the coffee berry borer bance. Comparing ant–scale interactions in two in shaded coffee farms than in sun coffee. Whereas types of agroecosystem provides some evidence for one species, Gnamptogenys sulcata, a shade-loving this. In intensive coffee plantations in Venezuela, ant, is an effective predator of this pest, but is rare Crematogaster and Camponotus species are considered in sun coffee systems. Finally, a number of arboreal pests because they tend scale insects (Coccus viridis) ant species are important biological control agents (Hanks and Sadof 1990). In contrast, in an organic, in coffee and cacao farms and will be entirely lost if shaded coffee plantation in Mexico, species of these the shade is eliminated (Perfecto and Castin˜eiras same two genera tend C. viridis, but the scales are not 1998). In particular, A. instabilis is a keystone spe- considered pests in the plantation. Instead, a mutual- cies, associated with the regulation of three main istic relationship between Azteca instabilis and C. vir- coffee pests (green coffee scale, coffee berry borer, idis effectively protects coffee plants from attacks by and coffee leaf rust) in Mexican plantations. The coffee’s most severe pest, the coffee berry borer (Hy- elimination of shade trees will most likely eliminate pothenemus hampei) (Perfecto and Vandermeer 2006). this species along with its biological control function However, A. instabilis is negatively affected by shade (Perfecto and Vandermeer 2008b). Aside from the tree pruning (Philpott 2005a), suggesting that even effects of individual ant species as predators, some minor habitat disturbances may influence ant–he- evidence demonstrates that behavioural diversity of mipteran–pest interactions. ANT BIODIVERSITY AND FUNCTION IN DISTURBED AND CHANGING HABITATS 153 ants may be important in the predatory role of ant 8.4.4 Ant effects on soil processes and assemblages (Philpott et al. 2008b), thus any distur- nutrient cycling as ecosystem engineers bance affecting ant diversity may also affect ecosys- Ants also act as agents of disturbance and modifi- tem function. cation to soils due to their role as ecosystem engi- neers. Ecosystem engineers are organisms that directly or indirectly modify the availability of re- 8.4.3 Ants as seed dispersers sources for other species by causing physical Generally, seed dispersal is affected by biotic fac- changes in biotic or abiotic material. Ants, by their tors (vegetation structure, ant composition, ant size, construction of nests, perturbation of soils and in- nest density, and competition for resources) and teractions with many other organisms are impor- abiotic factors (temperature and seed desiccation tant ecosystem engineers and keystone species rates) (Guitian et al. 2003; Ness 2004). As such, (Decae¨ns et al. 2002; Folgarait 1998). Although the both natural and human-caused habitat distur- impacts of ants on soil physical and chemical struc- bances can have strong impacts on seed dispersal ture is not as well known as for termites and earth- by ants. Because myrmecochory is a mutualism worms, their effect is likely to be substantial involving non-specific sets of partners, ant abilities (Vandermeer and Perfecto 2007). to disperse seeds, as well as rates and distances of The most visibly obvious effect of ants as ecosys- seed dispersal, will depend strongly on the ant tem engineers is bioturbation of soils through the species present (see Chapter 6). Several studies formation of mounds, subterranean galleries and have examined impacts of habitat modification chambers, and the movement of soil particles and fire on seed dispersal by ants. In highly dis- along the soil profile (Folgarait 1998). These soil turbed sites devoid of vegetation, dispersal rates modifications directly and indirectly affect the en- decrease drastically (Andersen and Morrison 1998; ergy flow, habitats, and resources for other organ- Guitian et al. 2003; although see Parr et al. 2007). isms, especially plants and soil micro-organisms. Similarly, Guitian et al. (2003) found lower seed Through formation of underground galleries, ants dispersal rates in open woodlots and hypothesized increase the drainage and aeration and reduce the that this was due to lower ant activity, and quicker bulk density of the soil. Through transformation of seed desiccation. Yet in highly disturbed crop areas organic matter by storing food and accumulating seed dispersal rates may increase (Heithaus and faeces and corpses, ants provide habitat for soil Humes 2003), perhaps due to lower quantities of micro-organisms and enhance soil nutrient condi- available seed resources compared with native ha- tions (Brian 1978; Folgarait 1998). Ants can move up bitats. In disturbed sites where species composition to ten tons of soil per hectare per year in moist shifts towards small-bodied ants, dispersal dis- subtropical and temperate systems (Paton et al. tances generally decline (Heithaus and Humes 1995). Leaf-cutting ants are among the most impor- 2003; Ness 2004), but where composition shifts to- tant agents of soil modification in the tropics, wards large-bodied ants with larger foraging moving biomass, altering chemical composition, ranges, dispersal distances can increase (Andersen and altering soil structure with complex galleries and Morrison 1998; Parr et al. 2007). Finally, seed (Folgarait 1998). In Brazil, a single colony of Atta dispersal in disturbed sites may be affected by high sexdens deposited 40 tons of soil on the surface abundance of exotic ant species (Ness 2004; see (Autori 1947). Perfecto and Vandermeer (1993) es- Chapter 15). Because seed predation also increases timated that Atta cephalotes could cause complete in disturbed and invaded habitats (Andersen and soil turnover in as little as 200 years in a lowland Morrison 1998; Ness 2004) care should be taken in rain forest in Costa Rica. In terms of global rates of equating seed removal with seed dispersal – some animal perturbation ants are considered second seeds removed by ants may not be dispersed far only to earthworms (Folgarait 1998), probably due enough to convey an advantage. to their high biomass. 154 ANT ECOLOGY

Many studies have investigated the roles of ants cation is delayed due to declines in Actinobacteria in creating nutrient and soil heterogeneity and abundance in mounds as compared to adjacent modifying soil structure (reviewed in Folgarait soils (Folgarait 1998). In Panama, Atta columbica 1998). Some studies in disturbed habitats provide increases the flux of 13 chemical elements by 38 information about how the interaction of habitat times compared to surrounding areas (Haines disturbance and ant activity affect soils. Both 1978), and in Puerto Rico leaf-cutting ant activity grazing and tillage of agricultural fields affect is associated with higher plant productivity, pre- nest densities in agricultural landscapes with sumably because of an increase in phosphorous subsequent effects on soil processes (Beever and availability (Lugo et al. 1973). Herrick 2005; de Bruyn 1999). Atta (a genus with All the disturbances that affect ant communities large impacts on tropical soils) are strongly affected can also alter the impact of ants as ecosystem engi- by edge effects, deforestation, and presence of neers. Tillage, pesticide use, and decreases in organ- roads (Vasconcelos et al. 2006), but resulting im- ic inputs associated with agricultural intensification pacts on soil processes have yet to be quantified. affect macrofaunal activity and diminish the contri- Decae¨ns and colleagues (2002) examined how con- bution of ants as soil modifiers (Decae¨ns et al. 2002). version of native savanna to crops and pasture Likewise, the increase of invasive species associated affects ants and subsequent influences on soil struc- with some disturbance events can also alter ecosys- ture. They found a significant decline in diversity tem engineering by ants. The effect of disturbance of bio-structures (e.g. tunnels and other structures on the ecosystem engineering activities of ants has created by invertebrates) in crop areas compared seldom been studied directly. with grasslands and that the presence of a high density of ants and other ecosystem engineers 8.5 Future directions maintained a diverse assemblage of soil particle sizes. In restored grasslands, ants and ant mounds There are several topics covered here that deserve generate soil heterogeneity (in variables such as soil attention in future studies. Given phylogenetic dif- texture, bulk density, soil temperature, and soil ferences among global regions, ant ecologists moisture) and create microsites for plant coloniza- should make it a priority to establish classifications tion in restored grasslands, but these effects may for different regions to assess which ant species, change with age since recovery (Lane and Bassir- species groups, and genera will be affected most iRad 2005). Thus, there are isolated examples of ant by different types of disturbance. As mentioned in effects of soils in some disturbed (and recovering) Box 8.2, functional groupings have been developed habitats, but there is still a general lack of under- for predicting community level responses to distur- standing of how changes in ant richness, composi- bance and stress at a global level. Yet, there have tion, and nest densities will affect soil processes in been some criticisms that these functional classifi- disturbed habitats. Many research objectives posed cations are somewhat specific to Australasia. For by Folgarait (1998), including comparing effects of most mutualisms and ecosystem services described ants in areas with impoverished diversity due to (fungus cultivation, hemipteran-tending, seed dis- the entrance of exotic species and consequent loss persal, and biological control of pests), the way in of native ant fauna, or the impact of habitat distur- which habitat disturbance affects the interaction bance are still valid topics for further research. relies very heavily on the composition of the ant Ants also alter soil chemistry and affect nutrient species present in the disturbed habitat. But as for immobilization, indirectly affecting plant and other taxa, it is now becoming increasingly recog- microbial communities (Dauber et al. 2006b; Lugo nized that functional classifications can provide et al. 1973). Most studies show an increase of organ- important tools for determining how species ic matter and N, P, and K in ant mounds, as com- groups with certain traits respond to disturbance pared to adjacent soil samples (Decae¨ns et al. 2002; or other ecological factors. Folgarait 1998). Decomposition processes by fungi A much more detailed understanding of how and ammonifying bacteria increase, while humifi- disturbed ant communities perform ecosystem ANT BIODIVERSITY AND FUNCTION IN DISTURBED AND CHANGING HABITATS 155 functions would be beneficial for both maintaining MacArthur and Wilson 1967), and that the size of or even increasing the ecosystem services provided the disturbance and distance to source populations by ants and for restoration. Outside of coffee agroe- affects colonization success. With suitable distribu- cosystems, few studies have examined how tions along disturbance gradients, or at increasing biological control services provided by ants are af- distances from source populations, studies of dis- fected by agricultural intensification. The studies persal and recruitment limitation of ants in dis- have also not examined whether habitat changes turbed areas may enlighten investigations into the cause behavioural changes in particular ant species assembly rules of ant communities generally. thus altering their relative impact as hemipteran- Finally, we lack research examining the synergis- tending ‘pests’ versus biological control agents. tic effects of multiple disturbances on ant commu- There is little work examining how habitat distur- nities and ecosystem services. Most of the work bance per se influences ant effects on soils. Other examining ants is limited to particular regions, than understanding that tillage has extreme im- habitat types, or commonly studied assemblages. pacts on soil ants, what other more minor distur- Although challenging, research that replicates ex- bances affect ants? How do the roles of ants differ in perimental designs in multiple habitat types or in conventional versus sustainable agricultural sys- multiple regions may reveal those processes impor- tems? Additionally, it would be very useful to tant for the maintenance of ant assemblages and the know whether ants could eventually be used as important services that they provide. functional agents in the recovery of degraded land- scapes via seed dispersal. A more detailed under- standing of the importance of increased seed 8.6 Summary dispersal distance in disturbed habitats should be investigated. Further, despite a relatively large There are some generalizations that can be made number of studies examining seed dispersal in dis- about effects of habitat disturbance and transforma- turbed habitats, few mention how dispersal is af- tion on ant communities, but several areas deserve fected with relative changes in seed densities across much more attention. Fire, flooding and inunda- disturbance gradients or how the relative propor- tions, forest tree-fall gap creation, hurricanes, log- tion of myrmecochorous seeds changes with distur- ging, fragmentation, agricultural intensification, bance. How might humans manipulate ants to grazing, mining, and urbanization can have very benefit restored grasslands or mines? drastic effects on ant assemblages. Generally, distur- Ecological studies relating to the assembly and bances that directly cause colony mortality will have maintenance of ant communities could be especial- different effects on ant communities than distur- ly fruitful in disturbed and transformed habitats. bances that have indirect effects through alteration For example, there are several mechanisms driving of plant biomass (Hoffmann and Andersen 2003). the observed changes in ant communities with dis- Some natural disturbances, such as fire, short floods, turbances, but a more detailed understanding of and tree-fall gaps, although drastic in their immedi- those factors (e.g. resource availability, microcli- ate effects, may have few long-term impacts on as- mate, ecophysiological conditions, changes in pred- semblages if entire colonies are not lost, or if ator or parasite populations) is needed. There is foundresses colonize disturbed sites quickly. Yet ample evidence that both fragmentation and habi- natural disturbances that occur more frequently tat disturbance influence ant assemblages. Working than colonization and establishment may signifi- in disturbed landscapes may help elucidate ques- cantly exclude ant species not adapted to disturbed tions such as, how do habitat configuration or other conditions. Generally, human disturbance result in landscape factors affect ants? What is the relative greater changes in ant species composition than nat- importance of local vs. landscape factors in deter- ural disturbances (e.g. mining, urbanization, and mining ant diversity and composition? It is repeated agriculture have especially severe effects). long known that dispersal of colonists is important As with other animal communities, disturbance im- for the recovery of original communities (e.g. pact will depend on frequency and intensity of 156 ANT ECOLOGY disturbance and the time over which habitats recov- bance and transformation have an impact on local er from perturbation. ant assemblages both indirectly through changes to The specific effects of disturbance on species com- habitat structure, and directly, through reduced position and functional groups based on the studies resource availability and removal of colonies. Al- presented here may seem specific to the type of dis- though some mechanisms have been examined, turbance and the study region. Yet, in most cases, much more work is needed to understand the disturbed sites are dominated by opportunistic or details. generalist species presumably because these species Finally, ants provide essential ecosystem functions can take advantage of changing resource bases, espe- such as biological pest control, seed dispersal, and cially when disturbance puts them at a competitive soil modification, many of which are affected by hab- advantage (Hoffmann and Andersen 2003). In most itat disturbance and transformation. Interactions of habitats affected by human disturbance, invasion by ants with hemipterans may significantly change in exotic or tramp ants is often reported and prevalent, disturbed habitats. Some evidence shows that but in ecosystems with frequent natural disturbance biological control capabilities of ants decline in dis- (fires and floods), only certain native ant species turbed habitats either due to a loss of diversity and seem adapted to local conditions, sometimes pre- change in vegetation structure, or because of shifts cluding invasion. in abundance or composition of ants. In some dis- Several mechanisms are implicated in changes in turbed habitats, seed dispersal capabilities increase, ant species richness, abundance, and composition whereas in other habitats they decrease. More work with habitat disturbance and transformation. Ants is needed to examine the implications for plant re- may be highly sensitive to changes in microclimate generation in disturbed and transformed habitats. brought about by changes to the dominant vegeta- Finally, ants have very strong impacts on soils, add- tion structure of a particular habitat and may be ing to nutrient enrichment, nutrient cycling, and to affected by changes in availability of food or nest- the biophysical structure of soils, but much more ing resources. Change in competitive interactions research is needed to understand the intersection of or in colonization processes may also affect ant ant effects as ecosystem engineers and habitat distur- assembly in disturbed areas. Thus, habitat distur- bance.