Is Empodisma minus the ecosystem engineer of the FBT (fen–bog transition zone) in New Zealand?

Abstract

The role of ecosystem engineers (EE) in the formation of ombrotrophic mires (bogs) from fens, called the fen–bog transition (FBT), can be best understood through categorization of the autogenic and allogenic processes causing bog initiation. Here we review these pathways, discuss the drivers of change in both cases, and tabulate an approach for distinguishing between them. We then compare the engineering ability of acknowledged and putative engineers against a number of characters which plants require to cross the FBT, and to stabilize occupancy on the bog side. While some Sphagnum spp. are accepted as the EE of the fen–bog transition in northern hemisphere bogs, they appear unimportant in New Zealand. Instead their role appears to be occupied by a restiad, Empodisma minus, a plant with leafless, wiry stems and capillaroid roots. Empodisma minus appears capable of engineering autogenic transitions from fen to bog across New Zealand, even more efficiently than Sphagnum.

Keywords:

• acidification
• decomposition
• drainage
• macrofossil
• litter
• microbes
• nutrients
• palatability
• peat
• peat precursors
• pH
• productivity
• water table

Introduction

The ecosystem engineer (EE; Jones et al. 1994; Charman 2002; Wright & Jones 2006) is a concept which has proved useful in elucidating some important ecological processes. Engineers are ‘organisms that directly or indirectly modulate the availability of resources (other than themselves) to other species, by causing physical state changes in biotic and abiotic materials’ (Jones et al. 1994, p. 374). While the ecological effects upon other species are described as ecosystem engineering (Kylafis & Loreau 2008), the evolutionary consequences of feedbacks to the EE have more recently been addressed in the concept of niche construction (Odling-Smee et al. 1996, 2003). While noting the comments of Reichman & Seabloom (2002a, b) and Wilby (2002), that engineering can be interpreted teleologically and can be all-encompassing, as Jones et al. (1994) previously noted, the concept has received widespread application in the mire literature, and increases our understanding of the processes involved.

Species from the bryophyte genus Sphagnum are widely considered to be EEs of mires in the northern hemisphere (Jones et al. 1994; Moore 1995; Svensson 1995; van Breeman 1995; Frankl & Schmeidl 2000). Although several Sphagnum species occur in the southern hemisphere, they appear to have no engineering role here (Moore 1995), restiad rather than Sphagnum bogs being the norm (Campbell 1975; Whinam et al. 2003). The only known southern hemispheric candidate for mire engineering is Empodisma minus(Hook. f.) Johnson & Cutler s.l.. A small member of the ‘jointed rush’ family, the Restionaceae, a southern hemisphere family closely allied to the Poaceae, Empodisma minus is widely distributed throughout New Zealand, Tasmania and the south and east of the Australian mainland (Campbell 1964; Moore 1995; Charman, 2002).

The process being considered here is the fen–bog transition (FBT), where an oligotrophic fen (terminology as per Wheeler & Proctor 2000) changes to an ombrotrophic bog. A peat accumulating system or zone is a prerequisite, but it is otherwise far from clear how the process occurs (Hughes 2000; Hughes et al. 2000, 2007; Hughes & Dumayne-Peaty 2002; Hughes & Barber 2003, 2004).

Since pathways across the FBT have been inadequately categorized, we here review the models of the pathways involved. We use this as background to discuss the role of ecosystem engineers in crossing the FBT, and review the characters EEs must possess to manipulate the environment in relevant ways. We go on to compare the abilities of Empodisma minus as EE of the FBT with Sphagnum spp., in addition to putative EEs in northern hemisphere and New Zealand mires, respectively the cyperad Eriophorum vaginatum L. and poad Chionochloa rubra Zotov.

Crossing the FBT

In its simplest sense, peat is the accumulated remains of undecomposed plant material. A peat deposit forms when the rate of deposition of dead plant matter produced by net primary production exceeds the rate of decomposition, i.e. Decomposition/Productivity = D/P <1 (Clymo 1983, 1984). The rate of decomposition is the more variable determinant of peat accumulation (Damman 1986; Charman 2002), and it is affected by depth of burial (Clymo 1965), temperature (Rosswall 1974), water table depth (Clymo 1983), water table fluctuation (Belyea 1996), plant litter type and quality (Linkins & Neal 1982; Johnson & Damman 1991; Kuder et al. 1998; Scheffer et al. 2001; Trinder et al. 2009), oxygen supply (Clymo 1983), and microbial populations (Linkins & Neal 1982; Sundh et al. 1997; Pankratov & Dedysh 2009). In a bog, primary productivity occurs atop the acrotelm (euphotic and aerobic decay layers; Clymo 1984, 1992), while decomposition occurs in both the acrotelm and catotelm (collapsing, anaerobic decay layers).

Accumulation of fen peat is in itself inadequate to engender a transition to a bog. Three other developments are required (Moore 1995; Hughes & Barber 2003):

• hydrological isolation of the peat surface from groundwater and surface water, so that the only nutrient source is nutrient-poor rainfall (hence inducing ombrotrophy);

• occurrence of mechanisms to retain water within the elevated peat mass;

• establishment of plant species capable of growth in the newly created oligotrophic and acidic conditions, which then perpetuate the trophic change.

The requisite developments are not achieved simply. A peat system where decomposition equals, if not exceeds, productivity (i.e. a fen) has to be replaced by one in which productivity exceeds decomposition (i.e. a bog). The transition between the two states is the crucial phase of this process, hence the name FBT. Changes in pH, nutrient status and water table are in part consequent upon those changes, and in part induced by the incoming bog species. When ombrotrophic specialist species establish and dominate the mire community, they demonstrate that its trophic status has changed to that of a bog (Moore & Bellamy 1974). However, the above developments are descriptors of events, but are not causal, and so are not in themselves sufficient as explanations for the mechanisms of transition from fen to bog.

There are two potential pathways across the FBT, based upon allogenic vs. autogenic forcing mechanisms (sensu Nicholson & Vitt 1990), which we outline here. We use “allogenic” to refer to processes which are external to the system concerned (omitting those which are anthropogenic in origin, e.g. Hughes et al. 2000, 2007; Hughes and Barber 2003), and so must be of abiotic origin. This includes processes such as climate change, or physical or hydrological events. “Autogenic” processes are those resulting from inherent features of systems, particularly the plants themselves, and the peat derived therefrom, and so do not occur in the absence of adapted vegetation.

Allogenic pathways across the FBT

Allogenic pathways to transit an FBT operate in two categories (Table 1). The first set of mechanisms isolate an existing peat outcrop by physical means, so that only meteoric water can impinge upon the incipient bog surface, increasing its hydrological isolation, and initiating ombrotrophic conditions:

• physical decrease in water table, which, when occurring allogenically, can only be by changes in catchment-level hydrology such as that caused by landslides (cf. Lowe et al. 1999), river channel adjustments (Kulczynski 1949; Campbell 1964), sea level changes (Newnham et al. 1995), isostatic rebound (Glaser et al. 2004), and tephra deposition (de Lange 1989; Shearer 1997);

• climatic change in rainfall leading to groundwater flow reversals, isolating the peatland surface (Winkler 1988; Foster & Wright 1990; Siegel et al. 1995; Glaser et al. 1996; McGlone et al. 1997; Halsey et al. 2000; Robichaud & Begin 2009).

The second set of mechanisms leads to an increase in height of a peat body by decreasing the peat decomposition rate (by implication, while maintaining D/P < 1):

• decrease in temperature, decreasing evapotranspiration, thereby decreasing the depth of the (aerobic) acrotelm, so that litter reaches the catotelm in a less decomposed state;

• increase in precipitation, so that the ground water influence on the peat mass is lessened, and ombrotrophy is promoted.

These allogenic mechanisms result in warmer and better-drained patches, which will immediately be exposed to heightened erosion and decomposition rates, reducing the projecting outcrop of peat, and tending to bring it into line with the main fen peat surface. However, the patch may be stabilized by timely establishment of ombrotrophic vegetation, completing the FBT.

Autogenic pathways across the FBT

A large-scale autogenic mechanism for crossing the FBT which comes into play in fens, but only over long expanses of time, is the lateral expansion of the mire surface outwards from its founder location, and across the landscape (Korhola 1992; Table 1). Eventually such a system will become sufficiently large spatially (Heinselman 1970) for meteoric water to become the only source of nutrients onto the inner parts of the mire (the ‘stagnation zones’ of Vitt 1994), as any lateral inflow of water is stripped barren of nutrients by competing organisms before reaching the centre. This creates centralized bog conditions into which ombrotrophic specialist species can invade.

Table 1  Parameters which allow differentiation between allogenic and autogenic mechanisms for crossing the fen–bog transition (FBT), including the role of ecosystem engineers (EE), and some examples of papers which apply these parameters to mire studies.

		Allogenic		Autogenic		Example of use in mires
Parameters‘	Tools	Fen → bog		Fen → bog
Climate	Prevailing climate	Not applicable or relevant		Moist		Winkler 1988; Foster & Wright 1990
	Regional climate change	Norm	Cooler and/or wetter	Not applicable		Tallis 1983; Kuhry et al. 1991; Langdon et al. 2003
Catchment history	Mire expansion	Consider		Not applicable		Korhola 1992
	Hydrological changes	Consider		Not applicable		Kulczynski 1949; Heinselman 1970
	Tephra or sediment layers	Absent	Present (buried)	Not applicable		Lowe et al. 1999
Water table indicators	Desiccation surfaces on peat	Present	Absent	Absent		Casparie 1993
	Charcoal fragments	Present	Absent	Absent		Hughes et al. 2000; Barber et al. 2003
	Diatom assemblages (in recent peat)	Differences, especially in Eunotia		Differences		Foster & Fritz 1987; Dell'uomo 1992
	Collembola	Differences		Differences		Slawska 2000
	Testate amoebae assemblage	Low and/or fluctuating water table	Stable, high water table	High water table	Stable, high water table	Charman 1997, 2002; Charman & Warner 1997; Wilmshurst et al. 2003; Lamentowicz & Mitchell 2005; Lamentowicz et al. 2008
	Macrofossil analysis	Low	High	Fluctuating	Stable, high	Johnson et al. 1990; Nicholson & Vitt 1990; Kuhry et al. 1993
Peat characters	Humification	Sapric	Sapric/Hemic	Sapric	Hemic	Shearer 1997; Ellis & Tallis 2001
	Carbon:Nitrogen ratio	Increasing		Stable, high		Kuhry et al. 1991; Turunen & Turunen 2003
Inferred pH		Abrupt transition from high to low		Steady decline		Nicholson & Vitt 1990; Kuhry et al. 1993
Vegetation structure	Presence of EE	EE absent or at low densities	EE absent or present	EE present	EE dominant	Payette 1988; Nicholson & Vitt 1990; Kuhry et al. 1993; Hughes & Dumayne-Peaty 2002; Barber et al. 2003

At smaller scales, the role of mire vegetation in generating autogenic pathways across the FBT has received much recent attention, such pathways requiring the presence of EEs, which “change the environment via their own physical structures, i.e. their living and dead tissue” ( Jones et al. 1994, p. 373). The physical consequences of such engineering on the growing peat body are (Zobel 1988; Malmer et al. 1994; van Breeman 1995):

1. a high internal water table within the peat bog, above the groundwater level of the surrounding minerotrophic fen peat;

2. development in the peat body of an acrotelm, engendering trophic isolation from the catotelm;

3. decline in water table fluctuations within the peat body, i.e. a dampening down of the effects of rainfall events and seasonal variations;

4. decreasing ash or inorganic content and increasing carbon content in peat as rheophilous inputs decline;

5. slowing of decompositional processes due to changes in soil chemistry, and to increasing anoxia and acidity, which result in a decline in decomposer (invertebrate, microbial and fungal) communities, slowing the decomposition process;

6. increasing proportion of recalcitrant material (which is slow to decompose) in the accumulating peat.

The process of engineering the FBT requires the presence of an EE, which differs from other species potentially contributing peat in its ability to grow in both the fen and bog situations, i.e., it is present before, during, and after the engineering event (visible in the peat profile; Table 2), and is increasingly competitive in the bog, so that it dominates. The EE alters the environment under some or all of the categories of peat accumulation, water table maintenance, and pH manipulations. We take accepted and putative EEs of the FBT from both hemispheres, and tabulate their engineering abilities under these categories (Table 2).

Table 2  Characters required for a species to be an ecosystem engineer (EE) across the fen–bog transition (FBT). The match to requirements of four potential EE taxa is categorized as good (√), bad (X), mixed (±), and unknown (?).

Character	Northern hemisphere Sphagnum spp.	Eriophorum vaginatum	Empodisma minus	Chionochloa rubra
Ecosystem engineering ability present
EE present both sides of the FBT	√ S. fuscum (Heinselman 1970; Payette 1988; Kuhry et al. 1993) S. fuscum has wide edaphic range (Bragazza 1997) S. imbricatum (Hughes & Dumayne-Peaty 2002)	X Temporary precursor only to Sphagnum-dominated raised bog (Hughes & Dumayne-Peaty 2002)	√ Common in mesotrophic to oligotrophic stages (Cranwell 1953; Shearer 1997; Clarkson 2002; McGlone 2009)	√ Wide edaphic range (Connor 1991)
Persistent macrofossil occurrence of EE throughout peat core	√ Preserved remains in 52% of samples in near continuous bands (Tallis 1994)	± Periodic peaks in leaf sheath bundles and spindles (Tuittila et al. 2007) May be a major peat former (Hughes pers. comm.)	√ EE persistent (Cranwell 1953; Campbell et al. 1973; Walker et al. 2001)	X Not a significant peat former in a mire largely lacking Empodisma (Walker et al. 2001)
Accumulation of peat precursors
Low palatability to herbivores	√ Unpalatable (Clymo & Hayward 1982)	X Palatable; grazed by sheep, cattle, lemmings, ground squirrels, geese and caribou (Chapin & Slack 1979; Grant et al. 1987; Wein & MacLean 1973)	√ Unpalatable (van Rees & Hutson 1983; van Rees & Holmes 1986; McDougall 2007)
	√ Unpalatable; ‘worthless’ character as sheep feed (Cockayne 1967, p. 121) Fertilization increases palatability (Connor et al. 1970) Organic matter digestibility is low (Fenner et al. 1993)
Inherent inertness of litter	√ Low rate of decomposition (Aerts et al. 1999) High phenolic content (Dickinson & Maggs 1974) High refractory nature of peat (Scheffer et al. 2001) Low N content impacts on decay (Aerts et al. 2001) Presence of sphagnan may cause a form of C limitation as consequence of humification (Thomas & Pearce 2004)	√ Low rate of decomposition (Coulson & Butterfield 1978; Cornelissen 1996; Hughes & Barber 2004) Non-significant differences between nutrient content in rain-protected and unprotected leaves (Jonasson & Chapin 1985) Litter of rhizomes, leaf bases and roots accumulates as peat (Hughes pers. comm.)	√ Inherently low tissue nutrients (Agnew et al. 1993; Sharp 1995; Clarkson et al. 2005) Decay resistant outer cortex, central strand and endodermis in root (Campbell 1981) High polyphenol and lignin in stems; anhydro-galactosan in root hairs (Kuder et al. 1998) Heavily lignified mature roots (Sorrell et al. 2000)	√ Lignin 6.8–10.3% (oven dry weight) in shoot material (Connor et al. 1970) High mechanical strength (Connor & Bailey 1972) High mean tiller weight and leaf dieback, and low mineral content in low nutrient soil (Lee & Fenner 1989)
Inhibition of microbial activity	√ Microbial inhibition (Dickinson 1983; Verhoeven & Toth 1995; Borsheim et al. 2001; Bonnett et al. 2006; Opelt et al. 2007; Pankratov & Dedysh 2009) Toxic organochemicals (Verhoeven & Liefveld 1997) Sphagnol and sphagnan inhibit decomposition (Smidsrød & Painter 1984; Rudolph & Samland 1985; Painter 1991; Stalheim et al. 2009)	?	√ Microbial inhibition due to allelopathic properties of tissues (Kuder et al. 1998) Low microbial biomass (Schipper et al. 1998) Litter and moss protected from degradation through incorporation into negatively geotropic root mass (Campbell 1981; Agnew et al. 1993)	?
Maintenance of high and stable water table
Reduced evapotranspiration	√ In hummocks, internal conduction of water re-wets hummocks (Titus & Wagner 1984) In hummocks, inrolled cucullate leaves and compact form reduce surface area for evaporation (Green 1968) In hollows, dehydrated S. cuspidatum hyaline cells have high albedo, reflecting light and reducing temperature (van der Molen & Wijmstra 1994)	√ Increased relative humidity under canopy (Lavoie et al. 2005)	√ Low evapotranspiration rates (<2.5 mm day^-1) due to thick litter and dense canopy (Campbell & Williamson 1997) Stomatal conductance maximized in morning (Sharp 1995)	√ Low transpiration losses c.f. Festuca novae-zelandiae and Chionochloa australis (Espie 1999) Few stomata, located in deep adaxial furrows, preventing desiccation (Espie 1999)
Water-holding capacity	√ High water holding capacity (10–25× own weight) (Andrus 1986) High effective porosity with depth (Holden 2009) Maintenance of macropore structure via litter quality (Turetsky et al. 2008)	√ Peat mass efficient at retaining water during periods of desiccation (Hughes et al. 2000)	√ High water holding capacity (15× own weight) in sponge-like root weft (Campbell 1964) Dense canopy retains moisture at deeper canopy levels (Campbell & Williamson 1997) Moisture retained in peat (Campbell et al. 1995)	?
Decreased vertical drainage	√ Hydraulic conductivity decreases with depth (Adema et al. 2006; Price et al. 2008) Sphagnum structure collapses and bulk density increases with depth (Hayward & Clymo 1982; Johnson et al. 1990) Variable infiltration rates with depth (Holden 2009)	X No change in infiltration rates with depth (Holden 2009)	√ Decreasing hydraulic conductivity with depth; most water exchange in surface 1.5 m of peat layer (Maggs 1997) Acrotelm drainable porosity halved within 200 mm of surface (Miller 1994)	?
Drainage barriers	√ Iron hardpan formation (Andrus 1986) Fe-Mg hardpan formation (Klinger 1996; Kusel et al. 2008)	√ Iron oxide deposits in soil surrounding roots (Koch et al. 1991)	?	?
Adaptations to water table fluctuation	√ Ability to grow under both wet and dry conditions of mire surface (Green 1968; Casparie 1972; Barber 1981; van der Molen & Hoekstra 1988)	√ Wide tolerance to water table depth (Gore & Urquhart 1966; Barber 1981)	√ Structural elements may protect chlorenchyma from collapse in dry spells (Linder 2000) Surficial roots avoid anoxia by negatively geotropic growth habit (Campbell 1981; Sorrell et al. 2000) Surface rhizomes maintain atmospheric O2 concentrations (Sorrell et al. 2000)	√ Wide tolerances (MacIntosh et al. 1983)
Development of mire microforms	√ Hummocks (Green 1968; Titus & Wagner 1984) S. imbricatum has two morphologically distinct ecads, allowing it to grow in both hummocks and semi-aquatic habitats (Green 1968)	?	√ Hummock and hollow topography forms near edges of mire (Rogers 1984)	X No microtopography known (pers. obs.)
Features altering pH
Acidification	√ pH of approximately 4 (Sonesson 1970; Braekke 1981; Glaser et al. 1981) Acidification via polyuronic acids (Clymo 1963; Bellamy & Reiley 1967) Polyuronic acid content 10–30% of dry weight; larger values for hummock species (Clymo 1964, 1967; Vile et al. 1999) Organic acid production acidifies subsurface waters (Hemond 1980)	± Low cation exchange capacity (Hughes & Dumayne-Peaty 2002) Acidification of substrate occurs under continental climate conditions (Hughes pers. comm.)	√ Acidification via polyuronic acids, at higher levels than for Sphagnum (Bannister 2000)	?

Enhanced accumulation of peat precursors from above- and below-ground tissues is due to reduced consumption and decomposition by:

• production of herbage of low palatability to both herbivores and detritivores (Clymo & Hayward 1982), or existence of physical barriers to herbivory such as hairs, spines or waxes (Cornelissen 1996), increasing litter retention on site;

• production of sclerophyllous tissues, high in lignin and in phenolics such as tannins (Cornelissen 1996), and low in nutrients (Clarkson et al. 2005), rendering litter inert and so reducing consumption;

• production of anti-microbial components such as sphagnan (Stalheim et al. 2009), reducing decomposition.

These features are fairly consistently found in species of mires (Table 2), especially in accepted EEs, though this may reflect availability of evidence. However, edibility and palatability render a species less successful as an EE.

Maintenance of a high and stable water table is attained by:

• reduced evapotranspiration from the vegetation canopy (Campbell & Williamson 1997), using a range of passive and active characters;

• heightened water-holding capacity within living and attached dead plant tissue mass due to high capillarity (Moore 1995), and the maintenance of macropores within the decomposing litter (Clymo and Hayward 1982; Turetsky et al. 2008);

• decreased vertical drainage down the peat profile due to increasingly dense packing with depth of partially decomposed material (Moore 1995), leading to lower hydraulic conductivity and permanent saturation (Clymo & Hayward 1982; Clymo 1992);

• drainage barriers due to mineral deposition with leaching (Andrus 1986);

• adaptation to water table changes, allowing growth in a range of hydrological conditions (i.e. both fen and bog);

• development of mire microforms such as hummock-hollow topography, which help inhibit the subsurface and lateral runoff (Ivanov 1981; Ingram 1983).

Water table manipulations appear to occur slightly less commonly in species with apparently limited engineering ability (Table 2) though evidence is mixed for Eriophorum and almost non-existent for Chionochloa. It appears that ability to use all of these mechanisms is important for an EE due to feedback loops. For example, the high water table imposes physical and chemical constraints (e.g. promotion of reductive geochemical pathways) on the decomposer communities in the substrate, further increasing peat accumulation, while in the wetter substrate, litter decay slows due to lower temperatures and limited rates of gas diffusion.

In addition to the more global effects of carbon dioxide levels, precipitation of acidic rain and the activities of sulphur-metabolizing bacteria on pH levels (Clymo 1964), high cation exchange capacity results in acidification of the substrate (Clymo 1963), lowering th pH, limiting competition and decreasing decomposition (Mitsch & Gosselink 2000). Thus pH manipulations may be a crucial characteristic for an EE (Table 2).

The beauty of an engineering process is that neither the species nor the system needs to be dominant or large, for the mechanism to apply (e.g. engineering can occur in small kettlehole bogs or even on quarry floors; Lamentowicz et al. 2008 and Andreas & Host 1983, respectively). Nor are long periods of time essential, providing a fen-based peat is present (Cranwell 1953; Zobel & Masing 1987; Shearer 1997; Robichaud & Begin 2009). Therefore it is possible for FBTs to occur in small, isolated patches of a larger mire system, as proposed by Vitt (1994). Indeed, spatial dispersion of successfully engineered FBTs might be the ultimate cause of the string or flark patterning in mires which has generated so much discussion in the literature (e.g. Moore 1982, 1991; Glaser & Janssens 1986; Foster & Fritz 1987; Rapson et al. 2006; Eppinga et al. 2008).

Persisting on the bog side of the FBT

A successful engineer must be dominant in the engineered environment. Thus the engineering event can be assumed to be part of the evolutionary strategy of the species for self-perpetuation (Jones et al. 1994; van Breeman & Finzi 1998). Once on a mire the EE's structures, in altering the flow of nutrient resources, create an oligotrophic environment in which it is the superior competitor (Table 3). All EE characters appear dedicated to this end, so that nutrient losses from tissues accumulating as peat are matched by nutrient inputs only obtainable from atmospheric sources. Thus competitive traits in the engineered environment include superior levels of nutrient access, and rates of nutrient uptake and retention.

Superior nutrient access, compared to potential competitors, is achieved by:

• water pre-emption, i.e. increased opportunity to intercept meteoric water, which is the primary incoming nutrient source in ombrotrophic conditions, so that any nutrients carried in the rainfall are disproportionately available to the intercepting species (Agnew et al. 1993);

• deep root systems extending to the base of acrotelm to enhance nutrient access (Adema et al. 2006).

Superior nutrient uptake is achieved by some or all of:

• enhanced rates of cation exchange, increasing acidification (Clymo & Hayward 1982) and nutrient supply;

• the development of root adaptations (such as cluster roots and root hairs) to low nutrient conditions (Lamont 1982; Lambers et al. 2006), which also increase below-ground productivity and hence the below-ground contribution to peat accumulation rates.

Superior nutrient retention is achieved by:

• the production of foliage which is long-lived, evergreen, and of reduced size, increasing nutrient retention times and decreasing nutrient requirements (Aerts 1999; Aerts & Berendse 1989);

• translocation of scarce nutrients to perennating organs prior to litter fall (Escudero et al. 1992; Aerts 1993), with large and long-lived rhizome and root systems through which scarce nutrients may be relocated or stored (Chapin 1980);

• enhanced rates of release of limiting nutrients from decomposing tissues for rapid resorption from within the upper peat layers (Moore & Bellamy 1974; Smith 1981), perhaps via capillary water flow on the exterior of the EE's tissues (Clymo & Hayward 1982).

Another competitive advantage may be obtained by suppression of competitors at germination, e.g. where recruitment may fail due to low light availability under a dense vegetation canopy or litter layer (following Xiong & Nilsson 1997 for riparian wetlands). In addition, perennial competitors must be capable of continuous upward growth as the peat surface rises, with new roots developing annually on a higher level (Malmer et al. 1994).

Identifying engineered systems

The two models for crossing of the FBT are differentiable, theoretically and mechanistically, historically and spatially, via a range of environmental, historical and peat characteristics (Table 1), no one line of evidence being sufficient (Langdon et al. 2003; Pellerin & Lavoie 2003). In physical terms, the regional climate is important in determining whether a mire can form, but, further, a change in climate can induce allogenic crossing of the FBT (Table 1), as can mire size increases and hydrological changes which result in more stable peat surfaces. The contribution of testate amoeboid assemblages (Charman 1997, 2002; Charman & Warner 1997; Woodland et al. 1998) as bioindicators of the FBT looks particularly promising, especially as they have no known role in creating the changes they indicate (McMullen et al. 2004; Payne & Pates 2009), though their close associations with plant species admits of circularity of logic (PD Hughes, pers. comm.). While fruitful, and easy to obtain, macrofossil evidence is also handicapped by circularity in determining causes of the FBT (Whinam & Kirkpatrick 1995), as is the use of level of degradation of plant polysaccharides as a proxy for bog palaeohydrology (Kuder & Kruge 1998, 2001). Decrease in desiccation indicators (e.g. Casparie 1993), decline in fire rates (e.g. Hughes et al. 2000), and declining fluctuations in the water table (e.g. Charman & Warner 1997), are also suggestive of allogenic crossings, while stable or gradual changes in water table indicators contribute to evidence of autogenic processes. The presence of a putative EE on a bog can be informative, but is not necessarily conclusive evidence of an ability to engineer the FBT, unless the species is also known to have been present in the fen stage (Table 1), e.g. occurs in quantity in the fen peat profile.

Though FBTs have been frequently recorded (e.g. Glaser 1992; Hughes & Barber 2004; Muller et al. 2008), most literature we are familiar with does not present the requisite information to allow us to unequivocally assign the studied mires to either autogenic or allogenic pathways across the FBT, impeding an assessment of the relative frequency of these models. Allogenic pathways are identified by Winkler (1988) and Nicholson & Vitt (1990), and autogenic ones by Lamentowicz et al. (2008). Kuhry et al. (1993) argue that even within the framework of a wetter climate shift, the FBT in Canadian boreal mires should be attributed to autogenic acidification by Sphagnum species, rather than to any particular climate variable. Both pathways, operating at different times, were inferred by Payette (1988) and Hughes & Dumayne-Peaty (2002). For example, climate change to increased precipitation and cooler temperatures (an allogenic process) may slow decomposition and favour ombrotrophic Sphagnum growth (autogenic) in poor fens, while further decreasing decomposition rates by raising regional water tables (allogenic; Langdon et al. 2003). An EE is probably also common as an accelerant of allogenically triggered pathways (Glaser et al. 1997; Hughes & Dumayne-Peaty 2002; Hughes & Barber 2004; Robichaud & Begin 2009), and such hybrid transition types, where “internal factors operate within the framework of external factors” (Winkler 1988, p. 1032), may well be the norm in the development of bog complexes. Hybrid situations would greatly increase the likelihood of crossing the FBT, autogenically stabilizing any allogenic crossing, while reducing the time elapsed in attaining the conditions necessary for a successful crossing.

Who is engineering the FBT?

A successful EE must have the ability to induce environmental changes under some or all of the three manipulations detailed above (i.e. enhanced accumulation of peat precursors, maintenance of high and stable water table, and substrate acidification). In the northern hemisphere Sphagnum is the classic plant engineer (Table 2), its abilities widely extolled in the literature (e.g. Damman 1986; Stoneman et al. 1993; van Breeman 1995; van Breeman & Finzi 1998; Charman 2002; McMullen et al. 2004). It is of extremely low palatability, and, though of modest productivity, is very slow to decompose, so that its peat accumulates. Almost all evidence regarding its engineering capacity is positive for an EE (11–12 of 12 possible engineering characters in Table 2), except for some facets of water evaporation from hummocks. Apart from the lack of storage organs, it is highly competitive, particularly for nutrients, and so demonstrates five out of eight characters known to stabilize occupancy of a bog (Table 3). However, species within the Sphagnum genus differ in their engineering ability, e.g. S. fuscum (Heinselman 1970; Payette 1988; Kuhry et al. 1993), and S. austinii (Hughes & Dumayne-Peaty 2002; Hughes et al. 2008) are reported as EEs. But species in New Zealand appear to have no role in engineering the FBT, despite their apparently high productivity in comparison with northern hemisphere species (Buxton et al. 1996; Gunnarsson 2005). Though Cockayne (1967) notes Sphagnum “bog” islands growing in Typha-Phormium swamp in lowland coastal North Island, this is probably Sphagnum falcatulum, a well-known swamp plant, but not a bog builder. A small montane mire possibly approaching the FBT, and reported by Walker et al. (2001) to have a peat profile composed of Sphagnum cristatum (other species being apparently unable to tolerate the fire regime), is the most plausible recorded engineering instance by New Zealand Sphagna.

Table 3  Characters required for a species to persist on the bog side of the fen–bog transition (FBT). The match of four potential EE taxa is categorized as good (√), bad (X), mixed (±), and unknown (?).

Character	Northern hemisphere Sphagnum spp.	Eriophorum vaginatum	Empodisma minus	Chionochloa rubra
EE dominance
Competitive?	√ Dominant (Moore & Bellamy 1974)	X Increased relative humidity under canopy facilitates establishment of moss species (Lavoie et al. 2005) Only temporary on raised bogs (Hughes & Dumayne-Peaty 2002)	√ Shrubs stunted by low P concentrations on restiad mires (Clarkson et al. 2005) Clonal resprouter (Meney & Pate 1999)	± Tussocks appear stunted in restiad mires (pers. obs.) and string fens (Rapson et al. 2006)
Nutrient access
Water pre-emption	√ Efficient capture of atmospheric water (Lee & Woodin 1988; Aldous 2002) External capture via capillary water flow (Clymo & Hayward 1982) Buoyancy-driven water flow replenishes nutrients (Rappoldt et al. 2003; Adema et al. 2006)	√ Exploitation of subsurface waters via deep roots overcoming nutrient limitation (Chapin et al. 1988)	√ Pre-emption by high stem flow (Agnew et al. 1993) Accesses atmospheric nutrient sources (Clarkson et al. 2009)	± By fog-water interception (Ingraham & Mark 2000)
Deep root systems	X No roots present	?	X Cluster roots retrieve N from only the top 5 cm of substrate (Clarkson et al. 2009)	√ Deep roots (Craine & Lee 2002)
Nutrient uptake
Cation exchange	√ High cation exchange capacity (Clymo 1963; Daniels & Eddy 1985) S. rubellum takes up amino acids as N source (Kielland 1997) Symbiotic methanotrophs oxidize methane for Sphagnum assimilation (Raghoebarsing et al. 2005) Incorporation of nitrate and phosphate rapidly from solutions of unusually low concentration (Clymo & Hayward 1982; Bragazza et al. 2003)	√ Absorbs organic N and has preferential use of amino acids as N source (Schimel & Chapin 1996) Methanotrophs in Eriophorum/Sphagnum dominant community similar to those of ombrotrophic peatlands (Chen et al. 2008) Very efficient in taking up nutrients in luxury nutrient conditions (Shaver et al. 1986) Root surface phosphatases contribute up to 69% of annual P demand (Kroehler & Linkins 1991; Moorhead et al. 1993)	√ Presence of uronic acids suggests uptake capability (Bannister 2000) High efficiency of nutrient uptake across a range of substrate values (Clarkson et al. 2005)	?
Root adaptations	X None	?	√ Scavenges nutrients with negatively geotropic roots and 1 mm root hairs (Campbell 1975, 1981; Agnew et al. 1993)	?
Leaf life span	√ Long leaf longevity (Chapin et al. 1995)	X Short leaf longevity (Jonasson & Chapin 1985)	√ Stems long-lived, dead shoots retained in canopy layer (pers. obs.)	√ Long-lived (Greer 1979)
Nutrient storage organs	X None	√ Seasonal variation in nutrients of tundra species' organs (Chapin 1980) Storage of limiting nutrients (especially P and K) in perennating organs (Goodman & Perkins 1959; Defoliart et al. 1988)	√ Rhizomes to 1 cm thick (pers. obs.)	√ High nutrient concentrations in leaf sheaths (Williams et al. 1978) High carbohydrate storage in stems and leaf bases of congeners (Payton & Brasch 1978)
Nutrient recycling
	√ Nutrient translocation (Rydin & Clymo 1989; Svensson 1995; Aldous 2002) Internal nutrient transport through perforations in stem parenchyma cells (Rydin & Clymo 1989)	√ 90% leaf P and 80% leaf N resorbed (Jonasson & Chapin 1985) High nutrient use efficiency under low nutrient conditions (Thormann and Bayley 1997) Competitive on cold, infertile sites with other sedges (Shaver & Laundre 2003) Vascular sclerenchyma in corm aids in nutrient recovery and redistribution (Cholewa & Griffith 2004).	√ Nutrient uptake occurs in upper 5 cm of root weft (Clarkson et al. 2005), so leachates can be scavenged	?

Another putative EE, a common northern hemisphere mire species, Eriophorum vaginatum (Table 2), is known as a colonizer or pioneer species (Wein & MacLean 1973). It has a number of characters which indicate reasonable engineering ability (5–6 out of 12 positive matches; Table 2). However, it is fairly palatable, and has low acidification capacity and relatively evenly draining peat (which may be mistaken for Sphagnum peat; Barber et al. 2003), reducing its engineering capacity. Additionally it has few characters which are able to stabilize a bog (4/8 characters; Table 3). Though providing habitat for Sphagnum invasion (Lavoie et al. 2005), against which it appears less competitive, Eriophorum is still able to persist in the new ‘pseudo-raised bog’ environment (Hughes & Barber 2004, p. 65; Korhola 1992), supporting a designation as EE. Since it is only a temporary precursor to Sphagnum-dominated raised bog (Hughes & Dumayne-Peaty 2002; Hughes & Barber 2004; McMullen et al. 2004), it acts more as a facilitator (Connell & Slatyer 1977; Callaway 1995). However, the current relative dominance of Sphagnum may be due to the shift to a wetter Holocene climate about 8000 yr BP, changing the balance towards less desiccation-tolerant peat builders (Hughes & Barber 2004). In past warmer, drier climates, changes towards ombrotrophy took place by the growth of sedge-rich communities raising the surface level of the peat above the groundwater table (Barber et al. 2003). In such climatic conditions, Eriophorum may have been more successful as an engineer than it is today.

In the southern hemisphere Empodisma minus is the most likely candidate for EE of the FBT (though two taxa may be involved; B. Clarkson, pers. comm.). A small rhizomatous perennial, Empodisma minus (Hook.f.) Johnson & Cutler grows in seasonally or permanently inundated habitats, mires, wet heathlands (including pakihi; Johnson & Gerbeaux 2004), and riparian zones with peaty soils throughout South-East Australia, Queensland, Tasmania and New Zealand (Campbell 1964, 1981; Johnson & Cutler 1973; Wardle 2002). The Empodisma genus is a part of the Winifredia group of the Australasian Restionaceae (Briggs & Johnson 2002). The name Empodisma refers to the much-branched, dark green, hollow and slender stems (culms), with extremely reduced leaves in whorls, hosting minute flowers. Stems are 12–200 cm long, forming dense, tangled masses (its local name is wirerush), which arise from bract-covered, glabrous dark brown rhizomes, buried up to 25 cm deep (pers. obs.) in the substrate. Empodisma is a strongly competitive clonal resprouter, with infrequent recruitment from seed (Meney & Pate 1999), except following disturbance such as fire (Clarkson 1997). At least in southern mires, Empodisma is associated with deep, well-decomposed peat, low pH, and low Ca: Mg ratios as well as high marine sodium inputs (de Groot 1999).

Empodisma minus has a dimorphic root system, with sturdy roots anchoring the rhizomes which presumably store nutrients and/or carbohydrates. A second root form, the capillaroid root which is covered with “closely crowded persistent root hairs” (Campbell et al. 1995, p. 9), develops in response to low soil nutrient levels, and is negatively geotropic in bogs (Campbell 1981). Capillaroid roots are not unique to Empodisma, being reported in other members of Restionaceae from Australia (e.g. Loxocarya spp. and Calarophus elongatus; Campbell et al. 1995) and South Africa (pers. obs.), though these areas are too dry for organic matter to accumulate (Campbell et al. 1995). In the wetter regimes of New Zealand, Empodisma's capillaroid roots accumulate as peat due to their abundance, capacity for water retention, chemical inertness and resistance to decay (Campbell 1981). In favourable conditions, the roots can intertwine into a dense, felt-like mat which may grow 20–50 mm above the mire surface, and remain live to a depth of 300 mm (Clarkson et al. 2009). These mats form the sole ground cover, building up around the shoots of adjacent plants and engulfing fallen litter (Campbell et al. 1995). The capillaroid roots have high hemicellulose, and low polyphenol and lignin contents, chemistry which in other species is associated with fine or amorphous detritus upon decomposition (following Kuder et al. 1998). The thicker axes of the roots and stems are high in lignin and polyphenols, which increase their resistance to decay by inhibiting microbial activity, and, with possible allelopathic properties, facilitates competitive exclusion of other mire species (Kuder et al. 1998). As a result of these capillaroid root mats, a “vertically displaced feeding root system” (Clarkson et al. 2009, p. 378) forms underneath the restiad canopy. Empodisma functions with low tissue nutrient contents (Agnew et al. 1993; Sharp 1995; Clarkson et al. 2005), and there is only limited colonization of roots by arbuscular mycorrhizae (Clarkson et al. 2005). Instead nutrient uptake from the low-nutrient peat substrate is by these negatively geotropic capillaroid roots and fine root hairs (Campbell 1981), or have limited stemflow (Agnew et al. 1993). The roots intercept nutrient-bearing rainfall via stemflow (Agnew et al. 1993), while other mire species can only access nutrients from within the substrate (Clarkson et al. 2009). The base-exchange capacity of the capillaroid roots is equal to that of a co-occurring New Zealand Sphagnum species (Agnew et al. 1993), suggesting efficient uptake of any intercepted nutrients. Further Bannister (2000) found significantly higher uronic acid levels in Empodisma than four northern hemisphere Sphagnum species, contributing to its high nutrient capture.

Empodisma has a range of characters which results in maintenance of a high and stable water table. Despite being a wetland species, its roots are ‘poorly adapted for growth in anoxic soils’ (Sorrell et al. 2000:682; Johnson & Brook 1998), though the water-holding capacity of the root mats is comparable to that of Sphagnum on a dry weight basis (Campbell 1964, 1981; Agnew et al. 1993). With low cortical porosity and a highly thickened endodermis and stele, the negatively geotropic habit of the capillaroid roots may be an adaptation to avoid anoxia (Sorrell et al. 2000). Further, the high volume of gas-filled Empodisma tissues near the surface may result in peat floatation, another engineering trait which may permit peatland surface oscillations (Fritz et al. 2008), as do occur in mire zones with high Empodisma cover. Campbell & Williamson (1997) indicate Empodisma exerts strong control on evaporative losses from its canopy, the dense shoots acting as a mulch, restricting movement of solar energy and water vapour between the substrate and atmosphere. Dead stems (including suspended litter), comprising up to 60% of the total canopy biomass (Hodges, unpublished data), are retained within the lower canopy. These intercept rainfall, so the moisture content of the lower canopy remains higher for longer than that of the upper canopy (Campbell & Williamson 1997).

Though much less well studied than Sphagnum, Empodisma minus has the expect ed characteristics of an EE of the FBT (Table 2; 11/12 characters), except that there is no information on its possible production of drainage barriers. It establishes early in the fen stage, and becomes dominant in later stages (Shearer 1997; Clarkson 2002; Clarkson et al. 2004). Then Empodisma dominates the water table and nutrient regimes of the mire, until the peat surface is hydrologically isolated, and a bog is formed, which it effectively stabilizes (Table 3; 7/8 characters), due to its unpalatable shoot material (van Rees & Hutson 1983), inert litter (Campbell 1981), and high nutrient retention capabilities. Due to ‘an elegant interaction between its morphology, substrate and rainfall’ (p. 107, Agnew et al. 1993), Empodisma is a world-class ecosystem engineer.

New Zealand's raised restiad mires form in areas with seasonal rainfall deficits, and lower annual rainfall than that required for the development of raised mires in the northern hemisphere (McGlone 2009), which may be why Empodisma is a better engineer here than Sphagnum species. Pollen and macrofossil data from four North Island mires of varying age provide support for an autogenically driven model of mire succession which can operate across a range of different climates (Clarkson et al. 2004). Empodisma may also be more versatile in terms of climate regime under which it can engineer than northern hemisphere Sphagna.

Another possible contender for southern hemisphere EE of the FBT is the red snow tussock, Chionochloa rubra (Tables 2 and 3), which includes four subspecific taxa (Connor 1991). Up to 1.5 m tall, it has a wide edaphic range, and occurred on the often peaty soils of south-eastern and southern New Zealand over far greater areas than today before land clearance for agriculture (Mark & McLennan 2005s ?n adaptive issue if the nutrients have already been harvested. issue if no ers which are able to stabilise a b). It is absent from northern lowland restiad bogs, where its position in the FBT is taken by sedges (e.g. Baumea spp). Chionochloa tends to be replaced in raised bog communities by Empodisma and other bog species, but persists on the more minerotrophic bog margins. Less well studied as an EE than the other species, it has no known characters which contradict assignment as an EE (5/12 in Table 2, and (3–4)/8 in Table 3), but it is not recorded as building peat and is of low productivity in bog situations (pers. obs.).

Empodisma's only mainland New Zealand bog-inhabiting relative, the rare giant restiad (Sporadanthus ferrugineus de Lange, Heenan et B.D. Clarkson, in the Lepyrodia Group of the Restionaceae: Briggs & Johnson 2002), contributes less than Empodisma to peat accumulation in its only locale, the peat domes of the Waikato region (Shearer 1997; Clarkson 2002). It has lower stem flow than Empodisma (Agnew et al. 1993), and is not found in fens, and so cannot be an EE. A recent segregate, Sporadanthus traversii (F. Muell.) F. Muell, is, however, reported as the major peat builder on the botanically related Chatham Islands (McGlone 2002; Clarkson et al. 2004) in the absence of Empodisma, and its shorter, finer, droopier canes may confer some as yet unstudied engineering abilities.

Conclusion

Work by Hughes & Barber (2003, 2004) has been pioneering in consistently addressing the FBT. Practical experimentation, such as that initiated in bogs by Clymo (1965) and Bellamy & Reiley (1967), and more recently demonstrated by Scheffer et al. (2001), Heijmans et al. (2002) and Adema et al. (2006), offers much needed insight into bog mechanics. Here we review the two mechanisms for crossing the FBT which are differentiatable, and encourage determination of the types of ecological, biological, palynological or stratigraphic evidence which can irrefutably distinguish between these causes of change in mire trophic status, perhaps considering the framework proposed in Table 1.

The importance of the EE in crossing the FBT cannot be underestimated. EEs certainly play pivotal roles in mires of both southern and northern hemisphere temperate regions, though these are not necessarily ecologically comparable to mires of other regions (Damman 1995). In New Zealand the apparent EE of the FBT is Empodisma minus. We encourage the identification, following Tables 2 and 3, of EE species in other parts of the world, an area where local knowledge will be crucial. We are undertaking some such work, focusing on Empodisma minus and the other putative engineer, Chionochloa rubra, in the New Zealand context.

Acknowledgements

We thank the Miss E.L. Hellaby Indigenous Grasslands Trust for support for research into the roles of Empodisma and Chionochloa in wet grasslands, and for conference support. The Department of Conservation and numerous landowners kindly allowed access to their mires. The Institute of Natural Resources, Massey University, provided many extra-ordinary resources. Thanks to Bev Clarkson for sharing her wirerush experiences. We also thank Paul Hughes and an anonymous referee for their comments on an earlier version of the paper, and Dame Ella Campbell, who first interested us in Empodisma's bag of tricks.

Dedication: We would like to dedicate this contribution to the memory of a pioneer thinker in restiad wetlands, Dame Ella Orr Campbell DSc, 28 October 1910–24 June 2003 (Rapson 2004)