Ant-hill heterogeneity and grassland management

1.Inmanygrasslands,someantsactasecologicalengineerstoproducelong-lastingsoil structures which have a considerable influence on the patterns and dynamics of plant, vertebrate and invertebrate species. They promote species richness and diversity. 2. The yellow meadow ant, Lasius flavus , is the most abundant allogenic ecological engineer in grazed European grasslands, producing vegetated long-lasting mounds. It is so frequent and abundant that it must be regarded as a keystone species. Grassland restoration projects frequently attempt to re-introduce grasslands on abandoned arable fields. Whenthis antdoes notcolonize naturally itshould beintroduced. Itprob-ably limits theabundance of grasses in a similar mannerto hemi-parasitic plant species. 3. Ant-hills make a distinctive contribution to grassland heterogeneity. Measurements on mounds in a single grassland over 45 years document the dynamics of the same 200 + ant-hills in volume, surface area and basal area. As the mounds aged, they increased in size and took over a higher proportion of the grassland surface. Occupied mounds continued to grow, abandoned mounds decreased in volume and some disappeared entirely. 4. Four plant species favoured by the soil heaped by the ants were also monitored. Two woody perennials grew up through heaped soil and two short-lived species col-onized its surface. As the mounds became occupied, some of these species significantly increased, and when they were abandoned some decreased. 5. In a grassland, the ant-hill population provides a fluctuating subset of plant and animal species which are characteristic of temporary habitats. This seems likely to reduce the rate of local extinctions which might otherwise result from fluctuations in grazing pressure. In conservation settings, ant-hills should be introduced or maintained where possible, and considered in planning grassland maintenance and management.

Some ant species, such as Lasius flavus (F.) in grasslands, produce long-lasting mounds which harbour a distinct flora and animal life. They frequently increase the species richness of the habitat, both on the mounds and between them. Their presence is important in the conservation of old grasslands and the creation of new ones (Jouquet, Dauber, Lagerhof, Lavelle, & Lepage, 2006;King, 2006). Their maintenance and long-term survival need to be carefully considered as an integral component of management plans, because they increase the range of micro-environments and may contribute to the ability of the grassland to withstand perturbations.
The yellow meadow ant, L. flavus, is the most important allogenic ecological engineer (sensu Jones, Lawton, & Shachak, 1994), apart from Homo sapiens, over much of its range, which stretches within the Palaearctic to northern Spain, and across most of Europe to China. Its mounds are the most abundant and persistent of all the extended phenotype structures created by organisms in the northern hemisphere.
In grazed grasslands this species can reach a higher fresh mass than any other ant species, 165 kg/ha, and shift up to 7 tonnes of mineral soil/ha/yr (Seifert, 2018). Its mounds sometimes cover over 20% of the ground area (Bushy Park SSSI. London, UK NGR TQ 154 708, grassland 519 years old, 26.33%, unpublished data) and contribute annually most of the bare soil within a grassland habitat. Despite this subterranean ant's reluctance to enter pitfall traps (Fagan, Pywell, Bullock, & Marrs, 2008a), it is likely to greatly outnumber other ant species in most European pastures.
They increase the range of insolation, soil temperature and moisture for mammals, birds, the soil mesofauna and many invertebrates which have the early stages of their life cycles in the soil (Haarlov, 1960: King, 2006. Ant-hills built by L. flavus display a variety of aspects, patches of bare soil ripe for colonization, and dryer soil with a lower content of organic matter, lower bulk density and altered level of mineral nutrients. They provide distinct island microhabitats, supporting small-scale meta-populations of plant species and invertebrates which would otherwise become locally extinct (King, 2006. They impart distinctive qualities to grasslands, providing 'alternative stable states' to grassland soils, fauna and vegetation (Beisner, Haydon, & Cuddington, 2003). The diversity of sizes, shapes and activities within an ant-hill population also contribute to heterogeneity, species richness and alpha diversity. This depends particularly on the grassland's grazing history. Although many static measurements have been published, the ways in which populations of ant-hills are born, grow and die have only sketchily been investigated (Ehrle et al., 2017Nielsen, 1992;Nielsen, Skyberg, & Winther, 1976;Waloff & Blackith, 1962). The implications of the presence of ant-hills for identifying ancient sites likely to be species-rich, and for grassland management, should also be taken into account in conservation practice.
This study encompasses 45 years on the same site, illuminates the fates of the ant-hills and discusses the relevance of ant-hills to grassland conservation. This ant species may be a useful tool in the restoration of typical species-rich grasslands. It tests the hypotheses that 1. as an ant-hill population grows, the mounds come to play an increasingly important role in the grassland, covering a higher proportion of its area, 2. the abandonment of ant-hills and their recolonization may play a significant role in the dynamics of grassland plants. The characteristic ant-hill species should expand when soil heaping by ants occurs and contract after ant colonies abandon a mound.
Four plant species were recorded on each ant-hill as indicators of soil heaping activity by the ants. The dwarf shrubs Thymus drucei ( = T. polytrichus = T. praecox) and Helianthemum nummularium are characteristic ant-hill species which respond to soil heaping by repeatedly branching and growing through the heaped soil (King, 1977c). The winter annual Arenaria serpyllifolia and the pauciennial Cerastium fontanum are largely restricted to ant-hills in continuous calcareous grassland (Dostal, 2007, King, 1977a.

MATERIALS AND METHODS
The progress of a single ant-hill population was measured by the same observer in 1970, 2007 and 2015 in the same calcareous grassland at Beacon Hill, Aston Rowant National Nature Reserve (UK NGR SU727972, 230-244 m altitude) at which the vegetation on the anthills has been investigated (e.g. King, 1977aKing, , 1977bKing, , 1977cKing, , 2007.
This plot (Figure 1) is described in detail in King (1977b). On each ant-hill, two measurements were recorded of its horizontal diameter and four of its height according to established methodology (King, 1977b(King, , 1981a

RESULTS
The

DISCUSSION
Since, in many grasslands over such a wide geographical area, L. flavus ant-hills make a major contribution to plant and animal species richness, this study of the dynamics of an ant-hill population has considerable implications for the conservation of existing grasslands and the establishment of new ones.
Some ant-hills probably continue to increase in volume over ten or hundreds of years (King, 1981a). Particular ant-hills have been followed over 62 years . In this study, as the ant-hill   King (1975King ( , 1977c. The results were consistent with the concept that those dwarf shrubs and short-lived species which are much more abundant on ant-hills than in the surrounding pasture will increase when the ant-hills are occupied by L.flavus and decrease when they are abandoned. It is reasonable to expect that as occupied ant-hills expand in volume and take over an increasing proportion of the grassland area, their populations of other distinctive plants and animal species will increase. Occupied ant-hills tend to be the only grassland sites for winter annuals (Grubb, 1976;Dostal, 2007;King, 1975King, ,1977a and acrocarpous bryophytes (King, 1981a particularly on their south-facing slopes (Streitberger et al., 2017), with pleurocarpous bryophytes on the north (King, 1981b(King, , 2003Streitberger et al., 2017).
Once they are abandoned, grasses, rhizomatous and rosette species invade from the surrounding grassland in concentric zones. The relative abundances of the species on their surfaces change to resemble that of the surrounding grassland (Dostal, Breznova, Kozlickova, Herben, & Kovar, 2005;King, 1977b;Woodell & King, 1991).
The bare soil on ant-hills (2.7% of the ground area in 2015 in the current study), annually replenished, would not otherwise exist on this scale. The bare soil provides oviposition sites for grasshoppers, butterflies and moths (King, 2006, Richards & Waloff, 1954, Streitberger & Fartmann, 2016. Many other invertebrates (listed in ) are confined to ant-hills or owe their presence in the grassland to the ants.
Furthermore, differences between ant-hill soils and the surrounding soils increase with increasing ant-hill volume (Ehrle et al., 2017).
As the mounds grow, their north and south-facing aspects constitute an increasing proportion of the grassland area. Significant (p < 0.05) differences in their soils from the surrounding grassland are welldocumented. For example, their soils have a lower bulk density (in occupied mounds), a lower water content, less organic matter, less total carbon, a higher pH, a greater cation exchange capacity and more exchangeable nitrate, potassium and phosphate ions (e.g. Blomqvist, Olff, Blaauw, Bongers, & van der Putten, 2000;Boots et al., 2012;Dostal et al., 2005;Ehrle et al., 2017). They also differ from the surrounding grassland in their microflora and nematodes (Blomqvist et al., 2000;Boots et al., 2012).
These long-term data from Beacon Hill, Aston Rowant suggests that the mounds, once built, exist as distinct structures which go through phases of abandonment and recolonization (Hypothesis 2) This is to be expected in view of the limited longevity of fertilized queens, and the presence of bare soil on top of the mounds into which newlyfertilized queens can burrow after the nuptial flight (Boomsma, Wright, & Brouwer, 1993). At least in captivity, existing colonies can accept additional queens (Waloff, 1957). Some established mounds have secondary summits, reflecting perhaps two or three successful invasions.
On abandoned ant-hills the vegetation closes over the bare soil and the species composition ultimately resembles that of the surrounding grassland (King, 1977b). Abandoned ant-hills may therefore become less suitable for recolonization with time. These Aston Rowant data illustrate that they may also decline in volume. Ant-hills containing vigorous colonies have a soil bulk density < 0.8 times that of the soil beneath the surrounding grassland (Dostal et al., 2005;Haarlov, 1960;Wells et al., 1976). When an ant-hill is abandoned, its chambers and channels cease to be excavated by the ant-colony and erosion exceeds growth. In abandoned ant-hills, characteristic root aphids and inquilines  should be lacking, soil organic matter, bulk density and the sizes of structural aggregates increase, and exchangeable cations such as K + are lower, than in active mounds (Dostal et al., 2005). Ultimately, the soil may be eroded so much that ant-hills disappear.
These data confirm that there is considerable heterogeneity amongst an ant-hill population. Many mounds may be occupied by colonies, but others will be in various stages of abandonment.
Ultimately, when the ant-hills become large, their summits expand and become closer together, allowing optimal dispersal of the characteristic plant and animal species which form metapopulations on their surfaces. The presence of ant-hills ensures that a site maintains a wider range of plant and animal species than if they are absent (Jouquet et al., 2006).

Established ant-hill populations
Because of the striking mounds it builds, the 'antscape' erected by L. flavus needs to be considered alongside the grazing regime in the conservation of grazed pastures. As the influence of the ant-colonies increases as the underground territories saturate the grassland, L.
flavus becomes a keystone species (Platner, 2006). The recent largescale expansion of abandoned pastures in central Germany (Poschlod & WallisDeVries, 2002) has been accompanied by a wholesale invasion of L. flavus colonies (Streitberger et al., 2017). They used to be characteristic of British grazed grasslands before they were eliminated from pastures in the 'improvements' between 1780 and 1820. The comment by Pitt (1809) is typical: 'ant-hills occupy a large proportion of the grazing land of this county, in some of which ant-hills are so abundant, that it is possible to walk over many acres, step by step, from one ant-hill to another, without coming upon the level ground . . . .' The ant-hills on this site have been growing for at least 70 years.
Sites with large ant-hills are likely to be old grasslands which have accumulated more species (Fagan, Pywell, Bullock, & Marrs, 2008b;Gibson, 2010;Gibson & Brown, 1991;Karlik & Poschlod, 2019;Poschlod & WallisDeVries, 2002;Wagner et al., 2019). Similar ant-hills are prominent features in many sand dunes, salt marshes where the ants can withstand periodic inundation by sea water, freshwater marshes, mountain pastures and acidic grasslands. In many of these areas, they cover 10-20% of the surface, and the dynamic heterogeneity created by the variation in their sizes, aspects, bare soil, degree of abandonment and nutrient accumulation contributes to the range of micro-habitats available for both plants and animals.
Old, well managed, calcareous grasslands are prized and in general, older mature grasslands support more species-rich vegetation and fauna (Gibson, 2010). In some places, this may be partly because the ant-hills are larger and contribute more environmental heterogeneity, as the data in the current paper suggests.
They appear particularly resistant to erosion by normal densities of sheep, cattle deer and rabbits. The number of established colonies stabilizes, as the available territory space becomes fully utilized. Just a few mounds continue to grow rapidly; on the Porton Ranges, the largest ant-hills probably grew at about a litre a year in volume (King, 1981a).
The extent of the influence of the underground L. flavus workers on the flora between the mounds, its energy flow and nutrient cycling, is unknown but is likely to be considerable (Bardgett & Wardle, 2014). Where the density of L. flavus colonies is considerable and the mounds have been established for some time so that the worker populations in the mounds are high, it is even possible that the ants compete effectively with mammalian grazers for plant net photosynthate. After all, the aphid populations on which they rely are clumped on grass roots between the mounds (Ivens, Kronauer, Pen, Weissing, & Boomsma, 2012;Langley, 1986;Pontin, 1978). They may limit grass growth between the mounds in a similar manner to yellow rattle, Rhinanthus minor, frequently sown in the early stages of grassland restoration to limit grass growth and increase the successful establishment of herbs (Pywell et al., 2004).

Ant-hill establishment in restored pastures
It is therefore desirable that grasslands restored from arable should accumulate ant-hills as soon as possible. A dynamic ant-hill population is worth establishing and managing in an active state for its positive influence on the species richness of the grassland. The invasion or re-invasion of many plant species to an isolated reserve or reversion locations may depend on the vagaries of chance dispersal by human agency (Fagan et al., 2008b;Gibson, 1986Gibson, , 2010. Now that suitable donor sites are rarer in the landscape (Ridding, Redhead, & Pywell, 2015), it is much less likely that desirable plant species reach isolated sites, unless sheep are introduced from elsewhere with suitable seeds in their fleeces (Manzano & Malo, 2006;Fischer, Poschlod, & Beinlich, 1996). A study on habitat fragments in Sweden suggested that ants are better dispersed than plant species (Dauber, Bengtsson, & Lenoir, 2006). It is likely that a grazed arable reversion site will accumulate root aphids and L. flavus colonies without help. Vigorous L. flavus colonies produce up to 410 queens a year (Pontin, 1963) dispersed over the local landscape for perhaps up to 5 km. Some specialist root aphids 'farmed' by L. flavus, such as Anoecia corni and Tetraneura ulmi, have winged morphs in their life cycles (Langley, 1986).
These ants sometimes invade early in grassland establishment (Woodell & King, 1991). In Dutch sand dunes and grasslands reverted from arable in Central Germany, L. flavus colonies begin to predominate amongst ant colonies in pastures more than 10-28 years old (Boomsma & van Loon 1982;Dauber & Wolters, 2005). The successful invasion and establishment of L. flavus probably depends to some extent on the prior colonization of the root aphids on which this species largely feeds (Pontin, 1978;Seifert, 2018). Once ant-hills become established they are likely to persist for at least as long as grazing continues, reducing grass growth and representing a continual source of environmental heterogeneity and bare soil.
If newer grasslands are far from source populations, existing anthills may have to be imported to establish this ant in the pasture, for example by the mechanical transfer method suggested by King and Balfour (2020), which is likely to introduce their myrmecophilous aphids at the same time. This technique seems more likely to succeed in establishing some species than artificial methods to create microtopographical heterogeneity (Wagner et al., 2016). Several plant species characteristic of mature calcareous grassland (Wagner et al., 2019) have anthill affinity indices over 60% (King, 1977a;Streitberger et al., 2017): Trisetum flavescens, Galium verum, Thymus drucei, Scabiosa columbaria, Asperula cynanchica, Campanula rotundifolia and Helianthemum nummularium. Establishing ant-hills could not only introduce these species, but winter annuals and essential myrmecophilous aphids, at an earlier stage than otherwise.

Conservation relevance of L. flavus
It is well known amongst reserve managers that in spring ant-hill veg- and cattle for ant-hills comes from a salt marsh off the Dutch coast (Veen et al., 2012) and China, where cattle (Bos taurus) preferred grazing on incipient L. flavus mounds instead of equivalent areas from which the ants had been removed (Li et al., 2018). A high grazing intensity by cattle or sheep during the initial establishment of new grassland is likely to reduce the rate at which vigorous ant-hills appear. Wright (1990) recommends that a sheep grazing intensity of 500-700 sheep grazing days year −1 hectare-1 is likely to achieve the optimal balance between vegetation height and ant-hill establishment.
In mown pastures, L. flavus ant-hills do not establish and increase in size. Once they reach a certain size in a grazed pasture, mowing becomes impossible, so that scything or clipping on and around the anthills, with removal of the clippings, is necessary to maintain the characteristic grassland species composition and prevent the establishment of tall grassland.
L. flavus appears likely to provide 'resilience' (Isaac et al., 2018) to maintain animal and plant species characteristic of the surrounding grassland during periods when grazing is relaxed and the taller vegetation eliminates species from the surrounding grassland. The lack of regular grazing is important in places which are in rotational systems to promote the survival of arthropods characteristic of tall grasslands (Morris, 2000). It may, however, increase the abundance of tall grasses, such as Brachypodium rupestre, B. pinnatum, Dactylis glomerata, Bromopsis erecta (Hawes, 2015) or Molinia caerulea, at the expense of low-growing forbs. This is the main risk to a flourishing ant-hill population. These grass species, especially if allowed to fruit, can invade the mounds, spread rhizomatously, eliminate lower-growing species, and shade out the ant colonies. This reduces the value of the high temperature sites and ant-hill bare soil which seem so valuable for the oviposition of insects such as grasshoppers (Richards & Waloff, 1954) and certain butterflies, reducing environmental heterogeneity. These grasses colonize the ant-hills more slowly than the surrounding pasture. When shaded, the ants heap bare soil more rapidly (as confirmed experimentally by Blomqvist et al., 2000). If grazing resumes in time, for example because of the resumption of a rotational or 'mob' grazing scheme, ant-hills are likely to act as sources of those low-growing grassland species which have been largely eliminated from the pasture.
Once regular grazing resumes, many plant species surviving on the tops of ant-hills are likely to recolonize the grassland between them.
Despite the high proportion of the grassland surface covered by these ant-hills in many National Nature Reserves, Sites of Special Scientific interest and local Nature Reserves in Britain, they are frequently ignored in management plans, despite their distinctive contribution to grassland ecology. The data in this paper from Aston Rowant about their heterogeneity, fluctuations in size and dynamics of their vegetation over 45 years suggest that they add an additional dimension to conservation, and that their importance increases with time.