Fairy circles (FCs) are remarkably circular and regular vegetation patterns covering an area of thousands of square kilometres in hyper-arid grasslands in Southern Africa (e.g. Namibia)1,2,3. They are typically made up of perennial tufts of Stipagrostis grass growing around barren circular patches roughly 2–10 m in diameter, with annual or perennial grasses in the area between circles, the matrix1,2. The edges of FCs are about 5–10 m apart and when dense they have a regular hexagonal arrangement3. Similar vegetation rings also occur in arid lands in Australia, Israel and elsewhere1,2,3,4,5,6,7. The processes leading to the formation of FCs have been the subject of debate for several decades8,9,10,11,12,13.
Aspects of FCs that require explanation include how and why individual circles form, how they persist and how the similar size and regular pattern is established at the landscape scale. The bare centres of individual FCs persist in situ almost permanently, although the peripheral plants delimiting the circles are much shorter-lived. Van Rooyen et al.8 noted no change in the position of five marked FCs over 22 years, despite intervening droughts8. Tschinkel14 analysed pairs of aerial photographs taken 4 years apart, and on the basis of how many FC positions were unchanged, died or emerged, estimated that average sized FCs persist in situ for an average of 75 years, implying some circles remain in situ for a century14. Similarly, using aerial photographs, Juergens2 noted a high survival (97%) of FCs in situ over a 50-year period (=0.06% pa mortality), implying an age of millennia for some FCs2.
Many alternative hypotheses have been proposed for FC spatial and temporal patterns, but without agreement. The first hypothesis is based on ecosystem engineering by termites that remove plants from the centres of circles2, facilitating localized underground water accumulation in circle centres. This moisture maintains the termites and the band of perennial grass on which termites feed year-round. The spatial patterning of the FCs is considered to result from competition between termite colonies2. However, the poor correlation of FCs with the presence of specific termites is an important concern with this hypothesis13.
The second hypothesis is based on the clonal mode of growth of individuals of many arid-land species that create vegetation rings4,6,7,15. For example, rings are formed by one of the Namibian FC species, Stipagrostis ciliata in the Negev desert. Here individual plants of this species send out underground rhizomes, which, with increasing age, results in a ring of ramets (i.e. sprouts from the same clonal colony or genet) around a barren central patch, which forms as the plant centre dies4. Such vegetation rings form and enlarge centrifugally due to competition between ramets, and as this process continues over time new ramets establish successfully only towards the periphery4,7. Globally, all of the many plant species that form circles do so by this type of clonal growth15. As the pattern of FCs is virtually fixed in situ for centuries, this suggests that the plants that indicate the pattern may also be long-lived. Individual clonal plants can be extremely long-lived15,16, which could then match longevities between the plants and the FCs they delimit. The clonality hypothesis could thus explain how circular shapes form (by self-thinning of ramets spreading from source plant), why bare centres occur (resource depletion and source plant death) and how they can persist over long periods of time (by continuous production of short-lived ramets). However, clonality has been disputed8 as an explanation for FCs for two reasons. Firstly, van Rooyen et al. suggested that the FCs referred to by Danin and Orshan are considerably smaller (about 2 m) than most FCs in Namibia4,8. Although this is correct that FCs tend to be much larger in Namibia, the mean size of FCs can be as small as 2.5 m in some areas2. Secondly, van Rooyen and colleagues suggested that clonality cannot explain why FC centres are bare8. Bare centres are now well known in rings and are commonly explained as being due to inter-ramet competition and resource depletion7.
The third hypothesis for FC spatial patterns is that it emerges through vegetation self-organization (the VSO hypothesis)1,5,12,17,18. Grasses in the peripheral band outcompete grasses in the FC centre and keep it bare and moist at depth1. The plants in the peripheral bands also compete with the matrix grasses and with the peripheral bands on adjacent FCs to maintain the regular pattern1. Mathematical vegetation models based on partial differential equations represent the theoretical basis of the VSO hypothesis17,18. These partial differential equations do not operate at the scale of individual-plant interactions, but at the level of local processes (largely rates of lateral water movement due to plant evapo-transpiration) in comparison to rates of lateral biomass spread. For these current VSO models, lateral water flow needs to be about 100 times faster than lateral biomass spread18. Thus, the mode and rate of how biomass spreads, whether via clonal growth or seed dispersal and population growth18, has an impact on the viability of vegetation-patterning models. In the absence of such information for FCs, the most recent VSO modelling study18 assumed a rate of biomass dispersal (1.2 m2 yr−1) derived from the Canadian woodlands19, for clonal spread or seed dispersal. These literature values may be unrealistic for FCs. For example, 1.2 m2 yr−1 is likely to be too high for clonal growth in the arid circumstances in which FCs occur. Stipagrostis individual plants across the landscape are typically only 0.005–0.13 m2 in canopy area with a mean area of 0.05 m2 (ref. 20). Similarly, clonal spread in other systems can also be lower than 1.2 m2 yr−1 by orders of magnitude (ref. 19). Alternatively, if Stipagrostis plants are not clones, their highly awned seeds will disperse much further than 1.2 m2 yr−1. Finally, if the assumed 1.2 m2 yr−1 biomass spread18 is due to the total growth of new recruits per parent individual, then this implies unrealistically high population growth rates (>20 new recruits, each with 0.05 m2 in canopy volume20 required per parent individual). Getzin et al.3 acknowledge that in the context of their VSO model it needs to be further investigated whether grass tufts experience a central dieback due to self‐thinning, i.e. the role of clonality33. Thus, clonality may be relevant to some VSO models.
Juergens has argued that there is a spatial mismatch in the root length and inter-FC distances10 and therefore that FCs cannot directly interact with each other to produce the regular spatial pattern. An extreme example of this mismatch is a study by Ravi et al. reporting that the roots of peripheral plants in FCs had a mean length of 5.9 cm, which is more than two orders of magnitude shorter than inter-circle distances (10–20 m)12. These short root lengths would make competition for water and other resources over long distances crucial for explaining FC patterning1. Most recently, Ravi et al.12 rejected the termite hypothesis due to an absence of termites and partially invoked clonal dynamics as well as the VSO to explain their FC patterning12.
In summary, there are critical issues with all the hypotheses explaining the formation of FCs and the role of clonality appears to be relevant to two of the hypotheses. Here, we use ddRAD-seq as a genetic test of clonality of peripheral grasses. Our analysis indicates that most individual grasses surrounding FCs are genetically distinct and does not support the clonality hypothesis.
Source: Ecology - nature.com
