New studies on fairy circles need to account for new observations on their spatial patterns

Submitted by editor on 28 January 2015. Get the paper!

 by Stephan Getzin and Thorsten Wiegand


The fairy circles of Namibia are currently subject to a lively debate on their origin, and they provide a textbook example of how progress is made in science. Hypotheses that were commonly accepted are challenged by a set of new observations that favor alternative explanations, and this forces the supporters of the earlier hypotheses to come up with an explanation for the new set of observations. In many cases, however, such attempts fail and the new hypothesis becomes eventually widely accepted.

                      With our new analysis (1) on the spatial patterns of fairy circles (selected as Editor’s Choice in Ecography), we have provided a new set of observations that challenge common ideas on their origin such as the termite hypothesis (2) or the hypothesis on a geochemical origin (3). Using advanced spatial point pattern analyses of aerial images, we found that the typical fairy circle pattern is characterized by very specific features, including an extremely regular distribution where hexagonal spacing spreads homogeneously over hundreds and thousands of meters. In our article we argue that such a unique spatial pattern, showing all the features observed in our study, is not known from any published study on termite or geochemical hydrocarbon distributions. Therefore, we cannot judge these two hypotheses as the likely cause of the fairy circles.

                      It is now widely accepted in modern ecology that spatial patterns of objects, plants, or animals contain a signature of the processes that have formed those patterns. For example, a clonally spreading fern will cause an aggregated growth pattern, and numerous studies have shown that insects colonize a landscape in a heterogeneously distributed pattern due to inherent population dynamics and preferential habitat choice. With sophisticated methods of spatial statistics (4), such as the scale-dependent pair-correlation function, we are now able to extract and describe the features of spatial patterns in great detail. From this perspective, which looks at the inherent “fingerprint” of a pattern, we can exclude processes that are unable to form the observed patterns, or we can argue that certain processes are able to form the observed spacing of objects. Hence, any new study that presents a hypothesis on the origin of fairy circles needs to demonstrate how this hypothesis can explain the complete set of characteristics of the spatial pattern we have identified in our study.

                      While studying in Namibia as an undergraduate, the first author of this blog visited the field several times and became familiar with the fascinating phenomenon of fairy circles. This experience resulted in a publication that favoured the termite hypothesis (5). Indeed, harvester termites were found feeding on grass tussocks within the fairy circle landscape, but was this narrow-scale observation really sufficient to explain that landscape scale? The spatial patterns of fairy circles likely contain crucial information that is essential to understand the underlying mechanisms that are responsible for their distinct pattern at the landscape scale. Of course, the fairy circles constitute a textbook example of a spatial pattern that calls for analysis with new, sophisticated methods!

                      Until recently, the long-known phenomenon of self-organization has been brought in context with fairy circles, but so far only vaguely (6, 7, 8). Self-organized pattern formation is a pervasive phenomenon in nature that can explain, for example, sand ripples in deserts or banded vegetation in water-limited environments. Underlying these phenomena are positive feedbacks operating at small scales that destabilize uniform states and lead to large-scale periodic patterns of a characteristic wavelength. Thus, an inherent feature of self-organization is that the interaction that causes the positive feedback operates usually over much smaller spatial scales than the emerging patterns. This mechanism is well understood in physics, but causes occasionally confusion among ecologists because it does not require that the roots of the plants reach under the bare vegetation-free patches to drain them. Instead the feedback can work based on laterally confined roots that deplete the water content in the vegetation matrix and induce hydraulic conduction from the vegetation-free gaps where the below-ground water content is higher.

                      Self-organized vegetation patterns do not occur everywhere, but are favored by certain climatic and geological/topographic constraints and are therefore well suited to be investigated within the methodological approaches of biogeography. In 2008, Deblauwe et al. (9) published such a study where they used data on the global occurrence of known periodic vegetation patterns together with data on environmental factors (e.g. climatic data and topography). This information allowed them to understand the environmental drivers of the world-wide distribution of periodic vegetation patterns. Indeed, the periodic patterned vegetation zones turned out to be omnipresent at the interface between arid and semi-arid regions and the predictors of the patterns are related to climate and/or topography (i.e. they included a humidity index, temperature seasonality, wilting point, mean annual precipitation, precipitation seasonality, slope, etc.). Interestingly, their model also correctly predicts the occurrence of periodic vegetation patterns in Namibia within the accuracy limits of the global data. This strong result suggests that fairy circles may be a climate-driven phenomenon. With this background a natural idea was for us to compare the observed fairy circle patterns of several aerial images with that predicted by self-organization models adapted to the specific conditions of Namibia.

                      While the global study by Deblauwe et al. (9) delimited the coarse areas suitable for self-organized patterns in Namibia, a field study by Cramer and Barger (10) published in 2013 revealed the more specific conditions required for the emergence of fairy circles on a regional scale. They could show that fairy circles indeed exist within a narrow range of climatic conditions. Their habitat model shows that the occurrence of fairy circles is strongly driven by mean annual precipitation and temperature seasonality and a measure of vegetation biomass. Interestingly, they also showed that the narrow band of sites with fairy circles is embedded within the much wider distribution of the component grass species and that fairy-circle size is inversely correlated to mean annual precipitation.

                       However, one month after we submitted our paper in 2013, a new study (2) appeared in Science that revitalized the termite hypothesis based on two observations: the sand termite (Psammotermes allocerus) was found in high frequencies at all fairy circle study areas and the circles contain in their center a perennial water reservoir.

                      The debate on the origin of fairy circles has gained particular momentum when the support for the termite hypothesis in Science (2) was immediately questioned by a number of insect or fairy circle experts such as Walter Tschinkel, Michael Cramer, Carl Albrecht or Vivienne Uys.

For example, Vivienne Uys stated in the commentaries of the Science article, “The link between foraging activity of the termite resulting in the formation of a perfect circle of bare soil is unclear.” Tschinkel agreed. “Juergens has made the common scientific error of confusing correlation - even very strong correlation - with causation.” These criticisms were repeated in response to our new Ecography paper in several major media discussions on CNN and BBC.

One of the main criticisms was that Juergens’ study has only shown a correlation of the presence of termites with fairy circles but no actual causation. In fact, the dependency of the sand termite Psammotermes allocerus on deep lying water reservoirs under fairy circles can be questioned based on previous studies on P. allocerus, which were not cited in the Science paper. In 1994, Grube and  Rudolph (11) found, “The only reliable source of water supply for P. allocerus is the water confined in the soil capillary system; to get access to this water the hypopharyngeal surface is firmly pressed to the soil surface.” This means that the sand termite of the Namib is obviously primarily dependent on water near the soil surface but not on the spatial distribution of fairy circles, with their underground water reservoirs, and it may explain why Vlieghe et al. (12) found in 2015 that “sand termites occur in relatively high numbers in both areas,” the matrix and at the periphery of fairy circles.

                      Now, a brief "News and Views" (13) on our Ecography article tries to defend the original termite hypothesis. Unfortunately, the key arguments and statements are based on severe misunderstandings on the nature of self-organization and spatial pattern analysis, and the commentary excludes a number of recent articles that would immediately question its interpretation of the facts it provides. In the following we try to respond to the major misunderstandings and omissions.

                      First, the "News and Views" claims that the termite species Microhodotermes viator would be known to cause the large mounds (heuweltjies) in South Africa and takes this as evidence that this termite would create similar patterns as observed for the fairy circles. However, latest studies on these heuweltjies by two independent research groups (14, 15, 16), provide strong evidence that these mounds are not caused by termites! In 2014, Cramer and  Barger (15) even provide a table with eleven mound characteristics and concluded that vegetation-erosion processes provide by far most evidence to form these characteristics. Hence, as for the fairy circles, yes there is correlation between termites and heuweltjies but there is no evidence that the termites could create these landscape structures (16).

                      Another misunderstanding of the "News and Views" is related with the emergence of underground water reservoirs in a vegetation self-organization context. The author asks, “It needs to be explained why competition should evoke the formation of bare patches in exactly that location that stores more water than found anywhere else in a desert habitat.” However, our Ecography article (1) provided a clear explanation on the underlying mechanism, and a subsequent article by Kinast et al. (17) published in 2014 showed in detail how the different feedbacks incorporated in the model, and also used in our study, can generate these patterns. In particular, they also showed (their Fig. 1B) that a feedback of laterally confined roots depletes the water content in the vegetation matrix and induces hydraulic conduction from the bare gaps where the water content is higher. This feedback of laterally confined roots leads to fairy circle patterns where soil water in the bare soil is higher than soil water in the vegetation matrix just as observed in several field studies. Additionally, the notion in the "News and Views" (13) that, “The matrix is uniformly dry,” is not supported by the results of the Science paper because soil moisture at continuous locations over the matrix has not been measured (and neither rates of water flow). In contrast, aerial images show that the matrix is not uniformly dry but contains gradients of green vegetation towards those places in the matrix, which are furthest away from the circles. These locations are on average hexagonally spaced and will eventually lead to a fairy circle. This can be seen on the photograph provided here.


Figure 1. This aerial image shows that there are positive short-range feedbacks of vegetation growth, leading to green biomass accumulation around fairy circles while these eco-hydrological feedbacks cause water depletion at larger distances away from the circles. In agreement with vegetation self-organization, this spatial gradient in water availability will lead to the emergence of new fairy circles in-between older ones. (Image: © Michael Fay, National Geographic Creative)


Another common misunderstanding of vegetation self-organization is the idea that the characteristic scales of the process causing a pattern should be identical with the characteristic scales of the pattern (see e.g. Juergens 2015) (13). However, by definition, self-organization is a phenomenon in which a global pattern emerges from local processes via positive feedback. This means that it is not the competition between individual grass plants that leads to gaps at larger scales, but it is the joint competition of a number of grass plants for water that lowers the amount of available soil moisture at places further away, where less competitive plants live. 

With our Ecography article and a more theoretically driven approach we have shown that no mechanism other than vegetation self-organization is currently known to be able to create the spatial patterns that agree in all features with that of the of fairy circles revealed in our study. Our conclusion is also supported by a recent field study of Cramer and Barger (10) who showed that the occurrence of fairy circles is strongly driven by climate, i.e. mean annual precipitation and temperature seasonality.

                      The sand termites of the genus Psammotermes are abundant in many arid regions of Africa up to Egypt and the species Psammotermes allocerus is not confined to Namibia only but lives in neighboring countries, too. Why would the sand termites undertake grass clearing only within the specific isohyet range (around 100 mm) that is needed for self-organization to work, but not elsewhere within the much wider distribution range of the component grasses?

                      In summary, we conclude that any new study on the origin of fairy circles needs to account for all patterns known, including the new observations on their spatial patterns that we provided in our Ecography article.



1. Getzin, S., Wiegand, K., Wiegand, T., Yizhaq, H., von Hardenberg, J. & Meron, E. (2015) Adopting a spatially explicit perspective to study the mysterious fairy circles of Namibia. Ecography, 38, 1-11.

2. Juergens, N. (2013) The biological underpinnings of Namib Desert fairy circles. Science, 39, 1618-1621.

3. Naude, Y., van Rooyen, M.W. & Rohwer, E.R. (2011) Evidence for geochemical origin of   the mysterious circles in the Pro-Namib desert. Journal of Arid Environments, 75, 446-456.

4. Wiegand, T. & Moloney, K.A. (2014) A Handbook of Spatial Point Pattern Analysis in Ecology. Chapman and Hall/CRC press, Boca Raton, FL.

5. Becker, T. & Getzin, S. (2000) The fairy circles of Kaokoland (north-west Namiba) – origin, distribution, and characteristics. Basic and Applied Ecology, 1, 149-159.

6. Tlidi, M., Lefever, R. & Vladimirov, A. (2008) On vegetation clustering, localized bare soil   spots and fairy circles. Lecture Notes in Physics, 751, 381-402.

7. Meron, E. (2012) Pattern-formation approach to modelling spatially extended ecosystems. Ecological Modelling, 234, 70-82.

8. Tschinkel, W.R. (2012) The life cycle and life span of Namibian fairy circles. PLoS One, 7, e38056.

9.  Deblauwe, V., Barbier, N., Couteron, P., Lejeune, O. & Bogaert, J. (2008) The global biogeography of semiarid periodic vegetation patterns. Global Ecology and Biogeography, 17, 715-723.

10.  Cramer, M.D. & Barger, N.N. (2013) Are Namibian ‘fairy circles’ the consequence of self-organizing spatial vegetation patterning? PLoS One, 8, e70876.

11. Grube, S. & Rudolph, D. (1995) Termites in arid environments: The waterbalance of Psammotermes allocerus Silvestri. Mitteilungen der Deutschen Gesellschaft fuer Allgemeine und Angewandte Entomologie, 10, 665-668.

12. Vlieghe, K., Picker, M.D., Ross-Gillespie, V. & Erni, B. (2015) Herbivory by subterranean termite colonies and the development of fairy circles in SW Namibia. Ecological Entomology, 40, 42-49.

13. Juergens, N. (2015) Exploring common ground for different hypotheses on Namib fairy circles. Ecography, 38, 12-14.

14. Cramer, M.D., Innes, S.N., Midgley, J.J. (2012) Hard evidence that heuweltjie earth mounds are relictual features produced by differential erosion. Palaeogeography, Palaeoclimatology, Palaeoecology, 350-352, 189-197.

15. Cramer, M.D. & Barger, N.N. (2014) Are mima-like mounds the consequence of long-term stability of vegetation spatial patterning? Palaeogeography, Palaeoclimatology, Palaeoecology, 409, 72-83.

16. McAuliffe, J.R., Hoffman, M.T., McFadden, L.D. & King, M.P. (2014) Role of aeolian sediment accretion in the formation of heuweltjie earth mounds, western South Africa. Earth Surface Processes and Landforms, 39, 1900-1912.

17. Kinast, S., Zelnik, Y.R., Bel, G. & Meron, E. (2014) Interplay between Turing mechanisms can increase pattern diversity. Physical Review Letters, 112, 078701.