Dr. Stephan Getzin | Scientist | Ecologist

True fairy circles only exist in the Namib and in Western Australia

In September 2023 appeared a new study in PNAS by Guirado et al. which showed that the global occurrence of circular vegetation gaps is associated with soil and climatic conditions. The authors used modern methodology such as deep learning algorithms and remote sensing to search for fairy-circle related vegetation gap patterns worldwide. They found these vegetation gaps at 263 sites in 15 countries across the three continents Africa, Asia and Australia. Guirado et al. also identified the main drivers of these gaps which mainly occur in grassland vegetation. According to their predictive model, these gaps are mainly found on sandy soils with a very low soil moisture and a low nutrient content, and in hot regions that are characterized by a high precipitation seasonality and a low mean annual precipitation ranging between 100 and 300 mm. Furthermore, the authors state that both termites and ants had a low importance as predictors of “FC-like” (fairy-circle like) vegetation patterns at a global scale. Overall, this study is interesting as it shows that fairy-circle related vegetation gaps are mainly driven by abiotic factors such as aridity and soil conditions. The conclusion is also plausible that these vegetation gaps should become more common under climate change because increasing drought and aridification should lead to less continuous vegetation cover and thus bare-soil gaps will more and more appear. This is also the reason why true fairy circles in Namibia and Western Australia are strongly confined to a narrow climatic envelope. For example, fairy circles start only to appear along the Namib when mean annual rainfall declines to less than 150 mm. But above that threshold value, there is enough moisture to allow a continuous vegetation cover.

The new paper, however, uses the term “fairy circles” in a way that it dilutes the original concept of that terminology. The authors imply with their title “The global biogeography and environmental drivers of fairy circles” that they would have found true fairy circles in 15 countries, which is not really the case because their found circles do not match the definition criteria of true fairy circles. Also, fieldwork has not been done to verify on the ground that the found gaps are indeed fairy circles. Another inconsistency is revealed by the fact that the authors use the term “fairy circles” only in the title but then they use the term “FC-like” throughout the article. There is certainly no copyright on the term “fairy circles” and in the recent past scientists have increasingly used that term for vegetation gaps which have very little in common with fairy circles. However, this distracts from an incisive discussion of the topic, which is about identifying the underlying mechanisms responsible for the formation of true fairy circles. For this reason, we have written a paper on the Definition of fairy circles, which identified three defining characteristics of genuine fairy circles. This paper won the Journal of Vegetation Science Editors’ Award in 2021 for its efforts to foster a precise and concise discussion of the fairy-circle topic. We wrote that fairy circles are defined by: (a) being “empty gaps” in grassland without a central insect-nest structure; (b) their ability to form spatially periodic patterns, which are regular hexagonal patterns with an extraordinary degree of spatial ordering; and (c) their strongly regional distribution confined within a narrow arid climatic envelope.

Guirado et al. refer to our definition paper: They followed our definition with regard to (a) because they discarded all automatically detected vegetation gaps from America which showed a central insect-nest structure in the satellite images. According to them, these are “samples of false positives (misinterpretation of fairy circle-like vegetation patterns) in the classification results” because “correctly classified FC-like vegetation patterns” show an “absence of shading to discard structures such as termite mounds”. Here we agree because the very common vegetation gaps that are found, for example, in North America are caused by harvester ants. Hence the causal process is obviously due to the disc-clearing behavior of harvester ants, and that’s why these gaps have never been called fairy circles before.

When it comes to the interpretation of spatial patterns, the study does not account for the fact that genuine fairy circles have a very unique spatial signature. That is (b), their ability to form spatially periodic patterns, which are regular hexagonal patterns with an extraordinary degree of spatial ordering. The importance of the unique spatial patterns of fairy circles has been emphasized by us in various papers before, as well as in a blog at Ecography: New studies on fairy circles need to account for new observations on their spatial patterns. Guirado et al. used the same spatial metrics that we used to describe the spatial patterns of fairy circles but they only selected fairy-circle patterns that are merely regularly distributed but that are not spatially periodic. The bare gaps of fairy circles were recognized as early as the 1970s, but the term “fairy circles” was first applied in the year 2000 to the highly ordered fairy circles of the Giribes Plains and Marienfluss Valley of northern Namibia, and later also to the similarly highly ordered circles in southern Namibia and the Wolwedans area. In 2014, new fairy circles were discovered by us in a small area in Western Australia. These typical fairy circles are distinguished by spatially periodic vegetation-gap patterns, which are not merely regular, but are significantly more ordered. For the most part, their defining signature is a sinusoidal fluctuation of the so-called pair-correlation function g(r) around the random null model (Fig. 1B). This pair-correlation function is the most powerful statistic to describe the spatial ordering of a point pattern. Typical fairy circles are additionally described by very high Clark-Evans R-indices (mean = 1.64) (Fig. 1C). It is this extremely high degree of spatial regularity that makes fairy circles so attractive. This spatial property is scientific and verifiable, independent of personal preference or interpretation and has served as the focus for seeking the causes of this phenomenon.

Thus, the term fairy circle should be reserved for gaps that meet these specific criteria. Guirado et al. claim that “FC-like vegetation patterns … resemble the FC spatial patterns … reported in Australia and Namibia”. A general resemblance does not by itself qualify these gaps as fairy circles, for even by Guirado et al.’s own spatial analysis, they are significantly less ordered than true fairy circles. None of the pair-correlation plots of their “FC-like” patterns show the key periodic spatial signature of true fairy circles. The study of Guirado et al. thus shows that despite their intensive global search for other fairy circles, they did not find these beyond the known true fairy circles that can be found in the Namib and in Western Australia. This is a remarkable result because one can assume that many sandy areas that they found in these 15 countries would have very comparable and homogeneous environmental conditions to allow the formation of similarly strongly ordered fairy-circle patterns as are found in the Namib. But while we can find hundreds of thousands of spatially periodic fairy circles along the Namib Desert, the global search using artificial intelligence did not detect any new true fairy circles. The conclusion of this blog is: True fairy circles only exist in the Namib and in Western Australia!

Figure 1

Figure 1. (A) In the exemplary Giribes Plains of northern Namibia, tens of thousands of spatially periodic fairy circles occur across thousands of hectares. (B) The typical signature of the pair-correlation function g(r) of these circles shows a sinusoidal fluctuation around the null model of complete spatial randomness, CSR. This is the most important spatial measure for identifying true fairy circles. The first pronounced peak with a strong amplitude at around 10 m distance indicates the mean nearest‐neighbor distance of the circles. It is immediately followed by a strong negative amplitude below the simulation envelope which indicates the empty space between the first and second row of nearest neighbors, which is only possible because the circles are periodically ordered as a hexagonal grid. None of the FC-like patterns in Guirado et al. show this signature. (C) True fairy circles in eight study plots of Namibia and Australia are additionally characterized by very high R-indices, ranging between 1.6 to 1.7. All the blue dots showing non-fairy-circles are significantly less regular and were classified as “common vegetation gaps” because they were unable to form spatially periodic patterns. (D) Shows Fig. S1 from Guirado et al. Their identified R-indices of “FC-like” vegetation gaps (blue dots) merely range between ca. 1.2 and 1.5. They lack the very high R-indices of 1.6 to 1.7 typical for true fairy circles in northern and southern Namibia (Marienfluss, Giribes, Wolwedans) or Western Australia shown in (C). The red dots in (D) are less ordered fairy-circle patterns of study plots that were affected by environmental heterogeneity and noise, and are not representative of the unique ability of fairy circles to form periodic patterns.

Welcome to FAIRY-CIRCLES.info! I am interested in the ecology of drylands, fairy circles, plant rings and all kinds of spatial vegetation and animal patterns, using a whole range of quantitative methods.