Vegetation ring formation by water overland flow in water‐limited environments: Field measurements and mathematical modelling

Abstract

Vegetation rings are a unique vegetation pattern found in drylands. Most examples are found in clonal plants growing in sandy soils with confined root zones. Using field measurements and numerical simulations, we found that water overland flow is a predominant mechanism that drives ring formation in the clonal species Asphodelus ramosus. In these rings, an infiltration contrast develops due to aeolian feedback between vegetation and wind‐induced particle transport. Fine particles settle at the patch's centre, reducing water infiltrability compared with that of its perimeter. In turn, this encourages the development of biological soil crust in the ring's centre, further decreasing water infiltration in this micro‐environment. These processes cause the development of surface run‐off source–sink relations between the ring's centre and perimeter. The outcome of this process is the formation of three different micro‐environments: the centre of the patch, characterized by low soil‐water content; its perimeter, characterized by higher soil‐water content; and the matrix, also characterized by higher soil‐water content. Competition for the water resource between the ramets at the perimeter and the ones at the centre leads to dieback, resulting in ring formation. Measurements of soil‐water content in rings of A. ramosus in the semiarid Negev of Israel throughout 5 years showed that the soil‐water content in the ring's centre is lower than that at its perimeter and matrix. These results are in accord with numerical simulations of a mathematical model for plant species with confined roots grown under highly seasonal rainfalls in water‐limited environments.

From nonequilibrium initial conditions to steady dryland vegetation patterns: How trajectories matter

“…The driest IHCs ( ) decay into bare states before three years have elapsed. Slightly wetter IHC ( ) evolve into transient ring patterns –supporting other modelling evidence (Bonanomi et al, ; Fernandez‐Oto et al, ; Sheffer et al, ; Tlidi et al, ; Yizhaq et al, )– which slowly break down into spotted, hybrid‐spotted and spiral patterns. In the central cases ( ) the biomass peaks smoothly develop into vegetation spots.…”

“…This theory is extensively supported by observations, mathematical analysis, and direct numerical results, which explain its wide acceptance, although it must be acknowledged that alternative and complementary hypotheses have been proposed, such as solely long‐range vegetation competition (Martínez‐García, Calabrese, Hernández‐García, & López, ) and interactions with fauna (Getzin et al, ; Tarnita et al, ), which remain under intense scientific discussion. Moreover, several feedback mechanisms of soil–vegetation–water interactions have been identified (Meron, ) and are supported by experimental evidence (Dunkerley, 2002jun, ; Ludwig, Wilcox, Breshears, Tongway, & Imeson, ; Tongway, Valentin, & Seghieri, ; Yizhaq, Stavi, Swet, Zaady, & Katra, ). Mathematical and numerical models are a key tool in exploring and understanding the dynamics of these systems (Sherratt, ) as the spatiotemporal scales make field studies and experiments expensive and with limited capabilities of producing insights into the dynamics (Meron, ).…”

“…Simulation duration varies largely among published studies: those based on the Rietkerk model report duration from 8 to 22 years (Rietkerk et al, ; Baudena & Rietkerk, ; Pueyo et al, ; Realpe‐Gomez, Baudena, Galla, McKane, & Rietkerk, ; Thompson et al, ; Thompson & Katul, ). Studies on the basis of other similar models report duration in the range of 37 to 166 years (Gilad et al, ; Gilad, Shachak, & Meron, ; Kletter et al, ; Sheffer, Yizhaq, Gilad, Shachak, & Meron, ; Yizhaq et al, ; Zelnik et al, ) and exceptionally up to 500 years (Baudena et al, ). In some studies, the duration is not reported (e.g., Guttal & Jayaprakash, ; Pueyo et al, ; Roitberg & Shoshany, ).…”

“…These metrics do not require any base assumption on the shapes or periodicity of patterns. We purposely avoid using metrics that assume periodicity (e.g., wavelength and frequency), which are often used to describe periodic patterns (e.g., Deblauwe, Couteron, Lejeune, Bogaert, & Barbier, ; Dagbovie & Sherratt, ; Yizhaq et al, ; Zelnik & Tzuk, ) because the transient states have nonperiodic patterns, which makes their interpretation unclear. Automatic classification of the patterns as bare, spots, labyrinths, and elongated spots or rings was based on circularity and patch perimeter (see supporting information for details).…”

“…Observed patterns have been reported not to coincide with simulated steady‐state patterns (Meron, ). Ring and spiral patterns have been identified as transient (Bonanomi et al, ; Fernandez‐Oto, Escaff, & Cisternas, ; Sheffer, Yizhaq, Shachak, & Meron, ; Tlidi et al, ; Yizhaq et al, ), likely triggered by excitable behaviour (Fernandez‐Oto et al, ; Tlidi et al, ). Discrete rainfall events have been observed to produce transients (Baudena et al, ).…”