The plant in the labyrinth: Adaptive growth and branching in heterogeneous environments

Five main phases of landscape degradation revealed by a dynamic mesoscale model analysing the splitting, shrinking, and disappearing of habitat patches

The ecological consequences of habitat loss and fragmentation have been intensively studied on a broad, landscape-wide scale, but have less been investigated on the finer scale of individual habitat patches, especially when considering dynamic turnovers in the habitability of sites. We study changes to individual patches from the perspective of the inhabitant organisms requiring a minimum area for survival. With patches given by contiguous assemblages of discrete habitat sites, the removal of a single site necessarily causes one of the following three elementary local events in the affected patch: splitting into two or more pieces, shrinkage without splitting, or complete disappearance. We investigate the probabilities of these events and the effective size of the habitat removed by them from the population’s living area as the habitat landscape gradually transitions from pristine to totally destroyed. On this basis, we report the following findings. First, we distinguish four transitions delimiting five main phases of landscape degradation: (1) when there is only a little habitat loss, the most frequent event is the shrinkage of the spanning patch; (2) with more habitat loss, splitting becomes significant; (3) splitting peaks; (4) the remaining patches shrink; and (5) finally, they gradually disappear. Second, organisms that require large patches are especially sensitive to phase 3. This phase emerges at a value of habitat loss that is well above the percolation threshold. Third, the effective habitat loss caused by the removal of a single habitat site can be several times higher than the actual habitat loss. For organisms requiring only small patches, this amplification of losses is highest during phase 4 of the landscape degradation, whereas for organisms requiring large patches, it peaks during phase 3.

The plant body as a network of semi-autonomous agents: a review

Plants can solve amazingly difficult tasks while adjusting their growth and development to the environment. They can explore and exploit several resources simultaneously, even when the distributions of these vary in space and time. The systematic study of plant behaviour goes back to Darwin's book The power of movement in plants. Current research has highlighted that modularity is a key to understanding plant behaviour, as the production, functional specialization and death of modules enable the plant to adjust its movement to the environment. The adjustment is assisted by a flow of information and resources among the modules. Experiments have yielded many results about these processes in various plant species. Theoretical research, however, has lagged behind the empirical studies, possibly owing to the lack of a proper modelling framework that could encompass the high number of components and interactions. In this paper, I propose such a framework on the basis of network theory, viewing the plant as a group of connected, semi-autonomous agents. I review some characteristic plant responses to the environment through changing the states of agents and/or links. I also point out some unexplored areas, in which a dialogue between plant science and network theory could be mutually inspiring. This article is part of the theme issue 'Liquid brains, solid brains: How distributed cognitive architectures process information'.

The effect of ramet mortality on clonal plant growth

Clonal plants grow horizontally by producing multiple physiological individuals (ramets). We studied clonal growth in a homogeneous environment using a dynamic spatial model based on a stochastic cellular automaton. We investigated different growth forms from the aspect of ramet mortality. Non-steady-state and quasi-steady-state cases were defined, and we determined the number of steps suitable for making a reliable difference between these two types of cases. This given number of steps was used when testing for the proportion of quasi-steady-state cases in 1000 repetitions. We also tested the efficiency of occupation in these cases. Our expectation was that higher occupation would be associated with lower ramet mortality. The results only partially verified this hypothesis. Though with increasing ramet mortality, the average number of ramets tended to decrease, it was not the lowest ramet mortality that resulted in the highest occupation. Our results showed that very low ramet mortality was unfavourable for the plant, as the spreading front and the area behind this front were so packed that the plant was not able to return and recolonize the vacated sites in the central area. This resulted in a lower proportion of quasi-steady-state cases and lower occupation in these cases. Our results may contribute to a deeper understanding of clonal plant growth and its limiting factors.

Fluctuations of random walks in critical random environments

Percolation networks have been widely used in the description of porous media but are now found to be relevant to understand the motion of particles in cellular membranes or the nucleus of biological cells. Random walks on the infinite cluster at criticality of a percolation network are asymptotically ergodic. On any finite size cluster of the network stationarity is reached at finite times, depending on the cluster's size. Despite of this we here demonstrate by combination of analytical calculations and simulations that at criticality the disorder and cluster size average of the ensemble of clusters leads to a non-vanishing variance of the time averaged mean squared displacement, regardless of the measurement time. Fluctuations of this relevant experimental quantity due to the disorder average of such ensembles are thus persistent and non-negligible. The relevance of our results for single particle tracking analysis in complex and biological systems is discussed.