Discussion
Bark beetle – tree and beetle- beetle interaction is a complex system, incorporating behavioral decisions of individual agents depending on their environment and decisions of previous agents. As such, it is impossible to study whole ecosystem and its interaction among components studying purely its individual traits or interaction between entities. ABMs afford opportunity to observe behavior and interactions among entities, and by including stochasticity into deterministic process, to observe emergent tree infestation pattern over time and space.
Beetles move randomly over landscapes formed by individual trees and search for suitable host. Firstly emerged beetles move mostly randomly, the next cohort is influenced by the decisions of the previous ones, as the number of beetles staying on the tree increase tree’s attractiveness for the subsequent beetle attack. Positive feedback is amplified. Tree is attractive for beetle and once infested; other beetles are more attracted to this tree to overpass its resistance. Once resistance threshold is overpassed, tree becomes no more attractive for beetle and dies. The decisions of first arrivals influence the subsequent beetle dispersion, aggregation and thus infestation success, and it is path dependent. By repetition of simulations runs (10), the uncertainty of the model is presented.
The emergent pattern is recorded by system response variables as % of beetle success in tree attack, number of infested trees and maximum tree infestation distance from the initial centered beetle spot. By sensitivity analysis, the influence of altering environment as clusters’ extend of weakened trees on these variables is recorded. Non-linearity of the complex system models and autonomous agents’ decisions making lead to not necessarily expected and predictable results.
Following my present knowledge about bark beetle – tree and beetle – beetle interactions, I expected increasing beetle success and tree infestation with increasing clustering level and lowering maximum tree infestation distance.
Results on (Figure 3) not fill these expectations. On the contrary, amount of infested trees decreases and maximum infestation distance grows with higher cluster level. Possible explanation of phenomena (besides eventual bugs in NetLogo coding) may be highly influenced by localization of clump of weakened trees in simulated forest stand (Figure 7). In lower radius values (i.e. r = 2), the weakened trees are distributed more uniformly over forested landscape. With increase of clustering level (i.e. r = 10), the amount of trees in the cluster increase but total number of clusters decreases.
Accordingly, the distances from centered patch from which beetles emerged are higher. Attractiveness of the weakened trees is however so important that beetles infest firstly weakened clustered trees, regardless the distance. Maximum distance of infested tree reached in this study is 500 m. Following these results and applying complexity aspect I can argue against argument of possible beetle flight over several kilometers (Zumr, 1992). This distances can be accomplish by a fragment of beetles, however the beetle aggradation need to overpass tree resistance remains.
The number of infested trees decreases with clustering level. This can be explained by evenly distributed beetles over edges of clusters of weakened trees (Figure 8), thus worst aggregation capacities and beetle mortality caused by starving or by tree resistance.
Figure 7 Comparison of clusters' distribution (dark) and their distances (red lines) from centered patch over simulated forest stand depending on different radius values (r)
In radius = 0, weakened trees are scattered over landscape. Spreading beetles may be are thus able to identify the unique target tree in small time span. In clustered stand, beetles react as individuals, thus randomly select suitable host tree leading to even tree occupancy and lowering beetle aggregation. The preferences of the beetles in tree selection depend on tree’s attractiveness value or inherence of previously arrived beetles will be evaluated in following study.
Figure 8 Beetles distribution on the edge of cluster of weakened trees (dark). Dispersing beetles (red) are searching for suitable tree host; staying beetles (orange) are localized mostly on the edge of the cluster and few inside of cluster. Because of higher availability of weakened trees, beetles are more regularly dispersed over. Aggregation necessary to overpass tree resistance is achieved later, leading to increased beetle mortality.