No abstract

Abstract. Soil organic matter (SOM) is key to maintaining soil fertility, mitigating climate change, combatting land degradation, and conserving above-and below-ground biodiversity and associated soil processes and ecosystem services. In order to derive management options for maintaining these essential services provided by soils, policy makers depend on robust, predictive models identifying key drivers of SOM dynamics. Existing SOM models and suggested guidelines for future SOM modelling are defined mostly in terms of plant residue quality and input and microbial decomposition, overlooking the significant regulation provided by soil fauna. The fauna controls almost any aspect of organic matter turnover, foremost by regulating the activity and functional composition of soil microorganisms and their physical-chemical connectivity with soil organic matter. We demonstrate a very strong impact of soil animals on carbon turnover, increasing or decreasing it by several dozen percent, sometimes even turning C sinks into C sources or vice versa. This is demonstrated not only for earthworms and other larger invertebrates but also for smaller fauna such as Collembola. We suggest that inclusion of soil animal activities (plant residue consumption and bioturbation altering the formation, depth, hydraulic properties and physical heterogeneity of soils) can fundamentally affect the predictive outcome of SOM models. Understanding direct and indirect impacts of soil fauna on nutrient availability, carbon sequestration, greenhouse gas emissions and plant growth is key to the understanding of SOM dynamics in the context of global carbon cycling models. We argue that explicit consideration of soil fauna is essential to make realistic modelling predictions on SOM dynamics and to detect expected non-linear responses of SOM dynamics to global change. WePublished by Copernicus Publications on behalf of the European Geosciences Union. 566 J. Filser et al.: Soil fauna: key to new carbon models present a decision framework, to be further developed through the activities of KEYSOM, a European COST Action, for when mechanistic SOM models include soil fauna. The research activities of KEYSOM, such as field experiments and literature reviews, together with dialogue between empiricists and modellers, will inform how this is to be done.

No abstract

No abstract

Ovipositing females of the cynipid waspPseudeucoila bochei discriminate between parasitized and unparasitized hosts, which results in a far more uniform distribution of eggs over the hosts than would be obtained if oviposition were random (Fig. 1,a -f).For the description of the distributions a few models were worked out, which rest on the assumption that the hosts are probed at random. The total number of effective probes made in a larva during the experiment is a random variable with a Poisson distribution and an expectation λ. The chance that at a certain probe an egg will be laid (δ) is dependent on the number of eggs present (j); 1=δ<δ≧δ≧δ.... In model I it was assumed that the female had only the ability to distinguish parasitized from unparasitized hosts. The chance that an egg will be laid in an unparasitized host when it is probed, δ, is considered to be equal to 1, while δ=δ=...=δ <1 (Fig. 2, 1,a -f). When the mean number of eggs present in a host was larger than about 1.1, this model did not describe the distribution of eggs satisfactorily (Fig. 3).It seemed that the ovipositing female is not only able to distinguish parasitized from unparasitized hosts, but also to distinguish thenumber of eggs present in a host. In model II it was assumed that the wasp could distinguish between hosts with 0, 1, and 2 or more eggs: the chance that an egg would be laid in a host containing 2, 3, 4, ... eggs was, hence, the same in this model δ<δ=δ=...=δ (Fig. 2, 1,c,e ,f). This model described the distributions of eggs much better (Figs. 4 and 5), but at mean numbers of eggs per host above 2 it was apparently inadequate.Two other models were then tried, in which the chance δ that an egg would be laid in a host decreased with the number of eggs already present (j). In model III (Fig. 2) the chance decreased according to the function δ =δ/j (δ=1, δ<1). Fig. 1,d ,e,f , gives some examples. In model IV the chance δ =δ (δ=1, δ<1) (see Fig. 1,d ,e,f ).From the comparison of Figs. 6 and 7 it is clear that model IV gives the best description of the distributions of eggs found.The value of these models is discussed, and plans for both an approach through experimental analysis and simulation models are given. In an Appendix the mathematical derivation of the models is presented.