Summary
Integrated weed management, aims to maintain the weed populations at a manageable level by using a diverse array of tools, i.e. biological, chemical, physical and cultural measures. The goal of this study was to identify the demography of the weed species Echinochloa crus-galli (L.) P. Beauv and the effect of post-dispersal seed predation in maize fields to test the weed control potential of seed predators as a conservational biological control measure. Not only do seed predators consume seed on the soil surface, before seeds burial into the soil; they also reduce the input of newly produced seeds into the seedbank, thus increasing seed mortality. To reach our study goal, three main obstacles were addressed: seed predators behaviour, weed species E. crus-galli demography and seed predators effect on the demography of the weed. Therefore, we tested seven main objectives. First, to identify the behaviour of seed predators, we estimated the level and response of seed predation to different weed seed densities of E. crus-galli in autumn. Second, we tested whether seed predation in autumn is a good estimate of seed predation from seed shed until the following spring (autumn plus winter). The third and fourth objectives aimed to understand E. crus-galli demography without seed predation. Specifically, the third objective tested whether relationship between the number of seeds per panicle dry weight or per panicle length can be used to simplify the measurement of the seed production of E. crus-galli. Furthermore, we included the effect of factors that could influence this relationship, such as the time of seedling emergence, the density of E. crus-galli, the control intensity of other weeds, seed predation and field. The fourth objective tested the role of density-dependent regulation in E. crus-galli demography. The last three objectives addressed the effect of seed predation on the demography of density populations of E. crus-galli: the fifth tested whether seed mortality by seed predation will lower the density of seedlings; the sixth, whether density-dependent seedling mortality and fecundity will compensate for the lower number of seedlings; and the seventh, whether seed predation will not affect the final constant level of seed production per unit area, but the level will vary between fields.

All Objectives were tested in an agroecosystem in north-eastern Germany. Here, two experimental approaches – one short-term and one long-term - were conducted on three minimally tilled maize fields that had a history of three consecutive years of maize. In the short-term experiment, which tested the first two objectives, different densities of seeds of E. crus-galli were applied on seed trays, such that seeds were exposed to seed predators during autumn in 2014 and 2015 (August until corn harvest in September - October). In the long-term experiment, which tested another part of objective two and remaining objectives, different densities of E. crus-galli seeds were applied to plots in autumn 2014. Halve of the plots were enclosed by a plastic frame to prevent the access of seed predators. In the following season in 2015, the number of seedlings, adult plants, and seed production m-2 were determined in the plots. To test the third objective, a few days before maize harvest, all panicles were removed from the fields and dried panicles were weighed and panicle length was measured, and for a subsample of panicles, the number of seeds was counted manually.

For the first two objectives addressing seed predators behaviour, results showed that in autumn 2014, the level of seed predation and the response to seed density differed between fields. In autumn 2015, in the three fields, a high number of seeds were removed via seed predation, when rodents dominated the seed predator assemblage. The response to seed density was density independent, as seed predation during the winter partially resulted in an increased level and a density-independent response to seed density in all fields. Thus, seed predation in autumn does not reflect seed predation from seed shed until the following spring. Analysis to test the third objective to simplify the measurement of seed production in E. crus-galli, showed that panicle dry weight (R2 = 0.92) predicted the number of seeds per panicle better than panicle length (R2 = 0.69). The other tested factors, except for “field” and “seed predation” had no effect on these relationships. The relationships between seed number and panicle dry weight found in this study closely resembled the results found in an earlier study. Based on our findings, we emphasize that both plant traits were appropriate to use in estimating seed production, depending on the users´ demand for precision and available resources for evaluating sustainable weed management strategies. Results on the fourth objectives showed that the demography of E. crus-galli was regulated by density-dependent processes. Density-dependent mortality hampers the seedling emergence, seedling survival and per capita fecundity of E. crus-galli and it results in a constant level of final seed production. Seed predators effect on the demography of E. crus-galli showed that while the number of seedlings was reduced, but density-dependent processes during seedling emergence, seedling survival and fecundity per capita compensated for the losses. Seed production per unit area was stable among all population densities of E. crus-galli. Results on the seventh objective showed that the final seed production was affected by in-field conditions. All initial population densities of E. crus-galli, either in the presence or absence of seed predators, increased from one vegetation period to the next.

In summary, the weed species E. crus-galli demography is regulated by density dependent processes and, thus, seed predation as a single control measure fails to limit the growth of population densities. In combination with other control measures that target the seedling survival and fecundity, however, seed predation may contributes to weed control by lowering the input into the seedbank and, thus, the distribution of the seeds. Furthermore, E. crus-galli is able to compensate weed seedlings that escapes from failed weed management in maize crops. The long-term effect of integrated control measures on the seedbank of E. crus-galli has not been simulation, however, this study provides data on the fully parametrized life-cycle of E. crus-galli that will support future applications to simulate the long-term effect.