Niche theory predicts that when two species exhibit major niche overlap, one will eventually be eliminated through competitive exclusion. Thus, some degree of niche specialization is required to facilitate coexistence. We examined whether two important seed bank pathogens on the invasive winter annual grass Bromus tectorum (cheatgrass, downy brome) exhibit niche specialization. These pathogens utilize seed resources in complementary ways. Pyrenophora semeniperda is specialized to attack dormant seeds. It penetrates directly through the seed coverings. Hyphae ramify first through the endosperm and then throughout the seed. Seed death results as the embryo is consumed. In contrast, the Fusarium seed rot pathogen (Fusarium sp.) is specialized to attack non-dormant seeds in the early stages of germination. It cannot penetrate seed coverings directly. Instead, it responds to a cue emanating from the radicle end with directional hyphal growth and subsequent penetration at the point of radicle emergence, causing seed death. Non-dormant seeds usually escape P. semeniperda through germination even if infected because it develops more slowly than Fusarium. When water stress slows non-dormant seed germination, both P. semeniperda and Fusarium can attack and cause seed mortality more effectively. The Fusarium seed rot pathogen can sometimes reach epidemic levels and may result in B. tectorum stand failure (‘die-off’). Stands usually re-establish from the persistent seed bank, but if P. semeniperda has also reached high levels and eliminated the seed bank, a die-off can persist indefinitely.

Temperature and water potential strongly influence seed dormancy status and germination of Bromus tectorum. As seeds of this plant can be killed by the ascomycete fungus Pyrenophora semeniperda, this study was conducted to learn how water potential and temperature influence mortality levels in this pathosystem. Separate experiments were conducted to determine: (1) if P. semeniperda can kill dormant or non-dormant seeds across a range of water potentials (0 to − 2 MPa) at constant temperature (20°C); and (2) how temperature (5–20°C) and duration at reduced water potentials (0–28 days) affect the outcome. When inoculated with the fungus at 20°C, all dormant seeds were killed, but fungal stromata appeared more quickly at higher water potentials. For non-dormant seeds, decreasing water potentials led to reduced germination and greater seed mortality. Results were similar at 10 and 15°C. Incubation at 5°C prevented stromatal development on both non-dormant and dormant seeds regardless of water potential, but when seeds were transferred to 20°C, dormant seeds evidenced high mortality. For non-dormant seeds, exposure to low water potential at 5°C resulted in secondary dormancy and increased seed mortality. Increasing incubation temperature, decreasing water potential and increasing duration at negative water potentials all led to increased mortality for non-dormant seeds. The results are consistent with field observations that pathogen-caused mortality is greatest when dormant seeds imbibe, or when non-dormant seeds experience prolonged or repeated exposure to low water potentials. We propose a conceptual model to explain the annual cycle of interaction in the Bromus tectorum–Pyrenophora semeniperda pathosystem.

The naturally occurring fungal seed pathogen, Pyrenophora semeniperda, reduces the seed bank of Bromus tectorum but the role of plant litter in this seed–pathogen interaction is unexplored. To investigate the direct and indirect effects of litter on this interaction, we first collected field seed-bank samples from low and high Bromus litter patches. From these data, we explored the relationship between litter depth, seed-bank density and seed mortality from P. semeniperda. Second, we manipulated the fungal stages (conidial spores and mycelium) in/on the litter through sterilization techniques, to measure the direct effect of litter on seed death. Third, for indirect effects, we manipulated litter levels and held seed density and inoculum constant to determine whether Bromus litter could modify the seed zone microsites to favour disease. We found that seed-bank samples from high-litter patches contained higher field-killed seed densities compared with low-litter patches, although the percent difference of disease between litter patch types varied among sites and years (e.g. 80% to 46%). In testing the direct effects of litter on the seed–pathogen interaction, we found that litter can act as a direct inoculum source for the pathogen in the early summer but decreases in disease transmission by the following spring when the litter naturally is in contact with seeds. Investigating indirect effects, we found four times as many pathogen-killed seeds in high-litter treatments as compared with low-litter treatments when inoculum loads and seed densities were held constant. In addition, we found that litter influences the seed–pathogen interaction through density-dependent disease transmission. Our findings demonstrate the ecological importance of litter in semi-arid environments as it influences disease levels of a seed pathogen by direct and indirect means.