Litter and its decomposition are crucial to terrestrial ecosystem functioning, and are responsible for carbon and nutrient cycles. Litter quality and quantity vary greatly with the changes of climate, soil parent materials, species distributions etc. Plants can regulate their physical and chemical properties to adapt and respond to karst habitats, i.e., shallow soil and drought, thereby produce litter with different quality, quantity and components, thus influence the litter decomposition. Previous studies have shown that the effects of water (or precipitation) and nutrient addition on litter decomposition during the decomposition period. However, the effects of soil thickness, water, and planting patterns during the growing period of plant on litter decomposition are less known. Our aim was to determine whether the decline of soil thickness and water, and different planting patterns during the growing period of two herbs would have effects on litter decomposition that affected by the initial litter quality, quantity, and components. Two perennial graminaceous grass species, Festuca arundinacea Schreb. (deep-rooted) and Lolium perenne L. (shallow-rooted) were chosen as the experimental materials. These two species are often chosen as pioneer plants in karst revegetation schemes. We manipulated three factors, soil thickness (control soil thickness:15 cm, and shallow soil:5 cm), water (normal irrigation:120 mL/3 d and drought treatment:36 mL/3 d), and planting pattern (monoculture and mixture) in a full three-way randomized block design. We collected aboveground and root litter of the plants, and placed these litters in litterbags to decompose in outdoor incubation beds for seven months. The results showed that:(1) compared with the control group (CK), the aboveground, root, and total mass loss rates, initial N and P contents of F. arundinacea and L. perenne increased in drought group (D) and shallow soil plus drought group (SD), and litter production, C/N, and C/P decreased significantly, whereas the aboveground component ratio of both species was not changed. All the above traits were not affected by the planting patterns; (2) the aboveground, root, and total mass loss rates of both species were positively correlated with the aboveground, root and total N contents, negatively correlated with C/N and litter production. The total mass loss rate was positively correlated with the aboveground component ratio for both species. The results indicated that drought and shallow soil plus drought could accelerate the litter mass loss rates of F. arundinacea and L. perenne by changing the initial litter quality, quantity, and components. Moreover, litter N content and C/N were the main factors that influenced the litter decomposition of the two species.