Resco de Dios et al. (2014) report that the effects of neighbour grasses on the recruitment of woody seedlings shift from negative, while the grasses are living, to positive, after the grasses are dead. Recruitment neatly parallels trends in soil water, which was reduced beneath live grasses and increased beneath dead grasses, relative to unvegetated plots.
The results were obtained from a 4-yr field experiment in Arizona conducted under large rain-out shelters. Seeds of Prosopis velutina (mesquite) were planted each year in a two-factor experiment (grass neighbours present or absent, precipitation greater or less than average), and the seedlings removed after 10 mo.
The finding of positive effects of dead neighbours contrasts with previous studies, which have generally found net negative effects of dead grass neighbours in semi-arid environments. For example, removing dead grass from Kansas prairie increased productivity (Hulbert 1988). The establishment of annual plants in experimental plots in Washington was significantly decreased by the presence of dead individuals of Poa annua (Bergelson 1990). The negative effect of neighbours on seedling growth in Saskatchewan prairie increased significantly with litter mass but not with standing crop (Wilson 2007). Thus, the finding of positive effects of dead neighbours is a novel result.
This paper builds on observations that neighbour interactions shift from negative to positive as environmental harshness increases (Choler et al. 2001). It adds a new layer of complexity by suggesting that outcomes may also depend on whether neighbour mass is living or dead. Higher soil moisture beneath dead grasses than in plots with bare ground suggests a mechanism whereby grasses facilitate shrub establishment. On the other hand, dead grasses could also have negative effects on soil water via interception of precipitation (Murray 2014), especially because natural rainfall in semi-arid regions is characterized by the dominance of very small events that rarely wet the soil (Sala & Lauenroth 1982). Further, dead grasses can immobilize nitrogen (Hulbert 1988; Köchy & Wilson 1997). Overall, it is not axiomatic that competition stops when neighbours die.
The strengths of the study of Resco de Dios et al. (2014) lie in its being conducted in field soils, presumably realistic, relatively undisturbed and stocked with relevant microflora. Another feature is that sowing was repeated in each of 4 yr. At a minimum, repetition among years provides insights into the generality of a study: was the result from 1 yr peculiar or representative? In this case, the repetition revealed changes in neighbour effects over time. Lastly, seedlings were followed over 10 mo, revealing negative effects on some stages (survival and recruitment) and positive or neutral effects on other stages (emergence). The outcome varied not only with the planting year, but also with target ontogeny.
Opportunities for improvement are present in every study. First, the realism of the neighbourhood in this study is unknown in terms of diversity, longevity and density. The experimental neighbourhood comprised a single species, the introduced perennial grass Heteropogon contortus, which died after 2 yr. I am unfamiliar with successional trajectories that result in the abandonment of occupied space by perennials, and am curious about the generality of this trajectory and the extent to which it is realistic. It is uncertain how the results apply to ecosystems with successional seres characterized by increasing mass, a constant presence of neighbours and species replacement.
Second, results are reported for each of the 4 yr. Relative interaction intensity was negative in the first 2 yr and positive (in some cases) in the last 2 yr. Instead, time could be included as a factor in the statistical analysis (Goldberg & Barton 1992). A significant time effect would explicitly test the hypothesis that interactions vary significantly among years. Hopefully, the growing trend towards a requirement for data to be archived as an accompaniment to publication will allow continued exploration of these kinds of studies by other researchers.
Third, the comparison of interest (live vs dead neighbours) is difficult to disentangle from other factors that vary among years, such as heat, herbivores and soil pathogens. Neither the cover nor mass of live or dead neighbours is reported. The repetition of studies among years is very worthwhile, but elucidation of the mechanisms underlying differences among years may require further experiments.
Lastly, two other factors (a second soil type, a second grass) were established at the same time as the reported experiment (English et al. 2005), but are not included here. Given the novelty of the results, it would be very interesting to know the extent to which they apply to other soil types and neighbors, especially since soil water may be linked to the shift from negative to positive effects, and the other factors (a clay soil, a neighbour species that dries soil faster) clearly influence water.
Overall, Resco et al. raise intriguing ideas. Is there a tipping point from positive to negative effects of dead neighbours, as there may be for live ones (Choler et al. 2001)? Does it depend on environmental harshness, as in the case for live neighbours (Madrigal-González et al. 2012), or on the amount of dead neighbours present (Van Zonneveld et al. 2012), or some combination of the two? Or does it depend on the type of resources (soil water, available nutrients) consumed by the neighbours? Species differences and their effect on decomposition and nutrient cycling are receiving ever-more attention (Freschet et al. 2013). This paper reports yet another link between ecosystem-level and community-level ecology. Resco et al. are careful to set their study in the context of woody plant invasion into grassland, but their results may well have broader implications for other systems and vegetation types.