• Serafini, John
  • Grogan, Paul
  • Aarssen, Lonnie W.



Water availability is a primary regulator of plant productivity and species richness in arid and semi‐arid ecosystems, but its influence in other habitats is much less clear partly because experimental manipulations of water are relatively rare. How important is soil moisture availability relative to other key environmental factors in determining plant species biomass and diversity in temperate mesic grasslands?


An old‐field meadow in southeastern Ontario, Canada.


We established a long‐term factorial field experiment to investigate how variation in soil water (achieved by rainout shelters and water additions), nutrients (NPK fertilizer additions), and above‐ground herbivory (deer fence exclosures), interact as regulators of plant production, species richness, and species composition. Above‐ground biomass for each species was determined in replicate plots (1 m2 = 10) after five years of treatment applications.


The rainout shelters reduced community shoot biomass by 35% relative to ambient (control) plots, with no influence of the fertiliser. To our surprise, however, the fertiliser enhanced community biomass in the ambient and water addition plots to a similar extent. Species richness and Shannon diversity were increased by water addition, but reduced by the rainout shelters and by fertilisation. Total species richness across all replicate plots was 60% higher in the water addition treatment, and almost halved in the water‐reduced treatment. In contrast to the above, the exclosures had negligible impacts on these community variables.


Together, these results suggest that in old‐field meadow grasslands: (a) shoot productivity is primarily limited by soil nutrient supply except under drought conditions when water availability becomes the primary constraint; (b) the two main soil resources for plants — water and nutrients — affect productivity and species richness/composition in fundamentally different ways; and (c) future increases in summer soil aridity and anthropogenic nitrogen deposition may interact to reduce vegetation productivity and diversity in temperate mesic ecosystems.


In 2010, 240 field plots were set up in 16 parallel rows (separated by 8 m), with 15 plots (separated by 7 m) per row, that were grouped into 10 equal‐sized blocks (24 plots per block). Each plot is 1.95 m in diameter (2.95 m2 ), delineating the circular treatment area that was centered over a 1 m × 1 m sampling/harvest area. The plots were randomly assigned to one of 12 treatments representing a factorial combination of 3 watering levels (soil moisture: ambient, reduced, and supplemented) × 2 soil nutrient levels (control and fertilised) × 2 herbivore exclosure levels (control and fenced), with two replicates per treatment designated within each block (i.e., n = 20 replicates per treatment in total; only half of the plots harvested for this study).

The three soil moisture levels were: reduced (using rainout shelters); control (receiving only ambient precipitation); and increased (through regular watering to double ambient precipitation — see details below). Rainout shelters consisted of a 2 m × 2 m peaked wood frame covered with clear polyethylene plastic (0.006 inches thick), positioned 1.85 m above the ground and centered over the circular treatment area (Appendix S2). The plastic was installed near the end of May and maintained in place until early September each year from 2011 to 2015. The water addition treatment was applied weekly each year (beginning in 2011), from the beginning of June to the end of August, using water from a nearby pond that was chemically similar to the region's natural precipitation (data not shown). This water was pumped into a transportable watering tank and delivered individually to each plot through a hand‐held hose and spray nozzle at the rate 50 litres per whole plot treatment area (i.e., 17 L/m2 ) per week (Appendix S2). The latter rate approximates the average local rainfall received during the three‐month (June– August) period across the years 2000–2010 based on climate data from Environment Canada's Kingston Pumping weather station for 2000–2007 and Kingston Climate weather station for 2009–2010 (Appendix S3). Accordingly, the total water (added plus natural ambient precipitation) received by the water addition treatment plots was approximately double that normally (on average) received by ambient precipitation only. Analyses of simultaneous light intensity and air and soil temperature recordings in the rainout shelters and nearby controls indicated that effects of the shelter apparatus on plot microclimate were minimal (Appendix S4).

The nutrient addition plots received 71.4 g/m2 of 14:13:13 NPK slow‐release (to restrict leachate and run‐off losses) fertilizer (Nutricote 14‐13‐13 Type 100; Plant Products, Arysta Life Science America, New York, NY, USA) near the end of May in each year beginning in 2010. These additions are equivalent to ~10 g N, P2O5 and K2O per m2 — annual fertilisation levels that are widely used to remove any potential limitation by those nutrients of plant growth.

For the large herbivore exclosure treatment, permanent wire fencing (122 cm high with 10 cm × 5 cm opening grid) — effective for excluding white‐tailed deer, rabbits, and groundhogs in particular, but also raccoons, porcupines, and skunks (less commonly found locally in old‐field meadows) — was installed around the perimeter of the 2.95 m2 circular treatment areas containing the plots (Appendix S2). Including the exclosure treatment therefore enabled not only a novel test of the direct impact (if any) of these mammals on the plant community biomass, but also allowed control for an important potential confound in the interpretation of our water and nutrient manipulation treatments. In other words, the latter — which we fully expected to stimulate plant growth — might in turn be expected to affect the extent to which resident herbivores (particularly deer and rabbits) may be attracted to graze on these particular plots (and thus limit the standing plant biomass recorded).