The effects of leaf litter nutrient pulses on Alliaria petiolata performance

Alliaria petiolata

A. petiolata is an invasive understory plant in the family Brassicaceae that is native to Eurasian temperate forests and is an obligate biennial in North America (Cavers, Heagy & Kokron, 1979; Dhillion & Anderson, 1999). A. petiolata was first recorded in North America in 1868 on Long Island, New York (Nuzzo, 1991). In Virginia, seedlings germinate in March and form a non-reproductive rosette consisting of several small leaves. The plant overwinters as a rosette, growing slowly until early spring when stems elongate rapidly with each individual shoot producing numerous small flowers, each of which produces a silique (Cavers, Heagy & Kokron, 1979). Silique maturation and seed set occurs primarily in June and plants senesce by late summer. Flowers are usually insect pollinated but can be self-pollinated (Anderson, Dhillion & Kelley, 1996; Cruden, McClain & Shrivastava, 1996). Seeds can remain dormant in the seedbank for up to five years (Baskin & Baskin, 1992), allowing A. petiolata to quickly exploit disturbances, especially along forest edges and flood plains (Nuzzo, 1999; Meekins & McCarthy, 2000; Stinson et al., 2006). Allelopathic root exudates may give A. petiolata an advantage over native species (Roberts & Anderson, 2001; Callaway et al., 2008).

Study site

The 17-hectare, second growth woodlot at Blandy Experimental Farm in Boyce, Virginia, USA (39°05′N, 78°03′W) had a well-established population of A. petiolata. This 100-year old forest had a canopy dominated by deciduous species, mainly Carya tomentosa, Celtis occidentalis, Quercus alba, Quercus rubra, Nyssa sylvatica, Prunus serotina and Liriodendron tulipifera. The shrub layer consisted predominantly of Cornus florida, Asimina triloba, Viburnum prunifolium, Sassafras albidum and Lindera benzoin. Several exotic species, including Lonicera maackii and Ligustrum spp., were also abundant. The herbaceous understory contained many perennial spring ephemeral species that perform most of their growth and reproduction between snowmelt and canopy closure (Rothstein, 2000). These spring ephemerals may be affected by shading from A. petiolata—which can reach 1.25 m in height (Cavers, Heagy & Kokron, 1979)—during this time of otherwise high light to the forest understory.

Nutrient manipulation

On 22 October 2006 we established 90 plots, each 0.5 m × 0.94 m, in a natural A. petiolata population. Naturally established A. petiolata rosette density ranged from 8 to 76 rosettes m⁻² within the plots. We surrounded each plot with a 15 cm high barrier using landscaping fabric (2 oz/yd², Greenscapes General Purpose Landscape Fabric, Calhoun, Georgia, USA), to prevent leaves and other debris from blowing onto the plot. Each plot was covered by a mesh net 60 cm above the ground in order to catch falling leaves. We cleared nets weekly to prevent shading and leaching of nutrients from debris and removed nets after leaf fall had ceased in late November. Plots were arranged in 15 spatial blocks, and each of the six treatments described below was randomly assigned to one plot in each block.

The experiment comprised six treatments. Five of these varied in the type and timing of nutrient addition: fall leaves (FL), fall artificial nutrients (FAN), spring leaves (SL), spring artificial nutrients (SAN), no inputs (NI) (Table 1). The sixth treatment (RM) was used as a control to determine the effects of A. petiolata on the abundance and diversity of other understory species. In RM plots we removed all A. petiolata rosettes. Forest floor leaf litter was left undisturbed in FL and RM plots, and litter caught above these plots was distributed evenly across the plot. Supplemental litter was added as required to match the leaf litter depth immediately outside the plot (typically 2.5–5 cm in depth) and was consistent among plots within a spatial block. Replicates of the four treatments that did not receive autumn leaf litter (FAN, SL, SAN and NI) were cleared of leaf litter beginning in October and continuing until leaf fall ceased in late November 2006. In FAN and SAN plots, artificial fertilizer was used to mimic the natural loss of nutrients from leaf litter decomposition and leaching. Artificial nutrients were added to plots to isolate the effects of nutrient addition from other properties of leaf litter that may positively or negatively affect plant growth (e.g., mutualistic or pathogenic bacteria or fungi associated with leaf litter). These nutrients were provided from slow release Osmocote^® fertilizer (13-13-13 NPK, Scotts Company, Marysville, Ohio, USA) at 0.994 g per plot on FAN and SAN plots on 19 November 2006 and 24 March 2007, respectively. This nitrogen addition was similar to the amount (0.283 g N m⁻² yr⁻¹) released from leaf litter in a mixed deciduous forest (Cromack & Monk, 1974). Litter that we collected in the fall 2006 and stored dry over the winter was applied to SL plots on 24 March 2007. NI plots were maintained without leaf litter or nutrient addition throughout the experiment. For each treatment in which leaves were removed (FAN, SAN, SL, and NI), we introduced artificial leaves into the plots in November 2006 to mimic physical properties of litter that may have had both positive (e.g., moisture retention and insulation) and negative (e.g., shading) effects on plant growth. We added ∼20 of these artificial leaves (∼20 × 20 cm squares), cut from landscaping fabric (Greenscapes General Purpose Landscape Fabric) to each plot. As with natural leaves, these artificial leaves were placed haphazardly within each plot. We carefully removed most artificial leaves from SAN plots prior to fertilization, however some leaves were left in place to avoid disturbing the vegetation.

Between our fall and spring nutrient additions, the mean high temperature was 10 °C and the mean low temperature was −2 °C. Over this 125 day period, the daily high temperature was above 0 °C nearly every day (113 times) and above 10 °C nearly half of the days (60 times). Blandy received 55 cm of snow during this period, most of which occurred in small events that melted rapidly and did not cover A. petiolata for long periods. Thus, 60 or more of the days between our fall and spring nutrient additions were likely conducive to at least minimal A. petiolata growth.

Initial measurements

We calculated A. petiolata rosette density on 28 and 29 October 2006, prior to any manipulations. These served as base measurements from which to compare individuals after manipulations.

Adult growth and reproduction

In June 2007 we harvested and separated plants into above-ground (vegetative) and reproductive biomass. Above-ground biomass was oven dried and weighed. A. petiolata plant density and above-ground biomass per plot were calculated. We harvested siliques from adult plants to quantify per capita and per plot silique production.

Competitor richness

To determine the impact of A. petiolata on its competitors we identified all herbaceous species in each plot (Table S1). We quantified species richness and the number of individuals of all species in each plot.

Statistical analyses

We tested all data for normality and homogeneity of variances. Competitor stem density was square root transformed to meet these assumptions. We performed repeated measures ANOVAs using PROC MIXED in SAS (SAS Institute Inc. Cary, NC Version 9.2) on A. petiolata plant density in October and June with nutrient treatment and time as fixed effects and block as a random effect. Autumn A. petiolata density (October, 2006) differed significantly among treatments (P = 0.031), although pairwise comparisons performed with Ryan’s Q failed to detect significant differences among any pairs of treatments. Nonetheless, we retained initial density as a covariate for subsequent analyses. We analyzed above-ground biomass and siliques per plot with ANCOVAs using SAS 9.2 PROC GLM with treatment as a fixed effect, block as a random effect and October A. petiolata density as a covariate. Plot above-ground biomass and siliques per plot were the only response variables for which the covariate effect had significant explanatory value. Because initial density did not contribute significantly to adult plant density or per capita silique production (i.e., ANCOVA was not appropriate), we performed two-way ANOVAs in SAS 9.2 using a general linear model with block as a random effect and nutrient treatment as a fixed effect on garlic mustard density, per capita silique production, and competitor species richness and plant abundance.

We performed five planned contrasts for all responses examining A. petiolata performance that produced a significant treatment effect: Fall leaves vs. Spring leaves; Fall artificial nutrients vs. Spring artificial nutrients; Fall nutrients (FAN and FL) vs. no inputs; Spring nutrients (SAN and SL) vs. no inputs; Fall nutrients (FAN and FL) vs. Spring nutrients (SAN and SL). We maintained a 5% experiment-wise error rate in each analysis with the Dunn-Sidak adjustment (5 contrasts, α′ = 0.01021). In addition to these five contrasts, we performed a sixth contrast for competitor analyses: A. petiolata removal (RM) vs. all other plots. Thus, the adjusted α′ for all competitor contrasts (6 contrasts) was 0.00851. When no contrasts showed significant differences among groups for a given response variable (e.g., silique production per plot), we performed all pairwise comparisons using the Tukey-Kramer test due to unequal sample sizes (data were lost for treatments FL and SL in Block 12).

Above-ground biomass

There was a highly significant difference among treatments in total above-ground biomass of A. petiolata at senescence in June 2007 (Table 2A) when initial density was included as a covariate (P < 0.001). Treatment FL produced significantly more above-ground biomass than SL (Fig. 1). Plants that received fall nutrients (FL and FAN) produced significantly greater above-ground biomass than plots that received either spring nutrient input (SL and SAN). Surprisingly, plots receiving spring nutrient additions had significantly lower A. petiolata biomass than plots that never received nutrients (NI). Total A. petiolata density was not significantly different among treatments (Table 3A) when plots were harvested in June (P = 0.39).

Reproductive output

Plots receiving FL had significantly higher per capita silique production than SAN plots (FL, 30.45 siliques/plant; SAN, 17.50 siliques/plant) (P = 0.009), while SL, FAN and NI plots had approximately equal per capita silique production (Fig. 2). There was a significant treatment effect on silique production per plot when treating initial density as a covariate (P = 0.014) (Table 2B), however none of the planned contrasts showed significant differences among groups (Fig. 3). When examining all pairwise comparisons, FL plots produced significantly more siliques than SAN plots (FL, 510.4 siliques/plot; SAN, 328.1 siliques/plot).

Competitor analyses

We found no treatment effects on competitor species richness in this study. However, competitor abundance (stem density, Fig. 4) was significantly affected by A. petiolata treatment (Table 4); plots from which A. petiolata was removed contained a significantly greater number of stems of competitor species (45.97) than plots receiving spring leaf-litter (24.70). There were no significant differences among any other treatments.