Decomposition of leaf litter is an important process in terrestrial ecosystems that affects nutrient cycling, including the availability of nutrients to plants (Chapin et al. 2002). In temperate ecosystems, leaves are deposited to the forest floor every fall. Nutrients in the leaf litter of fast-decomposing species can be available for new leaf growth in the next growing season, whereas nutrients in more slowly decomposing species are not released as fast. Because the nutrients released by decomposing leaf litter provide such a large part of nutrient demand by new leaves, decomposition rates are important in regulating a number of important ecosystem functions, including plant growth, nutrient cycling, and net primary production.

The introduction of invasive species can greatly alter nutrient cycling rates, with major effects on ecosystem and plant community processes. In many cases, leaf litter from invasive species decays more quickly (e.g., Ashton et al. 2005, Aragón et al. 2014, Hornbach et al. 2021), although this is not always the case (Jo et al. 2016). For example, the leaves of the nonnative shrub Amur honeysuckle (Lonicera maackii (Rupr.) Herder) decay much more quickly than those of the native woody plants that account for most of the leaf litter inputs in eastern deciduous forests (Blair and Stowasser 2009, Arthur et al. 2012, Poulette and Arthur 2012). This is because the leaf litter of L. maackii generally has higher levels of nitrogen (N), lower carbon (C):N ratios, and lower lignin levels than native trees. Studies have shown that L. maackii greatly affects the microarthropod and microorganism communities (Arthur et al. 2012, Poulette and Arthur 2012), in turn affecting rates of nutrient cycling.

Callery pear (Pyrus calleryana Decne.) is a relatively new woody invader in the eastern US. ‘Bradford’, the first horticultural cultivar, was introduced in 1960 (Vincent 2005, Culley 2017). Pyrus calleryana is self-incompatible; because cultivars are propagated by cuttings, the resulting trees cannot reproduce with one another (Culley 2017). However, problems with the Bradford cultivar led to the development of several other cultivars (Gilman and Watson 1994, Culley 2017). Many of these cultivars were planted next to existing plantings of Bradford and other cultivars. In addition, the rootstocks used for all cultivars can sometimes sprout and flower. This has led to the production of viable seed, and because the fruit is attractive to birds, seeds are readily spread (Culley and Hardiman 2009, Hardiman and Culley 2010, Culley et al. 2011, Culley 2017). This species is now reported to be invasive in 35 states (EDDMapS 2022; USDA NRCS 2022). Vincent (2005) and White et al. (2005) noted that this small tree can rapidly invade open and disturbed areas, forming dense thickets. However, its diameter distributions, which are typical of a shade-intolerant/midtolerant species, suggest that it is unlikely to invade forest understories (Boyce and Ocasio 2020), unlike L. maackii.

At present, leaf decomposition rates of P. calleryana, and thus its impacts on nutrient cycling, are unknown. Decomposition rates vary from site to site because of differences in many factors, including temperature and precipitation, and so a typical approach is to compare the rates of decomposition of various litters in the same site. In this case, leaves from two native and widely distributed trees were also used; the two species chosen were expected to have quite different litter decomposition rates. Red maple (Acer rubrum L.) litter has moderate to high levels of nitrogen, a C:N ratio higher than many other native species, low lignin levels, and moderate to fast decomposition rates (Day 1982, Blair 1988, Blair and Crossley 1988, Finzi and Canham 1998, Alexander and Arthur 2014, Jones and Swan 2015, Kay and Hone 2016). Little work has been done on leaf litter of American sycamore (Platanus occidentalis L.), but it has been reported to decay very slowly in aquatic environments and to have both high N and lignin levels (Ostrofsky 1997, Jones and Swan 2015). Both of these trees are common across the eastern USA (USDA NRCS 2022), and seedlings of both species establish in the same type of open and disturbed sites that are heavily invaded by P. calleryana (Burns and Honkala 1990). Leaves from L. maackii were also included, because it is an example of an invasive species with high leaf decomposition rates, and it often invades the same sites as P. calleryana (Boyce and Ocasio 2020). Thus, the overall objective of this study was to compare the leaf decomposition rates of P. calleryana against two native trees and one invasive shrub.

Materials and Methods.

Leaf litter was collected in the fall of 2014. Because the focus of this study was on invasive species, the native tree litter was collected from single trees in early November 2014 in the author's yard in Anderson Township, OH, whereas the litter from the two invasive species was collected on the campus of Northern Kentucky University in Highland Heights, KY (NKU), approximately 15 km away. The decomposition experiment was carried out at the latter site. The characteristics of this site are described by Castellano and Boyce (2007) and Boyce et al. (2014); it is on a road cut approximately 1 ha in area that overlooks Interstate-275. The part of the site used for this study was dominated by P. calleryana and L. maackii (personal observation).

The litter bags for this study were used in the EREN (Ecological Research as Education Network)/DATIS (Decomposition in Aquatic and Terrestrial Invaded Systems) project (EREN/DATIS 2022, Hornbach et al. 2021). These were ∼20 cm × 20 cm custom nylon bait bags obtained from Memphis Net & Twine Co. (Memphis, TN) with ∼1.5 mm openings on the bottom and ∼6.4 mm openings on the top.

Fresh litter samples of ∼20 g were loaded into litter bags and placed in the field on November 11, 2014. Samples of each plant type were also oven-dried for 3 days at 70 °C so that fresh sample weights could be converted into dry weight equivalents. Samples of each leaf type were placed at four locations within the P. calleryana-dominated stand on NKU's campus in a randomized-block design, and one bag of each species from each block was randomly sampled at 3, 9, 12, and 18 mo. Thus, there were 4 species × 4 blocks × 5 dates (0, 3, 9, 12, and 18 mo) for a total of 80 litter samples. After collection, drying, and weighing, samples were ground in a Wiley mill for homogenization. Portions were combusted at 500°C for at least 4 h to convert dry weights to ash-free dry weights (AFDW). In addition to determining AFDW, %C and %N were determined using a Carlo Erba CHN analyzer. Lignin was determined using the acid-detergent method as described by Hach Company (1990); two replications of each litter type were made. Dry weight and %N measurements across time were fit to the single exponential decay equation of Olson (1963), X(t_i)/X(t₀) = e^-kt, where X(t_i) was the measurement at time t_i, X(t₀) was the initial measurement, t = time (yr), and k is a decay coefficient (yr^–1). Data for each species were fit to this equation with nonlinear regression using the package nlstools in the statistical package R (R Core Team 2022). Data were also fit to the other decomposition curves proposed by Rovira and Rovira (2010), but the best fits were obtained with the single exponential decay equation. Differences between species for N and C:N levels were determined with a one-way ANOVA followed by Tukey tests, and 95% confidence intervals were calculated for mass loss and N loss k-values using nlstools and were examined for overlap.

Results.

At the start of the study, A. rubrum leaves had the lowest N levels, and P. occidentalis had the highest (Table 1). Lonicera maackii leaves had the lowest C:N ratio, whereas A. rubrum had the highest. Lonicera maackii also exhibited the highest decomposition rate (Table 1, Fig. 1a), although its k value did not differ statistically from those for P. calleryana and A. rubrum. Mass loss values of the four species were best correlated with lignin values (Pearson r = –0.86, P = 0.14), although lignin values for the three species other than Platanus occidentalis did not differ greatly from each other. Nitrogen decomposition rates followed a similar pattern for L. maackii, P. calleryana, and A. rubrum, in that L. maackii was the highest, but not differing statistically from the other two species (Table 1, Fig. 1b). However, the k value for P. occidentalis was negative, showing that N immobilization was occurring. N loss rates were highly correlated with lignin:N values (Spearman r = –0.96, P = 0.04). Lignin and lignin:N values followed similar trends, with P. calleryana having the lowest values and P. occidentalis having the largest.

Table 1.

Mass and nitrogen (N) loss k-values and initial litter N concentration, carbon (C):N ratios, lignin, and lignin:N ratios in leaf litter. Values for mass and N loss show best fit ± 95% confidence interval, obtained from the package nlstools (R Core Team 2022). Lignin values are means from two replications. Values for N and C:N are mean ± standard errors (n = 4). Lowercase letters indicate significant differences (95% confidence intervals are for mass and N loss k values; Tukey test for N and C:N).

Fig. 1.

(a) Decomposition (% initial litter mass remaining) and (b) nitrogen dynamics (% of initial litter nitrogen remaining). Error bars show mean ± SE.

Discussion.

Many studies have found that lignin concentration and lignin:N ratios are much better predictors of leaf litter decomposition rates than other factors (Shure et al. 1986, Blair 1988, Blair and Crossley 1988, Finzi and Canham 1998, Gartner and Cardon 2004, Alexander and Arthur 2014). High lignin concentrations can also hinder N mineralization (Finzi and Canham 1998, Shure et al. 1986). Measurements by Ostrofsky (1997) indicated that P. occidentalis has lignin:N ratios almost twice that of A. rubrum; aquatic mass lot rates of P. occidentalis were about one-third of A. rubrum, consistent with the terrestrial mass loss rates found in the current study. In the current study, mass loss rates were weakly correlated with lignin levels, although not significantly. On the other hand, N loss rates were significantly correlated with lignin:N ratios.

There were no statistically significant differences among L. maackii, P. calleryana, and A. rubrum, although leaf litter of L. maackii had the highest rates of mass and N loss of the four species in this study. In addition, L. maackii had the highest %N and lowest C:N ratio, as well as low %lignin and lignin:N ratio; values for N, C:N, lignin, and mass loss were similar to those found in other upland sites (Blair and Stowasser 2009, Arthur et al. 2012, Poulette and Arthur 2012, Jo et al. 2016), whereas N loss was lower (Jo et al. 2016). By comparison, leaf litter of P. calleryana had a lower %N and a higher C:N ratio; lignin and lignin:N levels were similar, but mass and N loss rates fell below those of L. maackii (although 95% confidence intervals of the k values did overlap). In fact, the mass loss rate of P. calleryana was most similar to that of A. rubrum (Fig. 1a). The N loss rate for P. calleryana was higher than that of A. rubrum and fell below that of L. maackii, although none of these were found to differ statistically (Fig. 1b). Overall, decomposition dynamics of P. calleryana were most similar to A. rubrum.

Despite the similarity of its mass and N loss rates to P. calleryana, A. rubrum had the lowest %N and the highest C:N ratio in this study, whereas lignin and lignin:N values were both higher than in P. calleryana. The %N and C:N ratio fell within the range seen in previous studies, while lignin, lignin:N, and the mass loss rate were slightly higher (Melillo et al. 1982; Aber et al. 1990, Delaney et al. 1996, Blair and Crossley 1988, Carreiro et al. 2000, Finzi and Schlesinger 2002, Ball et al. 2008, Alexander and Arthur 2014). The %N of P. occidentalis did not differ significantly from that of L. maackii, and its C:N ratio did not differ significantly from P. calleryana, but P. occidentalis had a very low rate of mass loss, and its negative k value for N loss indicates that it immobilized rather than mineralized N during the decomposition period of this study. This species also had the highest lignin and lignin:N values. Compared to Ostrofsky (1997), N and lignin were higher, C:N was lower, and lignin:N was very similar. Overall, N and C:N ratios were not useful in predicting decomposition rates in this suite of species. Lignin and lignin:N levels were better indicators, but they still did not completely predict these rates.

In summary, Pyrus calleryana had %N and C:N ratios that fell in the middle of the ranges for the species examined in this study, while lignin and lignin:N values were similar to those of L. maackii. Its mass loss rate was most similar to that of A. rubrum, but its N loss rate was somewhat higher. Thus, unlike L. maackii, P. calleryana appears to have litter decomposition dynamics that are similar to those of A. rubrum, an important native tree in eastern North America, and so P. calleryana is not predicted to greatly alter leaf litter decomposition in forests in which it invades. This is an example of an introduced species that does not display enhanced decomposition, as noted by Jo et al. (2016). A limitation of this study is that it only occurred in one location, and relative decomposition rates might differ at other sites. Also, many studies have shown that mixtures of leaf species do not behave additively (Gartner and Cardon 2004), so it will be important to look at P. calleryana litter decomposition dynamics in mixtures with native species. This is particularly important because its lignin and lignin:N values are similar to L. maackii, and L. maackii might accelerate mass and N loss rates when it occurs in mixtures (e.g., Blair and Stowasser 2009, Poulette and Arthur 2012).