BIOTROPICA 45(5): 541–548 2013 10.1111/btp.12048 Litterfall and Element Fluxes in a Natural Hardwood Forest and a Chinese-fir Plantation Experiencing Frequent Typhoon Disturbance in Central Taiwan Hsueh-Ching Wang1,2, Su-Fen Wang2, Kuo-Chuan Lin3, Pei-Jen Lee Shaner4, and Teng-Chiu Lin4,5 1 Department of Geography, National Taiwan University, Taipei, 10617 Taiwan 2 Department of Geography, National Changhua University of Education, Changhua, 50007 Taiwan 3 Taiwan Forestry Research Institute, Taipei, 10066 Taiwan 4 Department of Life Science, National Taiwan Normal University, Taipei, 11677 Taiwan ABSTRACT Chinese fir(Cunninghamia lanceolata) is the most important forest plantation species in subtropical Asia and is rapidly replacing natural forests. Such land-use change may affect ecosystem nutrient cycling through changes in litterfall nutrient flux. Tropical cyclones often cause pulses of litterfall. Previous studies, however, have mostly focused on the effects of a single cyclone with little effort examining the effects of repeated cyclones. We examined litterfall in a natural hardwood forest and a Chinese-fir plantation in central Taiwan expe- riencing an average of one typhoon per year. The natural hardwood forest had 54 percent higher annual litterfall (11,400 kg/ha/yr) than the Chinese-fir plantation (7400 kg/ha/yr). Four typhoon-affected months (typhoon period) contributed to approximately 60 percent of the litterfall and litterfall element flux in the natural hardwood forest and 80 percent in the Chinese-fir plantation, with contributions from individual typhoons varied by more than twofold. Litterfall N and P concentrations were significantly higher in typhoon period than in non-typhoon period, likely the result of limited retranslocation. Precipitation was a better predictor of quantity of typhoon-asso- ciated litterfall than wind velocity. Both types of forests in southeastern China beyond the reach of typhoons have litterfall peaks in the dry season. In contrast, we measured higher litterfall during the typhoon period than during the dry season, suggesting that in regions with frequent cyclones, cyclones drive temporal variation of litterfall. Global climate change is affecting the frequency and intensity of cyclones; therefore, knowledge of typhoon-litterfall dynamics is indispensable for understanding the effects of climate change on ecosys- tem nutrient cycling. Abstract in Chinese is available in the online version of this article. Key words: Chinese fir plantation; land use; litterfall; retranslocation; typhoon disturbance. LITTERFALL IS AN IMPORTANT COMPONENT OF NUTRIENT CYCLING IN replacement of natural forests by forest plantations, on nutrient FOREST ECOSYSTEMS (Brinson et al. 1980, Vitousek 1984, Aber & and carbon cycling is still unclear. Melillo 2001). Nutrient return from litterfall has been shown to Disturbances can greatly affect patterns of litterfall. Tropical be positively correlated with ecosystem productivity (Bray & cyclones (a generic term that has different names in different Gorham 1964, Chapin 1980). Numerous studies reported pat- geographic regions including typhoons, hurricanes and tropical terns of litterfall in temperate and tropical regions. Yet, empirical cyclones) often lead to sudden increases in litterfall followed by a data on litterfall patterns remain scarce in the subtropics of mon- period of low litterfall (Lodge et al. 1991, Lin et al. 2003, Beard soon Asia, although this region contains one of the world’s larg- et al. 2005). Studies in Hong Kong and Japan showed that a sin- est areal expanses of forest plantations. For example, 9.1 million gle typhoon event could contribute 20–45 percent of the annual hectares of Chinese-fir(Cunninghamia lanceolata Lamb.) plantations litterfall (Lam & Dudgeon 1985, Xu et al. 2004) and in Taiwan, already existed in southern and southeastern China in early 1990s six typhoons in 1 yr contributed to more than 50 percent of the and is now rapidly replacing natural forests (Ma et al. 2007, Guo annual litterfall (Lin et al. 2003). Unlike naturally senesced leaves et al. 2010). A few studies have compared primary productivity, in which the nutrient concentration is substantially reduced due nutrient cycling (e.g., litterfall and litterfall decomposition) and fine to retranslocation, typhoons deposit a large quantity of green and root dynamics between natural forests and Chinese-fir plantations immature leaves with limited retranslocation, leading to a high (Yang et al. 2004a,b, 2005), but few have taken into account cli- litterfall nutrient flux through forest floor (Lodge et al. 1991). mate conditions and disturbance regimes. Therefore, our under- Forest plantations can be more vulnerable to cyclone distur- standing of the effects of forest land uses, especially the bance than natural forests. The 1989 hurricane Hugo caused greater mortality to a 64-yr mahogany plantation than to a nearby natural forest of similar age in Puerto Rico (Fu et al. 1996). In Received 2 February 2013; revision accepted 1 April 2013. northeastern Taiwan, a Japanese cedar (Cryptomera japonica [L. f.] [This article was corrected after online publication on 17 August 2013. The spelling of author Dr. Su-Fen Wang’s name was corrected.] Don.) plantation had greater reduction in canopy cover than an 5Corresponding author; e-mail: [email protected] adjacent natural hardwood forest following a category 3 typhoon ª 2013 The Association for Tropical Biology and Conservation 541 542 Wang, Wang, Lin, Shaner, and Lin due to greater defoliation in the C. japonica plantation (Kang et al. ests in central Taiwan’s lowlands. The natural hardwood forests 2005). Both dead trees and fallen leaves contribute to litterfall. were dominated by Fagaceae and Lauraceae tree species, includ- Thus, the effects of tropical cyclones on litterfall dynamics could ing Schefflera taiwaniana (Nakai) Kanehira, Helicia formosana Hemsl., be very different between natural forests and forest plantations, Cinnamomum subavenium Miq., Cryptocarya chinensis (Hance) Hemsl., but it was rarely examined. Engelhardtia roxburghiana Wall., Cyclobalanopsis pachyloma (O. Seem.) Although litterfall dynamics driven by natural disturbances Schott., and Meliosma squamulata Hance. The majority of the are well documented, most studies have focused on the effect of remaining area in the experimental forest (192 out of 200 ha) is a single disturbance event. The intensity, duration, and timing of covered by conifer plantations. For the purpose of our study, the disturbance events can all affect ecosystem responses and feed- comparisons between the two forest types are only valid if they backs (Turner 2010) so that the results from a particular distur- have similar climate and disturbance regimes. Due to this con- bance event may not reflect the overall system responses to the straint, it is not possible to replicate at site level. Therefore, we type of disturbance. The lack of empirical studies on distur- focused the comparisons at the plot level. bance–ecosystem interactions driven by repeated disturbance The mean annual precipitation is 2200 mm and mean annual events is partly attributed to the unpredictability of most natural temperature is 20.8°C (lowest in January at 14.9°C and highest in disturbances (e.g., tropical cyclones and forest fires) in space and July at 25.4°C, 1961–2007). During our study period from May time. Therefore, a region with a high frequency of disturbance 2008 to April 2009, the annual precipitation was 3835 mm, and events (i.e., annual typhoons) might present great opportunities to mean annual temperature was 19.4°C (lowest in January at provide the much-needed empirical data. 13.2°C and highest in August at 23°C) (Fig. S2). There was a dry In this study, we examined patterns of litterfall in a Chinese- season between October and February with mean precipitation of fir plantation and an adjacent natural hardwood forest in Lienhu- 230 mm or approximately 10 percent of the annual precipitation achi Experimental Forest of central Taiwan (Table 1). The (Hsiao et al. 2007). experimental forest, on average, experienced 0.9 typhoon annually Litterfall patterns were examined at two paired (side-by- between 1994 and 2010 with as many as four typhoons in a year side) forest watersheds, an 8.4-ha natural hardwood forest, and (2008) (Lee et al. 2008, T. C. Lin unpubl. data) and a maximum a 5.9-ha Chinese-fir plantation. The plantation forest watershed of three consecutive years (1997–1999) without a typhoon. Spe- was originally a natural hardwood forest that was clearcut in the cifically, we tested the following two alternate hypotheses: (1) winter between November 1970 and March 1979 followed by typhoons influence temporal patterns of litterfall and litterfall planting seedlings of Chinese fir in 1981. Chinese-fir is a native nutrient flux; and (2) patterns of litterfall, litterfall element con- conifer widely distributed in southeastern China and Taiwan. centration and flux differ between natural hardwood forests and Between 1982 and 1985, weeds were cut several times a year Chinese-fir plantations. The results have important management and new seedlings planted to replace those that had died. There implications regarding forest nutrient cycling and carbon budgets have been no management activities in the watershed since not only for the study site but also for subtropical Asia where 1986. forest plantations are replacing natural forests. STAND PROPERTIES.—Diameter at breath height (dbh) and stand METHODS density of trees with dbh > 5 cm were measured in seven and four randomly located 20 9 20 m plots in the natural hardwood STUDY SITE.—The study was conducted at the Lienhuachi Experi- forest and Chinese-fir plantation, respectively. Aboveground bio- mental Forest (120o54′E, 23o54′N) in central Taiwan (Fig. S1). mass (ABG) of the two forests was estimated using allometric The Experimental Forest has a total area of 461 ha, of which models. Biomass of Chinese firs was calculated using an empirical 261 ha are covered by the only remaining natural hardwood for- model developed for Chinese fir (Lin et al. 2004). Biomass of natural hardwood trees was calculated using the Moist Model Without Height developed by Chave et al.