Landscape Ecol Eng (2010) 6:89–98 DOI 10.1007/s11355-009-0094-3
ORIGINAL PAPER
Impacts of land-use history on the diversity of a riparian forest landscape in warm-temperate Kyushu, southern Japan Tae Sato • Satoshi Ito • Yasushi Mitsuda Norihisa Soen
•
Received: 29 October 2008 / Revised: 13 July 2009 / Accepted: 29 October 2009 / Published online: 28 November 2009 Ó International Consortium of Landscape and Ecological Engineering and Springer 2009
Abstract We examined the impacts of land-use history on the species composition and diversity of a warm-temperate riparian forest landscape in Kyushu, southern Japan, focusing on the relationship between evergreen oaks and deciduous trees in natural and seminatural forests. The species composition of 59 plots was classified into four types (A to D). Type A, which showed a significant bias towards sites not subject to nonforest land use since 1947, had high species diversity consisting of (1) many lucidophyllous components of the region, including the rare indigenous oak Quercus hondae, and (2) summergreen tree species of varying dominance and number representing unique or locally rare elements of the riparian landscape in this warm-temperate region. Type B was dominated by a common species of oak, Q. glauca, and displayed less clear distribution bias with land-use history. In contrast to types A and B, types C and D, which were characterized by high dominance of deciduous trees, had negative bias away from sites that had been under forest land use in 1947. Presumably, intensive anthropogenic disturbances associated with nonforest land uses had expanded the habitats for deciduous trees. This phenomenon was represented by the establishment of forests (type D) dominated T. Sato N. Soen Graduate School of Agricultural Science, University of Miyazaki, Miyazaki 889-2192, Japan S. Ito (&) Faculty of Agriculture, University of Miyazaki, Miyazaki 889-2192, Japan e-mail:
[email protected] Y. Mitsuda Forestry and Forest Products Research Institute, Tsukuba 305-8687, Japan
by Ulmus davidiana var. japonica (UDJ) after it had been released from the suppression of evergreen forest trees during a period of nonforest land use that prevents the successful recovery of evergreen trees. From these results we conclude that the impacts of land-use history on the diversity of warm-temperate riparian forest landscape are multiphased: a period of nonforest land use has a strong negative impact on lucidophyllous forest trees represented by the rare indigenous oak Q. hondae; release from the suppressive effects of the lucidophyllous species then encourages establishment of locally rare deciduous tree flora represented by UDJ, which continue to persist for decades after abandonment of nonforest land use. Keywords Anthropogenic disturbance Nonforest land use Quercus hondae Species composition Summergreen forest trees Ulmus davidiana var. japonica (UDJ)
Introduction Riparian landscapes consist of a mosaic of varying physical environmental factors (substrate type and soil moisture, light and temperature regimes) under the influence of various types and regimes of natural disturbance according to the variations in riparian topography (Johnson et al. 1976; Masaki et al. 2007; Nakamura and Inahara 2007; Ito et al. 2006, 2008). Riparian forests established on these unique sites have a different vegetation structure than forests on more uniform topographic positions, such as hill slopes, and contain high biodiversity (Franklin 1992), including rare plant species that are dependent on the specific environmental conditions (Sakio and Suzuki 1997). Riparian areas also play an important role in
123
90
maintaining biodiversity by serving as ecological corridors for dispersion of vertebrates and gene flows (Naiman et al. 1993). Thus, riparian areas are a key element to be conserved for maximizing stability and biodiversity in landscapes and ecological contexts (e.g., Sakio and Tamura 2008). Riparian landscapes in Japan, however, have long been exposed to anthropogenic disturbances from agricultural and forestry land use because of their high productivity and utility for human activities (Ito et al. 2006; Sakio and Tamura 2008). In warm-temperate Japan, natural forests in riparian areas have been selectively converted to rice paddies on river terraces in lower reaches of rivers and to conifer plantations on the foot slopes of mountain streams (c.f., Ito and Nogami 2005; Ito et al. 2006). These land uses have resulted in fragmentation and decline in the distribution of natural riparian ecosystems and contribute to a decline in biodiversity through loss or modification of natural tree species composition. Furthermore, the impacts of land use on vegetation structure, particularly nonforest land uses such as pasture and cropping fields, have been reported to be quite severe and persistent even after the sites were returned to forest land use (Grashof-Bokdam and Geertsema 1998; Honnay et al. 1999; Graae et al. 2003; Osumi 2003; Ito et al. 2004). Thus, in riparian forests that have returned from nonforest land use, the current vegetation structure and diversity will be altered compared with the pre-existing natural forest. Sustainable management of riparian landscapes to promote ecosystem services based on sound biodiversity requires that both the natural conditions of riparian vegetation, and anthropogenic effects such as the impact of logging, conversion to plantation forests or agricultural land uses on them, be clarified. Recent studies in Japan have revealed the life histories and habitat characteristics of riparian component trees (e.g., Sakio 1997; Sakai et al. 1999; Sakio et al. 2002; Suzuki et al. 2002; Nakamura and Inahara 2007; Sakio and Tamura 2008). However, most of these studies were conducted in cool-temperate northern Japan; information on warm-temperate riparian forests is still limited (Ito and Nogami 2005; Ito et al. 2006). Furthermore, as most research on riparian forests has dealt with natural forests focusing on topographic effects of resource and disturbance regimes, the impacts of anthropogenic factors associated with land-use history on riparian forest structure have not been clarified. For warm-temperate Japan, riparian forests often consist of deciduous trees with disturbance-dependent traits and summergreen trees that are generally associated with cooltemperate forests (Ito and Nogami 2005; Ito et al. 2006). These deciduous trees contribute to increasing the biodiversity of warm-temperate riparian forests (Ito and Nogami 2005; Ito et al. 2006). Ulmaceae species such as Ulmus
123
Landscape Ecol Eng (2010) 6:89–98
davidiana var. japonica, Zelcova serrata, Celtis sinensis var. japonica, and Aphananthe aspera are typical components of warm-temperate riparian forests (Miyawaki 1981). Among these, Ulmus davidiana var. japonica (hereafter, UDJ) is one of the major components of cool-temperate natural riparian forests of hilly parts of northern Japan from Hokkaido to Honshu (Miyawaki et al. 1994; Kon and Okitsu 1995; Sakai et al. 1999; Wada and Kikuchi 2004) but also often occurs in low-elevation riparian forests in Kyushu (Nomiya 2008). Thus, UDJ might be a key species in interpreting the diversity and species composition of riparian forests in warm-temperate region. This species is also known to form Quercus–Ulmus forests in cool-temperate regions of the northern hemisphere, growing with deciduous oak trees mainly on river terraces (Lindsey 1961; Johnson et al. 1976; Miyawaki 1988). However, the phytosociological characteristics of UDJ in warmtemperate regions and its relationship with evergreen trees (especially evergreen Quercus spp.) are not fully understood. This paper reports research aimed at examining the impacts of land-use history combined with the effects of riparian topography on the species composition and diversity of a warm-temperate riparian forest landscape in Kyushu, southern Japan, focusing on the relationship between evergreen oaks and deciduous trees represented by UDJ in natural and seminatural (established after suffering severe anthropogenic disturbances) forests. In our study site, we found Quercus hondae Makino, an endangered indigenous oak of the region (Environment Agency of Japan 2000; Ito et al. 2007). As this species could be an additional key species to discuss the conservation of tree species diversity of the riparian landscape in the region, we pay particular attention to this species in the analysis.
Methods Study site The study was conducted in a riparian landscape along the Sakaigawa tributary (31°530 N, 131°140 E) of the Ohyodo River (Fig. 1). The study site was located on a sixth-order stream with an average longitudinal gradient of 1%. The investigated reach was 6.5 km long with a watershed area of 7,400 ha above the downstream end and 4,500 ha above the upstream end. The site has an elevation range from 40 to 90 m above sea level and is situated in the southern part of the warm-temperate (evergreen broadleaved forest) region. The annual mean temperature is 18.8°C and mean annual precipitation is 2,386 mm as measured at the nearest meteorological station (Miyazaki City). The base rock of the study site belongs to the Shimanto shattered belt
Landscape Ecol Eng (2010) 6:89–98
91
45N
Ohyodo River
Table 1 Breakdown of the distribution of the 59 surveyed plots by land-use history and dominant topography Type of land-use history (1947–1974)
40N
Forest––seminatural forest (FS) Forest––plantation (FP)
Sakaigawa tributary
35N
Nonforest––seminatural forest (NFS) Nonforest––plantation (NFP)
15 2 36 6
Topography 30N
Study plots 130E
135E
140E
Study site 0
500m
Terrace
17
Alluvial cone
4
Talus
2
Slope
36
Figures indicate the number of plots for each category Fig. 1 Location of the study site and vegetation assessment plots used for studying the influence of prior land use on riparian vegetation in southern Kyushu
and is dominated by shale associated with conglomerate, sandstone, and limestone. In the 6.5-km study reach, forest roads have been established along the river and protection works have been constructed in several places along the left bank. Some river terraces had been used as terrace paddies or building yards (abandoned 30–60 years before the survey), and the slopes of the opposite bank included conifer plantations. In this study, we investigated the natural and seminatural (unplanted) forest patches between the river channel and forest roads through a field investigation conducted in 2007. The average width of the vegetated belt between the edge facing the river channel and forest roads of the investigated reach was 17.3 m (maximum 28.7 m, minimum 5.8 m) and the average slope of the bank was 28.6°. The upper section of the studied reach (1.6 km) is currently assigned as a protected forest for environmental conservation by the local Forest Management Office and has been maintained with a relatively well-developed forest structure dominated by evergreen trees. The lower reach (4.7 km) is partially comprised of evergreen and deciduous seminatural forests, but significant parts displayed evidence (undeveloped secondary forest) of anthropogenic disturbance by alternative land use. Data collection For the field vegetation survey we selected 24 stands (20 stands from the left bank and 4 stands from the right bank of the upper reach) of natural and seminatural forest patches to cover the range of variation in topography (terrace, alluvial cone, talus, and side slope) (Fig. 1, Table 1). Conifer plantations, sites used for buildings, and banks affected by protection works at the time of the survey were excluded from the sampling. Several young stands
developing on sediments and side-cast spoil of forest roads were included in the samples. For each sample stand, two or three plots with dimensions of 10 m in the direction of flow and 5 m transverse to the direction of flow were located side by side, beginning at the vegetation edge at the river channel and continuing to the forest road running parallel to and approximately 10–15 m from the river. The total number of plots was 59 (Table 1). The vegetation survey was conducted during the period of September to October, 2007. All living trees of more than 1.3 m height were identified and their diameter at 1.3 m height (DBH) was measured. Diameter less than 0.5 cm was considered as 0.5 cm. The topography of each plot was recorded as the most dominant category (occupying more than 50% of the plot area). The land-use history of each plot was determined by interpreting aerial photographs taken in 1947 and 1974. Nonforest patches were detected and distinguished from forest patches for each plot location using monochrome photographs taken in 1947; more detailed patch classification than this was not possible because it was difficult to determine plantations and coppice woodlands from the photographs. Color photographs taken in 1974 were used to classify each plot location according to two land uses: seminatural forests and conifer plantations. From the combination of land-use classifications in these two periods, the land-use history for each study plot was classified into four categories: (1) forest (in 1947)––seminatural forest (in 1974) (FS), (2) forest––plantation (FP), (3) nonforest––seminatural forest (NFS), and (4) nonforest–– plantation (NFS) (Table 1). Analysis Altogether 72 species occurred in the 59 plots. The summed dominance ratio (SDR) (%) of each species in each plot was calculated from the relative tree density (RD) (%)
123
92
Landscape Ecol Eng (2010) 6:89–98
and relative basal area (RBA) (%) according to the following equation: SDR ¼ ðRD þ RBAÞ=2:
ð1Þ
This was adopted to reduce the bias of one large tree within plots including many small-diameter trees. The matrix of SDR for each species in each plot was used to perform a cluster analysis to detect species groups with similar occurrence traits. The cluster analysis applied a correlation matrix for similarity of occurrence and the group average (unweighted pair-group method with arithmetic mean, UPGMA) as the clustering method. Vegetation types were classified by TWINSPAN (Hill 1979) using the same SDR matrix. The multivariate analyses were performed by using PC-Ord version 4 (McCune and Mefford 1999), a personal-computer-based software package for vegetation analysis. The relationship between UDJ and other deciduous trees, all evergreen trees, evergreen oaks (Q. hondae, Q. glauca, Q. salicina, Q. gilva, and Q. myrsinifolia), and the single species of Q. hondae was examined by correlation analysis from total or individual SDRs, We applied Kendall’s rank correlation because the SDRs from 0 to 100 did not support normality. These statistical analyses were performed with R version 2.4.1 software (R Development Core Team 2005). The species compositions of the classified vegetation types were compared from the mean SDR of species groups determined from the interspecies cluster analysis. The SDR were also compared for the species functional types based on the combination of leaf habits (evergreen or deciduous) and life forms (tree or shrub), and pioneers in order to determine the key elements of the classified vegetation in terms of disturbance-dependent characteristics of trees, abundance of canopy species, and occurrence of deciduous leaf habit in the warm-temperate region. Leaf habits, life forms, and pioneer characteristics were determined using the descriptions of Kitamura and Murata (1971, 1979) and Okuda (1997). The SDR of UDJ, Q. hondae, Q. glauca, the commonest oak of riparian and hillslope forests of the region (Ito et al. 2007), and the total SDR of evergreen oaks were also compared separately between vegetation types. Species richness and Shannon’s species diversity index (H0 ) were compared for the species types of leaf habits and life forms, and pioneers between the classified vegetation types. The diversity index (H0 ) was calculated from X pi lnðpi Þ ð2Þ H0 ¼ where pi is the relative dominance of species i, for which SDR/100 for each species was used. The statistical comparisons between vegetation types were performed using the Kruskal–Wallis test with Bonferroni multiple-comparison correction (p \ 0.05).
123
To detect any bias from land-use history of the plots for each vegetation type, we tested whether the observed frequencies of the given land-use type were significantly higher or lower than expected given a null hypothesis of no relationship between vegetation type and land-use type, using the bootstrap method. We calculated the proportion of the number of times that the resampled frequency of land-use type from all plot samples was higher or lower than the observed frequency for a particular vegetation type among the total number of resamplings (10,000). This proportion could be regarded as the upper or lower probability that the observed result could have occurred by chance if the null hypothesis was true. A similar analysis was performed to detect any topographic bias of the plots in each vegetation type.
Results Interspecies relationships The 72 species recorded in the 59 plots were classified into four groups according to their occurrence similarity by cluster analysis (Fig. 2). Group 1 consisted of many evergreen trees such as Distylium racemosum, Q. gilva, Q. salicina, Castanopsis cuspidata, Ilex chinensis, and other major components of lucidophyllous natural forests of the region such as Symploco glaucae–Castanopsietum sieboldii Miyawaki et al. 1971 (Miyawaki et al. 1994), Lasiantho–Quercetum gilvae K. Fujiwara 1981 (Miyawaki et al. 1994), and Distrio–Querucetum Salicinae Nomoto et Suganuma 1965 (Miyawaki et al. 1994). Q. hondae, an endangered indigenous oak of the region which inhabits lower slopes along mountain streams (Ito et al. 2007), was included in this group. Group 2 consisted of Mallotus japonicus, a typical pioneer of the region, in association with several disturbance-dependent deciduous species (e.g., Rhus succedanea, Carpinus tschonoskii, and Deutzia scabra) (Okuda 1997) and evergreen trees common in evergreen coppice stands (e.g., Q. glauca, Machilus thunbergii, and C. sieboldii) (Miyawaki 1981). Group 3 was comprised of Morus bombycis, Celtis sinensis var. japonica, Aphananthe aspera, and Acer palmatum, which are riparian species that also occur in cool-temperate forests (Okuda 1997), associated with several evergreen shrubs or subtrees such as Aucuba japonica, Neolitsea sericea, and Cinnamomum japonicum. Group 4 included UDJ and deciduous pioneer shrubs (Rubus palmatus var. palmatus and Rosa multiflora). The SDR of UDJ was strongly negatively correlated with the total SDR of evergreen trees and evergreen oaks and the SDR of Q. hondae, but was not correlated with the SDR of other deciduous trees (Fig. 3). Of the four SDR
Landscape Ecol Eng (2010) 6:89–98 Fig. 2 Species grouping of riparian vegetation recorded in study plots using cluster analysis based on occurrence similarity. Figures in parentheses are the number of species
93
Distance 7.81 Quercus hondae, Distylium racemosum, Quercus salicina, Quercus gilva, Castanopsis cuspidata, Camellia japonica, Neolitsea aciculata, Machilus japonica, Eurya japonica, Ilex crenata, Meliosma rigida, Ilex chinensis, Callicarpa japonica, Cleyera japonica, Ehretia ovalifolia, Daphniphyllum macropodum, Litsea acuminata, Styrax japonica, Podocarpus macrophyllus, Hydrangea luteovenosa, Adina pilulifera, Euonymus alatus, Ilex latifolia, Wisteria brachybotrys, Boehmeria spicata, Cornus macrophylla, Gardenia jasminoides, Swida controversa, Xylosma congestum, Ilex rotunda, Ilex buergeri, Albizia julibrissin, Unknown1
Group 1 (33)
Quercus glauca, Castanopsis sieboldii, Carpinus tschonoskii, Machilus thunbergii, Cam ellia sasanqua, Mallotus japonicus, Rhus succedanea, Deutzia scabra, Tilia kiusiana, Pieris japonica subsp. japonica, Premna japonica, Symplocos kuroki, Ligustrum lucidum, Callicarpa mollis, Prunus spinulosa, Abies firma, Unknown2
Group2 (17)
Aucuba japonica, Celtis sinensis var. japonica, Pourthiaea villosa var. zollingeroi, Skimmia japonica, Acer palmatum, Diospyros kaki, Quercus myrsinifolia, Helwingia japonica, Actinodaphne lancifolia, Aphananthe aspera, Cryptomeria japonica, Morus bombycis, Neolitsea sericea, Cinnamomum tenuifolium, Ficus erecta, Ligustrum japonicum, Fatsia japonica
Group 3 (17)
Ulmus davidiana var. japonica, Ficus erecta var. sieboldii, Elaeagnus pungens, Rubus palmatus var. palmatus, Rosa multiflora,
100
100
a = -0.252 p = 0.005
80
Ulmus davidiana var. japonica (%)
Fig. 3 Relationship of summed dominance ratio (SDR) of Ulmus davidiana var. japonica (UDJ) with the SDR of a all evergreen trees, b other deciduous trees, c evergreen oaks, and d Q. hondae, indicating land-use history of plots. Statistics in the panels refer to Kendall’s rank correlation coefficient (s) and significance level (p) of the correlation for all plots
Group 4 (5)
b = 0.079 p = 0.378
80
60
60
40
40
20
20
0
10.4
0 0
20
40
60
80
100
0
20
100
c
100
= -0.393 p < 0.001
80
60
40
40
20
20
0 20
40
60
80
Total SDR of Quercus spp. (%)
60
80
100
d = -0.194 p = 0.030
80
60
0
40
Total SDR of deciduous tree (%)
Total SDR of evergreen tree (%)
100
0 0
20
40
60
80
100
SDR of Quercus hondae (%)
Forest – Semi natural forest
Non forest – Semi natural forest
Forest - Plantation
Non forest - Plantation
123
94
Landscape Ecol Eng (2010) 6:89–98
100
b b bc
B (25)
A (10)
C (17)
b
80
2 1
3 4
3
4
1
(%) SDR Mean
b
4
Group 4
bc
D (7)
60
a a
Group 3 Group 2
a
Group 1
40 a
20
b
ab
bc
b
b
3 2
1 4
4 3
0 A
C
c
D
Vegetation type
4 7 6 3 Fig. 4 Dendrogram obtained by TWINSPAN. Figures at terminal nodes indicate the number of plots. Figures in parentheses are the number of plots of vegetation types (A to D) as identified by the second division
Table 2 Summary of TWINSPAN analysis Divided plot group
Indicator species
Eigenvalue
AB
Castanopsis sieboldii, Eurya japonica
0.404
CD
Ulmus davidiana var. japonica, Aucuba japonica
A
Eurya japonica, Cleyera japonica
B
Quercus glauca, Deutzia scabra
C
Aucuba japonica, Quercus glauca, Helwingia japonica
D
Ulmus davidiana var. japonica, Elaeagnus pungens
0.284 0.354
relationships for UDJ, the correlation coefficient was highest with evergreen oaks (-0.393). At the plot level, UDJ and Q. hondae displayed clearly separate occurrence patterns. Vegetation classification The vegetation of the 59 surveyed plots was classified by TWINSPAN into four types (Fig. 4): type A (10 plots), type B (25), type C (17), and type D (7). At the first level division in TWINSPAN, types A and B were grouped against types C and D. The indicator species for each type were C. sieboldii for type A, Eurya japonica for type B, UDJ for type C, and A. japonica for type D (Table 2). At the second level, TWINSPAN adopted E. japonica and Cleyera japonica for type A against Q. glauca and D. scabra for type B; and A. japonica, Q. glauca, and Helwingia japonica for type C against UDJ and Elaeagnus pungens for type D.
123
B
Fig. 5 Summed dominance ratio of the species groups identified by cluster analysis for the four vegetation types classified by TWINSPAN. Species groups in different vegetation types displaying the same small letter are not significantly different (Scheffe, p [ 0.05)
Each of the classified species groups except for group 3 had a significantly higher mean SDR (p \ 0.05) in one or other of the vegetation types (Fig. 5): group 1 in type A, group 2 in type B, and group 4 in type D. The SDR of group 3 was significantly higher in vegetation type C than in types A and B but not type D. The species compositions of the leaf habit/life form types in each vegetation type are shown in Table 3. All evergreen trees and Quercus spp. other than Q. hondae and Q. glauca had a significantly higher SDR (p \ 0.05) in vegetation types A and B than in types C and D. The SDR of Q. hondae was higher in type A than in type C. Q. glauca dominated in type B, with a much higher SDR than in the other types. The SDR of deciduous trees was highest in types C and D and lowest in type B, with type A having an intermediate value. UDJ was dominant in type D and had a significantly lower SDR in type A and B than in type D. Deciduous shrubs had a higher SDR in type D than in type A. The species richness was higher in types A and B than in type D (Table 4). The richness of evergreen trees was significantly higher in types A and B than in types C and D. Type B had a significantly lower richness in deciduous trees than did type C. Similar trends were observed for the diversity index (H0 ). The H0 values of the vegetation types for all species were in almost the same order as for species richness, though the difference among them was not significant. For evergreen trees, types A and B had significantly higher H0 than did types C and D. The H0 of deciduous trees was significantly lower in type B than in types C and D. Land-use history and topography of the vegetation types Randomization tests performed for topographic attributes of the plots (Table 5) revealed that vegetation type A was
Landscape Ecol Eng (2010) 6:89–98
95
Table 3 Species composition (summed dominance ratio, SDR%) of the four vegetation types that were classified by TWINSPAN. Species are grouped by various leaf habits and life forms: evergreen trees, evergreen shrubs, deciduous trees, deciduous shrubs, pioneers, and unidentified. Selected individual species are also shown for evergreen trees and deciduous trees
Table 4 Average species richness and diversity index (H’) of species types grouped by leaf habits, life forms and pioneers in the four vegetation types classified by TWINSPAN
Species group/species
Species richness (/50 m2)
Evergreen trees Quercus hondae Quercus glauca Other evergreen trees Evergreen shrubs Deciduous trees Ulmus davidiana var. japonica Other deciduous trees
Vegetation types
Prob
A
B
C
50.9a
68.1a
18.6b
7.3a 9.7b
5.7ab 40.8a
0.0b 11.1b
D 4.7b 0.0ab 1.9b
Vegetation type A
Evergeen trees 0.000 0.007 0.000
34.0a
21.7a
7.5b
2.8b
0.000
20.5ab
7.6c
24.7a
6.1bc
0.000
6.0a
Prob.
B
C
D
4.7a
2.4b
0.6b
0.000
Evergeen shrubs
2.1a
2.2a
1.9ab
0.7b
0.009
Deciduous trees
1.4ab
1.0b
2.2a
2.0ab
0.018
Deciduous shrubs
1.9
2.7
2.3
1.4
0.052
Pioneers
0.2
0.3
0.6
1.0
0.144
Unidentified
0.0
0.0
0.0
0.1
0.327
Total
11.6a
10.9a
9.5ab
5.9b
0.007 0.000
21.6bc
6.4c
36.3ab
53.7a
0.000
7.3bc
1.4c
15.9b
49.1a
0.000
Evergeen trees
1.019a
0.832a
0.349b
0.108b
4.9
20.4
4.6
0.072
Evergeen shrubs
0.395a
0.238ab
0.385a
0.128b
0.002
0.320ab
0.141b
0.505a
0.448a
0.000
14.3
Diversity index (H’)
Deciduous shrubs
6.1b
16.4ab
16.9ab
28.3a
0.007
Deciduous trees
Pioneers
0.8
1.4
3.4
4.1
0.141
Deciduous shrubs
0.195b
0.387a
0.352a
0.362ab
0.023
Unidentified
0.0
0.1
0.0
3.2
0.313
Pioneers Unidentified
0.025 0.000
0.037 0.004
0.063 0.000
0.125 0.048
0.137 0.313
Total
1.954a
1.638a
1.654a
1.219a
0.012
SDR = (relative tree density ? relative basal area)/2 Prob denotes the significance level (Kruskal–Wallis test) Values followed by the same letter across vegetation types are not significantly different (Bonferroni correction, p [ 0.05)
biased away from alluvial cones and talus (i.e., depositional slopes) (p \ 0.001). Type B was biased away from terraces (p \ 0.05) and talus (p \ 0.001). In contrast, type D was biased towards terraces (p \ 0.05) and away from alluvial cone and talus (p \ 0.001). There was no topographic distribution bias for type C. With reference to land-use history, type A had a significant positive bias towards FS and FP types (p \ 0.05) and strong negative bias away from NFS (p \ 0.01) and NFP types (p \ 0.001). Type B had no clear distribution traits except for having no occurrence in FP (p \ 0.001). Types C and D showed a strong negative bias (p \ 0.001) away from FS and FP. Type C was positively biased towards NFP.
Discussion Vegetation types and their conditions In general, the structure of riparian forest strongly depends on topography (e.g., Nagamatsu and Miura 1997; Ito et al. 2006; Nakamura and Inahara 2007). The variation in species composition detected in the current study also reflected variations in riparian topography. Of the four vegetation types, types A and B were characterized by strong dominance of evergreen trees and shrubs (Tables 3) and were
Prob. denotes the significance level (Kruskal–Wallis test) Values followed by the same letter across vegetation types are not significantly different (Bonferroni correction, p [ 0.05)
mainly established on slopes (Table 5). Type C was characterized by high dominance of deciduous trees, though this type had no distribution bias in any topography. Type D was dominated by UDJ associated with pioneers (Figs. 2, 5) and had a strong distributional bias towards terraces (Table 5). However, the classification of species composition was more clearly related to plot land-use history (Table 5). Type A showed a significant bias towards sites free from intensive anthropogenic disturbances for at least the last 60 years (i.e., away from nonforest land uses NFS and NFP, as shown in Table 5). Type B showed a less clear distribution bias with land-use history than did type A. In contrast, types C and D were negatively biased away from sites with forest land use in 1947 (FS and FP). In particular, type C, which was dominated by deciduous trees other than UDJ, was concentrated in the NFS type, with no topographic preference. These results show that the forests dominated by deciduous trees in the investigated riparian landscape are associated with former nonforest land use, and suggest that the current vegetation is more strongly determined by previous human activities than by topography. Similar strong influences of previous nonforest land uses have also been reported in nonriparian forest ecosystems (Grashof-Bokdam and Geertsema 1998; Honnay et al. 1999; Graae et al. 2003; Osumi 2003; Ito et al. 2004). This study revealed that a period of nonforest land use, i.e.,
123
96
Landscape Ecol Eng (2010) 6:89–98
Table 5 Number and proportion (in parentheses) of plots located in different categories of former land use and topography for each of the four vegetation types classified by TWINSPAN Vegetation type
Total number of plots
A
B
C
D
10
25
17
7
Topography Terrace Alluvial cone
4 0
(40) (0)
NS ---
4 3
(16) (12)
NS
4 1
(24) (6)
NS NS
5 0
(71) (0)
? ---
Talus
0
(0)
---
0
(0)
---
2
(12)
NS
0
(0)
---
Slope
6
(60)
NS
18
(72)
NS
10
(59)
NS
2
(29)
-
Former land-use type Forest—semi-natural forest (FS)
6
(60)
?
9
(36)
NS
0
(0)
---
0
(0)
---
Forest—plantation (FP)
2
(20)
?
0
(0)
---
0
(0)
---
0
(0)
---
Non forest—semi-natural forest (NFS)
2
(20)
--
12
(48)
NS
16
(94)
??
6
(86)
NS
Non forest—plantation (NFP)
0
(0)
---
4
(16)
NS
1
(6)
NS
1
(14)
NS
Positive and negative symbols indicate values that were significantly greater (positive) or less (negative) than expected based on the null hypothesis that the total number of plots is distributed evenly among sub-categories (see text for further details). Three symbols represents p \ 0.001; two symbols represents p \ 0.01; and one symbol represents p \ 0.05
intensive anthropogenic disturbance, strongly influences the composition of recovered vegetation, even in riparian areas where natural disturbance regimes are more frequent and intensive (Ito and Nakamura 1994; Sakio 1997; Suzuki et al. 2002; Nakamura and Inahara 2007). The FS type could be expected to have received the least influence from anthropogenic disturbances among the four land-use types in this study. Thus, vegetation type A, showing the strongest association with FS, is assumed to have a species composition which is most similar to those of typical natural riparian forests of the region: high dominance by species group 1 consisting of lucidophyllous components represented by evergreen Quercus spp. (Q. gilva, Q. salicina, Q. hondae, and Q. myrsinifolia) and low dominance of group 3 characterized by cool-temperate summergreen forest components (Figs. 2, 5). Q. glauca, which dominated type B, has been reported to have greater disturbance dependence than other evergreen oaks (Ito et al. 2007) and is known to often occur in disturbed forests due to its vigorous resprouting ability after logging, especially on steep slope (c.f., Miyawaki 1981). Our analysis of species correlation also demonstrated a close relationship between Q. glauca and pioneer species such as Mallotus japonica (Fig. 2). Thus, the Q. glauca-dominated type B might be distinguished from type A as having a composition representative of secondary forest on steep slopes. Impacts of land-use history on the floristic biodiversity of riparian landscapes One important aspect of the ecological impacts of nonforest land uses detected in this study is the encouragement
123
of disturbance-dependent deciduous trees and summergreen forest species. These species are common in the cool-temperate zone but locally rare in warm-temperate forests, being important contributors to the diversity of regional tree flora in the warm-temperate riparian zone (Ito et al. 2006). In our study site, summergreen trees were included as unique or locally rare elements in vegetation type A (the closest representation of natural vegetation), which had high species diversity consisting of a variety of lucidophyllous forest trees (Tables 3 and 4). In contrast to this, severe anthropogenic disturbances by previous nonforest land use altered the original habitat–vegetation relationship in the riparian landscape. The habitats of deciduous trees had obviously been expanded to a wider range of sites from terraces to other topography, resulting in forests dominated by deciduous trees (types C and D) associated with decline in diversity of evergreen trees after abandonment of the nonforest land use (Tables 3, 4, 5). These changes could be partly explained by the similarity between site conditions created by nonforest land uses and natural riparian disturbances. Nonforest land uses such as agricultural fields, forest roads, and log landings lead to great changes in the physical and biological conditions of surface soils (Guariguata and Dupuy 1997; Pinard et al. 2000; Whitman et al. 1997). These would have impacts similar to those of intensive natural disturbances such as the earth surface movements that typically occur in riparian areas (Ito and Nakamura, 1994). The resulting compacted or degraded soil is likely to be detrimental to the survival of acorns and the root penetration of germinated oak seedlings. The loss of surface soils also means the loss of seedbanks (Guariguata and Dupuy 1997; Whitman et al. 1997; Pinard
Landscape Ecol Eng (2010) 6:89–98
et al. 2000). Furthermore, the evergreen species of this region include only a few species with wind-dispersed seeds which offers an advantage for colonization of extensively disturbed areas (Kominami et al. 1995). Therefore, especially in the case of disturbance over extensive areas, the impact is expected to restrict the recruitment of evergreen trees with limited dispersal capacity. This impact would be severe for acorns from oaks (cf., Utsugi et al. 2006). One or a combination of these factors must explain the unsuccessful establishment of evergreen forests dominated by oaks after the abandonment of nonforest land use. Conversely, the abandonment of nonforest land use advantages trees with wind-dispersed seeds such as UDJ. The unsuccessful establishment of evergreen trees also reduces the heavy suppression, allowing the return of deciduous trees which are generally more light-demanding than evergreen trees (Tadaki 1996; Ke and Werger 1999; Ito et al. 2008). Similar effects causing unsuccessful regeneration of evergreen trees and favoring deciduous trees were reported for disturbances from extensive and frequent natural sedimentation events (Ito et al. 2006, 2008). Among the summergreen forest trees, UDJ occurred widely in our study site and dominated the forest in places; this contrasts with its reported distribution in well-developed riparian forests of a warm-temperate region, which was restricted to near channels or the lower fan subject to frequent erosion and sedimentation, and the dominance was quite low compared with evergreen oaks (Ito et al. 2006, 2008). Thus, the UDJ-dominated forests in our study site (type D) were interpreted to be strongly associated with previous nonforest land use as well as terrace topography. The close relationship of UDJ with typical pioneer shrubs (Fig. 2) showed more disturbance-dependent characteristics of UDJ than the other summergreen forest trees. Furthermore, UDJ showed a clear negative correlation with evergreen trees, including oaks (Fig. 3). These facts suggest that UDJ is suppressed or outcompeted by evergreen trees. This might explain why typical Ulmus–Quercus forests as observed in cool-temperate riparian forests could not be found in warm-temperate region with heavy shade of evergreen oak canopy. We therefore presume that UDJdominated forests (type D) had established on terraces due to the suitable topography and soil conditions (Sakai et al. 1999; Wada and Kikuchi 2004) and the release from inhibiting effects of Quercus spp. and other evergreen trees caused by the period of nonforest land use. The habitat of UDJ was clearly segregated from that of Q. hondae (Fig. 3), which occurred in developed forests on foot slopes (Miyawaki 1981; Ito et al. 2007) and had the highest dominance in the most developed type A on foot slope and terrace in this study. This suggests a kind of trade-off relationship between the maintenance of the
97
population of rare indigenous oak and the establishment of naturally locally rare vegetation dominated by UDJ. Anthropogenic disturbances by nonforest land use might have controlled this relationship at our study site. We conclude that the impacts of land-use history on the diversity of the investigated riparian forest landscape were multiphased as follows: alteration from intensive nonforest land use and subsequent abandonment resulted in the encouragement of locally rare deciduous tree flora represented by UDJ. The period of land use itself had a strong and persistent negative impact on the lucidophyllous forest trees represented by the rare indigenous oak Q. hondae. Acknowledgments The authors wish to thank Hiroka Ito and Mika Sano for their assistance in the field survey. The Miyazaki District Forest Management Office is also acknowledged for providing the research site. This study was partly supported by a Grant-in-Aid for Scientific Research from JSPS (nos. 20380090 and 20241009).
References Environment Agency of Japan (2000) Threatened wildlife of Japan, red data book 8, vascular plants (in Japanese), 2nd edn. Japan Wildlife Research Center, Tokyo Franklin JF (1992) Scientific basis for new perspectives in forest and stream. In: Naiman RJ (ed) Watershed management. Springer, New York, pp 25–72 Graae BJ, Sybde PB, Fritzboger B (2003) Vegetation and soil differences in ancient opposed to new forests. For Ecol Manage 177:190–197 Grashof-Bokdam CJ, Geertsema W (1998) The effect of isolation and history on colonization patterns of plant species in secondary woodland. J Biogeogr 25:837–846 Guariguata MR, Dupuy JM (1997) Forest regeneration in abandoned logging roads in lowland Costa Rica. Biotropica 29:15–28 Hill MO (1979) TWINSPAN. A Fortran program for arranging multivariate data in an ordered two-way table by classification of the individuals and attributes. Ecological and Systematics Department, Cornell University, New York Honnay O, Hermy M, Coppin P (1999) Impact of habitat quality on forest plant species colonization. For Ecol Manage 115:157–170 Ito S, Nakamura F (1994) Forest disturbance and regeneration in relation to earth surface movement (in Japanese with English summary). Jpn J For Environ 36(2):31–40 Ito S, Nogami K (2005) Species composition and environments of riparian forests consisting of Lagerstroemia subcostata var. fauriei, an endangered species on Yakushima Island (in Japanese with English summary). Veg Sci 22:15–23 Ito S, Nakayama R, Buckley GP (2004) Effect of previous land-use on plant species diversity in semi-natural and plantation forest in a warm-temperate region in southeastern Kyushu, Japan. For Ecol Manage 196:213–225 Ito H, Ito S, Matsui T, Marutani T (2006) Effect of fluvial and geomorphic disturbances on habitat segregation of tree species in a sedimentation-dominated riparian forest in warm-temperate mountainous region in southern Japan. J For Res 11:405–417 Ito S, Ohtsuka K, Yamashita T (2007) Ecological distribution of seven evergreen Quercus species in southern and eastern Kyushu, Japan. Veg Sci 24:53–63 Ito H, Ito S, Nakao T, Kadomoto K (2008) Effects of fluvial and geomorphic disturbances on forest dynamics of a sedimentation-
123
98 dominated riparian forest in warm-temperate mountainous region in southern Japan (in Japanese with English summary). Jap J For Environ 50(1):17–24 Johnson WC, Burgess RL, Keammerer WR (1976) Forest overstory vegetation and environment on the Missouri River floodplain in North Dakota. Ecol Monogr 46:59–84 Ke G, Werger MJA (1999) Different responses to shade of evergreen and deciduous oak seedlings and the effect of acorn size. Acta Oecol 20:579–586 Kitamura S, Murata G (1971) Colored illustrations of woody plants of Japan (in Japanese), vol 1. Hoikusha, Osaka Kitamura S, Murata G (1979) Colored illustrations of woody plants of Japan (in Japanese), vol 2. Hoikusha, Osaka Kominami Y, Manabe T, Tanouchi H, Sato T, Niiyama K (1995) Fruit type composition of seed rain in Aya evergreen broadleaved forest (in Japanese). Bull Kyushu Br Jpn For Soc 48:111– 112 Kon H, Okitsu S (1995) Structure and Regeneration of Ulmus davidiana var. japonica Forest at Mt. Asama and Mt. Togakushi, Central Japan (in Japanese with English summary). Tech Bull Hort Chiba Univ 49:99–110 Lindsey AA (1961) Vegetation and environment along the Wabash and Tippecanoe Rivers. Ecol Monogr 31:105–156 Masaki T, Osumi K, Takahashi K, Hoshizaki K, Matsune K, Suzuki W (2007) Effects of substrate heterogeneity on seed-to-seedling process and tree coexistence in a riparian forest. Ecol Res 22:724–734 McCune B, Mefford MJ (1999) PC-ORD. Multivariate analysis of ecological data, version 4. MjM Software Design, Gleneden Beach Miyawaki A (ed) (1981) Vegetation of Japan, vol 2: Kyushu (in Japanese). Shibundo, Tokyo Miyawaki A (ed) (1988) Vegetation of Japan, vol 9: Hokkaido (in Japanese. Shibundo, Tokyo Miyawaki A, Okuda S, Fujiwara R (eds) (1994) Handbook of Japanese vegetation (in Japanese). Shibundo, Tokyo Nagamatsu D, Miura O (1997) Soil disturbance regime in relation to micro-scale landforms and its effects on vegetation structure in a hilly area in Japan. Plant Ecol 133:191–200 Naiman RJ, Decamps H, Pollock M (1993) The role of riparian corridors in maintaining regional biodiversity. Ecol Appl 3:209– 212 Nakamura F, Inahara S (2007) Fluvial geomorphic disturbances and life history traits of riparian tree species. In: Johnson EA, Miyanishi K (eds) Plant disturbance ecology: the process and the response. Elsevier, New York Nomiya H (2008) Germinatin traits of Japanese elm (Ulmus davidiana var. japonica) in lowland of Kyushu, Japan. Kyushu J For Res 66:67–68
123
Landscape Ecol Eng (2010) 6:89–98 Okuda S (ed) (1997) Wild plants of Japan (in Japanese). Shogakukan, Tokyo Osumi K (2003) Anthropogenic disturbances and the development of present cool-temperate forests in the Kitakami Mountain Range (in Japanese). Jpn J Hist Bot 11(2):53–59 Pinard MA, Barker MG, Tay J (2000) Soil disturbance and postlogging forest recovery on bulldozer paths in Sabah, Malaysia. For Ecol Manage 130:213–225 R Development Core Team (2005). R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna. ISBN 3-900051-07-0. http://www.R-project.org Sakai T, Tanaka H, Shibata M, Suzuki W, Nomiya H, Kanazashi T, Iida S, Nakashizuka T (1999) Riparian disturbance and community structure of a Quercus–Ulmus forest in central Japan. Plant Ecol 140:99–109 Sakio H (1997) Effects of natural disturbance on the regeneration of riparian forests in a Chichibu Mountains, Central Japan. Plant Ecol 132:181–195 Sakio H, Suzuki W (1997) Overview of riparian vegetation: Structure, ecological function and effect of erosion control works (in Japanese). J Jap Soc Erosion Cont Eng 49(6):40–48 Sakio H, Tamura T (eds) (2008) Ecology of riparian forests in Japan: disturbance, life history, and regeneration. Springer, Berlin, p 339 Sakio H, Kubo M, Shimano K, Ohno K (2002) Coexistence of three canopy tree species in a riparian forest in the Chichibu Mountains, central Japan. Folia Geobot 37:45–61 Suzuki W, Osumi K, Masaki T, Takahashi K, Daimaru H, Hoshizaki K (2002) Disturbance regimes and community structures of a riparian and an adjacent terrace stand in the Kanumazawa Riparian Research Forest, northern Japan. For Ecol Manage 157:285–301 Tadaki Y (1996) Forest environment science (in Japanese). Asakura, Tokyo Utsugi E, Kanno H, Ueno N, Tomita M, Saitoh T, Kimura M, Kanou K, Seiwa K (2006) Hardwood recruitment into conifer plantation in Japan: effects of thinning and distance from neighboring hardwood forests. For Ecol Manage 237:15–28 Wada M, Kikuchi T (2004) Emergence and establishment of seedlings of Japanese elm (Ulmus davidiana Planch var. japonica (Rehder) Nakai) on a flood plain along the Azusa River in Kamikochi, central Japan (in Japanese with English summary). Veg Sci 21:27–38 Whitman AA, Brokaw NVL, Hagan JM (1997) Forest damage caused by selection logging of mahogany (Swietenia macrophylla) in northern Belize. For Ecol Manage 92:87–96