Factors controlling vegetation succession in Kushiro Mire

e c o l o g i c a l m o d e l l i n g 2 1 5 ( 2 0 0 8 ) 225–236

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/ecolmodel

Factors controlling vegetation succession in Kushiro Mire Tadanobu Nakayama ∗ Asian Environment Research Group, National Institute for Environmental Studies (NIES), 16-2 Onogawa, Tsukuba, Ibaraki 305-8506, Japan

a r t i c l e

i n f o

a b s t r a c t

Article history:

The NIES integrated catchment-based eco-hydrology (NICE) model was expanded to include

Published on line 7 April 2008

the vegetation succession processes in the Kucyoro River catchment, Japan (NICE-VEG). The NICE-VEG simulated the water/heat budget and dynamic vegetation processes iteratively.

Keywords:

The simulation results indicated that the spatial occupation rate of alder invasion is posi-

Drying phenomenon

tively correlated with hydrogeological changes. Some discrepancies are attributable to local

Alder invasion

heterogeneity of changes in groundwater level, porosity, deposited sediments, and other

Mire shrinking

limiting factors (nutrients, soil moisture, light, temperature). Furthermore, simulation of

Two-species competition

channelized rivers showed that the recharge rate of Kushiro Mire decreases greatly. This

NICE-VEG

indicates that channelization will also cause an increase of sedimentation/nutrient load

Limiting factor

and flooding in the downstream area around the mire. The NICE-VEG reproduced excellently the invasion of alder in the mire over the last 30 years, which represents a dramatic advance in our understanding of the drying phenomenon associated with alder invasion. The reproducibility of these simulation results suggests that the NICE-VEG includes some of the important factors affecting vegetation succession in Kushiro Mire. © 2008 Elsevier B.V. All rights reserved.

1.

Introduction

Kushiro Mire (the largest mire in Japan) and the Kushiro River catchment (area: 2204.7 km2 located in northern Japan) (Fig. 1) (Digital National Land Information GIS Data of Japan, 1976, 1997) have undergone changes as a result of conversion to urban or agricultural use since 1884. Because of the channelization of meandering rivers in the northern part of Kushiro Mire in the 1970–1980s in order to smoothly drain runoff and protect farmlands from flooding, runoff containing nutrients from farmland and sediments from short-cut channels have flowed directly into the mire and deposited flood-borne sediment. The dominant species in this mire include alder (Alnus japonica), reed (Phragmites australis), moss (Polytrichum spp., Sphagnum spp.), sedge (Eriophorum vaginatum), willow (Salix spp.), Japanese ash (Fraxinus mandshurica var. japonica), and meadow sweet (Spiraea salicifolia). Alder has propagated widely

∗

Tel.: +81 29 850 2564; fax: +81 29 850 2584. E-mail address: [email protected]. 0304-3800/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.ecolmodel.2008.02.017

around Kushiro Mire since channelization, due mainly to lowering of the groundwater level and increased nutrient input, resulting in gradual shrinkage of the mire, as shown in Table 1 (Ministry of Environment, 2004). This drying phenomenon shows that human activities have changed the water cycle, and thus vegetation succession, in the mire (Nakayama and Watanabe, 2004, 2006). Some previous studies have investigated the environmental factors and primary succession of alder, mainly through field observations, field experiments, sampling and analysis. These have included: (i) the effect of environmental factors (soil texture, soil depth, litterfall, accumulation rate, root density, organic content, moisture, and nutrients) on primary succession (Chapin et al., 1994); (ii) experimental determination of relative growth rates under different relative addition rates and nitrogen uptake/fixation in seedlings (Burgess and Peterson, 1986); (iii) biomass increase about 1.2/1.7 times in

226

e c o l o g i c a l m o d e l l i n g 2 1 5 ( 2 0 0 8 ) 225–236

Table 1 – Vegetation change in the Kushiro Mire (Ministry of Environment, 2004)

Mire (A) Alder (B) Willow Forest Meadow Farm Residential Road Bare River&Lake Mire + Alder (C) A/C B/C

1947 (ha)

1977 (ha)

1996 (ha)

1947–1977 (ha/year)

1977–1996 (ha/year)

22,476 2,097 712 11,218 1,452 2,165 33 498 0 1,742 24,573 0.91 0.09

19,586 2,941 873 8,976 2,263 3,789 780 822 676 1,684 22,527 0.87 0.13

12,303 7,127 976 10,479 552 6,024 1,655 1,061 432 1,782 19,430 0.63 0.37

−96.3 28.1 5.4 −74.7 27.0 54.1 24.9 10.8 22.5 −1.9 −68.2

−383.3 220.3 5.4 79.1 −90.1 117.6 46.1 12.6 −12.8 5.2 −163.0

the presence of increased phosphorus/alkalinity (Ministry of Environment, 2004); (iv) growth conditions in submerged conditions, water level variations, pH, EC (electrical conductivity), and DO (dissolved oxygen) determined using Principal Component Analysis (PCA) (Yabe and Onimaru, 1997; Hotes et al., 2001); (v) oxygen uptake and nitrogen fixation based on chamber measurements and determinations conducted in the field (Hendrickson et al., 1990; Grosse et al., 1993); (vi) environmental factors related to reed and mire vegetation (Glaser et al., 1990; Wassen et al., 1990, 1995; Yabe and Onimaru, 1997). Because these previous studies allowed effective classification of the characteristics of alder and mire vegetation, it is powerful to combine these results with numerical simulation in order to reproduce the drying phenomenon in the mire, to evaluate the relationships among water, heat, nutrient, sediment and vegetation, and to simulate or forecast the influence of river channelization/meandering and

land-cover change on vegetation change downstream in the mire. The objective of the present study was to estimate the influence of river channelization and land-cover changes on vegetation change downstream in Kushiro Mire by adding the interrelationships among water, heat, nutrients, sediment, and vegetation. The author expanded the NICE (NIES Integrated Catchment-based Eco-hydrology) model series (Nakayama and Watanabe, 2004, 2006, 2007; Nakayama et al., 2006, 2007) to include the vegetation succession processes including competition between two species in the Kucyoro River catchment, Japan (NICE-VEG). The NICE-VEG simulated the water/heat budget and vegetation succession processes iteratively by adding the environmental limiting factor of controlling vegetation succession in relation to submerged depth, which is the newly developed part of this study. The author conducted a simulation to reproduce the alder inva-

Fig. 1 – Land cover and observation points of the study area (Kucyoro River catchment, a tributary of the Kushiro River catchment) in (a) 1976, and (b) 1997 (Digital National Land Information GIS Data of Japan). The simulation area in this figure is 22 km wide by 46 km long, covering the whole Kucyoro River catchment. The dark-blue line is the Kucyoro River, and the light-blue lines represent the Kushiro main river and other tributaries of the Kushiro River catchment. The black line is the border of the study area for vegetation succession simulation in the downstream area of the Kucyoro River catchment. In (b), the river has been channelized in the downstream area of the Kucyoro River.

227

e c o l o g i c a l m o d e l l i n g 2 1 5 ( 2 0 0 8 ) 225–236

sion that has occurred up to now, and evaluated the influence of river channelization and land-cover change on the vegetation changes that have occurred downstream in Kushiro Mire. The results showed that the new nature restoration project launched in this area would be very effective for recovery of the mire vegetation, which is now very seriously affected by the drying phenomenon associated with vegetation change caused by the invasion of alder.

2.

Study area

The Kucyoro River catchment (area: 123.0 km2 ), a tributary of the Kushiro River catchment, is located in Hokkaido, northern Japan (Fig. 1; Digital National Land Information GIS Data of Japan, 1976, 1997). The annual mean temperature is about 5–6 ◦ C, making it one of the coldest regions in Japan. Mean annual precipitation is about 1100 mm. In summer, the mean temperature is 17–19 ◦ C, and fog is common. Kushiro Mire, located in the downstream region of the Kucyoro River, has been protected under the Ramsar Convention since 1980 and was declared a national park in 1987. Nevertheless, the water cycle has recently changed and a drying phenomenon has occurred in the mire (Fig. 1), closely associated with vegetation change caused by inflow of increased sediment load from the surrounding areas due to river channelization, conversion of the surrounding areas to urban or agricultural use, and subsequent invasion of alder into the mire (Nakayama and Watanabe, 2004). These phenomena are closely related to river flooding, not only in the typhoon season but also during spring snowmelt (Nakayama and Watanabe, 2006). Some previous studies have investigated the responses of a gravel bed river to meander straightening and rectification (Brookes, 1985; Schilling and Wolter, 2000; Talbot and Lapointe, 2002). When reaches are shortened and slopes are steepened at meander cut-offs, increased sedimentation and flooding occur downstream (Talbot and Lapointe, 2002), causing drying of Kushiro Mire (Nakamura et al., 1997; Nakayama and Watanabe, 2004). In order to arrest sediment-load influx and allow recovery of the mire, the Japanese government started a new project in 2002, the Kushiro Mire Conservation Plan, to re-meander the channelized rivers in the catchment (Ministry of Environment, 2002), which included river restoration and assessment of ecological integrity (Jungwirth et al., 2002). Therefore, it is very important to assess the effect of river channelization on Kushiro Mire, in order to clarify the factors controlling species competition between alder, reeds, and willows, and to reproduce the alder invasion in the mire so far, in order to prevent sediment loading from riparian forests and to recover the mire in the future.

3.

Model description of NICE-VEG

3.1.

NICE model series

Up to now, the author has developed the NICE (NIES Integrated Catchment-based Eco-hydrology) model series (Nakayama and Watanabe, 2004, 2006, 2007; Nakayama et al., 2006, 2007), which includes surface-unsaturated–saturated water

processes and assimilates land-surface processes describing variations of LAI (leaf area index) and FPAR (fraction of photosynthetically active radiation) derived from MODIS satellite data (Fig. 2). LAI and FPAR are important parameters for evaluating vegetation growth (Justice et al., 1998). The NICE model connects several sub-models from the ground to the surface by considering water/heat fluxes, for example, (i) the gradient of hydraulic potential between the deepest layer of unsaturated flow and the groundwater level, (ii) effective precipitation calculated from the precipitation rate, infiltration of precipitation into the upper soil moisture store, and evapotranspiration rates, and (iii) seepage between river and groundwater. Details are have been described in Nakayama and Watanabe (2004).

3.2.

Vegetation succession model

The forest model used in this study is ZELIG (Urban et al., 1991, 1993; Cumming and Burton, 1994), a recent reformation of the basic JABOWA (Bonan, 1989) and FORET (Shugart and West, 1977) models. Although gap models have been applied to a variety of forests and differ in some details, they all share the same basic structure and logic (Urban and Shugart, 1992), simulating the establishment, annual diameter growth, and mortality of each tree. Studies comparing the simulation results of some models have shown that ZELIG is more accurate and suitable, despite having some limitations (Busing and Solomon, 2004). ZELIG simulates the establishment, annual diameter growth, and mortality of each tree on an array of model plots (Fig. 2). The primary zone of influence of a single canopy-dominant tree defines plot size. The plot is considered homogeneous horizontally, but vertical heterogeneity (canopy height and height to base of crown) is simulated in some detail. Adjacent cells interact through light interception at low sun angles. Seedling establishment, mortality, and regeneration are solved stochastically by using Monte-Carlo simulations, while the growth stage is largely deterministic. The competitive environment of the plot is defined by the height, leaf area, and woody biomass of each individual tree determined by allometric relationships with diameter as follows (Urban and Shugart, 1992) H = 137 + b2 D − b3 D2 ,

where b2 = 2 =

Hmax − 137 and b3 Dmax

Hmax − 137 D2max

(1)

where D (cm) is diameter at breast height (DBH) and H (cm) is height. The crown diameter CD (m) is derived in Eq. (2) (Garman, 2003). CD = exp[ln(D)b0 + b1 ].

(2)

The species coefficients were set as b0 = 0.5212634 and b1 = 0.6300169 for reed alder (Garman, 2003). CD was used to calculate the simulated predominant species at each mesh because vegetation cover of GIS data (Fig. 1), which was compared with the simulated value in the results of this study, was basically categorized from aerial photographs (Digital National Land Information GIS Data of Japan). The tree

228

e c o l o g i c a l m o d e l l i n g 2 1 5 ( 2 0 0 8 ) 225–236

Fig. 2 – Flow diagram of the NICE-VEG model in the expansion of the NICE model series (Nakayama and Watanabe, 2004, 2006; Nakayama et al., 2006). Components inside the double frame show environmental limiting factors for vegetation growth in Eqs. (4)–(10).

e c o l o g i c a l m o d e l l i n g 2 1 5 ( 2 0 0 8 ) 225–236

229

diameter growth is given in Eq. (3), assuming a logistic curve (JABOWA-derived gap model) and L = cD2 (Prentice and Leemans, 1990) where L is leaf area and c is a parameter (=0.160694):

Environment Preservation Foundation, 2001). In this study the author added the limiting factor by using previous data for reed (Ohmi Environment Preservation Foundation, 2001) and alder (Hotes et al., 2001):

dD GD(1 − DH/Dmax Hmax ) = dt (274 + 3b2 D − 4b3 D2 )

rr (SD )

(3)

where G is a growth rate scalar. The growth rate is calculated as a function of several growth reduction factors as described below. Several studies have pointed out that the multiplicative approach in ZELIG in the same way as that in JABOWA causes rapid convergence to zero and results in unrealistically low growth rates when many growth factors are considered (Bugmann, 1996, 2001; Bugmann et al., 1996; Yaussy, 2000). Though some studies have applied “Liebig’s Law of the Minimum” (Kienast, 1987; Bugmann, 1996, 2001) to use only the smallest of all the growth factors in view of these limitations, this approach is based on the unrealistic assumption that only the smallest factor limits tree growth and that one single environmental factor explains all the variability of tree growth during any given year. In this study, the author applied a stepwise procedure by Bugmann (1996) in order to satisfy the above two requirement for the environmental limiting factors r for light Qh , nutrients F, soil moisture M, temperature T, and submerged depth SD are given as follows (Urban and Shugart, 1992) G = Gmax [r(Qh )r(F)r(M)r(T)r(SD )]

1/3

(4)

r(Qh ) = c1 {1 − exp[−c2 (Qh − c3 )]}, where Qh = Q0 exp[−kL(h )]

(5)

r(F) = c4 + c5 F − c6 F2

(6)

r(M) =

r(T) =

 M ∗ −M 1/2

(7)

M∗ 4(T − Tmin )(Tmax − T) (Tmax − Tmin )

2

(8)

where Gmax is the maximum growth rate, Qh is the incident radiation at height h (1 = full sun), h is the height greater than h, k is a light extinction coefficient (=0.4), F is relative soil fertility (dimensionless on 0–1), M is soil moisture index (on 0–1), M* is maximum tolerable for a species, T is degree-day index (5.56 ◦ C base), Tmim and Tmax are minimum and maximum degree-day limits for species (Tmim = 1420, Tmax = 3084) used in this study, and SD (m) is submerged depth of vegetation from the ground surface. The available light Qh falls off negative-exponentially within the canopy according to the Beer–Lambert law (Urban and Shugart, 1992). The parameters ci (i = 1–6) are fitted constants dependent on shade-stress and nutrient-stress tolerance (c1 = 1.11, c2 = 2.52, c3 = 0.07, c4 = 0.2133, c5 = 1.789, c6 = −1.014). The author added the effect of submerged depth as a limiting factor r(SD ) into Eq. (4) because this effect is very different between growths of reed and alder (Hotes et al., 2001; Ohmi

ra (SD )

= 0.01 for SD < 0.0 or SD ≥ 0.8 = 0.01 + 4.95SD for 0.0 ≤ SD < 0.2 = 1.0 for 0.2 ≤ SD < 0.7 = 7.93 − 9.9SD for 0.7 ≤ SD < 0.8

(9)

= 1.0 for SD < 0.06 = 2.98 − 33.0SD for 0.06 ≤ SD < 0.09(= 0.06 + 2) . (10) = 0.01 for SD ≥ 0.09

The subscripts r and a in the above Eqs. (9) and (10) show reed and alder, respectively. Although the above Eq. (4) lacks a mechanistic basis, this approach yields intuitively reasonable results, probably superior to both the multiplicative approach and Liebig’s Law (Bugmann, 1996).

3.3.

Integration of models

Because the NICE-VEG simulates water/heat budget, mass transport (Itakura, 1984; Shimizu and Arai, 1988; Kushiro Branch Office, 2002; Toda et al., 2002), and vegetation succession processes iteratively (Fig. 2), it is possible to estimate the influence of river channelization/meandering and land-cover change on vegetation change downstream in Kushiro Mire by adding the relationships between water, heat, nutrients, sediment, and vegetation. It is very powerful to simulate the spatial and temporal variations of environmental factors controlling vegetation change, such as soil moisture, submerged depth, nutrient loading, and sediment accumulation, and to input the dynamic changes of these variables into the vegetation succession model as the boundary of model expansion (Fig. 2). In the first step, the model simulates water/heat values such as evapotranspiration, river flow discharge and depth, soil moisture, soil temperature, and groundwater level in each grid at each time step by inputting the meteorological forcing data. In the second step, the model simulates the vegetation succession processes by inputting the simulated results of the first step and the meteorological data. These simulations do not include feedback from vegetation change to climatic change because this study area is not so large, and therefore this process can be considered negligible.

4. Input data and boundary conditions for simulation 4.1.

Input data

The hourly observation data for downward short- and long-wave radiation, precipitation, atmospheric pressure, air temperature, air humidity, and wind speed at a reference level were calculated from the AMeDAS (Automated Meteorological Data Acquisition System) meteorological data (Japan Meteorological Agency, 1970–2003) obtained at Tsurui observation station in the catchment (Fig. 1; 43◦ 13 48 N, 144◦ 19 30 E,

230

e c o l o g i c a l m o d e l l i n g 2 1 5 ( 2 0 0 8 ) 225–236

mean elevation 42 m) collected by the Japan Meteorological Business Support Center for forcing simulation data, and these values were input at each grid because the study area is not large. Mean elevation data (Geographical Survey Institute of Japan, 1999), soil texture data (Hokkaido National Agricultural Experiment Station, 1985), vegetation class data (Environment Agency of Japan, 1993) and land cover data (Digital National Land Information GIS Data of Japan, 1976, 1997) were converted and input to the NICE-VEG with a resolution of 100 m. The simulation area in the vertical direction was divided into 20 layers with a weighting factor of 1.1 (finer at the upper layers). The upper layer was set at 2 m depth, and the 20th layer was defined as an elevation of −200 m from sea level. Geological structure was divided into four types on the basis of hydraulic conductivity (Kh and Kv ), the specific storage of porous material (Ss ), and specific yield (Sy ) by using soil samples taken at two depths (0.1 and 1.0 m) and previous data from 150 sample data points in the Kushiro River catchment (Ohara et al., 1975). Kushiro Mire consists largely of soils finer than silt, mainly peat. Details have been presented in Nakayama and Watanabe (2004). For the upstream boundaries, reflecting conditions on the hydraulic head were used assuming that there is no inflow from the mountains in the opposite direction. The hydraulic head values parallel to the ground level were input as the initial conditions for the groundwater flow model. For the model simulating hillslope hydrology, the flow depth and the discharge at the uppermost ridges of the mountains were set as zero throughout the simulation. In river cells, outflows from riverbeds of −1 m mean elevation from the ground surface were considered in the same way as in the study by Nakayama and Watanabe (2006). Important data for alder and reed in this study input to the NICE-VEG were summarized from previous studies (Burgess and Peterson, 1986; Chapin et al., 1994; Hotes et al., 2001; Ohmi Environment Preservation Foundation, 2001; Ministry of Environment, 2004) (Table 2). Because previous research has shown that the nutrient load to the mire has increased about 20% during the past 30 years (Ministry of Environment, 2004), the author increased the environmental limiting factor for soil fertility in Eq. (6) from 0.5 to 1.0 linearly by considering the relative growth rates (Burgess and Peterson, 1986).

4.2.

Observed data

The author set groundwater level meters at 10 points, flow depth meters at 5 points, and a water sampler/turbidity sensor at 1 point around the catchment for model validation. Measurements were made over a 3-year period in 2000–2002. The water level (KADEC, MIZU-II) was automatically recorded to data-loggers at hourly intervals. During winter, when the water level meters were not set up, river discharge data (supplied by Hokkaido Regional Development Bureau) at one point were used as the observed data. Details have been described by Nakayama and Watanabe (2004, 2006). The water turbidity concentration and water temperature (CTI, C105) were automatically recorded to data-loggers at 10-min intervals from July 2002 to June 2003. Furthermore, samples of the turbid water below the water surface were

automatically taken using sample bottles (ISCO, Model-3700) at hourly intervals six times during periods of high precipitation (July, August, October 2002, and May, June, July 2003) in order to validate the water turbidity concentration and observe the various nutrient concentrations (total phosphorus and total nitrogen).

4.3.

Running the simulation

The simulation area for water/heat budget and mass transport was 22 km wide by 46 km long, covering the whole Kucyoro River catchment (Fig. 1). This area is discretized into a grid of 226 × 460 blocks with a grid spacing of 100 m, which is sufficiently fine to describe the shapes of meandered channels. The time-step in the water/heat budget simulation was t = 30 min in order to facilitate numerical stability. The simulation was conducted on an NEC SX-6 supercomputer for the period 1970 (meandering channel, Fig. 1a) to 2003 (present channelized river, Fig. 1b) by using the same data for soil texture, vegetation class, geological structure, and the different shapes of the river channels. The first 6 months were used as a warm-up period until equilibrium conditions were reached, and parameters were estimated by comparison of simulated results using the observed values published in the literature. After simulation results for water/heat budget such as river discharge, soil moisture, groundwater level, and soil temperature, had been calibrated and validated by the observed data for 2001–2002 (Nakayama and Watanabe, 2004, 2006), the simulation was able to back-cast the water/heat budget during 1970–2000. Then, the mass transport was simulated at the hillslope and the river (Itakura, 1984; Shimizu and Arai, 1988; Kushiro Branch Office, 2002; Toda et al., 2002). The time-steps used in the hillslope and stream network models were t = 50 and 10 s in order to facilitate numerical stability. The simulated results of nutrient and sediment loads were validated by the observed data for 2002–2003. The simulation area for vegetation succession was 5 km wide by 13 km long in the downstream section of the Kucyoro River in the mire (Fig. 1). Mesh size was 10 m × 10 m, which was sufficient to represent a vegetation colony of plot size (Shinsho et al., 1988; Shinsho and Tsujii, 1996). Simulated results for soil moisture, groundwater level, submerged depth, nutrient loading, and sediment accumulation averaged at every month were input to the vegetation succession model in order to simulate vegetation growth for every year after a warm-up simulation of 300 years (Fig. 2). The simulated tree height was calibrated using values obtained in previous studies. Finally, the simulation of vegetation succession was conducted for the period 1970–2003, and the results were compared with the vegetation distribution of GIS data (Fig. 1), as described in the following section.

5.

Results

5.1. mire

Validation of ecohydrological characteristics in the

The simulated groundwater level was compared and validated by the observed value combining the NIES observation

231

e c o l o g i c a l m o d e l l i n g 2 1 5 ( 2 0 0 8 ) 225–236

Table 2 – Summarized inputted data to the NICE-VEG about alder and reed from the previous researches (Burgess and Peterson, 1986; Chapin et al., 1994; Hotes et al., 2001; Ohmi Environment Preservation Foundation, 2001; Ministry of Environment, 2004) Alder

Reed

References

Soil texture (% of total) Sand Silt Clay

64.8 ± 1.9 24.7 ± 1.5 10.5 ± 0.7

74.6 ± 4.3 15.8 ± 2.8 9.6 ± 1.8

Chapin et al. (1994) Chapin et al. (1994) Chapin et al. (1994)

Soil depth (cm/horizon) Litter Organic horizon A + B horizons

1.7 ± 0.2 2.8 ± 0.8 8.8 ± 0.2

0 0 5.2 ± 0.5

Chapin et al. (1994) Chapin et al. (1994) Chapin et al. (1994)

Litterfall (g/(m2 year)) Fine litter Wood Total litter

203–260 45–47 248–307

1.3–1.8 0 1.3–1.8

Chapin et al. (1994) Chapin et al. (1994) Chapin et al. (1994)

Biomass for limiting factor of seedlings growth (mg/plant) Control 298 N added 823 P added 1110 Survivorship of seedlings (% of total) 81

Chapin et al. (1994) Chapin et al. (1994) Chapin et al. (1994) Chapin et al. (1994)

Seedling growth rate Height growth (cm) Total biomass (g/plant) Aboveground production (g/(plant year)) Relative growth rate (g/(g year)) Height increment after 3-year (cm)

3.28–5.45 3.2–3.8 0.3–0.8 0.3–0.8 2.1–9.2

Chapin et al. (1994) Chapin et al. (1994) Chapin et al. (1994) Chapin et al. (1994) Chapin et al. (1994)

(Noninoculated) relative growth rates 5% N added 10% N added 15% N added

3.8–4.9 9.2–10.1 15.1–17

Burgess and Peterson (1986) Burgess and Peterson (1986) Burgess and Peterson (1986)

(Inoculated) relative growth rates 0% N added 5% N added 10% N added 15% N added

5.2–8.0 6.2–7.6 10.3–13.2 14.3–16.8

Burgess and Peterson (1986) Burgess and Peterson (1986) Burgess and Peterson (1986) Burgess and Peterson (1986)

Submerged depth for living (cm) Diameter at breast height (m)

Eq. (10) 2–16 (Peak:4–6)

Effect of P on living Monitoring site (total drying weight; g) P added (total drying weight; g)

0.05 0.067

Eq. (9)

data (Nakayama and Watanabe, 2004, 2006) and the scanned and digitized data from the previous research (Ministry of Environment, 2004) in the mire at the downstream area of the Kucyoro River in 2001–2002 (Fig. 3). The relative groundwater level takes a minus value downstream of the channelized river because sediment accumulation by coarser materials has increased the surface elevation (Nakamura et al., 1997; Nakayama and Watanabe, 2004). The groundwater level recovers further downstream of the channelized river because most of the sediment is deposited in the upper area. The annualaveraged simulation result reproduces these characteristics very well in the Kucyoro River catchment, although the simulated value overestimates the observed data in the Kottaro River catchment (in the northeastern area in this figure). This simulated distribution is very similar to that observed in a previous study by Nakayama and Watanabe (2004), which directly input changes in meteorological forcing data and vegetation classes (Fig. 1) between 1977 and 2001.

Hotes et al. (2001), Ohmi Environment Preservation Foundation (2001) Ministry of Environment (2004)

Ministry of Environment (2004) Ministry of Environment (2004)

The simulated tree height H (cm) was validated by values obtained in previous studies (Shinsho et al., 1988; Shinsho and Tsujii, 1996) (Fig. 4). The observed data are more scattered because alder thickets show different types of zonation and vegetation growth in natural river areas (with germinated alder thickets showing similar height and DBH) and improved river areas (with alder thickets raised from seed showing different height and DBH) (Shinsho and Tsujii, 1996). The simulated value depends greatly on the maximum tree height Hmax (m), and the following simulation uses Hmax = 9 (m) on the assumption that there are no differences in the types of zonation and vegetation growth of alder thickets in the mire.

5.2.

Evaluation of hydrogeological changes in the mire

Hydrogeological changes from the 1970 to 2000s were evaluated by long-term simulation (Fig. 5). The changes in elevation

232

e c o l o g i c a l m o d e l l i n g 2 1 5 ( 2 0 0 8 ) 225–236

Fig. 3 – Averaged groundwater level relative to ground surface in the mire around the downstream area of the Kucyoro River in 2001–2002: (a) observed value combining the NIES observation data (Nakayama and Watanabe, 2004, 2006) and the scanned and digitized data from the previous research (Ministry of Environment, 2004); and (b) the value simulated by NICE-VEG. The red colored region shows the higher submerged area. The white cell indicates the meandering river before channelization in 1976, and the black cell the straightened river after channelization in 1997, as shown in Fig. 1.

from during this period were evaluated using the GIS-database (Hokkaido Regional Development Bureau, 1975; Geographical Survey Institute of Japan, 1999) (Fig. 5a). During this 30-year period, the elevation increases predominantly in the downstream areas of rivers flowing into the mire (Kushiro main river and tributaries; Kucyoro River, Chiruwatsunai River, Setsuri River, Ashibetsu River, Hororo River, Onnenai River, etc.) particularly around the downstream area of the Kucyoro River, which indicates that the increase of sediment delivery has caused morphological changes in the mire (Nakamura et al., 1997; Nakayama and Watanabe, 2004). The NICE-VEG simulation shows that the groundwater level has decreased predominantly around the channelized rivers and the downstream area in the mire (Fig. 5b). The simulation result shows that the maximum value of groundwater degradation is more than 1 m. The increase of river discharge caused by channelization results in a decrease of seepage infiltration from the river to the aquifer, which accounts for the decrease in the groundwater level downstream of the channelized rivers. The change in groundwater level relative to the ground surface (Fig. 5c) was calculated by summation of both the elevation change (Fig. 5a) and the absolute groundwater level change (Fig. 5b). It was assumed that the alder invasion (Fig. 1) and the decrease in groundwater level relative to the ground surface (Fig. 5c) have a close correlation.

Fig. 4 – Comparison between simulated tree height H (cm) and observed value versus diameter at breast height (DBH). Lines are results of NICE-VEG simulation, and dots are the previously observed values (Shinsho et al., 1988; Shinsho and Tsujii, 1996). The Onnenai and Chiruwatsunai Rivers are outside the Kucyoro River catchment, and their data are also plotted as a reference to show the spatial heterogeneity of alder growth.

5.3. Reproduction of the drying phenomenon and alder invasion in the mire The long-term NICE-VEG simulation for the period 1970–2002 reproduces qualitatively and quantitatively the invasion of alder in the mire (Fig. 6a). The GIS data show that reed was predominant in 1976 (proportion of reed 85.3%; that of alder 14.7%), and that alder spread in 1997 (proportion of reed 52.9%; that of alder 47.1%) (Fig. 1), and this was reproduced well by the

e c o l o g i c a l m o d e l l i n g 2 1 5 ( 2 0 0 8 ) 225–236

233

Fig. 5 – Hydrogeological changes from 1970 to 2000s: (a) elevation change evaluated by using the GIS-database (Hokkaido Regional Development Bureau, 1975; Geographical Survey Institute of Japan, 1999); (b) simulated result for groundwater level change (above sea level) after river channelization; and (c) estimated value of groundwater level change (relative to ground surface). The white cell indicates the meandering river and the black cell the channelized river.

NICE-VEG simulation for both 1976 (proportion of reed 84.0%; that of alder 16.0%) and 1997 (67.0 and 33.0%, respectively) (Fig. 6a). The model forecasts that reed would be predominant and that alder would not have invaded the mire in 1997 if the river had not been channelized in the 1970s (dotted line in the figure; proportion of reed 85.0%; that of alder 15.0%). The NICE-VEG also simulated the spatial distribution of alder invasion up to the present time (Fig. 6b and c). Invasion of alder is very sensitive to submerged depth, nutrient input, and other factors. The simulation result indicated that

the degradation of groundwater level relative to ground surface and submerged depth are the most predominant factors associated with mire shrinkage, which is greatly affected by river channelization (Talbot and Lapointe, 2002) (Fig. 5). The increase in elevation agrees quantitatively with a previous in situ study using Cs-137 fallout concentration analysis, which indicated that the aggradation was about 2 m at the downstream end of the channelized reach of the Kucyoro River (Nakamura et al., 1997; Mizugaki et al., 2006). Although the simulation result for alder invasion in 1997 underestimates the GIS data (Fig. 1b) at the center of the mire, the simulation

Fig. 6 – Results of simulation of alder invasion in the mire: (a) proportion of alder and reed during 1970–2000, (b) and (c) spatial distribution of alder and reed in 1976 and 1997, respectively. In (a), the lines are actual simulated results for when the rivers were channelized in the 1970–1980s (solid line: alder, bold line: reed), and the area represented by dotted lines represents the simulated results that would have been expected if the rivers had not been channelized in the past (solid dotted-line: alder, bold dotted-line: reed). Circles and triangles represent the proportions of alder and reed calculated from the GIS data (Fig. 1). In (b) and (c), the dark-blue cell represents the Kucyoro River. In (c), the river has been channelized in its downstream area. The simulated results in (b) and (c) reproduce well the GIS data shown in Fig. 1(a) and (b), respectively.

234

e c o l o g i c a l m o d e l l i n g 2 1 5 ( 2 0 0 8 ) 225–236

values reproduce qualitatively the spatial distribution of alder invasion, similar to the GIS data (Fig. 1).

6.

Discussion

The drying phenomenon in Kushiro Mire (Fig. 1) is closely related to the increased influx of sediments from the surrounding area, where agricultural development, reclamation, and channelization of the river occurred (Nakayama and Watanabe, 2004). The spatial occupation rate of alder invasion is positively correlated with hydrogeological changes (Fig. 5). Some discrepancies in Fig. 6 are attributable to local changes in groundwater level (Nakayama and Watanabe, 2004), heterogeneity of porosity due to the particle size distribution of deposited sediments (Nakamura et al., 1997), and other limiting factors (nutrients, soil moisture, light, temperature) (Urban and Shugart, 1992). The higher nutrient concentration in the groundwater has also affected the vegetation change in the mire (Ministry of Environment, 2004). Furthermore, it was noteworthy that the simulation of river channelization showed that the recharge rate in the mire decreased greatly (Fig. 5). This indicates that channelization will also cause an increase of sedimentation/nutrient loading and flooding in the downstream area around the mire (Talbot and Lapointe, 2002). The NICE-VEG reproduced very accurately the invasion of alder in the mire over the last 30 years (Fig. 6) in comparison with the vegetation cover derived from GIS data (Fig. 1). This result represents a dramatic advance in our understanding of the drying phenomenon associated with alder invasion. The reproducibility of the simulation result implies that the NICE-VEG includes some of the important factors associated with vegetation succession in the mire. In particular, the model clarified that the change in submerged depth (Yabe and Onimaru, 1997; Hotes et al., 2001) is one of the crucial factors effecting alder invasion, accompanied by aggradation of sediment flowing in from the surrounding catchments due to urban or agricultural land use (Nakamura et al., 1997; Nakayama and Watanabe, 2004). Collecting data for various conditions affecting mire species such as alder, reed, moss, sedge, willow, Japanese ash, and meadow sweet, are important for reproducing more correctly the actual vegetation change. These would include hydrology (water level fluctuations; Hotes et al., 2001), water quality (nitrogen, phosphate, potassium, pH, electrical conductivity, dissolved oxygen; Burgess and Peterson, 1986; Chapin et al., 1994; Yabe and Onimaru, 1997) (Table 2), radiation and temperature, in order to simulate/forecast the competition among mixtures of plant species and to preserve biodiversity (UNEP, 2002) in the mire with high accuracy (Fig. 1). It is further necessary to clarify interactions among water, heat, sediment, nutrients and vegetation using statistical approaches such as canonical correspondence analysis and cluster analysis in order to evaluate more favorable environmental conditions for the mire and to reproduce the invasion of alder. Alder is distributed as thickets in the mire, which show differences in zonation and vegetation growth (tree height, crown diameter, and DBH) along the river, in the mire, and between the natural river and the improved river (Shinsho and Tsujii, 1996; Ministry of Environment, 2004). Germinated alder thick-

ets that consist of trees of similar height and DBH, together with bushes of meadow sweet on the forest floor, are found in natural rivers because of limited sediment supply. An alder thicket raised from seed, which consists of trees of different heights and DBH, together with reed and sedge on the forest floor, is found along improved rivers together with small-scale germinated alder thickets, because the sediment supply is frequent and the soil layer resulting from sediment accumulation is thick (Nakamura et al., 1997; Mizugaki et al., 2006). It is necessary to include these types of heterogeneity in Eqs. (1)–(10) of NICE-VEG as a function of hydrology, geology, and nutrient conditions in order to reproduce more accurately the alder invasion at the center of the mire (Fig. 6). The characteristics of alder thickets are closely associated with soil conditions (sand and silt, gravel, organic content, peat soil, etc.). Analysis of isotopes such as Cs-137, Ru-103, Ba-140, Be-7, and Th-232 is also an attractive approach for estimating the thickness of sediment accumulation (Mizugaki et al., 2006). It is further necessary to clarify the relationship between seedling establishment/mortality/regeneration and the expansion of alder distribution by both airborne and waterborne alder seeds. This is closely related to not only the above zonation heterogeneity but also the spatial localization of alder growth. Alder usually grows above the local mound of the mire in the shallow groundwater (Fig. 3). In some parts, this phenomenon is included as a limiting factor for submerged depth in Eqs. (9) and (10) (Yabe and Onimaru, 1997; Hotes et al., 2001; Ohmi Environment Preservation Foundation, 2001). Because the spatial scale of this mound is often smaller than the grid spacing of the NICE-VEG simulation (100 m), it will be necessary to simulate it by using a finer grid in the future. The NICE-VEG can also include the effect of airborne seedling distribution by using the forcing meteorological data for wind speed or coupling with the atmospheric model. This is also related to the differences in evapotranspiration between alder and other mire vegetation species such as reed. Investigations of the genetic diversity and population structure of alder (Huh, 1999) are very powerful for specifying the source and process of alder growth in combination with ground truth data in addition to plant maps, aerial photographs, and satellite data by using GPS (Global Positioning System) and GIS. In order to stop sediment loading and nutrient influx, and to assist the recovery of Kushiro Mire, several ministries in Japan initiated a new project in 2002 to re-meander the channelized rivers, to establish a riparian buffer, and to create a sediment retention pond (Ministry of Environment, 2002). This Japanese project will be very informative for clarifying certain aspects of the controversy associated with river restoration in the United States (Palmer and Bernhardt, 2006), because US practitioners rarely consider vegetation in their restoration efforts. For effective policy-making, it is necessary to forecast the effects of re-meandering of channelized rivers and to assess whether reed will recover if the current channelized rivers are re-meandered and sediment/nutrient loads are reduced by using the NICE-VEG, as currently there is a very serious drying phenomenon associated with the vegetation change caused by alder invasion (Ministry of Environment, 2002). Possible mitigation measures to preserve the mire would include the establishment of ripar-

e c o l o g i c a l m o d e l l i n g 2 1 5 ( 2 0 0 8 ) 225–236

ian forests along the river to filter suspended sediment from agricultural land (Jungwirth et al., 2002), to create channels or side-channel ponds for retaining suspended sediment, and to drain water into abandoned channels to dissipate stream power (Nakamura et al., 1997).

Acknowledgements Some of the simulations in this study were run on an NEC SX-6 supercomputer at the Center for Global Environment Research (CGER), NIES. Dr. M. Watanabe, Keio University, Japan, was helpful in many discussions about the model construction and the drying phenomenon in Kushiro Mire. Furthermore, the author would like to thank the Kushiro Development and Construction Department, Hokkaido Regional Development Bureau, for providing the discharge and agricultural data for the Kushiro River catchment.

references

Bonan, G.B., 1989. A computer model of the solar radiation, soil moisture, and soil thermal regimes in boreal forests. Ecol. Model. 45, 275–306. Brookes, A., 1985. River channelization: traditional engineering methods, physical consequences and alternative practices. Prog. Phys. Geogr. 9, 44–73. Bugmann, H.K.M., 1996. A simplified forest model to study species composition along climate gradients. Ecology 77 (7), 2055–2074. Bugmann, H.K.M., 2001. A review of forest gap models. Climatic Change 51, 259–305. Bugmann, H.K.M., Yan, X., Sykes, M.T., Martin, P., Lindner, M., Desanker, P.V., Cumming, S.G., 1996. A comparison of forest gap models: model structure and behaviour. Climatic Change 34, 289–313. Burgess, D., Peterson, R.L., 1986. Effect of nutrient conditions on root nodule development in Alnus japonica. Can. J. Bot. 65, 1658–1670. Busing, R.T., Solomon, A.M., 2004. A comparison of forest survey data with forest dynamics simulators FORCLIM and ZELIG along climatic gradients in the Pacific Northwest. U.S. Geological Survey of Scientific Investigations Report 2004-5078, 11. Chapin, F.S., Walker, L.R., Fastie, C.L., Sharman, L.C., 1994. Mechanisms of primary succession following deglaciation at Glacier Bay, Alaska. Ecol. Monogr. 64 (2), 149–175. Cumming, S.G., Burton, P.J., 1994. Zelig++ v0.9 documentation, user notes and installation guide. Univ. British Columbia, Canada, http://www.wiz.unikassel.de/model db/mdb/ftpmodels/zelig/zelig.tar.gz. Digital National Land Information GIS Data of Japan, 1976. Database of land cover in Japan. Ministry of Land Infrastructure and Transport of Japan. http://nlftp.mlit.go.jp/ksj/. Digital National Land Information GIS Data of Japan, 1997. Database of land cover in Japan. Ministry of Land Infrastructure and Transport of Japan. http://nlftp.mlit.go.jp/ksj/. Environment Agency of Japan, 1993. Vegetation class data in Hokkaido, Japan. Japan Integrated Biodiversity Information System. http://www.biodic.go.jp/J-IBIS.html. Garman, S.L., 2003. LandMod 2.0 documentation. Oregon State Univ., http://www.fsl.orst.edu/lter/pubs/webdocs/models/ land2doc.pdf.

235

Geographical Survey Institute of Japan, 1999. Digital map 50 m grid (elevation), Nippon-II (CD-ROM). Glaser, P.H., Janssens, J.A., Siegel, D.I., 1990. The response of vegetation to chemical and hydrological gradients in the Lost River peatland, northern Minnesota. J. Ecol. 78, 1021–1048. Grosse, W., Schulte, A., Fujita, H., 1993. Pressurized gas transport in two Japanese alder species in relation to their natural habitats. Ecol. Res. 8, 151–158. Hendrickson, O.Q., Fogal, W.H., Burgess, D., 1990. Growth and resistance to herbivory in N2 -fixing alders. Can. J. Bot. 69, 1919–1926. Hokkaido National Agricultural Experiment Station, 1985. Soil map of arable land in Hokkaido, scale 1:600,000, Hokkaido, Japan. Hotes, S., Poschlod, P., Sakai, H., Inoue, T., 2001. Vegetation, hydrology, and development of a coastal mire in Hokkaido, Japan, affected by flooding and tephra deposition. Can. J. Bot. 79, 341–361. Huh, M.K., 1999. Genetic diversity and population structure of Korean alder (Alnus japonica; Betulaceae). Can. J. Forest Res. 29, 1311–1316. Itakura, T., 1984. Investigations of some turbulent diffusion phenomena in rivers. Research Report of Civil Engineering Research Institute of Hokkaido 83, 1–91. Japan Meteorological Agency (JMA), 1970–2003. AMeDAS (Automated Meteorological Data Acquisition System) Annual Reports at 1970–2003, Japan Meteorological Business Support Center (CD-ROM). Jungwirth, M., Muhar, S., Schmutz, S., 2002. Re-establishing and assessing ecological integrity in riverine landscapes. Freshwater Biol. 47, 867–887. Justice, C.O., Vermote, E., Townshend, J.R.G., et al., 1998. The moderate resolution imaging spectroradiometer (MODIS): land remote sensing for global change research. IEEE T. Geosci. Remote 36, 1228–1249. Kienast, F., 1987. FORECE—A Forest Succession Model for Southern Central Europe. Oak Ridge National Laboratory, Oak Ridge, Tennessee, ORNL/TM–10575. Kushiro Branch Office, 2002. Agriculture of Kushiro in 2002. Hokkaido Prefecture (in Japanese). Ministry of Environment, 2002. Law for the Promotion of Nature Restoration, 148. http://www.env.go.jp/nature/saisei/law-saisei/law e.pdf. Ministry of Environment, 2004. Nature restoration project in Kushiro Shitsugen Wetland. http://www.kushiro.env.gr.jp/saisei/english/top e.html. Mizugaki, S., Nakamura, F., Araya, T., 2006. Using dendrogeomorphology and 137Cs and 210Pb radiochronology to estimate recent changes in sedimentation rates in Kushiro Mire, Northern Japan, resulting from land use change and river channelization. Catena 68, 25–40, doi:10.1016/j.catena.2006.03.014. Nakamura, F., Sudo, T., Kameyama, S., Jitsu, M., 1997. Influences of channelization on discharge of suspended sediment and wetland vegetation in Kushiro Marsh, northern Japan. Geomorphology 18, 279–289. Nakayama, T., Watanabe, M., 2004. Simulation of drying phenomena associated with vegetation change caused by invasion of alder (Alnus japonica) in Kushiro Mire. Water Resour. Res. 40 (8), W08402, doi:10.1029/2004WR003174. Nakayama, T., Watanabe, M., 2006. Simulation of spring snowmelt runoff by considering micro-topography and phase changes in soil layer. Hydrol. Earth Syst. Sci. Discuss. 3, 2101– 2144. Nakayama, T., Watanabe, M., 2007. Early View. Missing role of groundwater in water and nutrient cycles in the shallow eutrophic Lake Kasumigaura. Jpn. Hydrol. Process. 22 (8), 1150–1172, doi:10.1002/hyp.6684.

236

e c o l o g i c a l m o d e l l i n g 2 1 5 ( 2 0 0 8 ) 225–236

Nakayama, T., Yang, Y., Watanabe, M., Zhang, X., 2006. Simulation of groundwater dynamics in North China Plain by coupled hydrology and agricultural models. Hydrol. Process. 20 (16), 3441–3466, doi:10.1002/hyp.6142. Nakayama, T., Watanabe, M., Tanji, K., Morioka, T., 2007. Effect of underground urban structures on eutrophic coastal environment. Sci. Total Environ. 373 (1), 270–288, doi:10.1016/j.scitotenv.2006.11.033. Ohara, T., Matsushita, K., Futamase, K., et al., 1975. Explanatory text of the hydrogeological maps of Hokkado–Kushiro. Geological Survey of Hokkaido. Ohmi Environment Preservation Foundation, 2001. Research about vegetation condition for reclamation of reed in Lake Biwa, Research Report. http://nippon.zaidan.info/jigyo/2001/0000024546/jigyo info.html (in Japanese). Palmer, M.A., Bernhardt, E.S., 2006. Hydroecology and river restoration: ripe for research and synthesis. Water Resour. Res. 42, W03S07, doi:10.1029/2005WR004354. Prentice, I.C., Leemans, R., 1990. Pattern and process and the dynamics of forest structure: a simulation approach. J. Ecol. 78, 340–355. Schilling, K.E., Wolter, C.F., 2000. Application of GPS and GIS to map channel features in Walnut Creek, Iowa. J. Am. Water Resour. Assoc. 36, 1423–1434. Shimizu, Y., Arai, N., 1988. Numerical simulation of flow and bed variations in the river mouth region. Research Report of Civil Engineering Research Institute of Hokkaido 419, 5–36. Shinsho, H., Tsujii, T., 1996. Note on the alder thickets – Alnus japonica Steud – in Kushiro Marsh, eastern Hokkaido V. Research Report of Kushiro City Museum 20, 23–29 (in Japanese). Shinsho, H., Tsujii, T., Fujita, Y., 1988. Note on the alder thickets – Alnus japonica Steud – in Kushiro Moor, eastern Hokkaido III. Research Report of Kushiro City Museum 13, 25–34 (in Japanese). Shugart, H.H., West, D.C., 1977. Development of an Appalachian deciduous forest succession model and its application to

assessment of the impact of the chestnut blight. J. Environ. Manage. 5, 161–179. Talbot, T., Lapointe, M., 2002. Modes of response of a gravel bed river to meander straightening: the case of the Sainte-Marguerite River, Saguenay Region, Quebec, Canada. Water Resour. Res. 38 (6), 1073, doi:10.1029/2001WR000324. Toda, H., Nakasa, N., Hirano, K., Uemura, Y., Okino, T., Kawashima, H., 2002. Nitrogen cycling in the watershed of the Chikuma River. J. Jpn. Agric. Syst. Soc. 18 (2), 90–99 (in Japanese). UNEP (United Nations Environment Programme), 2002. The convention of biological diversity. http://www.biodiv.org/default.shtml. Urban, D.L., Shugart, H.H., 1992. In: Glenn-Lewin, D.C., Peet, R.K., Veblen, T.T. (Eds.), Individual-Based Models of Forest Succession. Plant Succession: Theory and Prediction. Chapman & Hall, London, pp. 249–292. Urban, D.L., Bonan, G.B., Smith, T.M., Shugart, H.H., 1991. Spatial applications of gap models. Forest Ecol. Manage. 42, 95–110. Urban, D.L., Harmon, M.E., Halpern, C.B., 1993. Potential response of Pacific Northwestern forests to climatic change, effects of stand age and initial composition. Climatic Change 23, 247–266. Wassen, M.J., Barendregt, A., Palczynski, A., de Smidt, J.T., de Mars, H., 1990. The relationship between fen vegetation gradients, groundwater flow and flooding in an undrained valley mire at Biebrza, Poland. J. Ecol. 78, 1106–1122. Wassen, M.J., Olde Venterink, H., de Swart, E., 1995. Nutrient concentrations in mire vegetation as a measure of nutrient limitation in mire ecosystems. J. Veg. Sci. 6, 5–16. Yabe, K., Onimaru, K., 1997. Key variables controlling the vegetation of a cool-temperate mire in northern Japan. J. Veg. Sci. 8, 29–36. Yaussy, D.A., 2000. Comparison of an empirical forest growth and yield simulator and a forest gap simulator using actual 30-year growth from two even-aged forests in Kentucky. Forest Ecol. Manage. 126, 385–398.