Chapter 4BIOMASS PRODUCTION IN RESPONSE TO POST-HURRICANE ENVIRONMENTAL GRADIENTS
Awake, the land is scattered with light, and see, Uncanopied sleep is flying from field and tree: And blossoming boughs of April in laughter shake Robert Bridges The thirsty earth soaks up the rain, And drinks, and gapes for drink again. The plants suck in the earth, and are With constant drinking fresh and fair. Abraham CowleyINTRODUCTION
In this chapter I assess the predictive ability of a gradient space, defined by primary gradients of soil moisture and solar radiation and the secondary gradient of hurricane damage, in determining rate of biomass accumulation. In addition, I examine the question: are species' energy profits maximized at the center of their distribution along gradients of abiotic environmental factors? Although the LEF is well studied, we have little information about exact values of the primary gradients, solar radiation, soil moisture, and temperature, across the LEF landscape. Therefore, secondary gradients and simulation models are used to approximate values of these primary gradients, as was suggested by Hall et al. (1992a). The resulting position or any geographic location in the gradient space is dependent on interactions among physical position in the landscape (elevation, topographic position, slope and aspect) and biotic controls (vegetation community, structure of the canopy), (Waide & Lugo 1992, see also Figure 3, Preface). I use the simulation models TOPOCLIM (Everham et al. 1991, Everham & Wooster 1991), and GEOPLT (Hall et al. 1992b) to simulate average solar radiation and soil moisture values, respectively. These two values are used to create a gradient space in which the BEW plots are located. Hurricane damage can be quantified as percent basal area damaged, or by examining the community composition (quantified as Average Pioneer Index API) of the plot following hurricane damage. Simulated solar radiation levels at the top of the canopy and hurricane damage are secondary gradients that are assumed to be correlated with the actual position of a plot in the primary gradient of solar radiation. I derived biomass changes from measured increases in diameter and using species-specific algorithms developed by Scatena et al. (1993) for the BEW and a general formula for the tabonuco forest (Weaver & Gillespie 1992). As with the analysis of gradients of hurricane damage that were used to predict recovery vectors, (Chapter 3), I developed isopleths of biomass increase within this gradient. The applicability of these isopleths is tested by repeating the analysis for plots from the HRP. In this case, each sample plot is placed within the gradient space and the resulting position in recovery isopleths is checked against measured rates of biomass increase. I did not measure rates of physiological processes, so the question of net energy profit could not be examined directly. I assumed that biomass increase reflects energy profits, and I examined the positions of the isopleths along the gradients to answer the question of whether maximum energy profits occur in the middle of the gradient. Various abiotic factors have been shown to influence rates of biomass accumulation in tropical forests recovering from disturbance, including: light (Bazzaz & Pickett 1980, Denslow 1987, Bazzaz 1991), soil water (Lugo et al. 1973, Murphy 1975, Ewel 1980, Medina & Klinge 1983, Denslow 1987, Brown & Lugo 1990, Bazzaz 1991), temperature (Lugo et al. 1973, Murphy 1975, Bazzaz & Pickett 1980, Ewel 1980, Bazzaz 1991) and nutrients (Lugo et al. 1973, Vitousek 1984, Denslow 1987, Uhl 1987, Brown & Lugo 1990, Bazzaz 1991). No measures of soil nutrients or temperature were made in this study, and a spatially-explicit simulation model was not available to simulate these gradients. TOPOCLIM does simulate temperature based on landscape changes in elevation, but the model does not include topography and canopy structure influences on microclimate. So, this analysis focuses on the role of solar radiation and soil water in controlling forest production.METHODS
I predict rates of biomass accumulation based on abiotic gradients following hurricane disturbance. My analysis is based on empirical data on growth (diameter accrual) after disturbance, and biomass equations to convert the field measurements to change in biomass. I then used geographical patterns of hurricane disturbance, and simulation models, to generate gradients of abiotic factors following disturbance. Finally, I developed isopleths of biomass accumulation. These predictive positions in gradient space are developed for the BEW and their generality for the LEF is tested on the HRP. Empirical Growth Data Following Hurricane Hugo, I remeasured plots in the BEW (calibration) and the HRP (validation) to examine the vegetation response to hurricane disturbance. Eighty three plots were established in the BEW before Hurricane Hugo. These are 5 m radius circular plots established on a 40 m grid across the watershed, at differing topographic positions. All stems ò 4 cm dbh were identified, tagged, and measured. These plots were re-sampled three months after the hurricane, and each stem was assessed for hurricane damage to establish hurricane damage severity (Scatena & Lugo in press). I resampled twenty five of these plots (Figure 22A, Chapter 3) a second time during the fourth year of recovery, from months 37 - 41 (Appendix C-XIV), again measuring the diameter of each previously-measured stem, and all new stems. TABLE 17 - Aboveground biomass equations used to convert diameter measurements to biomass and therefore diameter accrual to biomass change. D - diameter at breast height in cm, ln - natural log, exp - inverse natural log, LF - leaf biomass, BR - branch biomass, BO - bole biomass, BM - kg dry mass. c
Species Equation
Casearia arborea LF = exp(1.449 * ln(D)-3.370) a Casearia sylvestris BR = exp(1.786 * ln(D)-3.504) a BO = exp(1.471 * ln(D)-0.779) a BM = LF + BR + BO Cecropia schreberiana LF = exp(1.449 * ln(D)-3.490) a Shefflera morototoni BR = exp(2.062 * ln(D)-5.936) a BO = exp(2.543 * ln(D)-3.139) a BM = LF + BR + BO Dacyrodes excelsa LF = exp(1.701 * ln(D)-2.935) a Tetragastris balsamifera BR = exp(2.760 * ln(D)-5.138) a BO = exp(2.454 * ln(D)-2.653) a BM = LF + BR + BO Inga vera LF = exp(1.523ln(D)-3.022) a Inga laurina BR = exp(1.572ln(D)-2.786) a BO = exp(2.479ln(D)-2.355) a BM = LF + BR + BO Ocotea leucoxylon LF = exp(1.159 * ln(D)-3.593) a Ormosia krugii BR = exp(2.042 * ln(D)-4.609) a BO = exp(2.500 * ln(D)-3.039) a BM = LF+ BR + BO Palicourea riparia BM = exp(1.924 * ln(D)-1.601) a Sloanea berteriana LF = exp(1.916 * ln(D)-3.165) a BR = exp(2.053ln(D)-3.568) a BO = exp(2.501ln(D)-2.403) a BM = LF + BR + BO Musa paradisiaca and Heliconia caribaea LF = exp(1.792 * ln(D)-3.758) a BR = exp(1.982 * ln(D)-3.690) a BM = LF + BR All other species (including Prestoea montana) dbh < 5 cm BM = 0.321 * (D)1.3925 b dbh > 5 cm BM = 4.7306-2.8566(D)+0.5832(D)2 b
a Scatena et al. 1993 b Weaver and Gillespie 1992 c Equations incorporate the correction term for exponential conversion bias (Baskerville 1972, Beauchamp & Olson 1973) The HRP was established during the second and third years of recovery. All stems 10 cm dbh and larger were tagged, measured, identified, and assessed for hurricane damage. I resampled a subset of these plots during the fourth year after the hurricane (months 35-39), using a topographically- stratified random sample of 18 of the 400 20 m by 20 m quadrats in the HRP (Figure 22B, Chapter 3). To allow comparison to the BEW data set, I sampled a 5 m radius plot centered in each of these 20 m by 20 m quadrats. The values of diameter accrual were converted to above- ground biomass (kg dry mass) using species-specific equations developed by Scatena et al. (1993) using dbh (Table 17). Scatena et al. (1993) developed separate equations for leaf biomass, branch biomass and bole biomass. All three were calculated and summed to give the total aboveground biomass per tree. These equations were applied to twelve species in both the BEW and the HRP. For the remaining dicotyledonous species, and also the palm Prestoea montana, I used general equations for the tabonuco forest developed by Weaver and Gillespie (1992). Weaver and Gillespie (1992) developed two equations, one for stems less than 5 cm dbh and a second for stems 5 cm dbh or larger (Table 17). Scatena et al. (1993) also developed a general equation, but specifically for dicotyledonous stems. For Prestoea montana, Scatena et al. used an equation from Frangi and Lugo (1991) based on height of the palm. Additional species-specific equation were developed by Ovington and Olsen (1970) near the HRP, but these equations also utilized height, which was not recorded in my sampling. Biomass for two remaining monocotyledonous species, Heliconia caribaea and Musa paradisiaca was calculated by using Scatena et al.'s (1993) general equation for dicotyledonous trees, but summing only the leaf and branch totals. Simulating Primary Abiotic Gradients I used the simulation models, TOPOCLIM (Everham et al. 1991, Everham & Wooster 1991) and GEOPLOT (Hall et al. 1992b), to generate gradients of solar radiation and soil moisture. Since the growth data were collected in 5 m radius sampling plots, all simulations were performed using the topographic data set at a 10 m grid cell resolution. TOPOCLIM was developed to simulate climate factors of temperature, relative humidity, rainfall, and solar radiation over the landscape (Everham et al. 1991, Everham & Wooster 1991). For this study I used only the solar radiation simulation module. Solar radiation is simulated at the top of the canopy using slope, aspect, and position on the landscape. The position of the sun is tracked for each hour of each day and the direct and indirect radiation for each position on the landscape is calculated. I ran the model to give average monthly solar radiation values (watts/m2) for each plot in both the BEW and the HRP, then averaged these values over a year. This model simulates solar radiation incoming to the top of the canopy, but does not incorporate impacts of vegetation on the light environment. I modify these simulated solar radiation levels by incorporating the severity of hurricane damage (quantified as percent basal area damaged or as API of the recovering vegetation) that I assumed to be correlated positively with light penetration into the canopy. These two methods for quantifying damage should give a clearer position in the primary gradient of solar radiation and the change in this position resulting from the hurricane. GEOPLT was developed as a watershed simulation model and incorporates changes in vegetation and hydrology following disturbance (Hall et al. 1992b). The hydrology module uses rainfall input and subtracts canopy interception (as a function of LAI), surface runoff (as a function of slope and soil saturation), subsurface flow (based on soil characteristics), and evapotranspiration (based on above ground biomass). The structure of this hydrology module is based on the program SWRRB (Simulator of Water Resources in Rural Basins), (Williams et al. 1985, Arnold et al. 1989). LAI, aboveground biomass, and dead organic matter are adjusted following hurricane disturbance. Rainfall data for each site were used as input to the model and the simulated water availability (mm) was averaged for each plot in the BEW and the HRP over the first 12 months following the hurricane. The growth rate data for the BEW were clustered into three groups based on natural breaks in the data (Figure 32). When these groups were plotted in the two-dimensional gradient space of simulated solar radiation and simulated available water, isopleths were placed to maximize the accurate separation of these plots into growth categories. The same isopleths and growth rate categories were used to test the applicability of this model to the HRP. Figure 32 - Distribution of Bisley Experimental Watershed plots on a growth rate axis showing the clustering into three groups. To characterize the light environment better after hurricane disturbance, two methods were used to adjust the position along the solar radiation gradient for each plot. Fernandez and Fetcher (1991) and Walker et al. (1992) quantified solar radiation environments after Hurricane Hugo, but only the latter study compared post-hurricane light environments to pre-hurricane levels. Both studies quantified solar radiation in terms of photosynthetic photon flux density. Neither study presented results of solar radiation changes along a gradient of damage. Therefore, I plot the categories of growth in a gradient space of simulated solar radiation and damage, and fit a line along the border of high growth rate plots to quantify the increased solar radiation with damage. As discussed in Chapter 3, the 5 meter radius sampling plot methodology will not capture damage to trees outside the plot that may influence the microclimate significantly. Therefore, I use a second method to quantify the solar radiation environment. The API of the plot is assumed to reflect the solar radiation environment of the post-disturbance plot. I regressed API against damage, and used the resulting equation to convert API to damage. Then, using the relationship described above, I converted these damage values to adjusted solar radiation levels. These modifications to the simulated solar radiation levels provide three possible models for predicting biomass accumulation: 1) using the simulated water and simulated solar radiation levels, 2) using simulated water and solar radiation levels modified by empirical damage data, and 3) using simulated water and solar radiation levels modified by damage derived from API. For each approach, I determine isopleths to maximize the separation of BEW plots into growth categories, then test these isopleths on the plots in the HRP.RESULTS
The isopleths of growth within the abiotic gradient space of simulated available soil water and simulated solar radiation can correctly categorize growth in 88% of the plots in the BEW and up to 78% of the plots in the HRP. In the BEW, above ground woody biomass accumulation varied from 0.048 to 0.37 kg/m2/month (5.77 to 44.5 t/ha/yr). The distribution of growth rates clusters into three groups, separated at approximately 10 and 20 t/ha/year. That is, the low growth rate group has biomass accumulation of less than 10 t/ha/year (average=6.83, N=5), the medium growth rate group of plots has biomass accumulation of between 10 and 20 t/ha/year (average=13.7, N=12), and the high growth rate group has biomass accumulation rates higher than 20 t/ha/year (average=29.3, N=8). The individual plot results and summaries for each growth rate group are presented in Appendix C-XIV. In the HRP, above ground woody biomass accumulation rates showed similar heterogeneity, varying by a factor of ten. The minimum rate for any plot studied was 0.049 kg/m2/month (5.82 t/ha/yr) and the maximum rate was 0.459 kg/m2/month (55.1 t/ha/yr). Using the growth rate categories identified in the BEW, the 18 HRP plots are distributed as follows: two low growth rate plots (average 6.55 t/ha/yr), six medium growth plots (average 12.8 t/ha/yr), and ten high growth rate plots (average = 34.12 t/ha/yr). The individual plot results and summaries for each growth rate group in the HRP are presented in Appendix C-XIV. The values for simulated and modified abiotic gradient positions are presented in Appendix C-XV. Positions along a gradient of solar radiation are presented three ways: results of the TOPOCLIM simulation, simulation results modified by damage as quantified by percent basal area damaged, and simulation results modified by API. Figure 33 shows that results of graphing simulated solar radiation levels versus percent basal area damaged. The equation for that line was used to adjust solar radiation levels based on damage: adjusted solar radiation = simulated radiation + (percent basal area damaged / 2.55) The denominator in the equation, 2.55, represents the slope of the line that separates high and medium growth rate plots. This line was determined by regressing high growth rate plots with the lowest solar radiation levels (not including those with average radiation values below 250 watts). A line with the same slope is shown positioned next to the medium growth rate plot with the lowest solar radiation level. Figure 34 shows the results of regression analysis for API and damage. This regression equation was used to convert API to percent damage: percent damage = 31.6 * (average pioneer index) - 26.2 The isopleths created for the BEW from these three methods of quantifying solar radiation levels are displayed in Figure 35. Each approach results in 88% of the plots separated into the appropriate growth category (Appendix C-XV). However, these approaches differ in their success when applied to the independent data set from the HRP. The isopleths developed for the two simulated gradients, without modifying the solar radiation levels, has the best predictive ability (77.8% correct) when applied to the HRP (Appendix C-XV and Figure 36). Figure 33 - Positions of plots in a gradient space of solar radiation and hurricane damage. Used to develop a linear relationship to convert hurricane damage to adjusted radiation values. Figure 34 - Regression of Average Pioneer Index (API) with hurricane damage. Used to adjust radiation values based on damage as in Figure 32. Figure 35 - Biomass accumulation as a function of position in a gradient space of solar radiation and soil water for plots in the Bisley Experimental Watershed. Plots are clustered into three categories of growth. Three methods were used to calculate position along radiation gradient: A - using simulated radiation values from the model TOPOCLIM, B - adjusting the simulated values based on empirical measures of hurricane damage, and C - adjusting the simulated values based on Average Pioneer Index of recovering vegetation. Figure 36 - Biomass accumulation as a function of position in a gradient space of solar radiation and soil water applied to the plots in the Hurricane Recovery Plot. Plots are clustered into three categories of growth. Three methods were used to calculate position along radiation gradient: A - simulating the radiation values from the model TOPOCLIM, B - adjusting the simulated values based on empirical measures of hurricane damage, and C - adjusting the simulated values based on Average Pioneer Index of recovering vegetation.DISCUSSION
The rates of biomass recovery reported in this study (average for both sites = 20.1 t/ha/yr, BEW average = 17.3 t/ha/yr, HRP average = 23.9 t/ha/yr) are higher than those reported for other disturbances in the tropics. Uhl (1987) reported biomass accumulation of 12.6 t/ha/yr for recovering agricultural lands in Amazonia. Bazzaz and Pickett (1980) reported on total biomass accumulation for fields in Panama abandoned for two to six years, that averaged 10.6 t/ha/yr. Brown and Lugo (1990) report values of 4.6 t/ha/yr for forests recovering from logging in Venezuela. Jordan (1971) found an average biomass increment of 3.5 t/ha/yr in the first three years of recovery from irradiation in the LEF. Both Lugo et al. (1973) and Murphy (1975) report an overall average of 22 t/ha/yr Net Primary Productivity (above and below ground) from a variety of undisturbed tropical forest sites. Murphy (1975) includes two estimates of the LEF, 10.3 (Odum & Jordan 1970) and 12.3 t/ha/yr (Jordan 1971), again, not on recovering sites and for total biomass. Ewel (1983) states that growth rates are highest in successional tropical forests. My results might lead to a modification of that statement; growth rates are highest in forests recovering from natural disturbance, where the necromass (and its nutrient inputs) are not removed. Brown and Lugo (1990) suggest that the high rates of biomass accumulation during the first 15 years of recovery from disturbance will be independent of microclimate differences. My results show a distinct sensitivity of post-disturbance growth to gradients of solar radiation and soil water. The relationship developed in the BEW seems robust enough to transfer to the HRP, resulting in 78% correctly located in the gradient space, in the best case. This prediction of biomass recovery should be improved with a more direct measurement of solar radiation. Neither of the approaches used in this study to refine the simulations of solar radiation available after the hurricane were successful. Both resulted in lower percentages of plots in the HRP successfully located in the gradient space. The extremely high rates of biomass accumulation (seven plots had rates of 32 tons/ha/yr or greater) stimulated careful scrutiny of the techniques used to estimate biomass. None of these plots included stems that were larger than those used to develop the biomass equations. However, in two of the plots the majority of the biomass accumulation was due to one or two large stems. A small error in either the first or second diameter measurement could result in an erroneously large estimated biomass increase. An additional source of error in the biomass predictions may be inaccurate estimates of biomass of monocotyledons, Prestoea montana, Heliconia caribaea, and Musa paradisiaca. Four of the plots with extremely high biomass accumulation rates had many Prestoea montana stems that accounted for most of the estimated biomass increase. Species-specific equations, or empirical data that includes height, should result in more accurate biomass estimates, and possibly an improved predictive ability of this gradient approach. Plot 1013 in the HRP still accumulated biomass at a rate of 45 tons/ha/yr, due almost exclusively to Cecropia schreberiana stems that established after the hurricane. Maximum growth rates after disturbance are concentrated in intermediate soil moistures, but at the higher solar radiation levels. It is possible that solar radiation levels higher that those simulated for this study would result in lower growth levels, but these results to do not support the assumption that net energy profits will be maximized at the center of gradients of abiotic environmental factors. However, 'grow' in this study incorporates all species on a given plot. Treating species individually may identify species-specific optimal solar radiation ranges in the middle of the gradient for some species. Although these results are successful enough to incorporate into the simulation model of hurricane recovery, future efforts can be focused in two areas. First, it is necessary to quantify the relationship between severity of hurricane damage and the resulting shifts in solar radiation availability in and below the canopy. Second, future simulation efforts could concentrate on a spatially-explicit nutrient model to provide a nutrient availability gradient.
Return to the Table of Contents