INTRODUCTION AND OVERVIEW
"Nature's dice are always loaded ... in her heaps and rubbish are concealed sure and useful results." Ralph Waldo Emerson Catastrophic winds impact forests over most of the globe and the impacts of these storms have been reported in the scientific literature for over a century (Farrar 1818, Redfield 1831, Darling 1842, Perley 1891). Numerous studies have documented the immediate impacts or the short-term response to these events. Fewer investigators have tracked recovery over decades (Whitmore 1974, 1989a, Crow 1980, Weaver 1986), consequently little progress has been made toward generalizations concerning long-term dynamics of recovery or toward integrating catastrophic wind disturbance into more general disturbance theory (Glitzenstein & Harcombe 1988, Brokaw and Walker 1991, Tanner et al. 1991, Everham 1995). I use the impacts of Hurricane Hugo on the Luquillo Experimental Forest (LEF) in Puerto Rico as quantified in the Hurricane Recovery Plot (HRP) and Bisley Experimental Watersheds (BEW) as a case study to examine and develop generalizations about hurricane disturbance and recovery. In this case study I focus on how the forest responds to a hurricane over the spatial scale of the subtropical wet forest of the LEF. Within this spatial scale I examine the patch dynamics of disturbance and recovery using data gathered by myself and others. The empirical data collected for this analysis is limited to the first five years of recovery, but I was able to extend the analysis by utilizing previous studies of recovery from hurricane disturbance in the LEF, and extrapolate to a temporal scale of multiple cycles of succession. I simulated the severity of hurricane impacts and the recovery from this disturbance using, principally, topographic differences. These topographic differences are surrogates for the system inputs which directly influence the dynamics of recovery directly: solar radiation, water, and nutrients.THEORETICAL BASIS
The theoretical basis of this analysis is an extension of gradient analysis as described by Whittaker (1951, 1967), to the concept of disturbance. Although the term 'gradient' is used in the physical sciences as a rate of change, it has come to be synonymous with 'environmental variable' in its application in ecology (Jongman et al. 1987). In this study, 'gradient' indicates a change in an environmental variable over the landscape. In that sense, a response can be influenced by, or correlated with, a gradient. First, a distinction should be made between gradients of abiotic factors that influence physiological processes directly and those that impact these processes indirectly by changing the abiotic factors. I refer to the former as primary gradients, which include solar radiation, temperature, soil moisture, and soil nutrients. I refer to gradients of elevation, topographic position, and severity of hurricane damage as secondary gradients, because each influence primary gradients. These secondary gradients often are easier to measure and develop into maps of landscape gradients, but the key to understanding their influence is understanding their role in modifying the primary gradients. Harmon et al. (1983) have shown that disturbance patterns can be treated as gradients, that patterns of disturbance vary along other landscape gradients such as elevation, and also that the processes of recovery vary along these gradients and may influence further disturbance. However, Harmon et al. (1983) did not describe the role of disturbance in changing the position along these other gradients. They state that species respond to both environmental conditions and disturbance giving the example of pines colonizing bare soil after a disturbance. What is lacking in their analysis is a recognition that disturbance impacts can be quantified in terms of influence on abiotic gradients. It is likely that the pines are responding principally to changes in temperature and moisture that occur on the bare soil. Each point in the forest can be described in terms of its position in multiple gradient space; using either primary or secondary gradients. This is similar to: the concept of multiple dimension niche space introduced by Shelford (1951) and as described and developed by Hutchinson (1958, 1965); the cloud of probability in multidimensional space of either species or environmental factors (Kerner 1957, 1959); or more specifically the distinction of "habitat space" defined by Whittaker et al. (1973) as the gradients of physical and chemical environments and distinguished from "niche hyperspace", the interrelationships among species. My concept of this multiple dimension gradient space differs from these previous definitions in three ways. First, I make the above distinction between primary and secondary gradients (or axes in the hyperspace). Inherent in this view is the assumption that these primary gradients are the ultimate cause of patterns of species distribution. The implication of this assumption is that a multiple dimensional space based on relatively few primary abiotic gradients, rather than a potentially infinite number of biotic interrelationships among species and their environment; temperature, solar radiation, soil moisture and soil nutrients are assumed to be far more important than other possible gradients. As Krebs (1985) refers to temperature as the master regulator (though over a larger spatial scale), I view the variations of these primary abiotic environmental variables as the "master gradients". Second, it is important to include the feedback between the biota and the gradients of abiotic factors. The position along any given abiotic gradient is a function of both the landscape position and the vegetation cover. The latter influences solar radiation, temperature, humidity, soil water, and nutrients. As stated by Waide and Lugo (1992), recovery is influenced by a complex interaction among soil, biota, hydrosphere, and atmosphere. The separation into "habitat" and "niche" hyperspaces (Whittaker et al. 1973) is therefore potentially misleading, because they interact. Finally, I incorporate the critical role of disturbance in shifting the position along these abiotic gradients. Hall et al. (1992a) demonstrate the use of abiotic factors and disturbance in predicting the distribution of organisms (their Figure 3), but do not distinguish between primary and secondary gradients, nor elucidate the impact of disturbance on other gradients. I propose to quantify the impacts of disturbance through the resulting shifts in primary gradients - a multiple-dimension primary gradient approach (MDPGA). Although the LEF is well studied, we have little information on exact values of the primary gradients across the Luquillo landscape. Therefore, I found it necessary to use simulation models and secondary gradients to generate values of these primary gradients, as is suggested by Hall et al. (1992a). The resulting position in gradient space is dependent on interactions among physical position in the landscape (elevation, topographic position, etc.) and biotic controls (vegetation community, structure of the canopy) (Waide & Lugo 1992). That is, the secondary gradients influence the values of the primary gradients, and these primary gradient values are influenced further by biotic interactions such as the structure of the canopy. Within this conceptual framework (MDPGA), disturbance can be quantified as displacement in gradient space, and recovery is the process of returning to original position (Figure 1). Margalef (1969) illustrates one view of quantifying stability in an ecological system using regions in a hyperspace of state variable values that are interconvertible, and ranges of values that allow return to the stable region (Figure 2). My approach is similar, except I label these axes as primary abiotic gradients. For a given disturbance event (e.g. a category 4 hurricane), displacement in gradient space, can be compared between different places in the same system or between systems. Figure 1 - Impacts of disturbance on ecological space. 1 - pre-disturbance position in ecological space, a & c - displacement by two different disturbances, b & d - recovery vectors Figure 2 - Margalef's mapping of stability. Axes are state variable values. Area A is the region within which the variable values vary naturally. When stress results in displacements within area B, the system can return to the original state. With displacements outside of B the system seeks new stable points. (after Margalef 1969) The terms resilience, stability, and resistance have been used in a variety of conflicting ways in reference to disturbance. Westman (1978), in his review of the terminology, defines resilience as "the degree, manner, and pace of restoration", or more specifically "the ability of a natural system to restore its structure following acute disturbance". This definition is consistent with that of Clapham (1971), Denslow (1985), and Pickett et al. (1989) who state a system disturbed beyond its limits of resilience will return to a new domain - of altered composition and structure. This definition of resilience is synonymous with "stability" as defined by Holling (1973), May (1973), and Orians (1975); "adjustment stability" as defined by Margalef (1969) and used by Sutherland (1974); and "elasticity" as defined by Cairns and Dickson (1977). Westman (1978) defines stability in the narrow sense of the range or pattern of fluctuations in state variables. This is consistent with definitions of stability by Whittaker (1975) and synonymous with "persistence stability" as used by Margalef (1969) and Sutherland (1974). In my MDPGA, this characteristic of an ecosystem can be quantified as the total area of variation within the gradients of interest (Figure 2 - area A). Westman (1978) defines terms for five characteristics of ecosystems: 1) inertia, 2) elasticity, 3) amplitude, 4) hysteresis, and 5) malleability. The last four all are considered characteristics of resilience. The first, inertia, is often referred to as resistance. Inertia is the ability of a system to absorb a given intensity of disturbance without change or deformation in system components or interactions. Using the MDPGA, different systems can be compared by quantifying displacement in the gradient space in response to a given intensity of disturbance, or different disturbances might be compared as a ratio of intensity of disturbance to the magnitude of the vector of displacement. Elasticity is the time required for restoration (Westman 1978), and is easily quantified as the time to return to the original position in the defined gradients, when the displacement is within the system's ability to recover. Cairns and Dickson's (1977) "elasticity" is Westman's (1978) amplitude, the threshold beyond which ecosystem repair to the initial state can no longer occur. In the MDPGA, amplitude can be quantified as the total volume of space in which displacement of the system will result ultimately in the return to the original position (area b in Figure 2). This is defining the boundaries of resilience as described by Denslow (1985). In MDPGA, I would anticipate that the shape of the amplitude space would be asymmetrical, reflecting the disturbance history. If the system is adapted to recover from certain types of disturbance, the amplitude volume will encompass the regions of gradient space that reflect the results of these disturbances. As proposed by Johnston (1990), and supported by the work of Garcˇa-Montiel and Scatena (1994), certain types of anthropogenic disturbances in the LEF may not result in a return to the original vegetative communities. Crow (1980) proposed that anthropogenic disturbances tend to differ in spatial and temporal frequencies from the disturbances for which the systems are adapted. I believe the critical difference is that anthropogenic disturbances may involve displacements into regions of gradient space not previously experienced, and thereby outside of the amplitude of the system. Hysteresis, as defined by Westman (1978), is the degree to which the pattern of recovery is not simply the reversal of the pattern of initial alteration. Westman suggests this is most applicable to systems subject to chronic stress where the path of alteration can be quantified and compared to the path of recovery. Disturbance, defined as a relative discrete event in time (White & Pickett 1985), has paths of alteration which are therefore difficult to quantify. In the MDPGA, disturbance results in a displacement in gradient space. If the displacement is assumed to be a straight line, hysteresis can be quantified by tracking the path of recovery to the original position and measuring the maximum difference between this path of recovery and the assumed straight line path of alteration. Malleability is described as the ease in which the system can be altered permanently (Westman 1978). This sounds similar to resistance but is a measure of the difference between new, unique stable states and the original pre-disturbance system state. This can be measured directly as a straight line distance between the pre-disturbance and post-response positions with the MDPGA. Quantifying any of these characteristics of disturbance and ecosystems has proved difficult (Westman 1978). The MDPGA clearly allows quantification of each of these characteristics of systems: resilience, resistance, and stability. Quantifying these characteristics has the potential to allow comparisons of the impacts of different types of disturbance on and between different systems and could lead to more robust theories of the impacts of and response to disturbance.APPLICATION TO VEGETATION DISTURBANCE AND RECOVERY
Vegetation recovery from hurricane disturbance in the LEF has been quantified in a variety of ways including: biomass accumulation (Crow 1980, Weaver 1986, Hall et al. 1992b), recovery of physical structure (Crow, 1980, Brokaw & Grear 1991), or community composition changes (Crow 1980, Doyle 1981, Weaver 1986). I include all three of these measures in this analysis of recovery and integrate them into a spatially explicit simulation model. Figure 3 - Conceptual model of the application of primary and secondary gradients to recovery from hurricane disturbance. Secondary gradients affect primary gradients which can provide a mechanistic explanation for response, or secondary gradients can be correlated with response for a descriptive explanation. My field work involved quantifying recovery through collecting data on species composition, diameter accrual, foliage profiles, and canopy closure on plots located in the Hurricane Recovery Plot (HRP) and the Bisley Experimental Watersheds (BEW). I use simulated levels of abiotic factors and measured intensities of hurricane damage to describe the positions in gradient space following disturbance and test the predictive ability of this gradient approach in determining the dynamics of recovery (Figure 3). There are five specific goals to this project: 1) to describe the spatial patterns of hurricane damage and the factors that influence these patterns, 2) to predict the vegetation community response to gradients of hurricane damage, 3) to assess the role of gradients of solar radiation and soil moisture in predicting rates of biomass accumulation, 4) to analyze the dynamics of recovery of the canopy structure and the factors that influence this process, and 5) to develop a spatially explicit landscape simulation model of the hurricane disturbance and recovery that incorporates the above concepts. Due to the time frame of this project, recovery will focus on only the first five years following disturbance. Chapter 2: Factors Influencing the Spatial Pattern of Hurricane Damage Intensity of hurricane winds varies over the landscape and the resulting gradients of severity of damage impact the dynamics of recovery. I examine three aspects of the spatial pattern of hurricane disturbance: 1) how do the spatial patterns vary with different measures of hurricane damage, 2) what is the gap size of hurricane disturbance, and 3) what are the relative roles of abiotic environmental factors (topography, substrate features, and disturbance history) and biotic factors (stem density, basal area, and community structure) in influencing patterns of disturbance. Relative to the last issue, I test this hypothesis: HYPOTHESIS 1 Hurricane damage is correlated more highly to abiotic environmental factors than to biotic factors. Species in the LEF were impacted differently by Hurricane Hugo winds, (Basnet 1990, Dallmeier et al. 1991, Scatena and Lugo 1990, Walker 1991, Basnet et al. 1992, Walker et al. 1992, Zimmerman et al. 1994) so the distribution of species may be viewed as a proximate cause of spatial patterns of damage. However, it has been demonstrated that species distributions in the LEF are associated with changes in elevation (Weaver 1991), edaphic properties (Johnston 1992, Basnet 1993), topography (Crow & Weaver 1977, Crow & Grigal 1979, Heaton & Weaver 1986, Letourneau 1989, Weaver 1991, Basnet 1992, Johnston 1992), and historic landuse (Crow & Grigal 1979). In addition, topography influences the intensity of wind disturbance. Therefore, these abiotic gradients can be viewed as ultimate causes of spatial patterns of damage and should be better predictors. Elevation, topography, and soil characteristics can be used to predict species distributions and therefore the resulting impacts of a hurricane disturbance. Using the data for the entire HRP I examine the spatial patterns of damage over the plot, and determine how spatial patterns vary with different methods of quantifying damage. I use point pattern analysis techniques (Krebs 1989) to quantify clumping of damage, then the quadrat-variance methods described by Ludwig and Reynolds (1988) to quantify gap size of hurricane disturbance. Then I compare the relative roles of abiotic and biotic factors in relation to synthetic variables that incorporate all measures of damage, using canonical correlation analysis. Chapter 3: Hurricane Damage Gradients and Vegetation Community Dynamics There appear to be two principal vectors of response of the forest canopy. I refer to them as regrowth and recruitment (Figure 13). These two vectors differ in vegetation community composition. Regrowth is recovery of the canopy through the sprouting of surviving trees; what Muller (1952) and Hanes (1971) refer to as "autosuccession" and Boucher (1989) refers to as "direct regeneration". I believe this occurs where mortality is low and structural damage is low to moderate. The forest continues to be dominated by late successional species. Recruitment is recovery of the canopy by establishment of new early successional species. I predict this occurs where mortality is moderate or high, or where structural damage is high. The forest composition shifts to early successional species. Whitmore (1989a) describes an accelerated recovery of canopy gaps that includes the simultaneous colonization of the gap by both pioneer and climax species; a sequence of succession from small pioneer trees, to mature pioneer trees, directly to a canopy dominated by shade tolerant species. In my dichotomy, this path of recovery would fall into the recruitment category, because of the initial post-disturbance shift to early successional species. Two additional responses to disturbance are reported in the literature: release and repression. Release is recovery of the canopy by suppressed subcanopy trees whose growth is stimulated by loss of the canopy, referred to by Abrams and Scott (1989) as disturbance-mediated succession. This response was reported for the 1938 hurricane in New England (Spurr 1956), windstorms in Colorado (Veblen et al. 1989), treefall gaps in Venezuela (Uhl et al. 1988) and was projected for the recovery from a tornado in Texas (Glitzenstein & Harcombe 1988) and for cyclone disturbance in the Solomon Islands (Whitmore 1989a). Repression occurs when establishment of new trees is restricted by the dense herbaceous growth that invades after the disturbance. Weaver (1989) describes the formation of "alpine meadows" through a dense growth of ferns and grass following disturbance to the dwarf forest in Puerto Rico. Wolffsohn (1967) records the invasion of bracken fern and the subsequent restriction of hardwood regeneration after fire following a hurricane in Belize. These two latter paths of response, release and repression, appear to be uncommon following Hurricane Hugo, in the areas of the forest I studied. I assess the success of a two-dimensional secondary gradient space (severity of hurricane damage, quantified as compositional change - percent mortality - and structural change - percent of stems with severe damage or percent basal area lost - , Figure 12), in predicting the principal successional pathways to recovery: regrowth or recruitment. These two paths are distinguished based on the proportion of pioneer species present, quantified using the Average Pioneer Index (API) (Whitmore 1974). The position of each plot is determined in a two dimensional secondary gradient space of physical damage and mortality. Isopleths for response vectors (described above as regrowth and recruitment) were determined for a subset of plots. This approach is similar to the use of gradients of nutrient flow and degree of exploitation to distinguish theoretical regions of regeneration and system collapse in forests (Gatto & Rinaldi 1987). I use empirical data to determine the regions within the gradient space that indicate a particular response to disturbance. The predictive ability of these positions in gradient space is tested for other sites and at varying spatial scales. These paths of recovery, regeneration and recruitment, and their corresponding vegetation community differences, also are used in predictions of the vertical distribution of biomass. Chapter 4: Biomass Production in Response to Post-Hurricane Environmental Gradients In chapter three I use community composition to quantify response to disturbance; a second approach is measuring the rate of biomass accumulation. I assess the power of the primary gradients of soil moisture and solar radiation and the secondary gradient of hurricane damage, in predicting the rate of biomass accumulation. I examine the question: are energy profits maximized at the center of gradients of abiotic environmental factors? I use the watershed simulation model GEOPLT (Hall et al. 1992b) to simulate average solar radiation and soil moisture values. I use these two values, as well as a third, hurricane damage, to create a gradient space in which I position a subset of the plots I measured for biomass accumulation. Simulated solar radiation levels at the top of the canopy and hurricane damage are secondary gradients that correlate with the actual position of a plot in the primary gradient of solar radiation. Actual rates of physiological processes were not measured, so the question of net energy profit cannot be examined directly. I determine net energy profit as change in biomass, calculated from increases in diameter converted to biomass changes using species-specific algorithms developed by Scatena et al. (1993) and a general formula for the tabonuco forest developed by Ovington and Olsen (1970). Then, as with the analysis of gradients of hurricane damage predicting recovery vectors, isopleths of biomass increase within this gradient space were developed. The applicability of these isopleths is tested by repeating the analysis for the other plots. In this case each sample plot is placed within the two-dimensional gradient space (solar radiation and water) and the resulting position in recovery isopleths are checked against measured rates of biomass increase. Biomass increase is assumed to reflect net energy profits, so the positions of the isopleths along the gradients are examined to answer the question of maximum energy profits in the middle of the gradient. Chapter 5: Canopy Restructuring in Response to Hurricane Disturbance Finally, recovery can be quantified by analyzing changes in the canopy structure as the forest canopy recovers. For the purposes of this study, canopy structure refers to the distribution of plant material through vertical layers above the ground. Brokaw and Grear (1991) used foliage profiles to describe the impacts of Hurricane Hugo following methodology established by Karr (1971) to relate habitat structure to bird population distributions, and Hubbell and Foster (1986) to examine treefall gap dynamics. Brokaw and Grear (1991) found that the hurricane had impacted the upper canopy levels most strongly, lowering the average canopy height by as much as 50%, and predicted significant impacts on forest composition. Weaver (1986) noted changes in the canopy stratification and height distribution of trees from 1946 (14 years after the last hurricane) and 1976. What is needed now is to trace the closing of the canopy and the restructuring of canopy layers, particularly in the first decade of recovery. This study analyzes these processes during the first five years of recovery. These efforts to predict patterns of vertical distribution of structure also hold promise of predictive power for the distribution of herptafauna, avifauna, and invertebrates in the recovering forest. I evaluate the role of two secondary gradients in influencing the dynamics of canopy restructuring: 1) community dynamics (driven by gradients of hurricane damage) and 2) topographic position. I test two specific hypotheses: HYPOTHESIS 2 - Recovering forest stands have significantly different canopy structures when dominated by recruitment (early successional species) as opposed to dominated by regrowth (surviving late successional species). 2A Maximum canopy height is significantly more variable in regrowing forest sites than in those dominated by recruitment of early successional species. 2B Percent cover increases in each successively lower canopy interval in a regrowing forest. Maximum cover occurs at the lowest canopy interval - herb (0-1 m) 2C Forest sites dominated by recruitment have maximum cover at an intermediate canopy layer, therefore any lower layers have less vegetation. HYPOTHESIS 3 - Canopy structure is significantly different between sheltered valleys and exposed ridges. 3A Independent of the recovery vector, valley sites have significantly lower percent cover in the lowest two canopy layers: shrub (1-4 m) and herb (0-1 m), than ridge sites. 3B The difference between both of the two lowest canopy layers is significantly greater in valley sites than in ridge sites when comparing regrowth to recruitment. That is, the shading effect of the canopy in a recruiting site is amplified in valleys. I believe forest canopy structure differs between these two response vectors because of the difference in severity of damage and the resulting vertical gradients of abiotic factors, principally solar radiation. Regrowing forest stands have surviving stems that result in a variable upper canopy surface that allows some solar radiation to penetrate through to the lower canopy layers. A forest stand dominated by recruitment occurs where more severe disturbance results in initially higher levels of solar radiation at the forest floor. However, as a pulse of new trees is established, the resulting canopy is both even and dense, allowing less solar radiation to penetrate to lower levels as the canopy grows upward. The difference in solar radiation available to lower canopy levels is amplified by topographic position. Sheltered valleys receive less solar radiation due to shading of the terrain, so the lower canopy levels have less structure and the difference between these two vectors of recovery is increased. More diffuse solar radiation penetrates to the lower canopy levels on exposed ridges, so the vegetation structure at these levels is denser and the difference between the two vectors of response is diminished. I summarize foliage profile data by aggregating intervals into five canopy layers, designated as: emergent (surviving) 20-30 m, canopy 12-20 m, subcanopy 4-12, shrub 1-4 m, and herbaceous 0-1 m. Comparison of the resulting interval patterns is accomplished using paired t-tests, and 2 x 5 contingency tables.SIMULATION MODEL
I incorporate all of the above efforts into a computer model, driven by secondary gradients of topography and hurricane damage, and simulated primary gradients of solar radiation and soil moisture, which predicts patterns of recovery as quantified by community change, biomass production, and structural changes to the canopy (Figure 4). This model follows the concepts outlined in Everham (1991) and Everham et al. (1993). In general, each model function is derived from observed associations rather than measured physiological response. A more detailed explanation of each model routine is presented in Chapter 6, and complete documentation is found in Appendix E. Figure 4 - Conceptual diagram of hurricane recovery model: RECOVER Figure 5 - Flow diagram of RECOVER indicating input data, output data, and subroutines. Optional inputs are shown with dotted lines. Principal model subroutines are: TOPOGRPH - to determine topographic position; EXPOSE - to determine exposure to hurricane winds from a given direction; DAMAGE - to simulate damage severity; SUCCESS - to simulate recovery as measured by changes in API; BIOMASS - to simulate recovery of aboveground biomass; and CANOPY - to simulate the recovery of forest canopy structure. The principal input data set (TOPO.DAT) is elevation, slope, aspect, and region of the forest (Figure 5). This data set is used: 1) to determine topographic position (ridge, slope, bench, or valley - subroutine TOPOGRPH), 2) as inputs into a watershed model solved within the modeling shell GEOPLT (Hall et al. 1992b) to generate maps of soil moisture and 3) as inputs into the meteorology model TOPOCLIM (Everham & Wooster 1991) to generate maps of solar radiation. An additional data set (HURR.DAT) details information about a specific hurricane disturbance event: path, direction, and intensity. The hurricane data and topographic factors are used to determine exposure to hurricane winds (subroutine EXPOSE). Exposure then is used to simulate damage severity (subroutine DAMAGE). This simulation of damage can be refined based on other data (if available): soil type, tree community, or disturbance history; or can be replaced by empirical data on damage. The recovery of the damaged patches of forest is simulated using three separate approaches. First, gradients of hurricane damage are used to simulate the principal vectors of recovery: recruitment or regrowth (subroutine SUCCESS). Second, the rates of biomass accumulation are simulated based on position along gradients of solar radiation and soil moisture (subroutine BIOMASS). Third, the community dynamics associated with hurricane damage severity and topographic characteristics are used to simulate changes in canopy structure (subroutine CANOPY). The spatial patterns of recovery are displayed, including percent cover at each of the five canopy layers for each month during the first five years after the hurricane. This model can be used to investigate assumptions regarding the dynamics of recovery, or the impacts of changing the hurricane disturbance regime on landscape patterns and ecosystem processes. Typically, ecosystem processes such as biomass accumulation have been viewed as following an asymptotic curve during recovery from disturbance. This has been the approach used for many models applied to the LEF (Hall et al. 1992b, Everham et al. 1993). Bormann and Likens (1979a, 1979b) suggest an alternative perspective on the phases of recovery in a temperate hardwood forest: reorganization, aggradation, transition, and steady state (Figure 6). The reorganization phase is described by a continuing decline in ecosystem structure or function (e.g biomass) immediately following disturbance. Foster (1988b) observed periods of decline in biomass followed by rapid increases after a series of disturbances to a forest in Massachusetts. This may be evidence for the reorganization phase. During aggradation, the reorganized system achieves levels of structure or function greater than pre-disturbance, steady state levels. Eventually, the system goes through a transition phase to reach a sustainable (steady) state. The four phase model of recovery is similar to the pattern of a perturbation dependent system as described by Vogl (1980; his Figure 1), and the "stationary process" suggested by Loucks (1970; his Figure 7). I developed recovery algorithms based on Bormann and Likens phases and compared them to an assumption of asymptotic recovery in order to investigate the question: what would be the implications of recovery that follows the four phase Bormann and Likens model? If the LEF follows this pattern of recovery, the timing of subsequent disturbances may be particularly critical. Figure 6 - Two models of recovery after disturbance (after Crow 1980, Bormann & Likens 1979a, 1979b) Regardless of the path of recovery to a steady state, a second question is: does the LEF reach a steady state between hurricane disturbances? The answer may be dependent on the spatial scale of analysis, certainly as a fine spatial resolution individual stands on forest may reach an equilibrium, but this may not be true for larger spatial scales. Lugo and Scatena (1994) stated the LEF is in a "continued state of adjustment from the last major disturbance event". Basnet (1992) predicted that " a steady state is not possible". These views were supported by Weaver's (1986, 1989) and Crow's (1980) data for the tabonuco forest which shows a convex but not yet flat biomass recovery curve, and the colorado forest where the recovery is described as linear. Again, the slope of the curve of recovery will influence the impacts of subsequent disturbance, particularly if the disturbance regime changes. I investigate these possibilities by adjusting the time to reach steady state and running simulations of the entire LEF under various disturbance regimes.STUDY SITE
The LEF, a part of the Caribbean National Forest, is located in the northeast corner of Puerto Rico (Figure 7). The LEF varies from 100 to 1075 meters above sea level and includes four life zones in the Holdridge System: 1) subtropical wet forest, 2) subtropical rain forest, 3) lower montane wet forest, and 4) lower montane rain forest (Ewel & Whitmore 1973, Brown et al. 1983). This project focuses on the subtropical wet forest and is named for the tabonuco (Dacryodes excelsa Vahl. BURSERACEAE) tree, and is the lowest (200 - 600 m a.s.l.) of four vegetation zones occurring along an elevational gradient in the LEF. In addition to tabonuco, the forest in the area is dominated by the palm, Prestoea montana (R. Grah.) Nichols (ARECACEAE), and trees Manilkara bidentata (A. DC.) Chev. (SAPOTACEAE) and Sloanea berteriana Choisy (ELAEOCARPACEAE) (Odum & Pigeon 1970, Brown et al. 1983). Soils in the area are mostly zarzal clay, which are deep oxisols of volcanic origin (Huffaker in press). Figure 7 Luquillo Experimental Forest, Bisley Experimental Watershed, and Hurricane Recovery Plot The LEF is one of the National Science Foundation's Long- Term Ecological Research Sites, overseen by the Terrestrial Ecology Division of the Center for Energy and Environment Research, an affiliate of the University of Puerto Rico, and the International Institute of Tropical Forestry, U.S. Forest Service. The LEF is an ideal site for long-term tropical research for two reasons: 1) it has a history of ecosystem oriented research, including long-term study plots established in the 1930s and Odum and Pigeon's (1970) irradiation study; and 2) its disturbance history, natural and human induced, is well documented (Brown et al., 1983, Waide & Lugo 1992). On September 19, 1989 Hurricane Hugo passed over the eastern end of Puerto Rico (Figure 8). Hugo was a category 4 hurricane with maximum sustained winds of 166 km/hr (Scatena & Larsen 1991). Weaver (1987) presented the hurricane history of Puerto Rico over the last three centuries. During that period, 15 hurricanes have passed over Puerto Rico, of which five (including Hugo) have passed directly over the LEF. The last hurricane to most directly impact the LEF was San Cipriano in 1932, although San Felipe (1928), San Nicolas (1931), and Santa Clara (Betsy) (1956), may also have caused slight localized damage (Weaver 1989). Figure 8 - Historical hurricane tracks. Solid lines are known tracks, dashed lines are tracks estimated from descriptions. (after Salivia 1972, Weaver 1987, Scatena & Larsen 1991) Growth plots established by the Forest Service in the 1940's provide an excellent data set for following the recovery of the forest (Wadsworth 1951, Wadsworth 1970, Crow 1980), except for that critical first decade (Figure 9). This study is intended to help fill that empirical gap, to examine the application of a multiple dimension gradient space to the impacts of disturbance, and to develop simulation tools that can be used to investigate possible implications of changing disturbance regimes. Figure 9 - Sequence of disturbance and growth plot monitoring on the LEF. a data from Crow (1980), b data from Johnston (1990)
Return to the Table of Contents