Biodiversity at the LUQ Long-term Ecological Research Site

Authors: Robert B. Waide, D. Jean Lodge, Gerardo Camilo, and Whendee Silver

---

Abstract

Research on biodiversity has taken a prominent place at the Luquillo Long-Term Ecological Research (LTER) site because of the high species richness of some taxa at the site and the important interactions between richness and ecosystem function. A suite of studies of biodiversity are organized into eight themes: relating biodiversity and ecosystem function, relating biodiversity to biogeochemical cycles in regenerating forest, connecting diversity and primary productivity, effects of disturbance on microbial diversity, effects of land use on diversity, functional classifications of tree species, rare and endangered species, and the effects of disturbance on coadaptation of herbivores and plants.

Introduction

The development and maintenance of diverse biological systems is one of the most intriguing areas of study in ecology. Tropical ecosystems have long excited the interest of biologists both because of the high diversity of these systems and the unusual adaptations of individual organisms found in the tropics. Pioneer efforts to understand the development and maintenance of biodiversity focused on tropical islands, whose diversity in age, size, complexity and distance from source faunas provided ample fodder for the development of theory (McArthur and Wilson 1967). Subsequent scrutiny of these theories through experimental studies have often used island populations as easily-manipulated study subjects (Losos 1997). The intermediate level of isolation among islands in the Caribbean archipelago and between the islands and the mainland is responsible for fostering the evolution of new species in birds (Lack, Rickleffs, etc.) and some types of mushrooms (Lodge et al., work in progress).

The Luquillo Long-Term Ecological Research (LTER) site has the highest diversity of some taxa e.g., (plants) and the lowest diversity of others (e.g., mammals) among the 18 LTER sites. Other taxa (e.g., birds and ants) have similar numbers of species to other forested LTER sites when compared on an areal basis. The assemblage of species that occurs at the Luquillo LTER site is determined by the interaction of two broad biogeographical patterns, the well-known increase in species richness with decreasing latitude and the tendency towards reduced species richness with isolation and small island size (Waide and Reagan 1996). The relative strength of these two patterns varies among taxa, resulting in the paradoxical situation described above.

The frequency and variety of disturbances (hurricanes, landslides, floods, droughts) that occur at the Luquillo site add complexity to the patterns of diversity determined by latitude and isolation. Periodic disturbances expand the range of environmental conditions experienced by organisms at Luquillo (an expanded environmental space sensu Hall et al. 1992) and limit the degree of habitat specialization possible. The dynamic nature of Luquillo ecosystems result in high variability in the magnitude of energy flow through different pathways in the trophic structure. A key focus of biodiversity studies at Luquillo is to determine how disturbance affects the degree of redundancy in ecosystem processes and hence the resilience of the system. We couple these studies of ecosystem processes with research on the relationship between habitat complexity and community diversity, the long-term stability of communities, and the importance of pivotal species in restoring ecosystem function after disturbance. This suite of studies is described in more detail below and in Table 1.

Relating Biodiversity and Ecosystem Function

Studies of biodiversity in the tropics have largely ignored the effect that the loss of diversity has on ecosystem function. Research conducted at the Luquillo LTER focuses on critical junctures or pathways in key nutrient and energy cycling processes (termed interfaces).

These interfaces represent the points in the ecosystem where resources are concentrated and transferred between biotic and abiotic components (Silver et al. 1996a). Organisms influencing processes occurring at interfaces have a strong influence on ecosystem structure. We define functional diversity as a set of ecosystem processes that contribute to the maintenance of matter and energy flows through the ecosystem (Silver et al. 1996b). The few data that exist suggest that it is functional rather than species diversity that is most important in stabilizing nutrient and energy fluxes (Silver et al. 1996a). However, the existence of high species richness in an ecosystem may increase system resiliency after disturbance by providing multiple pathways for the flow of resources. These ideas are shown in Figure 1 (Silver et al. 1996a). Research at the Luquillo LTER is directed towards employing this conceptual framework to evaluate the rehabilitation of tropical ecosystems after disturbance.

Figure 1

The relationship of tree species composition and biodiversity to biogeochemical cycles in restored tropical moist forest

In this study, we are using a 60 year old tropical reforestation project to: (1) examine the relationship between biodiversity and ecosystem biogeochemistry and, (2) determine if species richness, composition, or dominance have a significant impact on soil C dynamics and storage. Changing land use in the tropics inevitably results in the loss of biodiversity at some spatial scales (Wilson 1988, Whitmore & Sayer 1992). The potential for negative ecological consequences from the loss of biodiversity has fueled considerable scientific debate regarding the role of species in ecosystem level processes, and the possible feed backs on ecosystem services (Schultz & Mooney 1993). Few empirical studies have addressed these issues, and those that have been conducted on relatively simple, natural ecosystems and in artificial microcosms, where conditions can be easily controlled and monitored (Naeem et al. 1994, Tilman & Downing 1994, Tilman et al. 1996). While these experiments are important and necessary for the development of sound ecological theory, they are difficult to apply to highly diverse and complex tropical ecosystems where the potential loses of biodiversity are greatest.

The diversity and complexity of tropical forests introduces some seemingly insurmountable problems for the design of experiments to link species with ecosystem-level processes. Manipulations that alter biodiversity introduce a disturbance to the ecosystem that confounds the effects of changes in species richness with the responses to disturbance. In forest ecosystems, recovery from disturbance events is likely to take years or decades. Furthermore, the diversity and complexity of tropical forests make species identification at the ecosystem scale quite challenging.

We are using long term forest plots with known species composition established by the USDA-Forest Service in Puerto Rico to determine the relationship of species richness, composition, and dominance to indices of ecosystem function. The study is being conducted in a 65 ha forest complex that was established from pasture and agricultural land 60 years ago as part of a well-documented program to restore tropical forest structure and function. In the mid 1930’s, the USDA-Forest Service planted native and non-native tree species, and removed grazing pressure to facilitate natural regeneration. In 1957 and 1993, 116 permanent plots were measured for growth, mortality, and changes in tree species composition. Today, the forest consists of distinct tracts of natural secondary forest, native species plantations, non-native species plantations, and mixed species plantations. All tree species have been identified, with a total of 75 species >9.1 cm dbh. These well documented plots with known land use history, growth rates, and tree species composition, provide an excellent opportunity to address the effects species richness, dominance, and composition on ecosystem-level processes in a tropical forest.

As part of the preliminary component of this study we examined the effects of tree species richness and species composition on indices of nutrient cycling. Soil and forest floor nutrients, bulk density, and soil carbon content were measured in 25 0.04 ha plots along a diversity gradient ranging from 4-16 species, with similar slope, aspect, tree density, and total basal area. Within a subset of 15 plots along the same diversity gradient, we estimated potential nitrogen mineralization and nitrification rates and estimated phosphorous retranslocation by Tabebuia heterophylla, a dominant canopy tree, in nine plots. We used linear regression analyses to examine relationships between species richness, species composition and several structural variables; and nutrient pools, nitrogen fluxes and estimates of retranslocation.

Results showed no correlation of species richness with most of our indices of nutrient cycling. However, the abundance of certain species and genera, and the abundance of exotic tree species were significantly correlated with nutrient pools and transformations. The basal area of the native tree Calophyllum antillianum was inversely correlated with both surface soil (0-10 cm) and subsurface soil (10-35 cm) phosphorous pools (Figure 2). The C:N ratio of forest floor material increased significantly with the basal area of exotic species (p < .01, r²=.52).. Potential nitrogen mineralization rates were found to be inversely correlated with the basal area of Inga species, suspected nitrogen fixing trees (p < 0.001, r²)

Figure 2a

Figure 2b

Phosphorous retranslocation by T. heterophylla does show a non-linear relationship with species richness. Using a post-hoc ANOVA we found significant differences between low, intermediate and high diversity plots, with those of intermediate diversity (7-11 spp.) having higher rates of retranslocation than either low (4-6 spp.) or high (12-16 spp.) diversity plots.

Connecting Diversity and Primary Productivity

The existence of a causal relationship between diversity and productivity has long been suspected by ecologists, but the mechanisms leading to such a relationship have yet to be identified (Rosenzweig and Abramsky 1993). Researchers from the Luquillo LTER organized a series of workshops on this theme at the National Center for Ecological Analysis and Synthesis. The purpose of these workshops is to use data from LTER and other sites to search for general patterns and the mechanisms causing them over broad ranges of productivity, diversity, spatial scale, and trophic level.

The LTER Network is particularly appropriate as a base for the proposed synthesis. The control of primary productivity is one of the five core research areas at each LTER site. Because of this emphasis, each site has available direct measures of productivity as well as information on surrogate measures such as soil fertility. The methodology used to measure productivity varies with the ecosystem characteristics at each site, but most sites have measurements spanning a range of environmental conditions. Moreover, at least 15 of the sites have manipulated productivity by varying site-specific controlling factors. These manipulations are mostly fertilizations, but they span a wide range of ecosystems including lakes, streams, forests, grasslands, and agriculture. At some sites, both aquatic and terrestrial manipulations have been done. In addition, the effects of these manipulations have been followed both in producers and higher-order consumers, which permits a comparison of responses in these two groups.

Results from the workshops suggest that the form of the relationship between primary productivity and diversity depends on the scale of study. In a survey of published studies, Mittlebach et al. (unpublished) found that the relationship between diversity and productivity changed from positive and monotonic at the continental scale to hump-shaped at the regional scale to unpredictable at the local scale. They found that theories developed to explain patterns at one scale were often unsuccessfully applied to explain patterns at a different scale. Data from a subset of the LTER sites show similar patterns in moving from local to regional scales (Gross et al. unpublished.)

Fungal diversity in the Luquillo Mountains

Polishook, Bills & Lodge (1996) cultured microfungi growing in decomposing leaves of two different tree species that occurred together on the forest floor at two widely separated sites in El Verde. They found that 58% of the 338+fungal species that were detected at least twice were only found on one of the two leaf species. Restricting the analysis to species cultured at least five times increased the degree of specificity for one of the two hosts. Similar patterns were found by Cornejo et al. (1994) in Panama. Lodge et al. (1996) and Laessoe & Lodge (1994) demonstrated that the fungal species with host preferences for certain types of forest floor litter are almost exclusively true decomposers that do not colonize the leaves before they fall from the tree. Species of decomposer fungi that show strong host preferences might have specific adaptations for decomposing litter from their host, and elimination of host-specific fungal species might result in slower decomposition and nutrient cycling. If so, this would represent a major link between species diversity and rates of ecosystem processes.

Bills (in Polishook et al. 1996) hypothesized that much of the host preference exhibited by microfungal decomposers may be related to physical and chemical characteristics of the leaves rather than host-specificity in the taxonomic sense (i.e., for a single host species, genus or family; Lodge 1997). In other words, microfungi may prefer "guilds" of leaf species that are defined by the concentrations of nitrogen, phosphorus and other nutrients in their tissues and the thickness of the leaf and waxy cuticles.

Effects of land use history on forest structure and composition in the Luquillo Mountains

The well-know land use history of the Luquillo Experimental Forest (Garcia-Montiel and Scatena 1994, Zimmerman et al. 1995) allows investigation of the impact of land use on forest biodiversity. Zimmerman et al. (1995) sampled three sites in the LEF and looked at forest structure and composition in minimally disturbed forest (these areas were selectively logged at worst), abandoned shade-coffee plantations, and abandoned pastures. These areas had been abandoned 40-60 years before they were sampled. The expectation was that the structure and composition of abandoned coffee would be similar to minimally-disturbed forest because the coffee shade would provide habitat for birds and bats, important disperser of tree seeds, and that forest would recover more quickly than abandoned pastures.

What we found was that forest structure (stand basal area and density) were largely identical among all land uses, but that the species composition of each land use was distinctive (Figure 3). The composition of the stands in abandoned shade-coffee differed from minimally-disturbed forest as much as did stands in abandoned pastures. Biodiversity (species richness) of the relatively unmanipulated stands is often higher than the secondary forests, indicating that human land use in the Luquillo Mountains causes a decline in alpha-diversity and an increase in beta-diversity of forest stands.

Figure 3

We believe that some ecosystem processes (litterfall and nutrient cycling) will be similar in these stands as found by Zou et al. (1995) in a comparison of forest stands that were previously clear-cut or minimally-disturbed. However, some differences may persist -- we hypothesize that nitrogen levels in soils of abandoned shade-coffee, where most shade had been provided by leguminous trees, may remain high for many years and may explain the current dominance of these stands by species (Cecropia schreberiana, Guarea guidonia; Figure 3) that can take advantage of the high nutrient availability (Silander 1979, Fernandez 1997).

Functional types of forest trees of subtropical wet forest in the LEF

Identifying functional differences among trees in tropical forests is an important step towards modelling their dynamics and developing management schemes. A single hectare of tropical forest may have over 50 species, most of which will be rare, and some coherent method of reducing the number of species to their functional equivalence, even if somewhat artificial, is necessary to model or manage these forests.

Swaine and Whitmore (1988) proposed the trees in the tropical lowlands could be divided into two functional groups, pioneers and nonpioneers. As shown by Swaine and Whitmore (1988), these two groups can be distinguished by a number of characteristics and they emphasized differences in seed and seedling behavior. Pioneers maintain a seed bank and seeds are physiologically tuned to respond to increased light levels produced by canopy disturbance. Seeds of nonpioneers germinate in shade and these species maintain a seedling bank ready to respond to any disturbance to the canopy. These species can also be distinguished by their demographic characteristics (Alvarez-Buylla et al. 1996): the most important transition in a pioneer life history is the survival of seeds in the seed bank and the growth rate of juveniles, while it is the survivorship of juveniles or adults that is most critical to the life histories of nonpioneers.

We analyzed the hurricane response of 26 species of trees (stem > 10 cm at breast height) occurring in the 16 ha Hurricane Recovery Plot in the LEF from the perspective of life history variation (Zimmerman et al. 1994). Five variables describing the type of hurricane damage each species sustained (frequency of uprooting, stem breakage, breakage of large limbs, and mortality) or their ability to recover following damage (frequency of branch sprouting 1 year post hurricane) were analyzed by Principle Components Analysis (Figure 4). In this way we hoped to see "breaks" in the responses of species indicating differing functional types. The first principle component summarized the important life history differences (Zimmerman et al. 1994) and indicated that three species fell to the highest end of the axis. These species, Cecropia schreberiana, Schefflera morototoni, and Miconia tetandra, were distinguished by having the highest levels of stem damage and mortality during the hurricane, as well as low levels of branch breakage and post-hurricane resprouting. Condit et al. (1996) found similar results for an analysis of the demographic characteristics of 142 species of tree located on the 50 ha forest dynamics plot on Barro Colorado Island, Panama; most species clumped together and a small handful (4 species) occupied the extremes of demographic space. These results, as well as ours, conformed to Swaine and Whitmore’s (1988) view of the world of tropical trees.

Figure 4

What is the value of this approach to functionality in tropical trees? It seems that the pioneer/nonpioneer dichotomy sets aside 3 - 11% of trees, pioneers, as having a unique function. What of the remaining nonpioneer species? Clark and Clark (1992) showed that there was significant variation among 6 nonpioneer species they studied at La Selva, Costa Rica. The implication is that we cannot lump all nonpioneers as the same functional type and some coherent scheme of summarizing the variation in their life histories must be found. However, it may be found that the variation among them is nonetheless continuous and any division of nonpioneers into smaller groups will be necessarily arbitrary.

Effects of disturbance on microbial diversity

Previous clearcutting of part of the forest over 70 years ago resulted in dominance of secondary forest tree species in the disturbed area. The secondary tree species were especially susceptible to damage from Hurricane Hugo eight years ago as compared to the mature forest species in the surrounding lightly disturbed area. Microbial diversity has been altered by either the effects of the original human disturbance that changed tree species composition, dominance and diversity, or the second disturbance from the hurricane (which mirrored the pattern of the previous human disturbance), or a combination of these. The microbial community on the forest floor processes large quantities of dead organic matter on an annual basis, and is responsible for recycling nutrients and making them available again for plant growth. Although the microbial community is small compared to the mass of organic material that is recycled, it is analogous to the eye of the needle through which the camel must pass. The microbes are critical in controlling the rate of flow of nutrients to both plants and animals in this forest.

Mike Willig and colleagues have attempted to assess the degree to which the spatial variability of soil bacterial communities is a consequence of historical land use or hurricane impacts at the Luquillo site. Based on the differential catabolism of a suite of 128 carbon sources by soil bacteria communities, they focused on four aspects of microbial functional diversity: total substrate activity, substrate richness, substrate evenness, and substrate diversity. Considerable microspatial heterogeneity characterized the functional diversity of forest soil communities, but the degree of hurricane damage to aboveground plant communities was positively related to all four indices of functional diversity five years after Hurricane Hugo. No differences in functional diversity were detected with respect to historical land use after controlling for the effects of hurricane damage. However, this lack of significance may be an artifact because the spatial distribution of hurricane damage is not independent of historical land use. As a consequence, contemporary studies of spatial heterogeneity that do not account for historical patterns of anthropogenic or natural disturbance may yield spurious or incorrect conclusions. Long-term studies help to rectify this problem and are especially important within the context of evaluating the impacts of increasing human demands on natural ecosystems.

Stevenson and Landolt (unpub.) also found high diversity of slime molds in the more highly disturbed area. Although Huhndorf & Lodge (unpublished data) found no differences in diversity of wood inhabiting ascomycete fungi between disturbance levels, disturbance greatly affected which species were present. Three quarters of the ascomycete species were restricted to either the less disturbed forest or the more disturbed forest, and only one quarter were found in both areas. It is remarkable that a human disturbance more than 70 years ago has left such an indelible mark on the decomposer community.

The status of the amphibian community of the Luquillo Experimental Forest

In order to quantify changes in the amphibian community of the LEF through time, Larry Woolbright and his students have established a variety of standard methods to monitor the appreciable amphibian component of the biota. These include permanent plots (established in 1987) that permit tracking individually-marked adults, auditory transects begun in 1988 that record the composition of the anuran chorus, and nocturnal auditory censuses of grids at El verde and Bisley sites to determine whether local results are general to the study area. The results of these studies are compared internally to themselves as well as to the wealth of information available on the amphibian community of the El Verde site since the early 1960's.

The major result of this work has been to document a 50% decline in species richness of the terrestrial amphibians. Of six species of Eleutherodactylus that were relatively common in the area surrounding the El Verde Field Station in the 1960's and 1970's, three no longer occur in the study area. Eleutherodactylus eneidae disappeared some time around 1980, and both E. portoricensis and E. richmondi disappeared in 1990. The three species have been absent for periods of time that exceed their life spans and thus can be concluded to be locally extinct. Because the are has been protected during the entire period spanned by these data, local extinctions do not appear linked to any specific human activity.

With the exception of the three species noted above, it is premature to draw conclusions about meaningful changes in relative abundance of the rest of the community. Earlier studies typical cannot be directly compared to the measurements being recorded now, but can only establish what species were present at what times. As the collection of current data sets continues. We expect to be able to speak more directly to questions of normal ranges of fluctuations among amphibian communities and what level of change allows for significant conclusions to be drawn.

Canopy invertebrate community response to hurricane disturbance

Study of canopy invertebrate responses to Hurricane Hugo began in February 1991. At that time six experimental blocks were designated in the 16-ha Big Grid at El Verde Field Station. Each block included a 30-50-m diameter treefall gap resulting from Hurricane Hugo and a similarly sized plot in which no trees fell. Canopy coverage was < 25% in the gap treatment and > 57% in the standing forest at the beginning of this study. Five tree species were selected for sampling on the basis of their representation of disturbed forest (Cecropia, Casearia) and undisturbed forest (Dacryodes, Manilkara, Sloanea) and their presence in all plots for comparison. A sixth tree, Prestoea, was included after 1992.

One tree per species was sampled in each plot during the wet season (samples collected June-November) and dry season (samples collected January-March) during 1991-92 and 1994-95. Trees were sampled using a long handled insect net with a closeable plastic bag inserted in the net, as described by Schowalter (1994, 1995).

Overall, invertebrate diversity did not differ significantly among tree species or disturbance treatments. However, several taxa and groups showed significant differences among tree species or hurricane treatments. Tree species was the most important factor affecting canopy invertebrates. Several taxa and functional groups (sap-suckers, omnivores, and detritivores) were significantly restricted to or more abundant on particular tree species. Disturbance initially influenced some taxa and functional groups. Lepidoptera, predaceous beetles and detritivores were more abundant in standing trees, whereas several Homoptera were more abundant in gaps during 1991 (Schowalter 1994, 1995).

Foliage losses to herbivorous insects appeared to be related to tree species abundance, with greatest losses found for Prestoea, followed by Cecropia (in gaps where it formed thickets after the hurricane), Manilkara and Sloanea; losses to Dacryodes were < 1%, probably due to terpenoid defenses (Schowalter 1994 and unpubl. data). Cecropia showed increasing loss to herbivores in gaps from 1991 to 1995. Foliage loss to herbivores generally was low (<20%) during the study period.

Field observation and preliminary analyses of the 4-year dataset indicate that initial differences due to disturbance largely disappeared by 1995. The dataset currently is being analyzed to assess seasonal patterns of abundance, i.e., differences between wet and dry season samples.

TABLE 2. Shannon-Wiener diversity by tree species and hurricane disturbance treatment at the Luquillo Experimental Forest, Puerto Rico).