System Cascades

Of all the guiding questions for this book, research on cascades in systems has been the most fruitful. This is because it strikes to the heart of explaining how the systems operate. Indeed, the cascades are the ecosystem responses of our title. The more we know about system operation, the more we will understand the true nature of the system. The complexity and extent of cascades in ecosystems caused by climate impact is due especially to process connections between the living components of ecosystems. In some cases abiotic components also affect the cascades. Initial and intermediate cascade elements may act as gateways, filters, and/or catalysts to the climate signal. Gateways can be open or closed. They can either permit, or not permit, the passage of material, energy, or information. Filters may pass a variable amount of material, energy, or information along through the cascade. The amount varies from all to none and includes all the possibilities in between. Thus, the filters in the system provide a buffering function to a climate disturbance. Cat alysts occur where the presence of one component greatly enhances the interaction of two or more other system components. Another consideration is that the climate event or episode having a potential effect on an ecosystem is often itself a part of a cascade in the climate system. For example, the variability in the values of many surface climate variables at the CAP LTER during ENSO events (chapter 7) occurs at the end of a cascade that started with anomalous sea surface temperatures (SSTs) in the equatorial Pacific Ocean. These SST anomalies entered a series of atmospheric processes that altered the upper part of the atmospheric flow and eventually affected the surface in the Southwest United States.

The complexity and extent of the cascades of climate effects within ecosystems is best illustrated through a detailed case study. The example selected here deals with the cascade of events in the hydrologic system of the Andrews LTER site following a large rain, or rain on snow, event such as that which occurred in February 1996. This is an interesting example for many reasons. First, it is well documented. Second, not only does the example deal with the abstract concept of a cascade, but it also is a very real cascade of water working in a real time sequence over a series of spatially linked channel and topographic components arranged on the landscape from higher to lower elevations. Third, the example focuses on how different processes may display some sequential linkage. Identification of the sequencing of different processes in a cascade is very important. Within the context of a global-scale temperature increase, Shaver et al. (2000) note that the dominant controls over ecosystem response will change over time as different processes change at different rates. Furthermore, the changing sequence will not necessarily be the same in all ecosystems. The following description of what the authors call a "disturbance cascade" is derived from the analysis by Nakamura et al. (2000).

The Andrews precipitation-forced disturbance cascade is a parallel cascade consisting of two initial drivers. The first is small, rapid debris slides from hillslopes. The second is large, slow-moving earthflows. In the first path, debris slides move into steep, headwater stream channels and move through the channels as debris flows. The movement delivers sediments and logs to larger streams. On entering fourth- and fifth-order channels, the debris can be entrained and float along on rafts of coarse woody debris (CWD) or can cause jams at the confluence with larger channels. The jams may break during floods, causing a surge that pushes the debris further downstream. The CWD transport may terminate in areas of accumulations of wood or may be dissipated gradually, as wood levees, along stream banks. In the second path, slow earthflows gradually constrict stream channels. This increases the potential for stream bank erosion and stream side slides during high-flow events. The slides can deliver sediment and trees that form temporary dams. The breakup of the dams triggers flood surges downstream. Associated CWD may move in a congested manner sometimes disturbing riparian vegetation.

The sequence of processes may be interrupted at any point along the flowpath. Sometimes preexisting conditions, such as a change in the channel slope, may halt or alter the nature of the sequence. Roads may intersect the cascade flow path and act as filters or have other effects (Wemple et al. 2001). Occasionally, the cascade sequence will not occur at all. Streamside slides may merely alter a channel location with few downstream consequences. Similar outcomes, such as flood surges, may result from different processes. These disturbance cascades were found to produce a gradient of decreased overall severity of impact on the ecosystem and increased variability of severity of stream and riparian disturbance in the downstream direction. For the February 1996 event, instances of changing processes from one to another and halting in particular parts of the cascade were quantified. Especially noteworthy is that only a very small fraction of all the initial mass movement events resulted in the full cascade sequence. Thus, filtering in this cascade was strongly marked.

This example suggests a case in which the climate event had to pass through a geomorphological cascade before beginning an ecosystem cascade affecting the flora and fauna. The immediate ecosystem effect varied from complete removal of alluvium, soil, and vegetation on steep, narrow, low-order channels to localized patches of toppled trees.

We learn much from the comparison of system cascades identified at the various LTER sites discussed in this book. To start with, cascades can be short or long, intuitively obvious or less obvious, linear or nonlinear. Some short cascades are found in the Antarctic. The remarkable responses of fauna in the glacial lakes of the McMurdo Dry Valleys LTER site to the freezing of the lake surface and the extreme low temperatures represents some extraordinary cascades through the aquatic ecosystem, as described by Fountain and Lyons (chapter 16). Although the responses are extraordinary, the cascades are short because the food chains are short. Fountain and Lyons (p. 334) point out that "the low biodiversity and short food chains make the ecosystem directly dependent on the physical environment such that few buffers exist and the response of the ecosystem to slight climate change is immediate." Short cascades are also seen at a daily timescale at the Arctic LTER, where there is approximately a direct response in Net Ecosystem Production (NEP) to an increase of photosynthetically active radiation levels, among other things (chapter 5). Cascades on agricultural crops also tend to be short (chapter 4). Many of the other cascades described in the preceding chapters are long ones. Often the more we learn about the way the ecosystem operates, the longer the cascades become. So, for example, LTER investigations have shown that a simple relation between high water flow events and increase productivity in the lakes of the arctic tundra actually involves changes in the degree of mixing of lake water and variations in available nutrient content (chapter 5).

Some cascades are simple and intuitively obvious such as the increase of glacial meltwater and streamflow in response to higher radiation input values at the Mc-Murdo Dry Valleys LTER site (chapter 10). Others, such as the increase of traffic accidents in and near Phoenix initiated by a La Niña event, gene switching and the production of new phenotypes in species at the Coweeta LTER site in response to drought (chapter 3), or the hurricane-initiated increase of forest fire danger and possible extensive logging that itself can create huge ecosystem effects (chapter 2), are certainly not intuitively obvious. Neither is the fact that some ecosystem processes may respond in different ways to a given climate episode, in the case of NEP levels in Oregon forests in relation to summer precipitation values (chapter 19).

Many cascades are linear, but we have also recognized nonlinear responses. For example, because summer temperatures are close to freezing point at MCM, the change between liquid and solid water in the hydrologic systems is delicately balanced. Welch et al. (chapter 10) report that small changes in temperature and radiant energy are amplified by large, nonlinear changes in the hydrologic budgets that can cascade through the system. Another example is the response at Luquillo, Puerto Rico, where 75% of the sediment export occurs during the 1% of the days that have the greatest rainfall (chapter 8). Such a nonlinear response can be exacerbated even more when the heavy rain events give rise to debris flows that course down hillslopes and through streams, as sometimes happens both at Luquillo and at the Andrews LTER site in Oregon. In an entirely different environment, the Sevil-leta LTER site of New Mexico, another surprising, nonlinear response to decadal-length drought is hypothesized to be economic collapse and a large number of changes of landownership toward the end of a drought (chapter 15). We learn something new about climate variability and ecosystem response from almost every different cascade. Drought in the corn crop ecosystem may lead to a cascade in which the plant system suffers mortality or becomes weakened and susceptible to insect herbivory or disease (chapter 4). A lesser recognized cascading effect, pointed out by Gage, for agricultural systems subjected to drought is the establishment of more irrigation systems with their subsequent effect on local and regional water tables. In these two last examples, the cascades existing in the ecosystem represent catalysts for later major changes in the human dimensions of local and regional change.

Scientists at the Palmer LTER site believe they have identified an important cascade in their ecosystem. Pygoscelid Penguins are representatives of the higher trophic level in this ecosystem. Population variations at quasi-quintennial and decadal timescales in Pygoscelid penguins have to be understood via a cascade that starts with an entrainment of phytoplankton in newly forming sea ice of the previous autumn. In some senses this could be regarded as a preexisting condition. The cascade continues with the growth of sea ice communities during the winter and the spring release of a potential bloom inoculum of particulate organic matter in the water column. These events are related, in turn, to the survival of larval krill that depend on the algal food source in the sea ice. In summary, there are strong linkages among sea ice, phytoplankton, and krill. The foraging ecology of the penguins is dependent on krill recruitment and abundance, indirectly through habitat changes that mediate the availability of krill (Smith et al. chapter 9). This type of cascade best illustrates the concepts of gateways and filters. Because the sea ice is necessary for phytoplankton development, the sea ice extent represents a gateway. Whether this gateway is open depends on the delicate balance of the sea water temperatures near freezing point. If the sea ice exists, it acts as an open gate. If sea ice is not present, its absence acts as a closed gate and does not permit the development of the phytoplankton that are the first level of the cascade. If the cascade is established then each step in the food chain filters the passage of chemical energy to subsequent steps in the cascade.

The concept of gateways existing in cascades has been developed very explicitly by Juday et al. (chapter 12) in the case of white spruce reproduction in Interior Alaska. These workers identify five climate-mediated gateways in the overall white spruce reproduction cascade. The first gateway is the preexisting condition of the need for a sufficient level of growth reserves. The second gateway is the need for a drought stress signal at the time of the formation of the bud primordia, which occurs at the end of vegetative shoot elongation. A third gateway is the requirement of a lack of severe pruning of reproductive shoots by wind and canopy snow loading in the fall and winter of the first year of seedling growth. A fourth gateway is the requirement of high growing degree-day heat sums to promote the maturation of the pollen and cone buds in time for the remainder of the steps to be completed before frost ends the growing season. Finally, a double fifth gateway requires both the survival of pollen and cone buds in early stages (e.g., lack of killing frost) and a heavy pollen flight (e.g., lack of persistent rains) to ensure high levels of cross pollination. In this example a suite of different aspects of climate variability plays a role in the final successful, or otherwise, species reproduction.

There are at least two temporal elements implicit in the concept of gateways. The first is sequencing. Second, the timing of the open gate must match the timing of other possible constraints. The gate will be open or closed at certain discrete times, and the timing of the opening is important if the ecosystem cascade is to be followed. So, for example, as mentioned previously in the context of preexisting conditions, in the Coweeta forest whether a drought has an effect related to the population dimensions and impacts of the Southern Pine Beetle (SPB) depends on the stage of beetle population. The drought has to coincide with the open gateway of a high SPB population for the cascade that ends in tree mortality to be completed.

The diversity of LTER sites presents a huge variety of potential cascades and the events occurring in these cascades. Nowhere is this more true that in the urban Central Arizona-Phoenix (CAP) site. Brazel and Ellis (chapter 7) list multiple resulting cascades and effects that are strongly driven by ENSO-related climate episodes. Some of the effects are very surprising. A case in point is the increase of traffic accidents associated with the frequent dust storms of La Niña years. ENSO events also partially control the intensity of the Phoenix urban heat island. Even more importantly, Brazel and Ellis point out that many of these cascading effects feed back into the urban ecosystem. The multiple effects of both El Niño and La Niña events on the CAP urban ecosystem suggest an extension of the cascade concept. Much of our attention in this book has been directed to a single cascade in the ecosystem following a climate event or episode. The ENSO climate driver establishes parallel cascades through its precipitation and temperature signals. The analysis in chapter 6 shows that sometimes the temperature effect results in an ecosystem response and sometimes the precipitation effect does. The CAP (chapter 7) case makes us recognize that there can be multiple, separate, parallel cascades to a single climate driver. Indeed, the reality is most likely that multiple climate drivers produce multiple parallel cascades, some of which interact and some of which do not. The MCM case (chapter 10) also identifies several parallel cascades. These relate first to algal mats and stream nitrogen uptake, second to salinity and stability of lake water columns as well as the type of phytoplankton species that are dominant, and third to soil invertebrates. The salmon catch case (chapter 13) refers to parallel cascades that can occur in both coastal and deep-sea ocean waters. The hurricane case (chapter 2) also can have parallel cascades associated not only with wind damage but also possibly to related primary effects such as river floods or salt water inundation, or secondary impacts such as landslides or fires. It is the task of the LTER, and other, researchers to identify the strands of the cascades and their interactions. This is particularly important because Shaver et al. (2000) suggest (in other words) that the longer the time of operation of the ecosystem the greater will be the chance of interaction of parallel cascades. In the case of ecosystem warming, these authors note many changes that will only take place during very long time periods. Such changes include soil profile development, organic matter accumulation, changes in fire regimes, or long-distance movement of herbivores or timberlines. Many such changes will be due to the interaction of parallel cascades.

An inherent characteristic of cascades is the temporal dimension. The importance of timing on the degree of effectiveness of the Southern Pine Beetle in killing trees at the Coweeta LTER site was mentioned in the context of cascade gateways. At the same site shoestring root rot fungus also has a greater impact during drought stress than at times of other climate conditions. The cross-site ENSO study (chapter 6) made it clear that the timing of a climate episode or event is critical if a subsequent cascade of events is to follow. Also the occurrence of hurricanes in New England in October and November when a minimum number of leaves is on the trees makes the forest less prone to hurricane damage (chapter 2). Thus, the timing of an event or episode may also act as a gateway. Studies at the North Temperate Lakes LTER (Robertson et al. 1994) clearly show that a difference of one month in the timing of an El Niño signal can be critical in determining whether that signal will have an effect on the ecosystem. In another example Gage (chapter 4) shows how important early growing-season precipitation is to the eventual corn yield of the North Central Region. The critical importance of the timing of snow-melt at the Arctic site and development and decay of sea ice at Palmer are additional examples (chapters 5 and 9). Whether the ecosystem response gateway is open often depends in these cases on the timing of the climate driver. An extension to the cascade principle and its temporal element is that an ecosystem response may be driven in sequence by two, or presumably more, climate drivers. For example, KNZ NPP is most highly correlated to air temperature in the early part of the growing season, whereas later in the growing season it is better correlated with precipitation values acting through soil moisture conditions (chapter 20).

The geography of the LTER network is important. Geographical considerations demonstrate the importance of cross-LTER site studies. The same initial climate driving function may have totally different effects in different areas. This is especially true when large-scale diving factors such as ENSO or PDO variability are considered. In the case of ENSO, the climate precipitation signal in the Pacific Northwest (PNW) is opposite that of the Southwest. This appears to be the case throughout the twentieth century at the decadal timescale (Schmidt and Webb 2001). In another example, McHugh and Goodin (chapter 11) give an analysis of the way in which the climate, particularly growing-season mean, maximum, and minimum temperature, is inversely associated between the Andrews and the Bonanza Creek LTER sites, respectively located in the Pacific Northwest and Alaska. Times of higher than average values of growing season maximum and minimum temperature in interior Alaska tend to correspond with times of lower than average growing season maximum and minimum temperatures in the Pacific Northwest. This is quite consistent with the salmon catch data referred to in chapter 13 and relates to the distinct manner of operation of the PDO. Intersite comparisons raise exciting new questions. A possible linkage between cone production in the PNW (chapter 19) and Interior Alaska (chapter 12) and the relation to the state of the PDO demands more investigation. In another geographical contrast, the same climate variable ENSO has a larger effect in the SW than in the PNW. In the PNW the difference in precipitation values between El Niño and La Niña years is not so significant as in the SW because the variation is around a higher mean value. As noted previously, the dry, forest-grassland types of vegetation in the SW seem to be "tuned" to ENSO variations, which stimulate grass and fuel production during wet phases and burning during dry phases (Swetnam and Betancourt 1998; chapter 15). Fuel variation is not so dynamic in the PNW conifer forests.

Yet another dimension to the consideration of cascades is exhibited at the Sevil-leta LTER site (chapter 15). Here it is hypothesized that an increase in creosote bush shrub is somewhat self-enforcing because the shrubs emit nonmethane hydrocarbons that act as local greenhouse gasses, keeping minimum temperatures as much as 4°C higher than they would be without the shrubs. Our discussion of cascades so far has been in unidirectional terms. This example shows that we must also consider the possibility of the cascade turning back on itself with a positive feedback. Cases of negative feedback are also conceivable. Yet, we tend to see the cases of cascade elements acting as catalysts in the situations of positive feedback.

0 0

Post a comment