Nội dung

chapter sixteen Monitoring the state of the lake trout resource: a landscape approach Nigel P. Lester Ontario Ministry of Natural Resources Warren I. Dunlop Ontario Ministry of Natural Resources Contents Introduction Evaluating the state of a lake trout population Criteria and indicators Lake trout abundance Angling CUE Mortality rate Angling effort Rapid assessment methods Mark–recapture estimates of lake trout abundance Index fishing CUE Angling CUE Mortality rate Angling effort Elements of a landscape-level monitoring program Selecting lakes Measuring state of the resource from a sample of lakes Selecting a sample of lakes Sampling through time Discussion Acknowledgments References © 2004 by CRC Press LLC Introduction Oligotrophic lakes … should be recognized for what they are, swimming pools carved out of granite, with low nutrient tributaries and a cold annual thermal regime. They are capable of having their environment and their communities severely altered — only too easily. Unless further increases of eutrophication and exploitation are brought to a halt, the lakes will be altered within the next three decades and their demise as producers of salmonid stocks may be irrevocable. Ryder and Johnson, 1972, p. 948 Thus spoke the prophets 30 years ago. Were they correct? Have oligotrophic lakes been seriously eroded by human-induced stress? The anecdotal evidence is overwhelming, indicating that lake trout Salvelinus namaycush populations in many lakes have been seriously degraded (Evans and Willox, 1991; Evans et al., 1991a, 1991b, 1996). The extent of the damage, however, cannot be assessed objectively because the required data are not available. Much of our lake trout data come from case studies instigated in response to a perceived problem. Such data have enhanced our understanding of the stressors that impact lake trout, but they are not very useful in addressing questions about the overall condition of the lake trout resource. What is the current state of the resource? How many lakes have been degraded? What is the spatial extent of the damage? How much change has occurred? What is the rate of change? These questions cannot be answered simply by compiling the available historical data. A data collection approach that acknowledges the spatial and temporal scales of the questions is required. Evaluating change in the state of a resource requires commitment to a long-term monitoring program that collects an unbiased sample of data. Such a program does not exist for the lake trout lakes of the southern Precambrian Shield. There are approximately 3000 lakes in this population. The Shield spans two Canadian provinces (Ontario and Quebec) and two American states (New York and Minnesota). In Ontario, there are approximately 2200 lake trout lakes; 20 of these are monitored by a network of Fisheries Assessment Units (FAUs). Data from these lakes are potentially useful, but the small sample size does not supply precise indicators of resource status at the landscape level. Other lakes in Ontario are sampled on an ad hoc basis by management offices. Because these inspections are often motivated by public concerns, the data cannot be used to monitor the overall condition of the lake trout resource. In Quebec, approximately 25 lakes are sampled on a regular basis (at least once every 5 years), but as in Ontario, a program that monitors lakes at the landscape level does not exist (D. Nadeau, 2000, personal communication). One reason for the lack of a large-scale lake trout monitoring program is the apparent cost. Traditional monitoring programs involve intensive measurement of a large number of environmental variables repeated through time at a small number of fixed sites. Applying the same methods on a larger spatial scale (i.e., many sample sites) would result in an impossibly expensive monitoring program. This does not have to be a problem. The cost per lake can be managed in several ways to increase the number of lakes that can be sampled. First, the program must be focused, monitoring only the variables that are needed to report on the condition of the resource. As Walters (1997) pointed out, scientists asked to develop a monitoring program will almost certainly identify a large set of variables to measure rather than focusing on key response variables. This strategy supplies data that may be needed to explain changes that occur, but it greatly increases the cost of the © 2004 by CRC Press LLC program. If the purpose of the program is to report change, not to explain it, then only key indicator variables should be monitored. Second, the program must be creative, seeking new cost-effective methods of measuring indicator variables. Traditional methods of fisheries assessment call for intensive studies to monitor key variables within a lake. Such methods cannot be afforded when the survey domain is a population of lakes rather than a population of fish. The development of a rapid assessment technology is needed to sample at the landscape level. “Rapid” implies a cost-effective, yet scientifically valid, method of measurement (Hoenig et al., 1987; Oliver and Beattie, 1993; Jones and Stockwell, 1995). Less-intensive sampling of individual lakes is one option that produces a more rapid assessment. Although this results in less-precise assessment of individual lakes, it allows more lakes to be assessed. Because the objective is to describe a population of lakes (not the state of individual lakes), a larger sample size in terms of lake number can compensate for a lack of precision in individual lakes estimates. In this chapter, we describe an affordable approach to monitoring the health of the lake trout resource. First, we discuss how the state of lake trout lakes can be evaluated based on a model that describes the expected response to fishing. Second, we use the model to identify a set of indicators and criteria for evaluating state. Third, we describe cost-effective methods of collecting data to provide estimates of these indicators. Finally, we discuss how to sample a population of lakes to obtain precise estimates of resource status and monitor changes over time. Evaluating the state of a lake trout population Major stressors of lake trout populations in the 21st century are expected to be angling exploitation, changes in fish community, and habitat changes. Angling pressure is already a major stress in areas where lakes are located close to urban centers and are easily accessed (Olver, 1991; Olver et al., this volume, Chapter 11; Lewis et al., 1990). In other areas, many lakes have been protected because they are remote and difficult to access, but this protection will disappear as human populations expand and further development of roads facilitates access. Already the network of forest access roads created in Ontario has resulted in a small portion of the Boreal landscape farther than 10 km from the nearest road (Gunn and Sein, 2000). Changes in the fish community and habitat are expected partly as a by-product of human encroachment on northern landscapes, but also because of global changes. Climate warming is expected to change the thermal and optical properties of Shield lakes (Schindler, 1998; Schindler and Gunn, Chapter 8, this volume), affecting the amount of habitat that supports production of lake trout and other species and driving changes in fish community structure. Changes in the fish community will be exacerbated by species introductions caused by humans (Vander Zanden et al., Chapter 13, this volume). Changes in habitat also affect lake trout abundance, but their avenue of effect is more variable and difficult to observe than that of angling (Evans et al., 1991b). Habitat changes can result from many factors (eutrophication, silt loading, acidification, climate warming, water-level fluctuations). Some of these factors can have acute effects on lake trout survival (mainly during early life stages), but their effect is often indirect, affecting the amount of habitat that is suitable for lake trout. For example, eutrophication shrinks summer habitat due to oxygen depletion in the deep layer of the hypolimnion, which provides a thermal refuge for the cold-adapted lake trout. Similarly, climate warming is expected to shrink summer habitat in small lakes (i.e., <500 -ha) due to deepening of the thermocline (King et al., 1999a, 1999b; Clark et al., Chapter 6, this volume). The effect of these habitat changes © 2004 by CRC Press LLC is increased crowding and thus higher levels of competition and cannibalism. In short, they reduce the carrying capacity and potential abundance of lake trout. Whereas the potential abundance of lake trout is limited by habitat and the biotic community, the observed abundance depends on angling pressure. Angling kills fish. Because anglers prefer large fish, sustained pressure reduces the number of adult fish and thus the production of eggs and new recruits. Although compensatory mechanisms (i.e., density-dependent growth and mortality) counteract these effects (e.g., Fabrizio et al., 2001; Negus, 1995; Rose et al., 2001), the degree of compensation is limited. The net result is that angling reduces the abundance of adult fish. In prescribing a healthy level of lake trout abundance, the inherent capability of a lake to support a lake trout population and the effect of different levels of fishing on the population’s ability to sustain itself must be considered. Also, one must be prepared to make compromises. Sustainability is maximized by eliminating the stress of fishing, but this option is generally not acceptable. Human exploitation of lake trout will persist, and the role of management is to choose a level of exploitation that will allow its persistence. This choice dictates which reference values will be used to evaluate the state of a population. For much of the 20th century, fisheries management was dominated by the single objective of achieving the maximum sustained yield (MSY) from a stock. Larkin (1977), in his now-famous paper, “An Epitaph for the Concept of Maximum Sustainable Yield,” argued that MSY was not attainable for single species and must be compromised to reduce the risk of collapse and to accommodate the interactions among species that comprise the aquatic community. Collapses of commercial fisheries worldwide have supplied ample evidence that an MSY-based approach makes stocks vulnerable to overfishing and collapse. These failures have fostered a more conservative “precautionary approach” (Food and Agriculture Organization, 1995, 1996; Mace, 2001) that defines a new role for MSY. MSY is now viewed as a threshold rather than a target. It defines a “limit reference point,” specifying a level of exploitation to be avoided to safeguard the long-term productivity of a stock. This concept is embodied in several U.N. Food and Agriculture Organization agreements and guidelines. Here, we use the concept to supply a reference point for evaluating the state of a lake trout population. We define the state of a lake trout population as healthy if lake trout are abundant and likely to remain so. Given this definition of a healthy state, diagnosis requires at least two things: a measure of abundance and a reference point for classifying abundance as high or low. If this test implies abundance is low, then diagnosis is complete: the population is not healthy. If the test says abundance is high, additional data and criteria are needed to evaluate stress. One must then decide whether current stress levels are likely to drive abundance down (below criterion). If the answer is yes, the population is at risk and is not deemed healthy. A 1998 model of lake trout exploitation (Shuter et al., 1998) supplies a framework for setting these reference values. The model describes the expected equilibrium relationship between lake trout angling yield (kg⋅ha−1) and fishing mortality rate (a direct measure of fishing stress). The relationship depends on lake characteristics that affect growth, reproduction, and natural mortality rates of lake trout and determine the carrying capacity of a lake. The example shown (Figure 16.1A) describes the expected relationship for one lake type (Area = 1000 ha, total dissolved solids [TDS ] = 26 mg⋅l−1). It demonstrates that angling yield Y has a dome-shaped relationship with fishing mortality rate F. When no fishing occurs (F = 0), yield is zero. As F increases, yield increases initially and attains a maximum level (MSY = 1 kg⋅ha−1) when fishing mortality rate Fmsy reaches 0.21 year−1. At higher levels of F, yield decreases and reaches zero at a fishing mortality rate (0.32 year−1) that drives the population to extinction. © 2004 by CRC Press LLC Figure 16.1 Predicted response to fishing stress for a 1000-ha lake with TDS = 26 mg⋅l−1 (based on Shuter et al., 1998): (A) how equilibrium yield changes as fishing mortality rate increases; (B) corresponding change in lake trout abundance; (C) and (D) effect of a change in habitat that reduces lake trout carrying capacity. This dome-shaped relationship is due to the progressive reduction in lake trout abundance N as fishing mortality increases (Figure 16.1B). The abundance of fish (vulnerable to angling) is about 7.4 fish⋅ha−1 when no fishing occurs. It drops to 5 fish⋅ha−1 when the fishing mortality rate reaches Fmsy , the value that maximizes yield. Then it decreases more rapidly as F approaches the extinction value. The peak of the yield curve MSY supplies a useful reference point. The corresponding value of fishing mortality rate Fmsy sets a prescribed upper limit on the level of exploitation © 2004 by CRC Press LLC (Caddy and McGarvey, 1996). Thus, it supplies a criterion for judging health based on the level of stress. The corresponding value of abundance Nmsy supplies an abundance criterion for judging health. Jointly, these criteria can be used to classify a lake into various stages of fishery development (Lester et al., 1991). These stages are identified as four quadrants in Figure 16.1B: • Stage 1 (healthy): low fishing mortality and high abundance. These conditions are expected during the early stages of fishery development and in rehabilitated fisheries. They indicate the population is healthy. The population is not overexploited, and abundance is roughly at the correct point given the current level of fishing. • Stage 2 (overexploited early): high fishing mortality and high abundance. These conditions are expected only during the early stages of overexploitation because stable combinations of fishing mortality rate and abundance do not exist in this quadrant. It represents a transient stage in an overexploited fishery and indicates a decline in abundance is expected if mortality remains high. • Stage 3 (overexploited late): high fishing mortality and low abundance. This state indicates that the lake is overexploited, and the expected decline in fish abundance has occurred. A reduction in mortality rate is expected to increase sustainable harvest. • Stage 4 (degraded): low fishing mortality and low abundance. This state indicates that the lake was probably overexploited in the past. It can result because management has imposed regulations to reduce fishing pressure on a Stage 3 fishery. It is also expected in the natural course of fishery development because anglers are likely to shift their effort to other lakes once their catch rates suffer due to the decline in abundance. The model predicts that stable combinations of abundance and mortality do not exist in this quadrant. If fishing mortality rate is kept low, a gradual transition to Stage 1, and the eventual reestablishment of stable, highabundance levels, should occur. The fishery would then be described as rehabilitated. However, changes in the fish community resulting from the reduced abundance of the harvested species could slow this recovery or prevent it (Walters and Kitchell, 2001). This schema suggests that measuring the state of the lake trout resource is a simple exercise. Estimates of abundance and fishing mortality for a sample of lakes could be compared to critical (MSY) levels predicted for each lake. The proportion of lakes judged as healthy (i.e., Stage 1) would supply an overall health index for a set of lakes. This index could be used to monitor changes in the condition of the resource if sampling was repeated at regular intervals. One complication is that a lake’s potential to produce lake trout may change, and this can affect criteria for judging its health (Shuter and Lester, Chapter 15, this volume). For example, if habitat changes caused by global warming reduce carrying capacity for lake trout, the yield–mortality curve will shrink (Figure 16.1C), the abundance–mortality curve will rotate downward (Figure 16.1D), and expected abundance at maximum sustainable yield Nmsy will be lower. Use of this abundance criterion is more likely to result in a healthy prognosis than would the higher abundance criterion discussed above (Figure 16.1B). It is important, therefore, to identify which benchmark will be used in diagnosing health. In this chapter we use a 20th century benchmark, basing reference levels on a current model that describes potential production of different types of lakes (Shuter et al., 1998). Because our objective is to describe ways of monitoring change in the condition of the resource, fixed criteria are needed. One consequence is that Stage 4 lakes may exist due © 2004 by CRC Press LLC to changes in habitat — historically high angling pressure does not have to be invoked as the cause. Either way, the lake would be classified as degraded, and this result would contribute negatively in reporting the health of the lake trout resource. Criteria and indicators Because lakes vary naturally in their ability to produce lake trout, appropriate criteria for evaluating state will differ among lakes. The expected yield curve depends on various factors, including the availability of suitable habitat, as well as the growth, maturation, and natural mortality of lake trout. Much of the variation in these factors is predicted by two easily measured lake parameters: surface area and total dissolved solids (TDS) (Payne et al., 1990; Shuter et al., 1998). TDS, an index of nutrient level (Ryder, 1964, 1965), has a positive effect on early growth rate (i.e., ω), measured as the slope of the growth curve at the origin (Figure 16.2A). Lake area is correlated with the asymptotic size of lake trout and maximum yields: Larger lakes tend to produce larger lake trout (Figure 16.2B) but support smaller maximum yields (Figure 16.2C). Size at maturity, size of first capture (Figure 16.2B), and natural mortality rate (Figure 16.2D) also vary with lake size and TDS. On larger lakes, fish mature and become vulnerable to angling at a larger size. Mortality rate in unexploited lake trout populations is positively correlated with early growth rate and negatively correlated with asymptotic length (Shuter et al., 1998), as predicted by Pauly’s (1980) empirical formula. Consequently, natural mortality decreases with lake size and increases with TDS (Figure 16.2D). The effect of these lake parameters on MSY levels of abundance and fishing mortality rate are described below for three lake sizes (100, 1,000, 10,000 ha) and three TDS values (13, 26, 92 mg⋅l−1, median and 5 to 90% range). We also develop critical values of other indicators (angling effort and catch per unit effort [CUE]) that could be used to evaluate abundance or fishing stress. Appendix 16.1 supplies formulae for calculating critical values for any combination of lake size and TDS. Lake trout abundance Critical levels of Nmsy (Figure 16.3A) refer to fish that are vulnerable to angling. Smaller lakes are expected to have a higher abundance of vulnerable fish when they are exploited at the MSY level. This result is somewhat misleading because the fish length criterion used to define the vulnerable population varies with lake size (Figure 16.2B). The lake size effect is less when critical abundance is calculated for a fixed minimum size of fish. A minimum size of 40 cm (the approximate size at first maturity) gives critical values that range from approximately 6 fish⋅ha−1 on small (100-ha) lakes to less than 2 fish⋅ha−1 on large (10,000-ha) lakes (Figure 16.3B). TDS has a positive effect on the critical abundance, but this effect is small compared to the lake size effect. Angling CUE The catch rate of anglers (CUE) is often used as an indicator of fish abundance and could be used to evaluate the state of lake trout populations. The CUE expected at MSY conditions can be calculated as CUEmsy = q N msy © 2004 by CRC Press LLC (16.1) Figure 16.2 Expected growth, yield, and natural mortality rate in lake trout populations (based on Shuter et al., 1998): (A) relationship between initial growth rate ω and TDS; (B) relationship between lake area and size parameters asymptotic length L∞, length at maturity, and initial length of capture by anglers; (C) relationship between lake area and MSY of lake trout; (D) predicted relationship between lake area and natural mortality rate M. In graphs B, C and D, results are shown for three levels of TDS corresponding to the points shown in graph A. where q is the angling catchability coefficient, and Nmsy is the abundance of lake trout vulnerable to angling. An estimate of q is available from Shuter et al. (1998). They found that q is not constant. It varies inversely with fish abundance: q= © 2004 by CRC Press LLC 0.14 (1 + 0.35 N ) (16.2) Figure 16.3 MSY levels of lake trout abundance and angling CUE: (A) Nmsy, (B) angling CUEmsy, (C) N40msy, (D) angling CUE40msy. Using this estimate of q, we calculated CUEmsy for different lake sizes and TDS levels (Figure 16.3C). We also calculated CUE40msy, the catch rate of fish larger than 40 cm CUE40msy = CUE40 N40msy N msy to supply a large fish abundance criterion (Figure 16.3D). © 2004 by CRC Press LLC (16.3) These calculations imply that CUEmsy ranges from 0.15 to 0.34 fish⋅h−1, depending on lake size. Small lakes are expected to support a higher CUE when exploited at MSY levels. The lake size effect disappears when only large fish are considered. At the median value of TDS (32 mg⋅l−1), CUE40msy ranges from 0.12 to 0.14 fish⋅angler-hour−1. Mortality rate Critical values of fishing mortality rate increase with lake size (Figure 16.4A) and TDS. As lake size increases from 100 to 10,000 ha, Fmsy (at the median TDS value) increases from 0.14 year−1 to 0.29 year−1. The lake size effect is partially due to differences in the natural mortality rates, which decrease with lake size (Figure 16.2D). Figure 16.4B shows that the range in critical values of total mortality rate (Zmsy = Fmsy + M) is less than the range in Fmsy. On small (100-ha) lakes, total mortality rate (at median TDS) is about 0.4 year−1 (33% annually) when sustained yield reaches a maximum. On large lakes (10,000 -ha), Zmsy is about 0.5 year−1 (40% annually). Angling effort In the same way that angling CUE is used as an indicator of abundance, angling effort can be used as an indicator of fishing mortality rate. Fishing mortality rate is related to angling effort E as F = qE (16.4) where q is the angling catchability coefficient discussed above. Thus, estimates of the critical effort intensity (angler-hour⋅ha−1) can be calculated as Emsy = Fmsy q (16.5) where q is given by Equation (16.2). Critical values of effort Emsy decrease with lake size (Figure 16.4C). As lake area increases from 100 to 10,000 ha, Emsy shrinks from around 6 angler-hour⋅ha−1 to values less than 4 angler-hour⋅ha−1. This result seems counterintuitive given that Fmsy shows the opposite trend, increasing with lake size. The difference is due to density-dependent catchability q. Because q increases as abundance decreases, strange things happen. Rapid assessment methods In this section we discuss various rapid assessment methods that could be used to measure lake trout abundance and fishing stress in a large set of lakes. Our intent is to identify a cost-effective suite of methods to allow assessment of multiple lakes. We discuss three methods of measuring lake trout abundance (mark–recapture, index fishing CUE, and angling CUE) and two measures of fishing stress (mortality rate, angling effort). We end with a brief discussion identifying which elements should be included in a landscapelevel monitoring program. © 2004 by CRC Press LLC