Complexity Uncertainty And Their Role In Shaping Management Decisions

From a conservation perspective, there were many desirable outcomes after the removal of the sheep and cattle from the western part of SCI. Aerial photographs indicated erosion rates had drastically declined, distribution and abundance of many endemic species increased (Schuyler 1993, Klinger et al. 2002), and recovery of woody species and shrub communities was occurring in many parts of the island (Wehjte 1994, Klinger et al. 2002). In these regards, few if any conservation practitioners would say sheep and cattle removal from SCI was not successful.

It was also clear though that the outcomes from the sheep eradication were more complex than anticipated and often depended on factors other than sheep. For example, relatively little attention had been given to outcomes from the cattle removal, but it was this action that contributed substantially to the expansion of fennel. Another example was pig abundance after the eradication. Densities were sometimes high, but dietary overlap between them and sheep is not extensive (Van Vuren 1981, Baber and Coblentz 1987) and variation in their abundance depended on factors other than sheep abundance (R. Klinger unpublished data). A final example was fire frequency. Plant biomass increased following the eradication of the sheep, but there was not an increase in fire frequency. The greater fuel loads certainly made this concern legitimate, but fire frequency on SCI appears to be related more to human ignition sources than fuel loads (Carroll et al. 1993, Keeley 2006, Klinger et al. 2006a), and human access to SCI has been highly regulated and controlled since at least the 1930s.

In part because of the complex and unpredictable outcomes and in part because of limited data, evaluation of the responses after the sheep were eradicated had to be done with an understanding that there was considerable uncertainty on why things happened, as opposed to what happened, following the eradication. Data were available on some aspects of post-eradication patterns, but they were primarily correlative so strong inferences could not be made.

Even with the successful outcomes of the sheep and cattle removal, TNC was clearly confronted with substantial management challenges. Despite the success of a trial pig eradication (1989-1991) in a 23 km2

fenced area of SCI (Sterner and Barrett 1991) and complete eradication on Santa Rosa Island (1990-1992) by the National Park Service (Lombardo and Faulkner 2000), in 1990 TNC decided not to pursue an islandwide pig eradication program for an indefinite period of time. Consequently, pigs continued to threaten rare plant populations and rooting was severe and extensive in years when their numbers were high. The continued presence of sheep on the east end of SCI made ongoing hunts and fence repair a regular activity (Van Driesche and Van Driesche 2000). Fennel was spreading and non-native annual grasses and forbs dominated the herbaceous layer over a large proportion of the island, resulting in displacement of native plants, including some rare endemic species. These situations resulted in the bulk of the management effort by TNC in the 1990s being focused on phased, long-term studies of the pigs, fennel, and non-native annual plants (Brenton and Klinger 1994, Klinger 1997, Klinger and Messer 2001, Brenton and Klinger 2002, Klinger et al. 2002, Klinger et al. 2004, Ogden and Rejmanek 2005). In addition, experiments related to fennel and other non-native plants were conducted by researchers from the University of California (Dash and Gliessman 1994; Wenner and Thorp 1994; Barthell et al. 2001, 2005).

The same degree of complexity and uncertainty that characterized outcomes of the sheep eradication typified management experiments during the 1990s. In virtually all of the experiments, variation in many plant and animal responses was due more to rainfall patterns than treatment effects (Klinger and Messer 2001, Klinger et al. 2004, Ogden and Rejmanek 2005), a pattern that data from the monitoring program indicated was occurring across the island. It was also apparent that simple treatments alone would rarely if ever result in long-term reduction of non-native plant abundance (Brenton and Klinger 1994, Klinger and Messer 2001, Ogden and Rejmanek 2005). The management experiments and monitoring data provided a great deal of insight on what outcomes could realistically be expected to occur and how long they could be expected to persist. But they also reinforced the difficulty involved in shifting systems dominated by multiple non-native species to ones dominated by native species. This was especially apparent in situations where native animals and plants had different patterns of abundance in the non-native-dominated systems (Figure 18.6; Ogden and Rejmanek 2005). It became clear that management actions favorable to some native species could be unfavorable to others.

Given the complexity and uncertainties associated with outcomes of the sheep eradication program, would application of the EE concept have changed any of the management considerations or decisions related to the program? In all likelihood the answer is no. There were few if any precedents for an eradication on the scale of the sheep from SCI (Schuyler 1993), so it was difficult to precisely know what to expect. The impacts from the sheep were severe and ongoing, the recommendations from scientists and managers were consistently in favor of immediate eradication, and delay could potentially have resulted in extirpation of rare species. Although it wasn't articulated in EE terms, TNC clearly recognized that there was a good chance that populations of non-native species would increase after the eradication. But erosion rates were so severe that establishing vegetation cover, even if much of it was non-native, was imperative. From an EE perspective or not, there is little argument from anyone that it would have been optimal to try to eradicate the pigs at the same time as the sheep (Byers et al. 2006). But TNC was constrained by legal agreements from doing this. And, in hindsight, it is easy to suggest that greater attention should have been paid to potential outcomes from the cattle removal. But given the removal of the sheep it would have been difficult to justify the presence of a couple thousand head of cattle, and it is unlikely anyone would have been able to predict how rapidly fennel would expand. There is little that the EE concept could have done to change these considerations.

In 1993 the EE concept was informally suggested as a potential conceptual basis for some of the management activities on SCI (J. Crooks, personal communication). It was quite obvious that many non-native species on SCI could potentially have ecosystem-level effects, but at the time the concept was relatively new, controversial, and not particularly well developed. While the EE concept may have provided a mechanistic model of what effects were likely from a given species, it had little if any utility predicting what species were likely to become an EE as a result of management actions. Management goals and planning were already relatively specific (Klinger 1997), and the integration of experimentation and monitoring was being used to try to understand the numerous complex interactions and improve predictions of outcomes from large-scale management activities. In addition, the concept of a transformer species was considered a useful and well-accepted mechanistic and phe-nomenological model (Richardson et al. 2000) for describing effects of the fennel, pigs, and non-native annual herbaceous species. For these reasons, while it was believed that the EE concept was interesting from an ecological perspective, it would have added little in a practical sense to what was already in place for planning, implementation, or evaluation of management actions.


Over the last 25 years there has been an upsurge in conservation activities on the Channel Islands and a substantial increase in research applied to management (Klinger and Van Vuren 2000). In large part this is because the islands have been invaded and impacted by so many non-native species. It is almost certain that management programs targeted at these species will be ongoing for many decades, and the EE concept does have the potential to make tangible contributions to these programs. In the last 10 years there has been considerable development of the concept (Wright and Jones 2006), including application to a number of invasive species problems (Crooks and Khim 1999, Byers 2002, Cuddington and Hastings 2004). There is also no reason the EE concept needs to be the only one underpinning management programs; it can be integrated with other ecological concepts such as trophic cascades or food web structure (Byers et al. 2006).

When thinking in terms of non-native species as ecosystem engineers it seems most appropriate to think of not just one species but multiple ones, and how they interact. Many systems have multiple invaders that have been present for long periods of time (Rejmanek et al. 2005b). The relationship between different phases of the invasion process, interactions among multiple invasive species, species abundance, and the strength of per capita effects may be the most helpful aspect of retrospectively applying the EE concept to management of non-native species on Santa Cruz Island. It is unlikely the EE concept in and of itself will be useful predicting what species will become problems on the Channel Islands, however this is an extremely difficult problem that is not inherent just to ecosystem engineering (Kolar and Lodge 2001, Underwood et al. 2004, Rejmanek et al. 2005a, Klinger et al. 2006b). But if linked to the invasion process and focused on species that either have known large per capita effects or are clearly in the spread phase of invasion, the EE concept may be extremely helpful in forecasting real and presumed effects of a species, and how their removal could change various ecosystem properties. Perhaps most important, it makes scientists and managers think about complex interactions and processes, which in and of itself is a crucial step in setting measures of success for management programs and designing potential follow-up programs.

The greatest challenge the EE or any ecological concept faces is the willingness of institutions to look at long-term benefits of applying them in management programs. For instance, 16 years after the end of the sheep eradication program, TNC and the National Park Service began a feral pig eradication program on SCI (Krajick 2005). The multimillion-dollar program has progressed very efficiently and effectively (Krajick 2005). Within 2 years the pigs have become functionally extinct (S. Morrison, The Nature Conservancy, personal communication), and persistent hunting of the remaining few will soon lead to their complete extinction. As with most eradication efforts, the majority of the resources have been targeted at the operational side of the program. Not without justification, success is primarily being measured as eradication (NPS 2003). Systematic monitoring designed specifically for outcomes from the eradication program has been limited to island fox (Urocyon littora-lis) populations, mainly because of the purported relationship between pigs and golden eagle (Aquila chrysaetos) predation on the foxes (Roemer et al. 2002, Bakker et al. 2005). Despite strong recommendations, monitoring of broader community and ecosystem processes has largely been ignored.

In contrast, the approach of monitoring desired outcomes rather than simply eliminating a species is the central focus of non-native species management programs by the Santa Catalina Island Conservancy (J. and D. Knapp, Catalina Island Conservancy, personal communication). Although they are not using the EE concept, they have adopted an adaptive management approach (Holling 1978) because they have observed similar complex patterns following eradication programs for goats and pigs (Garcelon et al. 2005, Knapp 2005) on Catalina as were observed on Santa Cruz. This includes the dominance of many areas by non-native annual plants (Laughrin et al. 1994), the spread of fennel after a lag period of several years, and increased fire frequency as a result of greater fuel loads and human ignitions (Knapp 2005). Their goals are to improve the ability to predict potential undesirable outcomes from their management programs, and to prepare management actions before undesirable outcomes become too extensive to effectively deal with.

More generally, effects from control and eradication programs are clearly landscape-scale phenomena. Therefore, it will be important to think beyond simplistic notions of restoration when setting management goals, and recognize that trying to control or eradicate non-native species will often have very broad ecosystem effects with few predictable trajectories (Zavaleta et al. 2001, Zavaleta 2002). These programs are not ends in themselves, but likely only the first in a long series of ongoing management programs. Just where the EE species concept fits into this situation is difficult to say with certainty, but it is likely that there will be many circumstances where it can be used effectively to help plan and forecast outcomes from non-native species management programs.

0 0

Post a comment