-Draft-

A Perspective On Ecologically Based Pest Management

Bruce D. Maxwell

Department of Land Resources & Environmental Sciences

Montana State University, Bozeman, MT 59717-0312

 

Public attitude associated with use of pesticides and other agricultural inputs has placed increased emphasis on development of ecologically based pest management (EBPM). The Weed Science discipline has reacted by calling for increased research on the basic biology and ecology of weed species in hopes that the "Achilles heel" of each weed can be discovered and exploited to reduce the reliance on management with herbicides and other environmentally invasive practices. The implicit assumption in this approach is that natural weed population regulating mechanisms can be augmented and then in some minimized way, supported with conventional practices where critically needed. In this paper I will state a hypothesis that reveals a potential critical flaw in our assumptions and approach to research and implementation of EBPM. In addition, I offer a starting point for increasing the potential success of research and adoption of EBPM.

Pesticides and intense tillage overwhelm biological systems with mortality, often at a single stage in the life-cycle of the pest. A basic premise of EBPM is to impose multiple and more subtle impacts on biological processes that regulate pest population dynamics toward a new lower equilibrium. I propose that subtle manipulation of biological processes to manage pests may be possible, but it is often not recognized that this approach, when applied in the field, will produce unpredictable responses in pest populations and subsequent crop impact during the early stages of conversion from conventional management. If the typical length of our experiments continues to be only 2 to 3 years then we may only see the stage in EBPM introduction where the pests are raging out-of-control. Organic growers have observed that conversion from conventional pest management is extremely difficult in the first few years until the system is allowed to rebuild ecological buffering capacity that results in natural pest population regulating mechanisms.

Fundamental Problems In Approach

EBPM relies on manipulation of the natural pest regulating processes to minimize system inputs. Natural regulation is complex and therefore often unpredictable especially in systems with a low ecological buffering capacity. Ecological buffering is thought to be highest in systems with a diversity of species at many trophic levels and many food web connections allowing for high resiliency and low temporal variability. The evidence is increasing that more diverse cropping systems have fewer pest problems (Liebman and Dyck, 1993; Andow, 1986, Altieri et al., 1978; Francis, 1990; Wagstaff, 1987). Therefore, buffering capacity may be relatively low in agroecosystems with few species. If this phenomenon exists, one may hypothesize that crop yield could be the same in a high versus low buffered system, but variation in yield may be lower in the more highly buffered system (Figure 1). This translates into a decrease in risk for the producer in the high ecologically buffered system.

Figure 1. Hypothetical crop yield and variance over a long period of time (+10 yr) in a monculture (low ecological buffering) and polyculture system (highly buffered system).monocultures with high pesticide inputs).

Taking this hypothesis one step further, one may speculate that in the transition from a high input monoculture system to a more diverse cropping system that yield may drop and variance in yield may increase drastically at the early stages of the transition. The decrease in yield would reflect pest population spikes and the lack of natural feedback regulating mechanisms (i.e. other organisms at all trophic levels) in response to a decrease in pesticide inputs (Figure 2). The high variance would reflect differential rates of increase and non-equilibrium dynamics in the pesticide influenced populations. Following this hypothesis, as the transition progressed, the variance should slowly decrease and as pest populations come under increased natural population regulation yields should increase.

Figure 2. Hypothetical crop yield and yield variance over a system in transition from monoculture high pesticide use system to a more diversified low pesticide use system.

Research and management implications associated with this hypothesis would include: 1) possible false interpretations of systems in transition, and 2) possible false expectations from systems in the early stages of transition. The solution may include longer-termed experiments that emphasize the study of variation rather than mean responses. In addition, a strategy may be adopted that seeks to spatially segregate and decrease the risk during the transition process. Small patches could be initially omitted from pesticide applications and even planted with a diverse assemblage of species with a gradual increase in the number of patches. In these areas, organisms including pests, have a refuge to begin a process of interacting and natural regulating processes can bring the community back to or near equilibrium. Spatial segregation avoids the problem of suffering potential major pest outbreaks that may accompany the initial stages of release from pesticide pressure. This plan for transition is dependent on good spatial representations and subsequent monitoring of the pest populations and as many known regulating factors as possible.

Upon monitoring of the pests, it occurs to me that there are two philosophical approaches to analysis that can lead to application of EBPM. The first, is the reductionist approach which assumes that only by close scrutiny of the variation and interactions among the pest population regulating variables, can we hope to identify and predict the pest dynamics and thus develop prescriptive management. The second approach, relies more on a study of the general patterns of variation and behavior in the pest communities and populations in response to general manipulations (e.g. polycultures decrease pest impacts relative to monocultures). The latter approach may be the more effective way to apply EBPM, at least initially.

Pest management scientists and extension specialists must change their research perspective from prescription to concept based approaches to management. Variability in pest populations under EBPM will doom this alternative approach if we try to apply it as prescription. Site specific approaches to farm field management may offer innovative solutions to understanding and managing around variability, but it leaves the prescriptions to be independently developed by applying the concepts within each field. Our challenge as a discipline is to identify ways to deliver the concepts and provide mechanisms where the producer or crop consultant working directly with the producer can apply the concepts into specific, within field, management practices. In addition, we must teach our clientele how to ask for concepts rather than prescriptions and let them apply there accumulated site-specific knowledge along with our concepts to realistically accomplish EBPM.

Solutions For The Success of EBPM

The solution to the impediment to EBPM begin with research focused on determining what natural regulating mechanisms exist and how they vary over a spectrum of production systems that range from intensive pesticide use to organic. A few long-term cropping system comparison experiments exist, however few of them are focused on studying the variation in crop and pest components. Instead, they report mean responses in their dependent variables.

I suggest one way for weed scientists to engage in the initial steps of developing an understanding of variation in weed population regulating mechanisms, during the conversion to EBPM, is to characterize the variation in weed impacts on crops. Crop response to density of pests is the basis for identification of pest density thresholds for management, which is the first principle of IPM. There are an increasing number of data sets which measure crop yield response to a range of weed densities over several years and on sites spread over cropping system regions (e.g. Lindquist et al., 1996). Analysis of regional data sets has indicated two salient features. First, in all of the systems there is significant site-to-site and year-to-year variation in the impact of weeds on the crops (for example, Figure 3).

Figure 3. Spring barley response to wild oats over two years (left) and two sites (right) demonstrating variability in response to interspecific interference (Maxwell, not published).

 

Differences in weather may explain a large proportion of the year-to-year variation and climate, soil differences and genetic variation may account for most of the site-to-site variation. However, it is not clear what specific processes, or how the processes that regulate the interaction between weeds and crops, are influenced by the weather or climate. When lists of candidate pest population regulating processes are identified, it must be recognized that these processes may function at very different spatial and temporal scales. For example, basic soil properties related to geologic parent material may vary at large scales (10+ km and 1000+ years), whereas biotic factors like plant pathogens may vary on relatively small scales (cm and minutes). This point alone indicates that an important feature of EBPM research and subsequent reliability of recommendations will be based on understanding the spatial and temporal distribution of pests and their population regulating factors at the smallest scales where significant variation in crop response occurs. Generally, research has indicated that the scale of crop impact spatial variability is within field and may be significant within a few meters for some species.

The second feature of the regional data analysis of crop impact by weeds, is that variation tends to increase at low weed densities (Figure 3). By decreasing the intensity of a population regulating factor (e.g. density), other factors are allowed to exert their influence on the dependent variable (e.g. yield) and thus increase the variation in response to the first factor (e.g. density in Figure 3). The increased variation decreases the ability to use density to predict the outcome of the relationship. The decreased predictive ability ultimately increases the risk for producers, because they usually make decisions at the low densities. Weed density thresholds for management are typically within the high variation region (low density) of the crop yield response. Understanding the interaction of the weed population regulating factors that become increasingly important in this region of the curve will be critical in order to predict weed thresholds and subsequent management response of the pest and the crop.

The interacting pest population regulating factors are so numerous and complex that we need an efficient way to identify hypotheses that give insight into the important interactions and factors. Several researchers offered approaches that utilize simulation modeling, factor analysis and other multivariate techniques. Another approach to accomplish the pertinent hypothesis generating goal is to monitor pest distribution by mapping and thus empirically characterize their spatial and temporal dynamics. The ability to concurrently (within the same growing cycle) map crop yield with a GPS linked yield monitor, provides the ultimate response of interest. In addition, inputs and some factors causing background spatial variability (soil types and qualities) can be mapped and key relationships identified.

The daunting complexity associated with predicting pest population dynamics and resultant variation in response of crops to pests seems to discourage extrapolation of experimental results across time and space (i.e. to farms not immediately adjacent to experiment stations and/or to different years). Thus, management recommendations based on ecological knowledge of the pest-crop ecosystem would seem highly risky and doomed. The new ability to capture ecological information in common farming practices at the sub-field scale through sampling, sensors and GPS/GIS technologies may offer site-specific information as well as repetition over time to provide local estimates of pest population dynamics and crop impacts and thus overcome the problems associated with extrapolation of information and predictive models based on research station experiments.

Selected References