I appreciate Chris Madson’s efforts to include some of the historical context and study surrounding sage-grouse population cycles in his recent WyoFile essay, “Circular reasoning,” and I largely agree with his characterization of the history of population cycle research and his presentation of species biology. I agree that population cycles were overemphasized for years and that for many species claimed to undergo cycles, the evidence is lacking when confronted with data. However, rather than attempt to understand and embrace the complexity of population cycling, Mr. Madson seems to miss what can be learned from nonlinear modelling and rather conclude in response to his subtitle (“Are greater sage-grouse populations really cyclic?”) that sage-grouse populations are not cyclic and are simply in linear decline.
A more accurate answer would be: some populations are cyclic, some are not. As with so many questions in ecology, it depends on the time frame and the spatial scale considered. Rather than oversimplifying complex systems, I believe we should use the tools at our disposal and within reach to understand natural variation in the distribution and abundance of organisms. All conclusions regarding wildlife population change are based on models either explicitly or implicitly. And there is always a tradeoff between explanatory power and model complexity (see: Occam’s razor). We have to simplify the world in order to understand it and choosing the balance between simplification and capturing every detail (i.e., overfitting a model) is one of the most difficult problems we face as research scientists. Mr. Madson’s presentation of our current understanding of population cycles in greater sage-grouse was oversimplified and represents a significant mischaracterization and misunderstanding of current literature. I began to review the article line by line, isolating each premise to assess the cogency of the arguments presented (as I would the dozens of peer-reviewed articles I review every year). However, I realized quickly that this approach would not make for good reading. So, here I present three examples of the most fallacious claims:
(1) “Over the last 70 years, the notion that sage-grouse populations have a 10-year cycle has been accepted with remarkably little technical rigor.“
The eminent scientist J.B. Haldane once wrote there are four stages of scientific acceptance.
- This is worthless nonsense;
- This is an interesting, but perverse, point of view;
- This is true, but quite unimportant;
- I always said so.
I thought that in the over 10 years since I published my first analyses of population cycles in sage-grouse in Wyoming, we had moved (at least) to step three in Haldane’s stages. When I submitted my first publications on population cycling in sage-grouse I was met with dismissive reviews from other professionals in the field. Several reviewers were convinced that it was worthless nonsense while others claimed that what I presented was already known. Since then, I have published six peer-reviewed publications on the topic in top journals that have received over 100 citations in the peer-reviewed literature. The modelling approaches presented in those manuscripts range from basic statistics to more sophisticated approaches including nonlinear modelling, wavelet analysis, and approximate Bayesian computation. Whether you are familiar with the approaches or not, I assure you they are statistically rigorous and represent some of the most technically advanced approaches to describing population trends. This research has been reviewed and critiqued in detail by many experts in the field before publication and I have presented the research at many professional conferences. The “notion” of population cycles in sage-grouse has been challenged at every step by the broader research community to ensure that we can collectively address the complexities of population dynamics across sage-grouse populations. I admit that some of the modelling applications are complex — it is my job to develop and implement unique approaches — and may be challenging to interpret. However, these types of population trend models are used by avian monitoring organizations around the world.
(2) “The “peaks” in sage-grouse counts that supposedly indicate a cyclic high can be anemic, so anemic, in fact, as to raise the question of whether they’re real peaks or just blips in the data, and the counts are often “smoothed” statistically to get rid of inconvenient year-to-year changes that might not be consistent with the expected pattern of the cycle.”
I appreciate the emotive impact of the word “anemic” in this description. It is evocative, but unfortunately, not very precise. I am a writer, but I am a scientist first, so I will aim for precision. Depending on the population in Wyoming, the difference between the low point in a cycle (i.e., nadir) and the high point (i.e., peak) can represent a 20% to 80% change in the population index. If the peak in consideration has not reached the same magnitude as a prior peak, we can quantify the disparity in terms of percent change in the population index (negative) along with the appropriate confidence intervals. These types of changes over relatively short time frames certainly do not seem anemic. Furthermore, cyclical does not imply that all the peaks and all the nadirs are of the same magnitude. For those populations where the model results support cycling, there is often a trending decline in population abundance. This negative trend could be captured even more simply with a linear model, and while that would simplify interpretation, it would miss the complexity of the population cycles that actually occur. It seems abundantly clear that anyone concerned with the conservation or management of a species should be concerned with populations trending toward lesser abundance, and would want to know where in the cycle the population has been in the past and may be in the future.
(3) “Three of the most recent attempts to analyze sage-grouse populations reach startlingly different conclusions.“
The three studies used similar data sets and this statement would be somewhat remarkable if the three studies were asking the same questions. But they were not asking the same questions. There are two related but different components when addressing long-term trends in sage-grouse (and all wildlife populations). One is to address the long-term, overall trends. As noted by Mr. Madson, this approach was applied to excellent effect in the work of Dr. Garton and colleagues in their assessment of long-term trends. I will oversimplify, but essentially Garton et al. were interested in long-term trends and therefore fit linear models to data to estimate long-term trends. I have also focused on long-term trends in Wyoming along with coauthors. In that paper we estimated the long-term decline in Wyoming sage-grouse using a combination of nonlinear modelling and change point analysis. Our results indicated a 54% decline with a 95% confidence interval of -64% to -41% from the peak in 1968 to the peak in 2006. Clearly a declining trend. On this we all agree.
The other important aspect of a trend is the variation around that trend, which in Wyoming, at the statewide level, has historically been cyclic. Oftentimes, when we conduct these types of analyses we use statistical techniques to specifically de-trend the data. That means we effectively remove the overall declining trend because we specifically want to isolate the variation around the trend. These results are clearly presented in my work. (An interesting side note is that in this paper we found sage-grouse cycles highly correlated with rabbit trends in Wyoming.)
The USGS publication cited by Mr. Madson and led by Dr. Pete Coates is an excellent example of how science moves forward. We know that sage-grouse populations are generally declining across the species’ range, and from my research on population cycles, we know there is cycling in some populations. Coates and coauthors have addressed these issues by explicitly modelling the low points (i.e., nadirs) of population trends across the range. This appears to be an eloquent approach to addressing long-term trends — highlighting the most concerning point in a cycle: the point at which a population is at its least — while concurrently addressing the variation around that trend as demonstrated in much of my research I have discussed here.
I have multiple other concerns. For example, Mr. Madson argues “The thing that makes a cycle a cycle, the one characteristic that sets it apart from other kinds of change, is that a series of events leads eventually back to a starting point.” For starters, this statement is patently false in both wildlife ecology and statistics. Perpetuating this myth is harmful to the public’s understanding of ecology and basic science. Furthermore, even if the statement was true, choosing a “starting point” that is ecologically relevant and statistically rigorous isn’t easy. There are dramatic differences in the sampling effort and approach for sage-grouse from the 1960s until today that will fundamentally influence how we understand these systems.
Clearly Mr. Madson is concerned with dishonest actors relying on the existence of population cycles in some populations as an excuse to delay action with the false assumption that all cycles return to their origin. I share this concern and I am always clear regarding the uncertainties in my analyses and the importance of spatial and temporal variation. There are many important management implications for understanding population cycles including the obvious one of trend assessment, but it is also relevant to study design, adaptive management, impact assessment, and others that are detailed in my relevant publications.
Relatedly (and in conclusion), Mr. Madson claims “The concept of the cycle does nothing to help us solve the fundamental problems facing sage grouse in the West.” Sage-grouse are a species of conservation concern. I would argue that one of the “fundamental problems” facing the species is the accurate description of the variation in population abundance. We must endeavor to understand why populations change. If we ignore populations’ cycles — where they exist — we are engaging in willful ignorance. Ecology is complex. Sometimes the answers will also be complex. But that does not mean they are wrong and should be dismissed because they make management and interpretation difficult, or even if the answers may be misused.
Finally, I take some personal affront with Mr. Madson dismissing the thoughtful research efforts my coauthors and I have painstakingly undertaken as a “comforting myth.” If Mr. Madson has data or analyses, or even an informed critique of the methods applied in my manuscripts, I would be happy to hear them. Otherwise, as a scientist and critical thinker, I have to rely on the data, analyses, and conclusions presented in rigorously reviewed and published manuscripts. I encourage others to do the same.