On theoretical projects and the role of theory and empiricism in my work
Background
I have been
often asked where I fit in standard categories like theoreticians vs.
empiricists or plant vs. animal ecologists. For such questions, there is no
simple answer. Since my first mountain tour, I have been driven by an interest
in the arctic-alpine terrestrial nature. The exact direction of this interest
has dependent on discoveries of open questions and unexplained patterns.
The first
thing I was amazed about was to find patches of spectacular vegetation – with
plants much taller and more showy than typical for the middle alpine tundra
– even in glacier-covered high alpine
landscapes. How these plants managed to thrive so well in such harsh
environments was difficult to understand.
To be
equipped to tackle such problems of adaptation ecology, I planned to combine
plant ecology with plant physiology and started my studies with such
combination. At that stage, animals seemed uninteresting. Their contribution to
community biomass and energy flow was negligible, and a summer visitor like me
could seldom see any signs of herbivory.
During the
lemming outbreak of 1970, the animal component of the system became suddenly
visible, giving some second thoughts….
…..but it
was only in the next year, when my perspective really changed. Winter grazing
by lemmings and voles had influenced the landscape in a way, which was totally beyond my
imagination. Not much was left of the moss cover and even the least edible
dwarf shrubs (like Cassiope tetragona,
shown in the plant picture above – so
loaded with terpenes that it makes perfect fuel even
in pouring rain), were clipped and destroyed over enormous areas. After having read Tihomirov’s
(1959) classical booklet on plant animal interactions on the tundra, I
understood that what I had seen was just business as usual in the arctic. I
realized that I had missed something really central about the tundra – and
about the potential of herbivorous animals to influence the terrestrial
vegetation in general.
To tackle
the issue, I first asked myself why corresponding events were so uncommon
elsewhere. To my judgment, the most plausible answer was provided by the Green
World Hypothesis (GWH) of Hairston, Smith and Slobodkin,
stating that the collective density of herbivores is regulated by the
collective action of predators. The critics of GWH seemed to be stuck with
peripheral semantics (Ehrlich and Birch) or were unaware of the enormous impact
of domestic grazers on
Exploitation Ecosystems Hypothesis (EEH)
In 1975, I
got exposed to Steve Fretwell’s then still
unpublished hypothesis concerning the role of energy in food web dynamics.
While some parts (e.g. the idea of marshes as “four link ecosystems”) seemed
dubious, the general idea that primary productivity and trophic dynamics have
something to do with each other appeared sensible, at least for endotherms,
which spend lots of energy to maintain their body temperatures and are thus
vulnerable to starvation if the density of their resources is too low. I thus
took Mike Rosenzweig’s three dimensional exploitation model, with fixed
consumer isoclines, as my point of departure, and deduced the position of the
plant isocline by assuming that plant biomass growth is logistic, herbivores
have type two functional response and, most importantly, the potential
productivity of the area determines the maximum rate of biomass growth in
plants and the maximum sustainable density of plant biomass. For terrestrial
systems, I regarded potential primary productivity as synonymous with AET
(Actual Evapo-Transpiration). (The size of the
nutrient pool is often important in local scale, but in the large scale
productivity gradients, AET rules.) The qualitatively different outcomes of
this exercise are summarized in the figure below (green, P = plants, blue, H =
herbivores, red, C = carnivores/predators); notice that subfigure 1 actually
consists of two subfigures; representing a very unproductive system (1a, the
pale green plant isocline, which does not meet the herbivore isocline) and a
slightly more productive system (1b, the bright green plant isocline that forms
a locally stable herbivore-plant equilibrium).
1 2 3
With these
premises, I found that there is a wide range of AET-values (subfigures 1b and
2), where plant and herbivore isoclines meet but the equilibrium point lies below
the predator isocline, implying that the intensive interaction is between
herbivores and plants. If the equilibrium is locally unstable (as in subfigure
2), generating violent plant-herbivore cycles, herbivore peaks are indeed
predicted to reach levels well above the predator isocline, making the system
periodically attractive for predators, but at that point, the herbivores are
already consuming the capital and are doomed to crash, due to inevitable forage
depletion. Only in productive areas (subfigure 3), all three isoclines meet,
creating systems with a stable predator-herbivore equilibrium or with cyclic or
chaotic dynamics, primarily driven by the predator-herbivore interaction.
Really unproductive areas, in turn (represented by subfigure 1a) are predicted
to be free from herbivorous mammals and to have plant biomass at the maximum
level sustainable by habitat conditions. In the spatial scale relevant for
herbivores, plant biomass should be low (that’s why they cannot make it in
these habitats), but if the habitat is heterogeneous in a small scale, it could
indeed harbor patches with fairly high plant biomass – as I had seen on the
ridges above the glaciers. In other words, the results of the analysis made
empirical sense.
The model –
known as Exploitation Ecosystems Hypothesis or EEH (Oksanen et al. (1981) -
has now been debated for a quarter of century. It does not suit for small
ectotherms – this was explicitly pointed out in the paper – as they do not meet either of the two
premises (dependence of predators on herbivorous prey, large daily energy
needs). Moreover, as recurrently pointed out by our critics, treating entire
guilds as if they were single populations is a strong, simplifying
assumption. For me, however, to simplify
is the soul of theoretical analyses. Formal models are attempts to identify the
dominating processes in nature and, thus, to capture the essence of the
enormously complex biotic community in a simple and tractable presentation. The
validity of a model is found by testing how well it manages to predict unknown
patterns and outcomes of experiments in the systems it attempts to represent,
not by arguing about details of assumptions.
Indeed, validity is unlikely if the underlying assumptions are taken
from thin air. In the case of EEH, they are not. For instance, my assumption
that plants act as a homogenous mass derives from the observation that in the
Other theoretical projects and future
perspectives
Except for some early papers, whose primary scope was to show that I
could tackle issues that were regarded as “hot” in those days (e.g. the Oksanen et al.
1979 paper on interference competition in bird communities and the Oksanen 1981 note on
reproductive strategies), my other theoretical contributions have been related
to trophic interactions along terrestrial productivity gradients, dealing with the kinetics of
consumer-resource interactions (Oksanen et al. 1992, Diehl et al. 1993)
complementing the EEH-framework with analyses on the impact of seasonality
(Oksanen 1990a, Oikos
57:14-24), plant strategies (Oksanen 1990b, pp.
445-473 in: D. Tilman and J. Grace, eds. Perspectives on plant
competition), impacts of spatial structure (T. Oksanen et al. Evol. Ecology 6:383-398, T. Oksanen et al. 1995),
evolutionary aspects of trophic dynamics (Oksanen 1988,
Oksanen. 1992, Evol. Ecology 6:15-33) and impacts of food and predation risk on the
reproductive strategies of small herbivores (Oksanen and Lundberg 1995, Evol.
Ecology 9:45-56). One question frequently asked by colleagues is why I have not
revised my approach more fundamentally. The reason is simple. My team has
during the past twenty years conducted a large set of experiments in the
mountain and tundra landscapes of northernmost Fennoscandia, and outcome of
these experiments has corroborated the predictions of EEH (follow the links to Joatkanjávri and
Iešjávri).
Old fashioned or not, but I have learned that before fixing anything, you should
check whether the thing is broken.
My recent theoretical contributions focus on population cycles. In our
empirical work, we accumulated a long-term record of rodent numbers at the
timberline and in high country where the lemming outbreaks build up. The
difference in fluctuation patterns was striking. In collaboration with Peter Turchin we checked how time trajectories for predators and
prey in limit cycle dynamics were supposed to look – and came to the conclusion
that the time trajectories of lemmings in the high country bore the “predator’s
sign”, (Turchin et al. 2000, Ekerholm
et al. 2001), indicating that at the willow
scrubland limit, the predation-driven boreal cycles give way to entirely
different kinds of cycles generated by herbivore-plant interactions (as indeed
predicted by EEH). Another interesting
issue is the relationship between the boreal cycles and the low amplitude
cycles, found in some temperate areas. The initiative to these studies was
taken by Tarja Oksanen, who had been bothered about
the assumption that the functional response of generalist predators would be
automatically stabilizing. This assumption did not emerge from the data cited
to support it, nor was it consistent with the theory of optimal foraging. We
thus started to reanalyze the issue (T. Oksanen et al. 2001)
and came to the conclusion that even generalist predators could drive
population cycles, albeit with lower amplitudes and much higher population
minima than found in specialist driven cycles. This issue was recently
re-actualized in the context of the debate concerning
vole dynamics in north English woodland (Korpimäki et al.
2003, Korpimäki
et al 2005).
Some of my recent contributions have dealt with methodological issues.
This “interest“ has been dictated by concrete problems. When I worked on
geographical patterns in the intensity of plant-plant competition, I realized
that the most widespread index used in this context had serious flaws that had
to be fixed. Otherwise, the patterns in index values would not faithfully
reproduce patterns in nature (see Oksanen et al. 2006).
Another, more fundamental methodological challenge is how to deal with
situations, where replication is not possible (at least in the sense that the
predictions of interest refer to contrasts created by large scale gradients of
primary productivity, which tend to be directional). While Hurlbert
made an important contribution against the sloppy use of statistics in ecology
during the 1970’s and early 1980’s, the total ban imposed by him on the use of
statistics (even in the form of presenting standard errors in graphs) in the
context of non-replicated experiments was, to my judgment, counterproductive.
Sometimes, the salient predictions concern processes (or contrasts) in large
spatial scales, making complete replication infeasible. In those cases, the
message of the experiment hinges on the a
priori likelihood of obtaining quantitatively corresponding contrasts (or
divergences of time trajectories, coinciding with the application of the
treatment) for reasons independent of the conjecture to be tested. This should
be admitted and alternative explanations should be discussed, but if this is
done, it is difficult to understand what is wrong with presenting explicit or
implicit statistics, demonstrating the reliability of estimates of population
means. In unreplicated experiments, the magnitude of the contrasts is crucial.
Thus reliable documentation of population means is then extremely important.
(The a priori probability of one
statistical population having larger mean than the other is 0.5; the a priori probability for a tenfold
difference is much smaller; for further viewpoints, see Oksanen 2001, Oksanen 2004).
My current theoretical work focuses on the connections between trophic
dynamics and biodiversity. The issue has turned out to be much more
multi-faceted than I initially thought, with all kinds of non-obvious feed back
relationships between evolution and trophic dynamics. The theoretical work is
connected to the empirical research on the dynamics of rare arctic-alpine
plants, carried out in the new Raisduottar site.
The above summary hopefully puts my theoretical and empirical projects
in the right context. I have indeed more wrinkles on my face than the little
boy shown in the first picture, who became enthusiastic about the life in the
open expanses at high latitudes and altitudes. Inside, however, I am still the
same person. What counts for me is to untangle mysteries of arctic and alpine
systems. In this process, I use whatever approach is needed. There is no doubt
that my talent lies primarily in theory. Math was always my favorite subject,
while practical work has been hard to learn. But that is irrelevant. A
carpenter cannot use just his favorite tools when building a cabin. He must
focus on the cabin, not on the tools, and use each tool when needed. Moreover,
while math is my favorite technique, and I do not share Goethe’s view “Grau ist alles Theorie,
nur grün des Lebens goldner Baum” , nature still beats theory 10-1 as my source of
inspiration. Those, who have ever seen the arctic spring and summer, with
marvelous flowers and bluethroats singing in midnight sun, know why. Those who
have not should take their chance.