Field gentian – when it’s good to be eaten

We tend to think of plants as victims – after all any interested herbivore can simply walk, fly or crawl over to its favorite plant, and begin munching. But not so fast! In reality, plants have a variety of ways they can make life difficult for potential herbivores. Plants can escape herbivores by simply growing in places that are not easily accessible (such as in cracks, or high enough to be out of a herbivore’s reach) or by growing at a time of year when herbivores are away from the plant’s habitat. Plants also use mechanical defenses such as thorns or a diverse array of chemical defenses to thwart overzealous herbivores. A third approach – tolerance – can take many forms. For example, following attack by a herbivore some plants can increase photosynthetic rates or reduce the time until seed production . Tommy Lennartsson and his colleagues were interested in a particular form of tolerance that ecologists call overcompensation, in which damaged plants produce more seeds than undamaged plants.


Herbivores in action. Notice the difference in vegetation height inside and outside the pasture. Credit: Tommy Lennartsson.

Overcompensation is an evolutionary puzzle, because undisturbed plants produce fewer offspring than partially eaten plants. That outcome seems to fly in the face of the scientific principle that natural selection favors individuals with traits that promote reproductive success. Lennartsson and his colleagues investigated this evolutionary puzzle by comparing two subspecies of the herbaceous field gentian Gentianella campestris. The first subspecies, Gentianella campestris campestris (which we’ll just call campestris), has relatively unbranched shoot architecture when intact, growing to about 20 cm tall, but produces multiple fruiting branches when the dominant apical meristem is eaten. The second subspecies, Gentianella campestris islandica (which we’ll call islandica), is much shorter (about 5-10 cm tall), and always has a multi-branched architecture.


Two subspecies of field gentian – campestris (left) and islandica (right).

Environmental conditions and soils can vary dramatically, even on a small spatial scale. The field site was a gently-sloped grassland in Sweden that had coarser, dryer soil on the ridge, and finer, wetter and richer soil in the valley. This created a productivity gradient, with taller vegetation in the valley. The average  height of all the vegetation was 15 cm in the high-productivity valley, 10 cm on the medium-productivity slope and 5 cm on the low-productivity ridge.

The researchers used this natural variation to set up an experiment that would allow them to explore hypotheses about why an undisturbed campestris is less successful than one that is partially-eaten. One hypothesis (the overcompensation hypothesis) is that campestris restrains branching to conserve resources, so that when it is grazed it has plenty of resources in reserve to be used for regrowth and the production of prolific branches, flowers and seeds. Limited branching and limited seed production of ungrazed campestris are simply a cost of tolerance, while overcompensation after damage maximizes reproductive success. A second hypothesis (the competition hypothesis) is that restrained branching allows the plant to grow tall, so it can compete better in ungrazed pastures than can the much shorter islandica. These two hypotheses are not mutually exclusive.

To test these two hypotheses, the researchers set up 2 X 2 meter experimental plots in the valley (18 plots), slope (12 plots) and the ridge (6 plots). They planted 2000 seeds per subspecies in each plot, which ultimately yielded about 20 plants of each subspecies per plot. Of course there were many other neighboring plant species in these plots. In the high productivity plots (valley), the neighboring plants in six plots were clipped to a height of 12 cm, six plots to 8 cm and six plots to 4 cm. In the medium productivity plots (which naturally only grew to 10 cm), the researchers cut neighboring plants to 8 cm in 6 plots and 4 cm in six plots. Finally, in the low productivity plots, the researchers cut neighboring plants to 4 cm in all six plots. In mid July, half of the gentian plants in each plot were clipped to the same height as the surrounding vegetation, while the remainder were not clipped.


Experimental plots from the valley (left), slope (middle) and ridge (right).  Black squares represent plots where neighboring plants were clipped to 12 cm, grey squares to 8 cm, and clear squares to 4 cm. Squares with slashes through them (left)  represent plots that were used for a different purpose.

The beauty of this experimental design, is that by counting seeds, the researchers could assess the reproductive success of both subspecies under conditions of high competition (when surrounded by tall neighbors) and low competition (when surrounded by shorter neighbors). At the same time, clipping the two subspecies allowed the researchers to simulate grazing in these different competitive environments. Lennartsson and his colleagues found that unclipped islandica did better than unclipped campestris when surrounded by short or medium height neighbors, but that islandica success plummeted when the neighbors were very tall (see the left graph below). Campestris reproductive success also dropped when surrounded by tall competitors, but not as much as did islandica, so that campestris produced twice as many seeds than islandica in the high competition environment (also the left graph).

When plants were clipped to simulate grazing, campestris outperformed islandica in all three competitive environments. Campestris actually produced more seeds when it was clipped than when it was not clipped in the low and medium competition environments. Thus campestris overcompensated for grazing under conditions of low and moderate competition (see the right graph below).


Mean (+ standard error) seed production for unclipped (left graph) and clipped (right graph) field gentian subspecies in relation to surrounding vegetation height.  Sample sizes are in bars.

The researchers collected data on growth rates, development, survival probabilities and reproductive success for both species under conditions of being clipped or unclipped at different levels of competition. They then used these data to create a population growth model in relation to the percentage of grazing (damage risk) at different levels of productivity. In these graphs, a stochastic growth rate of 1.0 (on the y-axis) indicates that the population is stable, above 1.0 indicates it will increase and below 1.0 indicates a declining population.


Population growth rate of both subspecies in relation to damage risk at different levels of productivity.  These models predict that the population will increase at growth rates above the dotted line (growth rate = 1.0) and decline below the dotted line.

This model shows that in high productivity environments, campestris always does better than islandica (top graph). However, the model predicts that islandica will decline at any damage level (note in the top graph that all islandica damage values yield a growth rate below 1.0), while campestris will also decline except for very high damage risks. In medium and low productivity populations (middle and bottom graphs), islandica does better than campestris when damage risk is low, but the reverse is true at high damage risk.

So how do these results relate to the two hypotheses for why an undisturbed campestris is less successful than one that is partially-eaten. Campestris overcompensated for damage by producing more seeds and having positive population growth under most levels of productivity. In contrast, islandica undercompensated when damaged, but produced more seeds than campestris when ungrazed, except for in the high productivity environment. These differences in responses support the hypothesis that restrained branching is favored by natural selection in environments where damage from grazing is common (the overcompensation hypothesis). But, the superior performance by campestris in productive ungrazed environments supports the competition hypothesis.

Can we generalize these findings to other plants? Lennartsson and his colleagues point out that many short-lived grassland plants can’t grow tall enough to be effective competitors for light. These plants are thus restricted to environments where the surrounding plants are not very tall. Two factors commonly create conditions where there are short neighboring plants: grazing and unproductive (low nutrient) soils. When grazing is widespread, tolerance mechanisms such as overcompensation are favored by natural selection. When soils are unproductive, unrestrained branching is favored. Therefore, Gentianella campestris provides us with a natural experiment for testing hypotheses about how natural selection acts on plants to promote their reproductive success in a variable environment.

note: the paper that describes this research is from the journal Ecology. The reference is Lennartsson, T., Ramula, S. and Tuomi, J. (2018), Growing competitive or tolerant? Significance of apical dominance in the overcompensating herb Gentianella campestris. Ecology, 99: 259–269. doi:10.1002/ecy.2101. Thanks to the Ecological Society of America for allowing me to use figures from the paper. Copyright © 2018 by the Ecological Society of America. All rights reserved.


Changing climate promotes prolific plants and satiated consumers

Plants in Sweden can have a difficult life, but climate change has provided a more benign environment for some of them, including the white swallow-wort, Vincetoxicum hirundinaria. This perennial herb grows in patches in sun-exposed rocky areas, in forests located below cliffs, and along the edges of wooded areas. The plant forms clumps that are heavily laden with flowers in June and July, and creates pod-like fruits in July and August


Vincetoxicum hirundinaria growing in rocky outcrop (top photo). Vincetoxicum pods releasing their wind-dispersed seeds (bottom photo).

Christer Solbreck has had a lifelong interest in insect populations, and he has been following the insects that eat Vincetoxicum’s seeds for the past 40 years. As he described to me, surprisingly few population ecologists actually measure the amount of food available to insects. I should add that very few people have the resilience to study the same population of insects for 40 years, either. And interestingly, though this paper discusses the effect of a changing climate on seed production and seed predation, it was not Solbreck’s intent to consider climate change as a variable when he began, as climate change was not a concern of most scientists in the 1970s.

But climate change has happened in southeastern Sweden (and elsewhere), and has affected ecosystems in many different ways. Ecologists can quantify climate change by describing its effect on the vegetation period, or growing season (days above 5°C), which has increased by about 20 days since the mid 1990s.


Length of growing season (vegetation period) in southern Sweden.

During the same time period the abundance of Vincetoxicum has increased sharply.


Vincetoxicum abundance, measured as area of the research site covered, during the study.

You will note that “Vincetoxicum” has the word “toxic” in its midst; the seeds are toxic to most consumers, and are important food sources for only two insect species. Euphranta connexa females lay eggs in developing fruits of the host plant, with the emerging larva boring through the seeds and killing most of them. Lygaeus equestris is an all-purpose seed predator; both larvae and adults suck on flowers, on developing seeds within the fruits, and on dry seeds they find on the ground up to a year later.


Euphranta connexa female lays eggs in an immature seed pod.


Lygaeus equestra larva feeds on a fallen seed.

Solbreck teamed up with biostatistician Jonas Knape to analyze his data. From the beginning of the study, Solbreck suspected that annual variation in weather – particularly rainfall – might influence Vincetoxicum seed production, and consequent population growth of the two insect species. They discovered something quite unexpected; the dynamics of seed production shifted dramatically in the second half of the study, alternating annually from very high to very low production over that period. This dynamic shift coincides with the extension of the growth season as a result of climate change.


Seed pod abundance by year.

The researchers argue that there is a non-linear negative feedback relationship of the previous year’s seed production on the current year’s seed production. Negative feedback occurs when an increase in one factor or event causes a subsequent decrease in that same factor or event. In this case, an increase in seed production uses up plant resources, leading to a decrease in seed production the following year. But the effect is non-linear, and does not come into play unless Vincetoxicum produces a huge number of seeds, as shown by the graph below,


Seed production in the current year in relation to seed production in the previous year. Note that both axes are logarithmic. The curve represents the expected seed pod density generated by the statistical model, with the shaded area representing the 95% credible intervals. Open circles are data for 1977-1996, while closed circles are data for 1997-2016.

The researchers also found that high rainfall in June and July increased seed production.

So how do these wild fluctuations in seed production affect insects and the plant itself? One important finding is that in high seed production years, the proportion of seeds attacked by insects plummets because the sheer number of seeds overwhelms the seed-eating abilities of the insect consumers. Ecologists describe this phenomenon as predator satiation.


Seed predation rates in relation to seed pod density.  Note that both axes are logarithmic. The curve represents the expected predation rate generated by the statistical model, with the shaded area representing the 95% credible intervals. Points are E. connexa predation rates while triangles are combined predation by both insect species.

As a result of predator satiation, there were, on average, seven times as many healthy (unattacked) seed pods in 1997-2016 than there were in 1977-1996. Presumably, this increased number of healthy seeds translates to an increase in new plants becoming established in the area. An important takehome message is that the entire dynamics of an ecosystem can change as a result of changes to the environment, in this case, climate change. More long-term studies are needed to evaluate how common these shifting dynamics are likely to become in the novel environmental conditions we humans are creating.

note: the paper that describes this research is from the journal Ecology. The reference is Solbreck, Christer and Knape, Jonas (2017), Seed production and predation in a changing climate: new roles for resource and seed predator feedback?. Ecology, 98: 2301–2311. doi:10.1002/ecy.1941. Thanks to the Ecological Society of America for allowing me to use figures from the paper. Copyright © 2017 by the Ecological Society of America. All rights reserved.

Light levels limit lake phytoplankton response to fertilization

One might naively think that because we humans are land-dwelling creatures, our impact on aquatic ecosystems might be relatively minor. Unfortunately, this assumption is incorrect, as human activities are changing aquatic environments in profound ways that influence how aquatic species survive and interact. Global warming is increasing lake and river temperatures, uncontrolled development is causing some streams to run dry and others to flood, and agricultural practices are adding nutrients to many lakes and streams. Because these human impacts occur simultaneously, it is difficult to evaluate how each factor contributes to the observed changes in species relations.

In northern Sweden, lakes vary naturally in the amount of dissolved organic carbon (DOC) they contain. DOC comes from runoff of decaying plant matter, so lakes surrounded by substantial vegetation, or that experience a great deal of water input (runoff) from the surrounding area, would have higher DOC than other lakes. DOC is potentially very important to lakes, because DOC tends to discolor a lake, which reduces light penetration and slows down photosynthesis. On the positive side, carbon may bond to other molecules such as phosphorus and nitrogen, which are important nutrients that may be in short supply in these relatively infertile lakes.   Anne Deininger and her colleagues focused their studies on two factors: DOC and nitrogen. Most lakes have too much nitrogen, a result of excessive use of nitrogen fertilizers that run off into lakes, so these relatively low-nitrogen lakes provided the researchers with a unique opportunity to see how these two factors, DOC and nitrogen, interacted in a natural ecosystem.

Screen Shot 2017-06-14 at 11.35.22 AM

Low DOC control lake. Credit: M. Klaus

The researchers selected six lakes that varied naturally in DOC levels: two low (~7 mg DOC/liter), two medium (~11 mg DOC/liter), and two high (~20 mg DOC/liter). In 2011 they measured everything possible about each lake: abundance of all of the life forms, DOC, temperature, light levels, nutrients and photosynthetic rates. In 2012 and 2013, they supplemented one of each pair of lakes with nitrogen compounds every one to two weeks. The added nitrogen was equivalent to the higher nitrogen inputs that are experienced by lakes in southern Sweden. And, as you might expect, the researchers continued measuring all factors of interest in both the experimental (fertilized) and control (unfertilized) lakes throughout the year – at least until the lakes froze over.


Anne Deininger (in orange) and Sonja Prideaux collect samples from a lake. Credit: M. Deininger.

Deininger and her colleagues were most interested in differences in the abundance of phytoplankton – small free-floating photosynthetic organisms, because these are the primary producers – the organisms that produce the chemical energy (via photosynthesis) that enters food webs. There are many different types or groups of these phytoplankton; some are flagellated, with hair-like processes that allow them to navigate in the water column. Some are exclusively autotrophs, producing their own energy from photosynthesis, some are primarily hetrotrophic, eating other organisms or the remains of dead organisms, while others are mixotrophs, using both strategies to produce energy. Cyanobacteria are photosynthetic bacteria, while picophytoplankton are phytoplankton of unusually small size.

Screen Shot 2017-06-14 at 11.36.12 AM

Flagellated phytoplankton (Cryptomonas). Illustration by Anne Deininger.

Many important findings are summarized in the graph below. “B” represents the year before fertilization (2011), while “A1” is 2012 (after fertilization – 1st year) and “A2” is 2013 (after fertilization – 2nd year). Remember only the N-lakes were fertilized; the control lakes were simply monitored all three years. One finding is that in 2011, the high DOC lakes had the lowest phytoplankton abundance.  A second is that the low and medium DOC lakes had both flagellated and non-flagellated phytoplankton, while the high DOC lakes were dominated by flagellated phytoplankton.

Moving to the years after fertilization (A1 and A2), you can see that nitrogen fertilization increased phytoplankton abundance, but more so for the low-DOC lake. However, fertilization had little impact on the types of phytoplankton found in each lake; rather it simply increased the abundance of already existing groups.


Mean biomass of major phytoplankton groups in relation to DOC.  Recall that B refers to 2011 (the year before fertilization), while A1 and A2 refer to the two years after fertilization (2012, 2013).

The data can be organized so we can get a better view of what is happening quantitatively. Fertilization increases phytoplankton biomass, but much more for lakes with low DOC levels. In addition DOC appears to decrease phytoplankton abundance.


Deininger and her colleagues conclude that in these northern lakes, phytoplankton production is nutrient-limited at low DOC levels, but becomes limited by light availability in more murky waters. So adding nitrogen increases phytoplankton abundance to a greater extent in low DOC lakes. High DOC lakes have more flagellated autotrophs, as these species can swim to the top of the water column where there is more light for photosynthesis. As needed, flagellated phytoplankton can move lower in the water column where nutrients are more abundant.

The researchers emphasize that the nitrogen experiments were only conducted for two years. They don’t know if, for example, the types of species would change if fertilization continued for more than two years. They also don’t know if after 2013, the communities reverted to their pre-fertilization state, or if biomasses remained higher when nitrogen fertilization stopped. These types of questions are important to pursue because we humans are making drastic changes to most of our aquatic systems in a very uncontrolled manner. We need to understand the effects of these changes to the aquatic environment, and also how we can reverse the effects should they prove to be highly detrimental.

note: the paper that describes this research is from the journal Ecology. The reference is Deininger, A., Faithfull, C. L., & Bergström, A. K. (2017). Phytoplankton response to whole lake inorganic N fertilization along a gradient in dissolved organic carbon. Ecology98(4), 982-994. Thanks to the Ecological Society of America for allowing me to use figures from the paper. Copyright © 2017 by the Ecological Society of America. All rights reserved.