Competition between Two Weeds

Joy Bergelson. American Scientist. Volume 84, Issue 6. Nov/Dec 1996.

Every gardener knows that plants compete with each other-especially if one of the plants is a weed. Although gardeners help their plants win competitive interactions by removing or poisoning weeds, plants in nature must fight on their own-finding ways to defeat rival plants. The question is: What causes competing plants to win or lose?

Competition between plants represents one of the best-documented interactions between biological species. Hundreds of examples show that plant species affect one another through the use of shared resources, including light, nutrients and water. A tomato seedling in your garden, for instance, may grow poorly in the shade cast by a mature oak tree. Another tomato plant may survive better in the shade, because some individual plants have different competitive abilities. The outcome of competition between plants may also depend on the composition of the local community and debris or chemicals left by plants from earlier generations.

Although botanists have assembled a virtual catalogue of examples of competition, we know very little about the mechanisms that govern differences in competitive ability. I shall describe a series of experiments that reveal an intricate-and somewhat surprising-web of competitive interactions between two particular plant species.

Garden Experiments

One of the best-studied examples of competitive dynamics involves two weedy annuals: common groundsel and annual bluegrass. These species thrive throughout Eurasia and the United States, where bluegrass grows more abundantly and persists admirably when faced with neighboring groundsel. About a decade ago, I started a series of experiments to unravel the factors that drive the interactions between these two species and to test the populationlevel repercussions of their competition.

I began by gathering 20 groundsel plants from the Seattle area. I collected seeds from each of those plants and established a “common garden” experiment to test the ability of the seeds to grow and reproduce with and without intense competition from bluegrass. In this experiment, I planted seeds from each group of siblings, or sibship, in a field plot. I planted some seeds alone, and I encircled others with 12 bluegrass seeds. The results from that experiment showed that some of the groundsel plants fared better than others when faced with competition from bluegrass. For instance, some of them produced twice as many flowers as others. Such differences in performance were not evident among the groundsel plants grown alone.

What caused some groundsel plants to perform better than others in the presence of competitors? A plant’s genotype could affect its competitive ability. Nevertheless, my initial experiment did not prove a connection between genetics and competitiveness, because the seeds used in that experiment also came from plants that had experienced different environmental conditions. A number of studies indicate that a mother plant’s environmental conditions can influence the characteristics of progeny-a phenomenon known as the maternal effect. Luckily, maternal effects can be controlled for rather easily, at least in this case. Groundsel produces seeds autogamously-meaning that a flower gets pollinated by its own pollen-and plants collected in the field can be propagated in a common environment for subsequent experiments. So I grew sibships from each plant for three generations in a greenhouse, and then repeated my original experiment. Again, different groundsel plants exhibited large differences in their ability to endure competition, but no differences appeared without competitors. In addition, if a field-collected mother plant performed well against competitors, so did her offspring. These results confirmed that there is a genetic basis for the varied tolerance to competition.

Early Winners

What features make a plant more competitive? That question can be addressed through a closer look at competitively superior and inferior groundsel genotypes. During my experiments, I tracked the date of germination and the height of each groundsel plant, which provided data that might explain differences in competitive ability.

Using a technique called path analysis, I assessed the relative importance of differences in early growth rates, late growth rates and emergence dates. Path analysis unravels the effects of several correlated characteristics, such as growth rates and date of emergence, on a plant’s subsequent performance. That analysis suggested strongly that competitively superior genotypes gain their advantage by emerging early In other words, a rapidly emerging groundsel plant is a competitive one. Moreover, the emergence date of the plants in this experiments explained more than 70 percent of the variation in their success.

That result led to another experiment: manipulating relative emergence dates by planting seeds at different times. I planted slow-germinating genotypes before fast-germinating ones, so that the seedlings would emerge simultaneously. The resulting plants possessed comparable competitive ability-confirming my hypothesis that competitive superiority comes from getting a head start on germination. Other plant systems operate similarly; small differences in emergence dates translate into large differences in performance. The effects can be dramatic. For example, Thomas Miller of Florida State University reported that a three-day head start in emergence results in more than a 1,900-fold increase in adult performance in common lambsquarters.

If emergence date determines a seedling’s success, we might expect to see the evolution of traits that enable seeds to modulate germination in response to competitors. In fact, seeds from some plant species reduce their probability of germination when adult plants are already growing in the area, evidently avoiding competition with established plants. Seeds also appear responsive to the presence of other seeds. In both groundsel and bluegrass, my colleagues and I have demonstrated a negative relation between seed densities and the probability that each seed will germinate, presumably a tactic for preventing high levels of competition among seedlings, which often leads to extensive mortality Moreover, seed-seed communication appears even more complex than a simple density-dependent germination. A series of greenhouse experiments showed that groundsel and bluegrass seeds accelerate their rates of germination when they are grown in soil that had previously contained germinating seeds. Apparently, the previously germinating seeds excrete chemicals, which have not yet been isolated, that accelerate the emergence of later seeds.

Neighborhoods and Gaps

In the groundsel-bluegrass system, competition portrays a race in which small differences in emergence produce large differences in performance, and seeds respond vigorously to clues that indicate the presence of competitors.

Apparently, the threat of competition shapes the activity of these seeds. That raises broader questions: To what extent does competition influence the population dynamics of competitors, and to what extent does community context influence competitive interactions? Although biologists commonly assume that competition affects the dynamics of plant communities, we are just beginning to explore the interplay between individual performance, which responds to competition, and populationand community-level factors. Such research provides an important arena for testing our understanding of the forces that structure plant communities.

At this interface between individuals and communities of plants, spatial patterning receives a disproportionate share of attention. Two distinct theories have been developed that predict that spatial patterning can exert a strong influence on competitive interactions. Since plant populations exhibit very patchy arrangements, as opposed to random distributions, the question arises as to whether patterning drives the dynamics of plant populations in nature.

Stephen Pacala of Princeton University and John Silander of the University of Connecticut champion the neighborhood-competition theory. That theory assumes that plant growth and reproduction depend on the density of nearby competitors and that competitors beyond a designated distance exert no impact on a particular plant. Combining that assumption with an elaborate mathematical theory suggests that patchiness in a distribution of competitors can profoundly alter competition at the population level. Intuitively, this theory can be understood by recognizing that patchiness leads to considerable variation in the amount of competition experienced by any individual plant. Some plants will experience lots of competition and experience little reproduction; other plants will be virtually free from competition and contribute a disproportionately large share to the next generation. In this way, the neighborhood-competition theory suggests that the spatial pattern of plants influences competition through interactions between contemporaneous plants.

An alternative theory for patchiness, gap colonization, considers the competitive impact of one generation on another. Traditionally, this theory has been applied to the dynamics of forests, where the germination and growth of seedlings depends on gaps left after trees fall. Nevertheless, it might be equally relevant to the dynamics of grassland plants. In nearly all cases, established vegetation can overpower seedlings. In fact, seedling emergence and survival can be totally inhibited by the presence of adult plants, or even by the litter left from previous generations. This type of inhibition can arise from several factors: chemicals, shading or structural interference produced by the established plants. In some cases, these factors may restrict seedling survival to areas that lack established vegetation. Moreover, the gaps in a patchy spatial pattern promote the persistence of competitively inferior plants. Gap-colonization theory, then, suggests that an area’s spatial pattern will influence competitive interactions because previous generations affect the survival of present seedlings.

Competition between common groundsel and annual bluegrass also depends on spatial patterns. Both of these species produce two generations each year-one in the spring and another in the fall. This rather unusual life cycle proves tremendously convenient for testing the neighborhood-competition and gap-colonization theories. In each fall generation, newly emerging seedlings compete with each other, which provides an opportunity for neighborhood competition, and they also interact with remnants from the spring generation, which could lead to gap colonization.

Litter Blockade

To explore the role of patchiness on competition between groundsel and bluegrass, I established artificial plots in the spring in which I created either a random or a patchy distribution of bluegrass. In the same plots, I also planted groundsel, but it was always distributed randomly at one-quarter the density of bluegrass. That relative abundance resembles natural communities, where groundsel is typically less abundant than bluegrass. I let the seedlings grow and compete, and then I counted the number of groundsel plants in the fall generation. The spatial distribution played a dramatic role in the growth rates of the groundsel: The fall generation contained nearly four times more groundsel plants when the bluegrass distribution was patchy rather than random. That experiment represented the first experimental manipulation of plant spatial patterns to assess competitive interactions, and it showed unequivocally that spatial patterning can have an enormous impact.

These results did not arise from neighborhood competition. Regardless of the distribution of the bluegrass, the spring generation of groundsel produced essentially the same number of seeds. That is, competition within the spring generation did not generate the differences that appeared in the fall. Pacala and Silander obtained similar results when exploring how patchiness affected the growth rates in a competitive system of velvetleaf and pigweed.

On the other hand, gaps did affect the groundsel’s growth. In the plots, a large fraction of all the surviving groundsel seedlings grew in areas of low bluegrass density. So patchy plots promoted the growth of groundsel populations by providing a greater number of gaps.

The difference in groundsel’s success in the spring and fall generations suggested that the bluegrass litterdead blades and roots from previous generations–could be a factor. To test that hypothesis, I repeated the above experiment with the patchy and randomly distributed bluegrass but removed the dead bluegrass from half of the plots and left the bluegrass litter intact in the other half. If litter drives the effect of spatial patterning on groundsel, then removing the litter should remove the effect, which is exactly what I found. With the litter intact, groundsel grew better where the bluegrass distribution was patchy; but with the litter removed, the groundsel grew about the same regardless of the distribution of bluegrass. This experiment indicated clearly that competition between generations, not within them, governs the dynamics of the groundsel-bluegrass system and explains the importance of the spatial pattern.

A series of greenhouse experiments revealed that the effect of bluegrass litter comes from the dead blades above ground. The presence of grass roots or chemicals that might have leached from the bluegrass did not affect the germination or survival of groundsel seedlings. Instead, litter inhibits groundsel seedlings, because emerging seedlings get trapped by the litter above them. The seedlings cannot penetrate the litter, which prevents them from capturing light or growing, and they die. This structural inhibition between bluegrass litter and groundsel seedlings provides the crucial competitive interaction. In addition, litter generates little trouble for the relatively slender morphology of a bluegrass seedling.

These investigations illustrate that the spatial pattern of bluegrass produces large effects on the success of the competitively inferior groundsel, and that the mechanism involves gap colonization, or interactions between generations. That conclusion has several additional implications. First, the interaction between groundsel and bluegrass includes a time lag-earlier generations affecting later ones. A variety of simple mathematical models illustrate that biological systems with time lags tend to have relatively more complex dynamics than systems without time lags. The second implication involves succession. As succession proceeds, a system’s litter accumulates, which can shift the competitive balance from litter-intolerant species to litter-tolerant ones. In that way, litter can qualitatively alter the outcome of competition. Again, models support such a conclusion, showing that bluegrass should dominate whenever litter accumulates, and that groundsel dominates if the litter decomposes quickly.

Modeling Trade-offs

These small-scale experiments showed that groundsel grows more successfully in a patchy plot. In the simplest terms, one might say that greater amounts of bare ground favor groundsel, because the plant requires such gaps for establishment. Then one might ask: Given a particular amount of bare ground, how does its spatial distribution influence the success of invading weeds? I approached that question in collaboration with Jonathan Newman of Southern Illinois University and Ernesto Floresroux, then of the University of Chicago. We performed experiments on a somewhat larger spatial scale-over a few meters-where we tried to determine how the dispersion of gaps influences how fast groundsel progresses through a field. These experiments reveal the community-level repercussions of between-generation competition.

We approached the effects of gap dispersion with a simple experiment. For each experimental plot in a field of ryegrass, we created six transects that were oriented like spokes on a wheel. On each transect, we created artificial gaps that covered one of three areas: 25, 225 or 900 square centimeters. To control the total amount of gap in a given transect, we created fewer large gaps than small ones. We distributed the gaps either uniformly or randomly, based on the distance between them. By analogy with the small-scale experiments described earlier, a large variance in the intergap distance corresponds to a patchy distribution, and equal intergap distances correspond to a uniform distribution. We introduced 12 invading groundsel plants in the center and then counted the number and position of all seedlings in two subsequent generations-hoping to determine whether the success of invasion depends on the spatial heterogeneity of the gaps.

One can assess a plant’s success of invasion in two different ways. The number of individuals that get established provides one index, and the distance between a parent and its offspring-the rate of spread-provides another. In our experiments, larger gaps increased the number of established groundsel seedlings, even though we controlled for overall gap area. In addition, the invading groundsel spread faster with large gaps. For example, large gaps produced nearly three times more distance between a parent and its offspring, as compared with small gaps. Moreover, the invading groundsel produced more established seedlings with patchy gaps than uniform ones, regardless of the size of the gaps. However, offspring traveled farther when the gaps were positioned uniformly, regardless of the size of the gaps.

We wondered if the way that a plant’s “shower” of seeds would fall on such gaps would lead to similar results. One can imagine that a plant produces a seed shadow, which depicts the proportion of seeds that fall relative to the distance from the plant. In our transect experiment, some seeds would fall in gaps and germinate, and others would fall in vegetation and not germinate. By knowing a plant’s seed shadow and a transect’s arrangement of gaps, one can predict the expected distance between parents and seeds that land in gaps. A simple mathematical model of this scenario produced results that resembled what we found in our experiments. In other words, how the seeds disperse and the strong competitive dominance of established grass over seedlings explains what we observed.

These results point to an interesting trade-off: An invading plant can progress faster in a field that contains a uniform distribution of gaps, but fewer seeds land successfully in uniform gaps. This tradeoff affects models of the persistence of competitively inferior species in patchy environments. In the past, such models suggested that a competitively inferior species can persist in a community by dispersing more effectively than its superior competitor, but models of that phenomenon ignore the spatial positioning of gaps. Our results, however, indicate that a competitively inferior species faces a more difficult challenge, because of the negative relationship between rates of dispersal and the probability that seeds land in gaps and establish successfully Dispersing seeds that travel a far distance, on average, require uniformly positioned gaps; but patchy gaps lead to more established seedlings. In other words, a plant can either widely disperse its offspring or produce lots of them, but it probably cannot do both. Future research should address how competitively inferior species persist in realistic, spatially heterogeneous environments.

My work with two common weedsgroundsel and bluegrass-shows that competition between these plants depends on many factors. Competition between generations-later groundsel seeds battling established bluegrass-is the primary factor that governs the dynamics of this system. Nevertheless, groundsel’s genotype determines largely when a seedling will emerge-a crucial factor in competitive success-and that suggests that contemporary plants must compete, as well. Moreover, the result of competition between groundsel and bluegrass also depends on the structure of the local environment, including the size and arrangement of gaps. In the future, ecologists hope to develop models and experimental systems that simultaneously examine how these factors contribute to plant competition.