In order to
provide prior knowledge for the machine learning algorithms,
we need to know what causes exotic plants to exhibit invasive
behavior to the degree that they might
be declared noxious. Therefore, we must identify a number of factors along with the list of declared noxious species. These factors will provide further insight to the learning model about invasive species. Although there is a widespread pessimism regarding the prospect of predicting which species are likely to become successful invaders, several authors have conducted studies to outline the general theory of plant invasiveness.
Rejmanec (1996) has summarized a number of factors causing invasiveness as a result of his discriminant analysis:
1.Geographic ranges—Species’ native geographic ranges are significantly correlated to invasiveness. If a species has a very large latitudinal range in its continent of origin, it is likely that it will have tolerance to a wide variety of climates and habitats, thus increasing its chances to thrive in newly found habitats.
2. Congneric native
species (native species within the same genus)—The number of native species from the same genus already in place. Species within a genus share many common traits. If an exotic plant finds a new habitat and there are already a good number of species from the same genus, this new exotic will have to compete with those. On the other hand, if there are very few or no similar species, the new exotic plant will have little competition. As summarized by Rejmanec, Charles Darwin originated this theory in his On the Origin of Species (1859). Rejmanec (1996) points out that the most successful and most influential invasive plant species seem to belong to non-native genera.
3. Seed characteristics and short juvenile period—Using a number of pine species, Rejmanec concluded that seed characteristics as well as juvenile period are important elements in predicting invasiveness. In particular, he noted that a short juvenile period and short interval between large seed crops mean early and consistent reproduction. His analysis reports that small seed mass, along with a large number of seeds and short period between large seed crops are responsible for invasiveness of many woody seed plants.
4. Genome size (nuclear DNA content)—Low nuclear DNA content seems to be a result of selection for short minimum generation time. More specifically, Rejmanec sees a direct correlation between low nuclear DNA content and small seed mass and short juvenile period.
and Hamilton (1997) have produced a framework for identifying
potential invaders among woody plants introduced into North
America using discriminant analysis and decision trees. Like
Rejmanec, they also conclude that prediction could be obtained
by using biological and geographic attributes. In addition to
those, Reichard points out that the most significant attribute
in her analysis was whether or not a plant is known to invade
elsewhere (i.e. “Invades elsewhere: Yes/No”). In agreement
with Rejmanec’s findings, she indicates that latitudinal
ranges of a species can have high predictive value—a fact
verified by other studies (Scott & Panetta 1993; Reichard
1994; Rejmanec 1995).