NICHE MODELING AND PREDICTIONS OF ALGAL BLOOMS IN AQUATIC ECOSYSTEMS

Background. One of the more elusive phenomena to predict in aquatic ecosystems is the blooming of planktonic algae (Smayda 1997); this aspect is interesting given that defining the distribution of a population in time and space is one of the most basic measurements one can make in biology and ecology (see Ricklefs 1990). If so, then why would this be the case? In part, algal blooms appear to be mediated by a series of complex (physical, chemical, biological) factors that individually are a challenge to understand, let alone in concert (e.g., particle transport mechanisms in freshwater, see below). Moreover, there is some urgency in developing predictive capabilities here, given that a number of algal species that form blooms are also capable of producing secondary metabolites toxic to vertebrates (Carmichael 1986, Plumley 1997). As a result, several large-scale research programs have been put into motion, dedicated to predicting the distribution (e.g., Tester and Steidinger 1997) and ecophysiology (e.g., Sellner 1997) of algal blooms and their consequences for the health of fish and vertebrates including humans (see Anderson and Garrison 1997). While significant strides have been made to characterize and quantify algal blooms, there still exist considerable gaps in the collective ability to predict the occurrence of blooms with any accuracy (see Millie et al. 1999). Bloom events may be analogous to weather predictions, which become markedly unreliable beyond envelopes of antecedent conditions that diverge from strict spatiotemporal scales that were essential in developing the original predictions (see Scofield et al. 1999). As such, blooms may be so context specific that generalization may not be possible, and furthermore, the exact physiological and ecological basis for toxicity is still largely unknown (Paerl and Millie 1996). With this in mind, perhaps there is something to be learned from other areas of ecology where researchers have struggled with similar challenges, but the object of study may be more tractable. Grinnell’s (1917) sentinel ornithological range analysis for the California thrasher (Toxostoma redivivum) offered important insights into the geospatial conditions that delimit species distribution patterns, as well as the role of using proxies in an attempt to describe niche space. Later, Hutchinson’s (1957) notion regarding the multidimensionality of species niche hyperspace was groundbreaking at the time it was proposed (Hutchinson 1980), and perhaps more important, this idea is still relevant today (see Slack 2011). So, it is generally accepted that the environment is a complex mosaic of variables that act in some combination to define the distribution patterns of species across a variety of spatiotemporal scales (Ricklefs and Miller 2000). That said, niche space is difficult to define in any practical sense, and until recently, most attempts to parameterize this entity have been piecemeal at best (see Weins 2004). The significance of the Millie et al. (2011 [this issue]) contribution lies in their use of recent advancements in computer and statistical technologies to evaluate complex data streams for pattern recognition and information synthesis regarding environmental-biotic interactions (e.g., termed ‘‘ecological informatics,’’ Recknagel 2003). Their analysis revealed an interesting and important set of community dynamics, whereby a transition between seasonal diatom blooms is antecedent to the occurrence of Microcystis (cyanobacterial) blooms in the nearshore region of Saginaw Bay; the Microcystis bloom has the potential to impart negative effects (toxicity, lower food quality) on the nearshore habitat in Lake Huron. This work dovetails with previous applications of niche modeling, where transfer functions have been derived to estimate niche space for diatom populations (Oksanen et al. 1988, Ponadera and Potapova 2007). Relevance of Millie et al. (2011): conceptual, computational, and practical strengths. Earlier applications of niche theory were used to evaluate resource utilization patterns among species and the likelihood of competition among sympatric populations (Lack 1940, 1947). The relevance of this early research, and accuracy of subsequent predictions, was largely borne out in later studies (e.g., Cody and Diamond 1975, Grant and Schluter 1982). More recently, the emerging subdiscipline of ecological informatics has J. Phycol. 47, 709–713 (2011) 2011 Phycological Society of America DOI: 10.1111/j.1529-8817.2011.01042.x