Sierra Nevada Aquatic Research Laboratory

01 Nov, 04 AM - 30 Apr, 05 AM

Abstract A major question in ecology is to what extent competition affects the assembly of species into communities. Using recently developed mathematical tools, Dr. Roland Knapp and I have shown that species of benthic insects in lakes in Yosemite National Park tend to assort aggregately into guilds, and independently into orders. While our current results point to these processes dominating either colonization or the assembly of species pools, they cannot rule out competition. To rectify this problem, I here propose to catalog the species pools. Overview For seventy years ecologists have debated to what extent competition structures ecological communities. At one extreme, the Gleasonian viewpoint posits that species assemble randomly, assorting by chance and abiotic factors (Gleason, 1926). At the other extreme, putative assembly rules suggest that interspecific interactions are responsible for the assortment of species (Diamond, 1975). Among the interactions, competition is often cited as key (Gotelli and Graves, 1996; Diamond, 1975). Differentiating between the two viewpoints has been challenging. At small spatial scales, experiments can provide answers; for instance the work of Connell (1961). However, at the spatial scale of communities, experimentation is impractical and unethical; extirpating or introducing species is difficult and disruptive to ecosystems. Moreover, interspecific interactions may take hundred of years to have measurable effects (Connor and Simberloff, 1986). Null model analyses constitute the popular approach to compensate for the lack of experimental evidence. The analyses ask what pattern would have been observed in the absence of competition. If the observed pattern differs from the prediction, competition is inferred (Gotelli and Graves, 1996; Connor and Simberloff, 1979). Fundamental to all null model analyses is the species presence-absence matrix, in which each row represents a site while each column represents a species. If at site i species j was observed, then the i,jth entry is a one; otherwise it is a zero. To perform an analysis, one generates a distribution of presence-absence matrices assuming no competition, from which the probability of the observed matrix or one more extreme can be calculated. A sufficiently small probability leads to rejecting the null model in favor of competitive structuring. Null model analyses are hence a form of statistical hypothesis testing (Gotelli and Graves, 1996). Like all other statistical tests, null model analyses should be chosen based on their power and robustness (Conover, 1999; Bradley, 1968) Power refers to the test's sensitivity, with the most powerful tests rejecting false null hypotheses most frequently. Robustness refers to the distortion of significance levels when assumptions are violated, with robust tests only negligibly affected by violations (Larsen and Marx, 1986). Thus, null model analyses should be powerful and, if assumptions cannot be independently verified, robust. Unfortunately, calculations show that no null model analyses are both adequately powerful and robust. Power varies, with no one analysis performing best for all presence-absence matrices [Ladau, in preparation; incorrect results are presented in Gotelli (2000)]. As for robustness, all null model analyses rely on a set of assumptions, the probabilities of each colonist belonging to each species and occurring at each site (Gotelli, 2000; Gotelli and Gaves, 1996) . No analyses are robust against these assumptions, even if only modest violations are considered. Moreover, robustness is a function of the dimensions of the presence-absence matrix, marginal totals, nominal level of significance, and assumption violations under consideration (Ladau, in preparation). The analyses are thus prone to falsely detect both independent and non-independent assortment of species. To fix the robustness problem, Dr. Steven J. Schwager and I have derived a null distribution based solely on assumptions of independence. Clearly, species can assort non-independently during two stages: when they assemble into the regional pool of species, or when they colonize from that pool (Simberloff, 1970). Let an "immigrant" be any species that joins the pool and colonizes, and let the ith immigrant be denoted with the ith Roman numeral. Define a "group" as a set of ecologically similar species (Fox, 1987) . If competition inhibits similar species from joining the pool or colonizing, new immigrants will most likely belong to species-poor groups -- implying for instance that III will be most likely to belong to the same group as II when II is the only member of its group. Overall, competition will thus produce a low variance in the number of species per group [VSG; see Fox (1987)] . On the other hand, unevenly distributed resources and a lack of competition will produce the opposite pattern; new immigrants will tend to be from species-rich groups. In the absence of either process, immigrants will belong to old immigrants' groups independent of membership; for example, the event that III belongs to the same group as II will be independent of whether II shares its group with I. Importantly, such independence can be proven sufficient to specify a distribution for VSG (Ladau and Schwager, in preparation). From a practical standpoint, an observed distribution of VSG may be shifted to the right or left of the null, or it may be unmoved. A right-shift implicates aggregation during colonization or pool assembly, because only this process can account for this shift. However, with strong aggregation in only one stage, competition or independent assortment could have occurred during the other stage; implying that a right-shift is only conclusive regarding aggregation. Likewise, a left shift implicates competition during colonization or pool assembly, but leads to no conclusion about aggregation and independent assortment. A distribution that matches the null is consistent with either complete independence or an exact balance between aggregation and competition. Dr. Roland Knapp and I have examined the distribution of VSG for aquatic insects in lakes in Yosemite National Park. These data were collected as part of a study aimed at documenting the effects of introduced trout on aquatic communities in the park, and they constitute a sample of 68,828 individuals taken from 295 randomly chosen lakes, each visited exactly once (study YOSE-00022; Knapp, Garton, and Ladau, in preparation). When groups were taken as orders, the observed and null distributions of VSG did not differ significantly, but when they were taken as guilds, the observed distribution was shifted to the right (Ladau and Knapp, in preparation). Hence, our current data are ambiguous regarding the existence of competition within orders and guilds. To further elucidate the matter, I propose to catalog the pools for each lake. Cataloguing the pools will first allow me to determine how they are assembled -- if the observed distribution of VSG is left-shifted, right-shifted, or unmoved, competitive, aggregative, or independent assortment will be suggested, respectively. As for colonization, if the distribution for the pools is shifted to the left of the post-colonization distribution, competition will be indicated, while a right-shift or stasis will indicate aggregation or independence, respectively. Besides addressing issues fundamental to ecology, these results will provide insight into how introduced trout affect the structure of benthic communities, because both lakes with and without fish will be sampled. The latter findings will have implications for management policies (see Sanders et al, 2003; Gotelli and Graves, 1996). Methods For a given lake, the pool can contain two types of species: colonists and transients. By definition, colonists establish stable populations and are consistently present in the lake. Transients fail to prosper, and may vary in their persistence depending on their rate of influx. It follows that if a lake is sampled repeatedly, the set of omnipresent species will approximate the set of colonists more closely than the cumulative set of species, and the cumulative set will more closely approximate the pool. Hence, by repeatedly sampling lakes, I plan to approximate the pools and sets of colonists. One way that I intend to improve the accuracy of these approximations is to maximize the number of visits. Sampling a lake only a few times will yield most of the colonists, along with the transients that occupied the lake at the time, but additional sampling will produce the remainder of the transients and allow all but the most persistent ones to be correctly identified. Underlying this plan is the premise that the pools and colonization sets will not change during the observation period, for fluctuations would complicate the predictions and ruin the colonist/transient dichotomy. To ensure that the results reflect a stable system, I will restrict the analyses to a time interval in which abundances remain relatively unchanged for all species. For example, if over two weeks all abundances remain relatively constant, I will presume that species were neither invading nor being excluded over the period. Because it may be necessary to choose a small time interval to ensure stability, it will be necessary to sample lakes frequently. Specifically, I plan to sample a total of approximately fifty lakes late in the summer of 2003 and throughout the summer of 2004. Roughly fifteen of those lakes will contain pre-existing trout populations. It is particularly important that the lakes be located within the Yosemite National Park because complete data exist on the trout in these lakes, and because it is from these lakes that the other data underlying this study were collected (study YOSE-00022). Each lake will be sampled over two weeks, every third day. Lakes will be chosen so as to be lightly used, accessible by no more two days of hiking, and at elevations between 2000 and 3500 m. While I have yet to settle on exact locations, I am considering working in the Otter Lakes Basin, Ten Lakes Basin, and upper Bartlett Creek area. All areas will be within designated wilderness. Each sample will be collected using an aquatic D-net, swept over approximately 10 square meters of benthic area. To avoid depopulating the sampled areas, different benthic areas with similar habitat will be swept on successive samples. All collected specimens will be preserved in the field using 70% ethanol, and later identified using Merrit and Cummins (1996). I plan to perform identification and reporting for this project at the Sierra Nevada Aquatic Research Laboratory (SNARL). At SNARL, I also plan to complete six additional manuscripts. I will require use of one cabin during my visit, and a small amount of laboratory space already volunteered by Dr. Roland Knapp. Needless to say, by sampling small areas, some species will go uncollected. One could hence argue that rather than being transient, species that occur rarely in the samples are simply spatially rare in the lakes. However, generally speaking, spatially rare species are subject to extirpation, making them likely to be transients (Ricklefs and Miller, 2000). Thus, while spatial rarity may reduce the precision of the approximations, the set of persistent species should still best approximate the set of colonists, and the cumulative set should still best approximate the pool. References Cited Bradley, J.V. 1968. Distribution-Free Statistical Tests. Prentice-Hall, Inc, Englewood Cliffs, New Jersey. Connell, J.H. 1961. The influence of interspecific competition and other factors on the distribution of the barnacle Chthamalus stellatus. Ecology, 42: 710-723. Connor, E.F and D. Simberloff. 1979. The assembly of species communities: chance or competition? Ecology, 60: 1132-1140. Connor, E.F and D. Simberloff. 1986. Competition, Scientific Method, and Null Models in Ecology. Am. Sci., 74: 155-162. Conover, W.J. 1999. Practical Nonparametric Statistics: Third Edition John Wiley & Sons, Inc., New York. Diamond, J. M. 1975. Assembly of species communities. In Ecology and Evolution of Communities, ed. M.L. Cody and J.M. Diamond, 342-344. Harvard University Press, Cambridge. Fox, B.J. 1987. Species assembly and the evolution of community structure. Evol. Ecology, 1: 210-213. Gleason, H.A. 1926. The individualistic concept of plant association. Bull. Torey Bot. Club, 53: 7-26. Gotelli, N.J. 2000. Null model analysis of species co-occurrence patterns. Ecology, 81: 2606-2621. Gotelli, N.J. and G.R. Graves. 1996. Null models in ecology. Smithsonian Institution Press, Washington DC. Larsen, R.J. and M.L. Marx. 1986. An Introduction to Mathematical Statistics and its Applications. Prentice-Hall, Inc, Englewood Cliffs, New Jersey. Merritt, R.W. and K.W. Cummins. 1996. An introduction to the aquatic insects of North America. Kendall/Hunt Pub. Co., Dubuque, Iowa. Ricklefs, R.E. and G.L. Miller. 2000. Ecology: Fourth Edition. W.H. Freeman, New York. Sanders, N.J., N.J. Gotelli, N.E. Heller, and D.M. Gordon. 2003. Community disassembly by an invasive species. Proc. Natl. Acad. Sci. USA, 100: 2474-2477. Simberloff, D. 1970. Taxonomic diversity of island biotas. Evolution 24: 23-47.

Visit #6087 @Sierra Nevada Aquatic Research Laboratory

Approved

Under Project # 3957 | Research

The Structure of Benthic Insect Communities in Lakes in Yosemite National Park

graduate_student - Cornell University

Reservation Members(s)

Joshua Ladau Nov 1, 2004 - Apr 30, 2005 (181 days)

Reserve Resources(s) | Create Invoice

Q8 1 Nov 1, 2004 - Apr 30, 2005