An exploration of chemical cues as mediators of marine predator-prey interactions
Loading...
Author(s)
Cepeda, Marisa R.
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
Chemical cues are the language of life (Meinwald et al., 2018), from the initial steps essential for reproduction (Gomez-Diaz & Benton, 2013) through death (Kats & Dill, 1998). The study of how these molecules mediate biological interactions, now known as chemical ecology, dates back to the late 19th century, with earlier research primarily focusing on terrestrial plants and insects (Hartmann, 2008). However, in more recent decades, pioneering work has been done with respect to chemically mediated interactions in aquatic environments (Ferrari et al., 2010; Hay, 2009), and within complex microbial communities (Schmidt et al., 2019), which are main themes of this thesis. Both macro and microorganisms that live underwater rely on chemical cues to inform important life events, especially in regard to locating food and avoiding predation (Hay, 2009). Some of these molecular messages are especially complex in that they are blends, released (i.e., excreted or secreted) into the water column by predators, prey, or even co-existing microorganisms. The effects of chemical mixtures are often much greater than those of individual molecules, but identifying all of the cues involved in specific biological phenomena is difficult (Poulin & Pohnert, 2019). This is especially true in cases where there is substantial naturally occurring chemical variation, the relevant bioassay is very laborious making multiple iterations impractical, the bioassay consumes a lot of material leaving little for structure elucidation methods, or the chemical cues are relatively unstable, such that they would survive an initial extraction, but not many rounds of chromatography.
Statistically driven metabolomics has proven to be an advantageous alternative to the traditional bioassay-guided fractionation method for deciphering important chemical cues because it allows for circumvention of these obstacles, thus facilitating the elucidation of molecules that mediate marine predator-prey interactions for a deeper understanding of community dynamics (Bayona et al., 2022). Although metabolomics is a relatively young discipline, efforts have been made to begin standardizing methods within the field (Broadhurst et al., 2018; Fiehn et al., 2007; Spicer et al., 2017). Careful consideration of experimental design, sample preparation, and data acquisition is critical to reduce systematic error and avoid batch effects (Alseekh et al., 2021). In preparation for multivariate statistical analyses, spectroscopic data sets (e.g., 1H NMR and mass spectra) are typically preprocessed via multiple mathematical transformations (Castillo et al., 2011; Dieterle et al., 2006; van den Berg et al., 2006) using various software (Chang et al., 2021; Misra, 2021). Statistical models are then built to pinpoint the most important spectral features that correlate with measured biological observations (Gromski et al., 2015; Xi et al., 2014). Machine learning and variable selection can be used to refine spectral feature lists and improve statistical models in cases where the majority of features are not predictive of the measured biological data (Andersen & Bro, 2010; Broadhurst et al., 1997; Liebal et al., 2020; Sorochan Armstrong et al., 2022). Additionally, many tools have been developed to accelerate the identification of corresponding metabolites using tandem mass spectra (Böcker, 2017; Cao et al., 2021; Dührkop et al., 2019; Kuhl et al., 2012; F. Wang et al., 2021).
Chapter 1 of this thesis explores whether two particular marine metabolites released in blue crab urine, homarine and trigonelline, are common fear cues, utilized by multiple prey to assess danger from this generalist predator. While both molecules were previously found to reduce mud crab feeding, the current investigation revealed that juvenile oysters also respond to these cues by making stronger shells, a clear defensive response to perceived predation risk. Dose response experiments spanning environmentally relevant concentration ranges revealed that oyster response to these two metabolites is nuanced; oyster shell strengthening significantly increased with trigonelline concentration, while homarine was the most potent at the lowest concentration tested. When exposed to blue crab urine, oysters made stronger shells in a dose-dependent manner; however, homarine and trigonelline urine concentrations did not correlate with induced oyster shell strengthening, indicating that they are not the only, or even the primary cues oysters utilize to assess predation risk. Nonetheless, these results support the idea that common fear molecules exist, and that they may serve as the base for more complex cue mixtures used by specific prey for detecting predators.
In Chapter 2, the chemical composition of blue crab urine was explored using statistically driven metabolomics to determine which cues in this blend are the most important for oyster assessment of predation risk, and whether any of these molecules are also potentially used by mud crabs. Twenty-four blue crab urine mixtures were profiled using 1H NMR spectroscopy and mass spectrometry, highlighting that urine of different diets (mud crab vs. oyster) is distinct as expected. Partial least squares regression (PLS-R) models revealed that oyster response to blue crab urine was unrelated to diet, unlike mud crabs which associated greater predation risk with urine of predators fed conspecifics. Low urine metabolite concentrations and poor cross-validation of PLS-R models made it impractical to pursue metabolite identification using 1H NMR spectra, however these data supported that trigonelline was a minor component of the blue crab urine blend related to oyster shell strengthening. Since mass spectral features can be treated as independent variables for modeling purposes, genetic algorithm variable selection was used to greatly improve the cross-validation of these analyses. Improved PLS-R models pinpointed seven metabolites that correlate with anti-predatory responses of both oysters and mud crabs, as well as additional cues specific to risk assessment by individual prey. These results reveal that there are similarities in fear cue blends used by different prey (i.e., common fear molecules exist), and that there are also blend components uniquely leveraged by specific prey. These distinct fear-inducing cues may be more reliable indicators of predation risk for oysters, therefore emphasizing the importance of their identification.
Chapter 3 investigates unique aquatic bacterial consortia for their potential to protect microalgal crops from pests. Since bacteria are known to produce their own chemical defenses, it was hypothesized that microalgal crops benefit from associations with specific bacterial species that produced protective secondary metabolites (i.e., natural pesticides). Chemical profiling was leveraged to explore the exometabolomes of these protective bacterial communities, but spectroscopic analysis showed very few differences distinguishing protective from non-protective consortia. Additionally, identification of these molecules was not pursued due to low concentrations, and cultures could not be scaled up for further analysis due to irreproducible growth of bacterial consortia. As a result, bacterial extracts were generated from 36 isolates with protective potential and a screening experiment revealed that six were toxic to rotifers. However, rotifer toxicity could not be reproduced, likely because the bacteria were in fact not clonal isolates, as revealed by sequencing results. Alternatively, induction of natural pesticide production by bacteria could require specific environmental conditions such as temperature, nutrients, the presence of competitors, or chemical cues released by predators. A subset of bacteria were later grown with conditioned media in an attempt to induce the production of protective compounds, but none of these extracts were toxic to rotifers. Bacterial extracts were also generated from the same isolates for testing against a predatory alga, but algal growth appeared to be stimulated rather than suppressed, perhaps by extracted media components. Additionally, attempts were made to cultivate and develop a pesticide screening assay for chytrids, but the selected species grew too slowly, making them non-ideal model organisms for future studies. As a whole, these investigations shed light on the many challenges presented by complex bacterial communities, and their great potential to be leveraged as a source of natural pesticides.
In summation, this thesis explores two distinct biological systems, where communication between predators and prey are mediated by complex blends of waterborne metabolites. The multi-functionality of previously identified chemical cues was explored to determine that taxonomically distinct marine prey use the same molecules to assess predation risk. Metabolomics was leveraged to further investigate the chemical composition of urine from a shared generalist predator, accelerating the identification of additional common fear cues and potentially more reliable, specific blend components utilized by individual prey. Furthermore, chemical profiling of microalgal exometabolomes combined with an examination of associated unique bacteria stressed the difficulties associated with dereplicating chemical cues from non-model predator-prey systems and the need for development of new methods to address them. Overall, this work provides insights and emphasizes major challenges with respect to identification of important waterborne chemical mediators of predator-prey interactions.
Sponsor
Date
2023-07-27
Extent
Resource Type
Text
Resource Subtype
Dissertation