Hello!
I’m a postdoctoral researcher in the Jones lab at the University of Notre Dame. Before this, I completed my Ph.D in Integrative Biology at UC Berkeley, my master’s in Wildlife Ecology at the Parc de Beauval, and my bachelor’s at the University of Michigan.
My research primarily concerns interactions among micro-organisms and their plant hosts. I use experiments and genome sequencing to investigate how plant-associated microbes impact their host’s performance and interactions with other community members. I also have interests in succession dynamics, coexistence theory, experimental evolution, and machine learning. My study systems of choice are aquatic and carnivorous plants, and environmental bacterial isolates.
I am developing the aquatic fern genus Azolla into a tractable model system for studying the causes and consequences of host-symbiont coevolution. All Azolla species host endosymbiotic Nitrogen-fixing cyanobacteria which are passed from parent-to-offspring. As part of the DOE Joint Genome Institute’s Community Science Program, I am sequencing the genomes of this cyanobacterial symbiont (and other undescribed taxa) to investigate the timing and order of genomic adaptations to endosymbiotic lifestyles. I will also be performing reciprocal transplant experiments to identify the fitness conseqeuences of host-symbiont ‘mismatch’. Future studies will also focus on the role of symbiotic N-fixation on the development of aquatic food webs, and on the roles of symbiotic bacteria in plant-herbivore and competititive interactions.
A primary goal of my disseration research was to demonstrate links between changing microbial communities and the functioning of their ecosystems (or hosts). To accomplish this, I studied how microbial communities develop and change over time within the digestive leaves of the carnivorous pitcher plant Darlingtonia californica. This unique plant relies on a microbial food web to break down captured insect prey in a manner somewhat analagous to our own microbiota. My research entailed enumerating all compartments of pitchers’ communities (viruses, bacteria, protists, & arthropods) and linking their dynamics to rates of carbon and nitrogen mineralization using stable isotope tracers and respirometry. My results indicated strong associations between community turnover, biomass degradation, and host nutrient uptake.
For my postdoctoral research, I am using both aquatic duckweed plants (Lemnaceae) and environmental bacterial isolates to experimentally quantify the importance of various coexistence mechanisms. For instance, I study how dormant life stages (e.g., spores, seeds, turions) allow organisms to persist in fluctuating environments via the storage effect and relative nonlinearities. This work is important for our understanding of communities’ responses to environmental perturbations.
Two species of duckweeds, Lemna minor (L) and Spirodela polyrhiza ® share similar habitats, yet their different overwintering strategies and temperature optima allow them to coexist.
Here is a video I recently made to showcase my work on duckweed coexistence at the Notre Dame Linked Experimental Ecosystem Facility (LEEF)
The coexistence of two species of duckweed appears to be due to a combination of relative differences in their thermal response curves (serving to equalize long-term average fitnesses), and stabilization from the effects of both dormant life history stages (fluctuation-dependent) and negative frequency-dependent growth (fluctuation-independent).
Bacterial isolates can be evolved to express biomarkers (e.g., antibiotic resistance) for the purposes of tracking low-abundance individual cell lines in mixed cultures.
The leaves of Darlingtonia californica are among the strangest of all plants. Although many of the leaf’s features are putative adaptations enabling the trapping of insect prey, they may serve alternative purposes. Using field and lab experiments, I tested alternative adaptive hypotheses for D. californica’s “forked tongue” appendage and window-like fenestrae. In addition, I found that bacteria can alter the rheological properties of a pitcher’s fluid to facilitate prey retention (see video below).
In the future, I plan to expand these types of studies to old-world pitcher plants (Nepenthes spp.) to investigate whether this speciose group has undergone parallel convergent radiations on a number of Southeast Asian islands.
I am currently using experimental microbial communities to investigate how competition constrains (or promotes) the evolution of antibiotic resistance and elevated mutation rates. I am also using fast-growing aquatic duckweed plants to experimentally test the roles of recombination and polyploidy in adaptative evolution.
Previously, I have used the Pseudomonas fluorescens model adaptive radiation to investigate whether time-integrated area and energy were better predictors of extant diversity than were contemporary ‘snapshot’ measures.
The relationship between time-integrated productivity and diversity.
My master’s research concerned the effects of fire periodicity in facilitating bat use of the endangered (and beautiful) longleaf pine - wiregrass ecosystem. I found that high-frequency prescribed fire allowed large-bodied, fast-flying bats to forage beneath the canopy due to the reduction of physical clutter in these habitats. I also tested and developed software for the automated processing and classification of bat echolocation calls recorded on ultrasonic detectors.
Surveys of microbial diversity often encounter thousands of species coexisting within a sample. One reason for this astounding diversity may be due to the fact that what researchers often consider a single habitat is likely partitioned by niche specialists at the micron-scale. By combining metagenomic and marker gene approaches, I documented marked spatial and temporal turnover of bacteria in a salt marsh microbial mat. Such mats have great value as teaching and research tools in microbial ecology.
Google Scholar page
Monson TA, Armitage DW, Hlusko LJ. (2018). Using machine learning to classify extant apes and interpret the dental morphology of the chimpanzee-human last common ancestor. PaleoBios 35: 1-20. PDF
Armitage DW. (2017). Linking the development and functioning of a carnivorous pitcher plant’s microbial digestive community. ISME Journal 11: 2439–2451. PDF & Appendix
Thompson LR, et al., Gilbert JA, and Knight R, and EMP Consortium (including Armitage DW). (2017). A communal catalogue reveals Earth’s multiscale microbial diversity. Nature 551: 457-463. Open access link
Armitage DW. (2016). Bacteria facilitate prey capture by the pitcher plant Darlingtonia californica. Biology Letters. 12: 20160577 PDF & News Coverage (2)
Armitage DW. (2016). Time-variant species pools shape competitive dynamics and diversity-ecosystem function relationships. Proceedings of the Royal Society B 283: 20161437 PDF
Armitage DW. (2016). The cobra’s tongue: rethinking the function of the “fishtail” appendage on Darlingtonia californica. American Journal of Botany 103: 780-785. PDF
Armitage DW. (2015). Experimental evidence for a time-integrated effect of productivity on diversity. Ecology Letters 18: 1216–1225. PDF
Doll HM, Armitage DW, Daly RA, Emerson JB, Goltsman DSA, Yelton A, Kerekes J, Firestone MK, Potts MD. (2013) Utilizing novel diversity estimators to quantify multiple dimensions of microbial biodiversity across domains . BMC Microbiology 3:293. Open access link
Armitage DW, Gallagher KG, Youngblut ND, Buckley DH, Zinder SH. (2012) Millimeter-scale patterns of phylogenetic and trait diversity in a salt marsh microbial mat. Frontiers in Microbiology 3:293. Open access link
Armitage DW, Ober HK. (2012) The effects of prescribed fire on bat communities in the longleaf pine-sandhills ecosystem. Journal of Mammalogy 93:102-114. PDF
Armitage DW, Ober HK. (2010) A comparison of supervised learning algorithms in the classification of bat echolocation calls. Ecological Informatics 5:465-473. PDF
Callis KL, Christ LR, Resasco J, Armitage DW, Ash JD, Caughlin TT, Clemmensen SF, Copeland SM, Fullman TJ, Lynch RL, Olson C, Pruner RA, Vieira-Neto EHM, West-Singh R, and Bruna EM. (2009) Improving Wikipedia: educational opportunity and professional responsibitlity. Trends in Ecol. & Evol. 24:177-178. PDF
Armitage DW, Jones SE. Negative frequency-dependent growth underlies the stable coexistence of two cosmopolitan aquatic plants. (In Revision)
Milton K, Armitage DW, Sousa WP. Successional loss of key fruit resources best explains marked decline in long-stable group sizes of wild howler monkeys (Alouatta palliata).
Bertolet BL, West WE, Armitage DW, Jones SE. Methanogen community assembly varies predictably across a pH gradient in temperate lakes.
Armitage DW. (2016). Microbes in time: incorporating bacteria into ecosystem development theory. Ph.D dissertation, University of California Berkeley.
Armitage DW. (2010). The effects of prescribed fire on bat activity in the longleaf pine sandhills ecosystem. Master’s thesis, University of Florida.
Population and Community Ecology (UC Berkeley; Fall 2015) — Upper-division undergraduate course in ecology.
Vertebrate Natural History (UC Berkeley; Spring 2013 & 2015) — Upper-division undergraduate natural history course consisting of weekly lectures, laboratories, and field trips.
Intro. Biology Lab Curriculum Development (UC Berkeley; Fall 2014) — Led the development of a new two-day laboratory exercise for the ecology section of the introductory biology course. During this course I developed two web-based platforms for illustrating complex population dynamics.
Predator-Prey app / Competition-Predator-Prey app
Mammalogy (UC Berkeley; Spring 2011 & 2014) — Upper-division course on mammalian evolution, ecology, and behavior consisting of lectures, labs, and field trips.
Intro. Biology Laboratory (UC Berkeley; Fall 2010 & Spring 2016) — Lower-division lab course introducing students to ecology, evolution, and botany.
Workshop on Generalized Linear Models in R (KMUTT, Thailand; Summer 2011) — Two-week long workshop on generalized linear (mixed) models taught to Thai graduate students and wildlife professionals.
Contact information
290B Galvin Life Science Center
Notre Dame, IN 46556
email: dave.armitage[at]gmail.com
The cobra’s tongue: Rethinking the function of the“fishtail appendage” on the pitcher plant Darlingtoniacalifornica1David W. Armitage 2PREMISE OF STUDY: Carnivorous pitcher plants employ a variety of putative adaptations for prey attraction and capture. One example is the peculiar forked“fishtail appendage”, a foliar structure widely presumed to function as a prey attractant on adult leaves of Darlingtonia californica (Sarraceniaceae). Thisstudy tests the prediction that the presence of the appendage facilitates prey capture and can be considered an example of an adaptation to the carnivo-rous syndrome.METHODS: In a field experiment following a cohort of Darlingtonia leaves over their growing season, before the pitcher traps opened, the fishtail append-ages from half of the leaves were removed. Additionally, all appendages were removed from every plant at two small, isolated populations. After 54 and104 d, prey items were collected to determine whether differences in prey composition and biomass existed between experimental and unmanipulatedcontrol leaves.KEY RESULTS: Removal of the fishtail appendage did not reduce pitcher leaves’ prey biomass nor alter their prey composition at either the level of indi-vidual leaves or entire populations. Fishtail appendages on plants growing in shaded habitats contained significantly greater chlorophyll concentrationsthan those on plants growing in full sun.CONCLUSIONS: These results call into question the longstanding assumption that the fishtail appendage on Darlingtonia is an adaptation critical for theattraction and capture of prey. I suggest alternative evolutionary explanations for the role of the fishtail structure and repropose a hypothesis on the mu-tualistic nature of pitcher plant–arthropod trophic interactions.KEY WORDS carnivory; Darlingtonia; pitcher plant; prey capture; Sarraceniaceae(e.g., Joel, 1988 ). For instance, certain morphological characters ofNepenthes (Nepenthaceae) pitcher organs have been shown toserve mutualistic, rather than predatory, roles with arthropods andmammals ( Clarke et al., 2010 ; Grafe et al., 2011 ; Bazile et al., 2012 ).Likewise, many leaf traits hypothesized to facilitate prey capture inSarracenia (Sarraceniaceae) serve either an alternative or no detect-able function upon empirical evaluation ( Cresswell, 1993 ; Greenand Horner, 2007 ; Bennett and Ellison, 2009 ; Schaefer and Ruxton,2014 ).Darlingtonia californica Torr. (Sarraceniaceae), the sole memberof its genus, is the only pitcher plant native to the western UnitedStates. It is associated with habitats containing a constant source ofcold, flowing water on or downstream of serpentine formations. Arecent phylogenetic analysis suggests Darlingtonia diverged fromthe clade giving rise to Sarracenia and Heliamphora sometime dur-ing the Oligocene (25–44 million years ago) ( Ellison et al., 2012 ).
Page 2
2 • AMERICAN JOURNAL OF BOTANYHowever, due to a paucity of fossil evidence, the phenotype of thecommon ancestor to these clades is unknown ( Wong et al., 2015 ).While the pitcher leaves of Darlingtonia clearly share morphologi-cal features with other members of Sarraceniaceae, they are argu-ably the most complex within the family ( James, 1885 ; Lloyd, 1942 ;Franck, 1975 ) ( Fig. 1 ). Foliar traits unique to Darlingtonia includethe presence of transparent light-transmitting fenestrations, adownward-oriented trap entrance, and the twisting of developingleaves around their vertical axes such that consecutive leaves on arosette end up facing different directions. Perhaps the most strikingfeature of the Darlingtonia leaf is its “fishtail appendage”—a uniqueforked structure hanging off the apex of the leaf’s pitcher entrance.This feature has inspired the taxon’s common name: the cobraplant or cobra lily. At its base, the appendage merges with the nec-tar roll, which encircles the margin of the pitcher entrance. Thenectar roll continues along the margins of the fishtail appendageuntil terminating distally. The fishtail appendage’s nectar glands,red venation, and short hairs oriented toward the pitcher entrancehave led researchers to treat the structure as a prey attractant( Austin, 1875–1877 ; Lloyd, 1942 ; Juniper et al., 1989 ; Schnell, 2002 ;Ellison and Farnsworth, 2005 ), but this function has not yet beenexperimentally verified.The leaves of Darlingtonia vary in size both within and amongindividual plants. Pitcher leaves can be classified into two morpho-logical types: juvenile and adult. Juvenile leaves are small and tubu-lar (less than 5 cm in length), and found on plants less than 2 yearsof age. These lack many of the specialized morphological features ofadult leaves (e.g., the bifurcated fishtail appendage) and may func-tion to trap small arthropods ( Franck, 1976 ). The more complexadult leaves produced by mature plants vary considerably in size(5 cm to >1 m tall), but are otherwise morphologically identical.Leaf height, mass, and chlorophyll content tend to increase withshade (D. W. Armitage, unpublished data ). Dur-ing the growing season (early June–October),an individual rhizome continuously producesleaves in a rosette, with each successive leaf de-creasing in size. The pitcher entrances remainclosed until the leaf reaches its maximumheight, then the fishtail appendage rapidly de-velops and begins producing nectar. Concur-rently, the semitransparent distal portion ofthe leaf swells and forces open the trappingorifice. Small adult pitcher leaves are often re-cumbent, and their fishtail appendage mayfunction as a ramp for small terrestrial arthro-pods to access the nectar roll ( Lloyd, 1942 ).If the presence of the fishtail appendageenhances the luring and trapping prey items,then its removal from an adult pitcher shouldresult in a decrease of successful capture eventscompared with a pitcher on which it is pres-ent. However, this effect may only manifestin small adult pitchers lying prone alongthe ground. Alternatively, since individualDarlingtonia ramets can produce upward of10 pitcher leaves per growing season, theremay be a ramet or genet-level benefit of theappendage that does not manifest at the levelof the individual pitcher leaf. For instance, ex-pression of the appendage by an isolated standof leaves may serve as a long-distance attrac-tant for prey insects, which otherwise may notcongregate in the vicinity of the plant. If theabsence of these appendages does not nega-tively impact a plant’s accumulation of prey,then they should not be considered a criticaladaptation for prey capture. I tested this hy-pothesis using a manipulative field experimentin which I removed the fishtail appendagesfrom individual large (erect) and small (prone)Darlingtonia leaves before their maturation.I tracked and compared the taxonomic com-position and accumulated biomass of capturedprey between manipulated and control leavesover a 3-mo growing season. Additionally, Iremoved all appendages from two small, isolatedFIGURE 1 Mature leaf of Darlingtonia californica with fully formed fishtail appendage.
Page 3
APRIL 2016 , VOLUME 103 • ARMITAGE— DARLINGTONIA PREY CAPTURE • 3patches of Darlingtonia to test whether the presence of the ap-pendage at the population level influenced rates of prey capture.MATERIALS AND METHODSThe study was done in Plumas National Forest (Plumas County,California, USA), where numerous, patchy populations ofDarlingtoniaoccur in spring-fed seeps ranging from between <10 tothousands of individual ramets, each possessing upward of 10pitcher leaves (exact coordinates are available on request). Duringlate May 2014, I tagged 60 individual developing, unopened pitcherleaves belonging to two size classes: large ( n = 40, height > 50 cm,erect) and small ( n = 20, height < 15 cm, growing recumbently).These leaves were located in a small (<1000 total leaves), isolated(>1 km from closest known patch) population of Darlingtoniagrowing in a 30 m 2 spring-fed seep in the Butterfly Valley BotanicalArea, Plumas National Forest. One single leaf was selected per ro-sette to minimize bias caused by a plant’s genotype and to decreaseimpacts on the organism. These focal leaves were spaced a mini-mum of 50 cm apart. I removed the developing fishtail appendageusing a pair of fingernail clippers from a random sample of half ofthe leaves. Because insects may also be attracted to the scent of apitcher leaf’s putrefying detritus, I removed fishtail appendages be-fore pitchers opened and commenced trapping.After 52 d, I harvested 10 large leaves from experimental andcontrol treatments. Prey accumulation in all small leaves was at thelimit of detection, and so none were removed. All other large andsmall leaves were harvested after 104 d. I removed the prey contentsof each leaf and stored them in vials of 70% ethanol. Most arthro-pod prey are rapidly dismembered and digested by the pitchers’ as-sociated aquatic food web, leaving only the exoskeleton behind.This digestion prevents the unambiguous identification of preyitems, although the remaining material (primarily head capsulesand wings) permitted identification of individuals to the level oforder, and many to suborder or family. This material was thendried at 60 ° C for 48 h and weighed. I tested for differences in preybiomass between leaves with and without their fishtail appendagesusing an analysis of variance (ANOVA) with pitcher size and sam-pling date as covariates. Additionally, I tested whether the age andpresence of the appendage on large pitchers had an effect on thetaxonomic composition of prey insects using the permutationalanalysis of variance algorithm adonis implemented in the vegan Rpackage ( Anderson, 2001 ; Oksanen et al., 2015 ). I performed thisanalysis twice using raw compositional data primarily at the subor-der and family levels and the same data grouped at the order level.I conducted a second experiment to test the hypothesis that fish-tail appendages play a role in prey attraction or capture at the levelof individual plants or small founding populations. Early in the2015 growing season, I located two small, isolated patches ofDarlingtonia comprising between 100 and 250 pitchers and locatedat least 500 m downstream from the nearest neighboring popula-tions. Because Darlingtonia populations tend to expand down-stream along spring-fed seeps, these small founding populationsprobably represented a single genet dispersed as a seed from largerpopulations growing upstream. From these populations, I removedthe fishtail appendage from all leaves. I marked 20 unopened leavesof equivalent developmental stage and returned to the site approxi-mately every 20–30 d to remove the appendages from developingleaves. After 104 d, I removed the 20 marked pitchers from eachplot and took them to the laboratory for prey biomass measure-ment. I compared the per-leaf prey biomasses from these “popula-tion-level” removal treatments to those of the 2014 “leaf-level”removals and control treatments using ANOVA. The 2014 datawere necessarily used as controls because no other unmanipulatedand similarly isolated plots of Darlingtonia could be located in theregion. However, there were no interannual differences in preycapture rates among unmanipulated leaves sampled from largernearby populations (D. W. Armitage, unpublished data) . To in-crease statistical power, I pooled data from the two experimentalpopulations because there were no significant differences betweenthe populations’ means and variances.Because Darlingtonia can grow in shaded habitats, its fishtailstructure may also serve a photosynthetic role. To investigate thispossibility, I removed an additional 20 fishtail appendages from 10leaves growing in full sunlight and 10 plants growing in heavyshade for chlorophyll analysis. These leaves were taken from amedium-sized population growing upstream of the 2015 experimen-tal plots. I estimated the total chlorophyll [Chl a + Chl b] concentra-tion of each appendage using an Apogee CCM-200 chlorophyllcontent meter (Apogee Instruments, Logan, Utah, USA). This de-vice outputs a chlorophyll content index (CCI) that I convertedinto total chlorophyll content (mg/cm 2) using a standard calibra-tion equation ( R2 = 0.96) ( Richardson et al., 2002 ). Each fishtail ap-pendage was measured in triplicate (at its midpoint and along bothwings), and I compared the averaged estimates using a Student’s ttest. I performed all data analysis in R (v. 3.1.1) ( R Core Team,2015 ). When necessary, data were log-transformed to meet the as-sumptions of homogeneity and normality of residual variances.RESULTSPrey biomass increased with leaf age, but removal of appendagesdid not impact prey biomass accumulation ( Table 1 , Fig. 2A ). Like-wise, I did not detect an effect of population-level appendage re-movals on prey biomass ( F2,56 = 2.1, P = 0.133) ( Fig. 2B ). Despitemarked interleaf variability in prey composition, I found no evi-dence for discrimination in prey taxa between Darlingtonia leaveswith and without fishtail appendages, but prey composition dif-fered significantly between 52-d- and 104-d-old leaves ( Table 1 ;Appendix S1, see Supplemental Data with the online version of thisarticle). The majority of prey in all large pitchers belonged to theorders Hymenoptera (11.4 ± 2.0/leaf; primarily families Vespidaeand Ichneumonidae), Coleoptera (9.9 ± 2.3/leaf; primarily familyCerambycidae), and Diptera (3.9 ± 0.9/leaf). However, individuallarge-bodied lepidopterans and orthopterans occasionally greatlycontributed to prey biomass. Small pitchers contained far fewerprey than large pitchers, and the prey assemblage contained onlysmall dipterans and coleopterans <3 mm in body length (AppendixS1). Chlorophyll content was relatively low in fishtail appendagesgrowing in both full sun and shade, but appendages from shadedplants contained significantly higher chlorophyll concentrationsthan plants in full sun ( t = 4.6, df = 16.9, P < 0.0005) ( Fig. 3 ).DISCUSSIONContrary to expectations, removal of the fishtail appendage fromleaves of Darlingtonia did not hinder the plant’s ability to lure and
Page 4
4 • AMERICAN JOURNAL OF BOTANYcapture prey items, nor did it change the taxonomic composition ofthe captured prey. This outcome was observed in both large andsmall adult pitchers and was not influenced by the leaf’s age. Fur-thermore, the complete removal of fishtail appendages from entireisolated patches of Darlingtonia did not negatively impact individ-ual leaves’ abilities to trap prey compared to both untreated popu-lations and experimental leaves surrounded by neighbors possessingthe appendage. In concert, these findings suggest that the presenceof fishtail appendages within relatively isolated, small populationsof Darlingtonia may not function to attract prey into the patch norinto a leaf’s trapping orifice. These results call into question the as-sumption that all nectary-bearing structures on pitcher plant leavesare critical adaptations to the carnivorous syndrome.The fishtail appendage of Darlingtonia may instead represent avestigial or exapted structure. Members of the sister genus Sarraceniapossess an offshoot of the nectar roll that forms an operculum“hood”, replete with extrafloral nectaries along its outer rim, redvenation along the medial axis, and hairs on its underside. Similar,convergent structures can also be found on the majority of Nepenthes(Nepenthaceae) and Cephalotus (Cephalotaceae) species. Inthese genera, the operculum shows remarkable interspecific varia-tion but is generally believed to function both in prey captureand the prevention of flooding and dilution of the pitcher chamberTABLE 1. ANOVA and PERMANOVA models and test statistics for the effects of appendage removal, pitcher leaf age (days), and pitcher leaf size (mg) on preybiomass (mg) and prey composition in Darlingtonia californica leaves. No significant interactions were found among any covariates.Response variableCovariatedfSSFPPrey biomassFishtail appendage15.711.9520.168Pitcher age1142.348.679<0.0001Pitcher size15.81.9840.165Residuals56163.7Prey composition (orders)Fishtail appendage10.2301.1310.324Pitcher age12.48912.236<0.0001Residuals367.324Prey composition (raw)Fishtail appendage10.4971.5580.101Pitcher age12.0336.377<0.0001Residuals3611.479from rainwater ( Juniper et al., 1989 ). Within Sarraceniaceae,Sarracenia purpurea and members of the neotropical genusHeliamphora lack an operculum, although derivatives of this struc-ture manifest in S. purpurea as a vertical extension of the nectar rollpossessing all of its characteristic traits and in Heliamphora as areduced appendage possessing enlarged nectar glands and no hairs( Adams and Smith, 1977 ). In the latter, flooding is mitigated via asmall slit in the leaf’s keel serving as a drain ( Lloyd, 1942 ). InDarlingtonia and the morphologically similar Sarracenia psittacina,the flooding problem is ameliorated by the species’ downward-facing pitcher openings. In these taxa, an inflated, semitransparent,hairless “hood” wraps over the entire upper portion of the pitcherleaf, and the operculum covering is either lost completely (as inS. psittacina) or remains as the fishtail appendage on Darlingtonia( Lloyd, 1942 ).The fishtail appendage of Darlingtonia and the opercula ofSarracenia and Heliamphora share many developmental and mor-phological features. The pitcher leaves of Sarraceniaceae undergonearly identical patterns of early histogenesis wherein the apicaltissue of the leaf primordium gives rise to the operculum in Sarrace-nia or the hood and fishtail appendage in Darlingtonia ( Lloyd,1942 ; Franck, 1975 ; Fukushima et al., 2015 ). Arber (1941) observedthat the leaves of Sarraceniaceae share a similar pattern of venationwherein the majority of veinsconverge at the midrib’s apex atexactly the point where the oper-cula and fishtail appendage origi-nate. The author hypothesizedthat this clustering of veins re-sults in a disproportionate supplyof nutrients and localized hyper-trophy, which, in turn, initiatesthe development of the pitcher’soperculum or fishtail appendage( Arber, 1941 ). On adult leavesof Darlingtonia and Sarracenia,nectar glands ring the margins oftheir appendages, and the posi-tioning and location of hairs onthe undersides of both structuresis identical, although the hairs arereduced in Darlingtonia ( Adamsand Smith, 1977 ). In Sarracenia,these hairs are hypothesized toFIGURE 2 (A) Prey biomass as a function of fishtail appendage presence ( x-axes), pitcher size (facets), andpitcher age (shading); (B) prey biomass as a function of appendage removal treatment. Boxes denote 25ththrough 75th percentiles; bolded lines denote median values.
Page 5
APRIL 2016 , VOLUME 103 • ARMITAGE— DARLINGTONIA PREY CAPTURE • 5FIGURE 3 Total chlorophyll content of fishtail appendages from sun- andshade-growing Darlingtonia populations. Boxes denote 25th through75th percentiles; bolded lines denote median values.inhibit the foothold of insects, which then fall directly into thepitcher trap ( Lloyd, 1942 ). In Darlingtonia, however, these hairswould cause potential prey to fall from the leaf. These observationsare consistent with the fishtail appendage of Darlingtonia being ho-mologous with the operculum of Sarracenia and the nectar spoonof Heliamphora.Given the morphological plasticity of the fishtail and opercularappendages in Sarraceniaceae, these structures may serve roles con-tingent upon their local environments. It was hypothesized thatplants’ investments in traits that facilitate carnivory come with acost to photosynthetic efficiency such that they can only thrive inhabitats where they will not experience strong competition withnoncarnivorous plants for light, water, or nutrients ( Givnish et al.,1984 ). Thus, plants should reduce their investment in carnivorousorgans in low-light environments and instead reallocate growth tophotosynthetic tissues ( Zamora et al., 1998 ; Thorén et al., 2003 ).Using the size of the fishtail appendage as a proxy for a plant’s in-vestment in prey capture, Ellison and Farnsworth (2005) concludedthat, contrary to their expectations, Darlingtonia plants were un-able to regulate the size of the fishtail appendage under changingphotosynthetic conditions. The authors also detected remarkablylow overall photosynthetic rates for all plants measured ( Ellisonand Farnsworth, 2005 ). Unlike other members of Sarraceniaceae,however, Darlingtonia can be found growing in shaded habitats(D. W. Armitage, personal observations), where the plants’ fishtailappendages contain greater chlorophyll concentrations than plantsgrowing in full sun. This finding suggests that the fishtail append-age might serve a photosynthetic role in light-limited habitats.Therefore, the structure should not be assumed to be independentof a plants’ photosynthetic abilities nor a proxy for investment incarnivory.The fishtail appendage of Darlingtonia may instead be involvedin supporting local insect populations—a phenomenon that hasalso been observed in the genus Nepenthes ( Bazile et al., 2012 ). Re-becca Austin (1875–1877) , the first botanist to study the naturalhistory of Darlingtonia in the wild, noted that flies would regularlyalight on the fishtail appendage and feed on its nectar, while only avery small minority were drawn into the leaf and trapped. The lowcapture probability of Darlingtonia was later estimated to be ap-proximately 1.7% of vespulid wasps that had already landed on apitcher leaf ( Dixon et al., 2005 ). The probability of successful preycapture (compared with the overall number of visits) in otherpitcher plant species is similarly low, and experimental and obser-vational studies have failed to support the hypothesized roles ofnumerous other foliar traits in prey capture ( Cresswell, 1993 ; Greenand Horner, 2007 ; Bennett and Ellison, 2009 ; Schaefer and Ruxton,2014 ). These findings, along with the common observation of in-sects feeding on pitcher leaves and avoiding capture, beg an alter-native explanation for the fishtail appendage’s role. Joel (1988)hypothesized that the relationship between pitcher plants and in-sects represents a mutualistic, rather than a predator–prey interac-tion. The author reasoned that because pitcher plants are commonlyfound in hydric, sunny habitats, their liberal nectar offerings do notnegatively impact a plant’s water and carbon budgets. Pitcherplants in the family Sarraceniaceae are often the dominant nectar-bearing members of their local communities and supply nectar toinsects continuously throughout the growing season. Nectivorousinsects quickly learn to exploit this resource, and the behavior ismaintained as long as insect mortality via entrapment is not com-mon enough to be selected against. With its bountiful nectar glands,the fishtail appendage on Darlingtonia may facilitate insects’ feed-ing on the plant’s nectar while remaining incidental to prey cap-ture. Additional sources of “unsafe” nectar along the margins of thepitcher entrance may tempt the occasional insect into entering thetrap while the “safe” nectar presented on the fishtail appendagemay promote and maintain visitation by nectivorous insects. Theplant would then benefit from the small proportion of insectslured into the pitcher trap. However, my data do not support thishypothesis, since the per capita prey biomass in isolated, appendage-free populations was not significantly lower than in control popula-tions, although a nonsignificant negative trend was observed. Analternative approach toward testing this hypothesis would involvequantifying the relative contributions of pitcher nectar to the dietsof the local insect community. Such a study could be carried outusing stable isotope ratios of nitrogen, which are anticipated to dif-fer between the nectars of carnivorous and noncarnivorous plantsand likewise between insects selectively feeding on either nectarsource (e.g., Zanden and Rasmussen, 2001 ).In conclusion, by experimentally removing the unique fishtailappendages on leaves of Darlingtonia californica, I demonstratedthat the absence of the structure does not hinder a leaf’s ability to
Page 6
6 • AMERICAN JOURNAL OF BOTANYcapture prey. This finding contrasts with over a century of descrip-tive literature categorizing the structure as an adaptation to the lur-ing and trapping of prey. There are at least three potential alternativeexplanations for this structure’s evolutionary persistence: (1) as avestigial homologous structure under little to no selection; (2) as afacultatively photosynthetic structure facilitating carbon acquisi-tion in shaded habitats; and (3) as a source of “safe” nectar encour-aging visitation by nectivorous insects and supporting their localpopulations while entrapping a small proportion of visitors. Thesealternative roles need not be mutually exclusive and likely dependon environmental factors. Adaptive explanations for the uniquemorphological and physiological traits possessed by carnivorousplants pervade the literature. However, barring experimental ver-ification, these adaptive hypotheses remain speculative. Goingforward, researchers are encouraged to take a more pluralistic ap-proach in their treatments of carnivorous plants by consideringalternative evolutionary explanations for the unique foliar featuresof these plants.ACKNOWLEDGEMENTSThe author thanks O. Cope and K. Saunders for assistance witharthropod identification, H. Uno and R. Leon for assistance withdata collection, and W. Sousa, N. Matzke, J. Benca, and two anony-mous reviewers for comments on the manuscript. This work wassupported by an NSF Graduate Research Fellowship to D.W.A. andby funding from the California Native Plant Society, the NativePlant Society of Oregon, and NSF DEB-1406524. Permits for fieldcollection of Darlingtonia leaves were issued by the U. S. ForestService (J. Belsher-Howe, Mt. Hough District).LITERATURE CITEDAdams , R. M. II , and G. W. Smith . 1977 . An S.E.M. survey of the five carnivo-rous pitcher plant genera. American Journal of Botany 64 : 265 – 272 .Anderson , M. J. 2001 . A new method for non-parametric multivariate analysisof variance. Austral Ecology 26 : 32 – 46 .Arber , A. 1941 . On the morphology of the pitcher-leaves in Heliamphora,Sarracenia, Darlingtonia, Cephalotus, and Nepenthes. Annals of Botany 5 :563 – 578 .Austin , R. M. 1875–1877 . Selected letters to W. M. Canby. Society of NaturalHistory of Delaware , Wilmington, Delaware, USA .Bazile , V. , J. A. Moran , G. Le Moguédec , D. J. Marshall , and L. Gaume . 2012 .A carnivorous plant fed by its ant symbiont: A unique multi-faceted nutri-tional mutualism. PLoS One 7 : e36179 .Bennett , K. F. , and A. M. Ellison . 2009 . Nectar, not colour, may lure insects totheir death. Biology Letters 5 : 469 – 472 .Clarke , C. , J. A. Moran , and L. Chin . 2010 . Mutualism between tree shrews andpitcher plants. Plant Signaling & Behavior 5 : 1187 – 1189 .Cresswell , J. E. 1993 . The morphological correlates of prey capture and re-source parasitism in pitchers of the carnivorous plant Sarracenia purpurea.American Midland Naturalist 129 : 35 – 41 .Dixon , P. M. , A. M. Ellison , and N. J. Gotelli . 2005 . Improving the precision ofestimates of the frequency of rare events. Ecology 86 : 1114 – 1123 .Ellison , A. M. , E. D. Butler , E. J. Hicks , R. F. C. Naczi , P. J. Calie , C. D. Bell , andC. C. Davis . 2012 . Phylogeny and biogeography of the carnivorous plantfamily Sarraceniaceae. PLoS One 7 : e39291 .Ellison , A. M. , and E. J. Farnsworth . 2005 . The cost of carnivory for Darlingtoniacalifornica (Sarraceniaceae): Evidence from relationships among leaf traits.American Journal of Botany 92 : 1085 – 1093 .Franck , D. H. 1975 . Early histogenesis of the adult leaves of Darlingtonia cali-fornica (Sarraceniaceae) and its bearing on the nature of epiascidiate foliarappendages. American Journal of Botany 62 : 116 – 132 .Franck , D. H. 1976 . Comparative morphology and early leaf histogenesis ofadult and juvenile leaves of Darlingtonia californica and their bearing on theconcept of heterophylly. Botanical Gazette 137 : 20 – 34 .Fukushima , K. , H. Fujita , T. Yamaguchi , M. Kawaguchi , H. Tsukaya , and M.Hasebe . 2015 . Oriented cell division shapes carnivorous pitcher leaves ofSarracenia purpurea. Nature Communications 6 : 6450 .Givnish , T. J. , E. L. Burkhardt , R. E. Happel , and J. D. Weintraub . 1984 .Carnivory in the bromeliad Brocchinia reducta, with a cost/benefit modelfor the general restriction of carnivorous plants to sunny, moist, nutrient-poor habitats. American Naturalist 124 : 479 – 497 .Grafe , T. U. , C. R. Schöner , G. Kerth , A. Junaidi , and M. G. Schöner . 2011 . Anovel resource–service mutualism between bats and pitcher plants. BiologyLetters 7 : 436 – 439 .Green , M. L. , and J. D. Horner . 2007 . The relationship between prey captureand characteristics of the carnivorous pitcher plant, Sarracenia Alata Wood.American Midland Naturalist 158 : 424 – 431 .James , J. F. 1885 . How the pitcher plant got its leaves. American Naturalist 19 :567 – 578 .Joel , D. M. 1988 . Mimicry and mutualism in carnivorous pitcher plants(Sarraceniaceae, Nepenthaceae, Cephalotaceae, Bromeliaceae). BiologicalJournal of the Linnean Society 35 : 185 – 197 .Juniper , B. B. E. , R. J. Robins , and D. M. Joel . 1989 . The carnivorous plants.Academic Press, London, UK.Lloyd , F. E. 1942 . The carnivorous plants. Chronica Botanica, vol. 9. RonaldPress, New York, New York, USA.Oksanen , J. , F. G. Blanchet , R. Kindt , P. Legendre , P. R. Minchin , R. B. O’Hara ,G. L. Simpson , et al. 2015 . vegan: Community Ecology R Package, version2.1-10. Website http://CRAN.R-project.org/package=vegan .R Core Team . 2015 . R: A language and environment for statistical comput-ing. R Foundation for Statistical Computing, Vienna, Austria. Available athttp://www.R-project.org .Richardson , A. D. , S. P. Duigan , and G. P. Berlyn . 2002 . An evaluation of non-invasive methods to estimate foliar chlorophyll content. New Phytologist153 : 185 – 194 .Schaefer , H. M. , and G. D. Ruxton . 2014 . Fenestration: A window of opportu-nity for carnivorous plants. Biology Letters 10 : 20140134 .Schnell , D. E. 2002 . Carnivorous plants of the United States and Canada , 2nded. Timber Press, Portland, Oregon, USA.Thorén , L. M. , J. Tuomi , T. Kämäräinen , and K. Laine . 2003 . Resource availabil-ity affects investment in carnivory in Drosera rotundifolia. New Phytologist159 : 507 – 511 .Wong , W. O. , D. L. Dilcher , C. C. Labandeira , G. Sun , and A. Fleischmann .2015 . Early Cretaceous Archaeamphora is not a carnivorous angiosperm.Frontiers in Plant Science 6 : 326 .Zamora , R. , J. M. Gómez , and J. A. Hódar . 1998 . Fitness responses of a car-nivorous plant in contrasting ecological scenarios. Ecology 79 : 1630 – 1644 .Zanden , M. J. V. , and J. B. Rasmussen . 2001 . Variation in δ 15N and δ 13C tro-phic fractionation: Implications for aquatic food web studies. Limnology andOceanography 46 : 2061 – 2066
rsbl.royalsocietypublishing.orgResearchCite this article: Armitage DW. 2016 Bacteriafacilitate prey retention by the pitcher plantDarlingtonia californica. Biol. Lett. 12:20160577.http://dx.doi.org/10.1098/rsbl.2016.0577Received: 6 July 2016Accepted: 26 October 2016Subject Areas:biomechanics, ecologyKeywords:bacteria, pitcher plant, rheology,surface tension, carnivorous plant,Darlingtonia californicaAuthor for correspondence:David W. Armitagee-mail: dave.armitage@gmail.comElectronic supplementary material is availableonline at https://dx.doi.org/10.6084/m9.fig-share.c.3571608.BiomechanicsBacteria facilitate prey retention by thepitcher plant Darlingtonia californicaDavid W. Armitage1,21Department of Integrative Biology, University of California Berkeley, 3040 Valley Life Sciences Building,Berkeley, CA 94720-3140, USA2Department of Biological Sciences, University of Notre Dame, 290B Galvin Life Science Center, Notre Dame,IN 46556, USADWA, 0000-0002-5677-0501Bacteria are hypothesized to provide a variety of beneficial functions to plants.Many carnivorous pitcher plants, for example, rely on bacteria for digestion ofcaptured prey. This bacterial community may also be responsible for the lowsurface tensions commonly observed in pitcher plant digestive fluids, whichmight facilitate prey capture. I tested this hypothesis by comparing the phys-ical properties of natural pitcher fluid from the pitcher plant Darlingtoniacalifornica and cultured ‘artificial’ pitcher fluids and tested these fluids’ preyretention capabilities. I found that cultures of pitcher leaves’ bacterial commu-nities had similar physical properties to raw pitcher fluids. These propertiesfacilitated the retention of insects by both fluids and hint at a previouslyundescribed class of plant–microbe interaction.1. IntroductionPlants have evolved a variety of strategies for living in nutrient-poor environments.One particularly widespread strategy is the close association between a host plantand one or more species of beneficial bacteria and fungi. These interactions com-monly involve a mutually beneficial exchange of the host’s photosynthates formineral nutrients scavenged from the soil or fixed from the atmosphere [1]. Inaddition to assistance with nutrient acquisition, plant-associated microbiota mayalso perform beneficial secondary functions. For instance, microbes may facilitatedisease suppression outside of the host’s own immune response [2] or aid in theremoval of growth-inhibiting compounds [3]. Outside of these cases, however,novel classes of beneficial plant–microbe interactions remain elusive.Carnivory is another adaptation to nutrient-poor habitats, and plants haveevolved a variety of methods to trap and digest arthropod prey [4]. Although themajority of carnivorous plants produce their own digestive enzymes to breakdown prey, members of the pitcher plant family Sarraceniaceae are hypothesizedto rely heavily on an associated microbial digestive community [5,6]. Theseplants possess modified, fluid-filled leaves in which prey are retained and digestedby a community of mutualistic aquatic invertebrates and bacteria [7,8].Fluid from these plants’ leaves can have lower interfacial (surface) tensions thanwater [4,5,9]—a property interpreted to facilitate the retention and drowningof prey. Because bacterial biomass can be very high in pitcher plant fluid(109–1011 cells ml21) [10], and because many bacteria produce biosurfactantshypothesized to aid the digestion of water-insoluble compounds such as lipids[11], it stands that bacteria may be causing a reduction of the pitcher fluid’s inter-facial tension. Based on the observation that insects added to the pitcher fluid ofadult Darlingtonia californica Torr. (Sarraceniaceae) leaves experience difficultyescaping, I measured the tensile and rheological properties of pitcher fluid andpitcher bacterial cultures and tested whether prey retention by pitcher plantbacterial cultures was comparable to that of natural pitcher fluid
rsbl.royalsocietypublishing.orgResearchCite this article: Armitage DW. 2016 Bacteriafacilitate prey retention by the pitcher plantDarlingtonia californica. Biol. Lett. 12:20160577.http://dx.doi.org/10.1098/rsbl.2016.0577Received: 6 July 2016Accepted: 26 October 2016Subject Areas:biomechanics, ecologyKeywords:bacteria, pitcher plant, rheology,surface tension, carnivorous plant,Darlingtonia californicaAuthor for correspondence:David W. Armitagee-mail: dave.armitage@gmail.comElectronic supplementary material is availableonline at https://dx.doi.org/10.6084/m9.fig-share.c.3571608.BiomechanicsBacteria facilitate prey retention by thepitcher plant Darlingtonia californicaDavid W. Armitage1,21Department of Integrative Biology, University of California Berkeley, 3040 Valley Life Sciences Building,Berkeley, CA 94720-3140, USA2Department of Biological Sciences, University of Notre Dame, 290B Galvin Life Science Center, Notre Dame,IN 46556, USADWA, 0000-0002-5677-0501Bacteria are hypothesized to provide a variety of beneficial functions to plants.Many carnivorous pitcher plants, for example, rely on bacteria for digestion ofcaptured prey. This bacterial community may also be responsible for the lowsurface tensions commonly observed in pitcher plant digestive fluids, whichmight facilitate prey capture. I tested this hypothesis by comparing the phys-ical properties of natural pitcher fluid from the pitcher plant Darlingtoniacalifornica and cultured ‘artificial’ pitcher fluids and tested these fluids’ preyretention capabilities. I found that cultures of pitcher leaves’ bacterial commu-nities had similar physical properties to raw pitcher fluids. These propertiesfacilitated the retention of insects by both fluids and hint at a previouslyundescribed class of plant–microbe interaction.1. IntroductionPlants have evolved a variety of strategies for living in nutrient-poor environments.One particularly widespread strategy is the close association between a host plantand one or more species of beneficial bacteria and fungi. These interactions com-monly involve a mutually beneficial exchange of the host’s photosynthates formineral nutrients scavenged from the soil or fixed from the atmosphere [1]. Inaddition to assistance with nutrient acquisition, plant-associated microbiota mayalso perform beneficial secondary functions. For instance, microbes may facilitatedisease suppression outside of the host’s own immune response [2] or aid in theremoval of growth-inhibiting compounds [3]. Outside of these cases, however,novel classes of beneficial plant–microbe interactions remain elusive.Carnivory is another adaptation to nutrient-poor habitats, and plants haveevolved a variety of methods to trap and digest arthropod prey [4]. Although themajority of carnivorous plants produce their own digestive enzymes to breakdown prey, members of the pitcher plant family Sarraceniaceae are hypothesizedto rely heavily on an associated microbial digestive community [5,6]. Theseplants possess modified, fluid-filled leaves in which prey are retained and digestedby a community of mutualistic aquatic invertebrates and bacteria [7,8].Fluid from these plants’ leaves can have lower interfacial (surface) tensions thanwater [4,5,9]—a property interpreted to facilitate the retention and drowningof prey. Because bacterial biomass can be very high in pitcher plant fluid(109–1011 cells ml21) [10], and because many bacteria produce biosurfactantshypothesized to aid the digestion of water-insoluble compounds such as lipids[11], it stands that bacteria may be causing a reduction of the pitcher fluid’s inter-facial tension. Based on the observation that insects added to the pitcher fluid ofadult Darlingtonia californica Torr. (Sarraceniaceae) leaves experience difficultyescaping, I measured the tensile and rheological properties of pitcher fluid andpitcher bacterial cultures and tested whether prey retention by pitcher plantbacterial cultures was comparable to that of natural pitcher fluids.© 2016 The Author(s) Published by the Royal Society. All rights reserved.on November 23, 2016http://rsbl.royalsocietypublishing.org/Downloaded from
Page 2
2. Material and methodsIn July 2014, I collected fluid from six individual D. californicaleaves growing in Plumas National Forest, California. Fourleaves belonged to the year’s first cohort, which began trappingprey in mid-June. Two additional samples were collected fromtwo-week-old leaves.To create ‘artifical’ pitcher fluid, I inoculated four glass tubeswith 10 ml of autoclaved, 0.2 mm-filtered water collected from myfield site. Into this water, I added 3 g l21 autoclaved powder fromfreeze-dried, ground crickets and a 50 ml aliquot from one of fourmonth-old 150 mm-filtered pitcher fluid samples. To simulate theoxygen environments of pitcher leaves [12], I briefly bubbledcultures every 2 h from 08.00 to 20.00. After 30 days, the cultureswere stored at 58C. Dilutions of each culture were also plated onR2A agar and colony-forming units (CFUs) were counted.To measure the interfacial tension of pitcher fluid samplesand bacterial cultures, I used a pendant-drop tensiometer [13]to photograph the profiles of individual 1.0 mm-filtered dropletssuspended from clean glass capillary tubes [14] (electronicsupplementary material, figure S1). I repeated this process for10 drops per sample. I estimated the interfacial tension, g, ofeach drop photograph using the method of Stauffer [15]. Thismethod requires the equatorial diameter, De, of the drop, andDs, the diameter of the drop at a vertical distance De from thedrop’s lower apex. The ratio Ds/De ¼ S relates to a correctionfactor H required by the equationg ¼ gD2eDrH,ð2.1Þwhere g is the acceleration due to gravity (9.81 m s22) and Dr isthe difference in densities between the droplet (997kgm23)and air at room temperature. The correction factor H wasestimated with the equation1H¼ keSÀkm þ k3S3 þ k2S2 þ k1S þ k0,ð2.2Þwhere ki are empirical constants [16]. I averaged six estimates froma single drop’s photographs for a per-drop estimate of g, and thenaveraged these estimates across 10 replicate drops per sample toestimate each sample’s interfacial tension. The measurementsof pure water and 70% ethanol were nearly identical to their stan-dard theoretical values. I used ANOVA and Tukey’s range test toinvestigate differences between sample means.I used a cone-plate viscometer (Brookfield Engineering) tomeasure the shear viscosities of 0.5 ml pitcher fluid samples atshear rates between 100 and 1600 s21 [17]. I plotted samples’viscosities against their log shear rates and fitted linear regressionlines to each series using R [18]. I expected shear-thinning orthickening fluids to have non-zero slopes [17].I conducted an experiment to test the effects of pitcher plantfluid and bacterial cultures on prey retention of the red harvesterant Pogonomyrmex barbatus—a close relative of species foundtrapped in plants at the sample collection site. Individual antswere dropped into 15ml centrifuge tubes containing 5ml ofone of the following substances: pure deionized water, naturalpitcher fluid and fluid pooled from the pitcher bacterial cultures.Escape behaviour of the ants was filmed for 10 min, at whichpoint they were removed and allowed to recover. The experimentwas repeated 30 times (using fresh ants for each trial) and Icalculated the frequency of escapes for each sample.Next, I tested the minimal concentration of bacterial culture(BACT) required for retaining ants. I created serial dilutions ofpooled cultures in pure water at regular increments from 1021 to1023, added ants into 5 ml of each dilution, and filmed their behav-iour for 10 min. This was repeated 12 times, using new ants for eachtreatment and replicate, and the number of successful escapes wasscored. I used logistic regression analysis to test whether an ant’sprobability of escape from the fluid was associated with its dilution.3. Results and discussionPitcher fluid from Darlingtonia had interfacial tensions signi-ficantly lower than water and higher than 70% ethanol(figure 1a; all statistical results are presented in the electronicsupplementary material, data). The same pattern was observedfor bacterial cultures seeded with Darlingtonia bacteria.These bacterial cultures had very similar mean interfacia
rsbl.royalsocietypublishing.orgResearchCite this article: Armitage DW. 2016 Bacteriafacilitate prey retention by the pitcher plantDarlingtonia californica. Biol. Lett. 12:20160577.http://dx.doi.org/10.1098/rsbl.2016.0577Received: 6 July 2016Accepted: 26 October 2016Subject Areas:biomechanics, ecologyKeywords:bacteria, pitcher plant, rheology,surface tension, carnivorous plant,Darlingtonia californicaAuthor for correspondence:David W. Armitagee-mail: dave.armitage@gmail.comElectronic supplementary material is availableonline at https://dx.doi.org/10.6084/m9.fig-share.c.3571608.BiomechanicsBacteria facilitate prey retention by thepitcher plant Darlingtonia californicaDavid W. Armitage1,21Department of Integrative Biology, University of California Berkeley, 3040 Valley Life Sciences Building,Berkeley, CA 94720-3140, USA2Department of Biological Sciences, University of Notre Dame, 290B Galvin Life Science Center, Notre Dame,IN 46556, USADWA, 0000-0002-5677-0501Bacteria are hypothesized to provide a variety of beneficial functions to plants.Many carnivorous pitcher plants, for example, rely on bacteria for digestion ofcaptured prey. This bacterial community may also be responsible for the lowsurface tensions commonly observed in pitcher plant digestive fluids, whichmight facilitate prey capture. I tested this hypothesis by comparing the phys-ical properties of natural pitcher fluid from the pitcher plant Darlingtoniacalifornica and cultured ‘artificial’ pitcher fluids and tested these fluids’ preyretention capabilities. I found that cultures of pitcher leaves’ bacterial commu-nities had similar physical properties to raw pitcher fluids. These propertiesfacilitated the retention of insects by both fluids and hint at a previouslyundescribed class of plant–microbe interaction.1. IntroductionPlants have evolved a variety of strategies for living in nutrient-poor environments.One particularly widespread strategy is the close association between a host plantand one or more species of beneficial bacteria and fungi. These interactions com-monly involve a mutually beneficial exchange of the host’s photosynthates formineral nutrients scavenged from the soil or fixed from the atmosphere [1]. Inaddition to assistance with nutrient acquisition, plant-associated microbiota mayalso perform beneficial secondary functions. For instance, microbes may facilitatedisease suppression outside of the host’s own immune response [2] or aid in theremoval of growth-inhibiting compounds [3]. Outside of these cases, however,novel classes of beneficial plant–microbe interactions remain elusive.Carnivory is another adaptation to nutrient-poor habitats, and plants haveevolved a variety of methods to trap and digest arthropod prey [4]. Although themajority of carnivorous plants produce their own digestive enzymes to breakdown prey, members of the pitcher plant family Sarraceniaceae are hypothesizedto rely heavily on an associated microbial digestive community [5,6]. Theseplants possess modified, fluid-filled leaves in which prey are retained and digestedby a community of mutualistic aquatic invertebrates and bacteria [7,8].Fluid from these plants’ leaves can have lower interfacial (surface) tensions thanwater [4,5,9]—a property interpreted to facilitate the retention and drowningof prey. Because bacterial biomass can be very high in pitcher plant fluid(109–1011 cells ml21) [10], and because many bacteria produce biosurfactantshypothesized to aid the digestion of water-insoluble compounds such as lipids[11], it stands that bacteria may be causing a reduction of the pitcher fluid’s inter-facial tension. Based on the observation that insects added to the pitcher fluid ofadult Darlingtonia californica Torr. (Sarraceniaceae) leaves experience difficultyescaping, I measured the tensile and rheological properties of pitcher fluid andpitcher bacterial cultures and tested whether prey retention by pitcher plantbacterial cultures was comparable to that of natural pitcher fluids.© 2016 The Author(s) Published by the Royal Society. All rights reserved.on November 23, 2016http://rsbl.royalsocietypublishing.org/Downloaded from
Page 2
2. Material and methodsIn July 2014, I collected fluid from six individual D. californicaleaves growing in Plumas National Forest, California. Fourleaves belonged to the year’s first cohort, which began trappingprey in mid-June. Two additional samples were collected fromtwo-week-old leaves.To create ‘artifical’ pitcher fluid, I inoculated four glass tubeswith 10 ml of autoclaved, 0.2 mm-filtered water collected from myfield site. Into this water, I added 3 g l21 autoclaved powder fromfreeze-dried, ground crickets and a 50 ml aliquot from one of fourmonth-old 150 mm-filtered pitcher fluid samples. To simulate theoxygen environments of pitcher leaves [12], I briefly bubbledcultures every 2 h from 08.00 to 20.00. After 30 days, the cultureswere stored at 58C. Dilutions of each culture were also plated onR2A agar and colony-forming units (CFUs) were counted.To measure the interfacial tension of pitcher fluid samplesand bacterial cultures, I used a pendant-drop tensiometer [13]to photograph the profiles of individual 1.0 mm-filtered dropletssuspended from clean glass capillary tubes [14] (electronicsupplementary material, figure S1). I repeated this process for10 drops per sample. I estimated the interfacial tension, g, ofeach drop photograph using the method of Stauffer [15]. Thismethod requires the equatorial diameter, De, of the drop, andDs, the diameter of the drop at a vertical distance De from thedrop’s lower apex. The ratio Ds/De ¼ S relates to a correctionfactor H required by the equationg ¼ gD2eDrH,ð2.1Þwhere g is the acceleration due to gravity (9.81 m s22) and Dr isthe difference in densities between the droplet (997kgm23)and air at room temperature. The correction factor H wasestimated with the equation1H¼ keSÀkm þ k3S3 þ k2S2 þ k1S þ k0,ð2.2Þwhere ki are empirical constants [16]. I averaged six estimates froma single drop’s photographs for a per-drop estimate of g, and thenaveraged these estimates across 10 replicate drops per sample toestimate each sample’s interfacial tension. The measurementsof pure water and 70% ethanol were nearly identical to their stan-dard theoretical values. I used ANOVA and Tukey’s range test toinvestigate differences between sample means.I used a cone-plate viscometer (Brookfield Engineering) tomeasure the shear viscosities of 0.5 ml pitcher fluid samples atshear rates between 100 and 1600 s21 [17]. I plotted samples’viscosities against their log shear rates and fitted linear regressionlines to each series using R [18]. I expected shear-thinning orthickening fluids to have non-zero slopes [17].I conducted an experiment to test the effects of pitcher plantfluid and bacterial cultures on prey retention of the red harvesterant Pogonomyrmex barbatus—a close relative of species foundtrapped in plants at the sample collection site. Individual antswere dropped into 15ml centrifuge tubes containing 5ml ofone of the following substances: pure deionized water, naturalpitcher fluid and fluid pooled from the pitcher bacterial cultures.Escape behaviour of the ants was filmed for 10 min, at whichpoint they were removed and allowed to recover. The experimentwas repeated 30 times (using fresh ants for each trial) and Icalculated the frequency of escapes for each sample.Next, I tested the minimal concentration of bacterial culture(BACT) required for retaining ants. I created serial dilutions ofpooled cultures in pure water at regular increments from 1021 to1023, added ants into 5 ml of each dilution, and filmed their behav-iour for 10 min. This was repeated 12 times, using new ants for eachtreatment and replicate, and the number of successful escapes wasscored. I used logistic regression analysis to test whether an ant’sprobability of escape from the fluid was associated with its dilution.3. Results and discussionPitcher fluid from Darlingtonia had interfacial tensions signi-ficantly lower than water and higher than 70% ethanol(figure 1a; all statistical results are presented in the electronicsupplementary material, data). The same pattern was observedfor bacterial cultures seeded with Darlingtonia bacteria.These bacterial cultures had very similar mean interfacial1.01.11.21.31.42005001000 1600shear rate (s–1)shear viscosity (cp)BACT (b=–0.02, p < 0.001)DACA 1 (b =–0.04, p < 0.001)DACA 2 (b=–0.04, p < 0.001)DACA 3 (b=–0.08, p < 0.001)water (b=–0.004, p = 0.526)01020304050607080interfacial tension (dynes cm–1)(a)(b)aedbbcbc bcbcbccdcddwaterDACA 1DACA 2DACA 3DACA 4DACA 5DACA 6BACT 1BACT 2BACT 3BACT 470% ethanolFigure 1. (a) Interfacial tension measurements for fluid samples. Samples sharing letters do not significantly differ from one another (p . 0.01). The dotted linedenotes mean value for aggregated Darlingtonia (DACA) fluid and bacterial culture samples. BACT sample numbers correspond to the DACA samples from which theywere inoculated. DACA samples 5 and 6 are from young leaves. 1 dyne ¼ 1Â1025 N. (b) Shear viscosity measurements on selected fluid samples. Significantly negativeslopes (b) of Darlingtonia fluid and bacterial cultures indicate very slight shear-thinning properties. 1 cp (centipoise)¼1023 Pa s. (Online version in colour.)rsbl.royalsocietypublishing.orgBiol.Lett.12:201605772on November 23, 2016http://rsbl.royalsocietypublishing.org/Downloaded from
Page 3
tensions to those of raw pitcher fluid (48.5+3.4 dyne cm21versus 47.9+1.7 dyne cm21 (1 dyne ¼ 1Â1025 N)), and pair-wise post-hoc analyses revealed individual Darlingtonia fluidsamples were nearly indistinguishable from their bacterialcultures (figure 1a). Furthermore, these cultures containedapproximately similar numbers of bacterial cells (approximately1011 CFUs ml21) to samples collected from natural pitcher plants(109–1011 cells ml21) [10]. I encountered significantly negativeslope estimates for shear viscosity measurements in Darlingtoniafluids and their bacterial cultures, suggesting shear-thinning(non-Newtonian) properties (figure 1b). These slopes, however,were weak in magnitude and may not be biologically relevant.In prey capture experiments, all ants were retained in rawDarlingtonia pitcher fluid, regardless of its leaf of origin(figure 2a). Similarly, 97% of ants introduced into pitchers’bacterial cultures were also unable to escape. Upon contactwith either fluid, ants immediately broke the liquid’s surfacetension and were completely submerged. This property wasalso observed—in informal field trials—for small ants andvolant insects introduced into pitcher fluid. Ants wereimmobile by the end of the 10min trial but recoverednormal motor function after a period of 10–30 min followingtheir removal from the fluid. While oxygen deficit is the mostlikely cause for this behaviour, the presence of some otherstunning compound cannot be ruled out. None of the antsintroduced into pure water broke the surface tension of thewater. Those that did not exit the water remained on its sur-face and were active upon removal. Using logistic regression,I determined that the ratio of water to BACT was significan-tly positively associated with an ant’s probability of escape(b ¼ 2.04+0.40, z ¼ 5.04, p , 0.0001). The response surfaceof this regression indicates that water:culture ratios below10% were sufficient to retain the majority of ants (figure 2b).In concert, these results suggest that the natural digestivebacterial associates of D. californica may additionally benefittheir hosts by facilitating prey retention by the pitcher fluid.A molecular survey of microbial diversity in Darlingtoniafluid has previously demonstrated a high abundance of puta-tive biosurfactant-producing genera [10] (e.g. Pseudomonas[19], Pedobacter [20], Serratia [21]), though further studyis required to demonstrate their biosurfactant productionin situ. A similar retentive property was described in Nepenthesrafflesiana [22], though its fluid was found to be highlyviscoelastic—a property hypothesized to be caused byplant-secreted mucilage, rather than by bacteria. While all mem-bers of the family Sarraceniaceae possess foliar structures(e.g. downward-facing hairs) purportedly functioning to directinsects into the fluid [23], the altered physico-chemical propertiesof the fluid help to drown insects that would otherwise failto break the surface tension and escape. Although the prey cap-ture role provided by the bacterial community may not be ascritical to the host plant’s fitness as its digestive role, these resultsnonetheless highlight a potentially novel class of beneficialplant–microbe interactions worthy of continued study.Ethics. This research did not make use of any organisms coveredby UC Berkeley animal care and use regulations.Data accessibility. All data are provided in the accompanyingsupplemental material.Competing interests. The author declares no competing interests.Funding. Funding was provided by an NSF Graduate ResearchFellowship.Acknowledgements. I thank W. Sousa, M. Badger and T. Dolinajec forinstrumentation and J. Belsher-Howe (United States Forest Service,USFS) for field collection permi
Time-variant species pools shapecompetitive dynamics and biodiversity–ecosystem function relationshipsDavid W. Armitage1,21Department of Integrative Biology, University of California Berkeley, 3040 Valley Life Sciences Building,Berkeley, CA 94720-3140, USA2Department of Biological Sciences, University of Notre Dame, 100 Galvin Life Science Center, Notre Dame,IN 46556, USADWA, 0000-0002-5677-0501Biodiversity–ecosystem function (BEF) experiments routinely employcommon garden designs, drawing samples from a local biota. The commu-nities from which taxa are sampled may not, however, be at equilibrium. Totest for temporal changes in BEF relationships, I assembled the pools of aquaticbacterial strains isolated at different time points from leaves on the pitcherplant Darlingtonia californica in order to evaluate the strength, direction and dri-vers of the BEF relationship across a natural host-associated successionalgradient. I constructed experimental communities using bacterial isolatesfrom each time point and measured their respiration rates and competitiveinteractions. Communities assembled from mid-successional species poolsshowed the strongest positive relationships between community richnessand respiration rates, driven primarily by linear additivity among isolates.Diffuse competition was common among all communities but greatestwithin mid-successional isolates. These results demonstrate the dependenceof the BEF relationship on the temporal dynamics of the local species pool,implying that ecosystems may respond differently to the addition or removalof taxa at different points in time during succession.1. IntroductionThe rates at which ecosystems cycle nutrients are predicted to be set predomi-nantly by the actions of their constituent organisms [1–3]. Over the past twodecades, this conceptual unification of communities and ecosystems has beenempirically evaluated using the biodiversity–ecosystem function (BEF) frame-work [4–6]. This research commonly reports a positive covariance betweenspecies richness and community biomass production and is hypothesized to bejointly driven by community members’ differential contributions to ecosystemproperties (selection effects) and their degree of niche overlap (complementarityeffects) [7].The relative importance of these effects is in large part a function of resourcecompetition among community members [8]. Many ecosystem functions areenabled by a single guild of competitors. If taxa within a guild vary in theircontributions to ecosystem function, then turnover resulting from interspecificcompetition should result in shifting BEF relationships. Communities, however,are naturally dynamic and can experience both gradual successional turnoverand rapid state transitions [9,10]. Such turnover is predicted to result, in part,from temporal variation in species interactions—particularly competition—asnew taxa arrive and changing local conditions lead to fitness differencesamong competitors [11]. Because the strength of resource competition amongcommunity members is predicted to vary over the course of primary succession[1,12,13] and also influence the magnitude and drivers of the BEF relationship,it stands that the BEF relationship should vary along a successional gradient.Thus, a comprehensive theory linking biodiversity to ecosystem function© 2016 The Author(s) Published by the Royal Society. All rights reserved.on September 14, 2016http://rspb.royalsocietypublishing.org/Downloaded from
Page 2
must explicitly account for the effects of community turnoverthrough time [14,15].The majority of BEF experiments track the productivity ofmonocultures and polycultures assembled from taxa randomlydrawn from a natural biota or from ad hoccombinations of tract-able organisms such as algae or protists. In these experimentalcommunities, the magnitude and drivers of the BEF relation-ship are often found to change over time [16–26]. Whilethese experiments have contributed fundamental insightsinto the temporal dynamics of BEF relationships, they donot account for a dynamic species pool. In other words, thegroups of species used to seed these communities representeither a snapshot of a natural community at a particular pointin time (figure 1a) or a collection of species that may be differ-entially distributed across time such that two species addedinto a community do not necessarily co-occur under naturalsettings (figure 1b). Communities assembled from a dynamicspecies pool, however, may show different BEF relationshipsover time owing to the shifting identities and interactions ofthe constituent taxa (figure 1c).Whereas BEF experiments are most commonly conductedusing primary producers, the framework has also been suc-cessfully extended to other groups. In particular, bacterialcommunities have been the subject of numerous BEF studies,owing to both their experimental tractability and importancein regulating ecosystem processes [18,24,27–30]. Becausenatural bacterial communities often exhibit marked turnoverthrough time [31,32], they provide an opportunity to investi-gate the strength and drivers of the BEF relationship over atemporal gradient.Carnivorous pitcher plants in the family Sarraceniaceaeare a group for which bacterial communities provide a par-ticularly critical function. These plants have evolved tocapture arthropod prey by means of a conical leaf in whichtrapped insects are drowned by fluid secreted by the host[33,34]. Digestion is facilitated both by enzymes producedby the plant and by a dynamic community of bacteria resid-ing in the fluid [32,35–37]. The pitcher plant Darlingtoniacalifornica (Torr.) is hypothesized to rely heavily on bacteriafor prey digestion [35]. The pitcher leaves of this species areproduced at regular intervals throughout the June–Octobergrowing season and are sterile prior to opening [32]. Oncethe leaves fully develop, they quickly begin trapping insects,and bacterial biomass skyrockets to over 109 cells ml21 [32].After approximately two months, a leaf ceases prey capturebut remains photosynthetically active for a second growingseason. Bacterial diversity in Darlingtonia pitchers changespredictably over time, as has been documented by bothculture-independent molecular approaches as well as amongbacterial cultures isolated from different aged leaves [32,38].These temporal isolates provide a unique opportunityto measure BEF relationships along a natural microbialsuccessional gradient.My goals for this study were twofold. First, I investigatedwhether the contribution of bacterial richness to rates ofcarbon mineralization changed over time along a natural suc-cessional gradient in Darlingtonia leaves. Second, I used thesedata to estimate the relative influences of individual strainsand their interspecific interactions (such as competition) onthe BEF relationship [39]. The strength of interspecific compe-tition among bacterial strains growing in a polyculture can beapproximated as the difference between the community’spredicted respiration in the absence of any interference(i.e. the sum of the strains’ monoculture respiration rates)and the community’s realized respiration rate, given themono and polycultures have equal total starting densities[40,41]. If strains in a polyculture do not inhibit one anotherthrough resource competition or direct antagonism, then thecommunity’s rate of carbon respiration will not significan-tly differ from the additive monoculture expectation [40].This measure of competitive inhibition is anticipated toincrease over time if, for instance, a competition–colonizationecosystemprocessspecies richnessspeciesrichnessecosystemprocessspeciesrichnessecosystemprocesstime(a)(b)©Figure 1. Species pools for BEF experiments are typically chosen either by sampling a community at a single point in time (a) or from a group of taxa that may notco-occur at a particular time point (b). Far fewer studies have taken the approach of measuring BEF relationships over a temporally dynamic species pool ©.rspb.royalsocietypublishing.orgProc.R.Soc.B283:201614372on September 14, 2016http://rspb.royalsocietypublishing.org/Downloaded from
Page 3
trade-off results in the dominance of early pitcher leaves byless competitive, ruderal taxa which are later excluded bysuperior competitors [11,13]. Alternatively, the bacterial taxadominating late-stage pitchers may be specialists on recalcitrantcarbon resources and therefore may not contribute significantlyto respiration, compared with early, fast-growing colonists [42].In this case, I anticipated a negative trend in competitiveinhibition over time. In order to experimentally test thesehypotheses, I assembled synthetic microbial communities,using pools of bacterial strains isolated from a cohort ofpitcher leaves at regular intervals and measured their ratesof carbon mineralization.2. Material and methods(a) Sample collection and strain isolationIn the field, I tagged five unopened Darlingtonia pitcher leaves of thesame approximate age at the beginning of the growing season andtracked them over their first year. I visited this cohort of leaves every11 days from June to September 2014 and once in June 2015 toremove 0.5 ml of pitcher fluid from each leaf. This fluid was dilutedand spread on R2A agar plates and incubated at 258C, and bacterialcolonies expressing unique colony morphologies, cellular mor-phologies and pigmentations were isolated in pure culture. The10 most abundant bacterial strains isolated from each pitcher ageclass were then used to inoculate experimental microcosms (elec-tronic supplementary material, table S1). Extended discussion ofthe sampling and isolation methods can be found in the electronicsupplementary material accompanying this article. Electronic sup-plementary material, figure S1 provides a graphical walkthrough ofthe experimental procedure.(b) Microcosm experimentI combined the 10 strains isolated from each time point into 1-, 2-, 5-and 10-strain communities using the random partitions designintroduced by Bell et al. [39]. My experiment consisted of four par-titions §, each containing four strain richness treatments ® and10/R randomized communities within each P x R treatment (elec-tronic supplementary material, figure S2). Every experimentalcommunity was replicated three times. This experimental designensures that all species are equally represented within andamong richness levels, giving each one an equal opportunity tocontribute to selection and complementarity effects and weakensstatistical artefacts such as the ‘variance reduction effect’ [43].It also permits the statistical separation of species effects and rich-ness effects on ecosystem processes without the need formeasuring the contribution of an individual species to the proper-ties of the polyculture, as is traditional in BEF studies using plantbiomass as a response. This enables the user to estimate species’contributions to emergent ecosystem properties (e.g. carbon min-eralization rates) that cannot be attributed to individual taxain polyculture. Furthermore, it relaxes the requirement for a full-factorial experimental design, which becomes intractable as thenumber of species increases. In total, I assembled 216 communitiesper time point, resulting in a total of 1944 cultures spanning ninesource community ages and four levels of richness.The bacterial microcosms consisted of 1.2 ml 96-well platescontaining a sterile artificial pitcher medium comprised M9 saltsolution and ground cricket powder. Individual bacterial strainswere grown to mid-log-phase in R2A broth, washed of theirmedium and starved for 2 h. Each strain was introduced intoits community at the volume required to keep the total numberof cells across richness treatments equal (100 ml, or approx. 104colony-forming units). Once assembled, plates were clampedonto 96-well MicroRespTM(James Hutton Institute, Inc.)respirometry plates containing a colorimetric CO2 indicator sol-ution [44]. All replicate communities for a single time pointwere incubated simultaneously at 258C for 3 days, after whichtime I estimated rates of CO2 –C entering each agar well on theMicroRespTM plate from its absorbance at 590 nm on a micro-plate reader. I measured the carbon metabolic profiles of each10-strain community using the GN2 microplate (Biolog, Inc.),which assays a community’s potential to metabolize 95 differentcarbon compounds. Each Biolog assay was run in triplicate at 258Cfor 3 days, and only substrates scoring positive for metabolismacross all replicates were scored as positive. I used ANOVA totest for differences in the mean number of compounds usedbetween community ages and principal coordinates analysis toordinate samples’ metabolic profiles based on their Jaccard dis-tances. Additional experimental procedures are detailed in theelectronic supplementary material.© Statistical analysesTo assess how drivers of the BEF relationship differed amongtime points, I fitted a linear model to community respirationrates [39]. This model took the formy = b0 + bLRxLR + bNLRxNLR +∑Sibixi(+ bQxQ + bMxM + 1,(2.1)where y is a community or ecosystem process (e.g. respirationrate), bLR is the effect of strain richness measured on a continuousscale (linear richness, xLR), bNLR is the effect of strain richnessmeasured on a categorical scale (nonlinear richness, xNLR), bi isthe impact of an individual strain’s presence on the productivityof its community, bQ is the effect of the particular taxon poolused in each P x R treatment, bM is the effect of a particular com-munity composition within each taxon pool, b0 is the interceptand 1 is the error term.Importantly, by estimating the linear richness term prior to thenonlinear richness and strains’ impact terms, the latter two termsbecome orthogonal. The species impact (bi) terms sum to zeroand reflect the relative influence an individual strain exerts on thecommunity’s respiration. The nonlinear richness term (bNLR) canbe interpreted as the magnitude of deviations from linear richnesseffects. Non-zero values of bNLR reflect the influence of facilitativeand competitive interactions on ecosystem processes. I used least-squares to estimate the model coefficients and anF-test to determinethe statistical significance of each variable. The denominator termfor the F-statistics of the bNLR and bi parameters were the parti-tioned mean-squares from the species pool (Q) or speciescomposition (M) factors, respectively. Model terms were enteredin the order in which they appear in equation (2.1): nonlinear rich-ness (bNLR) and species impacts (bi) were estimated from theresiduals of the model containing the linear richness (bLR) term.I estimated the effects of the source pitchers’ ages and exper-imental communities’ richness on the rates of CO2 respirationusing linear regression. To aid in the interpretation of interactions,predictors were centred to their mean values prior to model fitting.I assessed the pairwise differences among community ages usingTukey’s range test (a ¼ 0.05). Community age was treated as anordinal, discrete variable to account for the absence of samplingbetween days 88 and 365.I estimated the extent to which strains inhibit one another’spotential CO2 production in polyculture by calculating the differ-ence between a community’s predicted and observed respirationrates. The predicted values were calculated by summing all com-munity members’ average monoculture respiration rates. Thedifference between a polyculture’s predicted and observed res-piration values will equal zero if there are no inhibitory effectsbetween members of the community (i.e. all taxa in a polycultureperform as well they do in monoculture) [40,41]. Alternatively,rspb.royalsocietypublishing.orgProc.R.Soc.B283:201614373on September 14, 2016http://rspb.royalsocietypublishing.org/Downloaded from
Page 4
direct antagonism (e.g. antibiotic production) or resource compe-tition is anticipated to result in respiration rates less than theadditive prediction. I used ANCOVA to test the null hypothesisthat the mean differences between predicted and observed res-piration rates were equal among community ages, controllingfor richness effects. Pairwise differences between centred predic-tor variables were assessed using Tukey’s range test. All modelswere fitted using R v. 3.1 [45].(d) Pairwise antagonism assayI performed spot assays to determine whether a particular bacterialstrain directly inhibits the growth of a co-occurring strain. I createdlawns of focal strains by spreading log-phase broth cultures ontotwo plates containing R2A agar onto which I spotted 2 ml log-phase broth culture of each co-occurring isolate. Each spot wasreplicated four times on the same plate, resulting in eight cross-inoculations per strain pair (excluding sterile blanks). After 24 hat 258C, I searched for zones of clearing surrounding a colony.I considered the spotted strain to be inhibitory to the focal strainif unambiguous zones of clearing surrounded at least six replicates.3. ResultsI found an average of 6.9 (s.e. ¼ 0.18, range ¼ 5–9) bacterialstrains remaining in each 10-strain community, and there wereno significant differences in the proportions of surviving strainsamong source community ages (F8,27 ¼ 2.3, p ¼ 0.06). Thus,although the strains’ relative abundances changed throughoutthe incubation period, no single, dominant strain was able toexclude the majority of others. I detected significant differencesbetween the mean respiration rates of bacterial communities iso-lated from pitcher leaves of different ages (table 1 and electronicsupplementary material, figure S3). Post hoc analysis revealedrespiration rates to be greatest among bacterial communities iso-lated from pitcher leaves between 22 and 55 days old (electronicsupplementary material, figure S3). This pattern was consistentunder all four richness treatments, although there was a generaltendency for variance in respiration rates among treatments toincrease when more strains were present. Bacterial richnesshad a significantly positive effect on overall respiration rates,(bR ¼ 0.05+0.007; table 1 and figure 2), although there was asignificant interaction between richness and source communityage (table 1).The effect of linear richness (bLR) on respiration rates wassignificantly positive for all source community ages exceptthose from days 88 and 365 (figure 2). This positive effectof richness on respiration was greatest for isolates frompitcher leaves between 22 and 66 days old, and tended toincrease from days 11 to 22 and then slowly decrease towardszero throughout the rest of the pitchers’ lifespan (figure 3a).For each bacterial isolate pool, I detected individual nonlinearrichness effects (bNLR) and individual strain (bi) effects sig-nificantly greater or less than zero, but these effects werenot significant overall (electronic supplementary material,table S2). Despite this, the relative influence of nonlinear rich-ness effects was greater than overall strain effects for themajority of time points (figure 3b).The average differences between expected and observedrespiration rates initially increased between samples collectedfrom 11- and 22-day pitchers, and then declined with sourcecommunity age (figure 4). There were a number of instanceswhere the observed respiration rates of two-strain mixtureswere greater than their predicted values, but overall meanvalues were significantly greater than zero for all richnesstreatments (b0,R2 ¼ 1.04+0.26, b0,R5 ¼ 7.25+0.48, b0,R10 ¼16.05+0.62, p , 0.0001 for all cases). The magnitude of thisinhibitory effect increased with strain richness (figure 4 andtable 1). I detected only 12 antagonistic interactions betweeneight pairs of strains (out of 405 total). These interactionsoccurred only in 11- and 44-day source pools. Furthermore,there was no detectable temporal trend among source poolages in either the total number of carbon substrates used(electronic supplementary material, figure S4) or theirmultivariate Jaccard similarities (electronic supplementarymaterial, figure S4).4. Discussion(a) Dynamic species pools impact biodiversity–ecosystem function relationshipsI encountered a mid-successional peak in the rates of carbonmineralization, independent of taxonomic richness. Thisimplies that when placed into identical environments,bacterial strains isolated from leaves of intermediate agesTable 1. ANOVA and ANCOVA results for total respiration and respiration differences (i.e. interspecific inhibition). (Respiration rates were log-transformed to satisfyhomoscedasticity. Richness was treated as a continuous variable and age as a categorical variable with contrasts summing to zero. Marginal (type 3) sums-of-squares(SS) are presented.)responsecovariated.f.SSFp(<)R2log respiration rateintercept117.35317.350.0010.17source community age812.8912.890.001species richness147.0347.030.001interaction term814.523.3110.001residuals19261056expected–observed respirationintercept16891540.0010.88source community age8105029.30.001species richness174811675.00.001interaction term8113331.70.001residuals2701206rspb.royalsocietypublishing.orgProc.R.Soc.B283:201614374on September 14, 2016http://rspb.royalsocietypublishing.org/Downloaded from
Page 5
(22–55 days old) were better able to mineralize carbon in thegrowth medium. This result could not be explained by differ-ences in the taxon pools’ carbon metabolic profiles. Rather,the increase in strains’ average respiration rates during thisperiod coincided with the greatest rates of prey capture bythe pitcher leaf [32,46]. It is possible that the relatively lowrespiration rates of late-stage bacterial communities reflect anadaptive strategy for living in nutrient-poor pitcher environ-ments. This is supported by the observation of lower averageribosomal RNA copy numbers—a trait correlated withgrowth rate—as succession proceeds [32,47]. However, infor-mation on all strains’ relative performances across differentnutrient concentrations would be required to experimentallyverify this hypothesis. A recent study found that both BEFeffects and competitive interactions decreased in bacterialmicrocosms over time, as highly productive taxa were outcom-peted by specialists capable of efficient use of recalcitrantresources [42]. This observation mirrors and lends support to11 days22 days33 days44 days55 days66 days77 days88 days365 days1.21.62.02.4468102345674682345623451.52.01.01.21.41.61.51.82.11 25101 25101 2510strain richnesscommunity respiration (µg CO2– C ml–1d–1)Figure 2. Relationships between strain richness and community respiration for synthetic bacterial communities assembled from pitchers of different ages. Black linesdenote significant linear richness fits for individual communities within each age group (p , 0.05). Mean values for the response variables are presented for clarity.Bars denote standard error measurements.0.6101mean squaresbbbababaa*aa0.40.201122334455667788 365source community age (days)1122334455667788 365source community age (days)slope estimate (b LR± CI)nonlinear richnessspecies impactFigure 3. (a) Linear richness (bLR) regression coefficients as a function of source community age. Bars denote 95% confidence intervals and shared letters betweenages signify an overlap between the two estimates. Asterisks denote coefficients found to be significantly greater than zero (F-test, p , 0.05). (b) Log mean-squareestimates for the species impact (bi) and nonlinear richness (bNLR) parameters. These values represent the relative contributions of species-specific effects andspecies interactions, respectively, on respiration rates. None of these coefficients were significantly greater than zero.rspb.royalsocietypublishing.orgProc.R.Soc.B283:201614375on September 14, 2016http://rspb.royalsocietypublishing.org/Downloaded from
Page 6
my results, and suggests that the community dynamicsobserved in closed microcosms may approximate those frommore natural systems.The effects of a community’s richness on respiration rateswere generally positive, but varied over time such that theslope estimates peaked in pitcher leaves of intermediate age.These positive BEF relationships appeared to be driven bylinear, additive contributions of taxa, as evidenced by strongpositive linear richness terms, but weak nonlinear richnessand species impact terms. This observation implies that, onaverage, community members had similar relative respirationrates and low levels of niche overlap. This interpretation issupported by the lack of dominance by any one or morestrains in 10-strain communities, which would be predictedto lead to significant species impact terms. Inhibition of a com-munity’s potential additive respiration rates was common in allpolycultures and peaked in communities assembled fromintermediate-aged pitcher leaves. This observation, combinedwith an absence of direct antagonistic interactions, providesevidence for diffuse competition limiting a strain’s potentialrespiration in polyculture. Although I failed to detect signifi-cant negative nonlinear richness terms indicative of strongcompetition, I did encounter an increase in the effect ofnonlinear richness coinciding with the periods of highest respir-ation inhibition. This general pattern of diffuse competition inpolycultures is commonly found in bacterial microcosm exper-iments [40,48] but may not be typical of bacteria within pitcherplants owing to my isolation procedure. By using a singlemedium to isolate bacteria, it is likely that the strains I sampledwere more phenotypically similar to one another than to arandom sample of all bacteria in a pitcher leaf. Thus, the strainsused in this study should be considered members sampled froma guild of aerobic, heterotrophic bacteria and are expectedto compete with one another for resources and express similarrates of carbon respiration. However, this is no differentfrom most plant and microbial BEF studies, which commonlydraw inference at the guild level. A useful follow-up to thisexperiment would investigate the effects of increasing the phe-notypic diversity of the taxon pool by adding strains obtainedusing a broader range of media.Competition among isolates is predicted to decrease in bac-terial communities over time owing to divergent evolution andcan lead to changes in ecosystem functioning [49 – 51]. The rela-tively low levels of competitive inhibition among strains fromlate-stage pitcher leaves may represent indirect evidence ofdivergence. This scenario is plausible, given the rapid gener-ation times and population sizes of the isolates. A recentstudy by Fiegna et al. [48] showed that the experimental evol-ution of bacterial isolates over five weeks can alter the BEFrelationship via a relaxation of competition. Although suchan effect is possible in natural systems, its demonstrationwould require tracking individual bacterial lineages overtime and regularly assaying their competitive interactions.Miller & Kneitel [52] attempted this by measuring the degreeof competitive inhibition of four bacterial colony morphotypesisolated from the same pitcher leaves 7 and 42 days after open-ing. The authors found that the competitive abilities (relative toa common bacterial competitor) of two of the four strainsdecreased with pitcher age while two did not appear tochange [52]. These results match my observation of increasedcompetitive inhibition of potential respiration on a similartimescale (11- and 44-day leaves).4030201001122334455667788365source community age (days)predicted−observed respiration raterichness2510abbaabaaaababacabbabaaaaabecddbcdaaacacFigure 4. Relative inhibition of bacterial respiration in polycultures measured as the difference between additive predicted and observed rates. Values of zeroindicate that the sum of community members’ respirations in monoculture equalled the community’s performance in polyculture. Values greater than zero indicatethat observed rates were less than predicted rates and provide evidence for interspecific competitive or antagonistic inhibition. The y-axis has been reversed to moreclearly illustrate this inhibition. Letters shared by points within a richness group indicate that their means (white points) do not significantly differ from one another(Tukey’s range test, p , 0.05). Shading denotes richness treatments.rspb.royalsocietypublishing.orgProc.R.Soc.B283:201614376on September 14, 2016http://rspb.royalsocietypublishing.org/Downloaded from
Page 7
(b) Potential drivers of biodiversity–ecosystem functionrelationshipsTo date, few studies have directly estimated the impacts of natu-ral successional dynamics in the context of BEF [26,53,54].Using 15 years of observational data from regenerating tropicalforest plots, Lasky et al. [26] documented a decreasing effect ofspecies richness on rates of above-ground biomass productionin mid- and late-successional tropical forest plots. These resultsmatched both theoretical predictions [14] and experimentalstudies in which diversity effects were tracked over timewithin individual microcosms without immigration [18,22].My results conform to those of other BEF time-series exper-iments, despite marked differences in design. In concert, thesefindings challenge the common observation that the effects ofrichness on productivity become more positive over time [21],though further investigation is necessary to uncover themechanisms leading to these contrasting outcomes.One mechanism for generating non-positive BEF relation-ships is the negative selection effect [7,27]. This phenomenonoccurs when the competitively dominant taxa in a communityare those that contribute least to the measured ecosystem func-tion. Three lines of evidence from my experiments suggestthat the negative selection effect does not occur in late-stagesource communities. First, I did not detect any trends towardsincreasing rates of competitive exclusions in late-stage sourcecommunities. Second, these communities had some of thesmallest nonlinear richness (i.e. species interaction) terms andextents of inhibition. These lines of evidence signify a low contri-bution of negative species interactions to the diminishedrespiration in late-stage pitchers [39]. Further study, however,is needed to determine: (i) whether observed successionaldecreases in competition result from decreasing niche overlapwithin late-stage communities; and (ii) the relative influenceof competition versus habitat filtering during different stagesof ecosystem development and how these factors, actinghistorically, contribute to contemporary community structure.5. ConclusionAll previous experimental studies measuring the BEF relation-ship over time do so using communities with finite resourcesand no immigration. Consequently, the closed nature ofthese systems may have influenced the resulting commu-nity dynamics and ecosystem processes. My study, however,measured individual ‘snapshots’ of communities assembledfrom a temporal gradient of natural, open source pools. Further-more, my microcosms were assembled with equal startingconcentrations of bacterial strains and resources, which mayhave prevented communities from becoming resource limitedprior to measuring their respirations. Despite these differences,however, decreases in microbial BEF relationships of bothstatic species pools over time and dynamic species pools at asingle time point suggest that similar ecological processesmay govern these patterns in microbial communities.In leaves of the pitcher plant D. californica, bacterial degra-dation of organic matter is a process critical for the uptake ofprey-derived nitrogen and phosphorous in the nutrient-poorhabitats to which these plants are adapted. Using bacterialstrains isolated from pitcher leaves at regular intervals over a1 year period, I determined the magnitude of the BEF relation-ship to peak in mid-successional communities. This positiverichness effect on respiration was driven primarily by strains’relatively equivalent contributions to ecosystem function.At the same time, respiration was constrained by diffuse com-petition among strains in polyculture. This study represents aninitial attempt to integrate BEF effects over successional timeand concludes that the functional consequences of diversityloss on a host or ecosystem may vary along a successionalgradient. Future studies on BEF relationships are encouragedto adopt a dynamic species pool framework to improve thegeneralizability of their results.Data accessibility. The datasets supporting this article have beenuploaded as a part of the electronic supplementary material.Authors’ contributions. D.W.A. conceived of the study, collected all data,performed the analyses and drafted the manuscript.Competing interests. I declare I have no competing interests.Funding. Funding was provided by NSF DEB-1406524 and an NSFgraduate research fellowship.Acknowledgements. I thank H. Miller, A. Petrosky and R. Leon for assist-ance with data collection and E. Simms for providing keyequipment. W. Sousa, M. Power, M. Firestone and J. Lefcheck providedhelpful comments. Permits for field collection were provided by J.Belsher-Howe (USFS).References1. Odum EP. 1969 The strategy of ecosystemdevelopment. Science 164, 262–270. (doi:10.1126/science.164.3877.262)2. DeAngelis DL. 1992 Dynamics of nutrient cycling andfood webs. London, NY: Chapman & Hall.3. Loreau M. 2010 From Populations to ecosystems:theoretical foundations for a new ecologicalsynthesis. Princeton, NJ: Princeton University Press.4. Loreau M et al. 2001 Biodiversity and ecosystemfunctioning: current knowledge and futurechallenges. Science 294, 804–808. (doi:10.1126/science.1064088)5. Hooper DU et al. 2005 Effects of biodiversity onecosystem functioning: a consensus of currentknowledge. Ecol. Monogr. 75, 3–35. (doi:10.1890/04-0922)6. Cardinale BJ, Srivastava DS, Emmett Duffy J, WrightJP, Downing AL, Sankaran M, Jouseau C. 2006Effects of biodiversity on the functioning of trophicgroups and ecosystems. Nature 443, 989–992.(doi:10.1038/nature05202)7. Loreau M, Hector A. 2001 Partitioning selection andcomplementarity in biodiversity experiments. Nature412, 72–76. (doi:10.1038/35083573)8. Turnbull LA, Levine JM, Loreau M, Hector A. 2013Coexistence, niches and biodiversity effects onecosystem functioning. Ecol. Lett. 16, 116–127.(doi:10.1111/ele.12056)9. Connell JH, Sousa WP. 1983 On the evidenceneeded to judge ecological stability orpersistence. Am. Nat. 121, 789–824. (doi:10.1086/284105)10. Scheffer M. 2009 Critical transitions in natureand society. Princeton, NJ: Princeton UniversityPress.11. Connell JH, Slatyer RO. 1977 Mechanisms ofsuccession in natural communities and their rolein community stability and organization. Am. Nat.111, 1119–1144. (doi:10.1086/283241)12. Tilman D. 1988 Plant strategies and the dynamicsand structure of plant communities. Princeton, NJ:Princeton University Press.13. Grime JP. 1979 Plant strategies and vegetationprocesses, 1st edn. Chichester, UK: Wiley.14. Kinzig AP, Pacala SW. 2001 Successional biodiversityand ecosystem functioning. In The functionalconsequences of biodiversity: empirical progress andtheoretical extensions (eds AP Kinzig, SW Pacala,rspb.royalsocietypublishing.orgProc.R.Soc.B283:201614377on September 14, 2016http://rspb.royalsocietypublishing.org/Downloaded from
Page 8
GD Tilman), pp. 175–212. Princeton, NJ: PrincetonUniversity Press.15. Mouquet N, Moore JL, Loreau M. 2002 Plant speciesrichness and community productivity: why themechanism that promotes coexistence matters. Ecol.Lett.5,56–65.(doi:10.1046/j.1461-0248.2002.00281.x)16. Tilman D, Reich PB, Knops J, Wedin D, Mielke T,Lehman C. 2001 Diversity and productivity in along-term grassland experiment. Science 294,843–845. (doi:10.1126/science.1060391)17. Fox JW. 2004 Effects of algal and herbivore diversityon the partitioning of biomass within and amongtrophic levels. Ecology 85, 549–559. (doi:10.1890/03-0095)18. Bell T, Newman JA, Silverman BW, Turner SL, LilleyAK. 2005 The contribution of species richness andcomposition to bacterial services. Nature 436,1157–1160. (doi:10.1038/nature03891)19. Ruijven J, Berendse F. 2005 Diversity–productivityrelationships: initial effects, long-term patterns, andunderlying mechanisms. Proc. Natl Acad. Sci. USA102, 695–700. (doi:10.1073/pnas.0407524102)20. Spehn EM et al. 2005 Ecosystem effects ofbiodiversity manipulations in European grasslands.Ecol. Monogr. 75, 37–63. (doi:10.1890/03-4101)21. Cardinale BJ, Wright JP, Cadotte MW, Carroll IT,Hector A, Srivastava DS, Loreau M, Weis JJ. 2007Impacts of plant diversity on biomass productionincrease through time because of speciescomplementarity. Proc. Natl Acad. Sci. USA 104,18 123–18 128. (doi:10.1073/pnas.0709069104)22. Weis JJ, Cardinale BJ, Forshay KJ, Ives AR. 2007Effects of species diversity on community biomassproduction change over the course of succession.Ecology 88, 929–939. (doi:10.1890/06-0943)23. Doherty JM, Callaway JC, Zedler JB. 2011 Diversity–function relationships changed in a long-termrestoration experiment. Ecol. Appl. 21, 2143–2155.(doi:10.1890/10-1534.1)24. Gravel D, Bell T, Barbera C, Bouvier T, Pommier T,Venail P, Mouquet N. 2011 Experimental nicheevolution alters the strength of the diversity–productivity relationship. Nature 469, 89–92.(doi:10.1038/nature09592)25. Reich PB, Tilman D, Isbell F, Mueller K, Hobbie SE,Flynn DFB, Eisenhauer N. 2012 Impacts ofbiodiversity loss escalate through time asredundancy fades. Science 336, 589–592. (doi:10.1126/science.1217909)26. Lasky JR, Uriarte M, Boukili VK, Erickson DL, JohnKress W, Chazdon RL. 2014 The relationshipbetween tree biodiversity and biomass dynamicschanges with tropical forest succession. Ecol. Lett.17, 1158–1167. (doi:10.1111/ele.12322)27. Jiang L. 2007 Negative selection effects suppressrelationships between bacterial diversity andecosystem functioning. Ecology 88, 1075–1085.(doi:10.1890/06-1556)28. Salles JF, Poly F, Schmid B, Le Roux X. 2009Community niche predicts the functioning ofdenitrifying bacterial assemblages. Ecology 90,3324–3332. (doi:10.1890/09-0188.1)29. Langenheder S, Bulling MT, Sloan M. 2010Bacterial biodiversity–ecosystem functioningrelations are modified by environmental complexity.PLoS ONE 5, e10834. (doi:10.1371/journal.pone.0010834)30. Venail PA, Vives MJ. 2013 Phylogenetic distance andspecies richness interactively affect the productivityof bacterial communities. Ecology 94, 2529–2536.(doi:10.1890/12-2002.1)31. Redford AJ, Fierer N. 2009 Bacterial succession onthe leaf surface: a novel system for studyingsuccessional dynamics. Microb. Ecol. 58, 189–198.(doi:10.1007/s00248-009-9495-y)32. Armitage DW. 2016 Microbes in time: incorporatingbacteria into ecosystem development theory. PhDdissertation, University of California Berkeley,Berkeley, CA, USA.33. Lloyd FE. 1942 The carnivorous plants. In ChronicaBotanica. New York, NY: Ronald Press.34. Juniper BBE, Robins RJ, Joel DM. 1989 Thecarnivorous plants. San Diego, CA: Academic Press.35. Hepburn JS, Jones FM, John EQ. 1927 Thebiochemistry of the American pitcher plants:biochemical studies of the North AmericanSarraceniaceae. Trans. Wagner Free Inst. Sci. Phila.1927, 1–95.36. Lindquist JA. 1975 Bacteriological and ecologicalobservations on the northern pitcher plant,Sarracenia purpurea L. Masters thesis, University ofWisconsin, Madison, WI, USA.37. Butler JL, Gotelli NJ, Ellison AM. 2008 Linking thebrown and green: nutrient transformation and fatein the Sarracenia microecosystem. Ecology 89,898–904. (doi:10.1890/07-1314.1)38. Petrosky A. 2015 Local adaptation of bacteria andbacteriophage in California pitcher plants(Darlingtonia californica). Honor’s thesis, Universityof California Berkeley, Berkeley, CA, USA.39. Bell T, Lilley AK, Hector A, Schmid B, King L,Newman JA. 2009 A linear model methodfor biodiversity–ecosystem functioningexperiments. Am. Nat. 174, 836–849. (doi:10.1086/647931)40. Foster KR, Bell T. 2012 Competition, notcooperation, dominates interactions amongculturable microbial species. Curr. Biol. 22,1845–1850. (doi:10.1016/j.cub.2012.08.005)41. Keddy PA. 2012 Competition. New York, NY:Springer Science & Business Media.42. Rivett DW, Scheuerl T, Culbert CT, Mombrikotb SB,Johnstone E, Barraclough TG, Bell T. 2016 Resource-dependent attenuation of species interactionsduring bacterial succession. ISME J. 9,1235–1245.(doi:10.1038/ismej.2016.11)43. Huston MA. 1997 Hidden treatments in ecologicalexperiments: re-evaluating the ecosystem functionof biodiversity. Oecologia 110, 449–460. (doi:10.1007/s004420050180)44. Campbell CD, Chapman SJ, Cameron CM, DavidsonMS, Potts JM. 2003 A rapid microtiter plate methodto measure carbon dioxide evolved from carbonsubstrate amendments so as to determine thephysiological profiles of soil microbial communitiesby using whole soil. Appl. Environ. Microbiol.69, 3593–3599. (doi:10.1128/AEM.69.6.3593-3599.2003)45. R Development Core Team. 2015 R: a language andenvironment for statistical computing. Vienna,Austria: R Foundation for Statistical Computing.46. Fish D, Hall DW. 1978 Succession and stratificationof aquatic insects inhabiting the leaves of theinsectivorous pitcher plant, Sarracenia purpurea. Am.Midl. Nat. 99, 172–183. (doi:10.2307/2424941)47. Nemergut DR et al. 2015 Decreases in averagebacterial community rRNA operon copy numberduring succession. ISME J. 5, 1147–1153.(doi:10.1038/ismej.2015.191)48. Fiegna F, Moreno-Letelier A, Bell T, Barraclough TG.2015 Evolution of species interactions determinesmicrobial community productivity in newenvironments. ISME J. 9, 1235–1245. (doi:10.1038/ismej.2014.215)49. Lawrence D, Fiegna F, Behrends V, Bundy JG,Phillimore AB, Bell T, Barraclough TG. 2012 Speciesinteractions alter evolutionary responses to a novelenvironment. PLoS Biol. 10, e1001330. (doi:10.1371/journal.pbio.1001330)50. Harmon LJ, Matthews B, Des Roches S, Chase JM,Shurin JB, Schluter D. 2009 Evolutionarydiversification in stickleback affects ecosystemfunctioning. Nature 458, 1167–1170. (doi:10.1038/nature07974)51. Zuppinger-Dingley D, Schmid B, Petermann JS,Yadav V, De Deyn GB, Flynn DFB. 2014 Selection forniche differentiation in plant communities increasesbiodiversity effects. Nature 515, 108–111. (doi:10.1038/nature13869)52. Ellison AM, Gotelli NJ, Brewer JS, Cochran-Stafira DL,Kneitel JM, Miller TE, Worley AC, Zamora R. 2003The evolutionary ecology of carnivorous plants. Adv.Ecol. Res. 33, 1–74. (doi:10.1016/S0065-2504(03)33009-0)53. Wacker L, Baudois O, Eichenberger-Glinz S, SchmidB. 2009 Diversity effects in early- and mid-successional species pools along a nitrogengradient. Ecology 90, 637–648. (doi:10.1890/07-1946.1)54. Petermann JS, Fergus AJF, Roscher C, Turnbull LA,Weigelt A, Schmid B. 2010 Biology, chance, orhistory? The predictable reassembly of temperategrassland communities. Ecolo
Supporting Information. David W. Armitage, Stuart E. Jones. 2019. Negativefrequency-dependent growth underlies the stable coexistence of twocosmopolitan aquatic plants. Ecology.APPENDIX S1S1.1 Formulation of Growth MediumThis medium was originally formulated by Docauer (Docauer 1983) to approximate thebuffering system (bicarbonate), pH (6.5), conductivity (440 µS/cm), and nutrient contents ofnatural waters containing Lemna minor and Spirodela polyrhiza. Below, we reproduceDocauer’s original formulation from his dissertation.CompoundStockconcentrationStock perliter mediaFinal mediaconcentrationg/LmLmg/LmMNa2EDTA20.002.00.107NaNO325.002.08.2 N0.586K2HPO43.681.01.3 P0.042KCl50.001.026.0 K0.671CaCl225.002.018.0 Ca0.450MgSO4 • 7H2O37.500.513.5 Mg0.555MicronutrientsSee below1.0See belowNaHCO360.001.048.0 HCO30.785Micronutrient stock:CompoundStockconcentrationFinal mediaconcentrationg/100 mLmg/LmMFeSO4 • 7H2O0.9952.0 Fe0.036MnCl2 • 4H2O0.0720.2 Mn 0.004Na2MoO4 • 2H2O0.0440.1 Mo 0.001H3BO30.0570.1 B0.009ZnSO4 • 7H2O0.0440.1 Zn0.002Na2EDTA2.0000.054Mix the following separately and add 1 mL per 100 mL micronutrient stock:CuSO4 • 5H2O0.0040.0001 Cu 0.0000016CoCl2 • 6H2O0.4000.01 Co 0.00017Na2EDTA0.4000.00010Slowly add NaHCO3 while bubbling to desired pH; 1.0 mL gives pH 6.5.We used MilliQ ultrapure water as the base of our media. Stock solutions are added in theorder they appear to avoid precipitation. The sodium bicarbonate is added last. At this point, themedium is supersaturated with CO2 and is therefore more acidic than it would be when atequilibrium with air. To remedy this, the solution is vigorously bubbled with air for 30 minutes
Page 2
2until the pH stabilizes at 6.5. The solution is then autoclaved and can be used for growthexperiments.S1.2 Specifying the Inter- and Intraspecific Competition FunctionTo choose the appropriate functional form of our population dynamic models, we needed toappropriately specify a species’ growth responses to the densities of conspecifics andheterospecifics, and the interaction of these responses with the environment. For notationalsimplicity, we remove the time (t) subscripts but note that all states and parameters savemortality rates can be time-variant. For our duckweed model with vegetative (Nj) and dormantturion (Sj) stages and variable temperatures (T), we can model the total population growth ratesby summing the turion and vegetative sub-populations:where ( , ) describes the strength of intraspecific (when j=k) and interspecific (when j≠k)competition. We compared related forms for ( , ): the standard Lotka-Volterracompetition model (MacArthur 1970), and versions modified with log-transformed abundances(Turchin 2003). Likewise, a subset of models allowed for covariation between ambienttemperature and the interaction parameters. For two species, Spirodela and Lemna, our modelstook the general form:which, when combined with eq. S.1 and rearranged, yields the linear regression equation[ | , , ] = 0 + + ( + )log( + 1)+ ( + ) log( + 1) + .(S3)Here, the regression coefficients (β’s) relate to the coefficients of eqs. 5 and S.2 as follows:0 + = ( ( ) − ),+ = ( ) ( ),(S4)+ = ( ) ( ),and=(+)+=[ j( ) ( , ) − ]+, ( = 1,2)(S1)( , ) = 1 − ( ) − ( ) , ( ≠ )(S2)= ( + ),⁄( ) = ( ̅)+ ,(S5)( ) = ( ̅)+ .
Page 3
3We assume that the scaling parameter, , is equal to one for both species. Thissimplification did not significantly change our parameter estimates, as these values were veryclose to 1 for populations during most census periods. We then used our empirical estimates for( ) and to solve for the model parameters. For convenience, thermal responses ofcompetition parameters were assumed to be linear functions of temperature where ( ̅) arecompetition coefficients at average ambient temperatures (here, 20°C) and are thetemperature-dependent slope parameters in ( ). While this assumption has some empiricalsupport (reviewed in Amarasekare and Coutinho 2014), we note that our sampling design did notpermit the recovery of nonlinear or non-monotonic response surfaces. Regression models were fitto data from the fluctuating environment experiment using least squares (lm) in R (ver. 3.3).Best-fitting models were selected based on coefficients of determination (R2). We ultimatelyselected model 4, with log-transformed populations and temperature-dependent competition forour species interaction term. Parameters and fit statistics for each model are presented in TableS2.S1.3 Quantifying the temporal storage effectThe temporal storage effect (Chesson 1994) is one of two potential fluctuation-dependentcoexistence mechanisms. We used the Monte Carlo-based approach introduced by Ellner et al.(2016) to quantify the contribution of the temporal storage effect to invaders’ per capita growthrates in fluctuating environments. This technique begins by defining a function that describesspecies’ growth rates in terms of an environmentally-dependent parameter, Ej, and competitionparameter, Cj. Many formulations for ( , ) are possible, but here we use= ( , ) = ( − ),= (1,2),(S6)where= ( ), = 1 − ( ) − ( ) ,≠ .(S7)Using these equations, we generated environmental sequences for ( ) across a range ofdifferent average temperatures and amplitudes by simulating the dynamics of a resident speciesin monoculture and saving its model parameters and states at each time step once it had reachedequilibrium with its environment (i.e., ̅ \ = 0). These values were then used to calculate bothspecies’ long-term average resident and invasion growth rates, ̅ \ and ̅ \ , respectively,where ̅ = [ ( )]. Because a storage effect requires nonzero covariance between ( ) and( ), its contribution to an invader’s growth rate (and therefore coexistence) is proportional tothe contribution of cov( , ) to the invader’s growth rate, ̅ \ . To remove the signature ofcov( , ) from ̅ \ , we generated a second vector of environmental parameters,#( ), byrandomly subsampling without replacement from the original ( ) vectors. This step makes#and independent of one another, and allows us to estimate ̅# = [#( )]. With thesevalues, we estimated the contribution of the storage effect (∆ \ ) to the growth rate of invadingspecies ( ̅ \ ) using the equation
Page 4
4∆ \ = ( ̅ \ − ̅ \ ) − ( ̅ # − ̅ # ),(S8)where are species-specific scaling factors relating to the relative sensitivity to competitionexperienced by resident species compared to the invading species (Chesson 1994). These scalingfactors can be estimated by defining new environmental and competition parameters E and C :E = ( ,∗), C = − (∗, ),(S9)where∗and∗equal their baseline (i.e., mean or median) values. Thus, for our model, bydefining∗ = ( )̅̅̅̅̅̅̅, we arrive atC = ( − ( )̅̅̅̅̅̅̅[1 − ( ) log( + 1) − ( ) log( + 1)]).(S10)Finally, we used a regression approach outlined in Ellner et al. (2016, Appendix S1) toestimate the scaling parameters. For two species, Chesson’s (1994, 2000) definition of theseparameters is given by the equation=C \C .(S11)This partial derivative can be estimated by evaluating the slope of a nonlinear regression(here, a smoothing spline) of C \ ( ) on C \ ( ) at C \ = 0.We used this approach to identify thermal regimes where the temporal storage effectrescues a species from competitive exclusion by satisfying the inequality0 < ̅ \ < ∆ \ .(S12)We also identified regions where the storage effect was overall positive but not greaterthan an invader’s growth rate. In this situation, the storage effect positively contributes tocoexistence, but cannot be considered the sole operating coexistence mechanism.
Page 5
5Supplementary TablesTable S1. Model selection results for duckweed growth rates in monoculture and competition.Candidate models are listed for each experiment. Bolded models denote best fits based onparsimony, BIC, and R2. Temperatures and frequencies were centered at their median valuesprior to fitting.EnvironmentModel formBIC∆BICR2Static~ 1 ( − )-6903190temperatures~ species (S) (Spirodela = 1)-690319 0.02(monoculture)~ S + temperature (T) + temperature2-93870 0.60n = 72~ + + + ( )-10090 0.70Static~ 1 ( − )-6971420temperatures~ species (S) (Spirodela = 1)-694145 0.01(competition)~ S + temperature (T) + temperature2-8309 0.63n = 40~ + + + ( )-8400 0.67~ S + T + T2 + F + (S × F)-8373 0.67Fluctuating~ 1 ( − )-22967970temperatures~ species (S) (Spirodela = 1)-2295798 0.01(competition)~ S + census period ©-303954 0.80n = 55~ + + ( )-30912 0.83~ S + C + F + turion replace (TR) (Yes=1)-30930 0.83~ S + C + F + TR + (S × F)-30876 0.83~ S + C + F + TR + (S × TR)-30911 0.83~ S + C + F + TR + (S × F) + (S × TR)-30857 0.83Models with ∆BIC ≤ 4 were considered well-supported, and from these, we favored the model with the smallest number of parameters.
Page 6
6Table S2. Mean parameter estimates and fit statistics for our four competition models where theresponse variable is a species’ total growth rate (rj). The parameter ( ̅) and ( ̅) are theestimates for intra- and interspecific competition parameters at average temperatures (20 °C),respectively. The phi (φ ) parameters indicate the predicted change in competition ( ’s) withan increase or decrease of 1 °C from the average.Model formSpeciesα ( ̅)α ( ̅)φφR21( ) [1 − ∑2=1] −SpirodelaLemna7.27 × 10-55.15 × 10-53.99 × 10-55.83 × 10-5n/an/an/an/a0.550.612( ) [1 − ∑ log( + 1)2=1] −SpirodelaLemna1.14 × 10-11.07 × 10-16.38 × 10-26.16 × 10-2n/an/an/an/a0.730.763( ) [1 − ∑ ( )2=1] −SpirodelaLemna1.23 × 10-38.23 × 10-46.86 × 10-48.86 × 10-44.80 × 10-53.01 × 10-52.08 × 10-53.50 × 10-50.560.624( )[1 − ∑ ( ) log( + 1)2=1] −SpirodelaLemna1.07 × 10-11.01 × 10-16.46 × 10-25.91 × 10-26.22 × 10-36.34 × 10-33.09 × 10-33.04 × 10-30.770.80
Page 7
7Table S3. Parameter values for duckweed competition model. Unless otherwise noted, values areempirically derived from monoculture and competition experiments.Parameter DescriptionSpecies (j)Value (± 95% CI)Tmax,jSpecies’ maximumtemperature for growth (eq. 2)SpirodelaLemna38.93 (± 0.86)36.82 (± 1.43)Tmin, jSpecies’ minimum temperaturefor growth (eq. 2)SpirodelaLemna2.56 (± 4.53)0.00 (± 4.02)cjScaling constant for thermalgrowth model (eq. 2)SpirodelaLemna1.60 × 10-5 (± 4.33 × 10-6)2.18 × 10-5 (± 6.27 × 10-6)Td,jTemperature at which 50% ofgrowth is devoted to turions(eq. 3)SpirodelaLemna15 °Cn/aTg,jTemperature at which 50% ofturions germinate at 20 days(eq. 4)SpirodelaLemna25 °Cn/aα ( ̅)Intraspecific competitionparameter (at 20 °C) (eq. S.5)SpirodelaLemna0.1069 (± 0.016)0.1005 (± 0.015)α ( ̅)Interspecific competitionparameter (at 20 °C) (eq. S.5)SpirodelaLemna0.0646 (± 0.009)0.0591 (± 0.010)Effect of ±1°C temperaturechange on αjj (eq. S.5)SpirodelaLemna6.22 × 10-3 (± 1.93 × 10-3)6.34 × 10-3 (± 1.73 × 10-3)Effect of ±1°C temperaturechange on αjk (eq. S.5)SpirodelaLemna3.09 × 10-3 (± 1.12 × 10-3)3.04 × 10-3 (± 1.10 × 10-3)mjSpecies’ average per capitamortality rate (eq. 5)SpirodelaLemna0.0134 (± 7.46 × 10-4)0.0107 (± 6.86 × 10-4)* Values from Docauer (1983).
Page 8
8Supplementary FiguresFigure S1. (a) Geographic ranges of Lemna minor and Spirodela polyrhiza in theAmericas show broad, but not complete overlap (note: both species occur in Eurasia andAfrica, as well). (b-f) Multiple independent lake and pond surveys show that L. minor isfrequently encountered in the absence of S. polyrhiza, but S. polyrhiza rarely occurs inthe absence of L. minor. Data sources are as follows: Minnesota, Muthukrishnan Ranjanet al. (2018); Florida, Alahuhta et al. (2017); Connecticut, McCann (2015); and SEMichigan, Docauer (1983). Range maps were obtained from the BIEN 3.0 database(Enquist et al. 2016).
Page 9
9Figure S2. Population dynamics for duckweed cultures in fluctuating temperatureexperiment. Graph is faceted to more clearly display individual cultures.Figure S3. Low-abundance growth responses of each duckweed species when grown inmedia pre-conditioned by either conspecifics, heterospecifics, or a mixture of both. Withthe exception of Lemna growing in media conditioned by Spirodela, there is nosignificant effect of pre-conditioning on species’ growth rates.
Page 10
10Figure S4. Plot of observed versus model-predicted growth rates (from Table S2) foreach species. Perfect 1:1 relationship is denoted by the dashed black line.Figure S5. Plot of coexistence outcomes in static temperatures as a function of nicheoverlap, 1 − ( ) (Eq. 6), and competitive advantage ratio 2/ 1 = ( 2/1)√ 11 12/ 22 21, where temperature indices are dropped for notational simplicity(for details, see Barabás et al. 2018). This ratio compares the performance of both speciesin the absence of coexistence-promoting mechanisms. Coexistence occurs when 1 −( ) < 2( )/ 1( ) < [1 − ( )]−1, as illustrated by the shaded region. Points denotewhere on this plane Spirodela-Lemna communities fall at particular static temperatures,illustrating how coexistence critically depends on ambient temperature.
Page 11
11Figure S6. Contributions of the temporal storage effect to species’ growth rates in the presenceor absence of different interspecific differences. Comparing these values to species’ invasiongrowth rates (Fig. 4) suggests that, overall, the storage effect’s contributions to invader growthrates are minimal. However, removal of turion production weakens the contribution of thestorage effect to the growth of Lemna. Control = standard conditions with all species-leveldifferences (Eq. 5); –Tdiff = species’ thermal growth differences removed; – turion = turionproduction by Spirodela removed; – NFD = differences between inter- and intraspecificcompetition removed and replaced with average values.Figure S7. (a-d) Contributions of the temporal storage effect to species’ growth rates in thepresence or absence of different interspecific differences. Comparing these values to species’invasion growth rates (e-h) (censored to cases where rj\i > 0) suggests that, overall, the storageeffect’s contributions to invader growth rates are minimal. Control = standard conditions with allspecies-level differences (Eq. 5); –Tdiff = species’ thermal growth differences removed; – turion= turion production by Spirodela removed; – NFD = differences between inter- and intraspecificcompetition removed and replaced with average values. Here, removal of turion productionweakens the contribution of the storage effect to the growth of Lemna.
Page 12
12Figure S8. Illustration of covariance between environmental response and the strength ofcompetition for (a) Spirodela polyrhiza and (b) Lemna minor. Overall, these valuesremain close to zero for both species in most environments. More positive cov(E,C)values will increase the strength of the storage effect. This is consistent with ourobservation that the largest storage effects for both species occurred at low temperaturesand intermediate-to-high amplitudes.Figure S9. Illustration of subadditivity between environmental that response andcompetition for (a) Spirodela polyrhiza and (b) Lemna minor. Lines represent valuesobtained from oscillating equilibria in the standard interspecific competition model(Eq. 5). Models were fit to static environments with both species at starting densities of1. Results are robust to the addition of temperature fluctuations. Note that the X-axis isreversed to illustrate that the impact of competition increases from left to right. Theseresults imply that the impacts of competition on species’ growth rates decrease astemperatures move away from species’ optima.
Page 13
13Figure S10. Histogram depicting the correlation coefficients of species’ environmentalparameters, Ej, simulated under different average temperatures and amplitudes.Literature CitedAlahuhta, J., S. Kosten, M. Akasaka, D. Auderset, M. M. Azzella, R. Bolpagni, C. P. Bove, P. A.Chambers, E. Chappuis, J. Clayton, M. Winton, F. Ecke, E. Gacia, G. Gecheva, P.Grillas, J. Hauxwell, S. Hellsten, J. Hjort, M. V. Hoyer, C. Ilg, A. Kolada, M. Kuoppala,T. Lauridsen, E. H. Li, B. A. Lukács, M. Mjelde, A. Mikulyuk, R. P. Mormul, J.Nishihiro, B. Oertli, L. Rhazi, M. Rhazi, L. Sass, C. Schranz, M. Søndergaard, T.Yamanouchi, Q. Yu, H. Wang, N. Willby, X. K. Zhang, and J. Heino. 2017. Globalvariation in the beta diversity of lake macrophytes is driven by environmentalheterogeneity rather than latitude. Journal of Biogeography 44:1758–1769.Amarasekare, P., and R. M. Coutinho. 2014. Effects of temperature on intraspecific competitionin ectotherms. The American Naturalist 184:E50–E65.Barabás, G., R. D’Andrea, and S. M. Stump. 2018. Chesson’s coexistence theory. EcologicalMonographs 88:277–303.Chesson, P. 1994. Multispecies competition in variable environments. Theoretical PopulationBiology 45:227–276.Chesson, P. 2000. General theory of competitive coexistence in spatially-varying environments.Theoretical Population Biology 58:211–237.Docauer, D. M. 1983. A nutrient basis for the distribution of the Lemnaceae. Ph.D dissertation,University of Michigan, Ann Arbor, Michigan, USA.Ellner, S. P., R. E. Snyder, and P. B. Adler. 2016. How to quantify the temporal storage effectusing simulations instead of math. Ecology Letters 19:1333–1342.
Page 14
14Enquist, B. J., R. Condit, R. K. Peet, M. Schildhauer, and B. M. Thiers. 2016.Cyberinfrastructure for an integrated botanical information network to investigate theecological impacts of global climate change on plant biodiversity. PeerJ Preprints4:e2615v2.MacArthur, R. 1970. Species packing and competitive equilibrium for many species. TheoreticalPopulation Biology 1:1–11.McCann, M. J. 2015. Local and regional determinants of an uncommon functional group infreshwater lakes and ponds. PLOS ONE 10:e0131980.Muthukrishnan Ranjan, Hansel-Welch Nicole, Larkin Daniel J., and Nilsson Christer. 2018.Environmental filtering and competitive exclusion drive biodiversity-invasibilityrelationships in shallow lake plant communities. Journal of Ecology 106:2058-2070.Turchin, P. 2003. Complex Population Dynamics: A Theoretical/Empirical Synthesis. PrincetonUniversity Press, Princeton, New Jersey, USA
Experimental evidence for a time-integrated effect ofproductivity on diversityDavid W. ArmitageDepartment of Integrative Biology,University of California Berkeley,Berkeley, CA, USACorrespondence: E-mail: dave.armitage@gmail.comAbstractThe time–area–productivity hypothesis is a proposed explanation for global biodiversity gradients.It predicts that a bioregion’s modern diversity is the product of its area and productivity, inte-grated over evolutionary time. I performed the first experimental test of the time–area–productivityhypothesis using a model system for adaptive radiation – the bacterium Pseudomonas fluorescensSBW25. I initiated hundreds of independent radiations under culture conditions spanning a vari-ety of productivities, spatial extents and temporal extents. Time-integrated productivity was thesingle best predictor of extant phenotypic diversity and richness. In contrast, ‘snapshots’ of mod-ern environmental variables at the time of sampling were less useful predictors of diversity pat-terns. These results were best explained by marked variation in population growth parametersunder different productivity treatments and the long periods over which standing diversity couldpersist in unproductive habitats. These findings provide the first experimental support for time-in-tegrated productivity as a putative driver of regional biodiversity patterns.KeywordsAdaptive radiation, area, diversification, productivity, Pseudomonas fluorescens, time.Ecology Letters (2015)INTRODUCTIONThe differential distribution of biodiversity across biogeo-graphic regions is simultaneously striking in its apparency andperplexing in its origins (Dobzhansky 1950; Hutchinson 1959;Pianka 1966). A remarkable number of hypotheses have beenproposed to explain why such regional differences in diversityexist – many of which place primacy on the strength and fluc-tuations of physical and biotic processes over evolutionarytime (e.g. Fischer 1960). In particular, the time–area–productivityhypothesis presents a compelling mechanism for explainingglobal gradients in biodiversity (Mittelbach et al. 2007; Jetz &Fine 2012; Belmaker & Jetz 2015). This hypothesis stresses thesimultaneous interacting roles of spatial extent and energyavailability integrated over evolutionary time in driving regio-nal diversification. Here, I follow Jetz & Fine’s (2012) defini-tion of bioregions as ‘evolutionary arenas’ – geographicalregions sharing few species and generally encompassing a sin-gle or multiple similar climatic biomes. Time integration of abioregion’s area and/or productivity allows their historical sig-natures on diversity dynamics to influence contemporary pat-terns of species diversity. In other words, time integrationimplies that communities equilibrate to a changing environ-ment relatively slowly compared to the timescale over whichdiversification occurs.The number of species a bioregion can support is predictedto positively correlate with its spatial extent (Chown & Gas-ton 2000). This prediction fits with the observation that thelarger tropical bioregions generally contain many more speciesthan smaller, globally discontinuous temperate regions(Terborgh 1973; Rosenzweig 1995). Originally, the effects ofarea were attributed to larger bioregions supporting greaterpopulation sizes than smaller regions. All else being equal,these larger populations have a greater probability of peri-patric speciation (Nuismer et al. 2012) and a decreased proba-bility of extinction due to buffering from catastrophicdisturbances (Rosenzweig 1995; Kisel et al. 2011). Largerbioregions are also anticipated to contain more boundaries togene flow, which should result in the evolution of reproductiveisolation (Kisel & Barraclough 2010). These hypotheses havebeen met with mixed results when observational biodiversitydata are fit to models incorporating their bioregional extents(Willig et al. 2003), suggesting that these contemporary ‘snap-shot’ measures of area may not be the dominant factors driv-ing global diversity patterns, especially among assemblagescontaining high levels of endemism.As with geographic extent, a bioregion’s energy supply canset the upper limit on population sizes (Wright 1983; Hurlbert& Jetz 2010). Regions receiving more energy per unit area,such as those in the tropics, can support larger populations ofindividual species than their temperate counterparts, makingthem less vulnerable to extinctions and more prone to specia-tion (Preston 1962; Srivastava & Lawton 1998; Hurlbert 2004).Through its effects on organismal metabolic rates, productivityis predicted to set limits on the total number of diverging pop-ulations that an ecosystem can support (Allen et al. 2002). Ifthe assumption of diversity-dependent speciation rates holdstrue (see Rabosky 2013), then this theory provides an avenuefor a positive feedback loop between standing and incipientdiversity mediated by energy input. Productive environmentsare also predicted to be more heterogeneous with regard toresource availability and thus may also permit diversificationand coexistence via niche partitioning and resource specialisa-tion (Abrams 1995). It is widely believed that the fitnessadvantages of many life history specialists should manifestonly above a minimum resource base (Wilbur et al. 1974).© 2015 John Wiley & Sons Ltd/CNRSEcology Letters, (2015)doi: 10.1111/ele.12501
Page 2
Until now, I have ignored the importance of time as bothas a driver of diversity gradients and as a mediator of arealand productivity effects on diversification rates. All else beingequal, older bioregions are anticipated to have accumulatedmore diversity than younger ones (Fischer 1960; Ricklefs &Schluter 1994). This hypothesis is supported by the observa-tion that many extant clades originated in older, tropicalbioregions (Mittelbach et al. 2007). Furthermore, because thetempo of diversification occurs on timescales encompassingmajor climatic and tectonic events, a bioregion’s area and pro-ductivity are expected to fluctuate to varying degrees whichshould in turn influence the region’s diversification dynamics(Fischer 1960; Terborgh 1973; Ricklefs & Schluter 1994).Opponents of contemporary area–productivity explanationsoften cite as a counterargument small and/or resource-poorhabitat patches containing high biodiversity (McGlone 1996;Fine & Ree 2006). If these patches are relics of once-large andproductive habitats, then their contemporary diversities maysimply be due to past events promoting diversification and thesubsequent long-term maintenance of that diversity in the faceof environmental change. By scaling a bioregion’s time-inte-grated area by its average productivity, Fine & Ree (2006)and Jetz & Fine (2012) were able to predict with remarkableaccuracy the species richness of their ‘evolutionary arenas’ – astrong argument for the joint roles of historical area, produc-tivity and temporal stability in promoting and maintainingdiversity. Crucially, the authors were unable to make similarlyaccurate predictions using contemporary ‘snapshot’ measuresof area and productivity for trees, endemic vertebrates andectotherms – the majority of species included in their analy-ses.The environmental stability of a bioregion can theoreticallyboth promote and hinder the generation and maintenance ofdiversity over evolutionary time. First, greater stability impliesmore time is available for niche specialisation to evolve andselects for continuous, rather than discrete, generations, accel-erating rates of recombination (Klopfer 1959; Fischer 1960;Connell & Orias 1964). Likewise, organisms adapted to fluctu-ating environments are often generalist phenotypes adapted totolerate ephemeral resources and fluctuating abiotic condi-tions. Alternatively, environmental fluctuations can lead tonegative frequency-dependent population dynamics and createtemporal niche opportunities, leading to a greater number ofcompetitors able to coexist (Levins 1979; Petraitis et al. 1989;Chesson & Huntly 1997). These contrasting effects of stabilityin both promoting and inhibiting coexistence and diversifica-tion are not mutually exclusive, and likely depend on spatialand temporal scales.The bacterium Pseudomonas fluorescens SBW25 is a modelsystem for the experimental study of adaptive diversificationand coexistence (Rainey & Travisano 1998). Under homoge-neous (i.e. shaken) culture conditions, colonies of the strainremain uniform in morphology (called smooth spreaders). Ifcultured under static conditions, however, this ancestral typequickly diversifies into a variety of morphologically distinctniche specialists. These mutants can be categorised into twodistinct groups based on colony morphology: the wrinklyspreaders and fuzzy spreaders, each of which encompasses sev-eral unique subtypes including wheel-like, lobate, filamentousand undulate forms (Fukami et al. 2007). Wrinkly spreaderspredictably evolve to exploit the air–water interface by excret-ing acetylated cellulose and forming a thick biofilm (Spierset al. 2003). This niche construction results in a sharp oxygengradient, paving the way for additional diversification. Thegenetic basis for these changes is well understood, with theparadox of random beneficial mutations leading to predictableindependent radiations explained by large population sizesand rapid growth rates (Spiers et al. 2002). Smaller culturevolumes are said to inhibit predictable radiations due toreduced population sizes (Rainey & Travisano 1998). Simi-larly, the productivity of the growth medium (carbon sub-strate concentration) also affects the population sizes, growthrates and relative fitness of P. fluorescens morphotypes, andtherefore can also constrain adaptive radiation (Kassen et al.2000). Finally, the stability of the culture habitat also influ-ences its diversity dynamics. Intermediate rates of disturbancevia imposed population bottlenecks drives diversification ratesand resulting community composition in a unimodal pattern,indicating that negative frequency-dependent selection atintermediate productivity and disturbance was equalizing thepopulation sizes (and relative fitness) of competing phenotypes(Buckling et al. 2000; Kassen et al. 2004).Here, I use the P. fluorescens SBW25 model system toexperimentally test the time–area–productivity hypothesis as itrelates to the tempo and extent of adaptive diversification.Because the air–water interface habitat is crucial for diversifi-cation, I used air–water interface area as a proxy for biore-gion extent, though it scales exactly with culture volume inthis study. I tested the hypothesis that time-integrated, pro-ductivity-scaled area (TimeAreaProductivity) is the best predic-tor of P. fluorescens phenotypic diversity owing to the positiveeffects of area on population size and the positive effects ofproductivity on both population size and growth rate. Fur-thermore, I investigated whether the temporal stability of theculture conditions experienced by each cell line eitherincreased or decreased the extent of P. fluorescens diversifica-tion. My goal was to compare the utility of these time-inte-grated variables as predictors of diversity to ‘snapshots’ of theecosystem taken at the time of sampling. I anticipated time-in-tegrated measures would outperform ‘snapshot’ measures ifthe time it took diversity to equilibrate to changing abioticconditions was long relative to the frequency of disturbance. Iperformed this test in order to experimentally verify the mech-anism and relative importance of time-integrated productivityand area as drivers of P. fluorescens diversity dynamics.MATERIAL AND METHODSTime–area–productivity experimentAncestral Pseudomonas fluorescens SBW25 cells were grown in20 mL KB broth under shaken conditions at 26° C for 2 days(attaining a population density of 3.4 9 1010 cells mLА1).Approximately 1 mL of this culture was spun at 10 0009g for5 min. The supernatant was replaced with sterile phosphatebuffer and the pelleted cells re-suspended. The centrifugationand resupply of fresh buffer was repeated three times to washcells of any residual medium. Cells were then diluted to© 2015 John Wiley & Sons Ltd/CNRS2 D. W. ArmitageLetter
Page 3
105 mLА1and starved for 2 h. I prepared three growth mediaencompassing a 100-fold difference in nutrient availability –a proxy for productivity. This order of magnitude approxi-mates the variance of primary productivity experiencedacross all terrestrial biomes (Yuan et al. 2010). Media wereprepared by diluting 19 M9-KB broth (NH4Cl1gLА1,Na2HPO4 6gLА1, KH2PO4 3gLА1, NaCl 0.5 g LА1, glyc-erol 6 g LА1, proteose peptone #3 20 g LА1) 10-fold and 100-fold in M9 salts solution. These three media were aliquotedinto flat-bottomed culture vessels spanning two orders ofmagnitude in volume: 10 mL (6-well plates), 1 mL (48-wellplates) and 0.1 mL (96-well plates). This experimental designallowed for growth of bacterial populations in nine differentcombinations of habitat volume (0.1, 1 and 10 mL) andresource availability (0.019, 0.19, 19).Into each habitat, I inoculated 50 lL of starved ancestralP. fluorescens cells (approximately 1000 isogenic cells). Eachof the nine culture conditions were replicated in triplicate andstored at 26° C under static conditions for 24 h. Aliquots ofthese cultures were then sampled, briefly vortexed and stored20% glycerol at А20° C. From two of the three replicate setsof cultures, I removed, washed and starved 50 lL, and inocu-lated them into three different randomly assigned vol-ume 9 productivity treatments, for a total of 54 randomisedhabitats (Fig. S1). These cultures were grown for 2 days andsampled as previously described. Aliquots were once againtaken from 27 of these cultures, washed and starved beforeusing each one to seed three new randomised habitats for atotal of 81 cultures. These were incubated for 3 days and thensampled and preserved. This schedule forced bacterial popula-tions to remain in or near the exponential phase of growthfor the duration of the experiment and did not allow thickbiofilms to form and collapse inside of the culture medium,nor allow cultures to deplete the medium and starve. Trans-fers contained approximately 106 cells, which did not repre-sent a significant genetic bottleneck. Dilutions of each 1-, 3-and 6-day population were made in phosphate buffer andspread on KB agar plates for enumeration. I identified bacte-rial morphotypes by scoring 100 random colonies per plate.All questionable colonies were re-plated to ensure that theywere genetically distinct.Time-to-equilibrium experimentI conducted a second experiment to estimate the time scaleover which diversity equilibrates to a particular productivity.Washed and starved ancestral bacteria were inoculated intowells containing 1 mL of high, medium, or low-resource brothas previously described. Every 3 days, a 50 lL aliquot of eachculture was used to seed a new well of the same productivityand the rest of the culture was preserved and frozen. After24 days, 50 lL aliquots from the eight cultures containing thegreatest number of unique morphotypes (all of the high andtwo of the medium productivity cultures) were then washed ofmedia, starved, added into 1 mL low-productivity wells andserially transferred and preserved every 3 days for an addi-tional 21 days. Samples from the entire 45-day time serieswere then plated and scored for morphotype richness anddiversity.Growth curve measurementI measured growth parameters of P. fluorescens SBW25 grow-ing in 100 mL batch cultures in the three productivity treat-ments. Population growth was approximated as the opticaldensity at 600 nm (OD600) on a Molecular Devices Emaxspectrophotometer. I fit Gompertz growth functions to thesetime series and estimated the 95% confidence intervals for theparameters l (growth rate), A (carrying capacity) and k (lagtime) using nonlinear least squares. The growth function tookthe formy ¼ Aexp Аexplm Б eAрk А tЮ ю 1[]{},where t is the culture age (hours), e is base of the natural loga-rithm and y is the change in OD600 values from t0. Once eachculture reached stationary phase, I estimated its populationdensity by plating dilutions of the culture onto solid mediumand counting the colonies.Data collection and statistical analysesI estimated diversity as previously described (Kassen et al.2004). Alongside morphotype richness, I estimated the com-plement (1 А k) of Simpson’s index, k, where k = Σp2i and piis the frequency of each morphotype in the 100 censused colo-nies. This metric takes a value between 0 and 1 and representsthe likelihood that any two randomly selected colonies belongto different morphotypes. This index was logit-transformed tofulfil the assumptions of linear regression. Time-integratedarea (TimeArea) and productivity (TimeProductivity) weremeasured by integrating the total areas and productivitiesexperienced by each cell line over their 1-, 3- or 6-day evolu-tionary histories. Similarly, TimeAreaProductivity is the time-integrated product of culture area and productivity (Jetz &Fine 2012) (Fig. S1). I estimated the stability of TimeArea,TimeProductivity and TimeAreaProductivity using the coeffi-cient of variation (CV = standard deviation/mean). I fit a ser-ies of generalised linear (GLM) and generalised additivemodels (GAM) to the hypothesised drivers of diversity. I usedthe slope parameters of GLM fits to interpret the magnitudeand direction of covariance, while GAM fits were used to esti-mate the shape of the response surface and estimate the pro-portion of variance explained. Additive models wereconstrained to a maximum of five basis dimensions to avoidoverfitting while permitting quadratic and logistic-likeresponse surfaces. The response variables were richness (mod-elled as Poisson distributed) and logit-transformed diversity(modelled as Gaussian distributed). Best-fit models wereselected using AICc and coefficients of determination (R2). Toavoid pseudoreplication, I only analysed endpoint communi-ties – those that were not used to seed new microcosms at 1-and 3-day transfers.To estimate whether equilibrium diversities were reached inthe productivity treatments and whether they differed, I fit aseries of generalised additive models to the 24-day time seriesbegun with the isogenic ancestral cells. I visually compared© 2015 John Wiley & Sons Ltd/CNRSLetterHistory affects microbial diversification 3
Page 4
the shapes of each productivity treatment’s curves andassessed whether the 95% confidence intervals surroundingeach estimated mean differed between productivities. I com-pared these curves to intercept-only null models using likeli-hood ratio tests and R2 values. I then performed the same setof analyses on the high-diversity, low-productivity treatmentsstarted at day 24. I anticipated that these cultures’ diversitieswould decrease over time, as less fit morphotypes are drivento rarity or extinction and that their diversities would eventu-ally equal those of low-productivity cultures started withancestral cells. All statistical operations were performed usingR (R Development Core Team 2014) and code is availablefrom the author on request.RESULTSTime–productivity drives diversity dynamicsA total of 116 independent microcosms were scored for diver-sity. In total, I identified eight distinct, heritable colony mor-photypes, though no single culture contained more than fourand all but three were variants of the wrinkly spreader pheno-type (see Fukami et al. 2007 for descriptions) (Fig. S2). Inonly one case was a morphotype observed to go extinct.I confirmed a relatively weak relationship betweenTimeAreaProductivity and diversity, but not for morphotyperichness (Table 1). However, the covariate TimeProductivitywas the strongest single predictor for both diversity and rich-ness (Fig. 1). Furthermore, linear models incorporating time-integrated productivity explained approximately 33% (diver-sity) and 26% (richness) more variance than did models usingonly ‘snapshot’ productivity measured at the time of sampling(Table1). Neither contemporary ‘snapshots’ of area norTimeArea were found to be associated with diversity or rich-ness. As anticipated, culture age per se (Time) was also a goodpredictor of culture diversity. The stability of area over time(CVarea) was positively associated with diversity and richness.Inclusion of CVareainto the TimeProductivity linear andnonlinear models delivered the best predictive accuracies(Fig. 2). For logit-transformed diversity and richness, thisincreased the percentage variance explained by the best-fitGAM models to 72 and 60% respectively.Growth rates vary by productivityBy analyzing the strain’s growth kinetics under varying nutri-ent concentrations, I determined that differences exist in thestrain’s growth rates (l) and carrying capacities (A) amongproductivity treatments (Table 2). Specifically, diluting thestandard KB+M9 medium by 100 has the effect of decreasingthe bacterium’s growth rate and carrying capacity approxi-mately 10-fold (Fig. 3). At steady state, the cultures had esti-mated population densities of 2.3 9 109, 1.8 9 109and2.8 9 108 cells mLА1for the high, medium and low-produc-tivity treatments respectively.Historical signatures of productivity are slow to disappearDiversity and richness in all cultures initially increased over aperiod of approximately 6–9 days (Fig. 4 and Fig. S3). Inhigh and medium cultures, these values initially increased andthen appeared to remain unchanged after 9 days. Diversity inlow-productivity cultures steadily increased throughout the24-day period and ended lower than those in high and med-ium productivities. Average morphotype richnesses for high(five morphotypes) and medium (four morphotypes) produc-tivities were both greater than morphotype richness in low-productivity cultures (two morphotypes) (Fig. 4). For culturesstarted with ancestral smooth cells, AICc and R2 metricsfavoured time-variant models of diversity over their intercept-only counterparts (Table S1). Morphotype-rich culturesmoved into low-productivity media did not decrease in diver-sity nor richness. For these cultures, time-dependent modelswere not improvements over intercept-only null models (TableS1). The expected values from these intercept-only modelswere within bounds of the endpoint mean diversity and rich-Predictor variablesRichnessDiversityGLMmodelsGAMmodelsGLMmodelsGAMmodelsDAICcR2DAICcR2DAICcR2DAICcR2Time170.21 200.18930.21 1140.21Area280.01 310.00 1200.00 1410.00Productivity130.28 150.30690.36890.36AreaProductivity260.05 280.08 1050.12 1210.17TimeArea290.00 320.00 1200.00 1410.00TimeProductivity00.4900.5650.6360.69TimeAreaProductivity250.06 280.06 1070.11 1260.06CVarea220.12 250.10 1080.10 1290.10CVproductivity280.02 280.10 1160.04 1230.10CVarea + CVproductivity240.12 260.16 1090.11 1180.16TimeProductivity + CVarea00.5400.5900.6500.72TimeAreaProductivity + CVarea210.18 240.16950.22 1150.16Intercept-only null270.00 300.00 1180.00 1390.00GAM, generalised additive models; GLM, generalised linear.Table 1 Relative performances of time-integrated, snap-shot and additional covariates in predicting morphotyperichness and diversity. Best models (DAICc ≤ 2) are boldedand R2represents the shrinkage-adjusted coefficient ofdetermination of observed vs. model-predicted values© 2015 John Wiley & Sons Ltd/CNRS4 D. W. ArmitageLetter
Page 5
ness values from the high and medium productivity culturesstarted with the ancestral strain but greater than the meanendpoints of the low-productivity cultures (Fig. 4).DISCUSSIONUsing a model microbial adaptive radiation, I have shownthat time-integrated productivity is a primary driver of diver-sification dynamics in the P. fluorescens model system. Thesedata represent the first experimental test of the time–area–pro-ductivity hypothesis and are consistent with results obtainedobservational studies (Jetz & Fine 2012; Belmaker & Jetz2015). Contrary to their findings, however, I failed to identifyspatial extent as an important predictor of diversification inthis system. A reason for this discrepancy may be that theprevious studies were unable to explicitly model TimeProduc-tivity separately from TimeArea due to a lack of pre-Holoceneproductivity data. Instead, the authors scaled TimeArea bythe bioregion’s modern productivity. Therefore, this studyrepresents the first to independently test for TimeProductivityand TimeArea effects. It remains to be determined whethertime-integrated productivity is able to explain more variationthan time-integrated area in global datasets.The differences in diversification observed among experi-mental cultures stemmed from the maximisation of populationcarrying capacities and growth rates during periods of high(a)(b)©(d)Figure 1 Extent of Pseudomonas fluorescens SBW25 morphotype diversity and richness plotted against time-integrated productivity (a and b) and ‘snapshot’productivity (c and d). Solid lines denote fitted generalised additive models, dotted lines denote linear models.© 2015 John Wiley & Sons Ltd/CNRSLetterHistory affects microbial diversification 5
Page 6
productivity. Because cultures were maintained at or nearexponential phase of growth during the experiment, the posi-tive effects of productivity on growth rates allowed for benefi-cial mutants to accumulate more rapidly and persist in moreproductive environments. Contrary to anecdotal evidence(Rainey & Travisano 1998) neither culture volume (i.e. area)nor its time-integrated measure was positively associated withthe extent of Pseudomonas diversification. This occurreddespite a two order-of-magnitude difference in populationsizes between 10 mL and 0.1 mL cultures. Thus, for P. fluo-rescens SBW25, time-integrated air–water interface area (orculture volume) cannot be considered an important driver ofdiversification in the presence of other environmental vari-ables affecting population growth.One explanation for why productivity, rather than area,drove diversification in this experiment is the competitive sup-pression of de novo niche specialists at low nutrient concentra-tions. Similar to Kassen et al. (2004), even small (0.1 mL),low-productivity cultures at 48 h contained populations ofP. fluorescens large enough to produce wrinkly spreadermutants. However, the relative fitness of these morphotypesrely on a nutrient supply large enough to allow coordinatedexpression of cellulosic polymer (Spiers et al. 2003; Kassenet al. 2004). Lacking a surplus of growth substrate, wrinklyspreader populations cannot exploit the air–water interfaceniche and are either driven extinct or maintained at low-frequencies via competition with the ancestral smooth mor-photype (Kassen et al. 2004). At intermediate and high pro-ductivities, all culture sizes contained nutrient concentrationsnecessary for wrinkly spreaders to invade, despite the pleiotro-pic fitness costs to biofilm production via decreased carbon(a)(b)Figure 2 Predicted vs. observed values for best-fit generalised additive models for (a) logit-transformed morphotype diversity and (b) morphotype richness.Table 2 Ninty-five per cent confidence intervals for Gompertz growthmodel under varying productivities. Parameter l is growth rate(absorbance units hА1), A is carrying capacity (absorbance units) and k istime lag (hours)ProductivityParameter estimate (95% CI)lAk190.037–0.0551.054–1.22516.5–19.80.190.025–0.0300.830–0.92713.8–16.00.0190.004–0.0060.112–0.1324.5–10.3Figure 3 Growth kinetics of Pseudomonas fluorescens SBW25 under thethree productivity treatments. Lines denote Gompertz growth curve fits.© 2015 John Wiley & Sons Ltd/CNRS6 D. W. ArmitageLetter
Page 7
metabolism (MacLean et al. 2004). In contrast, populations ofmacro-organisms are often constrained by geographic barriersto sizes much smaller than those observed in this experiment(Preston 1962). Such populations should not diversify, partic-ularly if they are simultaneously being constrained by a low-productivity environment and/or competition (Wright 1983;Rosenzweig & Abramsky 1993). Additionally, larger biore-gions are anticipated to contain more barriers to gene flow,which can promote non-adaptive diversification (Terborgh1973; Gittenberger 1991; Rosenzweig 1995). Such allopatricdiversification is not permitted in the relatively homogeneousenvironment of microbial culture vessels due to their lack ofgeographic boundaries. However, it is worth noting that nosingle culture contained all eight morphotypes, and a summa-tion of morphotypes in 10 isolated 1 mL (or <100 isolated0.1 mL) wells was unanimously greater than or equal to thediversity of any single 10 mL well, given similar productivi-ties. This finding lends support to the role of geographic bar-riers in promoting regional diversity through the ecologicalsaturation of convergent morphotypes (Terborgh & Faaborg1980; Ricklefs 2004). The large effective population sizes (106cells) and overall magnitude and replicability of diversificationindicate that divergence via neutral drift is not the cause ofthis pattern, and it is instead due to selection. In a culturewithout geographically isolated populations and limitedresources over which to compete, clonal interference couldlimit the number of coexisting niche specialists (Gerrish &Lenski 1998).My results are similar to those of Kassen et al. (2000, 2004)in demonstrating inhibition of P. fluorescens diversification inlow-productivity habitats. Unlike these previous studies, how-ever, I did not encounter a negative-quadratic shape to theproductivity–diversity relationship. Instead, diversity andrichness appeared to asymptote at high time-integrated pro-ductivities. This difference may be because Kassen et al.’s‘high-productivity’ cultures contained nutrient concentrationsfar greater than the standard 19 King’s medium formulation.Furthermore, the extent of P. fluorescens diversification tendsto follow a sigmoidal trend over the time periods used in thisexperiment – a pattern observed elsewhere (Fukami et al.2007). Because two sources of environmental variation areincorporated into the TimeProductivity variable, it is probablethat the saturating dynamics I observed represent a combina-tion of constraints on diversification imposed by culture age(sigmoidal) and productivity (negative-quadratic). However,equivalent values of TimeProductivity can be achieved in avariety of ways. For instance, an intermediate productivityculture running for 3 days can equal a high-productivity cul-ture at 1 day, or a low-productivity culture at 6 days. Produc-tivity, however, is one constraint on the diversificationdynamics of P. fluorescens SBW25. A number of other fac-tors, most notably habitat heterogeneity, temperature, distur-bance and community interactions (e.g. predation andcompetition) also mediate this model radiation (Rainey &Travisano 1998; Kassen et al. 2004; Fukami et al. 2007;Meyer & Kassen 2007).I detected an effect of culture area stability on morphotypediversity and richness, but I did not anticipate the associationto be positive in direction. Moreover, inclusion of this covari-ate in my models only resulted in an improvement of 5% vari-ance explained. Models of species accumulation over timeoften cite the importance of climate-driven habitat stability indriving niche specialisation and speciation (Klopfer 1959; Fis-cher 1960; Connell & Orias 1964). My finding that morpho-(a)(b)Figure 4 (a) Diversity and (b) richness dynamics for cultures in high (red), medium (blue) and low (green) productivity media. Dotted line denotes day atwhich the eight highest richness cultures were moved into low-productivity environments. Solid lines indicate best-fitting generalised additive models.Shaded regions denote 95% confidence regions around predicted means.© 2015 John Wiley & Sons Ltd/CNRSLetterHistory affects microbial diversification 7
Page 8
type richness and diversity decreased with environmental sta-bility does not support the prediction that environmental sta-bility per se promotes diversification. Rather, this observationsuggests that disturbance in terms of periodically imposedpopulation bottlenecks promotes the maintenance of incipientniche specialists. However, these bottlenecks were not espe-cially strong, since approximately 106 cells were used to seednew cultures and transfers were made at time intervals overwhich both community interactions and adaptive diversifica-tion can influence diversity dynamics. Buckling et al. (2000)obtained a similar result and concluded that intermediatelevels of disturbance acting on diversifying P. fluorescens cul-tures allowed rare genotypes the opportunity to invade anotherwise resistant community. Using the same system, Tanet al. (2013) found evidence that temporal variation in nicheavailability promoted coexistence among morphotypes. How-ever, the positive covariance of CVarea with both Time andtransfer frequency hinders the interpretation of this effect inmy study. Nonetheless, whichever stability metric was used inmy models, the qualitative result was consistent: a negativetrend exists between diversity and environmental stability inthe model P. fluorescens radiation.Although I only detected a single extinction event duringmy experiment, it is probable that many undetected de novomorphotypes arose in all cultures and were rapidly drivenextinct by established competitors. In both experiments, nei-ther diversity nor richness decreased when diverse, high-pro-ductivity assemblages were transplanted into low-productivitymedia. In other words, niche specialist phenotypes unable toestablish in low-productivity cultures could persist at their his-torical relative frequencies despite the reduced habitat quality.Extinction debt is a hypothesis used to explain the mainte-nance of hyper-diverse assemblages in deteriorating or shrink-ing habitats (Tilman et al. 1994). In both this experiment andmany observational studies, present day area and productivitywere poor predictors of regional variance in diversity (e.g.Fine & Ree 2006; Jetz & Fine 2012). Rather, time-integratedarea and productivity were much better explanatory variables.It is noteworthy, however, that Jetz & Fine (2012) found evi-dence for the primacy of non-time-integrated ‘snapshot’ mea-sures in explaining the richness of non-endemic bird andmammal assemblages. This suggests that time-integrated mea-sures may be less useful when applied to assemblages capableof crossing bioregional boundaries (e.g. via migration). Fur-ther work is needed to determine whether biota better mod-elled with time-integrated variables share certain key traitspredisposing them to endemicity.The primacy of snapshot vs. time-integrated variables inexplaining regional diversity patterns depends in part on therelative rates at which these systems equilibrate following adisturbance. Historical signatures of area and productivityshould manifest if communities return to equilibrium slowly,but should be erased if communities either do not reach asteady state or return to it very quickly. My finding that high-richness cultures placed into low-productivity environmentsdo not lose their derived morphotypes confirms the impor-tance of the cultures’ historical conditions in explainingcontemporary diversity patterns. The time it took for high,medium and low-productivity cultures to reach initialequilibria was on the order of 1 week, whereas once the com-munities had developed, they were capable of remainingunchanged for 3 weeks, despite being moved into low-produc-tivity media. While this timeframe is partly a product of thestudy system’s large population sizes, rapid generation timesand relative simplicity, it nonetheless verifies that given certainconditions, time-integrated effects on community structurecan persist over evolutionary timescales. Fitness differencesthat precluded de novo morphotypes from establishing inunproductive medium did not appear to affect their long-termpersistence if they had originally diversified in a more produc-tive habitat. In other words, the trajectory of diversity in aculture experiencing increasing productivity was not along thesame response surface as a culture experiencing a decrease inproductivity. Therefore, it is likely that the direction of envi-ronmental change can influence the length of time over whichtime-integrated effects can persist – environments increasingin productivity can be invaded by de novo mutants, whereasdeteriorating environments may prevent such invasions viatheir suppression by established competitors.In concert, these results suggest that time–energy effectsmanifest most strongly during early stages of diversificationand persist longest in deteriorating environments. Once nichesare sufficiently saturated with de novo phenotypes, competitiveexclusion of similar low-frequency phenotypes sets the upperlimit on richness. In the simple physicochemical habitats usedin my experiments, niche saturation occurred relativelyquickly and persisted indefinitely. Time–productivity effectsmay not have been as pronounced had cultures been allowedto remain at equilibria for a longer stretch of time prior tomodel fitting. However, doing so would have forced the sys-tem away from modelling natural diversity dynamics. There iscurrently limited consensus on whether a strict asymptoticdiversification model holds for natural systems, primarily dueto the fact that key evolutionary innovations and mass extinc-tions tend to keep diversification rates from reaching pro-longed steady states (Rabosky 2013). Whether or not theobservation of hyper-diverse biotas in sub-optimal habitats orrefugia represents extinction debt or evolutionary acclimationrequires further investigation, though it is clear from thesedata that if extinction debt responsible for this observation, itis occurring over a period at least three times as long as thetime required for diversity to first appear. Answering thisquestion requires long-term experiments and new methods toidentify and characterise novel phenotypes.In conclusion, high historical energy availability drove theevolution of niche specialists, which were unable tosuccessfully establish in resource-poor environments. Low-productivity cultures inoculated with diverse assemblages fromhigh-productivity habitats, however, did not experience extinc-tions, implying that a habitat’s standing diversity can bedecoupled from its contemporary environmental conditions.These results confirm that modern day ‘snapshot’ ecosystemmetrics are at best proxies for explaining regional variation inextant diversity, particularly among endemic species. Atworst, these variables can mislead analyses on drivers ofbiodiversity. Further, these results extend the domain of time-integration hypotheses to bacteria – organisms rarely consid-ered bound by historical biogeographic constraints (but see© 2015 John Wiley & Sons Ltd/CNRS8 D. W. ArmitageLetter
Page 9
Hanson et al. 2012). Going forward, these findings supportthe need for more historical data on both area and productiv-ity to explain patterns of biodiversity at large spatial and tem-poral scales.ACKNOWLEDGEMENTSI thank P. Fine, T. Fukami, W. Sousa, D. Quiroz, H. Kurk-jian, A. Hurlbert, two anonymous reviewers and the IB 250graduate seminar on latitudinal gradients for inspiration,insight and feedback. Laboratory instrumentation was kindlyprovided by E. Simms. P. Rainey generously donated the bac-terial strain used for this study. Support for this researchcomes from UC Berkeley’s Wang Family Fellowship and theDepartment of Integrative Biology.AUTHORSHIPDWA conceived this work, performed data collection andanalysis and wrote the manuscript.REFERENCESAbrams, P.A. (1995). Monotonic or unimodal diversity-productivitygradients: what does competition theory predict? Ecology, 76, 2019–2027.Allen, A.P., Brown, J.H. & Gillooly, J.F. (2002). Global biodiversity,biochemical kinetics, and the energetic-equivalence rule. Science, 297,1545–1548.Belmaker, J. & Jetz, W. (2015). Relative roles of ecological and energeticconstraints, diversification rates and region history on global speciesrichness gradients. Ecol. Lett., 18, 563–571.Buckling, A., Kassen, R., Bell, G. & Rainey, P.B. (2000). Disturbanceand diversity in experimental microcosms. Nature, 408, 961–964.Chesson, P. & Huntly, N. (1997). The roles of harsh and fluctuatingconditions in the dynamics of ecological communities. Am. Nat., 150,519–553.Chown, S.L. & Gaston, K.J. (2000). Areas, cradles and museums: thelatitudinal gradient in species richness. Trends Ecol. Evol., 15, 311–315.Connell, J.H. & Orias, E. (1964). The ecological regulation of speciesdiversity. Am. Nat., 98, 399–414.Dobzhansky, T. (1950). Evolution in the tropics. Am. Sci., 38, 209–221.Fine, P.V.A. & Ree, R.H. (2006). Evidence for a time-integrated species-area effect on the latitudinal gradient in tree diversity. Am. Nat., 168,796–804.Fischer, A.G. (1960). Latitudinal variations in organic diversity.Evolution, 14, 64–81.Fukami, T., Beaumont, H.J.E., Zhang, X.-X. & Rainey, P.B. (2007).Immigration history controls diversification in experimental adaptiveradiation. Nature, 446, 436–439.Gerrish, P.J. & Lenski, R.E. (1998). The fate of competing beneficialmutations in an asexual population. Genetica, 102–103, 127–144.Gittenberger, E. (1991). What about non-adaptive radiation? Biol. J.Linn. Soc., 43, 263–272.Hanson, C.A., Fuhrman, J.A., Horner-Devine, M.C. & Martiny, J.B.H.(2012). Beyond biogeographic patterns: processes shaping the microbiallandscape. Nat. Rev. Microbiol., 10, 497–506.Hurlbert, A.H. (2004). Species–energy relationships and habitatcomplexity in bird communities. Ecol. Lett., 7, 714–720.Hurlbert, A.H. & Jetz, W. (2010). More than “more individuals”: thenonequivalence of area and energy in the scaling of species richness.Am. Nat., 176, E50–E65.Hutchinson, G.E. (1959). Homage to Santa Rosalia or why are there somany kinds of animals? Am. Nat., 93, 145–159.Jetz, W. & Fine, P.V.A. (2012). Global gradients in vertebrate diversitypredicted by historical area-productivity dynamics and contemporaryenvironment. PLoS Biol., 10, e1001292.Kassen, R., Buckling, A., Bell, G. & Rainey, P.B. (2000). Diversity peaksat intermediate productivity in a laboratory microcosm. Nature, 406,508–512.Kassen, R., Llewellyn, M. & Rainey, P.B. (2004). Ecological constraintson diversification in a model adaptive radiation. Nature, 431, 984–988.Kisel, Y. & Barraclough, T.G. (2010). Speciation has a spatial scale thatdepends on levels of gene flow. Am. Nat., 175, 316–334.Kisel, Y., McInnes, L., Toomey, N.H. & Orme, C.D.L. (2011). Howdiversification rates and diversity limits combine to create large-scalespecies–area relationships. Philos. Trans. R. Soc. Lond. B Biol. Sci.,366, 2514–2525.Klopfer, P.H. (1959). Environmental determinants of faunal diversity.Am. Nat., 93, 337–342.Levins, R. (1979). Coexistence in a variable environment. Am. Nat., 114,765–783.MacLean, R.C., Bell, G. & Rainey, P.B. (2004). The evolution of apleiotropic fitness tradeoff in Pseudomonas fluorescens. Proc. Natl Acad.Sci. USA, 101, 8072–8077.McGlone, M.S. (1996). When history matters: scale, time, climate andtree diversity. Glob. Ecol. Biogeogr. Lett., 5, 309–314.Meyer, J.R. & Kassen, R. (2007). The effects of competition andpredation on diversification in a model adaptive radiation. Nature, 446,432–435.Mittelbach, G.G., Schemske, D.W., Cornell, H.V., Allen, A.P., Brown,J.M., Bush, M.B. et al. (2007). Evolution and the latitudinal diversitygradient: speciation, extinction and biogeography. Ecol. Lett., 10, 315–331.Nuismer, S.L., MacPherson, A. & Rosenblum, E.B. (2012). Crossing thethreshold: gene flow, dominance and the critical level of standinggenetic variation required for adaptation to novel environments. J.Evol. Biol., 25, 2665–2671.Petraitis, P.S., Latham, R.E. & Niesenbaum, R.A. (1989). Themaintenance of species diversity by disturbance. Q. Rev. Biol., 64, 393–418.Pianka, E.R. (1966). Latitudinal gradients in species diversity: a review ofconcepts. Am. Nat., 100, 33–46.Preston, F.W. (1962). The canonical distribution of commonness andrarity: part I. Ecology, 43, 185–215.R Development Core Team (2014). R: a Language and Environment forStatistical Computing. R Foundation for Statistical Computing, Vienna,Austria.Rabosky, D.L. (2013). Diversity-dependence, ecological speciation, andthe role of competition in macroevolution. Annu. Rev. Ecol. Evol. Syst.,44, 481–502.Rainey, P.B. & Travisano, M. (1998). Adaptive radiation in aheterogeneous environment. Nature, 394, 69–72.Ricklefs, R.E. (2004). A comprehensive framework for global patterns inbiodiversity. Ecol. Lett., 7, 1–15.Ricklefs, R.E. & Schluter, D. (eds.) (1994). Species Diversity in EcologicalCommunities. University Of Chicago Press, Chicago.Rosenzweig, M.L. (1995). Species Diversity in Space and Time. CambridgeUniversity Press, Cambridge, UK.Rosenzweig, M.L. & Abramsky, Z. (1993). How are diversity andproductivity related? In Species Diversity in Ecological Communities:historical and Geographical Perspectives. (eds Ricklefs, R.E., Schluter,D.). University of Chicago Press, Chicago, pp. 52–65.Spiers, A.J., Kahn, S.G., Bohannon, J., Travisano, M. & Rainey, P.B.(2002). Adaptivedivergenceinexperimental populations ofPseudomonas fluorescens. I. genetic and phenotypic bases of wrinklyspreader fitness. Genetics, 161, 33–46.Spiers, A.J., Bohannon, J., Gehrig, S.M. & Rainey, P.B. (2003). Biofilmformation at the air–liquid interface by the Pseudomonas fluorescensSBW25 wrinkly spreader requires an acetylated form of cellulose. Mol.Microbiol., 50, 15–27.© 2015 John Wiley & Sons Ltd/CNRSLetterHistory affects microbial diversification 9
Page 10
Srivastava, D.S. & Lawton, J.H. (1998). Why more productive sites havemore species: an experimental test of theory using tree-holecommunities. Am. Nat., 152, 510–529.Tan, J., Kelly, C.K. & Jiang, L. (2013). Temporal niche promotesbiodiversity during adaptive radiation. Nat. Commun., 4, 2102.Terborgh, J. (1973). On the notion of favorableness in plant ecology. Am.Nat., 107, 481–501.Terborgh, J.W. & Faaborg, J. (1980). Saturation of bird communities inthe West Indies. Am. Nat., 116, 178–195.Tilman, D., May, R.M., Lehman, C.L. & Nowak, M.A. (1994). Habitatdestruction and the extinction debt. Nature, 371, 65–66.Wilbur, H.M., Tinkle, D.W. & Collins, J.P. (1974). Environmentalcertainty, trophic level, and resource availability in life historyevolution. Am. Nat., 108, 805–817.Willig, M.R., Kaufman, D.M. & Stevens, R.D. (2003). Latitudinalgradients of biodiversity: pattern, process, scale, and synthesis. Annu.Rev. Ecol. Evol. Syst., 34, 273–309.Wright, D.H. (1983). Species-energy theory: An extension of species-areatheory. Oikos, 41, 496–506.Yuan, W., Liu, S., Yu, G., Bonnefond, J.-M., Chen, J., Davis, K. et al.(2010). Global estimates of evapotranspiration and gross primaryproduction based on MODIS and global meteorology data. RemoteSens. Environ., 114, 1416–1431.SUPPORTING INFORMATIONAdditional Supporting Information may be downloaded viathe online version of this article at Wiley Online Library(www.ecologyletters.com).Editor, Mark VellendManuscript received 7 July 2015First decision made 29 July 2015Manuscript accepted 4 August 2015