Classifying lifespan in perennial tallgrass : can we use readily available trait data?

Justin Meissen | University of Minnesota Effective restoration action relies on predicting species response to management

Informs: . Establishment outcomes (e.g. long lived species establish slowly) . Efficient use of resources (e.g. proper herbicide timing reducing repeat treatments) . Benefits of management actions (e.g. spring burns boosting warm season species) . Risks of management actions (e.g. wild seed harvest depleting donor species) Functional traits can be useful for predicting response to management

Seed production . Regeneration potential Clonality . Persistence after disturbance height . Light requirements Lifespan . Persistence and regeneration potential

(Roberts et al. 2010, Spasojevic et al. 2010, Clark et al. 2012)

Functional traits must be measured to be used

Most traits are readily measured within a field season with standard lab equipment

Examples: Seed production . Seed counts Clonality . Presence/absence of vegetative organs Plant height . Meter stick However, measuring lifespan is uniquely difficult

By observation . Extremely long timeframes necessary . Dormancy, genets make it complex Weaver and Anderson 1936

By growth rings . Highly destructive . Only works for some taxa

von Arx and Dietz 2006 Related traits may inform lifespan estimates more easily than observation

Does not require many years to measure, not as destructive

Often already available in databases

Established use in estimating other ecological parameters in plants (e.g. ecological strategies) Related traits may inform lifespan estimates more easily than observation

Can we use other life history traits to indirectly estimate lifespan?

Approach

1. Reviewed literature to find traits that reflect lifespan 2. Developed a decision tree model for lifespan classification using traits 3. Classified common tallgrass prairie species using decision tree 4. Tested model using published lifespan studies 1. Literature shows three important traits influencing lifespan

Life form . Graminoids= longer lived (Chu et al. 2013)

Age of maturity (AoM) . Higher AoM= longer lived (Bender et al. 2000)

Clonality . Clonal= longer lived (Bender et al. 2000) 2. Lifespan classification system decision tree

AoM= Age of maturity

Class Lifespan (years)

1 1-5 2 5-10 3 10-15 4 >15 2. Lifespan classification of common northern prairie species

Species Age of Maturity Graminoid Clonal LifeClass Andropogon gerardii Yes Yes 4 Anemone canadensis 3 No Yes 4 Calamagrostis stricta Yes Yes 4 ligulistylis 9 No No 4 Rosa arkansana 4 No No 4 Apocynum cannibinum 2 No Yes 3 laevis 2 No Yes 3 Bromus kalmii Yes No 3 Galium boreale 2 No Yes 3 Thalictrum dasycarpum 3 No No 3 Aster ericoides 1 No Yes 2 Dalea purpurea 2 No No 2 nemoralis 2 No No 2 Solidago rigida 2 No No 2 Zizia aurea 2 No No 2 Ratibida columnifera 1 No No 1 Rudbeckia hirta 1 No No 1 Senecio paupercaulus 1 No No 1 3. Is this classification system accurate?

Tested using independent long term lifespan studies . Lauenroth and Adler 2008, Woodson 1962, Levin 1973 . Classified species in these studies with this ? system . Assessed whether classes reflected observed Many other studies also lifespan used to find age of maturity 3.Trait-based lifespan classes reflect observed lifespan in the field

Correctly predicted Predicted Observed Observed Species LifeClass LifeClass lifespan the true lifespan Amorpha canescens 4 4 17 Asclepias tuberosa 4 4 26 class 72% of the time Aster ericoides 2 2 6 Aster oblongifolius 2 2 6 Calylophus serrulatus 2 2 10 Dalea purpurea 2 2 7 Liatris aspera 4 4 41 Misclassified to a Liatris cylindracea 4 4 29 3 3 12 non-adjacent class Liatris spicata 4 4 22 Ratibida columnifera 1 1 3 only once Sphaeralcea coccinea 2 2 6 Echinacea angustifolia 2 3 11 Mirabilis hirsuta 2 4 20 Psoralea tenuiflora 2 3 11 Scutellaria resinosa 2 3 13 Solidago rigida 2 1 3 Conclusions

1. A trait-based classification system can inform estimates of maximum lifespan in perennial prairie plants

2. Life-form, age of maturity, and clonality determine 4 lifespan classes (1-5 yr, 5-10yr, 10- 15yr, >15yr)

3. Lifespan classes are accurate especially when determining very short or very long lived species

Implications

Lifespan classification system

. Could help predict time to establishment

. Could help predict response to management (e.g. seed harvest)

. Can help inform seed mix design

. Useful for further life history and trait-based research

Acknowledgments

Meredith Cornett and Sue Galatowitsch, Advisors

Joe Fargione, Tony D’Amato, and Bob Haight, Committee Questions?