Demographic Model: Structure

1. Demographic Model:Structure Mary C. Christman (UMD/UFL) Danny Lewis (UMD) Jon Volstad (Versar)

2. Overview • Yearly time step starting in October each year • Parameters and structure are modified according to alternative under consideration • Done on a per bar basis and aggregated to desired spatial scales

3. Spat Fall • Predictions based on: 1: Projected number of spat surviving to 6-40 mm (mode 30 mm) 2: Spatial distribution based on larvae settlement projected from hydrodynamic modeling (Dr. Elizabeth North, HPL) • Part 1: predicted # spat per spawner (standardized fecundity to 77 mm oyster) based on stock-recruit regressions: • Use the DNR spat fall survey data (1991-2003) • Estimate regression parameters w/ standard deviation by regions and type of weather year (dry, average, wet)

4. Parameterization ofdemographic model based on currently available data Jon H. Vølstad, Jodi Dew and Ed Weber Versar, Inc., Columbia, MD and Mary Christman, UFL

5. Data sources for estimating growth parameters (C. Va) • Dr. Paynter (unpublished) • Individual growth data from 25 MD sites with sufficient sample sizes • Coakley (2004) • Growth parameters based on cohort analysis (29 MD sites) • Virginia growth data from James River

6. VBGC growth Assume growth (length) during a time step is a function of size not age

7. Growth equation for oysters at size (not age) • VB function • L1 is the size class in the current year • L2 is the mean size class after growth in a single time step.

8. VB Growth Parameters K L-infinity (mm) (1) corr(L_inf, K) = -0.6815

9. Growth of diploid Crassostrea ariakensis • apply the growth rate for C. virginica, but with an extended growing season through the winter months

10. Data for estimating Mortality • Maryland fall survey, 1991+ • Used ‘recent’ and total box counts • The category of “box” includes dead oysters with shells still articulated • “recent” include gapers, in which tissue is still found within the shell, as well as boxes with no fouling or sedimentation on the inner valve surfaces • “old” (boxes in which fouling and/or sedimentation is found on the inner valve surfaces and no tissue remains).

11. Assumptions when using box counts for estimating mortality • ‘Old boxes’ -- • assumed to represent mortality within the last year prior to October survey; • ‘Recent boxes’ -- • assumed to represent mortality for ~2 weeks period prior to October survey; • Yearly mortality mostly occur from May to October (20 weeks)

12. Limitations of using box counts or size-class cohort analysis: • Older boxes -- • may represent mortality over 1+ years; • Transition between size classes due to growth not accounted for; • Less separation between disease tiers • Recent boxes – • Time since death can only be defined approximately • Cohort analysis • Lack info on age; cohorts overlap

13. Classified years by disease level

14. Mortality: Small and market sizedCrassostrea virginica • Empirical estimates of mortality by salinity (ppt) and disease tier • Salinity classes: high ( 15 +), medium (11-15), low (<=11) • Disease intensity: Tiers 1-3 • Likelihood of disease Tier determined by type of year (Wet, Average, Dry)

15. Mortality of small oysters Crassostrea virginica

16. Mortality of market sized oysters Crassostrea virginica

17. Mortality by year for MD Crassostrea virginica (across salinity zones)

18. Disease intensity tier of year by type of weather

19. The disease tier for a given year is randomly assigned based on the type of weather regime

20. MSX events forCrassostrea virginica • MSX event assigned when two or more dry years occur in a row • estimated from Maryland DNR historic disease data

21. Mortality, high MSX intensity

22. Increased mortality due to MSX events Crassostrea virginica • Increased baseline mortality by 10% -points for bars with high MSX events

23. Density-dependent mortality • Under development, with input from scientific review panel • Currently assume a maximum of 300 oysters per m2 • If the density at a bar exceeds this following spat-fall, the oysters will be assigned a uniform density dependent mortality across all size groups to scale back the density to the threshold.

24. Natural Mortality of Disease TolerantCrassostrea virginica “Standard” =Tier 3 From Calvo et al. (2003)

25. Natural Mortality of Disease TolerantCrassostrea virginica From Calvo et al. (2003) “Standard” =Tier 3

26. Mortality of disease tolerant C.va (Calvo et al. 2003) in high salinity

27. Natural mortality of disease tolerantCrassostrea virginica: limitations • Estimates are based on off-bottom cage experiments • Mortality due to predation is not fully accounted for • Is it reasonable to assume that the disease tolerance is maintained in future generations, after cross-fertilization with standard oysters?

28. Natural Mortality for aquacultured triploid C. ariakensis (off-bottom)

29. Natural Mortality for introducedC. ariakensis • Apply tier 3 mortality rates for C.virginica • Assume minimal mortality due to Dermo and MSX • Assume similar predation mortality as for C.viriginca • Sensitivity analysis will involve increased predation mortality due to thinner shells

30. Harvest Mortality • Exploitation rates by spatial area and year for each alternative • Provided by DNR

31. Empirical Stock-recruitment function forCrassostrea virginica, Maryland

32. Estimating Recruitment for Crassostrea ariakensis • The number of eggs produced per oyster by shell height: • Based on data from Taylor Hatchery and Allen and Merritt (2004)

33. Estimating Recruitment for Crassostrea ariakensis • Estimate the standardized spawning stock: • Divide the total number of eggs for spawning stock by the average number of eggs produced by a 77mm C. virginica oyster • Apply stock-recruitment function to estimate # spats • Assume that cumulative natural mortality from egg to spat (in October) is the same as for C. va

34. Spatial distribution of spat • The larval transport model (North et al.) provides estimates of the spatial distribution of spat that survive from eggs released from each bar in the Chesapeake Bay

35. Starting population of oysters for model projections out to 2015 • Survey data from 2004 used to define population for C.va: • MD survey data used for spatial distribution by size • VA survey data by bar • Number of stocked oysters & locations by year as provided by agencies

36. Linked Modeling Strategy North et al. Juvenile/adult demographic model Circulation models Predictions river flow mean abundance high Larval transport model low Settlement at each oyster bar time

37. Outside suitable habitat: continue swimming Larval Transport Settlement Model Dead Incorporates habitat data from MD DNR’s Bay Bottom Survey Inside: settle Oyster bars in 1980s Present day oyster bars Choptank River (Smith et al. in press)

38. Distribution of spat Particles will be released from 2,000+ habitat polygons in circulation model boundaries Modeled particle behaviors will be based on C. virginica and C. ariakensis laboratory experiments Simulations will be conducted with predictions from two Chesapeake Bay hydrodynamic models (ROMS and QUODDY) Blue line is QUODDY model boundaries Black shapes are oyster habitat polygons

39. C. virginica C. virginica C. virginica Larval Transport Model Strategy Step 1: Release particles from each oyster polygon Step 3: Determine which particles settle successfully on polygons Step 4: Determine the number of particles that start and end on each polygon for input to demographic model Step 2: Track change in location due to currents and larval behavior

40. C. virginica C. virginica C. virginica Larval Transport Model Strategy Step 1: Release particles from each oyster polygon Step 3: Determine which particles settle successfully on polygons Step 4: Determine the number of particles that start and end on each polygon for input to demographic model Step 2: Track change in location due to currents and larval behavior

41. Larval transport model will be run for 1995 – 1999 to capture years with different physical conditions 1995 1996 1997 1998 1999 dry wet ave dry wet

42. Forward projections in demographic model replicate past weather patterns by simulations Scenario 1: bootstrap, 5 year blocks from 1935-2005 Scenario 2: random selection from recent 10 years

43. Model Runs • Start with baseline run for c. virginica • Scientific review • Additional runs for alternatives with C.ariakensis after review

44. Model output • Number and biomass of oysters by size class; • By habitat polygon • By NOAA code/Chesapeake Bay segment • By State

45. Acknowledgements • Tom O’Connel and Phil Jones, DNR, for technical support and project management

46. Acknowledgements • Chris Judy and Mitchell Tarnowski (Maryland DNR) provided information on available oyster habitat & survey data for estimating mortality and recruitment; • Elizabeth North et al. for info on larval distribution • PIs on MDNR funded research; • Kelly Greenhawk, GIS analysis to delineate habitat

47. Independent Oyster Advisory Panel Panel’s Charge: • Review the adequacy of data and assessments used to identify the ecological, economic, and cultural risks and benefits, and associated uncertainties for each EIS alternative; • Provide advice on the degree of risk that would be involved for each EIS alternative if a decision were made in 2005 based on the available data and assessments; and • Recommend additional research, and associated timeline, that could be obtained to reduce the level of risk and uncertainty. Membership • Brian Rothchild • Jim Anderson • Mark Berrigan • Maurice Heral • Roger Mann • Eric Powell • Mike Roman