# Age Structure

It is desirable to describe populations by age structure in that this provides running quantitative indices of both recruitment and mortality. Unlike the vast majority of bivalves oysters do not have typical symmetry in their growth form, indeed the aggregated growth in reefs generally insures a variety of final growth shapes. Signature growth increments in the shell matrix, as seen in sections of the shell and/or hinge, are also variable and require care in interpretation. The approach employed herein uses predominantly demographic analysis bolstered by very large n values. It is, notably, in agreement with blind testing using both growth signature and isotope analysis (for discussion see Harding et al., 2008, 2010b).

As an historical note, demographic plots prepared for each year in the 1998-2006 period for each reef surveyed in the James River for both live oysters and boxes using 5 mm bins provided limited guidance on age versus length relationships. Distinct year classes of live oysters that could be followed for a minimum of three years were rare in these plots. The period of recruitment to the benthos (also commonly termed spatfall) in the James results in a broad size range within each year class such that interannual junctions are not distinct with a 5 mm size bin. Data for 2004 –2006 were, as mentioned earlier, available in 1 mm size bins. All data for these years for reefs with densities exceeding 100 oysters m^{-2} are in close proximity to each other and were aggregated on a single size frequency plot with 2 mm bins. The individual cohorts (not year classes, there being one or more cohorts in a single year class) were identified by the method of Bhattacharya (1967). The range and modal length of each cohort was identified by counting cohorts and relating the cohort settlement dates to long term recruitment patterns developed from annual spatfall reports for the James River over the study period (annual reports by Southworth et al), The cohorts were thus assigned to years and a linear age at length relationship fit to the data (y = mx + c). The fit for James River observations (Mann et al 2009b, Figure 6) is:

y = 30.22 + 21.6x; R^{2} = 0.93, n> 22,000

Using a July 1 birth date and noting that current data is for a fall survey then lengths on November 1 represent ages of 0.33, 1.33, and so on with annual increments, although for clarity throughout this text these are referred to as age classes 0 (= YOY = spat), 1, 2, 3 and >3 year olds. Corresponding lengths are 37.3 mm at 0 y, 58.9 mm at 1 y, 80.5 mm at 2 y, 102.1 mm at 3 y and 123.7 mm at 4 y. This age at length model was used to recast the length demographic as an age demographic and to estimate age specific mortality. Mann et al. (2009b) argue that a linear age versus length fit is appropriate for early years given the life expectancy of an oyster (10-15 years in undisturbed populations, Powell and Cummings 1985), the previously mentioned plastic form and the lack of adherence to isodiametric form with allometric b values generally nearer 2 than 3 (including those presented above). James River oysters are also distinctively long and thin rather than typically cupped in shape – they have been locally described as “pencils”. Such a form approximates to a tube rather than a sphere, again allowing for a linear fit. The adoption of an age-at-length plot using the von Bertalanffy (1938) model is not considered appropriate for these oysters. In addition to the linear (best fit) model for James River data a quadratic length versus age descriptor was developed for oysters in the Great Wicomico River by Southworth et al. (2010) and the Piankatank River by Harding et al. (2010). In both instances the quadratic fit gave a higher R^{2} value than the linear fit. Unlike the James River oysters, individuals collected from both the Great Wicomico River and the Piankatank River tend to be more cup shaped (not “pencils”) and a non-linear age versus length fit is supported. The quadratic fit is of the form SL = a*(Age)^{2} + b*(Age) + c.

For the Great Wicomico River: included years are 2003-2009. Values of a, b and c are as follows: a =-3.76, b = 38.077, c = 8.51. The corresponding SL values are 17.5 mm at 0 y, 52.1 mm at 1 y, 75.4 mm at 2 y, 87.4 mm at 3 y, 98.3 mm at 4 y.

For the Piankatank River: included years are 2003-2009. Values of a, b and c are as follows: a =-2.947, b = 32.6565, c = 14.456. Corresponding lengths are 24.9 mm at 0 y, 52.7 mm at 1 y, 74.5 mm at 2 y, 90.5 mm at 3 y, 100.4 mm at 4 y.

The age at length estimates for the Great Wicomico and Piankatank Rivers are slightly lower than those for the James River linear fit. Caution is required when extrapolating these fits in that the very nature of linear and quadratic fits are that they will diverge with increasing age. In this report the river specific fits are used to estimate age structure from demographic data for only those specific locations except when otherwise stated. We have not yet developed age at length estimators for the Rappahannock River, and Pocomoke and Tangier Sounds.