Length-Weight Models for Intertidal Clams in Puget Sound (Bivalve Regions 1, 5, 6, 7, and 8)

On Washington State public tidelands, intertidal clams are jointly managed by the Washington Department of Fish and Wildlife (WDFW), treaty tribes, and the Washington Department of Natural Resources. The state is divided into eight Bivalve Regions for management purposes (Figure 1). Annual plans negotiated between the state and tribes for each of these Regions define the biological survey methods, management procedures, and annual TACs (Total Allowable Catch) for the sport and commercial fisheries. Currently, TACs are calculated for only two clam species -- Manila clams (Tapes philippinarum) and native littleneck clams (Protothaca staminea) â€" but several other â€�"passively managed” clam species are included in the survey estimates of biomass on these beaches. From the standpoint of sport and commercial fisheries, the most important of these other species are butter clams (Saxidomus giganteus), cockles (Clinocardium nuttallii), and eastern softshell clams (Mya arenaria). Two other passively managed intertidal species, geoducks (Panopea abrupta) and horse clams (Tresus spp.) live deep in the substrate and are thus not likely to be sampled in proportion to their abundance.

The annual TACs for Manila and native littleneck clams are calculated based on biological survey estimates of the clam biomass on individual beaches. WDFW and the treaty tribes conduct these surveys using similar but not identical methods. WDFW survey methods are detailed in a departmental Procedures Manual (Campbell 1996), and tribal methods are outlined in at least two technical reports and memoranda: Point No Point Treaty Council (1998) and Fyfe (2002).

All state and tribal surveys currently performed in Bivalve Regions 1, 5, 6, 7, and 8 involve a systematic random sampling design that covers either the entire public beach or the "productive" portion of the beach (i.e., the portion of the beach containing significant clam resource). Numerous sample digs â€" either 0.0929 m² (1 ft² ) for state surveys or 0.1858 m² (2 ft² ) for tribal surveys â€" are taken from the beach at low tides, during which most or all of the Manila and native littleneck habitat can be sampled. The mean density of clams per unit area is estimated for the beach, and total biomass is then estimated as the product of mean density, average weight per clam, and the area of the beach.

The only survey method that differs significantly between WDFW and tribes is the procedure used to estimate the average weight per clam. WDFW transports sampled clams to the laboratory and individually records the shell length and weight of each clam; mean weight per clam is therefore estimated as the average of all unbroken clams in the sample. Most tribal surveys, on the other hand, involve measuring the shell length of all sampled clams on the beach, and later applying an allometric length-to-weight model to estimate the weight of each clam.

Tribes that estimate clam weights from lengths rely on two models. For Manila clams, tribes use a length-to-weight model originally published as Appendix E of a Sea Grant Technical Report on Manila clam aquaculture in Puget Sound (Anderson et al. 1982; reproduced here as Appendix Table 1). The exact same model appears as Appendix F in the revised edition of the Sea Grant report (Toba et al. 1992). In this model â€" which henceforth we refer to as the "Sea Grant model" -- live weight of a clam is estimated as W = 0.0001433L^3.11, where W is weight in grams and L is length in millimeters. Unfortunately, neither the geographic source of the Manila clam samples, the sample size, nor the statistical error terms for this model are cited in either report.

For native littleneck clams, tribes rely on a computer spreadsheet developed by the Northwest Indian Fisheries Commission (Appendix Table 2). This spreadsheet was based on "…length/weight data for littleneck clams provided by the Washington Department of Fish and Wildlife…" (Point No Point Treaty Council 1998). Unfortunately, no details are cited, and WDFW biologists can no longer provide the data upon which this model is based. Presumably, it was estimated from native littleneck length-weight data taken during routine beach surveys. But, as with the Manila model, no information is available on the geographic source of the native littleneck samples, the total sample size, or the statistical error terms of the model.

When using these length-to-weight models, managers make the tacit assumption that the model reliably estimates the weight of sampled clams on each surveyed beach. If this assumption is violated, the result will be either an overestimate or underestimate of the clam biomass on a given beach. In turn, unreliable biomass estimates produce TACs which are either undesirably high or low. In recent years, several tribal biologists have made empirical comparisons between Manila clam weights which have been measured in the field or laboratory and those derived from Anderson et al. (1982) model cited above. These comparisons raised doubts as to the reliability of the model on at least some beaches. The Skokomish Tribe abandoned the use of both the Sea Grant Manila clam model and the NWIFC native littleneck model beginning in 2002. Skokomish biologists began weighing sampled Manila and native littleneck clams taken during their beach surveys in order to estimate mean weight per clam. The Squaxin Tribe followed suit in 2003. Most tribes, however, are still using the Manila and native littleneck models cited above to estimate clam biomass on surveyed beaches.

In this report, we examine length and weight data from clams routinely sampled by WDFW in public beach surveys from 1994 through 2002 and within Bivalve Regions 1, 5, 6, 7, and 8. Our objective is to estimate beach-specific length-to-weight models for Manila clams, native littleneck clams, butter clams, cockles, and eastern softshell clams. We also examine the statistical confidence bounds of these beach-specific models and produce pooled regional models for each species where appropriate. We then make empirical comparisons using actual beach survey data for Manila and native littleneck clams; here we compare biomass estimates based on actual weighed samples against biomass estimates generated using length-weight models. We also test at selected beaches our assumption that length-weight relationships did not change significantly over the nine-year sampling period of this study. Finally, we make practical recommendations on how to make the most reliable estimates of mean weight per clam from survey data.