|Stock Assessment of Connecticut River Shad: Examination of Fishing and Predation Effects on the Recent Stock Decline
Victor Crecco, Thomas Savoy and Jacqueline Benway
Connecticut Marine Fisheries Division
333 Ferry Rd.
Old Lyme CT 06371
December 21, 2006
Report to the American Shad Stock Assessment Subcommittee
American shad (Alosa sapidissima) landings and run sizes have fallen steadily in the Connecticut River since 1993 despite relatively high and persistent juvenile production since 1990. In this assessment, several analyses are conducted to determine the effects of fishing and predation on the recent stock decline of Connecticut River shad. In addition, ocean recreational shad landings and ocean commercial discards were included in the fishing mortality estimates, and equilibrium and non-steady state overfishing thresholds (Fmsy, Nmsy) were re-estimated for Connecticut River shad. Total age aggregate (ages 4+) fishing mortality (FT) estimates on shad, based on the ratio of combined riverine and ocean landings to stock size, were highly variable from 1981 to 1994, ranging from 0.14 to 0.47. Only in 1986 and 1987 did the FT levels slightly exceed the F30% overfishing threshold of 0.43 established for Connecticut River shad from the last peer reviewed stock assessment (ASMFC 1998). After 1995, the FT estimates on ages 4+ shad fell by 40 to 50%, to below 0.20 in most years, culminating in the lowest FT of 0.11 in 2001. All fishing mortality (FT) rates from 1996 to 2005 were 50% below the overfishing threshold (F30% = 0.43). This indicated that the recent drop in shad run size was not due to overfishing. From 1981 to 1993 adult shad run size in the River remained high and relatively stable (> 600,000 fish), but after 1993, shad run size fell steadily thereafter to the historic (since 1965) low level of 226,000 fish in 2005. Note that the preliminary 2006 run size of 290,000 adult shad is the second lowest ever recorded since 1965.
Shad juvenile indices monitored annually in the River have varied without trend from 1981 to 1992. Since 1996, juvenile indices have been at or above the long-term median index despite the persistent drop in adult stock size from 1996 to 2005. The sudden and unexpected lack of coherence between recent juvenile indices and subsequent adult recruitment from those year-classes indicated the emergence of a recruitment bottleneck after 1996. Since fishing mortality rates on adult shad have fallen to low levels since 1995, Savoy and Crecco (2004) hypothesized that this recent failure in shad productivity was largely due to a systematic rise in striped bass (Morone saxatilis) predation on adult shad in the River. Results from 2005 and 2006 striped bass dietary studies (Davis 2006 in prep) in the River are wholly consistent with the Predation Hypothesis. Davis (2006 in prep) reported that large (> 90 cm) striped bass sampled in the River south of the Holyoke Dam during spring of June, 2005 and 2006 fed extensively on pre-spawned adult shad, whereas smaller stripers (size range: 50-89 cm) consumed large numbers of pre-spawned blueback herring. Some 46% of the diet in weight (gm) of stripers between 90 and 99 cm was composed of adult shad, whereas 82% of the diet (gm) of 100 cm+ stripers was made up of adult shad.
The age aggregated Steele and Henderson (S-H) production model indicated that striped bass consumption rates (Mp) on adult shad rose in the River fourfold after 1994 coincident with a steady rise in striped bass abundance. Estimated adult shad consumed annually by striped bass from 1999 to 2005 were 5 to 15 times greater than the in-river landings during those years. The approximate equilibrium Fmsy level of 0.39 for shad based on the S-H model easily exceeded the total fishing mortality (FT) levels on adult shad in all years from 1989 to 2005. Non-equilibrium Fmsy levels approached 0.50 in most years from 1981 to 1993, but annual Fmsy levels fell steadily thereafter to 0.02 in 2004 following a steady rise in striped bass consumption rates (Mp). Statistical and empirical evidence given here strongly suggests that the recent emergence of a recruitment bottleneck and the subsequent decline in adult shad run size in the Connecticut River were linked mainly to predation effects from striped bass. The management implications of successful stock rebuilding of Connecticut River shad in the presence of rising predatory mortality are discussed.
The most recent peer-reviewed assessment of American shad (ASMFC 1998) indicated that adult run size in the Connecticut fell steadily from 1992 to 1996. Average aggregated (ages 4+) fishing mortality (FT) estimates on Connecticut River shad, based on the ratio of combined coastal and riverine landings to stock size, varied without trend from 1988 to 1996. Moreover, all FT levels except the 1987 estimate were below the F30% overfishing threshold of 0.43 from the last peer reviewed assessment. Thus the most recent stock assessment (ASMFC 1998) concluded that the recent drop in shad run size in the Connecticut River was not due to overfishing. Although this assessment (ASMFC 1998) clearly demonstrated a pronounced decline in shad abundance after 1992 among several shad stocks along the Atlantic coast, this and all previous assessments have yet to examine the potential impact of non-fishing effects on the recent drop in shad productivity.
Nearly all anadromous finfish assessments, including previous American shad assessments (ASMFC 1988, 1998), assume that natural mortality is constant across all ages (ages 3+) and years. The constant M assumption is thought to be unrealistic under most conditions, particularly for American shad that are susceptible to a wide variety of finfish predators (Savoy and Crecco 2004) and that usually experience very high post-spawning mortality (Leggett et al 2004). Yet the constant M assumption is widely accepted in virtually all stock assessments, because time varying M is often difficult to estimate with confidence and because a constant M assumption greatly reduces the number of parameters to be estimated in conventional VPA models. Although not explicitly stated in most assessments, the constant M assumption implies that predation, inter-specific competition and environmental effects on stock productivity are fixed in time. Thus, the constant M assumption greatly limits our ability to explore the significance of non-fishing effects on the recent failure in stock productivity. Non-fishing effects include enhanced predation that may result in a systematic rise in M (Wahle 2003), as well as temporal shifts in environmental factors that can adversely affect recruitment, somatic growth and maturation. Even subtle shifts in trophic and environmental factors (Link 2002) have been shown to confound stock rebuilding strategies of other depleted fish stocks (Rose et al 2000, Shelton et al 2006).
In this assessment update, several analyses are conducted on Connecticut River shad to determine the effects of fishing and predation on the recent stock decline. Our updated fishing mortality estimates from 1981 to 2005 include shad landings from the ocean recreational fishery, as well as discards from the Atlantic herring (Clupea harengus) commercial fishery. Although a thorough test of the Overfishing Hypothesis is our primary goal of this assessment, there is now an increasing body of evidence that suggests that enhanced predation by striped bass in the River has played a major role in the recent drop in shad run size. Savoy and Crecco (2004) recently used empirical and statistical evidence to show that the recent drop in shad population size in the Connecticut River was consistent with enhanced striped bass (Morone saxatilis) predation of adult shad. Moreover, Davis (2006 in prep) recently reported that adult shad dominated the diet of large (> 90 cm) striped bass in the River during spring 2005 and 2006. The striped bass is regarded as a voracious finfish predator from the Mid and North Atlantic on menhaden, gizzard shad, American shad and river herring (Walter and Austin 2003; Nelson et al 2005; Rudershausen et al 2005). Moreover, unlike many other marine finfish predators, adult (> 70 cm) striped bass have recently increased in New England waters to record high levels (ASMFC 2005), and have been sampled in large and increasing numbers well above the salt wedge in the Connecticut River since 1993 (Savoy 1995). Therefore, it would be useful to summarize these findings and present an analytical model that merges the population dynamics of American shad with the foraging characteristics of striped bass.
The age aggregated Steele and Henderson (S-H) production model (Steele and Henderson 1984) was used to further examine the effects of fishing and predation effects on the recent decline of Connecticut River shad. The S-H model has extensive theoretical appeal since it incorporates compensatory stock dynamics of the prey with fishing effects plus a sigmoid Type III foraging response by the predator that may lead to critical depensation at low shad abundance (Spencer and Collie 1997). Since the S-H model can easily accommodate environmental variables (Spencer 1997), this modeling approach represents a modest but straightforward attempt at ecosystem modeling. The S-H model was also used to estimate fixed (equilibrium) and time varying overfishing thresholds (Fmsy, Nmsy) for Connecticut River shad under temporal shifts in predation mortality. In addition, more robust and precise estimates of the overfishing thresholds (Fmsy, Nmsy) from the S-H model were derived through iterative reweighted least squares regression.