The need for quantitative descriptors of temporal change has been increasingly recognized as a fundamental requirement for ecological assessment and global fisheries management initiatives. Scientific studies that use novel syntheses of past and largely anecdotal data sources to reconstruct ecological conditions on decadal and century scales are now landmark contributions to ecosystem and conservation science (e.g., Jackson et al. 2001). The development of proxies for assessing ecological change has its roots in paleobiology, oceanography, and climatology, where discrete measurements (e.g., trace chemicals, abundance, presence—absence of fossilized organisms or parts) have enabled characterization of environmental and biological conditions on century to millennial scales. Development of similar rigorous measurements of past ecological conditions has proven more difficult, especially for large animals. Syntheses of anecdotal observations has led to more qualitative measures of past ecological conditions (e.g., top consumers were more abundant in the past; Jackson et al. 2001; Pandolfi et al. 2003) that are of limited use for establishing restoration and stock rebuilding targets and have been viewed with skepticism by many scientists (e.g., Boesch et al. 2001; Aronson et al. 2003). Extrapolation of quantitative measurements from anecdotal sources, surveys (Mora et al. 2005), or public interviews can be biased by generational changes in personal perceptions of environmental baselines (Pauly 1995; Sáenz-Arroyo et al. 2005) and often provide only limited quantitative metrics for assessing change. Re-examination and parameter estimation of past fishery-dependent (Baum et al. 2005), fishery-independent (Myers et al. 2007), and ecological (Paddack et al. 2009) data sets not originally collected for quantitative assessments have indicated declining trends for many fisheries but have also been met with substantial criticism (Hilborn 2006). Ideally, decadal or century data sets that can be analyzed in a similar context as the original intended use would be the most valuable. Although, few such data sets have been identified by marine ecologists and fisheries scientists, records of size or weights may fulfill this criterion for many species that are exploited (e.g., McClenachan 2009).

A key goal of restoration and stock rebuilding plans for many species is the proliferation of older (and hence larger) individual in populations. For many species, particularly marine fishes, conservation of larger individuals can be expected to increase per capita reproduction rates because larger fish can put more energy into reproduction and may produce eggs with higher probabilities of survival (Berkeley et al. 2004a, b). Reversing the decades-long trend of reduction in average size of fish will require aggressive management intervention (Birkeland and Dayton 2005). Unfortunately, many efforts have been hampered by disagreement among fisheries scientists, fishermen, and conservationists on the magnitude, timing, and consequences of these declines (Baum et al. 2003; Baum and Myers 2004; Burgess et al. 2005; Hilborn 2006). Detailed records of exploitable fish and shellfish populations that could be used to resolve these disagreements and set definitive goals are unavailable or unreliable to reconstruct the history of most fisheries — a point multiple scientists agree upon (Baum and Myers 2004; Burgess et al. 2005). The lack of quantitative data also hampers the establishment of specific quantitative conservation and stock rebuilding goals.

Although systematic and rigorous collections of fisheries dependent and independent data are a relatively recent endeavor in the USA, fishermen have collected fish from coastal and marine waters for millennia. Size and weight are often measured by fishermen to gauge their success among their peers. When these competitive interactions are organized into fishing tournaments, a potential historic record for the sizes of the largest catchable fishes is recorded in fishing journals and newspaper articles. We used data sets generated annually by generations of fishermen participating in the three oldest and largest fishing rodeos in the northern Gulf of Mexico to examine whether the size (weight or length) of large sharks has decreased over time. These data sets do not meet the rigorous criteria for incorporation into traditional stock assessment because they are fundamentally biased. In tournaments, fishermen do not sample in a random manner, they target a specific size (largest) and use experience and traditional knowledge to direct their efforts. Although these biases must be considered in any analyses, many of these biases may facilitate ecological investigations. For example, rigorous fisheries independent sampling rarely captures sufficient quantities of large and older fish to resolve trends in the largest size-classes. In fishing tournaments, a tremendous effort is focused on sampling this largest size-class and, hence, could provide important information on trends while standardizing many aspects of fishermen behavior (e.g., time of year, geographic area, etc.) via tournament rules.

METHODS

We used 80 years of records (1929–2009) from the three oldest fishing rodeos within the USA, all of which occurred in the Gulf of Mexico: (1) the Alabama Deep-Sea Fishing Rodeo (ADSFR; spanning 79 years), (2) Mississippi Deep-Sea Fishing Rodeo (MDSFR, 60 years), and (3) Texas Deep-Sea Roundup (TDSR; 76 years). We examined trends in the weight (ADSFR, MDSFR), length (TDSR) and species identity of the largest sharks caught by fishermen as an indicator of the relative abundance and size of large sharks within the Gulf. All three rodeos have similar characteristics, including timing (late July to early August), length (3 d), and rules (all sharks aggregated into a single category and three largest fish win prizes). Effort can also be assessed by examining trends in ticket sales.

Long-term tournament records for sharks and other big game fish from the three rodeos, as well as NOAA commercial fisheries data, were gathered and synthesized into one database for analysis. Rodeo records were reconstructed from daily leaderboards (Figure 1) published in the Mobile Press-Register (ADSFR) and Mississippi Sun-Herald (MDSFR), as well as archived weigh-station registry cards (TDSR). Final leaderboard and registry-card data were entered in a digital database, and the largest (TDSR) or three largest (ADSFR, MDSFR) sharks from that year's rodeo were entered. For the ADSFR and MDSFR, weights were used to determine winners, whereas length was used for the TDSR. Shark identifications for the last 15–25 years of each tournament were made by fish biologists on site, whereas earlier identifications were based on the examination of newspapers and archived photos and additional taxonomic experts as needed. In addition to trends in shark sizes, we also compared long-term trends in ADSFR data between sharks and other big game fishes (Blue Marlin Makaira nigricans, King Mackerel Scomberomorus cavalla, Wahoo Acanthocybium solanderi and Yellowfin Tuna Thunnus albacares) that require specialized (heavy) tackle and experience. Weights for the three largest specimens of each species for each year were averaged. To make direct interspecific comparisons, weights were standardized (within each species) by scaling records against the year in which the heaviest fish was observed. Therefore, annual data for each species scale between 0 and 1. This comparison was used to evaluate whether observed shark trends were specific to these elasmobranch fishes or rather part of a larger phenomenon resulting from changes in either (1) rodeo practices, or (2) the ability of the Gulf of Mexico to support large apex fishes. Finally, annual commercial landing statistics for all sharks within the Gulf were obtained by querying the NOAA-Fisheries database.

FIGURE 1.

Example accounts from Mobile Press-Register in (A) 1962 and (B) 1998 describing deep-sea fishing rodeo results of prize-winning sharks, which were the data source used to construct temporal shark size data sets for the last eight decades in the Alabama and Mississippi rodeos. Newspaper accounts were compiled from the final rodeo leaderboards at the end of each year.

Individual long-term trends in shark catches were fitted for ADSFR, MDSFR and TDSR data. In each instance, we used the Akaike information criterion (AIC) to determine the model order that provided the best goodness of fit for the data (balancing model specificity and generality): AIC = 2k + n[log_e(RSS/n)], where k is the model order, n is the number of observations, and RSS is the residual sum of squares between the observed and fitted data. The annual standardized catch records for the four big game fishes were averaged, and these data were compared with the long-term data for sharks. As before, we used AIC analyses to determine the best fit for the big game fishes time series.

Our historical reconstruction of the sizes of coastal sharks was supplemented by interviews of anglers participating in the 2008 and 2009 ADSFR. Approximately 400 anglers were indiscriminately selected upon exiting the weigh-in booth at the rodeo. Overall, selection was driven by the desire to achieve dispersion across ages and recreational fishery experience. The objective of the interviews was to measure perceptions of size trends in winning species of fish. In addition to a series of demographic and socioeconomic questions, fishermen were asked several species-specific questions (see Appendix) regarding size trends in major fish categories for the ADSFR and asked to use one of five different trend lines to describe their opinion of how size has changed over time.

RESULTS

Despite increasing fishermen effort over time, dramatic changes were seen in the size of the winning sharks in recent years. The average size of the three largest sharks captured in all three rodeos increased from the rodeos' inceptions until the early 1980s, but decreased by >60% in the late 1980s and remained low through 2009, the last year included in the data set (Figure 2A–C). Choosing the fit producing the lowest AIC value, rodeo trends based on weight were best fit by a cubic polynomial with R² values of 0.56 for ADSFR and 0.48 for MDSFR (P < 0.01). The TDSR data showed the same trend based a quadratic polynomial of length data (R² = 0.53, P < 0.01). Effort in the two rodeos that documented ticket sales (ADSFR and MDSFR) has increased linearly (R² > 0.79, P < 0.01) since their inceptions and increased from hundreds of fishermen to greater than 4,000 today. The substantial decrease in size of the prize-winning sharks was not seen in other big game fish. A linear increase in size of big game fish is seen in the ADSFR data (R² = 0.78, P < 0.01; Figure 3). In contrast, shark size increased until the late 1980s and then abruptly decreased.

Species composition of the winning large sharks also changed over the 80 years of rodeo records. Large Tiger Sharks Galeocerdo cuvier and Bull Sharks Carcharhinus leucas predominated the winner boards until 1990, whereas much smaller Bull Sharks, Hammerheads Sphyrna spp., and Blacktip Sharks Carcharhinus limbatus predominated during the last 20 years (Table 1). Species-level identifications for sharks were rarely included in the newspaper accounts of the rodeos prior to 1970. Winning sharks for the ADSFR were almost exclusively Tiger Sharks prior to the 1990s. Post 1990 Tiger Sharks occasionally made the leaderboard, but their weights were considerably less. For the nearby MDSFR, large Tiger Sharks followed by Bull Sharks composed the majority of the winning sharks until 1990. From 1990 to 2007, each species made the winners board only once. Instead, smaller Blacktip Sharks and Hammerheads made the leaderboard, but their weights were considerably less than that of the prize winning sharks captured during the 1970s and 1980s. The Texas rodeo showed a similar trend, predominated by Bull Sharks and occasionally Tiger Sharks pre-1990 and a larger diversity of smaller species of sharks post-1990.

Interviews with over 400 fishers (>90% participation rate) conducted at the ADSFR indicate that the retrospective memory of most fishermen would have predicted an increasing or stable trend in size of winning sharks. Less than half the fishermen would predict a decreasing size of sharks based on their aggregate knowledge of the fishery (Figure 4). Among the 215 anglers that answered the shark-related questions and provided their age, the percentage of anglers predicting a declining trend increased with age of the respondent. The only group that had a majority of respondents describing a negative trend was composed of individuals over 60 years of age.

DISCUSSION

The inability to examine long-term (decadal) trends for shark populations has imposed a high degree of uncertainty on the status of sharks. Our fishermen-generated data are evidence for a drastic reduction in the size of large sharks in the Gulf of Mexico by the early 1990s. Reduction in the size of these sharks has substantial population-level and ecosystem-level implications. Several aspects of shark life history indicate a greater sensitivity to overfishing and prolonged population recovery time (long maturation period, low fecundity, older ages at first reproduction; Music 1999). For example, none of the Tiger Sharks caught in the last 20 years would have been classified as reproductively mature (> 150 kg; Branstetter et al. 1987; Whitney and Crow 2007). For those species of sharks that mature at smaller sizes, recovery of shark populations may be delayed if a positive correlation between size and reproductive fitness exist, as is the case for bony fish (Berkeley et al. 2004a, b). Reduction in the abundance of large sharks also has the potential to cascade down the food web by releasing mesopredators from top—down control (Myers et al. 2007).

FIGURE 2.

Temporal trends for the three heaviest or longest sharks landed annually during the last eight decades in the (A) Alabama (ADSFR), (B) Mississippi (MDSFR), and (C) Texas (TDSR) deep-sea fishing tournaments. For each state, the annual records (symbol ± 1 SE)and overall trends (solid lines representing cubic [ADSFR, MDSFR] and quadratic [TDSR] fits for the data) are shown. (D) National Marine Fisheries Service (NMFS) records showing changes in commercial landings of sharks during the 1980s in the Gulf of Mexico (GOM), which coincides with a decline in the size of prize-winning sharks available to tournament fishermen from around 1990 to the present.

FIGURE 3.

Temporal trends in the weights of prize-winning sharks and big game fishes captured each year in the Alabama Deep-Sea Fishing Rodeo. Big game fishes include Blue Marlin, King Mackerel, Wahoo, and Yellowfin Tuna. Weights were standardized (within each species) by scaling the records against the year in which the heaviest average weight of prize fish was observed. The decrease in shark weights following the mid-1980s is counter to this expectation.

FIGURE 4.

Perceptions among three generations of fishers regarding the long-term trend in sizes of sharks available to be caught in the northern Gulf of Mexico (n = 215); a value >0.5 would indicate that more than half of fishermen surveyed believed that shark size has declined.

TABLE 1.

Species identification of winning sharks in the Alabama (ADSFR), Mississippi (MDSFR), and Texas (TDSR) deep-sea fishing tournaments, 1947–2009.

The pattern of declining size of sharks is largely driven by decreases in the occurrence and size of Bull and Tiger Sharks. Whereas the leaderboards were predominated by these two species prior to the late 1980s, the winning sharks came from a larger pool of species that included smaller coastal species (e.g., Blacktip Sharks, Spinner Sharks, and Scalloped Hammerheads S. lewini) as well as smaller Tiger and Bull Sharks. Because so few species besides Tiger and Bull Sharks won the earlier tournaments it is difficult to assess whether our pattern is exclusive to these two species or encompasses other coastal sharks. The occurrence of large Hammerheads in the earlier records of the TDSR does suggest that the pattern encompasses more than just Tiger and Bull Sharks. From an ecological perspective, the answer may be somewhat irrelevant given that Tiger and Bull Sharks as well as Hammerheads are common large (>150 kg) sharks that fill the niche of large mobile predators in the coastal foodweb in the Gulf of Mexico (see Drymon et al. 2010), and their reduced prominence would be troublesome in and of itself.

The interpretation of our data sets requires reasonable assumptions be made regarding the efficiency of fishermen and tournament behavior to make inference concerning trends in shark size. The initial 50-year increase in weight is most likely a function of increasing effort and improved fishing technology as evidenced by a similar pattern in big game fish (Figure 3). Substantial fisheries literature exist documenting increases in the efficiency of fishermen with both increases in technology and communication among fishermen. For fisheries assessments, catchability (q) is used to quantify the relationship between the efficiency of a fishery (in our case, recreational fishermen) and population abundance (Arreguín-Sánchez 1996). Given an increasing or unchanging efficiency, decreases in landings would be a function of changes in abundance. The substantial decrease in size of prize-winning sharks is not seen in other big game fish, species that would require similar size boats and specialized tackle. The departure of sharks from the increasing or asymptotic relationship provides further evidence that large sharks were in decreased supply to anglers.

Fishermen behavior cannot be explained exclusively by q. The popularity of sharks as a targeted species, as well as the financial cost of fishing these species, must also be considered during the period over which decline has been documented. The number of fishermen in these multispecies tournaments increased twofold over the period of the decline (1990–2009). Ticket sales do not fully resolve trends for the sector of fishermen targeting sharks. The popularity of shark fishing increased in the mid 1970 through 1980s after the movie “Jaws” (Babcock 2008) and may have resulted an increased fishermen participation in this sector for a period of time. As populations declined and conservation of sharks received increasing public attention, shark-kill tournaments probably declined (Hueter 1991), although some tournaments adopted tag-and-release rules. None of our three tournaments would be considered major shark tournaments; instead their multispecies categories and modest entry costs and prizes (<$200 in value) attracted local fishermen who opportunistically targeted sharks. Although most relied on chance encounters while fishing for other finfish species, some anglers did routinely target sharks within the tournament and were less successful at capturing large sharks later in the time series. Fuel prices, which are often a consideration in evaluating offshore fishing effort, sharply increased in the mid-1970s and again in the late 2000s (U.S. Department of Labor, Bureau of Labor and Statistics). In the intervening years between these two spikes, a modest and steady increase was seen in fuel prices and, hence, did not cause an abrupt and major change in shark fishing in the tournaments, a notion supported by increased tickets sales and increased size of big-game fish.

Similar to our conclusion regarding fishermen behavior (i.e., no plausible explanation for changes could parsimoniously explain the abrupt decline in shark size seen in the early 1990s), no major rule changes to the tournaments occurred during the period of decline that could account for the shark size pattern. Regulation and harvest prohibitions for shark species in the Gulf of Mexico are relatively recent. Prohibitions on recreational harvest of Sandbar Sharks Carcharhinus plumbeus, Dusky Sharks Carcharhinus obscurus, and Silky Sharks Carcharhinus falciformis did not occur until early in the 2000s. Further, the dramatic decline in shark size in our three data sets occurred before stock assessments were commenced for large coastal sharks in the northern Gulf. Hence, no major changes in fisheries regulation could explain the decline.

Given the basin-wide nature of the pattern we detected, climate change or longer-term oscillations of climate could be plausible drivers. For instance, recent studies have shown that significant variability in blue crab Callinectes sapidus recruitment within the Gulf of Mexico could be explained by the oscillation between wet and dry periods driven by the a combination of the effects of the Atlantic Multidecadal Oscillation and the North Atlantic Oscillation (Sanchez-Rubio 2011a,b), but the major shifts in these indices occurred in 1994, which postdates the rapid decline in size of sharks in the late 1980s. The findings of another recent study in the region suggested that the structure of northern Gulf of Mexico seagrass fish communities may have already been affected by warmer water temperatures (Fodrie et al. 2010). In any respect, an altered prey base could have potentially slowed a recovery of sharks.

While changes in angler behavior, tournament rules, fishing regulation, and climate oscillations fail to provide strong temporal correspondence with the precipitous decline in shark size evident in the three fishing rodeos, the decreasing trend in shark size shows a distinct temporal correspondence with increased landings in the shark longline fishery in the Gulf of Mexico. The spike in landings of sharks occurred in the late 1980s and early 1990s as a result of increasing effort in the commercial longline shark fishery in the Gulf. Many of the older tournament anglers noted the temporal correspondence between decreased landings of large sharks and increases in the commercial harvest: “Long liners have made it progressively more difficult to catch big sharks” (quote from an unidentified fisherman interviewed in the Mobile Press Register, 1986).

Differences among generations were consistent with predictions of the “shifting baseline” theory in that older fishermen viewed changes in size of sharks more accurately than younger fishermen who felt that shark sizes had not changed over the last 9 decades. Misperceptions of such trends by many fishermen not only illustrate the need for assessing age and experience level in social surveys of ecological conditions, but more importantly point to the need for quantitative measures of change. We anticipate that the source of these data (i.e., fishermen) will serve to promote greater acceptance of the current status of shark populations by resource managers and the general public, and it illustrates the importance of incorporating traditional knowledge in a quantitative way.

After a long history of encouraging harvest (Barrett 1928) and an absence of harvest regulations, significant improvements have been made in stock assessment and have resulted in regulations to rebuild shark stocks. It is important to note that the changes we document in the average size of sharks occurred prior to stock assessments for most coastal sharks (e.g., Cortés et al. 2006; NMFS 2006, 2008; Hayes et al. 2009). Improved management is critical to recovery of these stocks. Our data set provides discrete and tangible rebuilding targets that could be achieved. It is somewhat ironic that fishing tournaments that target sharks, which may be part of the reason for the decline, are one of the best indications of their decline; however, this irony extends to most exploited species. Our data suggest that the commercial long-line fishery was the primary driver for the decline in size of sharks and that the decline is relatively recent (1990s). Declines in apex predators can alter entire food webs (Estes et al. 2011), and increasing the abundances of the largest members of the apex predator community should be a fisheries and ecosystem management priority.