Assessments of threatened and endangered sea turtle populations evaluate their status and trends, as well as effects of conservation actions and anthropogenic and natural threats that influence their recovery (Seminoff and Shanker 2008; Committee on Sea Turtle Population Assessment Methods–National Research Council [CSTPAM-NRC] 2010). The Deepwater Horizon (DWH) oil spill (Belter 2014; Trustees 2016) in the northern Gulf of Mexico (GoM) during 20 April to 15 July 2010 heightened awareness of the need for assessing sea turtle population trends and demographic processes that drive them (Bjorndal et al. 2011; Putman et al. 2015).

Kemp's ridley (Lepidochelys kempii) is the most endangered of all sea turtle species. Its female population in the GoM suffered a major setback sometime between the ends of the nesting seasons in 2009 and 2010 (Crowder and Heppell 2011; Caillouet 2014). Prior to that, annual nests at the female population's index beach (Fig. 1) on the GoM coast of Tamaulipas, Mexico, were increasing exponentially (National Marine Fisheries Service [NMFS] et al. 2011), the result of more than 4 decades of cumulative conservation efforts on land and at sea. Annual nests dropped 35.4% in 2010 and remained well below predicted levels through 2014, and annual hatchlings released were lower in 2010–2014 than in 2009. Causes of the setback have not been determined with certainty (Caillouet 2014; Trustees 2016).

View Figure

• Download as Powerpoint
• Download Figure

The foundation provided by demographic modeling of the Kemp's ridley female population prior to 2010 (Heppell et al. 1996, 2005, 2007; Turtle Expert Working Group [TEWG] 1998, 2000; Coyne and Landry 2007; NMFS et al. 2011) facilitated subsequent demographic and stock assessment modeling aimed at elucidating the causes and consequences of the setback (Crowder and Heppell 2011; Gallaway et al. 2013, in press a, in press b; Gallaway and Caillouet 2014; Gallaway and Gazey 2014, 2015; Heppell 2014, 2015; NMFS and US Fish and Wildlife Service [US FWS] 2015). To date, all demographic and stock assessment models of the Kemp's ridley female population have been applied to time series of annual nests (i.e., clutches laid by multiple year-classes of nesters) and annual hatchlings released (by year-class, both sexes combined) on this population's index beach between Playa Dos-Barra Del Tordo and Barra Ostionales-Tepehuajes on the GoM coast of Tamaulipas (Fig. 1) (Márquez M. et al. 1989b, 2005; Márquez-Millán et al. 2014). These models were based on the principle that addition of female hatchlings to the GoM is the major agent of population growth, loss through deaths (attributable to natural and anthropogenic causes combined) of females within all life stages is the major agent of population decline, and emigration and immigration can be ignored (Heppell et al. 2007). In a given year, documented annual nests on the index beach reflect annual numbers of nesters, which reflect abundance of accumulated year-classes of mature females in the population. Adult females represent a small proportion of the female population, and those that nest in a given year an even smaller proportion (Heppell 1997; Bowen and Karl 2007; Seminoff and Shanker 2008; Bjorndal et al. 2011; Crowder and Heppell 2011; NMFS et al. 2011; NMFS and US FWS 2015). During their lives, individual Kemp's ridleys (both sexes) experience widely varying spatial-temporal exposures to natural and anthropogenic factors which affect their survival, growth, maturation, and reproduction. A relatively small proportion of any year-class survives to reproduce.

In this article, we examine 1) the relationship between time-lagged annual nests during 1986–2014 and cumulative hatchlings released (both sexes) by 1976–2004, respectively, assuming female minimum age at maturity (MAaM) of 10 yrs, and 2) the time-series of time-lagged annual nests during 1986–2014 divided by cumulative hatchlings released by 1976–2004, respectively, under the same assumption. We cumulated annual hatchlings released from year 1966 onward. These two novel metrics reflect the effects of cumulative hatchling inputs on future outputs of annual nests. Our purposes are to show that these metrics are very sensitive indicators of impacts of combined natural and anthropogenic factors on adult females, and large subadult females that mature between the ends of consecutive nesting seasons, and to emphasize the importance of nesting beaches as well as conservation efforts on nesting beaches to Kemp's ridley recovery and sustainability.

Bjorndal et al. (2011) and Finkbeiner et al. (2011) emphasized the need for assessing cumulative effects of individual threats on sea turtle populations (see also Crowder and Heppell 2011). We believe it is equally important to assess cumulative effects of conservation efforts on sea turtle populations (Caillouet 2010), both on land (nesting beaches) and at sea. Here we focus on effects of cumulative conservation efforts on the Kemp's ridley female population's index beach. Our novel approach integrates the effects of all endogenous (genetic) and exogenous (environmental) factors that affected survival, growth, maturation, and reproductive output of individual females, as well as all factors that affected monitoring and documentation of annual nests and hatchlings released on the index beach during 1966–2014. In other words, Kemp's ridley females are demographically heterogeneous. Kendall et al. (2011) reviewed demographic heterogeneity among individuals in natural populations and its influence on cohort (i.e., year-class) selection and population growth. They concluded that heterogeneous survival in long-lived species can increase the long-term growth rate in populations of any size. We consider the positive feedback loop between cumulative hatchlings released and time-lagged annual nests (Caillouet 2010) to be a form of cumulative causation of population growth. In addition, our approach dampens the effects of variations in annual hatchlings released and survival in the oceanic stage over years.

There is a parallel in concept between our use of post-1965 cumulative hatchlings released from the index beach to represent the number of hatchlings ever released into the GoM and Cohen's (2014) use of cumulative human births to represent the number of people ever born up to calendar year t. Cohen (2014) also estimated the fraction F(t) of those ever born up to calendar year t who were alive at t. He then examined conditions under which F(t) rises or falls. Ediev et al. (2015) extended Cohen's (2014) methodology to estimate the proportion of those ever born who reached a certain age. These approaches (Cohen 2014; Ediev et al. 2015) may have application in future modeling of sea turtle populations. For purposes of our article, we consider cumulative hatchlings released from the index beach to be very large percentages of the number ever released into the GoM after 1965 but not 100% (see “Methods”). Although our analyses are not designed to elucidate causes of the extraordinary downward departures of observed annual nests from those predicted for 2010–2014 by NMFS et al. (2011), they provide insights as to the extraordinary magnitudes of those departures as well as sensitivity of our two novel metrics.

Historical Background

Kemp's ridley was headed toward extinction when Hildebrand (1963) called for preventative conservation efforts. At that time, a declining and aging remnant (Fig. 2) was all that remained of the apparently bountiful population that in 1947 produced the largest ever recorded mass-nesting (i.e., arribada, Spanish for arrival from the sea) at this species' primary nesting beach near Rancho Nuevo (Fig. 1) (Carr 1963, 1977; Hildebrand 1963; Chavez et al. 1968; Pritchard and Márquez M. 1973; Márquez-M. et al. 1989a, 2005; Marquez-M. 1994; Márquez-Millán et al. 2014). Very high levels of mortality attributable to human overexploitation and natural predation of eggs, intentional take of turtles, and incidental capture of turtles in shrimp trawls caused this decline (Committee on Sea Turtle Conservation [CSTC] 1990; US FWS and NMFS 1992, 2015; Marquez-M. 1994; Heppell 1997; TEWG 1998; Heppell et al. 2007; NMFS et al. 2011).

View Figure

• Download as Powerpoint
• Download Figure

During 1963–1977, Mexico's government promulgated protective actions to reduce exploitation of sea turtles and their eggs, including actions specific to Kemp's ridley, its nesting beach near Rancho Nuevo, and adjacent waters (Márquez-M. et al. 1989a, 2005; Heppell et al. 2007; Márquez-Millán et al. 2014). In April 1966, Mexico's Instituto Nacional de Investigaciónes Biológico-Pesqueras (INIBP) began routine patrols of the stretch of Kemp's ridley nesting beach between Barra del Tordo, at the mouth of the San Rafael River, and Barra del Carrizo to its north (Fig. 1) (Chavez et al. 1968). INIBP also conducted studies of nesters, eggs, and hatchlings and nester capture–mark–recapture (CMR), and protected eggs and hatchlings (Chavez et al. 1968). INIBP also initiated annual monitoring and documentation of nests, eggs, and hatchlings (Chavez et al. 1968; Márquez-M. et al. 1989b). In 1971, the Instituto Nacional de Pesca (INP) succeeded INIBP (Márquez-Millán et al. 2014) and continued the work in Tamaulipas (CSTC 1990; US FWS and NMFS 1992; Marquez-M. 1994, 2001; NMFS and US FWS 2015). As resources available for conservation, monitoring, and research increased, and as nesting spread to the south and north, the length of the index beach was extended stepwise to its current span by 1996 (Fig. 1) (Márquez et al. 1999; Márquez-M. et al. 2005, 2014; Burchfield and Peña 2014). In 2001, the Comisión Nacional de Áreas Naturales Protegidas (CONANP), within the Secretaría del Medio Ambiente y Recursos Naturales (SEMARNAT), succeeded INP and continued the work in Tamaulipas (Márquez-Millán et al. 2014). Prior to 1974, Mexico gradually expanded and enforced jurisdiction over its GoM shrimping grounds, thereby excluding US shrimp trawlers from many of its productive shrimping areas (Iversen et al. 1993). In 1976, a treaty between the US and Mexico required phasing out all shrimp trawling by the US fleet off Mexico's coast of the GoM by the end of 1979 (Cicin-Sain et al. 1986; Richards and Juhl 1987; Condrey and Fuller 1992; Iversen et al. 1993; TEWG 1998; Gillett 2008). These changes reduced shrimp trawling effort by the US fleet in Mexico's GoM waters, thereby increasing survival of neritic stage turtles at sea (CSTC 1990; TEWG 1998). However, annual nests continued to decline, prompting Carr (1977) to write:

The species is clearly on the skids, and if present conditions continue it will shortly—in two years perhaps, or three, or five—be gone. The dramatic drop during the 1950's was caused by overexploitation combined with very heavy natural predation pressures. The terminal decline now in progress has been brought about by incidental trawler catch. When ridleys were many and shrimping was less intensive this factor was negligible. Today it is wiping out the species. Lepidochelys kempi [sic] can possibly be saved, but it will surely disappear unless drastic action is taken.

Government agencies in the United States and Mexico responded by initiating the Kemp's Ridley Restoration and Enhancement Program (KRREP) in 1978, to 1) reestablish nesting on Padre Island National Seashore (PAIS) near Corpus Christi, Texas (Fig. 1), as a precaution against possible catastrophic natural or anthropogenic impacts on the nesting beach near Rancho Nuevo; and 2) to enhance hatchling releases on the nesting beach near Rancho Nuevo (Berry 1987; Marquez-M. 1994; Márquez-Millán et al. 2014; Caillouet et al. 2015b; Shaver and Caillouet 2015).

Beginning in 1978, gear research programs under the auspices of NMFS and Sea Grant in cooperation with the shrimping industry led to development of several turtle excluder devices (TEDs) that allowed sea turtles caught incidentally in shrimp trawls to escape (Conner 1987). A successful prototype TED was developed by 1981, and during ensuing years, TEDs became smaller, lighter, and collapsible (Conner 1987). An annually recurring seasonal closure to shrimp trawling along the Texas coast (referred to as the Texas Closure) was initiated in 1981 to allow brown shrimp (Farfantepenaeus aztecus) to grow to larger sizes before the shrimping season opened, but it also led to reduction in sea turtle strandings (Lewison et al. 2003). A number of NMFS-approved TED models were available for voluntary use by 1983, but fewer than 3% of active shrimpers had tried TEDs by the end of 1986 (Conner 1987). A threat to sea turtles posed by skimmer trawls developed during the early 1980s (Price and Gearhart 2011; Pulver et al. 2012, 2014; Scott-Denton et al. 2014). There exists a federal tow-time restriction for skimmer trawls, butterfly nets, and similar gears but no federal requirement for TEDs in these gears; some states require TEDs in skimmers or prohibit such gear (M. Barnette, pers. comm., December 2015). In year 2000, an additional, annually recurring seasonal closure to shrimp trawling was established by Texas Parks and Wildlife Department (TPWD) along the south Texas coast as a shrimp fishery management measure and to protect sea turtles (Lewison et al. 2003).

Despite reductions in shrimp trawling effort (Condrey and Fuller 1992; Iversen et al. 1993; TEWG 1998), Kemp's ridley annual nests continued to decline through 1985 (Fig. 2) (Frazer 1986), but this decline reversed in 1986 (Fig. 3). In 1987, NMFS established the first regulations requiring use of TEDs in shrimp trawls within the southeastern US shrimp fishery (Condrey and Fuller 1992; Yaninek 1995; Epperly 2003; Jenkins 2010, 2012; Damiano 2014; National Oceanic and Atmospheric Administration [NOAA] 2014). CSTC (1990) designated shrimp trawling as the major cause of sea turtle mortality associated with human activities and the most important human-associated source of deaths of adult and subadult sea turtles. Byles (1993) acknowledged the importance of the nesting beach (in Tamaulipas) and nesters, TED regulations, and reversal of the decline as follows:

View Figure

• Download as Powerpoint
• Download Figure

I believe the protection of the [Kemp's ridley] nesting beach and the turtles nesting there is the single, most important action we can take to protect the species from extinction. We also have regulations to alleviate the single greatest cause of mortality in the Gulf of Mexico, which is incidental mortality in shrimp trawls. Optimism for the recovery of the species comes from an apparent reversal of the decline in nesting females with an increase in nests of about 8% per year since 1985. Concurrent broad compliance with TED regulations should result in large increases in the recruitment of nesting adults in the next decade.

Over the years, nests have been characterized according to levels of protection afforded the clutches (US FWS and NMFS 1992; Heppell et al. 2005; Caillouet 2006; Wibbels 2007; NMFS et al. 2011; Bevan 2013; Bevan et al. 2014; NMFS and US FWS 2015): 1) translocated to corrals; 2) translocated to Styrofoam^TM or polystyrene boxes; 3) left in situ; and 4) predator control. Egg-to-hatchling survival in clutches incubated in Tamaulipas egg hatcheries (i.e., corrals and boxes) is higher than that in clutches left in situ (Heppell et al. 2005; Caillouet 2006; Wibbels 2007; NMFS et al. 2011; Bevan 2013; Bevan et al. 2014; NMFS and US FWS 2015). Beginning in 2004, increasing proportions of annual nests were left in situ (NMFS et al. 2011; NMFS and US FWS 2015). This was a response to the exponential increase in annual nests and limitations on resources available for translocation and protection of clutches. It also was a move toward reducing manipulative conservation practices and human intervention (TEWG 1998, 2000; Heppell et al. 2005; National Fish and Wildlife Foundation [NFWF] 2009; NMFS et al. 2011; Bevan 2013; Burchfield and Peña 2013; Bevan et al. 2014; Caillouet et al. 2015b). NMFS et al. (2011) established two demographic criteria for downlisting Kemp's ridley from endangered to threatened status; viz., 10,000 females nesting in a season, and a minimum of 300,000 hatchlings released per season; these criteria were restricted to nesters and hatchlings at the index beach (Fig. 3). The first criterion is equivalent to 25,000 nests, assuming 2.5 nests per nester per season. The downlisting criterion for hatchlings includes both sexes.

Demographic models (Heppell et al. 1996, 2005, 2007; TEWG 1998, 2000; Coyne and Landry 2007; Crowder and Heppell 2011; NMFS et al. 2011; Heppell 2014, 2015; NMFS and US FWS 2015) and stock assessment models (Gallaway et al. 2013, in press a, in press b; Gallaway and Caillouet 2014; Gallaway and Gazey 2014, 2015) have been used to evaluate effects of various natural and anthropogenic factors thought to have had the most influence on the Kemp's ridley female population trajectory over the years. All of these models incorporated various time series of annual hatchlings released and annual nests on the index beach, as well as estimated or assumed vital rates including female age at maturity, adult female remigration interval, nests per female per nesting season, eggs per nest, egg-to-hatchling survival, temperature-dependent sexual differentiation (TSD) sex ratios, age at which oceanic juveniles become neritic, age-specific or stage-specific mortality, and time-phased multipliers that reduced mortality rates in neritic stage turtles at sea. The time-phased mortality-reducing multipliers in all the models were applied to start “knife edge” in selected years, with their effects continuing thereafter. In the demographic models, these mortality-reducing multipliers were applied to instantaneous total mortality rate, recognizing that their effects represented any combination of factors in aggregate that decreased neritic stage mortality at sea (e.g., TEDs, favorable environmental factors, reduction in anthropogenic mortality from other sources, changes in fishery patterns driven by non-TED factors, etc. [TEWG 1998; Caillouet 2006]). In the stock assessment models (Gallaway et al. 2013, in press a, in press b; Gallaway and Caillouet 2014; Gallaway and Gazey 2014, 2015), the mortality-reducing multipliers were applied only to instantaneous shrimp trawling mortality (estimated from shrimping effort by the US fleet), under a hypothesis that all anthropogenic mortality at sea was caused by shrimp trawling by the US shrimping fleet, except in 2010. Shrimp trawling effort by Mexico's fleet was not available (Gallaway et al. 2013, in press a).

With one exception (NMFS and US FWS 2015), female age at maturity has been applied “knife edge” in demographic and stock assessment models (Heppell et al. 1996, 2005, 2007; TEWG 1998, 2000; Coyne and Landry 2007; Crowder and Heppell 2011; NMFS et al. 2011; Gallaway et al. 2013, in press a, in press b; Gallaway and Caillouet 2014; Gallaway and Gazey 2014, 2015; Heppell 2014, 2015). The updated demographic model (NMFS and US FWS 2015) incorporated the following modifications: 1) variable age at first nesting with a minimum of 10 yrs to provide for gradual rather than “knife edge” maturation of females; 2) separation of nesters into neophyte and remigrant (i.e., those returning from prior nesting seasons) categories; and 3) multipliers applied to reduce mortality rate of adult females at sea during various intervals of years.

The combination of restored and enhanced annual releases of hatchlings on the index beach (beginning in 1966), decreases in shrimp trawling effort (Nance 1992; Griffin et al. 1997; Lewison et al. 2003; Caillouet et al. 2008; Nance et al. 2010), and use of TEDs led to post-1985 exponential increase in annual nests and annual hatchlings through 2009 (Fig. 3). Annual additions of hatchlings restored the population's age structure and momentum (TEWG 1998, 2000; Heppell et al. 2007; Caillouet 2010, 2011, 2014; Gallaway et al. 2013, in press a, in press b). Declines in shrimp trawling effort and use of TEDs reduced mortality of neritic stage turtles at sea, and thereby increased the proportion of females that reached maturity and nested (TEWG 1998, 2000; Lewison et al. 2003; Heppell et al. 2005, 2007; Caillouet 2010, 2011, 2014; Crowder and Heppell 2011; NMFS et al. 2011; Gallaway et al. 2013, in press a, in press b; Gallaway and Caillouet 2014; Gallaway and Gazey 2014, 2015; Heppell 2014, 2015; NMFS and US FWS 2015). Most model-predicted annual nests for years 2010–2014 have not been very close to observed annual nests for those years. An exception was the updated demographic model (NMFS and US FWS 2015) fitted to annual nests on the index beach in years 1978–2009. This model predicted annual nests for years 2010–2014 that were remarkably close to observed annual nests for those years. Dixon and Heppell (2015) conducted regression analyses of various time series of log(annual nests) for three beach categories: 1) the index beach; 2) all Tamaulipas nesting beaches combined; and 3) all Tamaulipas and Texas nesting beaches combined. Annual hatchlings released were not included in these models. Dixon and Heppell (2015) chose 1991 as the first year of the time series for the index beach, and 1996 or 1997 for the first year of the time series for all Tamaulipas nesting beaches combined and for all Tamaulipas and Texas nesting beaches combined. All of these time series ended with year 2014. All of the regression models showed clear “loss” of nests in years 2010–2014, represented by downward deviations in observed from expected log(annual nests). In other words, following the positive trend through 2009, a change occurred in 2010 that resulted in large reductions in annual nests compared with expected annual nests, and annual nests have not recovered since (Dixon and Heppell 2015). These analyses also showed that variance of deviations from regression was approximately constant (i.e., homogeneous) within each of the regression models. This meant that the variance of deviations from exponential trends in annual nests was heterogeneous, increasing as annual nests increased.

METHODS

We used the best available 1966–2014 time series of annual nests and annual hatchlings released (Fig. 3), obtained from table 1 in NMFS and US FWS (2015). They were also used by Selina Heppell, Oregon State University, in the updated demographic model in NMFS and US FWS (2015). Typically, most of the clutches found on the index beach have been translocated to protective corrals for incubation (Heppell et al. 2007; Wibbels 2007), but some have been translocated to Styrofoam or polystyrene boxes (US FWS and NMFS 1992; Marquez-M. 1994; Burchfield and Peña 2013) and others left in situ (Wibbels 2007; Bevan 2013; Bevan et al. 2014). Surprisingly, neither NMFS et al. (2011) nor NMFS and US FWS (2015) mentioned that annual nests and annual hatchlings released on the index beach included clutches incubated in Styrofoam or polystyrene boxes. In any case, all nests found on the index beach and hatchlings produced from them were included in the documented annual nests and annual hatchlings released (J. Peña, pers. comm., September 2015). Therefore, we assumed that the nests and hatchlings data we analyzed (Fig. 3) represented combined data for clutches incubated in corrals, boxes, and in situ. Use of annual nests as an index of female population size is based on assumptions of proportional relationships between numbers of annual nests and nesters, and between numbers of annual nesters and adult females in the population, both of which can vary over years.

Stepwise extensions of the length of the index beach (Márquez et al. 1999; Márquez-M. et al. 2005; Burchfield and Peña 2013, 2014; Márquez-Millán et al. 2014) no doubt affected the annual nests and hatchlings data series (Fig. 3). Additions of hatchlings to the population from beaches other than the index beach (i.e., in Tamaulipas, Veracruz, Texas, and elsewhere) could also have affected the annual nests series if any females resulting from such hatchlings nested on the index beach. Although such effects on the data series cannot be ruled out, they are assumed to have been limited because of the strong propensity for natal homing in this species (Putman and Lohmann 2008) and the preponderance of nesting on the index beach. On the other hand, there must be some plasticity in nesting site selection (Shaver and Caillouet 2015) because the Kemp's ridley's nesting range has expanded since conservation efforts began.

Intermittently estimated sex ratios of hatchling year-classes released at the index beach have generally been dominated by females (Coyne and Landry 2007; Wibbels 2007; Eich 2009; NMFS et al. 2011; LeBlanc et al. 2012; Bevan 2013; Bevan et al. 2014; NMFS and US FWS 2015). As far as we are aware, no complete 1966–2014 time series of annual estimates of hatchling year-class sex ratios exists from which we could have estimated the number of female hatchlings released annually. The major reason is that the most reliable method of determining sex of sea turtle hatchlings is gonad histology, which requires sacrificing living specimens (Mrosovsky and Godfrey 2010). It is unlikely that hatchling year-class sex ratios have been constant over years. Variation and trends in year-class sex ratios would not only affect our results, but also those of all previous modeling, since sex ratios have generally been assumed constant for purposes of modeling. The number of hatchlings released from a year-class is the greatest abundance that year-class will ever have, except for the year-class' abundance of eggs. Survival rates of both sexes during the oceanic stage likely varied by year-class (Putman et al. 2010, 2012, 2013), thereby contributing additional variation to the proportions of each sex that entered the neritic stage.

We assumed a female MAaM of 10 yrs, based on annual nests data from table 1 in NMFS and US FWS (2015) and anecdotal evidence in Marquez-M. (1994). Marquez-M. (1994) noted that “young” Kemp's ridley nesters appeared on the nesting beach near Rancho Nuevo in 1976 and “old” nesters disappeared by 1984 (Fig. 2; see also Caillouet 2010; Caillouet et al. 2011). For purposes of Fig. 2 only, which shows the 1966–1985 decline in annual nests, we interpreted “young” nesters to be survivors from INIBP's and INP's post-1965 annual hatchling releases, and “old” nesters to be carryovers from the declining and aging remnant population of adult females that existed prior to 1966.

Although we selected 10 yrs as the female MAaM, we concede that some “young” nesters may have nested before 1976 without being observed (see Pritchard and Márquez M. 1973; Pritchard 1990). We note that annual nests declined more rapidly during 1966–1971 than during 1972–1975 (Fig. 2). This slowing in rate of decline could have resulted from early protective actions taken by Mexico to reduce sea turtle mortality (Márquez-M. et al. 1989a, 2005; Heppell et al. 2007; Márquez-Millán et al. 2014). If any “young” nesters nested before 1976, some of their nests could have been included in the available data series (Fig. 3). Some demographic models also incorporated female age at maturity of 10 yrs (TEWG 1998; Heppell et al. 2005; Coyne and Landry 2007), or a female MAaM of 10 yrs (NMFS and US FWS 2015).

Heppell et al. (2007) suggested that emigration and immigration of Kemp's ridleys can be ignored in demographic modeling, “because we have data for the entire species.” Indeed, the geographic range of Kemp's ridley includes the GoM and North Atlantic Ocean (see review by Caillouet et al. 2015b), but most of the population is restricted to the GoM (NMFS et al. 2011; NMFS and US FWS 2015). No known Kemp's ridley nesting colonies exist outside the GoM, although a few isolated nestings have been reported on the US east coast (NMFS et al. 2011; Caillouet et al. 2015b; NMFS and US FWS 2015; Shaver and Caillouet 2015). Kemp's ridleys that emigrate from the GoM to the North Atlantic Ocean do so in significant numbers as oceanic juveniles (TEWG 1998, 2000; Márquez M. 2001; Putman et al. 2010, 2013, 2015; NMFS et al. 2011; Witherington et al. 2012; NMFS and US FWS 2015), and at least one neritic Kemp's ridley has been tracked from the GoM into the western North Atlantic Ocean (Renaud and Williams 2005). Most if not all Kemp's ridley that enter the North Atlantic Ocean originate as hatchlings released on the index beach in Tamaulipas, but significant numbers of hatchlings originate at other beaches in Tamaulipas, Veracruz, and Texas (Márquez-Millán et al. 2014; Shaver and Caillouet 2015). An important question remains as to whether most if not all of the oceanic stage juvenile Kemp's ridleys carried by surface currents from the GoM into the North Atlantic Ocean remain there for the rest of their lives and, therefore, are lost to the population in the GoM (Caillouet et al. 2015b). Evidence that Kemp's ridleys in the North Atlantic Ocean return to the GoM is limited. Fewer than 20 tag returns have been reported for individuals known to have been tagged in the North Atlantic Ocean and later documented to have returned to the GoM; all of them were documented on the nesting beach near Rancho Nuevo (Caillouet et al. 2015b). Except for the few that might originate from rare nestings on the US east coast, any Kemp's ridleys that migrate from the North Atlantic Ocean to the GoM are simply returning to the GoM where they originated.

The fundamental concept underlying our analyses is that it takes cumulative hatchlings released from multiple year-classes at the index beach to cause a change in time-lagged annual nests produced at the index beach by surviving nesters representing multiple year-classes. We stipulate that the available annual nests and hatchlings data (Fig. 3) were samples (albeit the largest for any of the Kemp's ridley nesting beaches) and that nesting occurred on beaches beyond the index beach in Tamaulipas as well as elsewhere outside of Tamaulipas (Heppell et al. 2007; Gallaway et al. 2013, in press a; Caillouet et al. 2015b). Hatchlings released in a given year depended on the number of nests, eggs per clutch, and all factors that affected egg-to-hatchling survival. We make no assumptions concerning environmental carrying capacity or density dependence (Heppell et al., 2007; Gallaway et al., in press b).

Below we list additional concepts and concessions relevant to our analyses, most but not all of which are also relevant to demographic and stock assessment modeling:

• 1. 

  Beginning in the egg, each female experiences unique exposures to environmental conditions during its lifetime; i.e., each female has a unique life history (Coyne and Landry 2007).

• 2. 

  Sample sex ratios of hatchlings released have generally been dominated by females (NMFS et al. 2011; NMFS and US FWS 2015) and influenced by variation from year to year in incubation temperatures in corrals, Styrofoam or polystyrene foam boxes, and in situ (TEWG 1998; Wibbels 2007; LeBlanc et al. 2012; Bevan 2013; Bevan et al. 2014).

• 3. 

  Longevity of individual females is unknown, but average longevity has been estimated to be ≈ 50–100 yrs, assuming absence of anthropogenic mortality (Gallaway et al. 2013, in press a).

• 4. 

  Adult females represent more year-classes in the population than do all other life stages combined, because longevity of females is several times greater than female age at maturity (Caillouet 2010). This concept can be tested by using demographic and stock assessment models to predict the annual number of year-classes representing adult females in the population in any year. Only adult females lay eggs; hence, the unknown annual number of nesters is the only source of annual nests on the index beach in any given year. Although this is obvious, we emphasize it to make clear that abundance of all earlier life stages of females, except for large subadult females that mature between the ends of consecutive nesting seasons, cannot have any effect at all on annual nests in a given year, unless they reduce resources (e.g., prey; Shaver et al. 2013, 2016; Servis et al. 2015) needed by large subadults and adults to provide energy for migration to the index beach and development of eggs.

• 5. 

  All vital rates vary.

• 6. 

  Not all adult females in the population nest in a given year.

• 7. 

  Nesters lay one or more clutches per season.

• 8. 

  Some nests on the index beach are not found (Pritchard 1990).

• 9. 

  Adult males may influence fecundity of adult females, thereby affecting reproductive output of adult females (Kichler et al. 1999; Bowen and Karl 2007; Coyne and Landry 2007; Rostal 2007; Boyle et al. 2014). Females that lay larger clutches are more likely to have been multiply-mated than have those that lay smaller clutches (Neff et al. 2002), and clutch weight increases with body weight of nesters (Witzell et al. 2005; Bjorndal et al. 2014). Abundance of males has been adequate so far, but could become limiting (Coyne and Landry 2007).

RESULTS

During 1966–1976, the highest annual hatchlings released (32,970 in 1970 and 36,100 in 1976) were lower than all annual hatchlings released thereafter, with two exceptions (30,100 in 1977 and 32,921 in 1983, respectively) (Fig. 3). For the period 1966–1983, Marquez-M. (1994) noted that “old” nesters exhibited high fecundity until overall fecundity of nesters (“old” and “young” combined) began to decline as increasing numbers of “young” nesters reproduced (Fig. 2). The 1966–1992 time series of maximum, average, and minimum annual eggs per nest (Fig. 4; based on data from table 8 in Marquez-M. 1994) provided additional evidence of decline in overall fecundity of nesters. Maximum annual eggs per nest showed the highest variability but no trend (slope = 0.1402, intercept = −115.0, SD from regression = 12.92, p = 0.6707, n = 26). Average annual eggs per nest were the least variable and declined (slope = −0.3953, intercept = 886.7, SD from regression = 2.585, p < 0.0001, n = 26). Minimum annual eggs per nest were intermediate in variability and showed the highest rate of decline (slope = −1.170, intercept = 2352, SD from regression = 8.232, p < 0.0001, n = 26). We suggest that the decline in numbers of “old” nesters during 1966–1983 could have contributed to the reduction in overall fecundity of nesters during those years.

View Figure

• Download as Powerpoint
• Download Figure

Cumulative hatchlings released increased smoothly and gradually before becoming exponential through 2009 (Fig. 5). They reached 260,035 by 1976 (Fig. 5), 10 yrs prior to 1986 when time-lagged annual nests equaled 744 (Fig. 3). They reached 2,099,026 by 1999 (Fig. 5), 10 yrs prior to 2009 when time-lagged annual nests reached its peak of 19,163 (Fig. 3). During 2010–2014, slowing of the rate of increase in cumulative hatchlings released reflected downward departures in annual nests from their pre-2010 exponential trend (Fig. 3). Annual nests during 2010–2014 (Fig. 3) were well below levels that would be expected from cumulative hatchlings released by 2000–2004 (Fig. 5), respectively, assuming female MAaM of 10 yrs. The lowered annual hatchlings released during 2010–2014 will affect annual numbers of adult females from these year-classes in years 2020–2024, respectively, assuming female MAaM of 10 yrs.

View Figure

• Download as Powerpoint
• Download Figure

The relationship between time-lagged annual nests in years 1986–2014 and cumulative hatchlings released by years 1976–2004, respectively, became slightly sigmoid as it approached a peak in 2009 (Fig. 6). The sigmoid shape suggested gradual development of density dependence prior to 2010 (see Heppell et al. 2007; Gallaway et al., in press b). It may also reflect effects of declining fecundity of nesters and increasing proportions of nests left in situ. However, large downward departures from this sigmoid relationship after 2009 (Fig. 6) would not be expected from gradual development of density dependence (see Heppell 2014, 2015). They also are not expected based on cumulative hatchlings released 10 yrs prior to 2010. For consecutive pairs of post-2008 data points (Fig. 6), the largest drop between years 2009 and 2010 was followed by an increase between 2010 and 2011, then further drops between 2012 and 2013, and 2013 and 2014.

View Figure

• Download as Powerpoint
• Download Figure

Time-lagged annual nests during 1986–2014 divided by cumulative hatchlings released by 1976–2004, respectively, were proportions that time-lagged annual nests represented of hatchlings ever released on the index beach after 1965 (Fig. 7). They provided estimates of how many time-lagged annual nests resulted from one released hatchling. The time series for these estimates appeared roughly sigmoid in shape up to 2009, initially declining from 0.00286 in 1986 to a minimum of 0.00206 in 1989, then increasing followed by apparent slowing of its rate of increase as it approached its peak of 0.00913 in 2009 (Fig. 7). This peak was followed by a sharp downturn to 0.00502 in 2010, a rebound to 0.00661 in 2011, and a decline thereafter to 0.00277 in 2014, which was lower than in 1986 (Fig. 7). Multiplied by 1000, these estimates would represent the number of time-lagged annual nests resulting from 1000 cumulative hatchlings released. If annual proportions of female hatchlings released were available for years 1966–2004, these estimates could be adjusted to represent female hatchlings only. We did not convert time-lagged annual nests to time-lagged annual nesters, because reproductive output per adult female and overall annual egg-to-hatchling survival appear to have declined over the years (Marquez-M. 1994; Witzell et al. 2005; Caillouet 2014).

View Figure

• Download as Powerpoint
• Download Figure

Figure 8 facilitates visual comparison of trends in numbers of cumulative hatchlings released, annual hatchlings released, annual nests, and annual hatchlings released divided by annual nests. The trend in cumulative hatchlings released was upward, but its rate of increase slowed after 2009 (Fig. 8). The trends in annual hatchlings released and annual nests were also upward, except for the downward departures after 2009 (Fig. 8). The trend in annual hatchlings released divided by annual nests was upward through 1989, then generally downward thereafter (see Caillouet 2014).

View Figure

• Download as Powerpoint
• Download Figure

The NMFS et al. (2011) criterion of 300,000 hatchlings (both sexes) released per season at the index beach includes hatchlings produced in clutches translocated to corrals and those left in situ. Clutches incubated in Styrofoam or polystyrene boxes apparently are not included in this criterion. Annual hatchlings released first exceeded 300,000 in year 2000 (at 365,479), but dropped to 291,268 in 2001, then remained above 300,000 until reaching a peak of 1,025,027 in 2009 (Fig. 3) (NMFS and US FWS 2015). It dropped to 663,614 in 2010 and to 630,182 in 2011, then rose sharply to 927,002 in 2012, followed by decline thereafter to 519,273 in 2014 (Fig. 3).

DISCUSSION

Under conditions of very high anthropogenic and natural mortality, the obviously viable Kemp's ridley population in 1947 became extinct ≈ 37 yrs later (Fig. 2), as indicated by disappearance of “old” nesters by 1984 (Marquez-M. 1994). We come to this conclusion because no “young” nesters were observed during 1966–1975, suggesting that there were few if any turtles maturing within the residual population to replace “old” females that were dying off through 1983.

The decline in Kemp's ridley annual nests at the index beach through 1985, reversal of this decline in 1986, subsequent exponential growth in annual nests through the end of the nesting season in 2009, and downward departures from exponential growth after 2009 are all good examples of how the balance between annual additions of female hatchlings and losses of females in all life stages changed in this population over years 1966–2014. Under the assumption of female MAaM of 10 yrs, it should be clear that reversal (Fig. 3) of this population's pre-1986 decline (Fig. 2) could not have occurred unless cumulative additions of hatchlings exceeded losses (attributable to all natural and anthropogenic causes) in all life stages, and that limited voluntary use of TEDs (Conner 1987) could not have contributed much if at all to this reversal (Caillouet 2010). It should also be clear that exponential growth in this population could not have occurred unless cumulative additions of hatchlings overwhelmed losses (attributable to all natural and anthropogenic causes) in all life stages before the population's setback in 2010 (Caillouet 2010, 2011, 2014). More than 4 preceding decades of cumulative conservation efforts proved incapable of preventing the population's setback in 2010. Population momentum may have been responsible for temporarily elevating annual nests in 2012 and 2013 compared with annual nests in 2010, but annual nests in years 2010–2014 were still well below predicted levels (NMFS et al. 2011; Caillouet 2014; NMFS and US FWS 2015; Gallaway et al., in press a, in press b). For 2015, annual nests at the index beach were ≈14,000 (Preliminary Data, T. Shearer, pers. comm., December 2015), remaining well below the level predicted from pre-2010 exponential growth.

In retrospect, Mexico's early actions, which prohibited intentional take of sea turtles and their eggs, established and conducted conservation practices on the nesting beach near Rancho Nuevo, and reduced shrimp trawling effort by the US fleet in Mexico's waters of the GoM, were the major contributors to reversal of the pre-1986 decline in annual nests (Márquez-M. et al. 1989a, 1989b, 2005; TEWG 1998; Caillouet 2006, 2010; Márquez-Millán et al. 2014). It is remarkable that only 260,035 cumulative hatchlings released by 1976 led to reversal of the pre-1986 decline in annual nests, assuming female MAaM of 10 yrs, especially because an unknown proportion of theses hatchlings were males. This demonstrated the population's resilience and responsiveness to Mexico's early conservation actions and efforts. It was not until 1987 that NMFS established the first regulations requiring use of TEDs in shrimp trawls within the southeastern US shrimp fishery and not until 1994 that NMFS established regulations requiring use of TEDs in shrimp trawls at all times in this fishery (Condrey and Fuller 1992; Yaninek 1995; Epperly 2003; Jenkins 2010, 2012; Damiano 2014; NOAA 2014).

If our novel approach were applied using assumed female MAaMs greater or less than 10 yrs, it would produce results different from those we obtained (Figs. 6–8). For example, under an assumed female MAaM of 12 yrs, cumulative hatchlings released by 1978 would coincide with time-lagged annual nests in 1990. Under an assumed female MAaM of 8 yrs, cumulative hatchlings released by 1974 would coincide with time-lagged annual nests in 1982. Neither would coincide with reversal of the decline in annual nests in 1986.

Despite considerable amounts of sampling, monitoring, analyses, and modeling conducted since the DWH oil spill began, major causes of the setback and their relative contributions to the setback remain to be determined with certainty (Caillouet et al. 2015a). Trustees (2016) concluded:

DWH oil did not arrive on the continental shelf of the northern Gulf of Mexico until late May or early June 2010. By that time, adult Kemp's ridley turtles that were going to breed in 2010 would likely have already departed the northern Gulf for their breeding and nesting areas in the western Gulf. Therefore, DWH oil was unlikely to have had a direct impact on Kemp's ridley nesting in 2010. However, DWH oil could have contributed to the reduced numbers of nests in subsequent years (2011–2014) through direct and indirect pathways.

Data on migrations of large subadult and adult Kemp's ridleys from foraging grounds to nesting beaches are rare (Shaver et al. 2016). On the other hand, satellite tracking data for post-nesters (e.g., Shaver et al. 2015, 2016) could provide a basis for relating migrations from foraging areas to nesting beaches to environmental variables such as ocean circulation features, salinity, temperature, and bathymetry, which in turn could be used to develop homing migration models designed to bracket a range of plausible scenarios for travel times from northern GoM to the index beach in Tamaulipas. Likewise, analyses of time-sequences of daily nest counts on the index beach and adult strandings in Tamaulipas in years 2010–2014, compared with those in preceding years 2005–2009, might provide insights as to the timing of homing migrations in those years.

The following paraphrased objectives of the Kemp's ridley stock assessment (Gallaway et al. 2013) are suggested as guides for further investigation of possible causes of the population's setback: 1) examine Kemp's ridley population status, trend, and temporal-spatial distribution within the GoM (including Mexico and United States); 2) examine status, trends, and temporal-spatial distribution of shrimping effort in the northern GoM; and 3) examine other factors that may have contributed to increased Kemp's ridley–shrimp fishery interactions or otherwise caused Kemp's ridley strandings, injuries, or deaths in the northern GoM in 2010 and beyond, to include but not be limited to abundance of shrimp and Kemp's ridley prey species, river outflow (especially from the Mississippi River), 2010 oil spill and mitigative actions associated with it, surface circulation, hypoxic zones, cold stunning, locations and characteristics of nesting beaches, weather patterns, tropical storms and hurricanes, droughts, red tide, harmful algae blooms, etc. The forthcoming report covering the Gulf of Mexico Sea Turtle Workshop held in October 2015 by Harte Research Institute and Texas A&M University, Corpus Christi, Texas will provide recommendations for sea turtle research, monitoring, conservation, collaboration, and data sharing that may help elucidate causes of the population's setback.

TEWG (1998, p. 4), citing CSTC (1990) and US FWS and NMFS (1992) as their sources, stated that: “It is unlikely that any major non-shrimping cause of sea turtle mortality could have escaped detection during the many years of study of factors causing sea turtle injury and mortality at sea.” The fact that declining shrimping effort and use of TEDs were major contributors to exponential growth in annual nests is confirmation that incidental capture of Kemp's ridleys in shrimp trawls was the most important human-associated cause of mortality before 2010. However, well before the DWH oil spill, the threat of shrimp trawling was greatly reduced and declining (Lewison et al. 2003, 2013; Gallaway et al. 2013, in press a, in press b). For purposes of Kemp's ridley stock assessment modeling, Gallaway et al. (in press a) hypothetically treated GoM shrimp trawling by the US fleet as the only cause of anthropogenic mortality in neritic stage Kemp's ridleys at sea during years 1966–2012 except for year 2010. Under this hypothesis, the difference between estimated annual deaths attributable to shrimp trawling and annual total deaths represented deaths attributable to natural causes in all years except 2010. Estimated shrimp trawling deaths of females ≥ 9 yrs old (i.e., adults and large subadults) exceeded those attributable to natural causes in each year 1980–1989, but estimated deaths of females ≥ 9 yrs old attributable to natural causes exceeded those attributable to shrimp trawling in each year 1990–2012. The data series did not extend beyond 2012 (Gallaway et al., in press a). Shrimp trawling accounted for 12.5% (3,339 turtles) of the estimated 26,626 total deaths of females ≥ 9 yrs old in 2010, leaving 87.5% (23,287 deaths) unexplained (Gallaway et al., in press a). These unexplained deaths in 2010 were caused, hypothetically, by other anthropogenic as well as natural causes combined. In 2009, deaths attributable to shrimp trawling were 24.1% (3,705 turtles) of estimated total deaths of 15,371 females ≥ 9 yrs old, leaving 75.9% (11,666 deaths) explained by natural causes only. In other words, estimated shrimp trawling–related deaths of large subadult and adult females dropped in 2010 as compared with 2009 (Gallaway et al., in press a). For females ≥ 9 yrs old, estimated natural deaths in 2009 were 3.15 times higher than estimated shrimp trawling–related deaths; applying this multiplier to shrimp trawling–related deaths in 2010 gives an estimate of 10,514 natural deaths, leaving 12,773 deaths attributable to undetermined anthropogenic causes. NOAA (2014) concluded that continued implementation of sea turtle conservation regulations under the Endangered Species Act, and continued authorization of the southeast US shrimp fisheries in federal waters under the Magnuson-Stevens Fisheries Conservation and Management Act, were not expected to cause an appreciable reduction in likelihood of survival and recovery of Kemp's ridleys in the wild. Incidental capture of Kemp's ridleys in shrimp trawls, skimmer trawls, and hook-and-line fisheries were included in the assessments that led to this biological opinion (NOAA 2014). Kemp's ridleys caught incidentally in skimmer trawls and by recreational hook-and-line pier fishing between the ends of the nesting seasons in 2009 and 2010 were predominantly juveniles. In each year 1980–2011, annual carapace length histograms for Kemp's ridleys (both sexes) stranded along the US coast of the GoM contained mostly juveniles (Gallaway et al. 2013). This was expected because abundance of adults and large subadults is low (Heppell 1997; Bowen and Karl 2007; Seminoff and Shanker 2008; Bjorndal et al. 2011; Crowder and Heppell 2011; NMFS et al. 2011; NMFS and US FWS 2015). It appears that shrimp trawling in the GoM by the US fleet was not a major contributor to the population's setback in 2010.

Cold temperatures and lowered salinities of northern GoM waters, related to high Mississippi River outflows during the winter of 2009–2010 (fig. 2 in Carmichael et al. 2012), could have blocked or otherwise interfered with the departure of subadult and adult females from northern GoM foraging grounds (Caillouet 2011; Gallaway and Gazey 2014; Gallaway et al., in press b) long enough to allow their exposure to DWH oil and dispersants (Crowder and Heppell 2011). Cold temperatures and low salinity during March through May 2010 were mentioned frequently by Trustees (2016). In April 2010, as a response to the DWH oil spill, fresh water from the Mississippi River was released by opening salinity control structures in 9 Louisiana locations: Davis Pond, Caernarvon, Bayou Lamoque, West Pointe a la Hache, Violet Siphon, White Ditch, Naomi Siphon, Ostrica Lock, and Bohemia (Trustees 2016). These structures were opened at or near maximum capacity for extended periods of time to repel the approaching DWH oil (Trustees 2016). By the time the flow of DWH oil was shut down and the salinity control structures were closed in late 2010, the highly atypical flow of fresh water over a sustained period had greatly reduced salinity levels in Louisiana coastal areas (Trustees 2016). The nesting season appears to have been delayed in 2010 (Bevan et al. 2014; Gallaway et al., in press b). Limited data on timing of adult female Kemp's ridley nesting migrations were provided by three satellite-tracked individuals that departed northern GoM foraging areas in late November 1994, 1 January 2007, and 27 February 2014, and arrived near western GoM nesting beaches on 10 March 1995, late March 2007, and 11 April 2014, respectively (Renaud et al. 1996; Shaver et al. 2016). They suggested that adult females left the foraging ground in autumn or early winter. An examination of seasonal sequences of daily nest numbers at the index beach during years 1966–2014 would be useful, if compared with Mississippi River outflows during autumn through each nesting season in those years. The cold winter of 2009–2010 also could have had indirect effects on subadult and adult Kemp's ridley females on northern GoM foraging grounds (Shaver et al. 2013; Wibbels and Bevan 2015). For example, the blue crab (Callinectes sapidus) population was already declining (Bourgeois et al. 2014) before the oil spill, and cold temperatures and low salinity could have reduced availability of this important prey species. Gallaway et al. (in press b) discussed possible density-dependent effects related to high abundance of Kemp's ridleys and reduced abundance of important prey species, including but not limited to blue crab and other portunid crabs, with consequent effects on migration, production of eggs, and remigration interval.

Hurricane Alex (Pasch 2010) has not received adequate attention with regard to its possible interference with adult and large subadult female migration from southern GoM foraging grounds to the index beach as well as nesting on the index beach in 2010 (Bevan et al. 2014). Alex became a tropical depression in the Caribbean on 25 June, crossed the Yucatan Peninsula on June 27, and as a category-2 hurricane made final landfall near Soto la Marina on the coast of Tamaulipas on 1 July (Pasch 2010, figs. 1, 4, 5).

The decline in annual hatchlings released divided by annual nests (Fig. 8) should receive greater attention (Caillouet 2014). Such a decline could be the result of increasing proportions of neophyte nesters on the index beach over the years, with consequent lowering of overall fecundity of nesters. Increasing proportions of nests left in situ also contributed to the recent decline in annual hatchlings released divided by annual nests (unpubl. data, S. Heppell, pers. comm., April 2015). These trends can be expected to reduce annual nests in the future. In this regard, the separation of neophyte nesters from remigrant nesters in the updated demographic model was an important innovation (NMFS and US FWS 2015). The combination of decline in fecundity and possible increase in remigration intervals of adult females (Gallaway and Gazey 2014, 2015; Heppell 2014, 2015; Gallaway et al., in press b) is also cause for concern, because it could reduce reproductive output of adult females in the population. The effect of declining fecundity likely was offset to an undermined extent by increasing average annual hatchlings per nest through 1989. However, the post-1989 decline in average annual hatchlings per nest will make restoration of population growth more challenging (Caillouet 2014). Consideration should also be given to possible long-term cumulative effects of contaminants that may reduce reproductive fitness (Rowe 2008).

Emphasis should be placed on restoring exponential growth of the female population as soon as possible. Even though annual hatchlings released remain above 300,000 per season, the practice of leaving increasing proportions of annual nests in situ should be revisited (Caillouet 2006). The exponential increase in annual hatchlings released was a major contributor to pre-2010 exponential growth in the population. Conservation practices that enhance annual hatchlings released on nesting beaches in Tamaulipas, Veracruz, and Texas probably would be the most expedient ways to restore exponential growth of the population, if resources for doing so are available (Caillouet et al. 2015b). We suggest that more clutches be translocated to protective corrals, leaving fewer in situ. Even this approach could take at least 10 yrs following its implementation before effects on annual nests are evident. Admittedly, corral hatchery operations are highly manipulative conservation measures (Caillouet et al. 2015b), but they had very positive effects on the Kemp's ridley female population (NMFS et al. 2011; NMFS and US FWS 2015). Before 2010, a major goal of NMFS et al. (2011) was replacement of Kemp's ridley hatchery operations in Mexico with natural beach conservation for long-term stability without human intervention (NFWF 2009). In addition, Bevan (2013) and Bevan et al. (2014) speculated that losses of clutches attributable to tidal inundation of nests left in situ in low lying areas could improve long-term fitness of the species.

Adult females have the highest reproductive value of any life stage, and eggs have the lowest (Heppell 1997; NMFS et al. 2011; NMFS and US FWS 2015). If no eggs were produced, there would be no hatchlings released from nesting beaches, and eventually there would be no turtles to protect via other conservation measures. Effects of such a scenario were demonstrated by the detrimental effects of overexploitation of eggs prior to 1966 (Heppell et al. 2007). Indeed, protection of neritic stage Kemp's ridleys at sea would have no effect on the population were it not for beneficial effects of protecting nesters, eggs, and hatchlings on nesting beaches in Tamaulipas, Veracruz, and Texas.

The importance of capture–mark–recapture (CMR) studies to dynamical and statistical modeling of vertebrate population dynamics was emphasized by Lebreton (2006), who stated: “Marking individuals at birth or later, and sampling them through time and space is indeed the closest counterpart that one may reach of the exhaustive registration of births and deaths in human populations.”

Mass tagging of hatchlings, by year-class, with life-long tags at the index beach could provide a wealth of data useful to estimating vital rates applicable to modeling, if combined with concerted efforts to detect and document recaptures (Eckert et al. 1994; Caillouet and Higgins 2015; Caillouet et al. 2015b). NMFS and INP personnel tagged a total of 43,885 hatchlings at Rancho Nuevo with internal coded wire tags (CWTs) over yrs 1996–1997 and 1999–2000, but relatively few recaptures were reported (Caillouet et al. 2015b); the population could still contain some adult survivors from these releases. Research is needed for development of easily detected, life-long tags applicable to emergent hatchlings (Caillouet and Higgins 2015). During the 2014 nesting season, with funding provided by the Gulf States Marine Fisheries Commission (GSMFC), large samples of Kemp's ridley nesters were examined for external metal tags and internal PIT tags, and their carapace lengths were measured on the beach near Rancho Nuevo (unpubl. data, B. Gallaway, pers. comm., August 2015). These activities were repeated during the 2015 nesting season, with funding by the same source (unpubl. data, B. Gallaway, pers. comm., August 2015). Carapace length appears to be a better predictor of female maturity and fecundity than does age (Márquez-M. 2001; Witzell et al. 2005; Caillouet et al. 2011; Bjorndal et al. 2014).

We applaud the practices of sharing and analyzing available time series of annual nests and annual hatchlings released at the index beach, because analytical methods and models are diverse, and multiple uses of available data encourage development, examination, and comparison of different analytical and modeling approaches. Based on our results and information in “Historical Background,” we make the following recommendations related to future demographic and stock assessment modeling of the Kemp's ridley female population:

• 1. 

  There seems little if any reason to include annual nests documented during the period of decline through 1985 in time series data used in demographic or stock assessment modeling, except to increase sample size; including pre-1986 annual nests contributes to overall variability in fitting the models.

• 2. 

  For demographic and stock assessment models, time series of annual hatchlings released, or cumulative hatchlings released should always begin with 1966, because additions of hatchlings restored the population's age structure and momentum.

• 3. 

  With regard to interruption of exponential growth in annual nests, various models can and should be used to predict age structure of the female population following the ends of nesting seasons in 2009 and later years; thus, these annual age structures can be compared (Caillouet 2014). In other words, model-predicted annual age structures in years following the 2009 nesting season are associated with interruption of the exponential increase in annual nests; predicted annual age structure in 2009 represents a prior condition. Insights regarding causes of interruption of the exponential trajectory may be gained by comparing the predicted annual age structure for 2009 with annual age structures in years thereafter (e.g., Gallaway et al. 2013, in press a, in press b).

• 4. 

  Develop a maturity schedule (ogive) for application in future models (Gallaway et al. 2013, in press a; NMFS and US FWS 2015) in lieu of “knife edge” age at maturity.

• 5. 

  Develop and apply methods of testing the demographic assumptions made in various models.

• 6. 

  Compare the time sequences of daily nest numbers (and especially arribadas) at the index beach during the 2009 nesting season with those of years before and after 2009 to determine the extent to which nesting was delayed in 2010 (Bevan et al. 2016). If archived data are available, it would be worthwhile to compare the time sequences of daily nests numbers for all seasons beginning in 1966.

• 7. 

  Determine annual proportions of nests incubated in corrals, boxes, and in situ.

• 8. 

  Estimate annual reproductive value of the female population based on model-predicted annual age structure over the years. Expectations are that reproductive value of the female population has declined over the years, primarily attributable to exponential increase in the annual proportion of neophyte nesters, as the population increased exponentially through 2009.

• 9. 

  At the index beach (or the Rancho Nuevo segment of the index beach), conduct annual sampling of clutch sizes (number of eggs) and carapace lengths of nesters, and conduct CMR of nesters, to estimate annual a) distribution of clutch size, b) proportions of neophyte or putative neophyte nesters and remigrant nesters, c) remigration interval, d) nests per nester, and e) relationship between carapace length and clutch size.

• 10. 

  Mine archived records and literature for data and information useful to estimating annual proportions of neophyte (or putative neophyte) nesters and sex ratios of hatchlings released over the years.

• 11. 

  Examine archived records for shifts in distributions of annual nests, eggs, and hatchlings among various Tamaulipas beach segments (Burchfield and Peña 2013, 2014; Dixon and Heppell 2015) over the years.

• 12. 

  Mine archived records and literature on strandings, bycatch, in-water studies, etc., for data and other information on annual CMR, carapace lengths, and sex ratios.

• 13. 

  If a compatible time series of annual shrimp trawling effort in the GoM by Mexico's fleet can be acquired, adapt and combine it with the time series of annual shrimp trawling effort by the US fleet, for development of a time series of instantaneous shrimp trawling mortality to be used in future Kemp's ridley stock assessment models (Gallaway et al. 2013, in press a).

• 14. 

  If applicable, consider modifying future models based on the demonstration by Dixon and Heppell (2015) and Gallaway et al. (in press b) that variance in annual nests increases as annual nests increase.

• 15. 

  Use available models to estimate annual carapace-length distributions of the neritic female population in 2009 and beyond, and compare them to sample carapace-length distributions derived from various data sources (strandings, CMR, in-water studies, samples of nesters on nesting beaches, etc.). This is a way to compare and validate different models (Caillouet 2014).

• 16. 

  Use available models to estimate annual size of the neritic female population in 2009 and beyond, to determine how well each model explains the lowered annual nests in post-2009 years.

• 17. 

  Mass tag and release hatchlings at the index beach over a number of consecutive years and increase efforts to detect and document recaptures from these and previously tagged and released year-classes (see Eckert et al. 1994; Caillouet et al. 2015b).

Enhanced production of hatchlings on beaches of Tamaulipas, Veracruz, and Texas will be needed to restore the population's exponential growth. If exponential growth is restored, it eventually will decelerate as the population approaches carrying capacity or recovery (Heppell et al. 2007; NMFS et al. 2011; NMFS and US FWS 2015). Kemp's ridley has exhibited considerable plasticity by nesting on widely separated beaches in the GoM and to a very limited extent on the US east coast (Caillouet et al. 2015b; Shaver and Caillouet 2015). Additional nesting colonies could provide safeguards against losing the species to environmental changes and catastrophes in the future (Mrosovsky and Godfrey 2010; Caillouet et al. 2015b).