The federally threatened American crocodile (Crocodylus acutus) is a flagship species and ecological indicator of hydrologic restoration in the Florida Everglades. we conducted a long-term capture-recapture study on the South Florida population of American crocodiles from 1978 to 2015 to evaluate the effects of restoration efforts to restore historic hydrologic conditions. The study produced 10,040 crocodile capture events of 9,865 individuals and more than 90% of captures were of hatchlings. Body condition and growth rates of crocodiles were highly age-structured with younger crocodiles presenting with the poorest body condition and highest growth rates. Average body condition was 2.14±0.35 SD throughout South Florida. Crocodiles exposed to hypersaline conditions (> 40 psu) during the dry season maintained lower body condition scores and reduced growth rate by 13% after one year, by 24% after five years, and by 29% after ten years. Estimated hatchling survival for the South Florida population was 25% increasing with ontogeny and reaching near 90% survival at year six. Hatchling survival was 34% in NE Florida Bay relative to a 69% hatchling survival at Crocodile Lake National Wildlife Refuge and 53% in Flamingo area of Everglades National Park. Hypersaline conditions affected survival, growth and body condition and was most pronounced in NE Florida Bay, where the hydrologic conditions have been most disturbed. The American crocodile, a long-lived animal, with relatively slow growth rate provides an excellent model system to measure the effects of altered hydropatterns in the Everglades landscape.  Restoration efforts targeted toward returning freshwater flow and salinity targets of < 20 psu  to reflect a more historic state in NE Florida Bay will ensure improved health of the Everglades and illustrates the need for continued long-term monitoring projects to assess system-wide success.

We grouped survey routes into hydrologically distinct areas based on proximity to each other and potential crocodile use: Crocodile Lake National Wildlife Refuge, North Key Largo, Barnes Sound, and Manatee Bay (CRL); NE Florida Bay from US1 to Alligator Bay (including Long, Little Blackwater and Blackwater Sounds, Joe Bay, Davis Cove, Deer Key, and Alligator Bay) (NEFB); Little Madeira Bay, Taylor River, and Madeira Bay (MADB); West, Cuthbert, Long, Seven Palm, Middle, and Monroe Lakes, and Terrapin Bay (WEST); North and South Biscayne Bay and Card Sound (BBC); Cape Sable beaches, East Cape Canal, Lake Ingraham and associated creeks (CAPE); Flamingo, Buttonwood and Homestead canals, Coot Bay, Mud and Bear Lakes (FLAM; Fig 1). Northeast Florida Bay was split into NEFB and MADB to detect any effects of freshwater flow from Taylor River that empties directly into Little Madeira Bay, however geographically Madeira Bay is contained within NEFB.

We used mean daily salinity to calculate annual minimum and maximum salinity, number of days above a high salinity threshold (≥40 psu = hypersalinity), and number of days below a low salinity threshold (< 20 psu) as parameters in regression analyses to investigate the relationship between salinity and crocodile indicators of body condition, growth, and survival.

We conducted capture surveys by boat along accessible coastal and estuarine shorelines between East Cape at the western boundary of ENP, to south Biscayne Bay (Fig 1) from February 1978 to December 2015 (we conducted no surveys 1982–1985 due to lack of funding). Crocodiles were captured by hand, tongs, net, or noose and individually marked by notching caudal scutes according to a prescribed sequence [5]. We collected morphometric data, including total length (TL), snout-vent length (SVL), body mass, and determined sex when possible. We assigned crocodiles to size classes based on TL measurements, size classes are defined as follows: hatchling (TL < 65cm), juvenile (65 ≤ TL <150cm), subadult (150≤ TL < 225cm), and adult (TL ≥ 225cm) [5]. We categorized hatchling crocodiles based on time of year, differentiating between individuals observed within the hatching season (June–September) from those observed outside of the hatching season. We limited adult captures to between September 15^th and March 15^th, to minimize impacts on reproductive activities. We released crocodiles at the site of capture, and recorded date, time, location (measured by global positioning system, GPS), salinity (measured by a refractometer in psu) and habitat type (i.e., canal, cove, pond, creek, river, or exposed shoreline) for each crocodile capture event.

To assess body condition, we calculated Fulton’s condition factor (K) as follows:

K=100*WL³                                                                Equation (1)

 where W is body mass (g) and L is SVL (cm).

We used multivariate linear regression analysis to investigate relationships of biotic and abiotic factors with body condition. Three models were compared: model parameters in the “basic” model included year and season (wet season: May to September, dry season: October to April), location of capture (i.e., NEFB, FLAM, CAPE, etc.), size class, and habitat type. The “salinity” model assessed the relationship between salinity and crocodile body condition, including minimum and maximum salinity, and number of days above and below salinity thresholds. The “location” model assessed the combined effects of capture location and salinity. We compared models based on their Akaike Information Criteria (AIC), and selected the best model with lowest AIC as the most parsimonious.

We first modeled the general shape of total length as a function of age using generalized linear models for growth curve analysis [see 40]. Using all crocodiles of known age captured from 1978 to 2015, we compared a constant model with three models including linear, quadratic, and cubic terms of age incrementally (i.e., first-, second-and third-order polynomials) on a (natural) log–log scale. We included longitude (easting) (to indicate physical location of capture) and its quadratic effect as additive and multiplicative effects on growth. After selecting the most parsimonious model for the general shape based on their lowest AIC, we investigated effects of salinity (minimum and maximum salinity, and average salinity during the wet and dry seasons), and location of capture on crocodile growth.

We performed capture-recapture analyses (CR; [41]. This analysis estimates two parameters in the survival rate model: estimated annual survival rate (proportion of crocodiles that survived between time t and t+1, later referred to as Φ) and recapture rate at time t (later referred to as p, [42,43]).

            Most hatchlings (94%) were captured between June and September, peaking in July (SI Figure 1); thus, we defined June 1st as the starting point (noted t) for each year in the CR analysis [41]. We first modeled constant, time- and age-dependent recapture rates. We then fit a full age-dependent model for survival because we had no a priori knowledge about the age-structure of crocodile survivorship. From the observed pattern of age-specific survival (survival between age a and age a+1), we pooled age-classes with similar survival rates to reduce variation in survival rate estimates. This action resulted in four age-classes, defined as follows: 0–1 year, 1–2 years, 2–6 years, and ≥ 6 years. We then investigated the effects of average salinity, and area of capture by adding them as linear effects on estimated crocodile survival rates. We selected models with the smallest AIC as the best performing model.

All analyses were performed using R 3.6.0 [44], with the notable help of the packages RMark [45] for survival analysis (based on Program Mark software; [46], and ggplot [47] and cowplot [48] for graphs.