Survival of Adult Female Northern Pintails in Sacramento Valley, California

Introduction

California supports the largest winter population of pintails in North America and most are found in SACV (Bellrose 1980). Pintails migrate to SACV beginning in early August (Miller 1985), large numbers are present by mid-September, peak numbers occur October-February, and many pintails remain through March (Bellrose 1980:270). Pintails face mortality risks from hunters, disease, contaminants (Hunter 1970, Rosen 1971, Ohlendorf and Miller 1984, Miller et al. 1988, Gilmer et al. 1989), and predators. However, the relative effect of these factors on winter survival, and the effect of winter mortality on abundance of this pintail population has not been assessed.

Pintail populations declined in North America between 1979 and the early 1990s (U.S. Fish and Wildl. Serv. and Can. Wildl. Serv. 1994). Pintail breeding population size is related to spring habitat conditions (Raveling and Heitmeyer 1989), and low recruitment during extended drought over much of the pintail's nesting range likely was the major cause for this decline (Raveling and Heitmeyer 1989, U.S. Fish and Wildl. Serv. and Can. Wildl. Serv. 1989). However, dry habitat conditions during winter in California may also limit recruitment (Raveling and Heitmeyer 1989).

The U.S. Fish and Wildlife Service began to restrict harvest in 1985 to increase annual survival of pintails (Molini 1989). However, Rienecker (1987) estimated average annual survival of AHY females banded preseason in SACV during 1964-79 at 0.639 (SE = 0.028), and Hestbeck (1993) estimated annual survival of females banded postseason in the Central Valley (SACV and San Joaquin Valley [SJV]) (Fig. 1) during 1962-67 at 0.712 (SE = 0.040) and during 1974-78 at 0.647 (SE = 0.025). Annual survival of AHY female pintails banded preseason in Alberta during 1974-90 and in Saskatchewan during 1970-80 ranged from 0.621 (SE = 0.018) to 0.650 (SE = 0.075), respectively (F. A. Johnson, U.S. Fish and Wildl. Serv., Laurel, Md., unpubl. data). Thus, the relationship between harvest restrictions and annual survival is not clear. Managers need to assess the effect of winter versus nonwinter (nesting-migration) and hunting versus nonhunting mortality rates to implement appropriate management measures (Raveling and Heitmeyer 1989).

Female pintails generally are more productive in their second and succeeding breeding seasons than first-year nesters (Duncan 1987), and recruitment is not thought to be affected by a shortage of males. Therefore, we estimated winter survival and identified mortality causes of AHY female pintails radiotagged after their arrival in late summer in SACV, 1987-89.

Study Area

We monitored pintails in SACV, Sacramento San Joaquin River Delta (Delta), Suisun Marsh, San Francisco Bay (SFB), SJV, and northeast California (Fig. 1). Rice farming dominated central SACV (about 150,000 ha annu.); harvested and reflooded rice fields provided important wintering waterfowl habitat (Miller et al. 1989). State wildlife areas, NWRs, and private duck clubs provided the majority of wetlands in SACV (36,500 ha) (Miller 1987, Heitmeyer et al. 1989). Waterfowl habitat in Delta included wetlands and dry and flooded grain fields (Heitmeyer et al. 1989), whereas freshwater wetlands predominated in the SJV (Heitmeyer et al. 1989). Suisun Marsh consisted of brackish wetlands (Miller et al. 1975, Rollins 1981), and SFB provided tidal marsh, diked marsh, and salt ponds (Josselyn 1983, Houghten et al. 1989). Northeast California provided freshwater wetlands and dry grain fields (Klamath Basin NWR, Tulelake, Calif., unpubl. data).

Methods

We captured and radiotagged pintails during late August and early September 1987-89. This was prior to arrival of peak numbers of pintails, but our objective was to have a radio-tagged sample present in the main pintail wintering region throughout the period when pintails were present to reflect the broadest potential variety of mortality sources (logistic constraints precluded marking birds throughout winter). We assumed that our sample would mix with wintering ducks, using the same areas in similar chronological fashion. We further assumed that our sample consisted of migratory pintails; small numbers nest in California but are virtually absent from SACV (A. E. H. Perkins, California Waterfowl Assoc., Sacramento, unpubl. data).

We used rocket-nets and welded-wire funnel traps (Schemnitz 1994) baited with rice to catch AHY female pintails on Sacramento and Delevan NWRs and a private duck club, all in Colusa and Glenn counties in central SACV (Fig. 1). We started and completed trapping during the last week of August 1988 and 1989 but continued until 10 September in 1987. We legbanded, weighed (to the nearest 5 g), measured (flat wing, culmen 1, total tarsus [Dzubin and Cooch 1992] to the nearest 0.01 mm), radiotagged (n = 190 during the 3 winters; n = 154 [33 in 1987, 48 in 1988, 73 in 1989] at Sacramento NWR, n = 33 [21 in 1987, 12 in 1988] at Delevan NWR, and n = 3 [1989] at the duck club), and released pintails where captured < 12 hours later. We used 21-24-g back-mounted (Dwyer 1972) radio transmitters (3.0-3.2% of mass), each with unique signal characteristics and life expectancy of 150 days in 1987-88 and 180 days in the final 2 winters. Transmitters initially had a minimum range of 3.2 km ground-to-ground using 150-db receivers and dual 4-element Yagi antennas mounted on pick-up trucks.

We recorded pintail status (location, alive, or dead) once during daylight hours and once at night (not all pintails were located every night) each day in SACV and Delta, and ≥1 day/week in other California areas from 25 August 1987 to 12 February 1988 (172 days), 21 August 1988 to 24 February 1989 (188 days), and 23 August 1989 to 22 February 1990 (184 days). We used standard aerial tracking procedures (Gilmer et al. 1981) to search for missing ducks throughout California, western Nevada, southern Oregon, and western Mexico. We censored pintails with which we lost contact after transmitter signals indicated impending failure (erratic, fast, slow, intermittent, atypical, or weakening) at the last day the duck was known to be alive. We did not include for analysis 1 female surviving < 5 days following release.

Radio transmitters did not have mortality sensors. Therefore, if a pintail did not move for 2 consecutive days, we either flushed the duck or retrieved the carcass. At mortality sites, we determined cause of death by looking for predator sign, other waterbird carcasses (indicative of disease outbreak), and damage to harness or transmitter. By taking cause of death into consideration with bird history, we ascertained time of death.

We estimated date of nonhunting death to within 2 days for pintails in SACV and Delta. Outside this area, we estimated death date as midway between last known date alive and date first located at the mortality site. We shipped carcasses to National Wildlife Health Research Center in Madison, Wisconsin, for necropsy to confirm suspected disease. We also recorded deaths reported by hunters. Without contrary evidence, we assumed that 2 transmitters found intact and undamaged had been removed by surviving birds (Derleth and Sepik 1990).

The hunting season was a continuous 79-day period in 1987-88 (24 Oct-10 Jan) but was split into 2 segments totaling 59 days in 1988-89 (22 Oct-13 Nov and 4 Dec-8 Jan) and 1989-90 (28 Oct-19 Nov and 2 Dec-6 Jan). The daily bag limit for pintails was more liberal in 1987-88 (4 pintails [1 F], 5 total ducks) compared with 1988-89 and 1989-90 (4 ducks, 1 pintail); but only 1 female/day could be taken each year.

Data Analysis

We used univariate analysis of variance (ANOVA) (PROC GLM; SAS Inst. Inc. 1989a) to test for winter effects on body mass and morphometric variables. We preceded ANOVAs with a multivariate ANOVA (MANOVA) as a check against possible joint effect of winter on these variables (Johnson and Wichern 1988:169). We used Fisher's protected LSD value to isolate pairwise differences in means following effects in ANOVAs (Milliken and Johnson 1984:31).

We estimated survival with the Kaplan-Meier method to avoid difficulties associated with assumptions of constant survival distributions (Kaplan and Meier 1958, Cox and Oakes 1984: 48-50). We estimated survival during 3 intervals (preseason, hunting season, and postseason) and for winter. We estimated survival for all pintails radiotagged in SACV, irrespective of subsequent winter distribution (n = 190). To pool across years, we used a 180-day wintering period, the mean length of the 3 field seasons.

We used PROC LIFETEST (SAS Inst. Inc. 1989a) to estimate Kaplan-Meier survival rates. Because LIFETEST requires all subjects to enter the experiment simultaneously (White and Garrott 1990:235), we standardized all pintails to begin entry into the study on 25 August (date of earliest trapped ducks in 1987) in each of the 3 winters, by left censoring birds caught prior to 25 August (earliest was 21 Aug) and adjusting birds caught after 25 August (latest was 10 Sep) back to 25 August as if they had been caught and released 25 August. We believe standardizing to 25 August was reasonable because there were no mortalities prior to release of the last ducks each year, we caught all ducks within a short span relative to the 180-day winter period, and as with staggered entry (Pollock et al. 1989a), we assumed that females caught after 25 August were in the same population and at the same risk as those already caught and released.

We computed mortality rates for hunting and nonhunting by considering nonhunting mortalities as censored observations for hunting mortality rate. Conversely, we estimated mortality rate for nonhunting deaths of considering hunting mortalities as censored observations (Conroy et al. 1989). We compared hunting mortality rates to nonhunting mortality rates, using a Chi-square test following Scott and Seber (1983), which accounts for nonindependence of mortality rates. We computed an overall 3 degree-of-freedom Chi-square test by summing the separate Chi-square values for each winter. If the overall Chi-square test was significant (P = 0.05), we then examined the 3 within winter single degree-of-freedom Chi-squares.

We compared survival rates among winters and among preseason, hunting season, and postseason intervals (and interactions) in a 3 × 3 configuration, following Sauer and Williams (1989), using PROC IML (SAS Inst. Inc. 1989b). We used a log-rank test (Pollock et al. 1989b) to make pair-wise comparisons among the 3 winter survival distributions. We examined effects of body mass at time of capture as a continuous covariate and molting status at time of capture on winter survival probabilities, using proportional hazards modeling (Cox and Oakes 1984:91-110). We used PROC PHREG (SAS Inst. Inc. 1992) and followed Dobson (1990:98) and Milliken (1984:990-999) to assess importance of explanatory variables and their interactions on survival.

We approximated the statistical power of (1 - β) of detecting hypothetical 0.05, 0.10, 0.15, and 0.20 decreases in survival rates. We subtracted the above hypothetical values from the largest overall survival rate estimate among the 3 winter periods, among intervals, and between pintails less than and greater than the median bodymass. We used the number of radio-tagged pintails as sample sizes for the power tests, and we paired the largest sample size with the smallest in each comparison. We also conducted similar power analyses between hunting and nonhunting mortality rates, with the exception of using the actual difference as the hypothetical effect size, and determined what sample size would have been required to attain a power of 0.8. We used the program code in White and Garrott (1990:31-35) to calculate 1 - βs.

We estimated survival during the spring-fall migration-nesting period (Mar-Aug) by dividing estimated mean annual survival (unweighted of estimates in Rienecker [1987], Hestbeck [1993], and F. A. Johnson = 0.654) of banded AHY female pintails by the winter survival estimate we generated.

Results

Morphometric variables differed among years (MANOVA, Wilks' λ = 0.8335; F = 5.88; 6, 370 df; P < 0.001). Follow-up ANOVAs indicated flat wing (F = 12.01; 2, 187 df; P < 0.001) and culmen (F = 3.58; 2, 187 df; P = 0.030) lengths differed by year, but tarsus length did not (F = 0.73; 2, 187 df; P = 0.485) (Table 1). These annual variations were negligible. Body mass varied by year (ANOVA); 1989-90 females weighed less than females in the other 2 winters (F = 6.03: 2, 187 df: P = 0.003)(Table 1).

Radio Transmitter Performance

Transmitters failed rapidly, most without warning, throughout 1989-90: 18% (14) failed in the first 60 days, 20% (15) in the second 60 days, and 38% (29) in the last 64 days. Fewer failures occurred in the first 2 winters, and these radios emitted signal changes that gave warning of impending failure (1987-88, 4% [2] failed in the first 60 days, 4% [2] in the second 60 days, and 26% [14] in the last 52 days [these radios designed for 150-day life only]; 1988-89, 8% [5] failed in the first 60 days, 8% [5] in the second 60 days, and 15% [9] in the last 68 days). Additional failed radios occurred in the last third of each field season as batteries gave out toward the end of predicted life.

Hunter returns of expired transmitters from 1989-90 revealed that these pintails were alive in the study area after we had lost contact and that battery failure and antenna loss caused transmitters to fail; failures did not result from ducks leaving the search area or from mortality. Pintails with functioning transmitters in 1989-90 yielded usable data because they distributed themselves throughout the study area as did our radio-tagged sample during previous field seasons, and total radio exposure-days exceeded those of the first 2 winters (Table 2). We did not include deaths of pintails carrying nonfunctioning transmitters reported by hunters because similar natural deaths would not have been detectable, but we collected data until time of censoring.

Pintail Distribution

Radio-tagged pintails remained in the SACV during the study, with 92.9% of all radio exposure-days occurring there (92.1 % in 1987-88, 91.9% in 1988-89, 94.5% in 1989-90). The SFB and Delta (3.8%), SJV (2.9% ), and northeast California (< 0.5%) accounted for the remainder of exposure days.

Causes, Location, and Timing of Mortality

We confirmed 17 pintail deaths (Table 2); 15 of these occurred in SACV, as did all known natural mortality (7). Eleven deaths occurred on private land (6 legal hunting, 1 avian botulism [Clostridium botulinum], 1 avian cholera [Pasturella multocida], 2 illegal shooting, 1 unknown), and the remainder on public lands (2 mammalian predator, 2 avian predator, 1 avian cholera, 1 legal hunting). All predator kills occurred on Sacramento or Delevan NWRs during preseason, and both illegal kills occurred postseason. All confirmed hunting season mortalities were by legal shooting, all occurred during the latter part of the seasons (after the split in 1988-89 and 1989-90), and 5 of 7 occurred ≥10 km from sanctuaries. Five pintails were shot in SACV, 1 in SFB, and 1 in SJV.

Survival Rates

Survival estimates did not vary by winter (χ² = 0.43, 2 df, P = 0.808) among preseason, hunting season, and postseason (χ² = 1.09, 2 df, P = 0.579), or from interaction between winter and each interval (χ² = 3.25, 4 df, P = 0.517); overall pintail survival in winter ranged from 0.854 to 0.901 (Table 2), and the pooled value is 0.874 (SE = 0.030). Most mortality occurred beginning in the latter portion of hunting seasons, continuing to the end of winter (Fig. 2), and we found no differences among pintail winter survival distributions, using the log-rank test (all P > 0.34). Because we found no indication of differences in pintail winter survival estimates, we conducted the modeling effort by pooling across the 3 winters. Body mass had no detectable effect on survival probability for either molt status group (χ² = 0.12, 1 df, P = 0.732), and body mass alone was not related to survival (χ² = 2.32, 1 df, P = 0.127). Ducks that had not yet molted (wing) when we caught them (n = 10 used for analyses here) tended to survive at a lower rate (0.622, SE = 0.178) than did ducks that had already molted (n = 180) (0.887, SE = 0.030) (χ² = 0.286, 1 df, P = 0.091). There were no differences between hunting and nonhunting mortality rates among years (χ² = 3.42, 3 df, P = 0.332), or within year (all P > 0.149).

We could detect a 0.05, 0.10, 0.15, and 0.20 decrease from the 1989-90 survival rate estimate of 0.901 (max. of the 3 winters, largest n = 76, smallest n = 54) with 1 - β = 0.14, 0.36, 0.62, and 0.82, respectively. We could detect a 0.05, 0.10, 0.15, and 0.20 decrease from the pooled preseason interval survival rate estimate of 0.972 (max. of pooled intervals, largest n = 190, smallest n = 128) with 1 - β = 0.51, 0.92, 0.99, and 0.99, respectively. We could detect a 0.05, 0.10, 0.15, and 0.20 decrease from the pooled survival rate estimate of 0.92 for pintails with body mass larger than the median (largest n = 97, smallest n = 93) with 1 - β = 0.20, 0.54, 0.82, and 0.95, respectively. True mortality rate differences as small as those observed would have required a minimum of 2,700, 1,390, and 255 pintails for winters 1987-88, 1988-89, and 1989-90, respectively, for detection at 1 - β = 0.8.

The estimated nonwintering period survival rate of AHY female pintails was 0.748 (0.654/0.874). Thus, point estimates of winter survival exceeded the combined rate of spring and fall migration plus the nesting season.

Discussion

Pintails radiotagged in SACV survived winter at a relatively high rate, as did northern pintails in Sinaloa, Mexico (0.91) (Migoya and Baldassarre 1995), adult female canvasbacks (Aythya valisineria) in Louisiana (0.95) (Hohman et al. 1993) and Chesapeake Bay (1.00) (Haramis et al. 1993), and mallards in Arkansas (0.84-0.96) (Reinecke et al. 1987, Dugger et al. 1994). However, shooting pressure was low in Mexico, canvasback studies were conducted under closed hunting seasons, 1 mallard study (Dugger et al. 1994) was conducted after the hunting season, and winter periods in all studies started later or ended earlier (or both) and lasted for shorter periods than our winter periods in California; 180-day survival probably would have been lower in these studies to account for mortality before and after their experimental periods. Our estimated survival rate for AHY females seems high given high shooting pressure in California (Gilmer et al. 1989) and inclusion of premolting pintails. Actual survival could have been higher, because ducks carrying backpack radios may be more susceptible to harvest (and possibly predation) than would unmarked ducks (Reinecke et al. 1992).

Ducks with lower body mass often have lower survival probability than those with greater mass (Greenwood et al. 1986, Haramis et al. 1986 [but not for ad F], Hepp et al. 1986, Conroy et al. 1989, Bergan and Smith 1993, Dufour et al. 1993). However, we found no difference in pintail survival by mass. Similarly, survival of radiotagged AHY female pintails wintering in Sinaloa, Mexico, 1989-92 (Migoya and Baldassarre 1995), mallards in Arkansas (Dugger et al. 1994) and Colorado (Jeske et al. 1994), and black ducks (Anas rubripes) in Maine (Krementz et al. 1989) did not differ by body mass (or condition), and Haramis et al. (1993) found no difference in body mass among radio-tagged canvasbacks that were shot, found dead, or that survived in Chesapeake Bay.

The lack of association between body mass and survival of SACV radio-tagged pintails may have resulted from low mortality rates or difficulties in associating capture mass with mortality later in winter when AHY female pintails undergo changes in body mass (Miller 1986). We did not test survival relative to a condition index because in pintails structural size is not a determinant in explaining body fat and protein content variation (Miller 1989). Mild climate and abundant food, typical in California and other southerly wintering regions (Bellrose 1980), and high survival of our sample, may have precluded our ability to detect body mass as a factor in pintail survival, although low power of statistical tests, resulting from few mortalities, reduced our ability to detect real differences.

Hunting mortality was low and inconsistently related to nonhunting mortality among winters in SACV radio-tagged pintails. We speculate that in addition to restrictive bag limits and season lengths, AHY females may have reacquainted themselves with sanctuary location during the nearly 2 months between their arrival in SACV (when we marked them) and the beginning of hunting season; no radio-tagged pintails were shot during the first half of any hunting season. By implication, pintails arriving in the SACV closer to the start of, or during, hunting season may be at more risk from hunting (and natural mortality).

The routine of daytime roosting on sanctuaries and night feeding in harvested rice or newly flooded duck club habitat was established before the start of hunting season and was disrupted little by the onset of hunting (Miller 1985). In the latter part of hunting seasons, pintails left sanctuaries to seek outlying feeding and roosting habitats (Miller, unpubl. data). Hence, most shot pintails were taken later in the season and at locations away from refuges.

Survival of pintails during hunting season did not differ among years, even with the more liberal daily bag limit and season length in 1987-88. Thus, harvest of AHY female pintails was similar under a long season and 1 female within a 4 pintail bag limit (4 chances/day to bag a F), compared with a shorter season and a 1 pintail daily bag limit (1 chance/day to bag a F). Data from the U.S. Fish and Wildlife Service (F. A. Johnson, unpubl. data) show that direct recovery rates of AHY female pintails legbanded before hunting seasons 1985-87 in southern Alberta and southwestern Saskatchewan (major sources of Calif. wintering pintails; Bellrose 1980), when pintail bag limits ranged from 4 to 5 with 1 female, did not differ from those banded in 1988-91, when the 1 pintail (either sex) bag limit was in place (χ² = 6.14, 6 df, P = 0.41 for Alta.; χ² = 3.36, 6 df, P = 0.76 for Sask.). The pooled recovery rate was 0.0085 and the kill rate was 0.032 (assumes band reporting rate of 0.32 and a crippling loss of 0.20), resulting in hunting mortality accounting for < 10% of total annual mortality (F. A. Johnson, unpubl. data).

Illegal kill of AHY pintails in SACV occurred because of indiscriminate (1 shot during closed duck season but open goose season) or planned (1 shot at night during closed season) shooting, not as the result of identification error of the kind reported for canvasbacks (Haramis et al. 1993, Hohman et al. 1995). Illegal kill has not been reported in other studies of radio-tagged waterfowl.

Natural Mortality

The absence of differences in survival rate among preseason, hunting season, and postseason, and between hunting and nonhunting, demonstrates that pintails faced similar risks from natural and hunting mortality factors, but at different times. Predator kills were limited to preseason, suggesting that predators responded to the relative abundance of pintails appearing on the limited wetlands available in late summer by increasing capture rates (Krebs 1978:256). In California, predator kills of pintails occurred only on public refuges, perhaps because wetlands were available earlier there than on private wetland and agricultural habitats (Heitmeyer et al. 1989), and respective predator populations may have differed.

Premolt pintails were more vulnerable to predation than were pintails that had already replaced flight feathers when we caught them, although the small sample size requires caution in interpretation. The number of premolt pintails present on the refuges was small each year (Miller et al. 1992), but they represented a predictable component of the population present in late summer, and we included them in overall survival estimates for pintails present at that time, although their inclusion marginally depressed winter survival estimates.

Radio-tagged pintails lost to disease died during large outbreaks of avian botulism (Oct 1987 near and on Sacramento NWR) and avian cholera (late Jan-Feb 1989, near and on Gray Lodge Wildl. Area) (U.S. Fish and Wildl. Serv., Portland, Oreg., unpubl. data), rather than from isolated incidents. For example, carcass retrieval by NWR staff normally ranged from 0 to 147 birds/month on and in the vicinity of NWRs in 1987-90. In October 1987, however, field crews found > 1,000 dead waterfowl (confirmed botulism), including 1 of our radio-tagged pintails.

Management Implications

Winter survival of AHY female pintails radiotagged in SACV was relatively high. We caution, however, that estimates are needed for hatching year females, which, in ducks, typically suffer higher mortality than do AHY ducks (Krementz et al. 1989, Reinecke et al. 1987, Conroy et al. 1989, Haramis et al. 1993; but see Bergan and Smith [1993] and Migoya and Baldassarre [1995] for contrasting results). Survival information is also needed for other pintail wintering regions in California, including SJV, Delta, Suisun Marsh, SFB, and Salton Sea, and for pintails entering the Central Valley from other regions, particularly northeast California, shortly before, during, and after hunting season.

Efforts to reduce loss to disease and illegal kill might achieve only marginal returns because of low mortality of radio-tagged pintails from these sources. However, losses to cholera exceeded that to hunting in 1989-90, suggesting that managers should continue to focus on large disease incidents for carcass removal and disposal to reduce pintail mortality.

Survival of our AHY female pintails in winter exceeded that during the remainder of the annual cycle, assuming that annual survival estimates of Rienecker (1987), Hestbeck (1993) and the U.S. Fish and Wildlife Service (F. A. Johnson, unpubl. data) are representative of 1987-90. Thus, if gains in annual survival are desired for pintails arriving in SACV in early fall, managers should first examine the nesting-migration period for opportunities to achieve increases.