| The Use of Constant-effort
Mist-netting to Monitor Demographic Processes in Passerine Birds: Annual Variation in
Survival, Productivity, and Floaters Nadav Nur1, Geoffrey R. Geupel1, and Grant Ballard1
|
INTRODUCTION
Populations of many North American passerine species appear to be declining at an alarming rate (reviewed in Hagan and Johnston 1992; Askins 1993; Finch and Stangel 1993; Martin and Finch 1995). Migratory species, especially those wintering in the Neotropics, appear particularly threatened, though it is hard to determine the extent of the problems, because data are sparse for many species and for many regions of the continent. As a first step in formulating effective management and conservation responses, we are faced with the challenge of identifying the factors precipitating this decline. This effort requires detailed demographic information that cannot be provided by long-established national monitoring programs, specifically the Breeding Bird Survey.
The four most critical demographic processes underlying population growth and decline are: (1) Adult survivorship, (2) Reproductive success (i.e., production of young or "productivity"), (3) Survival and recruitment of young into the breeding population, and (4) Balance between immigration and emigration (i.e., net immigration). These four components are the most critical because the change in breeding population size from one year to the next, representing decline or recovery of a species, can be directly attributed to a combination of these and only these components. Thus, they are the necessary and sufficient components to explaining population dynamics.
The need for researchers, managers, and agencies to assess primary demographic processes has been enunciated frequently (e.g., Temple and Wiens 1989, DeSante et al. 1993) and is the thrust of several papers in this volume (e.g., Robinson and Morse; Sherry and Holmes; Rosenberg et al.). To obtain this sort of information on passerine birds, biologists have used a variety of approaches. The three used most commonly (with examples drawn from this volume in parentheses) are: (1) Observations of color-banded birds (Sherry and Holmes), (2) Mist-netting (Rosenberg et al.), and (3) Nest-monitoring studies (Conway and Martin; Robinson and Morse; Hejl and Holmes). Of these, we focus in this paper on Constant-effort mist netting (CEMN).
Mist-netting appears to be a potentially powerful and efficient means of collecting critical data on demographic parameters such as annual survival and reproductive success, and is the cornerstone of several monitoring programs, including the Constant Effort Sites (CES) Scheme of the British Trust for Ornithology (Baillie et al. 1986, Bibby et al. 1992, Peach 1993), and more recently, the Monitoring Avian Productivity and Survivorship (MAPS) program of the Institute of Bird Populations (DeSante et al. 1993; Rosenberg et al. this volume). However, the accuracy and validity of inferences derived from mist-netting capture data remain in question, and we are ignorant of the limitations and qualifications of data derived CEMN. Finally, in the absence of knowledge of the population from which mist-netted birds originate, it is impossible to develop methods of data collection and data analysis that best measure their intended targets.
For these reasons, we have begun a project to investigate the accuracy and validity of inferences drawn from mist net capture data for assessing certain demographic parameters. Knowledge of these demographic parameters, in turn, is of critical importance for interpreting population trends detected by other national monitoring programs, such as the Breeding Bird Survey. The crux of the project is a long-term study of several year-round resident species breeding at the Palomarin Field Station (Point Reyes National Seashore) whose populations have been simultaneously studied using two different methodologies: CEMN and detailed observations of color-banded individuals. In this way, we can directly compare parameter estimates (e.g., regarding productivity or survival) obtained with CEMN with estimates obtained using more intensive methods (i.e., observations of color-banded individuals).
Additional objectives of this paper are: (1) To demonstrate the value of CEMN for studying population processes, even when conducted at a single site (in this case the Palomarin Field Station); in this way, this paper complements that by Rosenberg et al. (this volume) who report on a large-scale, multi-site study of survival using CEMN; and (2) To demonstrate the additional insights gained by studying bird populations using both CEMN and nest-monitoring, supporting our contention that the two techniques are complementary rather than redundant.
METHODS
Study species and study site
The two principal study species are the Song Sparrow (Melospiza melodia) and the Wrentit (Chamea fasciata). Ecologically, both species are similar: they are open-cup nesters, insectivorous, and, at our study site, often nest in coastal scrub habitat. Wrentits, unlike Song Sparrows, are primarily restricted to coastal scrub for breeding (Geupel and DeSante 1990); Song Sparrows also breed in moister areas (Shuford 1993). Both species are year-round residents at Palomarin, and in both species, adults appear to disperse very little between years (Nice 1937, Erickson 1938, Johnson 1972, Geupel and Nur, unpublished data) and thus would be well suited for estimating survivorship on the basis of capture/recapture data. For productivity estimates we also include analyses of an additional year-round resident species breeding in the coastal scrub, the White-crowned Sparrow (Zonotrichia leucophrys nuttalli).
The field work has been conducted at the Point Reyes Bird Observatory's (PRBO's) Palomarin Field Station, located just within the southern boundary of the Point Reyes National Seashore and adjacent to the Pacific Ocean (see Figure 1 in Nur and Geupel 1993a; also Figure 1 in Johnson and Geupel 1996). On our study site of 36 ha, CEMN has been carried out since 1976, and intensive observations of color-banded individuals have been conducted since 1980. Thus in one study area we simultaneously have carried out CEMN, nest-monitoring, intensive spot-mapping, and behavioral observations of color-banded individuals.
CEMN was conducted using 20, 12-m mist nets erected at 14 netting sites. Eight sites (14 nets) were located on the edge of mixed, evergreen forest habitat. The other six single nets were located in successional coastal scrub habitat, interspersed with introduced grasses. For further description of the study area see DeSante and Geupel (1987).
The net locations are adjacent to and extend across approximately 25% of the 36 ha study plot. The rest of the study plot is within 700 m of the mist-net complex. The standardized operation of mist nests is described by DeSante and Geupel (1987) and has continued with little change. In summary, nets were run daily for 6 hours, beginning 15 minutes after local sunrise (weather permitting) from 1 May (1 April prior to 1989) to approximately 25 November, and 3 days per week from December through the end of March (or the end of April, since 1989); further details are in Johnson and Geupel (1996).
Field methods
The detailed monitoring of individuals was conducted on the 36 ha study plot and has been described elsewhere (Geupel and DeSante 1990). In summary, identities and territory boundaries of color-marked individuals were determined from detailed spot-mapping censuses conducted on a minimum of 3 days per week during the breeding season (1 April - 31 July) throughout the 15 years of the study. Concentrated efforts were made to locate and monitor all nest attempts of all territorial pairs on the study area from 1981 through 1985 and 1988 through 1994. No attempt was made to locate nests in 1986, therefore no direct estimates of productivity are available for that year. In addition, nest-finding effort was reduced in 1980 and 1987 compared to other years. However, identification of breeding individuals (territory holders) was not reduced in 1986 or 1987 compared to 1981-1985 or 1988-1994. In 1980, however, fewer than 50% of breeding adults were color-banded, and because the sample of breeders may not be representative, we did not use data from 1980 for adult survival estimation (whether derived from color-banding or mist-netting data). Nearly all successful nests (those fledging one or more young) are found before fledging and their young are individually color-banded (see further details in Geupel and DeSante 1990). Additional individuals are color-banded when first caught in mist nets as hatching year (HY) or as after-hatching year (AHY) birds.
Locals are defined as juveniles banded as nestlings and fledged from nests on the study plot. Nonlocals are juveniles first captured and banded in the standard operation of mist nets, and are assumed to have dispersed onto the study area. We attempted to identify all territory-holders on or adjacent to the study grid. This allowed us to classify all adults caught in mist nets (as well as those identified by color-bands only) as territory holders or non-territory holders. We refer to these two groups of individuals as "breeders" and "floaters," respectively (Stutchbury and Zack 1992).
Analysis of productivity and survival
The number of HY birds caught as part of a CEMN protocol reflects at least five factors: (1) Productivity, as measured in terms of fledged young per breeding pair (or per adult); (2) Post-fledging survival of fledglings (note that HY birds are usually not caught immediately after fledging but at least several weeks after fledging. For example, Song Sparrows at this site were first caught in mist nets 10 to 40 days after fledging [Nur et al. 1995]); (3) Breeding density in the area being sampled by the mist nets, which we refer to as "catchment area;" (4) Capture probability for HY birds in the catchment area; and (5) Flux of HY birds, i.e., movement of HY birds into and within the vicinity of the mist nets. Note that only (1) and (2) above reflect differences in productivity, broadly defined.
To infer patterns in productivity, mist-net data have been used in different ways: (a) analysis of total number of HY birds caught; (b) analysis of number of HY birds caught per AHY (After Hatching Year); or (c) analysis of % of all birds who are HY. Note that (c) is just a transformation of (b) and vice versa, assuming that all birds are classified as HY or AHY (i.e., assuming that the total = AHY + HY).
If one uses (b) or (c) to estimate productivity, then differences in capture rates among AHY individuals (and not just HY individuals) will influence the productivity index. The number of AHY individuals caught reflects: (1) Breeding density in the catchment area of AHY individuals; (2) Abundance of non-breeders in the AHY catchment area; (3) Capture probability for AHY birds in the catchment area, and (4) Flux of AHY birds, both breeder and non-breeders.
As a result of the factors enumerated above, there are hazards with including the number of AHY birds caught as a measure of productivity. First, the catchment area of HY and AHY can differ markedly (Nur and Geupel 1993b; Nur et al. in press). Hence if breeding density in the HY catchment area should change, without a change in breeding density in the AHY catchment area (or vice versa), the productivity index would be affected, even though no change in actual productivity would have occurred. Second, many AHY birds caught are either non-breeders (Nur and Geupel 1993a; see below) or transient, i.e., not breeding locally. Hence a change in the capture rates of nonbreeders would be reflected in a change in the productivity index. Pitfalls such as these should be kept in mind in the following analyses of HY and HY/AHY captures.
For both Song Sparrows and White-crowned Sparrows we compare indices of productivity determined from direct observation of nesting and fledging success with indices of productivity determined from mist-net capture data. Specifically, we compare HY captures per year with the total number of fledglings produced on the study plot (i.e., the number of young known to have fledged). Additionally, we evaluate whether the proportion of locally fledged young that are subsequently captured in mist nets varies significantly from year to year. If the aim of a monitoring program is to assess the production of fledged young (per year, at one or more sites) by way of mist-netting juveniles, then it would be of concern if the proportion of locally fledged young that are captured should vary between years.
Finally, for Song Sparrows we evaluate whether HY captures alone or HY/AHY captures together provide a better measure of temporal variation in productivity on the Palomarin study plot. We use linear regression analysis for statistical comparison of productivity indices. To analyze capture probability of fledged young we use logistic regression (Cox and Snell 1989), which is appropriate when the outcome is binary (here there are two states: captured or not captured). The statistical test used for logistic regression was the Likelihood Ratio Test (LRT) and its associated test statistic, the Likelihood Ratio Statistic (LRS; Lebreton et al. 1992). Under the null hypothesis, the LRS is distributed as a chi-squared random variable with degrees of freedom corresponding to the number of parameters being tested.
To analyze survival rates of adults, we used the statistical program SURGE (Lebreton et al. 1992, Cooch et al. 1996) and compared survival estimates derived from color-band observations of breeding individuals with estimates derived from mist-net capture/recapture data. Capture/recapture analyses such as SURGE allow one to separately estimate a survival probability, as well as a recapture (or, equivalently, a resighting) probability. The latter is defined as the probability that an individual is recaptured (or resighted) at a given time period, given that it has survived to that time period (Nur and Clobert 1988). Statistical tests of survival and recapture probabilities were carried out using the Likelihood Ratio Test. Note also that in this study, mist-net data alone cannot establish whether an individual is a breeder (Nur and Geupel 1993a); whereas individuals with well-developed brood patches can be assumed to be breeders, and those without well-developed brood patches are a mixture of breeders and floaters (unpublished data).
RESULTS AND DISCUSSION
Evaluation of productivity indices
Productivity of Song Sparrows and White-crowned Sparrows
The greater the number of Song Sparrows known to have fledged from the study plot in a given year, the greater the number of HY birds caught in mist nets that year (Figure 1A; R
2=0.802, P=0.0001, N=12 years; capture rates adjusted in all cases for number of net-hours each year). The relationship was remarkably tight. For White-crowned Sparrows, as well, the number of fledglings produced on the study plot correlated well with the number of HY birds caught (Figure 1B; R2=0.673, P=0.0011, N=12 years). For both species, data from 1980 and 1987 were excluded because of low effort at nest-finding (see above); however, the relationships still were significant, even when such data were included.| Figure 1. Comparison of Local Productivity of Song Sparrows (part A) and White-crowned Sparrows (part B) in relation to number of total HY caught for each species in each breeding season (whether local or not) per 1000 net-hours. X-axis for each panel shows number of young that fledged on the study plot (see text). '80' refers to 1980 breeding season, etc. In each Figure panel, the solid line is the best least squares fit to the data. Regression statistics are: R2 = 0.802, P = 0.0001 (part A); R2 = 0.673, P = 0.0011 (part B). |
 |
 |
Among locally fledged young of both species, the proportion that were subsequently captured in mist nets varied significantly among years (LRT, using logistic regression; Table 1). This finding is ambiguous: annual variation in the proportion of locally fledged young that are subsequently captured may reflect differences in the probability of capture among years, or it may reflect differences in post-fledging survival (i.e., in some years fledglings may have died before they could be caught). Examination of Table 1 indicates that 1990 was unusual: for both species, it was the year with the lowest capture proportions. For example, among Song Sparrows, only 5% of fledglings were subsequently caught in mist nets, compared to the median percentage of 28%. Even if 1990 is excluded from the analyses, significant year-to-year variation in capture proportions still was evident among locally fledged Song Sparrows and White-crowned Sparrows (Table 1).
Table 1. Capture probability of locally fledged young, captured in first summer, and proportion of all Hatching Year (HY) birds caught that were of local origin, in relation to year.
A) Song Sparrows
| Year | Fledglings N | Percent Captured in Mistnests1 | Total HY Caught N | Percent of HY that are local |
| 1980 | 67 | 31 | 183 | 12 |
| 1981 | 49 | 22 | 83 | 13 |
| 1982 | 137 | 32 | 188 | 23 |
| 1983 | 105 | 15 | 144 | 11 |
| 1984 | 103 | 22 | 123 | 19 |
| 1985 | 164 | 29 | 157 | 31 |
| 1986 | - | - | 59 | - |
| 1987 | 12 | 33 | 119 | 3 |
| 1988 | 56 | 41 | 66 | 35 |
| 1989 | 100 | 27 | 116 | 23 |
| 1990 | 39 | 5 | 51 | 4 |
| 1991 | 52 | 29 | 43 | 35 |
| 1992 | 66 | 30 | 84 | 24 |
| 1993 | 69 | 19 | 76 | 17 |
| 1994 | 69 | 33 | 89 | 25 |
B) White-crowned Sparrows
| Year | Fledglings N | Percent Captured in Mistnets2 | Total HY Caught N | Percent of HY that are local |
| 1980 | 70 | 36 | 106 | 24 |
| 1981 | 53 | 43 | 64 | 36 |
| 1982 | 71 | 24 | 70 | 24 |
| 1983 | 43 | 19 | 32 | 25 |
| 1984 | 54 | 19 | 57 | 18 |
| 1985 | 46 | 20 | 21 | 43 |
| 1986 | - | - | 41 | - |
| 1987 | 20 | 15 | 43 | 7 |
| 1988 | 49 | 14 | 24 | 29 |
| 1989 | 45 | 22 | 40 | 25 |
| 1990 | 38 | 3 | 22 | 5 |
| 1991 | 44 | 21 | 29 | 31 |
| 1992 | 44 | 14 | 11 | 55 |
| 1993 | 32 | 9 | 11 | 27 |
| 1994 | 40 | 8 | 18 | 17 |
1
Significant year-to-year variation in capture probability: P = 0.0006 (LRT; excludes 1987 because of low sample size); P = 0.026 (LRT; excludes 1987 and 1990)2 Significant year-to-year variation in capture probability: P < 0.0001, all years (LRT); P = 0.0006 (LRT; excluding 1990); P = 0.0004 (LRT; excluding 1990 and 1987 [low sample size])
Results for Wrentits previously were presented in Nur and Geupel (1993a, 1993b). No relationship existed between the total number of HYs caught (which includes local and non-local young) and the number of young known to have fledged from the study plot (see Figure 4B in Nur and Geupel 1993a), but a significant relationship did exist between the number of locally fledged young that subsequently were caught and the total number of young that fledged from the study plot (Figure 2 in Nur and Geupel 1993a). The proximal explanation for this result is that the number of non-local Wrentit HYs caught is uncorrelated with productivity as measured on the study plot (Nur and Geupel 1993a). Nur and Geupel (1993a) were unable to distinguish between two competing hypotheses to explain this paradoxical result: (1) productivity outside the plot is not correlated with productivity on the plot, or (2) year-to-year differences in the number of non-local HYs caught does not reflect productivity differences, but rather differences in dispersal or capture probability.
Thus, the ability of mist-net capture data to index productivity appeared to be good for two species, Song Sparrows and White-crowned Sparrows; for the third species, the results are mixed. In previous studies, Feu and McMeeking (1991) found differences in the ability of mist nets to track annual variation in productivity among four British passerines. Bart et al. (1998) concluded that mist nets could adequately index annual variation in productivity of Kirtland Warblers (Dendroica kirtlandii).
Which criterion of productivity to use?
For Song Sparrows, we compared two different mist-net based measures of productivity (HY caught per 1000 net hours; HY caught per AHY caught) with two different direct measures of productivity on the study plot (total number of young fledged from the study grid; number of young fledged per breeding pair). HY captures alone were better correlated with total number of fledglings produced on the grid than was the ratio of HY to AHY birds (r = +0.915, P < 0.001; r = + 0.459, P = 0.16, respectively). In addition, HY capture rates were better correlated with the number of fledglings produced per pair than was the ratio of HY to AHY birds caught, though the difference was slight (r = +0.674, r = +0.669, respectively).
That the HY/AHY index of productivity did not perform better than the HY-only index may seem surprising. Incorporating information on capture rates of adults, as a means of estimating breeding density, did not improve our ability to estimate, for the study site, productivity per breeding pair. The problem is not, in our view, that AHY captures are unrelated to breeding density. On the contrary, in a separate analysis, we determined that capture rates of AHY Song Sparrows were significantly correlated with breeding density (as determined by observations of color-banded birds; Silkey et al., in press). Rather, the problem may be that the number of AHY birds caught is not a sensitive index of breeding density in the area that is producing the HY birds (i.e., the catchment areas for AHY and HY birds likely differ). Furthermore, at the Palomarin study site there were no marked fluctuations in Song Sparrow breeding density during the course of the study (as determined by observations of color-banded territorial birds; e.g., Helzer and Geupel 1996). In the absence of marked fluctuations in breeding density, we recommend the use of HY captures per net hour to index productivity. However, where breeding density fluctuates greatly, the HY/AHY index may prove a better index of productivity than simply using the number of HY captures per net hour. In the latter situation, incorporating a rough measure of breeding density (reflected in AHY captures) may be better than ignoring variation in breeding density.
Breeders and floaters
Two groups of adults are caught in CEMN: site-faithful (this includes, but is not limited to, territory-holders, i.e., breeders) and transient individuals. Failure to distinguish between these two groups can lead to serious bias in estimating survival (Peach 1993, Nur and Geupel 1993a). In this study we attempt to distinguish these two groups, first, on the basis of whether individuals held a territory, and second, on the basis of whether individuals were caught once or more than once per breeding season.
The number of breeders (i.e., territory holders) and non-breeders (i.e., non-territory holders or "floaters") varied among years. In most years, 40% or more of adults caught were non-territory holders (average = 53%). This has important implications. First, when estimating productivity, if one divides HY captures by AHY captures to estimate productivity (see above), the latter is reflecting abundance of the non-breeding segment of the population as well. Second, the large number of non-territory holders caught in the nets has important implications for survival estimation, because non-breeders tend to be much more transient than breeders (see below). Furthermore, the percent of adults caught that were floaters varied significantly among years (LRS = 26.39, df = 12, P = 0.01).
That the number of non-breeders caught in the nets equaled or exceeded the number of breeders caught does not imply that non-breeders equaled or outnumbered breeders in the population as a whole. The non-breeders were, for the most part, just "passing through" the study site, whereas breeders stayed put. Thus, over the course of the breeding season there was a high turnover of non-breeders, but not of breeders. The transience of these non-breeders is indicated by the fact that few were ever recaptured (in contrast to breeders, see below). Also, most (83%) of these non-breeders were first caught in the winter or spring of their after-hatching year (i.e., they were neither locally fledged nor caught in the nets in their first year of life). In contrast, 81% of breeders caught in nets had been locally fledged or were caught in nets during their first year of life. Thus, most non-breeders caught during the breeding season never had been seen before nor ever were seen again. We would expect transients to be an especially large component among captures of migrant species, a result that has been confirmed at the Palomarin field station for Swainson's Thrush (Catharus ustulatus, Johnson and Geupel 1996) and Wilson's Warbler (Wilsonia pusilla, Chase et al. 1997).
Our work has revealed that floaters are a large component of adult captures of Wrentits. In fact, Wrentit floaters constitute an even larger fraction of adult captures than is the case in Song Sparrows: X = 18.9 floaters caught per year vs. X = 9.1 breeders caught per year, n = 10 years (PRBO unpublished data; see also Nur and Geupel 1993a). As with Song Sparrows, this difference in capture rates may not necessarily reflect the numerical dominance of floaters in this Wrentit population. Rather, the difference in capture rates reflects the fact that breeders show very high site fidelity, whereas floaters are usually very transient. Only breeders within 200 m of the nearest mist net were caught at least once during the breeding season (Nur and Geupel 1993b; Nur et al. in press); those breeding farther away almost never were caught. In contrast, floaters were very transient, and thus over the course of the breeding season a high number "passed through" our study site. Once caught, most Wrentit floaters were never seen again. Only 52 out of 189 floaters (28%) were seen again in the same season, either caught in mist nets or observed and identified by means of their color-bands. In contrast, most breeders were caught repeatedly during the breeding season. Sixty-two percent of breeders were recaught in the same breeding season (n = 91), some repeatedly. The highly transient nature of floaters has important implications for estimation of adult survival using CEMN, as discussed below.
In Wrentits, even more so than Song Sparrows, the number of floaters caught each breeding season varied markedly, between 3 and 34. We examined demographic influences on variation in Wrentit floater abundance (Geupel et al. in prep.). Using a stepwise multiple regression analysis we found that floater capture rate (floaters caught per 1000 net-hours) in year x varied most strongly in relation to two demographic parameters: (1) breeding density in year x (i.e., the more pairs breeding, the fewer the vacancies and hence the greater the number of floaters), and (2) productivity in year x - 1 (i.e., the greater the production of young during the previous year, the more floaters are present the next year). Once productivity in the previous year was controlled for statistically, there was a significant relationship between floater capture rate and breeding density (P=0.006, Figure 2); conversely, after controlling for the effect of breeding density, the effect of productivity was significant (P = 0.01; Geupel et al. in prep.).
| Figure 2. Relationship of Wrentit floater abundance in relation to Wrentit breeding density, after both are statistically controlled for effect of production of Wrentit young during the year before. The figure, referred to as an "Added Variable Plot" (STATA Corp. 1997), graphically presents the results of a multiple regression analysis (dependent variable is floater capture rate; independent variables are breeding density and productivity the year before) and illustrates the effect of breeding density on floater capture rate. Each data point represents data from one year; the solid line is the best least squares fit to the data. On y-axis is shown floater capture rates (number of floaters caught per 1000 net-hours, standardized to a mean of zero); on x-axis is shown number of breeding pairs on the study site (standardized to a mean of zero). The effect of breeding density is highly significant, P=0.004, after controlling for the effect of productivity. |
|
Survival estimation
Analyses of survival and recapture probability (i.e., the probability that a bird that has survived to year x is caught in year x), were conducted using the statistical program SURGE on the Song Sparrow mist-net capture data (data from 323 different individuals caught between 1981 and 1994) and the results were compared with detailed observations on individually color-banded Song Sparrows (420 different individuals).
We used SURGE to estimate survival and recapture probability for adult Song Sparrows based on observations of territory-holding color-banded birds (i.e., individuals that did not hold territories are not included in this data set). An individual that was actually observed in a given year, but not as a confirmed territory holder, is considered to be "not observed." Thus failure to be "resighted" incorporates failure to hold territory among experienced breeders. Adult survival for color-banded Song Sparrows is shown in Table 2A: males survived better than females (64% vs. 56%); mean survival was 60.0%. Resighting probability was high for both sexes (87% to 92%). There was an indication of annual variation in survival, but this was not significant for either sex (Table 2B).
Table 2. Song Sparrow Adult Survival in Relation to Sex and Year.
A) Survival and resighting probabilities for color-band observations of breeding males and females, from SURGE (N = 420 individuals).
| Survival (+ S.E.) | Resighting (+ S.E.) | |
| Male: | 0.644 + 0.021 | 0.918 + 0.020 |
| Female: | 0.556 + 0.026 | 0.874 + 0.031 |
| P = 0.0081 | NS1 |
B) Year-to-year variation in survival as estimated from color-band observations of breeding males and females (columns 2 and 3; N = 420 individuals) and as estimated from SURGE analyses of mist net capture/recapture data (column 4; N = 323 individuals, both sexes combined). For the latter analysis, adults were classified as single- or multiply caught, and separate recapture probabilities were allowed for each class of adult. Recapture probability did not vary significantly between years (LRT) and was assumed constant with year for all analyses.
| Color-band observations | Mist-net observations | ||
| Year | Male2 | Female2 | Adults2 |
| 1981 | 0.540 + 0.24 | 0.527 + 0.11 | 0.500 + 0.22 |
| 1982 | 0.648 + 0.23 | 0.576 + 0.12 | 0.274 + 0.13 |
| 1983 | 0.714 + 0.071 | 0.706 + 0.14 | 0.728 + 0.18 |
| 1984 | 0.569 + 0.069 | 0.383 + 0.12 | 0.410 + 0.12 |
| 1985 | 0.584 + 0.073 | 0.386 + 0.15 | 0.915 + 0.17 |
| 1986 | 0.642 + 0.066 | 0.671 + 0.15 | 0.530 + 0.15 |
| 1987 | 0.749 + 0.071 | 0.652 + 0.16 | 0.519 + 0.16 |
| 1988 | 0.624 + 0.075 | 0.568 + 0.11 | 0.608 + 0.19 |
| 1989 | 0.767 + 0.13 | 0.555 + 0.20 | 0.512 + 0.18 |
| 1990 | 0.670 + 0.081 | 0.546 + 0.11 | 0.608 + 0.20 |
| 1991 | 0.601 + 0.15 | 0.717 + 0.11 | 0.576 + 0.24 |
| 1992 | 0.528 + 0.11 | 0.546 + 0.12 | 0.442 + 0.34 |
1 LRT, df = 1, comparing the two sexes with respect to survival and resighting probabilities.
2 No significant year-to-year variation (LRT, df = 11, P>0.1) for males or females or adults.
Use of double-capture criterion to distinguish transients
In most CEMN protocols it would not be possible to distinguish breeders (or territory-holders) from non-breeders (or floaters). Therefore, we have considered a mist-net based criterion for distinguishing transients from site-faithful individuals. That criterion is the following: individuals caught at least two times during the breeding season (and at least 7 days apart) are presumed to be "resident," the others are presumed transient (similar criteria adopted by Peach et al. 1991, Peach 1993, Chase et al. 1997). In our analyses, an individual that is "resident" in any one year is assumed to be resident in all years. Indeed there is a significant association between multiple capture and breeding status (P < 0.0001, LRS = 44.43, df = 1). There were 79 double-caught Song Sparrow adults, of these 57 (72%) were known breeders. However, most breeders were not double-caught: among 157 known breeders that were caught at least once during the study period, only 57 (36%) were double caught in at least one breeding season. Thus the double-capture criterion is not very efficient at identifying Song Sparrow breeders. Nonetheless, this criterion appears to provide a good means of obtaining estimates of adult survival that accord well with estimates based on color-band observations (Table 3).
Table 3. Survival and Recapture Probability of Song Sparrows. Results of SURGE Analyses on mist-net capture data for adults, 1981-1994, with respect to multiple-capture criterion (caught more than once in at least one year vs. never caught more than once in a season).
A) Separate estimates of survival and recapture for multiply caught and for not multiply caught (N = 79 individuals and N = 244 individuals, respectively).
| Survival Probability | Survival S.E. | Recapture Probability | Recapture S.E. | ||
| Multiply caught | 0.6211 | 0.044 | 0.4752 | 0.065 | |
| Never multiply caught | 0.4131 | 0.066 | 0.1842 | 0.054 |
B) Separate estimates of recapture probability for multiply caught and for not multiply caught; pooled estimate of survival (sample size as in part A).
| Survival Probability | Survival S.E. | Recapture Probability | Recapture S.E. | ||
| Multiply caught3 | 0.508 | 0.064 | |||
| Pooled estimate (multiply and not multiply caught) | 0.564 | 0.038 | |||
| Never multiply caught3 | 0.109 | 0.025 |
C) No differentiation between multiply caught and never multiply caught (N = 323 individuals).
| Survival Probability | Survival S.E. | Recapture Probability | Recapture S.E. | ||
| All Adults | 0.475 | 0.038 | 0.302 | 0.046 |
1 - Comparison of multiple vs. single capture: LRS = 6.21, df = 1, P = 0.013.
2 - Comparison of multiple vs. single capture: LRS = 9.43, df = 1, P = 0.002.
3 - Comparison of multiple vs. single capture: LRS = 44.00, df = 1, P < 0.0001.
Differentiating birds according to the multiple-criterion yields estimated survival of 62% for double-caught birds (assuming different recapture probabilities for single-caught and double-caught birds) and 41% for birds that never were multiply caught, a significant difference (P = 0.013, LRT). A pooled estimate of survival for both classes of adults is 56% (assuming different recapture probabilities for the two groups; Table 3). The survival estimate for double-caught birds only and the pooled survival estimate for both classes of birds agree closely with estimates from color-band observations. In fact, any discrepancy between the color-band estimate and the estimate for double-caught individuals is due to 1993/1994 data included for the latter and not for the former; when both data sets cover exactly the same years (1981-1993), the two survival estimates yield nearly identical estimates: 60.0% and 59.9% survival for color-banded and mist-net data. Note that double-caught birds are a mixture of mostly known-territory holders (comprising 72% of double-caught birds), with some presumed floaters (the other 28%). Despite differences in breeding status, it appears that both types of double-caught individuals demonstrate relatively high site fidelity between years.
The double capture criterion excludes many individuals who are likely resident, and not transient. An alternative criterion is to carry out analyses on individuals caught in at least two different years. The assumption is that such individuals are site-faithful, not transient (Pradel et al. 1997). We conducted an analysis of survival using individuals who were first caught as HYs and then caught at least once as AHY individuals. For the Song Sparrow population there were 110 such individuals. Adult survival of these individuals (from their first capture as AHY individuals, and excluding 1993/94 to make the estimate compatible with analyses of color-banded individuals) was estimated to be 52.8% (S.E. = 4.6%); less than the color-band derived estimate of survival, and less than the double-capture derived estimate. Hence, use of this criterion (survival of individuals caught at least once as a AHY individual and previously caught as a HY individual), rather than the double-capture criterion, increases sample size—thus improving precision—but decreases accuracy.
Year to Year Variation in Adult Survival
SURGE analyses of color-band observations and mist-net capture/recapture data indicated between-year variation in adult survival (Table 2). Annual survival estimates using mist-net capture data varied by more than two-fold, as did estimates of annual adult survival based on color-band observations (Table 2). However there was no significant correlation between estimates of adult survival in each year obtained from the mist-net data and estimates of survival obtained from observations on color-banded individuals (linear regression, P > 0.2). This was the case whether the data sets for color-banded individuals included only males, only females, or both sexes pooled. The lack of concordance between year-specific estimates from the two types of data collection calls into question the ability of mist nets to track annual variation in adult survival. However, we must consider the possibility that the color-band observational data are insufficient to accurately estimate survival on a year-by-year basis (as opposed to a long-term average, see above). Marshall et al. (this volume) discuss whether observations of color-banded individuals can be used to accurately estimate adult survival.
Other demographic parameters
Constant-effort mist-netting can be used to investigate a number of other demographic parameters, discussion of which lies outside the scope of this paper. Examples include age structure of adults (where Second Year adults can be distinguished from After Second Year individuals), dispersal (of juveniles and adults), and offspring survival and recruitment (see Nur et al. 1995). Offspring survival and recruitment is especially important, yet information is sorely lacking (Perrins 1991, Clobert and Lebreton 1991, Sherry and Holmes this volume).
CONCLUSIONS AND IMPLICATIONS FOR AVIAN MONITORING PROGRAMS
These results demonstrate that temporal variation in productivity can be estimated well using CEMN. This is not a universal conclusion (Wrentits being a possible exception), but does appear to be well justified for Song Sparrows and White-crowned Sparrows. This study did not address the practice of using CEMN data to assess spatial differences in productivity (e.g., DeSante et al. 1993), which should be the target of a future validation study. Furthermore, we do not recommend that differences in HY capture rates, or comparisons of HY/AHY ratios, be used to assess between-species differences with respect to productivity. For example, Vansteenwegen (in press) compared HY/AHY ratios for 24 species with large sample sizes, as part of a national CEMN program in France. There was a 100-fold difference in this ratio among species! It is unreasonable to maintain that such differences in HY/AHY ratios reflect mainly differences in productivity; as our own data suggest, most of the differences surely reflect different capture probabilities of adults and juveniles.
Mist-net capture/recapture data appear to give good estimates (estimates with small bias) of adult survival, provided that transients and non-transients can be distinguished. In the absence of any distinction between these two classes, survival estimates will be substantially biased (by 14 percentage points for Song Sparrows [i.e., 0.62 vs. 0.48, Table 3]; by 30 percentage points for Wrentits [Nur and Geupel 1993a]). The double-capture criterion (two or more captures in the same breeding season, at least 7 days apart) appears to be a good means to make this distinction, at least for Song Sparrows. The double-capture criterion also produced much improved estimates of Wilson Warbler survival, compared to not distinguishing transients at all (Chase et al. 1997). Note that the bias in estimating survival if transients and site-faithful individuals are not distinguished lies not so much in survival differences between the two groups, but arises because recapture probability is poorly estimated (being very different for the two groups of individuals), which results in poor estimates of survival. When recapture probabilities are separately estimated for each group and a combined survival probability estimated, there is some bias but not that great (0.62 for double-caught individuals only vs. 0.56 for the combined survival estimate [Table 3], a difference of six percentage points).
We recommend a two-pronged approach: for obtaining estimates of absolute survival, we recommend confining analysis to double-captured individuals (Table 3). However, for obtaining an index of year-to-year variation in adult survival, we recommend use of statistical models that incorporate all individuals, with separate recapture probabilities estimated for single-caught and multiply caught individuals, and a pooled estimate of survival probability for each year (as done in Table 2). If a study can obtain large sample sizes of double-caught individuals each year, there would be little reason to include single-caught individuals as well. An implication of this approach is that, in establishing a constant-effort mist-netting program, one goal would be to maximize the number of adults recaught, not just caught. Running nets often (at minimum, two or more times per 10-day period) would further that goal.
We make the following additional recommendations regarding implementation of monitoring programs on the basis of results from our intensive study of demography of several passerine species. (1) We recommend that calculations of adult survivorship from mist-net capture/recapture data always attempt to distinguish transients from non-transients. Transients (mostly floaters) appear to be a large component of adult catches of both year-round resident species and migrant species. (2) More work needs to be done to establish the best means to distinguish transients from non-transients. The double-capture criterion appears to yield valid estimates of survival, but, for some species, may be too restrictive; other criteria should also be investigated. (3) We recommend running mist nets more than once every 10 days. This will increase the number of adults which are double-captured, and recaptured in the following year, thus improving the accuracy and precision of survival estimates. Running nets more often also will increase the sample size of juveniles that are caught, improving our ability to infer trends in productivity and juvenile survival/recruitment. It is likely that recapture probabilities differ between sites (Peach 1993) and thus adequate sample sizes are required to estimate site-specific parameters. The optimal frequency with which to run mist nets should be the topic of future investigation. (4) While mist nets appear to provide a good means of indexing year to year variation in productivity, at least in coastal scrub habitat, more work is needed to evaluate the use of mist-net capture data to compare productivity at different sites and in different habitats.
We conclude that there is great value to combining CEMN with intensive observations of color-banded individuals. The use of several techniques in concert, rather than being redundant, provides one with information that could not otherwise be attained. An example of the power of a multi-level, integrated approach is the ability to track abundance of non-breeders (floaters). Through spot-mapping or point counts we can potentially track the number of breeders, but floaters are an important component of avian demography, yet are hard to observe in the field due to their secretive nature. Mist-netting allowed us to track the total number of adults (whether breeding or not), but by itself could not tell us which were breeders. Putting both together (total adult abundance and breeder abundance) allows inference about non-breeder abundance. This allowed us to correlate the abundance of non-breeders with several demographic parameters. The ability of a population to respond to perturbation might well be related to the existence and abundance of non-breeders (as exemplified by the Northern Spotted Owl Strix occidentalis caurina, Verner 1992).
ACKNOWLEDGMENTS
The research presented here was made possible, in part, by funding through Grant Agreement 14-48-0009-94-1272 with the U.S. Department of Interior Fish & Wildlife Service, Office of Migratory Bird Management. We thank the numerous intern field biologists who helped collect data reported here, and thank B. Denise Hardesty and Oriane E. Williams for help in preparing and analyzing field data. We thank the Point Reyes National Seashore for their continued cooperation, and Dr. Jean Clobert for his guidance at an early stage of the work. We are grateful to Drs. L. Richard Mewaldt, C. John Ralph, and David F. DeSante for their foresight in establishing a long-term monitoring and research program at Palomarin. This is PRBO Contribution Number 771.
LITERATURE CITED
Askins, R. A. 1993. Population trends in grassland, shrubland, and forest birds in eastern North America. Current Ornithology 11:1-34.
Baillie, S. R., R. E. Green, M. Boddy, and S. T. Buckland. 1986. An evaluation of the constant effort sites scheme. Unpublished report to the British Trust for Ornithology.
Bart, J., C. B. Kepler, P. W. Sykes, Jr., and C. I. Bocetti. 1998. Evaluation of mist-net sampling as an index of productivity. Paper presented at the 1998 North American Ornithological Conference, St. Louis, MO, April 1998.
Bibby, C. J., N. D. Burgess, and D. A. Hill. 1992. Bird census techniques. Academic Press, London.
Chase, M., N. Nur, and G. R. Geupel. 1997. Survival, productivity, and abundance in a Wilson's Warbler population. Auk 114:354-366.
Clobert, J., and J. -D. Lebreton. 1991. Estimation of demographic parameters in bird populations. pp. 75-104 in Bird Population Studies. (C. M. Perrins, J. -D. Lebreton, and G. J. M. Hirons, eds.) Oxford Univ. Press, Oxford.
Cooch, E., R. Pradel, and N. Nur. 1996. A Practical Guide to Capture/Recapture Analysis using SURGE. 125 pages. Centre d'Ecologie Fonctionelle et Evolutive - CNRS, Montpellier, France.
Cox, D. R. and E. J. Snell. 1989. Analysis of Binary Data. Chapman & Hall, London.
DeSante, D. F., O. E. Williams, and K. M. Burton. 1993. The Monitoring Avian Productivity and Survivorship (MAPS) Program: Overview and Progress. pp. 208-222 in Status and Management of Neotropical Migrant Birds. D. M. Finch and P. W. Stangel, eds. USDA Forest Service Publication: GTR RM-229, Fort Collins, Co.
DeSante, D. F. and G. R. Geupel. 1987. Landbird productivity in central coastal California: the relationship to annual rainfall and a reproductive failure in 1986. Condor 89:636-653.
Erickson, M. M. 1938. Territory, annual cycle, and numbers in a population of Wrentits Chamaea fasciata. Univ. Calif. Publ. Zool. 42:247-333.
Feu, C. du and McMeeking, J. 1991. Does constant effort netting estimate juvenile abundance? Ringing and Migration 12:118-123.
Finch, D. M. and P. W. Stangel. 1993. Status and Management of Neotropical Migrant Birds. USDA Forest Service Publication: GTR RM-229, Fort Collins, Co.
Geupel in Prep??????????
Geupel, G. R. and D. F. DeSante. 1990. Incidence and determinants of double brooding in Wrentits. Condor 92:67-75.
Hagan J. M. and D. W. Johnston (eds.) 1992. Ecology and Conservation of Neotropical Migrant Birds. Smithsonian Institution Press, Washington, D. C.
Helzer, D. P. and G. R. Geupel. 1996. 1995 Breeding Bird Census: Coastal Scrub. Journal of Field Ornithology 67 (Supplement):77-78.
Johnson, N. K. 1972. Origin and differentiation of the avifauna of the Channel Islands, California. Condor 74:295-315.
Johnson, M. and G. R. Geupel. 1996. The importance of productivity to the dynamics of a Swainson's Thrush population. Condor 98:133-141.
Lebreton, J. -D., K. P. Burnham, J. Clobert, and D. R. Anderson. 1992. Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecological Monographs 62:67-118.
Martin, T. E. and D. M. Finch, eds. 1995. Ecology and management of Neotropical migratory birds: a synthesis and review of critical issues. Oxford Univ. Press, New York.
Nice, M. M. 1937. Studies in the life history of the Song Sparrow. I. Trans. Linn. Soc. New York 4:1-247.
Nur, N. and J. Clobert. 1988. Measuring Darwinian fitness in birds: A field guide. Acta XIX Congressus Internationalis Ornithologici. H. Ouellet (Ed.) University of Ottawa Press. 2121-2130.
Nur, N. and G. R. Geupel. 1993a. Evaluation of nest monitoring, mist-netting, and other methods for monitoring demographic processes in landbird populations. Pp. 237-244 in Status and Management of Neotropical Migrant Birds. (D. M. Finch and P. W. Stangel, Eds.). USDA Forest Service Publication: GTR RM-229, Fort Collins, Co.
Nur, N. and G. R. Geupel. 1993b. Validating the use of constant effort mist-netting to monitor avian populations. Report of the Point Reyes Bird Observatory to U. S. Fish and Wildlife Service, Office of Migratory Bird Management (available from PRBO, Stinson Beach, CA 94970).
Nur, N., G. R. Geupel, and G. Ballard. 1995. Validating the use of constant effort mist-netting to monitor avian populations. II: Studies of Song Sparrows, Wrentits and other species. Report of the Point Reyes Bird Observatory to U. S. Fish and Wildlife Service, Office of Migratory Bird Management (available from PRBO, Stinson Beach, CA 94970).
Nur, N., G. R. Geupel, and G. Ballard. In press. Estimates of adult survival, capture probability, and recapture probability: evaluating and validating constant effort mist netting. In: Proceedings of First International Mistnet Conference, Point Reyes, CA. (ed. by C. J. Ralph and W. Peach). USDA Forest Service Gen. Techn. Rep.
Peach, W. J., S. R. Baillie, and L. Underhill. 1991. Survival of British Sedge Warblers Acrocephalus schoenobaenus in relation to west African rainfall. Ibis 133:300-305.
Peach, W. J. 1993. Combining mark-recapture data sets for small passerines. pp. 107-122 in Marked Individuals in the Study of Bird Populations. (J.-D. Lebreton and P. M. North, Eds.) Birkhauser, Basel.
Perrins, C. M. 1991. Constraints on the demographic parameters of bird populations. pp. 190-206 in Bird Population Studies. (C.M. Perrins, J.-D. Lebreton, and G. J. M. Hirons, eds.) Oxford Univ. Press, Oxford.
Pradel, R. and J. –D. Lebreton. 1993. User's manual for Program SURGE Version 4.2. CEFE/CNRS, Montpellier, France.
Pradel, R., J. E. Hines, J.-D Lebreton, and J. D. Nichols. 1997. Capture-recapture survival models taking account of transients. Biometrics 53:60-72.
Shuford, D. 1993. The Marin County Breeding Bird Atlas. Bushtit Books, Bolinas, CA.
Silkey, M., N. Nur, and G. R. Geupel. In press. Do mistnets adequately monitor population trends in abundance? a validation study. Condor.
STATA Corporation. 1997. STATA Statistical System, Release 5.0. STATA Corporation, College Station, Texas.
Stutchbury, B. and S. Zack. 1992. Delayed breeding in avian social systems: the role of territory quality and floater tactics. Behaviour 123:194-219.
Temple, S. A. and J. A. Wiens. 1989. Bird populations and environmental changes: Can birds be bio-indicators? American Birds 43:260-270.
Vansteenwegen, C. In press. Monitoring breeding bird populations by capture in France. In Proceedings of the First International Conference on Use of Mistnets to Monitor Avian Populations. (C. J. Ralph and W. Peach, Eds.). USDA Forest Service Gen. Technical Report.
Verner, J. 1992. Data needs for avian conservation biology: Have we avoided critical research? Condor 94:301-303.
1Point Reyes Bird Observatory
4990 Shoreline
Highway
Stinson Beach, CA
94970