Abstract Size 7-B aluminum bands were recovered from Laysan (N=124) and Black-footed Albatrosses nesting at Midway Atoll in 1979 and 1992-93. Each band was weighed to the nearest 0.001 g and etched if illegible. Bands recovered from Laysans were carried from 0.6 to 30.4 years and lost an average of 2.42% of their weight per year (range 0.79-7.13%/yr). Band loss commenced about year 13 for most series, but as early as eight years in one corroded 807- prefix series. The oldest legible band was carried only 19 years; all bands worn longer required etching to recover the number; half were illegible by 14 years. When the average Laysan band had lost 54% of its original weight, it fell off. Once band loss begins, it adds an error of 9.45%/year to the estimated annual mortality rate. Aluminum bands on Black-footeds averaged 1.4%/year weight lost (range 0.22-2.95%./yr.), with just 17 of 217 bands losing >2% of their weight/year. One band was recovered in the field shortly after it fell off; it had been carried 32.9 years and had lost 54% of its initial weight. Black-footeds will retain aluminum bands 23 years, and 50% of banded Black-footeds will retain aluminum bands for 37 years. Band loss will begin to bias data seriously after year 23 and be lost at the rate of 5.4%/year thereafter. Although many bands carried by Black-footeds longer than 15 years required etching to recover the number, 23% of bands carried 25.7-33.7 years were legible. Aluminum bands are not suitable to provide long-term survivorship and life table data for Laysans but are marginally suitable for Black-footeds. No difference was found in band wear rates among Black-footed Albatrosses banded as chicks and those banded as adults. Rebanding of albatrosses carrying aluminum bands should be a priority.

The SERE Group, Ltd. Box 138 Concession 2 Kingsville Ont. N9Y 2K7 The SERE Group, Ltd. 2633 Willowbrook Rd. Elsie, MI 48831 4471 Hippin Lane Freeland, WA 98249 World Wildlife Fund 1250 24th St. NW Washington, DC 20037 2864 Military St. Port Huron, MI 48060 4471 Hippin Lane Freeland, WA 98249 INTRODUCTION The nature of band loss makes measuring loss rates difficult (Kadlec 1975). Abrasion may complicate the interpretation of data sets since band numbers may be recovered by professionals using etching techniques, but only if they have access to the illegible band (F.E. Ludwig 1941). For very long-lived species like albatrosses, loss of conventional aluminum bands is virtually certain to influence survivorship data (J.P. Ludwig 1981). Numerous authors have analyzed data for one species of interest to them (Anderson 1980, Bailey et al. 1987, DiCostanzo 1980, Hickey 1952, Poulding 1954, Spear 1980, Ludwig 1967), often speculating that band loss was an important source of error in their life table analyses (Hatch and Nisbet 1983, Ludwig 1967, Paynter 1966). Others had too few recovery data to correct life tables (e.g., Schreiber and Mock 1988, Fordham 1967). Various models of band loss have been proposed (Brownie 1973; J.P. Ludwig 1967, 1981; Nelson et al. 1980). Reported wear rates of aluminum bands range from a minimum of 0.57%/year in Sooty Terns (Stema fuscata) (Bailey et al. 1987) up to a maximum of >9.5%/yr in Ring-billed Gulls (Larus delawarensis) (Ludwig 1967) and Eared Grebes (Podiceps nigricollis) (Jehl 1990). Corrosion can cause the loss of monel and aluminum bands (Ludwig 1981). The mass of band worn away before loss occurs is also widely variable from 14% in Sooty Terns to 60-70% in Caspian Terns (Stema caspia) and Ring-billed Gulls. Each species seems to have a characteristics mass loss rate for each metal type and a typical end point. Empirical studies of bands recovered from birds offer the clearest picture of the ways that band loss influences life tables (Hatch and Nisbet 1983; Ludwig 1967, 1981 ). A typical survivorship curve is produced for data from birds banded as chicks that lose bands (Ludwig 1967, 1981; Bailey et al. 1987). This semi-logarithmic curve is character- ized by four phases: (1) An initial rapid downturn phase in the first year post-fledging reflecting high first- and second-year mortality rates; (2) a linear section of varying length in the early adult life (four years in Ring-billed Gulls [Ludwig 1967]; 10-22 in Caspian Terns [Ludwig 1981]; 17-24 in Sooty Terns [Bailey et al. 1987]); (3) a curvilinear section where the apparent ra.e of death rises as band loss begins; and (4) a final steeper decline section where the rate of band loss adds to the real mortality (Ludwig 1967, 1981). Mortality estimates of older birds are typically grossly inflated by the effect of band loss. For short-lived species, this effect is largely irrelevant. However, these errors can lead to serious underestimation of age- specific survivorship and highly inaccurate population models for long-lived species that reproduce for decades. Two north Pacific albatrosses, Laysan (Diomedea immutabilis) and Black-looted (Diomedea nig#pes), were banded with aluminum butt end style bands beginning in the late 1930s. Large numbers were tagged after 1955. Experimental monel metal bands were applied to chick cohorts in 1967. Cohorts of Laysan and Black-looted chicks were banded with stainless bands in 1978 and 1979, respectively. After 1986, many previously banded albatrosses were rebanded with incaloy and stainless steel bands, and a few chicks were banded with these newer"hard" bands after 1984, as well as the readily available aluminum butt-end style bands. Because most of the albatrosses were banded in the north Pacific colonies with conventional aluminum bands, it is essential to study their wear patterns if those data are to be used to model populations or generate life tables. In 1979, while rebanding several thousand breeding adults of each species, the senior author recovered 173 aluminum bands from albatrosses at Midway. From November, 1992 through Page 158 I December, 1993, an additional 223 bands were recovered. Five monel bands were recovered in this same period from Laysan Albatrosses found dead. We describe patterns and rates of wear in the 346 of these 401 bands for which banding data were available. Our goal is to determine when band loss begins and to measure how it affects albatross survivorship data. MATERIALS AND METHODS We used the simple empirical band loss estimation model developed by Ludwig (1967, 1981). This model requires a sample of bands with 95% of the wear rates in a range of the mean value + 2/3 of the mean, and a normal distribution of wear rates to be used. Only 2 of 112 uncorroded Laysan bands and four of all 217 Black-footed bands fell outside of these distribution and the range requirements. This model derives estimates from bands with particular wear rates and a known endpoint and projects the additive error for a life table calculated from raw banding data. We compared used albatross bands with 1226 new 7-B bands from three series; 1,000 from a 587-prefix series manufactured in the mid-1950s, 26 from a 1227- series manufactured in 1977; and 2OO from a 1247-series manufactured in 1981. These reference bands averaged 1.846, 2.110, and 1.832 grams respectively (Table 1). Within each series, the variance was quite small (SDs 6.1- 22.1 ). The newest aluminum bands manufactured from 1980 were heavier but more variable. We used 1.846 g for the base weight of calculations, except for the five bands made after 1981 for which 2.11 g was used. First, bands placed on Black-footed chicks were separated from those placed on adults, but we Table 1. Weights of Representative New 7-B Butt-end Aluminum Bands. Mean Weight Band Series (g) Range S.D. 587-67001-68000 1.846 1.793-1.867 11.3 1227-26675-2670 1.832 1.813-1.848 6.1 1247-87001-8720 2.110 2.049-2.139 22.1 I could not detect measurable differences in wear rates between these groups (chick mean 1.38%/ yr, SD- 0.27; adult mean 1.41%/yr, SD- 0.36; 0.1055% weight. Since we recaptured thousands of banded birds of these species including 137 birds known to be 30 years or older, if bands were surviving the loss of more than 55% of initial mass regularly, we would have encountered them. From these observations, we selected the average end points of 54% and 56% of metal loss for Laysan and Black-looted Albatrosses, respec- tively. Nearly half the bands (186) carried longer than 10 years by Laysan Albatross or 18 years by Black- looted Albatross were partially or completely illegible and were etched to recover the number. Every Laysan band older than 19 years required etching, but some Black-looted bands as old as 33.7 years were still legible. Numbers were found on all these bands, but up to four etchings were required for a few bands. Experimental etching of Oct. - Dec. North American three bands revealed an average loss of 2.3 mg weight per treatment with acid etch solution to the whole numbered surface of the band. This could add an error of up to 1% to the wear calculations made for a band that had lost 40% of its base weight and was completely etched four times. This potential error was ignored since it was close to the vadance of weights of new bands. Most bands required only 1 or 2 etchings on part of the band surface to recover one to three digits of the number. RESULTS Laysan Albatross bands averaged 2.46%/year of metal loss with a wide range of rates (0.78 - 7.06%). The fastest rates, 5.39 - 7.06%/year were by 12 bands all in a series of 807-prefix bands which had corroded: these were the only corroded aluminum bands encountered (Table 2). These 12 bands were eliminated from the analysis as atypical. The average 112 uncorroded Laysan band lost 2.42%/year (range 0.78 - 4.75%); 98% of this band sample fell within the range required by the Ludwig (1981) model. With a 54% endpoint and an average wear rate of 2.42%/year, half of the bands would last 22 years. Band loss for a cohort of Laysans, calculated from the mean value (2.42%/yr + 1.61 = 0.81 to 4.03%/year wear rates) would begin at 13 years, and the last band will be lost at 52 years (Tables 3 and 4). Many Laysan bands were already illegible after 13 years; thus, the effects of wear on the data will be even greater than this calculation unless all the bands found are etched for data used to calculate life tables. Sixty of the 124 bands in this sample were etched. One subsample of 50 Laysan bands in an 1117-prefix series applied by the senior author in 1979 was recovered after 12.8-14.0 years of wear; 32 (64%) were etched to recover the full number. Once band loss begins, the proportion of surviving birds retaining bands decreases very rapidly (Table 4). Following Fisher's (1976) estimate of a mean survivorship of 94% for adult Laysans, by age 40 when 65 survivors from a cohort of banded chicks would be alive, only four will retain bands. Another way of stating this is that the probability of recovering an aluminum banded 40-year-old Laysan is only 6% (1-61/65=0.06)of what it would be if durable bands had been used (Table 4). Table 2. Wear Characteristics in Aluminum Bands on Albatrossses. Percent Per Year Wear Rates Number in Estimated Mean End Species Sample Mean Maximum Minimum Point Laysan - all 124 2.46% 7.13% 0.78% 54% of weight lost Laysan - 807 Series 12 6.23% 7.13% 5.39% " Laysan Less 807 112 2.42% 4.75% 0.78% " Series Black-fooled 217 1.40% 2.95% 0.22% 56% of weight lost Five legible monel metal Laysan bands were recovered. The mean weight of two carried 12.2 years compared to three carried 26 years was only 7.16% different, for an estimated ARWL of only 0.52%. No unused bands of this type were available to compare with these five bands. If this wear rate is representative for monel bands on Laysans, then loss of monel bands would not begin until year 62 on this species (Table 2). However, we saw monel bands (none in our sample of five) in the field with significant corrosion and pitting around the numbers that may well lead to unpredictable loss rates in the future, similar to Eared Grebes (Jehl 1990) and Caspian Terns (Ludwig 1981 ). Black-footed Albatross bands wore out at much slower rates, averaging 1.4%/year; only 12 bands in the 217 band sample had lost >2%/year of their weight. Four bands (2%) fell outside the required range criterion of the model (1.4 + 0.93 = 0.47 - 2.33%/year). With an endpoint of 56% weight lost, band loss is estimated to begin at 23.1 years. Half the bands would be lost at 37 years, and the last band would fall off at 104 years (Table 3). The oldest Black-foot band recovered for this analysis was 33.8 years, but 77 of the 217 bands (35%) had been carried >25 years. Of these oldest 77 bands, 18 (23%) were still legible. The effect of band loss on life table estimates of Black-looted is much less severe than for Laysan Albatrosses. Serious 'Table 3. Distribution of Band Loss in Albatrosses. Start of Band Effect on Survivorship Species/Band Loss End of Band Loss Rate Metal and Series Endpont/max rate I LaysanlAluminum Except 5414.75 = 11.4 y. 807 Series , Laysan 807 Series - 'Aluminum 54/7.13 = 7.8 y. Laysan Model for 98% of 5414.03 = 13.4 y. Sample Laysan Monel Black-looted all Alum. 56/2.95 = 19 y. Black-footed Model for 5612.33 = 23.1 y. 98% of Sample Page 160 End poJnt/mJn rate Endpoint/mean Increased Rate 54/0.78 = 69 y. 54/2.42 = 22.3 y. 9.95%/year after year 11 54/5.39 = 10 y. 54/6.23 = 8.67 y. 30.8%/year after year 9 5410,81 = 66.7 y. 5412.42 = 22.3 y. 9.45%/year after year 13 54/0.52 = 104 y. 1.92%/year after year 62 56/0.22 = 255 y. 5611.40 = 40 y. 5.6%/year after year 22 56/0.47 = 119 y. 46/140 = 40 y. 5.6%./year after year 22 Table 4. Life Table Calculations Including the Timing of Impacts of Band Loss on Data. Number Alive Still Carrying Underestimation of Age- Number Alive Bands Specific Survivorship Age of Birds in Years LAAL BFAL LAAL BFAL LAAL BFAL FLEDGING 1,000 1,000 1,000 1,000 NONE NONE 7 500 500 500 500 NONE NONE 13 345 345 345 345 NONE NONE LAAL band loss Pegins at 13 years. Tag loss of 9.45%/year + 6% natural mortality. 20 224 224 107 224 117 0 BFAL band loss begins at 23.1 fears. Tag loss of 5.6%!year + 6% natural mortality. 25 164 164 46 155 118 9 30 120 120 20 84 100 36 35 88 88 9 46 79 42 40 65 65 4 25 61 40 45 47 47 2 13 45 34 50 35 35 I 7 34 28 55 26 26 0 4 26 22 60 19 19 0 2 19 17 65 14 14 0 1 14 13 70 10 10 0 I 10 9 104 I 1 0 0 I I biases do not occur until year 23, and the rate of loss is lower at 5.6% (Table 4). At year 40, 38% will still have their bands compared to 6% of Laysans. The last Black-footed from a banded cohort would lose its band at age 72. DISCUSSION The patterns of band wear are virtually identical for these species. Abrasive wear is concentrated on the outside of bands, often causing loss of the number before the band falls off. Many other long- lived birds tend to abrade bands most rapidly on the inside (Ludwig 1981). More wear may occur on the outside because they nest on coral sand. Their eggs are laid directly on the ground in unlined or poorly lined nests. It is common to encounter bands with one or two of the eight digits Oct. - Dec. North American readable and still a significant thickness of metal remaining. All bands weighing 60% or less of their initial weight had to be etched. This abrasion pattern will further bias survivorship and life table data contributed by the public which is less likely to send in unreadable bands to an unknown. address. The fact that etching allows profession- als to recover these numbers is somewhat reassuring. Clearly, aluminum bands are not suitable for collecting survivorship data for Laysan Alba- trosses breeding first at ages 6-9 and having a 6%/ year adult death rate (Fisher 1976). Assuming half of a banded fledged Laysan chick cohort were to die before first nesting at age 7 and a 6% adult death rate, then band loss begins when two-thirds of chicks that survived to become breeding adults were still alive (Table 4); 38% of these adults would be alive when half the bands were lost (Table 4). Band loss is estimated to proceed at 9.45%/year, the apparent death rate after year 14 would be over 15% (real mortality of 6% + 9.45% band loss). Because abrasion obliterates so many band numbers, the effect on life table data is surely even greater than we calculate. Band loss projections for Black-tooted Albatross are more encouraging. The lower wear rates in this species indicate that band loss will not bias the survivorship data until after year 23 when 37% of the adults are still alive (Table 4). Half the bands should be left when only 13% survive, if model parameters are accurate. The constant. band loss error (5.6%/year) added to the real Black-tooted mortality is significantly less than the 9.45% for Laysan Albatross data. Roughly 1% of the Black- tooted Albatross bands could be expected to survive to age 72 when nine of a cohort of 1,000 banded Black-tooted Albatrosses fledged would be alive. However, the probability of recovering a legible band on aged birds of either species is very low. It is somewhat surprising that Laysans wear out their bands at close to twice the rate of Black- footeds. Black-footeds prefer to nest on open beaches in nests scooped out of coral sand. Laysans prefer to nest under vegetation and often line nests with leaf litter. Given the abrasivehess of coral sand, it might be expected that the reverse would be true. However, we have noted that Laysans are more restless when incubating. Laysan nests are often heavily infested with Argus argus ticks, but few Black-tooted nests had ticks. Laysans were often captured with numerous tick bites on their feet and brood patches. Ticks may stimulate Laysans to move legs frequently, thus wearing out bands more rapidly. It was also a surprise that wear rates did not differ among those Black-tooted Albatrosses banded as chicks and those banded as adults. Intuitively, we expected that chicks would wear bands more slowly because they spend much of their juvenile life at sea where abrasion is thought to be minimal. The model we use does not make allowances for senescence or increasing mortality rates in aged birds of these species and is not realistic for very old birds (Table 4). However, during our studies we encountered banded Laysan Albatrosses nesting at ages 43 and 45. A Black-tooted banded as an adult in 1956 was recaught on a nest in 1994, making this bird at least age 43 and possibly significantly older. In three years of field studies, we encountered 137 banded birds of both species nesting that were 30 or more years ot age. Because albatrosses are among the longest lived of all birds, their bands should be suited to their life span and extended breeding period. Other novel marking techniques of producing mark-recapture data are reported. These include web tags (Hatamis and Nice 1980), plastic neck collars (Raveling et al. 1992), radiotelemeters (White 1983), and specially modified leg bands (Blums et al. 1994). However, each of these techniques has been applied to relatively short- lived species with much higher natural death rates. For species with lower survivorship rates, loss of their conventional metal bands will be less important to the integrity of the data collected on populations since very few birds would be alive when band loss begins (Ludwig 1981 ). Some very desirable techniques, such as satellite radiotelem- etry, have been used successfully on Wandering Albatross (Diomedea exulans), for relatively short periods at a very high cost per individual monitored (Weimerskirch et al. 1993). Banding is likely to remain the method used to mark large numbers of birds for population studies. Band retention is critical to interpretation of all mark-recapture data sets. Several studies used metal bands to vedty retention of other types of tags in the short term of weeks to a few years (Nichols and Hines 1993). However, each of these thoughtful approaches was applied to species for a few years or less rather than the five to seven decades that some albatrosses live. Conventional banding will be the method of choice to study the very long-lived albatross populations for the foreseeable future. The challenges are to render the available banding data as accurate as possible and to use durable bands on albatrosses. Abandoning banding in favor of other unproven methods which are very expensive or are unsuited to the long-term performance required of albatross tags is not recommended. CONCLUSIONS Standard butt-end aluminum bands are not durable enough to produce accurate life table data for either albatross species directly. However, aluminum bands may be good enough to produce useful data for Black-looted Albatrosses, provided illegible bands are etched and these data segregated from those produced by recoveries obtained from the general public which is less likely to send in illegible bands. For these two albatrosses which had large numbers banded with conventional aluminum tags during the 1960s and 1970s, it is now critical to reband these adults to retain at least part of the very old banded cohorts for future research. Aluminum bands are better than no bands. But, considering the high cost to reach albatrosses for banding, stainless steel "hard"bands with very low rates of metal loss must be used. Banding authorities should encourage the use of bands made of the appropriate metals with deeply inscribed numbers to lessen number loss from abrasion of the band surface. ACKNOWLEDGMENTS During the 1979 year work on Midway by Ecological Research Services, Inc., Craig Harrison of the USFWS remote Pacific islands refuge program, and U.S. Navy CWO Don Williamson were exceptionally helpful. In 1992-1994, bands were recovered during research on the global distribution of chlorinated contaminants that used these albatrosses as the monitored species by the SERE Group, Ltd. Administration for the project was provided by The World Wildlife Fund with funding from the U.S. Environmental Protection Agency under Grant #8200227010 (A Baseline Assessment of Global Madne Contamination). In 1992-94 important logistic assistance was provided by Ken McDermond. Mike Nishimoto and Ken Neithammer of the USFWS remote Pacific refuges program, and U.S. Naval personnel LCDR Michael Driggers and LCDR James Ray. John Tautin and two unidentified persons reviewed an early version of this report, and Chandler Robbins reviewed the final manuscript. We are grateful for the help from all these persons. LITERATURE CITED Anderson, A. 1980. Band wear in the fulmar. J. Field Ornithol. 51:102-109. Bailey, E.E., G.R. Woolfenden, and W.B. Robertson, Jr. 1987. Abrasion and loss of bands from Dry Tortugas Sooty Terns. J. Field Ornithol. 58:413-424. Blums, P., A. Mednis, and J.D. Nichols. 1994. Retention of web tags and plasticine-filled leg bands applied to day-old ducklings. J. Wild Manage 58:76-81. Brownie, C. 1973. Stochastic models allowing age-dependent survival rates for banding experiments on exploited bird populations. Ph.D. thesis, Cornell Univ., Ithaca, NY. 114 pp. 1976. Models allowing for age- dependent survival rates for band return data. Biometrics Unit, Cornell Univ., Ithaca, NY. 18 pp. DiCostanzo, J. 1980. Population dynamics of a Common Tern colony. J. Field Ornithol. 51:229-243. Fisher, H.I. 1976. Some dynamics of a breeding colony of Laysan Albatrosses. Wilson Bull. 88:121-142. Fordham, R.A. 1967. Durability of bands on Dominican Gulls. Notornis. 14:28-30. Haramis, G.M. and A.D. Nice. 1980. An improved web-tagging technique for waterfowl. J. Wild Manage 44:898-899. Hatch, J.J. and I.C.T. Nisbet. 1983. Band wear and band loss in Common Terns. J. Field Ornithol. 54:1-16. Hickey, J.J. 1952. Survival studies of banded birds. U.S. Dept. of Interior, Special Scientific Report #15. 177 pp. Jehl, J.R. 1990. Rapid band wear in Eared Grebes and other saline lake birds. J. Field Ornithol. 61:108-110. Kadlec, J.A. 1975. Recovery rates and loss of aluminim, titanium and incaloy bands. Bird-Banding 46:230-235. Ludwig, F.E. 1941. An improved method of determining band numbers on illegible bands. Bird-Banding 13:103-104. Ludwig, J.P. 1967. Band Loss--its effect on banding data and apparent survivorship in the Ring-Billed Gull population of the Great Lakes. Bird-Banding 38:309-323. 1981. Band wear and band loss in the Great Lakes Caspian Tern population and a generalized model of band loss. Colonial Waterbirds 4:174-186. Nelson, L.J., D.R. Anderson, and K.P. Burnham. 1980. The effect of band loss on estimates of annual survival. J. Field Omithol. 51:30- 38. Nicholas, J.D. and J.E. Hines. 1993. In Marked individuals in the study of bird populations. Birkhauser Verlag. Basel, Switzerland. Pp. 229-243. Paynter, R.A. 1966. A new attempt to construct lite tables for Kent Island Herring Gulls. Bull. of the Mus. of Comp. Zoo., Harvard Univ., Boston, MA 133(11):489-528. Poulding, H.R. 1954. Loss of rings by marked Herring Gulls. Bird Study 1:37-40. Raveling, D.G., J.D. Nichols, J.E. Hines, D.S. Zezulak, J.G. Silveria, J.C. Johnson, T.W. Aldrich, and J.A. Weldon. 1992. Survival of Cackling Canada Geese, 1992-1988. J. Wild Manage 56:63-73. Schreiber, R.W. and P.J. Mock. 1988. Eastern Brown Pelicans: what does 60 years of banding tell us? J. Field Omithol. 59:171- 182. Spear, L. 1980. Band loss from the Western Gull on southeastern Farallon Island. J. Field Omithol. 51:319-328. Weimerskirch, H., M. Salamolard, F. Sarrazin, and P. Jouventin. 1993. Foraging strategy of Wandering Albatrosses through the breed- ing season: a study using satellite telemetry. Auk 110:325-342. White, G.C. 1983. Numerical estimation of survival rates from band recovery and biotelemtry data. J. Wild. Manage. 47:716-728.