(POPTOOLS is a set of functions that can be added to Microsoft Excel and which contains algo2. 366. 4 æ. Richard B. HARRIS: Population dynamics of pandas ...
动物学报 50 ( 4) :662 - 668 , 2004 A ct a Zoologica S i nica
Insights into population dynamics of giant pandas gained from studies in North America Richard B. HARRIS 3 Wildlife Biology Program , College of Forestry and Conservation , University of Montana , Missoula , M T USA 59812
Abstract Alt hough population dynamics of giant pandas A il uropoda melanoleuca remain poorly known , basic principles can be safely transferred from studies of , and experience wit h , similar species. Wit hout minimizing t he importance of con2 tinued study of demograp hics , genetics , and behavior of pandas , I offer t he following generalizations from our understand2 ing of carnivore population biology in Nort h America. First , elasticity analyses confirm t hat pandas have evolved life2histo2 ries t hat prioritize high survival of adult females. In comparison , reproductive rates are unimportant . Increases in survival of adult females will yield approximately 5 times t he conservation benefit of proportional increases in reproductive output . Second , under circumstances likely to characterize most panda populations , survival rates of males ( even adults) are also relatively unimportant . Thirdly , notwit hstanding its well2deserved reputation as a slow breeder , giant panda populations are mat hematically capable of growing surprisingly quickly , if habitats ( and associated survival rates) allow. Finally , rein2 troductions of endangered species in western Nort h America remind us of t he critical role of maintaining large patches of unfragmented habitat . Wolves Canis l upus , extirpated from t he western U1 S. before t he mid220 t h century , have made a rapid and surprisingly painless recovery because wild areas and prey populations remained abundant . In contrast , black2 footed ferrets M ustela nigripes , lost from t he wild more recently , have encountered great difficulty in becoming reestab2 lished , despite tremendous scientific efforts. It appears t hat ferrets may simply no longer have sufficient wild habitat ( i. e. , prey) to persist . The best science will not be capable of saving pandas if sufficient habitat is not available [ A cta Zoo2 logica S i nica 50 ( 4) : 662 - 668 , 2004 ] . Key words Giant panda , A il uropoda melanoleuca , Elasticity , Population dynamics , Reintroduction , Survival
从北美的研究看大熊猫的种群动态 Richard B. HARRIS 3 Wildlife Biology Program , College of Forestry and Conservation , University of Montana , Missoula , M T USA 59812
摘 要 尽管人们对大熊猫的种群动态了解很少 , 但是我们可以从对相似物种的研究和经历中了解一些基本原 理 。在不减少对后续的有关大熊猫种群特征 、遗传学和行为研究的前提下 , 根据对北美食肉动物种群生物学的 理解 , 我提出了下面一般性的结论 。首先 , 弹性分析确认 , 大熊猫演变出了确保雌性个体高存活率的生活史 。 比较而言 , 繁殖率并不重要 。成年雌性个体的存活率增加 , 比相应的繁殖输出要导致 5 倍的保护效益 。第二 , 在可能表现大熊猫种群特征的假设前提下 , 雄性 ( 甚至成年个体 ) 的存活率相对而言也是不重要的 。第三 , 尽 管都认为大熊猫繁殖很缓慢 , 但是从数学上来说 , 如果生境 ( 以及与其相关的存活率 ) 允许 , 大熊猫的种群能 够比较快地增长 。最后 , 北美西部对濒危物种再引入的经验提醒我们 , 保留大片尚未破碎化的生境非常重要 。 狼在 20 世纪中叶就在美国西部灭绝了 , 但是目前由于有广阔的生存区域和丰富的食物 , 种群恢复很快 。对比而 言 , 最近从原野中消失的黑足鼬 , 在种群重建过程中遇到了很大的困难 , 尽管人们在科学上做出了巨大的努力 。 看来黑足鼬可能只是没有足够的野外栖息地 ( 猎物 ) 以维持生存 。如果没有足够的栖息地 , 再好的科学也不能 拯救大熊猫 [ 动物学报 50 ( 4) : 662 - 668 , 2004 ] 。 关键词 大熊猫 弹性 种群动态 再引入 生存
There is little doubt t hat t he giant panda A il 2 u ropoda mel anoleuca is among t he world’s most en2 dangered species , and t hat preventing it s extinction is
Received Apr. 09 ,2004 ; accepted J une 23 ,2004 33 Corresponding aut hor. E2mail :rharris @montana1com ν 2004 动物学报 A cta Zoologica S i nica
a high priority. The task is multi2disciplinary , and will require energy and cooperation f rom expert s , politicians , and non2governmental groups f rom a wide
4期
Richard B. HARRIS : Population dynamics of pandas
variety of fields. Of course , t he reason for t he pandaπs plight has little to do wit h t he animals t hem2 selves , and much to do wit h humans : conserving pandas will f undamentally be a job t hat requires man2 agement of people and t heir activities. Pandas will be only too happy to do t heir own part . That said , maintaining a healt hy and vigorous panda population can , at least in t heory , be reduced to t he mat hematics of ensuring t hat reproduction bal2 ances2or preferably , more t han balances2mortality. What humans do (or fail to do ) can affect birt h and deat h rates , but it is t hese rates t hat ultimately dic2 tate whet her any given panda population will in2 crease , remain stable , or decline. In recent years , bot h t heoretical and practical work has enhanced our understanding of t hese forces , and provided insight into t he management and conservation of ot her wild populations. In t his paper , I apply some lessons learned f rom ot her cases in Nort h America to t he pop2 ulation dynamics of giant pandas. I have chosen four concept s t hat have recently been investigated or discussed in conservation of rare mammals in Nort h America , all of which appear to have somet hing to teach us about giant pandas. First , I briefly outline t he concept of elasticity , arising f rom analysis of population mat rices. I argue t hat elasticity analysis suggest s t hat effort s focused on assuring high survival of reproductive females will be much more richly rewarded t han will similar effort s placed on im2 proving reproductive rates. Next , I reiterate a con2 cept well2known to t heoretical population biologist s but often difficult to fat hom for t hose of us more com2 fortable envisioning populations as aggregations of in2 dividual animals t han abst racting t hem into life2tables or mat rices ; namely , t hat under a broad range of re2 alistic circumstances , survival rates of males are irrel2 evant to population growt h rate. Third , I attempt to lessen t he pessimism t hat often surrounds discussion of t he f ut ure of pandas by arguing t hat t here is not h2 ing inherent in t heir life2history st rategy of low repro2 ductive rates t hat dooms t heir populations to slow , or no increase. If adult female survival can be kept suffi2 ciently high , even slow2reproducing pandas can in2 crease rapidly enough to help t hemselves. Finally , I leave t he t heoretical world behind to provide an overview of some recent reint roductions of rare species in Nort h America. Where nat ural habitat has been conserved , such reint roductions can succeed. Howev2 er , where habitat reductions have been t he primary reason for t he species being rare and such reductions have not been reversed , even t he most scientifically and caref ully planned reint roductions have met wit h f rust ration. I conclude wit h some lessons for giant panda conservation.
1 Elasticit y : investing conservation
663
resources in panda reproduction is poor st rategy compared wit h investing in panda survival Elasticity analysis is an extension of t he com2 monly2applied Leslie mat rix t hat reveals t he effect t hat a proportional change on a given life2history rate has on t he population ’s rate of increase (λ) ( Benton and Grant , 1999 ; de Kroon et al. , 2000) . It mea2 sures how an infinitesimal change in an assumed fixed rate would affect population growt h ; it becomes less reliable if used to infer t he relative effect of large changes on t hose rates. Formally , it is defined as : a ij 9λ 9logλ eij = λ 9a ij = 9log aij where eij = elasticity of t he element corresponding wit h t he i t h row and jt h column , a ij = t he element corresponding wit h t he i t h row and j t h column , and λ = t he population ’s finite rate of increase. Thus , under certain assumptions ( Mills et al. , 1999 ) , it measures t he “importance ”of one life2history stage relative to anot her. Application of elasticity analyses to brown bear U rsus arctos populations in Nort h America has generally revealed t hat adult ( and in some cases , sub2adult ) survival rates st rongly influ2 ence λ, whereas reproductive rates ( or t heir con2 stit uent element s , probability of breeding and litter size ) are relatively unimportant ( Harris et al. , 2004) . In general , it has been found t hat for long2lived species wit h low2reproductive rates , elasticity of sur2 vival rates is much greater t han for reproductive rates ( Taylor et al. , 1987 ; Eberhardt et al. , 1994 ; Hovey and McLellan , 1996 ; Boyce et al. , 2001 ) . Popula2 tion information for giant pandas is still sparse and not necessarily reliable ( Harris and Metzgar , 1993 ) , but attempt s have been made to use Leslie mat rix and simulation models to project panda population growt h rates. To conduct a preliminary elasticity analysis for giant pandas , I used 2 alternative life2table formula2 tions: t hose f rom st udies in Wolong Nat ure Reserve ( Wei et al. , 1989 , 1997 ; Li et al. , 2003) and t he Foping Nat ure Reserve ( Zhou and Pan , 1997 ; Pan et al. , 2001 , 2004) . Alt hough bot h survival/ fecundity schedules were based on small sample sizes and may have been influenced by biases in data collection or in2 terpretation ( Harris and Metzgar , 1993 ) , I assume here t hat t hey provide a reasonable approximation to panda life histories. I used t he life table and mat rix projection mod2 ules of POP TOOL S ( www. cse. csiro . au/ cdg/ pop2 tools/ index. ht m ) to calculate deterministic ( and t herefore approximate only ) estimates of λ and elas2 ticities. ( POP TOOL S is a set of f unctions t hat can be added to Microsoft Excel and which contains algo2
动 物 学 报
664
50 卷
rit hms usef ul in population analysis) . To generate e2 lasticities for t hese 2 schedules of giant panda repro2 duction and survival , I first used POP TOOL S to t ransform reproductive rates ( summarized by m x ) to t he Fx values required for a Leslie mat rix ( Taylor and Carley , 1988 ; Harris and Metzgar , 1993) Elasticity analyses confirmed t he importance of female survival and relative unimportance of reproduc2 tive rates in cont ributing to λ for giant pandas. For t he Wolong2based population mat rix ( Wei et al. , 1989 , 1997 ; Li et al. , 2003) , elasticities for survival rates ( s x ) of subadult s ( ages cub27) and adult s ( ages 8 + ) were 0155 and 0137 , respectively , whereas e2 lasticity of t he Fx ( recruit ment ) was only 0108. For t he population mat rix built for t he Foping population ( Zhou and Pan , 1997 ) , elasticity for survival of subadult ( cub25) pandas was 0141 and for adult s was 0149 , whereas for recruit ment ( Fx ) it was only 0110. I also used life2table analysis to estimate t he ef2 fect onλ of increasing ( and decreasing) isolated vital
rates by 10 % ( Table 1) . These result s again demon2 st rated t hat proportional changes in panda reproduc2 tion ( in t his case , parameterized by m x rat her t han Fx ) have minor influences compared to t he same pro2 portional change in survival. For example , if it were possible to increase reproductive output of a popula2 tion similar to t hat modeled for Wolong by 10 % over it s documented value ,λ would increase by 017 %. In cont rast , were it possible to increase survival of adult s by 10 % over it s documented value ,λ would increase by 415 %. For t he Foping2based model ( Zhou and Pan , 1997 ) , t he deterministic λ associated wit h a stable2age st ruct ure is 110396. Wit h t his growt h rate , t he population would double in 19 years. Were t he reproductive rate ( m x ) of 0125/ female cubs/ fe2 male/ year reduced by 10 % , such a population would still increase , albeit at a slower rate , and double in 25 years. However , were t he high survival of 01977 re2 duced by 10 % , population t rajectory would reverse , and in 19 years rat her t han be twice it s initial size would instead be only half it s initial size ( Table 1) .
Table 1 Vital rates summarized from 2 published life2history schedules of giant pandas ( Zhou and Pan , 1997 ;Li et al. , 2003) , and sensitivities of the rates of increase (λ) to 10 % increases and decreases in each rate separately ( indicated in bold type) Reproductive
Subadult female
Adult female
rate
survival rate
survival rate
01330
01980
01867
110046
Reproduction
01363
01980
01867
110123
017 % +
58 +
Adult survival
01330
01980
01954
110502
415 % +
15 +
Reproduction
01297
01980
01867
019962
011 % -
182 -
Subadult survival
01330
01826
01867
019571
417 % -
17 -
Adult survival
01330
01980
01780
019620
412 % -
19 -
01250
01707
01977
110396
Reproduction
01275
01707
01977
110495
110 % +
15 +
Subadult survival
01250
01778
01977
110597
210 % +
13 +
Reproduction
01225
01707
01977
110290
110 % -
25 +
Subadult survival
01250
01636
01977
110185
210 % -
39 +
Adult survival
01250
01707
01879
019658
711 % -
19 -
λ
Percentage
Doubling/
change inλ
halving time
—
153 +
Li et al. , 2003 Published rate Increased by 10 %
Decreased by 10 %
Zhou and Pan , 1997 Published rate
—
19 +
Increased by 10 %
Decreased by 10 %
Change is calculated as ( (λoriginal - λmodified) /λoriginal) , expressed as a percentage. Also shown are t he number of years required for a doubling (whereλ > 1) or halving (whereλ < 1) of t he population (assuming no change in vital rates) .
In general t hen , reduction of mortality rates will reap conservation dividends roughly 5 times t he size of
proportional increase in panda reproduction. It is t rue t hat altering one life2history rate may be more diffi2
4期
Richard B. HARRIS : Population dynamics of pandas
cult t han altering anot her. J ust because sub2adult or adult survival rates have higher elasticities t han repro2 duction does not necessarily mean t hat improving sur2 vival will be easy or even possible ( e. g. Li et al. , 2000) . However , reproductive rates of bears are usu2 ally relatively invariant wit hin any given location , i. e. , are not readily influenced by management , so t his is probably not a sit uation in which relatively low e2 lasticity misrepresent s t he potential for t his life2histo2 ry stage to cont ribute to population growt h. I suspect t hat survival rates , particularly of adult females , would be quite high in t he absence of humans. Thus , any reduction of artificial (i. e. , human2caused) mor2 tality or removal will reap large dividends.
2 Male survival may be unimportant for population growt h It is well known among managers of hunted pop2 ulations2usually ungulates2t hat populations are much more responsive to mortality of females t han males. Less often appreciated by practitioners ( even if well2 known to mat hematically2inclined t heorist s) is t hat , under a broad range of reasonable assumptions about how populations operate in t he real world , t he long2 term t rajectory of populations wit h a polygynous mat2 ing system is completely unaffected by t he survival rate of males. That is , unless certain conditions ap2 ply , t he rate of growt h (λ) is entirely ( not merely primarily) dictated by t he survival and reproduction of t he female segment of t he population. As long as males are numerous enough to impregnate all t he fe2 males available to breed in any year , t heir survival rate does not affect population t rajectory. It is t he survival rate of females t hat will determine how fast t he population will grow. This is easily shown mat hematically but can still be difficult to embrace int uitively. Even if we have no problem understanding t hat females are t he only ani2 mals t hat become pregnant , must not t he mortality rate for males have some influence on λ, even if small ? To see why it might not , I offer t he following heuristic example ( Fig11) . Imagine a highly2productive population of imagi2 nary bears ( U rsus spp . ) in which females produce , on average , 014 female cubs/ year beginning in t heir 3rd year. After survival in t he initial 2 years at 70 %/ year , constant annual survival is 0192 for females and 0185 for males until age 16 , at which point survival declines. Such a population will increase at about 513 % per year , and be about 55 % females ( Fig11A) . Now imagine t hat annual survival of 2 + year2old males suddenly plummet s f rom 0185 to 0150 (i. e. , each year half t he subadult and adult males die) . It is nat ural to imagine t hat such high mortality for a long2lived animal would have a depressing effect
665
on t he rate of increase. And indeed it does2for a few years only , as t he male : female ratio equilibrates to t he new reality ( Fig11B) . But as soon as t his equili2 bration has taken place , t he overall population growt h rate revert s to t he original 513 % per year ( Fig11C) . True , it does so f rom a lower overall level t han would have been t he case had annual male survival remained at 0185 ( t he population now consist s of about 65 % females) , but wit hin a few years , t he population is once again t hriving ( despite half t he adult males dy2 ing each year ) . As long as t here are sufficient males to impregnate all eligible adult females , cubs keep getting produced ( 513 % more each year t han t he previous year ) , and half of t hese will be females ( whose annual survival , in t his hypot hetical example , did not change) . So t he female portion of t he popula2 tion is entirely unaffected by t he calamitous mortality affecting males , and t he male portion of t he popula2 tion , after having it s age2st ruct ure t runcated by t he increased mortality , follows along at t he same 513 % yearly increase ( Fig12) . The irrelevance of males to λ breaks down under 3 plausible scenarios t hat could apply to panda popula2 tion dynamics. First , if t he number of males becomes so low t hat females t hat are available for breeding are simply not bred , t hen clearly t he number of males present matters. Such an effect , producing a positive correlation between population size and growt h rate , could occur if survival of males is low enough t hat adult s simply cannot find each ot her during est rous. Secondly , t he simple model breaks down if reproduc2 tive rates or female survival is an inverse f unction of male density , i. e. , if density2dependence ( popula2 tion growt h a negative f unction of population size ) is felt across gender lines. In t his case , lower survival of males might act ually lead to higher λ. Finally , mor2 tality of males could affectλof t he entire population if pandas practice infanticide , and if t he probability of infanticide is , in t urn , a f unction of social st ruct ure among males ( Swenson , 2003) . In very small populations , or where males have become ext remely rare , insufficient number of males is a t rue concern. Where pandas have been st udied , however , t here is little evidence t hat eligible females have lacked mates during t he appropriate time of year ( Pan et al. , 2004) . For most populations of bears in which anyt hing about density2dependence can be de2 termined , density is likely to affect growt h rate only where populations are relatively dense , close to t heir ecological carrying capacity. There is now some evi2 dence t hat where bears are totally protected f rom hu2 man exploitation , survival of cubs may decline ( Miller et al. , 2003) . Whet her t his is a current con2 cern for pandas is unclear. Finally , sexually2selected infanticde ( Swenson , 2003) is a t heoretical possibili2
666
动 物 学 报
50 卷
Furt her , males make genetic , not merely demo2 grap hic , cont ributions to populations. A highly im2 balanced sex ratio , in which only a few males procure a highly disproportionate number of available mating opport unities , can dramatically decrease t he effective population size ( N e ) , and t hus increase t he likelihood of inbreeding or genetic drift ( Harris and Allendorf , 1989) . As well , mortality of males may indicate un2 detected mortality of females , which2as I argued above2is t he driving force behind panda population t rajectory. Thus , of course males are important . But for populations facing an immediate demo2 grap hic t hreat , i. e. , t hat t here simply aren ’ t e2 nough animals to sustain t hemselves , t he immediate task is usually to obtain population growt h. That is , we desire to achieve λ > 1 , t he quicker t he better. And for t he immediate question of increasing λ, as long as males are numerous enough to impregnate all females available to breed in any year , t heir survival rate is unimportant . It is t he survival rate of females t hat will determine how fast t he population will grow.
3 Does low reproductive rate doom panda populations to slow increase ?
Fig11 Smoothed histograms , showing the age2specif ic abundance of the male segment of a hypothetical bear pop2 ulation Survival rates allow bot h t he female and male segment to increase at approximately 5 % yearly. A. In year 1 , male survival is 017 for ages 021 , and 0185 t hereafter. The population consists of approxi2 mately 45 % males , 55 % females , and t here are 154 males in t his population. B. In year 4 , survival of males age 2 + suddenly changes to 015 , and t he number of males declines for a few years. C. Wit h survival of males age 2 + still at 015 , by year 9 t he age distribution of males has reached a new steady state , and t he popu2 lation of males is increasing at about 5 % yearly , just as t he female segment has done during all years.
ty in pandas , and should be considered , but is not known f rom all bear populations , and even if it ex2 ist s , may not be altered should t he survival rate of males change.
There is no question t hat pandas are relatively slow breeders , and t hat rates of increase of panda populations are t herefore const rained. But a con2 st rained or slow rate of increase does not necessarily mean t hat t here is a“genetic problem”( Cao , 2004) , t hat no increase at all is possible , or t hat , wit h t ruly effective measures in place , we cannot expect to see substantial increases. For example , when I combined survival rates used by Zhou and Pan ( 1997 ) wit h a mean reproductive rate of 0125 female cub/ adult fe2 male/ year for females aged 5 - 19 ( and half t hat for females aged 4) , t he expected growt h rate was almost 4 %/ yr. Obviously t his rate would only be achieved under conditions in which t he available habitat was capable of supporting more t han t he existing popula2 tion. But under such a growt h rate , a population of 100 would increase to about 142 by t he 10t h year , and double in about 18 years.
4 If a panda population is lost , can it be successf ully reint roduced using cap2 tive animals ? One rationale for emp hasizing captive breeding of pandas is t hat t hey may event ually be released back into t he wild. Experience wit h large mammal reint ro2 ductions in China is limited (Jiang , 2004 ) . Howev2 er , two recent carnivore reint roductions in t he U1 S. offer some usef ul lessons. Bot h t he wolf Canis l u pus and t he black2footed ferret M ustel a ni gri pes are con2
4期
Richard B. HARRIS : Population dynamics of pandas
Fig12 The number of males , females , and total animals in the same hypothetical population as in Fig11 , graphed on year Initially , bot h females and males increase at about 5 % yearly. J ust before year 4 , survival of males age 2 + changes abruptly from 0185 to 015 , causing a temporary reduction in male abundance. Howev2 er , once males equilibrate at t he new survival schedule , t hey join fe2 males in increasing at t he original rate ( ～5 % yearly) , albeit from a lower level. Bottom ( dotted ) line represents males ; middle (dashed) line represents females ; top ( solid) line represents total population.
sidered endangered in t he U . S. , and receive federal protection. Bot h species were almost completely ex2 tirpated in t he western states by t he mid220 t h cent u2 ry. However , t he prey base for wolves ( large artio2 dactyls , such as O docoileus spp . , Cerv us el aphus , and A lces alces ) had been well conserved , not only in national parks but also in publicly owned areas used for ot her resources and private lands. In cont rast , ro2 dent s of t he genus Cy nom ys —prey species upon which black2footed ferret s are wholly dependent —had experienced a dramatic decline , to t he point where one species , C1 l u dovician us , had even been catego2 rized as wort hy of endangered species protection it2 self . In t he mid 1990πs , U 1 S. federal agencies spon2 sored reint roductions of bot h wolves ( in Yellowstone National Park , and in wilderness areas of cent ral Ida2 ho ) , and ferret s (in t he best protected Cy nom ys spp . populations in Arizona , Montana , Sout h Dakota , U2 tah , and Wyoming) . Wolf reint roductions were con2 ducted quickly and wit hout much scientific prepara2 tion ; wolf packs f rom Canada were capt ured by heli2 copter , and simply let go (in Idaho ) or kept in on2site pens for a few mont hs prior to release ( in Yellow2 stone) . In cont rast , ferret reint roductions were pre2 ceded by years of caref ul captive breeding , and all reint roduced animals were caref ully screened. By 2002 , it was clear t hat t hese 2 reint roduc2 tions had fared dramatically differently. Alt hough wolves were feared and disliked by many rural resi2 dent s , t heir populations had increased dramatically ( U . S. Fish and Wildlife Service , 2003 ) . In Yel2 lowstone , 31 wolves int roduced in 1995 - 1996 had by t he year 2002 increased to 217. In Idaho , 35
667
wolves int roduced during t he same 2 years had in2 creased to 263. Including wolves in Montana ( some of which originated f rom resident s) , t he total popula2 tion in 2002 was approximately 663 , and government officials had begun t he process of changing t heir sta2 t us f rom endangered. In cont rast , t he ferret reint roduction programs experienced continual f rust ration. Despite exhaustive2 ly2researched captive breeding programs ( Vargas et al. , 2000 ) , new populations formed f rom captive2 bred individuals did not sustain t hemselves. At only 1 of 7 reint roduction sites in t he U1 S. had t he ferret population increased. At t he 6 ot her sites , t he release of a total of 1 272 captive2reared ferret s ( in addition to ferret kit s born in t he wild to t hese animals) had resulted in a total wild population of only 137 in t he year 2002 ( R. Matchett , U 1 S. Fish and Wildlife Service , Lewistown , Montana , U SA , personal com2 munication , 2003 ) . The one successf ul population had among t he largest Cy nom ys populations. There are numerous possible reasons for t he dif2 ferent outcomes of t hese 2 reint roduction programs. In particular , success in ferret reint roduction was complicated by t he advent of plague Yersi nia pestis , which had not been native to Nort h America but by t he late 20t h cent ury had become a major mortality factor for Cy nom ys spp . on which ferret s depend. Even accounting for t he effect s of plague , however , a st rong correlation was found between survival rates of ferret s and t he extent of t heir habitat ( i. e. , size of colonies of Cy nom ys spp . ) . In ot her words , where habitat for ferret s ( Cy nom ys spp . populations) re2 mained in large , relatively contiguous patches , t he reint roduced ferret populations could succeed. Where wild habitat patches were small , t he best laboratory science was incapable of overcoming t hose nat ural limitations. By cont rast , wolves increased - despite local fears , occasional poaching , and t he removal each year of a number of livestock2killing wolves by federal agent s - because t heir habitat ( native ungulates) re2 mained healt hy. Wolves released were also wild ani2 mals , and had spent negligible time in captivity. In cont rast , released ferret s ( even t hose subjected to a pre2release“acclimatization”period) were all captive2 born.
5 Implications for pandas I conclude f rom t hese examples t hat emp hasis on panda reproduction is misplaced. In general , it re2 quires about 5 additional cubs to compensate for t he loss of a single adult female. Put anot her way , effort s put toward allowing a given female to live in t he wild until it s nat ural deat h would produce 5 times more positive effect on a population t han would producing a single additional cub. There has been too much em2
668
动 物 学 报
p hasis placed on t he panda’ s“low”reproductive rate ( Cao , 2004) . It certainly is low , but pandas have e2 volved a life2history st rategy in which limited repro2 ductive capability is balanced by relatively long life2 span. Research on captive pandas is certainly valuable and has cont ributed important insight s for conserva2 tion of wild populations. However , if reint roduction or augmentation of wild populations is envisioned , Nort h American experience suggest s t hat even t he most rigorous captive rearing program may encounter f rust ration if wild habitat s have become too limited. Focus on captive rearing is understandable : result s may come relatively quickly , and one has t he satisfac2 tion of working wit h t he animals t hemselves. But bot h t heory and practice suggest t hat t he largest con2 t ributions toward ensuring preservation of wild pan2 das will come f rom limiting human2caused mortality or removal of females , and ensuring t hat panda habi2 tat s remain as large and healt hy as possible ( Hu and Wei 2004 ) . This will require creative and energetic effort s , because merely creating legal protection is ev2 idently not sufficient ( Liu et al. , 2001 , 2004 ) ; t he social and economic conditions of local people also need to be addressed. Acknowledgements I t hank R. Swaisgood for invit2 ing t his paper. R Matchett provided data on black2 footed ferret s. The manuscript was improved by sug2 gestions f rom D. Garshelis and two anonymous re2 viewers. References Benton T G , Grant A , 1999. Elasticity analysis as an important tool in evolutionary and population ecology. Trends in Ecology and Evolu2 tion 14 : 467 - 471. Boyce MS , Blanchard BM , Knight RR , Servheen C , 2001. Population viability for grizzly bears : a critical review. International Associa2 tion for Bear Research and Management . Monograph Series Num2 ber 4. Cao DS , 2004. “Pandas gift people wit h more birt hs ”. China Daily , J une 11 , 2004 , 1. Eberhardt LL , Blanchard BM , Knight RR , 1994. Population trend of t he Yellowstone grizzly bear as estimated from reproductive and survival rates. Canadian Journal of Zoology 72 : 360 - 363. Harris RB , Allendorf FW , 1989. Genetically effective population size of large mammals : assessment of estimators. Conservation Biology 3 : 181 - 191. Harris RB , Metzgar L H , 1993. On some common errors in analysis of age - structured populations. Acta Theriologica Sinica 13 ( 3 ) : 217 - 222 ( In Chinese) . Harris RB , Schwartz CC , Haroldson MA , White GC , 2004. Trajec2 tory of t he greater Yellowstone ecosystem grizzly bear population under alternative survival rates. Wildlife Monographs ( In press) . Hovey FW , McLellan BN , 1996. Estimating population growt h of griz2 zly bears from t he Flat head River drainage using computer simula2 tions of reproductive and survival rates. Canadian Journal of Zoolo2 gy 74 : 1 409 - 1 416.
50 卷
Jiang , ZG , 2004. Wild horse Equus przew alskii . Chinese Journal of Zoology 39 : 100 - 101 ( In Chinese) . De Kroon H , Groenendael J V , Ehrlen J , 2000. Elasticities : a review of met hods and model limitations. Ecology 81 : 607 - 618. Li YM , Gao ZW , Li XH , Wang S , Niemelg J , 2000. Illegal wildlife trade in t he Himalayan region of China. Biodiversity and Conserva2 tion 9 : 910 - 918. Li YM , Guo ZW , Yang QS , Wang YS , NiemelgJ , 2003. The impli2 cations of poaching for giant panda conservation. Biological Conser2 vation 111 : 125 - 136. Liu J G , Linderman M , Ouyang ZY , An L , Yang J , Zhang HM , 2001. Ecological degradation in protected areas : t he case of Wo2 long Nature Reserve for giant pandas. Science 292 : 98 - 101. Liu J G , Ouyang ZY , Linderman M , An L , Bearer S , He GM , 2004. A new paradigm for panda research and conservation : integrating e2 cology wit h human demography , behavior , and socioeconomics. In : Lindburg D G , Baragona K ed. Giant Pandas : Biology and Conservation. Berkeley , California : University of California Press ( In press) . Miller SD , Sellers RA , Keay J A , 2003. Effects of hunting on brown bear cub survival and litter size in Alaska. Ursus 14 : 130 - 152. Mills L S , Doak DF , Wisdom MJ , 1999. Reliability of conservation ac2 tions based on elasticity analysis of matrix models. Conservation Bi2 ology 13 : 815 - 829. Pan WS , L üZ , Zhu XJ , Wang DJ , Wang H , Long Y , Jiang FD , Zhou XJ , 2001. A chance for lasting survival. Beijing : Beijing Univer2 sity Press ( In Chinese) . Pan WS , Long Y , Wang DJ , Wang H , L üZ , Zhu XJ , 2004. The fu2 ture survival of giant pandas A il uropoda melanoleuca in t he Qin2 ling Mountains of China In : Lindburg D G , Baragona K ed. Giant Pandas : Biology and Conservation. Berkeley , California : Universi2 ty of California Press ( In press) . Swenson J E , 2003. Implications of sexually selected infanticide for t he hunting of large carnivores. In : Festa2Bianchet M , Apollonio M ed. Animal Behavior and Wildlife Conservation. Island Press , Washington , DC , 171 - 190. Taylor M , Carley J S , 1988. Life table analysis of age structured popu2 lations in seasonal environments. Journal of Wildlife Management 52 : 366 - 373. Taylor M , DeMaster DP , Bunnell FL , Schweinsburg RE , 1987. Mod2 eling t he sustainable harvest of female polar bears. Journal of Wildlife Management 51 : 811 - 820. US Fish and Wildlife Service , Nez Perce Tribe , National Park Service , USDA Forest Service , 2003. Rocky Mountain Wolf recovery 2002 Annual Report . In Meier T ed. USFWS Ecological Services. Hele2 na , M T : Vargas A , Gober P , Lockhard M , Marinari P , 2000. Black2footed fer2 rets: recovering an endangered species in an endangered habitat . In : Griffit hs HI ed. Mustelids in a Modern World. Management and Conservation Aspects of Small Carnivore : Human Interactions. London : Backhuys Publishers , 97 - 105. Wei FW , Hu J C , Xu GZ , Jiang MD , Deng Q T , Zhong ZM , 1989. A study on t he life table of wild giant pandas. Acta Theriologica Sini2 ca 9 (2) : 81 - 86 ( In Chinese) . Wei FW , Feng ZJ , Hu J C , 1997. Population viability analysis comput 2 er model of giant panda population in Wuyipeng , Wolong Natural Reserve , China. International Conference Bear Research and Man2 agement 9 (2) : 19 - 23. Zhou ZH , Pan WS , 1997. Analysis of t he viability of a giant panda population. Journal of Applied Ecology 34 : 363 - 374. Harris RB , Metzgar L H , 1993. 在年龄结构化动物种群的分析过程 中常见的错误 . 兽类学报 13 (3) : 217 - 222. 潘文石 , 吕 植 , 朱小健 , 王大军 , 王 昊 , 龙 玉 , 付达莉 , 周 欣 , 2001. 继续生存的机会 . 北京 : 北京大学出版社 . 魏辅文 , 胡锦矗 , 许光瓒 , 江明道 , 邓启涛 , 钟肇敏 , 1989. 野生 大熊猫生命表初编 . 兽类学报 9 (2) : 81 - 86.
