S. H33A Atti della Società Italiana di Scienze Naturali e del Museo Civico di Storia Naturale di Milano Voi. 133 (1992), n. 12, pag. 141-151 Milano, novembre 1993 THE NATUhA H1STORY (ViUSEUivi Giuseppe Bogliani (*) 2 7 JUN 1994 EXCHANGED GENERAL LIBRARY Population structure and loss of heterozygosity in relation to management in Sardinian semi-feral ponies Abstract — Semi-feral horses which roam on thè Giara di Gésturi under partial manage¬ ment were censused (513 individuals) and their population structure was compared with thè daughter population living on Capo Caccia (27 individuals) where horses are left unmanaged. The sex ratio was dose to 1 at Capo Caccia, but was female biased on thè Giara (1 male per 8 females), where most yearling males are removed. The unbalanced sex ratio on thè Giara greatly reduces thè Effective Population Size and will cause a rapid loss of heterozygosity. In order to maintain a large amount of genetic variation thè sex ratio on thè Giara would have to be closer to 1. Riassunto - Struttura della popolazione e perdita di variabilità genetica dovuta alla gestione dei cavallini sardi viventi allo stato semi-selvatico. Sono stati censiti i cavallini della Giara di Gésturi (513 individui) sottoposti a manipola¬ zioni da parte dell’uomo, e di Capo Caccia (27 individui), non manipolati in modo sistema¬ tico, e ne sono state comparate le strutture di popolazione. Il rapporto-sessi è prossimo a uno a Capo Caccia; è fortemente sbilanciato sulla Giara (1 maschio ogni 8 femmine), nella quale la maggior parte dei giovani maschi di un anno viene rimossa regolarmente. Il rapporto sessi sbi¬ lanciato sulla Giara riduce di molto le dimensioni della popolazione efficace (Effective Popu¬ lation Size) da utilizzarsi per le stime di perdita di eterozigosi in generazioni successive, e cau¬ serà una rapida perdita di variabilità genetica nel corso di poche generazioni. Per un efficace mantenimento del massimo di eterozigosi e della vitalità della popolazione nella Giara occor¬ rerebbe portare il rapporto-sessi più vicino aH’equilibrio. Key words: Equus caballus, Population structure, Heterozygosity, Management, Sardinia. (*) Dipartimento di Biologia Animale, Università di Pavia, Piazza Botta 9, 27100 Pavia, Italy. © Soc. Ital. Sci. Nat. Museo Civ. Storia Nat. Registrato al Tribunale corso Venezia 55, 20121 Milano di Milano al n. 6574 ISSN 0037-8844 Dir. resp. Giovanni Pinna Stampa Fusi-Pavia 142 G. BOGLIANI Introduction The wild dose relatives of some domesticated animals have become extinct in recent times, and thus thè remaining genetic variation of thè spe- cies can only be found in captive populations. However modera breeding techniques are oriented towards thè selection of breeds that are well adapted to captivity and to economie goals. This is usually achieved by enhancing homozygosity for some characteristics, and in this way much genetic variation is being lost through artificial selection. Such loss of genetic variability in domesticated animals will reduce thè possibility of selecting breeds for characteristics which might be desirable to future generations. The loss of genetic variation might further be conside- red a negative feature because, overall, thè population involved faces redu- ced viability in thè short term and loss of potential adaptability to changing environmental conditions in thè long term (Soulé and Wilcox 1980, Allen- dorf and Leary 1986). Allendorf and Leary (1986) and Scribner (1991) have shown that thè loss of heterozygosity is of great concern for population conservation and that thè health of many populations is deteriorating for this reason, whereas Lande (1988) focused thè attention on demographic stochasticity, which could be of more concern than genetic problems. Two main methods exist for thè maintenance of genetic variation in domestic livestock. The fìrst is thè maintenance of native breeds that con- tain a well diversified set of characteristics and whose reproduction can be controlled (Maijala et al. 1984; Henson 1992); thè second is thè conservation of feral populations (domesticated animals that have returned to thè wild state). According to Van Vuren and Hedrick (1989), feral populations may have two main categories of attributes which make them desirable for con¬ servation: «First, feral populations may have relict characteristics or genetic variants that either are absent in modera breeds (...) Second, feral popula¬ tions may have novel or rare characteristics or adaptations. These traits may include adaptations to extreme environmental conditions such as tempera¬ ture stress, drought, high parasite load, or other characters of potential com¬ mercial or scientifìc value». The horse Equus caballus has become extinct as a wild species, but feral horses exist in many places around thè world, where they show great adap¬ tability to different environmental conditions, ranging from thè arid plains of Australia to thè cold mountain ranges of western North America (Berger 1986). In Europe there are no feral horses, but in a few areas free roaming populations exist, managed by man to different degrees, eg. thè New Forest (Tyler 1972) and Exmoor (Gates 1979) in England, thè Camargue, France (Wells and von Goldschmidt-Rothschild 1979) and thè Bialowieza forest and thè Popielno reserve, Poland (Kownacki 1984). Feral horses existed in Sardinia, Italy, at least as late as thè end of thè eighteenth century, when Cetti (1774) described thè populations of thè Island of Sant’Antioco and of thè Nurra, thè region at thè North-western corner of Sardinia. These feral populations have not survived, and there are no completely feral horses. However many free ranging ponies live under moderate human interference in an area of centrai Sardinia, thè Giara di Gésturi. The presence of ponies on thè Giara can only be documented POPULATION STRUCTURE AND LOSS OF HETEROZYGOSITY 143 since thè nineteenth century, but it is possible that a free ranging population existed there in earlier times (Cancedda, pers. comm.). These horses are pri- vately owned by thè inhabitants of towns surrounding thè Plateau. Giara ponies are well adapted to a diffìcult environment, characterized by drought, high temperature and food shortage during summer. The main- tenance of a viable population of these ponies is desirable because they have been selected by an extreme Mediterranean climate, and thus can be a source of rustie characteristics for breeding purposes, for instance for thè establishment of new controlled herds in formerly cultivated and recently abandoned agro-ecosystems in thè Mediterranea area. The Giara ponies are at present of great economie, cultural and aesthetic value, and for thè Sardi- nian people, as well as for other Italians, they are also an important part of our heritage (Ruiu 1988). Despite thè fame of thè Giara ponies in Italy, no data are available on their number, population structure or thè management problems they are faced with. In about 1976, a group of Giara ponies (2 males and 3 females) was translocated to a 12 km2 protected area, managed by thè Forest Service in thè Capo Caccia peninsula, in North-western Sardinia, and gave origin to a naturally structured population. In this paper an estimate of thè population size in thè Giara is given, as well as a description of thè population structure and some aspeets of thè so¬ cial behaviour of Giara ponies as compared with thè ponies of Capo Caccia, focusing on thè conservation problems of thè Giara population under thè current management regime and particularly on loss of heterozygosity as an indicator of genetic viability (Allendorf and Leary 1986). Study area The Giara di Gèsturi is a 45 km2 plateau of volcanic origin with a sharp edge, dominating thè intensively cultivated Marmilla region. Altitude varies between 493 and 609 m a.s.l. The soil is thin but very fertile and allows thè growth of good pasture; some areas are not however covered by soil and thè basaltic rock is exposed. About 46% of thè plateau is covered by oak woods, where thè prevailing tree species in thè cork oak Quercus suber. The woods are characterized by a low canopy cover value: only 18% of thè total surface area being occupied by wood whose canopy cover is at least 60%. 32% of thè area is covered by various types of Mediterranean macchia, 10% by grassland, 9% by garrigue and 3% by temporary ponds (De Martis and Mossa 1988). The area has a Mediterranean, mesomediterranean subregion, climate (Tomaselli et al. 1973) characterized by 759 mm of rain, mainly concentrated in autumn and winter and mean annual temperature of 15°. Generally speaking, this area experience a period of drought in summer and early au¬ tumn of about 4 months, whose effeets are enhanced by frequent winds. This aflfects thè growth of new pasture but does not prevent horses from drinking since water remains in thè deepest ponds; furthermore there are a few permanent springs in thè Giara. There are some hundred free ranging cattle and goats which compete for food with horses. The plateau is uninhabited by man, apart from a few shepherd’s huts. A recently built road enables vehicles to reach and cross 144 G. BOGLIANI thè plateau, and in thè summer months an increasing number of tourists visit thè area. Capo Caccia is a calcareous peninsula, mainly covered by garrigue and Mediterranean macchia, with some recent pine plantations. Methods Census The census at Giara was performed by counting of all thè horses in 3 large plots of known surface area (14.3 km2 overall) and extrapolating thè re- sult to thè total area. In all, 31.8% of thè plateau was surveyed (Tab. 1). This is well over 10%, which is assumed to be thè minimum sample which allows thè extrapolation of census results to total area (Fattorini 1992). Preliminary observations indicated that herds were quite regularly spaced through thè Giara, so that estimation in sample areas was feasible; indeed thè coefficient of variation of density in thè areas sampled was relatively low (14%). Howe- ver, since thè census plots were of unequal size and not chosen at random, it was impossible to calculate thè confidence interval of thè estimate (Krebs 1989). Recognizable individuals were repeatedly observed around thè same spots at each visit, indicating that they did not move from one sample area to thè other. In areas A and B (fìg. 1), two observers, in mutuai visual con¬ tact, thoroughly surveyed thè sampled area by foot and counted thè ponies they encountered. Doublé counts of thè same individuals or bands were avoided by accurately describing thè physical features of each horse. In C, horse bands were located at sunrise from thè top of Mt. Zeppara Manna. Within three hours, each band was then approached by two observers, and an accurate count of individuals was effected. Observations on thè social structure were carried out during 5 visits to thè Giara, from 28th September 1989 to 21st June 1990, for a total of 12 days of field work. The area was walked by foot, and thè sex, age and a detailed description of physical features was registered for every individuai observed, as well as thè type of social unit. Much care was taken to ascertain thè mo- ther of foals and yearlings, as could be observed from mutuai interactions (Tyler 1972, Wells and von Goldschmidt-Rothschild 1979). In all 520 indivi¬ duals were observed; some individuals may have been encountered more than once during different visits, but this would not have caused a bias in thè estimated population structure, as thè resighting probability was thè same for all age and sex classes. Tab. 1 — Census results for thè sample areas on thè Giara di Gèsturi. plot area (km2) no. ponies density (inds. / km2) A 3 40 13.3 B 2.8 29 10.4 C 8.5 89 10.5 Total 14.3 158 11.4 POPULATION STRUCTURE AND LOSS OF HETEROZYGOSITY 145 margin of thè plateau. Thin line show ponds. Lower left figure indicates thè position of thè Giara di Gèsturi (dot) and of Capo Caccia (arrow) on Sardinia. On Capo Caccia all thè horses were individually known, as they were thè object of intensive fìeld work (Bogliani, in press). Data on population structure were tested for homogeneity using thè G-test for goodness of fìt with Williams’ correction (Sokal and Rohlf 1981). Analysis Effective population size and heterozygosity The main feature of population structure in thè Giara was unequal numbers of adult males and females. In this case thè most suitable formula to calculate thè Effective Population Size (Ne) (Wright 1969) is: Ne = 4M*F/(M + F) (1) where M = number of males F = number of females Other population parameters, such as variance in progeny (males or fema¬ les), unequal numbers in successive generations and non-random distribu- tion of family size should be accounted for to have a thorough estimation of Ne (Falconer 1989), but these data are unavailable for thè Giara or for any other feral population. Harris and Allendorf (1989) state that formula (1) might produce a large overestimation of Ne. However thè estimate made with (1) can be used hearing in mind that Ne is probably small than calculated. 146 G. BOGLIANI The loss of heterozygosity was calculated following Wright (1969) as: Ht = H0 (1 - l/(2Ne))‘ (2) where t = time in generations H0 = initial heterozygosity (assumed to be 1) Estimation of generation length Feral and free ranging horses can live more than 20 years, while thè fìrst reproduction is usually at 3 years for females. This produces a great overlap between generations, and mean generation length could therefore be calcu¬ lated from thè net maternity function lx mx, which is thè product of thè fe- cundity m at thè age x and thè probability of survival 1 to thè age x during thè fertile period of adult females (Eberhardt 1985). Neither lx or mx were available for Giara ponies, however data on fecundity are available for 5 North American populations (Sable Island - Welsh 1975, in Berger 1986; As- sateague Island - Keiper and Houpt 1984, in Berger 1986; Montana, free ranging - Speelman, Dawson and Phillips 1943, in Berger 1986; Great Basin - Berger 1986; Montana, feral - Garrott and Taylor 1990) and data on survival are given by Garrott and Taylor (1990). The medians of lx mx of females dur¬ ing their fertile period for thè above mentioned populations are 6, 7, 6 and 8, respectively. It is therefore reasonable to assume that thè mean generation of Giara ponies is within thè range of 6-8 years. Results Population size On thè Giara thè extrapolation of thè calculated mean density at thè three census areas (11.4 individuals per km2; Tab. 1) led to an estimated population size of 513 individuals. On Capo Caccia 25 individuals were present, with a density of 2.1 per km2. Population structure The estimated sex ratio did not change through thè study period (G = 0.02, df = 3, n = 230, n.s.), with an average of 0.12 males per female, or 1 male every 8.17 females, and a proportion of males among adults of 0.11. The number of foals per female and of yearlings per female changed over thè study period (Tab. 2). The relative number of foals was higher in autumn and winter and declined in spring, increasing during summer (G = 39.9, df = 3, n = 102, P < 0.001). Births mainly took place in spring, and to a lesser extent in summer. After Aprii most young were classifìed as yearlings. The peak in number of yearlings recorded was reached during summer, following thè presumably normal rhythm of birth in previous year, then thè number sharply declined (G = 74, df = 3, n = 80, P < 0.001) following thè capture of a large number by pony owners. Most captured yearlings were re- moved and sold, but a small fraction of females was released after branding. A few individuals escaped capture. On Capo Caccia, in spring and summer 1991, ponies were subdivided into fìve bands composed of one stallion and one or more mares with foals POPULATION STRUCTURE AND LOSS OF HETEROZYGOSITY 147 Tab. 2 — Observations on population structure on thè Giara di Gèsturi in 1989-90. date 28-29 Sept. 12-14 Dee. 28 Feb.-2 Mar. 14 Apr. 19-21 Jun. Total males (adults) 5 — 7 2 10 24 females (adults) 44 64 54 17 81 260(196)* males/females (adults) 1/8.8 — 1/7.71 1/8.5 1/8.1 1/8.17 foals 30 39 4 29 102 54** yearlings 6 7 10 57 80 foals/females 0.68 0.61 — 0.23 0.36 Q 49*** yearlings/females 0.14 0.11 — 0.59 0.70 Q 29*** * thè figure in parentheses indicates thè nuraber of females excluding those for December when males were not censused. ** foals and yearlings were not distinguishable. *** these ratios were calculated excluding data for February-March. and yearlings; there were also fìve bachelors. These horses are left unmana- ged as regards thè sex ratio and thè social structure. The sex ratio of thè population was dose to one, but if only stallions holding a harem are con- sidered, thè ratio was 0.5. Loss of heterozygosity The Effective Population Size calculated, Ne, equals thè adult popula¬ tion size when thè sex ratio is 1 : 1, but can be very low when there is an unbalanced sex ratio (fig. 2). The estimated Ne of thè Giara, calculated using M = 56 and F = 457 in formula (1), is Ne = 199. The loss of heterozygosity, calculated with formula (2), is shown in fig. 3. If thè current sex ratio management is continued in thè future, a 10% loss of heterozygosity may be expected in 42 generations (294 years). 50% of heterozygosity will be lost in 276 generations (1932 years). Discussion Comparison of thè population structure of thè ponies in thè Giara with that of Capo Caccia, whose sex ratio and age ratio are not altered by man, shows some differences (Tab. 3). There is some evidence in feral horses that thè proportion of males siring foals is higher than might be expected if only harem stallions have access to fertile females (Bowling and Touchberry 1990); in fact some adult bachelors or low-rank males in multi-male bands are likely to copulate with oestrous females. Therefore thè reai sex ratio in Capo Caccia is probably closer to 1 : 1. In The Giara thè sex ratio is heavily altered by man, and almost all yearling males are removed. During our ob¬ servations, only 3 yearling and 1 two-year-old males were observed after thè autumn round-up by horse owners. It seems unlikely that any foal be sired by very young males (Berger 1986), so that thè observed sex ratio is reliable. The number of foals per year per female was thè same in both areas (0.49-0.50), and was consistent with birth-rate fìgures for North American 148 G. BOGLIANI feral populations: Assateague Island, 0.57 (Keiper and Houpt 1984, in Berger 1986), Sable Island, 0.60 (Welsh 1975, in Berger 1986), Great Basin, 0.54-0.60 (Berger 1986), Montana 0.49 (Garrott and Taylor 1990). proportion of males among adults Fig. 2 — The Genetically Effective Population Size as a function of thè proportion of males in thè reproductive population of Giara ponies. Fig. 3 — Scenario of variation of heterozygosity (assuming thè present value of H0 to be one) through time with diflferent Effective Population Sizes on thè Giara. 1 - N = 513 (sex-ratio = 1 male/8.17 females) Ne = 199 2 - N = 1000 (sex-ratio = 1 male/8.17 females) 3 - Ne = N = 513 (sex-ratio = 1 male/l female) 4 - Ne = 1000 5 - Ne = 2000 POPULATION STRUCTURE AND LOSS OF HETEROZYGOSITY 149 Tab. 3 — Features of free ranging pony populations in Sardinia. Giara di Gèsturi Capo Caccia population size 513 27 males/females (adults) 1/8.17 1/0.91 (1/2)* foals/females 0.49 0.50 yearlings/females 0.70-0.11 0.57 * thè figure in parentheses refers to males owning a harem. There are two mechanisms which counteract thè effect of a small Ne in free ranging horses. The first is naturai selection, by which individuals carry- ing deleterious homozygote alleles should show a lowered fitness. The second mechanism is thè tendency to exogamy, which results in a non- random choice of mating partners (Duncan et al. 1984). In captive propagation schemes a common goal is thè maintenance of 90% of thè average heterozygosity for 200 years (Soulè 1986). These values are arbitrary, and were chosen in thè belief that, within this timespan, human population will have stabilized is growth and new technologies will be of help in maintaining genetic diversity. The Giara ponies face a serious threat of thè rapid loss of genetic variability especially if minor changes in thè wrong direction occur. Two alternative measures may be taken in order to reduce thè loss of heterozygosity. The first is thè maintenance of thè present total population size, but with a sex ratio of 1, which would maxi- mize thè eflfective population size. Under this regime, 10% of heterozygosity would be lost in 105 generations (735 years) and 50% would be lost in 693 generations (4851 years). These values are more than twice as high as with thè current unbalanced sex ratio. An alternative measure to reduce thè loss of heterozygosity would be an increase in numbers (N). With 1000 individuals and thè current sex ratio, thè estimated Ne would be 392. In this case, 10% of heterozygosity would be lost in 82 generations (574 years) and 50% in 543 generations (3801); these values are lower than those achievable where N = Ne = 513, which is thè maximum estimated N in thè case of perfectly balanced sex ratio. With a sex ratio of 1 : 1 and N = Ne = 1000, 10% of heterozygosity would be lost in 228 generations (1596 years) and 50% in 1505 generations (10,536 years). With N = 2000 and a sex ratio of 1 : 8.17, Ne would be 748; in this case 10% of hete¬ rozygosity would be lost in 165 generations (1155 years) and 50% in 1086 generations (7602 years). With a sex ratio of 1 : 1 and N = Ne = 2000, 10% of heterozygosity would be lost in 415 generations (2905 years) and 50% in 2736 generations (19,152 years). In 200 years, or about 29 generations, at least 7% of heterozygosity will be lost under thè current management regime. This figures are optimistic because of thè probability of thè overestimation of Ne discussed above. The goal of maintaining a large portion of thè present level of heterozygosity for thè future could be achieved 1) by maintaining thè present population size unaltered and achieving a sex ratio closer to 1 : 1, or 2) by increasing thè po- 150 G. BOGLIANI pulation size maintaining thè current unbalanced sex ratio. Increasing thè population size might cause a heavy impact on vegetation. The first solution seems to be more suitable but would mean lower income for horse owners because thè number of fertile females will be reduced. This problem could be solved by means of aid to owners given by public agencies interested in wildlife management and conservation or in rare stock maintenance, such as thè Sardinian Autonomous Regional Authority. Manipulations which would mitigate naturai mortality, such as supple- mentary feeding and veterinary help, should be discouraged because they could enhance thè survival of individuals hearing recessive deleterious cha- racteristics in homozygosity which could spread in thè population as a result of genetic drift. Acknowledgements - I wish to thank thè director and thè staff of thè Istituto di Zoologia, University of Sassari, for their hospitality during my residence in Sardinia. Thanks are due also to Marcello and Luisella Sequi for supporting in various way thè research. Assistance in thè fìeld by Vit¬ torio Baglione, Claudio Celada and Antonio Torre was greatly appreciated. The manuscript has benefìtted from criticai comments on earlier drafts by Marco Apollonio, Luigi Boitani, Gustavo Gandini, Rita Lorenzini, Sandro Lovari, Alberto Meriggi and Ettore Randi. Mark Pearce revised thè English. References Allendorf F. W. and Leary R. F., 1986 - Heterozygosity and fitness in naturai populations of animals. pp. 57-76, in Soulè M. E. (ed.) «Conservation biology. The Science of scarcity and diversity». Sinauer, Sunderland, MA. Berger J., 1986 - Wild horses of thè Great Basin. Univ. Chicago Press, Chicago. Bogliani G. (in press) - Social behaviour of coexisting feral horses and feral donkeys. Ethol. Ecol. Evol. Bowling A. T. and Touchberry R. W., 1990 - Parentage of Great Basin feral horses. J. Wildl. Manage. 54: 424-429. Cetti F., 1774 - I quadrupedi della Sardegna. Sassari. De Martis B. and Mossa L., 1988 - Giara di Gesturi. pp. 143-173, in Ca- marda I. and Cossu A. (eds.) «Biotopi di Sardegna. Guida a dodici aree di rilevante interesse botanico». Duncan P., Feh C., Gleize J. C., Malkas P. and Scott A. M., 1984 - Reduction of imbreeding in a naturai herd of horses. Anim. Behav. 32: 520-527. Eberhardt L. L., 1985 - Assessing thè dynamics of wild populations. J. Wildl. Manage. 49: 997-1012. Falconer D. S., 1989 - Introduction to quantitative genetics. Longmann, London. Fattorini L., 1992 - Modelli probabilistici per la valutazione del numero di animali in popolazioni selvatiche. Suppl. Rie. Biol. Selvaggina, 21: 465-477. Garrott R. A. and Taylor L., 1990 - Dynamics of a feral horse population in Montana. J. Wildl. Manage. 54: 603-612.