NAT. NAT. HIST. BULL. SJAM Soc. 53 (1): 71-78 ，2005

ARE HILL (G 此4. CULA RELIGIOSA) MORE COMMON IN DISTURBED THAN IN VIRGIN FOREST HABITATS?

Walter Walter A. Sontag ，Jr. 1

ABSTRACT

Habitat Habitat selection of Hill Mynas (Gracula religiosa) of the race intermedia was studied on on five visits to Kh ao Ang Rue Nai Wildlife Sanctuary ，southeastem 百 ailandτ 'h e tended tended to occur in pairs and were often found spatially clumped. Th eir distribution in two dry seasons seasons was analyzed. Th ey we 児 recorded significantly more frequently in moderately disturbed than than in 釦 lly intact prim 紅 y fores t. The wぉ confrrmed in all moderately disturbed sites but not not but in seve ra1 of the undamaged ones. On the other hand ，it wぉ found much more infrequently infrequently or not at all in greatly disturbed forest habitats and a reforested 釘 ea. Th is study suggests 白紙 good-quality fo 閃 st combined with a portion of open habitat may be highly at 町active for HilI Mynas during the non-breed 泊g season.

Key Key words: forest habitats ， Gracula religiosa ，habitat use ，Hill ， Khao Ang Rue Nai Wildlife Wildlife Sanctuary

INTRODUCTION

Wh ile there is a considerable amount of information on the dis 凶bution of and resource use use in South Asian 蜘 mids living in open habitat (e.g. ALI & RIPLEY ， 1972; Tu NHIKO 悶， 1990) ， such knowledge is much more limited with respect to forest-living stumids of 血e region ，including the Golden-crested Myna Ampeliceps coronatus and Hill Myna Gracula religiosa religiosa (e.g; FEARE & CRAIG ， 1998). Although records of these two forest-dwelling species 釘 e reported in more comprehensive avifaunistic literature (e.g. EVANS ET AL. ， 2000) ，detailed studies are lacking. Rough information on the distribution of both species in 官 lail 叩 d，p紅 ticul 紅 ly 泊 protected areas ，is presented by ROUND (1 988). Severe threats to to the Hill M 戸la originate from the pet trade (ROUND ， 1988; ARCHAWARANON ， 2003). Particularly Particularly in view of the ongoing habitat loss 加出 e species' dis 位ibutional ranges ，there is is a 蹴 ong need for data on the role various forest types ，i.e. virgin to differentially degraded ， play for the birds. In In an earlier study ， stumids were surveyed in Kh ao Ang Rue Nai Wildlife Sanctuary ， southeastem 百lailand ，血 two dry seasons (SONTAG ，1998). 百 is area is part of the 1紅 gest lowland lowland evergreen forest complex in Th ailand (CUBπr & STEWART-Fox ， 1995). Most of the the observations were done in three forest categories defined according to human impac t. Su 中risingly ，forest-living stumids appe 紅 'ed to be more common 泊 moderately disturbed 白m 泊 unaffected fores t. However ， no quantitative data were available for statistical analysis.

'Konrad 'Konrad Lo renz In stitute for Comparative Ethology ，Savoyens 回 sse la ，A-II60 Vienna ，Austria. Correspondence: Trubelgasse Trubelgasse 19 /2 4，A ・1030 Vienna ，Aus 甘ia. Received Received 18 April 2003; accepted 25 February 2∞ 5.

71 71 72 WALTER A. SONTAG ，JR.

Special Special methodological problems arose 合om habitat classification and defming comp 紅 'able sltes. sltes. More recently ， 1 revisited Kh ao An g Rue Nai Wildlife Sanctuary on 伽 ee occasions (SONTAG ， 1999). In the present paper 出e forest sites are critically re-evaluated based on all all five visits and an attempt is made to quantify the propo 武ional dis 回bution of one of the observed observed species ，出e Hill Myna (Gracula religiosa intermedia) ，泊 three forest types classified classified accord 加g to human impacts. In addition to 血e da 飽食om two dry seasons (1 994 ， 1995) ，some findings from the later visits are also reported where usefu l.

STUDY AREA

Kh ao An g Rue Nai Wildlife Sanctuary is located 泊 Chachoengsao ，Chonburi ，Rayong ， Chanthaburi Chanthaburi and Pr achinburi Pr ovinces in Southeast 百lailand. It covers a small part of the

Pr ach 泊b凶 floodplain and extends across the Chonburi 司 Chan 血aburi watershed (CUBITT & Sτ 芭WART-COX ， 1995; ROUND ，1988). 百 e sanctuary belongs to 血e1 訂 gest primary forest complex in Southeast 官lailand. Its topography is characterized by well-watered lowlands and several hills ， such as Kh ao Takrup (661 m) and Kh ao An g Rue Nai ， the highest point in in the sanc 知的 (777 m). Precipitation is variable accord 均 ωthe specific 間 as and is estimated estimated to be 1，50ι2 ，000 mm per year in most places relevant here (so 町 'ces: Industrial Pr omotion Center Region 9; Wildlife Research Station; see also Acknowledgements). Regardless Regardless of more subtle sub-distinctions ， the vegetation is generally classified as lowland evergreen evergreen forest. Extensive sections of the 釘 ea consist of selectively logged semi-evergreen forest forest on level ground and in 血 e vicinity of larger s住'eams at 10 0- 200 m elevation (ROUND ， 1988). 百 e most dominant tree is Lagerstroemia calyculata; the other 加 portant tree species include include Afzelia 勾Il ocarpa ， Pterocarpus macrocarpus ， Tetrameles nudiflora ，and Irvingia malayana (K 町'AWAC 臥 K 叫， 1998). Th e observations were made 白血e sanc 飢紅y ，including a buffer zone 紅 'ea 恒出e north-west ，泊 November 1994 and December 1995. Two more visits took place 泊 1997/ 1998 1998 (December/January) 佃 d 1998 (November /D ecember). Al l血e fo 町 stays were in 白e dry season. An additional visit took place in April 1996 when the climate was hot and we t. Count data from the latter period 紅 e used here only for discussion and when specifically stated. stated.

METHODOLOGY

Sites Sites

Th e study sites were si 卸 ated within approx. 13 ・15' ー13 ・28'N latitude and 101 0 40' -101 ・58'E longitude. Al l of them were located in the lowlands except for Site 17 on a gentle gentle slope and the steeper Site 25 in the Kh ao Takrup range (Fig. 1). 1 chose the sites randomly ，al 血ough ， for practical reasons ，accessib 出.ty played a role in the selection process. All 白e analyzed plots ，however ， were considered to be 加dependent ，加 cluding those 出at were in 由e same 訂 'ea (i.e. Sites 8a-c ，13a-c ， and 22/1-22/3). 官 le 29 available forest sites include include seven plots where former nesting-sites of forest-living mynas were known ARE HTLL MYNAS (G RAC ULA RE Ll GIO SA) MORE COMMON JN DISTURBED FOREST ? 73

Khao Khao

.25 I km A Takrllp

Range Range

24 内勾 8a . ・U 8b ・ 22β 6 62 212 8c ・7・ 5' マ ... main road 186 企三 4 22 /16 620 ・ 司唖ー マ6・ 企 9 "'1 8 21 • 企 10 rQut e lo 11 企企12 •19 D ・マ13c Sanam Sanam Chai Kh el 13. ママ 13b ・28 16'6 ・14 ' ・27 圃 Type A 15"' . 企 Type B マ Type C

Khao ・Ang Ru e Nai • Type 0 /

Figllre Figllre 1. Study sit e locat ion s in Kh aoA ng Ru e Na i Wildlife Sanctllary. T ype A- C: primary fores t. Typ e 0 forested reforested area. ASleri sk indi ca tes “n巴stin g- tr ee s it e"

(i .e. “nestmg-tree Slt 巴s"; Fig . 1) (P. PONSENA ，pers . comm.) but birds were not actually detected. detected. Al l sit es are attributed to one of s巴veral di stinct habitat typ 巴s according to the classification classification given below. As the sites other than “n巴sting-tree sites" were roughly comparab le in size (approx. 16 ha) ， only these 22 sites wer 巴 taken into account for 出e quantitative quantitative analyses. Additiona lly ，a r巴forested site was checked. This plot was deem 巴d as an appropriate contro l sit e becau se (1) other songb irds wer 巴 seen h巴I巴 ，(2) according to P .PONSENA (pers. comm.) ， Hill Mynas approached the se mi-refore sted area situated nearby ， and (3) Hill Mynas hav e be 巴n reported from 巴nvironments drastically changed by humans (ALI & RIPLEY ， 1972).

Sampling

Th 巴 aim of this inv es tigation was to record all H ill Mynas present in the sites as complete ly as pos sible in order to gain ar 巴liab le measur 巴 for th 巴 relative abu nd a nce of the speci 巴s in different habitat type s. The study does not ，howeve r， attempt to d巴termine exact overall overall densities. Thus ，specia l considerat ion was given to the comprehens iv eness of the surv 巴illance of the respective sites . For this purpose ，a combination of lin e transects and point point co unt s was applied covering walking di stances of approximate ly 500 m (range c. 400-1 ，000 m). As total visual overvi 巴w u sually was not pos sible ，acoustic evidence was 74 74 W 札叩R A. So 附 AO ， JR. crucial crucial except for larger groups (i .e. flock size > 6 individuals). Hence ， only in Site 22/1 did did visual cues play the key role. Due to 血e res 甘icted sizes of the plots 白.e c'O un 血19 distance distance was clearly less 白an the detection distance 泊 all habitat types. Birds more 白血 200 m away were excluded from counting. Sites were visited 'O n one t 'O four days (Ar ea 22: 22: four times on three days and a partial visit on another day). Coun 曲19 t加 ew 剖 c.3 0-4 0 minutes minutes per site (exception: Si 総 17 - roughly 60 min for 1，000 m). 官lU S， the same am 'O unt of effort w 剖 applied at all sites. 官官 time spent 泊 Sites 27 加 d 28 (no birds found) situated situated in an extended forest area was shorter. Th ese data ，however ， were included because adjoining ，much larger forest parts of simil 紅 quality were visited directly before and afterwards afterwards and no birds were detected there either. Additional Additional confidence in the results was provided by observations made at the sites bef 'O re an d/ or after the pr 'O per counting sessions whenever possible. F 'O r the difficult Site 23 ，useful information from the later stays was als 'O taken into accoun t. As suggested by direct observations and confirmed on later visits ，two individuals usually usually occ 町 at/ close to a specific sound source ， even if only one of the birds may be detected. detected. Such records were therefore taken 出向presenting two individuals ， except for one case case in which definitely only one individual was documented. On a few occasions it was not not possible to determine the number of birds with certainty due to p 'O or visibility. Hence ， two different types of data are presented: confirmed records ，i.e. minimum individual numbers ， and estimated numbers. Independent Independent of all p釘沼田ters including site ，habitat type and observation peri 'O ds ， in total ， between 67 to 128 individual Hill Mynas were recorded. 百le wide range between minimum and maximum numbers was largely due t 'O flocking birds observed in an 紅 eaon 'O ne day under difficult viewing conditions.

Habitat Habitat Classification

Four Four habitats 釘 e di 首位entiated here according to human impac t. Thr ee of them 紅 e primary primary forest habitats (Types A ，B ， and C ，respectively); 出ey could be qualitatively distinguished distinguished by sight. Th e reforested control site was designa 飽d Type D. Type A: 白lly 泊tact prim 紅 y forest; undamaged pristine ， dense vegetation wi 由 l訂 ge trees; trees; n 'O l'O gged 紅 'ea or (in s'O me cases) few limited forest patches or s凶 ps damaged by man ， such as paths and selectively removed 紅白s (see also below). Type B: m 'O derately disturbed primary forest; largely dense ，original vegetation with high high trees; however ， extended patches of l'O gged or otherwise damaged vegetation or clear 加gs; up to 50% of the 釘 'ea damaged or logged. Type C: severely devastated habitat; much open area ，“ savarmah ・like"; a few isolated 'O r scattered large trees 'O r 仕'ee-remnants 紅 e left 佃 d 由e surrounding f 'O rest edge is usually present; present; in 住'O duced 住ees may 'O cc 町; less 血an 20% of 血.e area c'O vered by orig 泊al fores t. Type D: reforested area containing few 佐田species (such 錨 Caesalp 泊iaceae; Leucaena leucocephala ， Mimosaceae); estimated tree heights 7 m. 百le present habitat classification partially differs from the initial scheme (SONTAG ， 1998). 1998). Originally ， the at 凶 butes had been based 'O n the change of af 'O rest section by humans regardless of the actual consequences of human activity (considering the main r'O ad per se as neu 仕al ， where applicable). For instance ，Sites 8a ，8b ， 8c situated 泊 a buffer zone zone were characterized by very large trees ，much dense vegetation and extended patches ARE HILL MYNAS (GRACULA RE Ll GIOSA) MORE COMMON IN DISTURBED FOREST? 75 of of a Type A character despite some former human encroachmen t. Thi s forest part now appe 釘 'ed undisturbed by humans ， and the presence of a well-established wild elephant population population suggested that the area was successfully protected. Th us ，it is appropriate to at 凶bute Type A quality to those sites 血血 is investigation.

In In total ， 13 forest sites were 出 sessed to be Type A habitats ， 11 Type B ，組 d5 Type C. C. Excluding the “nesting-tree sites ヘ11 sites were Type A ，8 Type B ， and 3 Type C.

Data Analysis

百le f同quencies of recorded Hill Mynas per site were compared (a) between all 由民e prim 釘 y habitat types A ，B ，C (Kru skal-Wallis test) and (b) solely between A and B (Mann- Whi tney U test) ， which by f: 紅 were the most frequent habitat types. SPSS software (version 7.5 7.5 for Windows) was used.

RESULTS

H 出 Mynas occurred 泊 all primary forest types studied (Table 1) ， whereas no individuals were were found in the reforested Type D habitat. An overall comp 剖 son between the primary forest types ，including the unde 町 epresented Type C sites (i.e. devastated habitats) ，clearly revealed significant frequency differences for both both the minimum individual numbers and the estimated numbers (Kru skal-Wallis tests: Chi-Squ aIち= 8.894 ，P = 0.012 ， and Chi-Square = 7.842 ，P = 0.020 ，respectively; df = 2， N 1 = 11 ，N 2 = 8，N 3 = 3). 官le same was 甘ue for the respective comparisons between Type A and B sites alone (two-tailed U tests: z = -2.637 ，P = 0.008 ， and z = -2.382 ，P = 0.017 ， respectively). respectively). Frequencies in Type B habitat were clearly higher compared to the fully intact intact sites (Type A). Th e minimum Hill Myna records and estimated numbers di 首ered

Table Table 1. Recorded f詑quencies of Hill Mynas in 出ree primary forest types (excluding “nes 由19-tree sites") defmed accord 加g to human impac t. Minim um individual numbers and estimated record numbers (related to 16 hectares ，i.e. individual site site size) 釘 e given (see text).

TypeA TypeB Type C

Number of sites (n) 11 8 3 Sites Sites with birds present 7 8

Number of b廿ds: Minimum individual numbers range range ι2 1-16 0- 2 median 2 4 0 Est. Est. record numbers range range 0-4 1-24 。ー 2 median 2 4 0 76 76 WALl 芭R A. SONTAG ， JR.

only only slightly. Th e values of both categories deviated for 0 叫y one Type A and B site. S住注 ingly ，no Hill Mynas were found 加 4 of 11 Type A sites ， whereas in all Type B sites individuals individuals were recorded. 官le considerable variation of individual numbers among the sites (Table 1) has 由民e explanations. explanations. First ， Hill Myn 回， if present ， tended to occur 泊 pairs and not as single birds. Only once was a single ， flying documented. In a second case ， a conspecific was probably probably nearby or only briefly separated. Second ，pairs were often dis 肘buted 泊 a spatially clumped m 釘mer ，especially apparent 泊Ty pe B habitats. 官官d，a patchy dis 佐ibution pat 飽m was caused by flocking birds. In one case ，possibly up to 40 Hill Mynas may have aggregated. aggregated. In血 is instance ， at least one rich feeding resource ，a large fruit-bearing Lagerstroemia Lagerstroemia tree with fruiting Ficus around it ，was available. Th e most heavily used sites were Type B 22-1 to 22-3 with tall trees but a lot of open habitat 泊白ec 即位e(cle 訂 ing). Hill Mynas were also very common 泊出is 訂 ea on the three later later visits. Pairwise Pairwise statistical analyses between Type C habitat and each of the other two prim 紅 y forest forest habitat types are less indicative because the number of available C sites was very low. low. Never 仕leless ， the single comparison between Type B (much utilized) and C showed significant a significant difference 泊 individual 合'equencies (two-tailed U test: z = -2.084 ，P = 0.037 for for both minimum and estimated record numbers). Recorded bird numbers at bo 也 A 佃 d C sites were too small to statistically comp 訂 e the two habitat types. Only one pair of Hill Mynas was actually confrrmed 加 Type C habita t.

DISCUSSION AND CONCLUSION

Although Although Hill Mynas occurred in all three primary forest types ，白 ey were found in significant significant numbers only 加 Type A and B habitats. Both habitat types were characterized by many 1釘 ge trees ， such as Lagerstroemia ，which 訂 e hig hI y at 住active for 血is arboreal species species (pers. obs.) and used for nesting (P. PONSENA & C. KAPASUWAN ，unpub l. report; ARCHAWARANON ，2003). 百le birds were ，however ，reported distinctly more often from Type B 由加 Type A forest. Remarkably ， Hill M 戸las intensively utilized a buffer zone section section (including Type A Sites 8a ，8b ，8c) ， where limited logging had formerly t紘 en place ， but whose overall aspect was 白紙 of an open to moderately dense Type A fores t. Even the Golden-crested Myna was observed here. In con 住ast ，no Hill Mynas were found in in a steep valley characterized by dense ，白 lly intact vegetation and very high trees 加出e north north of the sanc 制御"y (est. 30 0-4 ∞0o m al t.ふ Likewise ，on a 凶P to Kh ao Yai National Park (November 1998) ，no Hill Mynas were discovered in a similar habitat ， a steep-sided valley (est. (est. 550 m al t.ふ Note ，however ， that Hill Myna records 企om mountainous areas are inconsisten t. In another part of Southeast Th ailand (Khao Soi Dao Wildlife Sanctuary) ， Hill Hill Mynas were observed in such forest in December 1998 obs.). (pers. For the same race (intermedia) ，BERTRAM (1970: 90) classifies ravines and steep slopes as typical Hill Myna habitat habitat in Assam. On Kh ao An g Rue Nai ， Hill Mynas are common at least on the lower third third of the mountain (see below). 百lere may be a difference between races. In South Th ailand ， the southem race religiosa appears to be restricted to level lowland (ROUND ， 1988). 1988). In In dia and Sri Lanka ，populations (and races) differ in altitudinal distribution (compiled (compiled by BERTRAM ， 1970). ARE HILL MYNAS (GRACULA REUGIOSA) MORE COMMON 町 DISTURBED FOREST? 77

Two arguments might be raised against 血e interpretation 白紙 Type B habitat is more attractive attractive for Hill Mynas than virgin forest (Type A). First ， sample sizes 釘 e rather small. Both minimum 泊dividual numbers and estimated numbers ，however ，indicated significant differences. differences. Second ，one could 訂 gue that it might be more difficult to detect Hill Mynas in in dense Type A fores t. Accordingly ， the result would rather reflect the dis 凶 bution of records records than the ac 加 aloccu 町 ence of the birds. As counterarguments ， Hill Mynas can be localized localized by ear relatively easily and counting distances were less than possible detection distances. distances. Th e case of Sites 22-1 to 22-3 also clearly demonstrates 白紙 open habitat (cle 紅泊g with recently plan 旬d tree saplings) and its surrounding forest sections can strongly attract attract pairs as well as flocks (cf. SONTAG ， 1998). Th e combination of available fruit 位'ees and and relatively 1釘 ge scanning space the for birds to avoid predation (see ROB 別 ETIE & HA ， 2001: 2001: 448) might have con 凶 buted to this preference. Finally ，accumulations of numerous individuals individuals were spotted at another Type B site in the sanctu 釘 y in 血e dry se 拙 on in 1998. According According to ALI & RlP LEY (1972) ， small flocks occur 鉱山e edge of closed forest or in cultivation cultivation clearings with sp 釘 se sprinkling of standing 位ees 加 the non-breeding season. In In my study area ， Hill Myna records were clearly underrepresented in Type C habitats (i .e. devastated devastated forest) ，especially when considering 白紙 (a) in this environment birds were relatively relatively easy to see and (b) much better forest ，inhabited by Hill Mynas ，was situated nearby. nearby. If site numbers ，especially of Type C ，had been 1紅 ger ， then the minor use of Type C habitat might have been statistically significant (Type A vs. C) or more markedly significant 由加 shown here (Type B vs. C).官 lat no sightings were made in the reforested

釘 ea with trees measuring less than 10 m high (Type D) is in line with the observed “trend" in in Type C habitat. 百le observations provide evidence for patchy habitat utilization.τ 'h is becomes even clearer clearer when forest 釘 'eas qualitatively surveyed outside the proper study sites are also taken 加to accoun t.官 le fruit production of 仕'ees (i.e. non-fruiting ，合 ui 由19 ，di 妊erent stages of of ripeness) might be much more important for Hill Mynas 出an the structural difference between between Type A forest and high quality B forest (cf Au & RIPLEY ， 1972: 192) ， at le 槌 t in in the dry season when 血ere is no need for breeding-sites. BER' 百 AM (1 970: 84) points out that that large-scale or long-distance migrations probably do not occur ， but that regular seasonal movements and altitudinal changes 泊 range 紅 'e known ， probably dependent on chang 泊g food food supplies and the availability of nesting-sites. 百le flocks observed 泊出e sanc 旬紅y outside outside the breeding season also suggest that the distribution of individuals varies with the time time of ye 釘. Two fmdings 仕om April 1996 would fit with potential changes in local distribution. distribution. At 白紙 time ，precipitation was high in the sanc 伽 ary. During a short visit ，Hill Mynas were spotted several times on Kh ao An g Rue Nai mountain ，which supports a large virgin virgin forest 釘 'ea. 百le dis 住ibution of these records was clearly uneven. Furthermore ， close to 血e park station at 血e foot of the mountain were four neighboring 住ees containing cavities cavities that had b関 n used by Hill Mynas as nesting-sites. Th is additionally suggests that Hill Hill Mynas do 'not necessarily avoid the proximity of humans. In In conclusion ，Hill Mynas utilized moderately disturbed more 出佃fully intact primary forest forest habitats 泊白.e northem part of Kh ao An g Rue Nai Wildlife Sanc 旬紅y outside the breeding breeding season. Habitat choice in other places and the role of different forest types (e.g. ， open vs. dense; moderately damaged vs. virgin) for 血is stumid in the breeding season require require further rese 釘 ch. 78 W 札花R A. So 町 'AG ，JR.

ACKNOWLEDGMENTS

1 am grateful to Dr. Schwann Tunhikom and Dr. Viroj Pimmanrojnagool (bo 出 Royal Forest Forest Department) for allowing me to stay in the sanc 加訂y. 1 白創lk Pongsak Ponsena ， Sawai Wanghongsa (bo 血Kh ao An g Rue Nai Wildlife Sanctuary Research Station) and Narong Koonkhunthod (F ield Assistant in the Sanctuary) for suppo 民泊g my research most effectively. effectively. Pongsak Ponsena provided information on former nesting-sites of Hill and Golden-crested Golden-crested Mynas. Special thanks go to Dr. Pilai Poonswad (Mahidol University) and Dr. Dr. Schwann Tunhikom for their selfless and continuous support of my studies in 官lail 佃 d. V 司ak Chimchome (Kasetsart University) helped me in many respects. 1 am deeply grate 釦l to to Dr. E. Nemeth (Konrad Lorenz In stitute for Comparative E 血ology ， Vienna) for detailed discussions discussions and profound suggestions on the manuscrip t. Dr. H. Hoi (same 加stitute) and two anonymous refere 四 provided highly valuable comments. 1 also thank a reviewer for providing providing additional rainfall information. Dr. M. Stachowitsch (In stitute of Ecology and Conservation Conservation Biology ， University of Vienna) critically read various versions of the manuscript manuscript and improved the English and Dr. A. Seidel (same institute) helped in statistical and software matters.

RE 四 RENCES

Au ，S. ， AND S. D. RIPLEY. 1972. Handbook of the birds of India and Pakistan together with those of Nepal ， Sikkim ， Bhutan and Ceylon. Vo l. 5. Oxford University Pre ss ，Bombay ，Lo ndon & New York. AR 四 AWARANON ， M. 2003. 四 e 加 pact of human interference on Hill Mynahs Gracula religiosa breeding 血 ailand. Th ailand. Bird Conservation International 13 (2): 139-149 BERTRAM ， B. 1970. 百le vocal behaviour of the In dian Hi1l Mynah ， Gracula religiosa. Behaviour Monographs 3 (2): 81-192. C田口T ，G. ， AND B. STEWART-COX. 1995. Wild Thailand. Asia Books ，Bangkok. FEA 胆， C. ， AND A. CRAIG. 1998. and Mynas. Christopher Helm Ltd ，A & C Black Ltd ，Lo ndon. EVANS ，T. D. ， H. C. TOWLL ， R. J. 百MMINS ， R. M. THEwus ， A. J. STONES ， W. G. ROBICHAUD ，AND J. BARZEN.

Ornithological 2000. Ornithological records from the lowlands of southern Laos during Dec ember 1995 ・ S巴ptember 1996 ，including 眠ぉ on 白e 司副組 d Cambodian borders. Forktail 16: 29 ー52. KITAWACHAKUL ， N. 1998. Some findings and outstanding issues regarding community forestry in bufferzone management: management: ac 鎚 e study on the An g-Rue-Nai Wildlife Sanc 佃 ary in e鎚 tern Th ailand. Pages 275-284 in M. M. Victor ，C. L組 g and J. Bornemeier (eds.) ，Community forestry at a crossroads: reflections and future directions directions in the development of community forestry ・RECOFTC R 里port 16 ，Bangkok. ROB 別町布，R. L.， AND J. C. HA. 200 1. Social and ecological factors influencing vigilance by northwestern crows ， Corvus Corvus caurinus. Anim. Behav. 62 (3): 447 -4 52. ROUND ，P. D. 1988. Resident forest birds 泊百ailand: their sta 佃 sand conservation.ICBP Monograph 2. In temational Council Council for Bird Pre servation ，Cambridge ，U.K. So 町AG ， W. A. ，JR. 1998. Devastated ，damaged ，and fu Il y intact forest habitats and the Sturnidae 伽 nily in a dry lowland lowland everg 問 en biome in Southeast Th ailand. Nat. Hist. Bull. Siam Soc. 46 (1): 43-53. SONTAG ， W. A. ，JR. 1999. Whi ch forest categories do “forest mynahs" really prefer? A study revisited and new findings 仕om 百 ailand. Zoology 102 ， Supplement II: 86. Urban & Fischer ，Jena. Tu NHJKORN ，S. 1990. Resource p紅白tioning of four sympa 凶 cmyn 悩飢ds 凶 1泊.gs (Stun 副総) in 百凶land. Unpub l. Ph. Ph. D. thesis. Oregon State University.