萬(wàn)雅瓊，郭偉波，吳 軍，李 莉，徐海根①

(1.南京師范大學(xué)生命科學(xué)學(xué)院，江蘇 南京 210046；2.環(huán)境保護(hù)部南京環(huán)境科學(xué)研究所，江蘇 南京 210042)

物種占域模型的基本原理及其研究進(jìn)展

萬(wàn)雅瓊1，郭偉波2，吳 軍2，李 莉1，徐海根2①

(1.南京師范大學(xué)生命科學(xué)學(xué)院，江蘇 南京 210046；2.環(huán)境保護(hù)部南京環(huán)境科學(xué)研究所，江蘇 南京 210042)

占域模型是用于估算某個(gè)區(qū)域被目標(biāo)物種所占據(jù)的比例，從而進(jìn)一步估算物種的豐度、預(yù)測(cè)物種的分布范圍和了解群落結(jié)構(gòu)的一種模型。主要介紹了占域模型的基本原理，包括數(shù)據(jù)類(lèi)型、基本參數(shù)、模型選擇及適用范圍，總結(jié)了占域模型的研究和應(yīng)用進(jìn)展。目前，對(duì)該模型的研究與應(yīng)用仍以國(guó)外學(xué)者居多，但近年來(lái)國(guó)內(nèi)一些調(diào)查、監(jiān)測(cè)工作中也有初步應(yīng)用。在此基礎(chǔ)上，分析了該模型實(shí)際應(yīng)用中可能出現(xiàn)的一些問(wèn)題，如協(xié)變量的引入、探測(cè)率的異質(zhì)性、不完全的探測(cè)和觀測(cè)值的缺失等，并對(duì)這些問(wèn)題作出了一定的解釋。最后，提出進(jìn)一步完善占域模型并將其應(yīng)用于我國(guó)生物多樣性調(diào)查與監(jiān)測(cè)計(jì)劃中，從而為生物多樣性保護(hù)提供有效的技術(shù)支持。

占域模型；占域率；探測(cè)率；協(xié)變量

近幾十年來(lái),全球生物多樣性下降趨勢(shì)顯著,各國(guó)紛紛開(kāi)展了一系列生物多樣性調(diào)查與監(jiān)測(cè)項(xiàng)目,以期掌握生物多樣性的動(dòng)態(tài)變化趨勢(shì),對(duì)其開(kāi)展有針對(duì)性的保護(hù)。占域模型(site occupancy model)是近年來(lái)發(fā)展起來(lái)的一種新的模型方法,該模型最早由MACKENZIE等[1-2]提出,用于根據(jù)物種調(diào)查中的出現(xiàn)數(shù)據(jù)估算物種空間占有率及其動(dòng)態(tài)變化?？臻g占有率(occupancy rate)指某樣點(diǎn)(site)或某斑塊(patch)被目標(biāo)物種占據(jù)的概率[3],即該物種在此棲息地中分布的比例(proportion of area occupied,PAO),這在某種程度上可以作為物種種群數(shù)量的替代參數(shù)。對(duì)特定物種來(lái)說(shuō),空間占有率可以使研究者快速地了解種群密度情況,進(jìn)一步評(píng)估影響物種分布的因素。占域模型可以在目標(biāo)物種占據(jù)某樣點(diǎn)但不是每次都被探測(cè)到的情況下,通過(guò)多次重復(fù)調(diào)查,科學(xué)地估算單次調(diào)查對(duì)目標(biāo)物種的探測(cè)概率(detection probability),進(jìn)而估算出特定樣點(diǎn)或區(qū)域被目標(biāo)物種占有的概率[4]。該模型只需得到目標(biāo)物種在樣點(diǎn)是否出現(xiàn)的數(shù)據(jù),而不需要調(diào)查物種數(shù)量,與之前物種豐度或種群大小估算方法,如標(biāo)志重捕法等相比,該模型方法更加簡(jiǎn)單實(shí)用[5]。因此,近年來(lái)占域模型在多個(gè)領(lǐng)域得到了研究和應(yīng)用[6-7],也解決了一些生態(tài)學(xué)問(wèn)題,如棲息地與物種分布的關(guān)聯(lián)[8-10]、物種的分布狀態(tài)[11-12]、集合種群動(dòng)態(tài)[13-18]等。同時(shí),由于占域模型不需要投入大量的調(diào)查工作(sampling effort),因而在野生動(dòng)物調(diào)查和長(zhǎng)期監(jiān)測(cè)中得到廣泛應(yīng)用[19-20]。為讓讀者更系統(tǒng)地理解并應(yīng)用占域模型,筆者詳細(xì)介紹了占域模型的基本原理、所需數(shù)據(jù)類(lèi)型、基本參數(shù)和適用范圍,并分析了應(yīng)用中可能出現(xiàn)的一些問(wèn)題。

1 占域模型的基本原理

2 模型假設(shè)

所有模型都有其應(yīng)滿足的條件或假設(shè),占域模型也不例外。MACKENZIE等首次提出的占域模型需要滿足以下假設(shè)[1]:

(1)在目標(biāo)物種整個(gè)調(diào)查期間樣點(diǎn)的占據(jù)狀態(tài)保持不變,也即之前被占據(jù)的樣點(diǎn)仍然被占據(jù),未被占據(jù)的樣點(diǎn)仍然未被占據(jù),不存在物種遷入或遷出的可能。

(2)在任何一個(gè)樣點(diǎn)目標(biāo)物種的調(diào)查均獨(dú)立于其他樣點(diǎn),沒(méi)有樣點(diǎn)與樣點(diǎn)間的交叉或樣點(diǎn)與樣點(diǎn)間物種的相互交換。如果2個(gè)相鄰樣點(diǎn)離得太近,就有可能發(fā)生目標(biāo)物種在2個(gè)樣點(diǎn)間的來(lái)回移動(dòng),從而使得目標(biāo)物種的同一個(gè)體在多個(gè)樣點(diǎn)被重復(fù)調(diào)查,在調(diào)查方案的設(shè)計(jì)過(guò)程中就需要避免出現(xiàn)這種問(wèn)題。

(3)探測(cè)率在每個(gè)樣點(diǎn)的每次調(diào)查中都恒定,也即探測(cè)率在時(shí)間或空間上不存在異質(zhì)性(heterogeneity),或者即使存在異質(zhì)性,但能夠用協(xié)變量(如棲息環(huán)境、探測(cè)特征)進(jìn)行解釋。目標(biāo)物種在每個(gè)樣點(diǎn)內(nèi)的重復(fù)調(diào)查中占域率都是恒定的,或者即使存在不同的占域率,但能用棲息地協(xié)變量進(jìn)行解釋。

(4)目標(biāo)物種在每個(gè)樣點(diǎn)的出現(xiàn)/未出現(xiàn)記錄都是正確的,即物種的識(shí)別都是正確的。

3 模型的協(xié)變量與選擇

3.1 協(xié)變量

占域模型中涉及到2個(gè)最主要的參數(shù),即占域率(ψ)和探測(cè)率(P),但要更準(zhǔn)確地求出每個(gè)樣點(diǎn)的占域率和探測(cè)率,就需要引入?yún)f(xié)變量(covariates)[29-31]。協(xié)變量可分為2類(lèi)[32]:一類(lèi)是樣點(diǎn)協(xié)變量,也叫棲息地協(xié)變量(site-specific covariates),在同一時(shí)間段每個(gè)樣點(diǎn)的棲息地協(xié)變量是恒定的,但在調(diào)查過(guò)程中如果時(shí)間跨度比較大,如前期調(diào)查在夏季,后期調(diào)查在冬季,棲息環(huán)境就會(huì)發(fā)生變化,如棲息地類(lèi)型、植被、海拔、樣方大小、是否被干擾等;另一類(lèi)是探測(cè)協(xié)變量(sampling-occasion covariates),可因每個(gè)樣點(diǎn)的每一次調(diào)查而不同,如調(diào)查時(shí)的環(huán)境條件(氣溫、霧霾、云層)、一天中的不同時(shí)段、不同調(diào)查者等。樣點(diǎn)協(xié)變量和探測(cè)協(xié)變量都可能會(huì)影響探測(cè)率,而占域率只受樣點(diǎn)協(xié)變量的影響,不隨探測(cè)協(xié)變量而改變。

一些占域模型的研究考慮了協(xié)變量。MACKENZIE等[1]將環(huán)境氣溫、不同棲息地類(lèi)型(池塘、湖泊、沼澤、濕地)作為協(xié)變量用于美國(guó)馬里蘭州2個(gè)無(wú)尾兩棲動(dòng)物的占域模型中;BAILEY等[22]在研究美國(guó)大霧山國(guó)家公園的7種陸棲蠑螈時(shí),將時(shí)間和4個(gè)棲息地特征(是否有干擾、植被類(lèi)型、海拔、溪流)作為重要的協(xié)變量參與模型分析,確定影響蠑螈物種生存最重要的因素;LI等[33]在估算四川王朗和臥龍自然保護(hù)區(qū)雉類(lèi)占域率和探測(cè)率時(shí),引入8個(gè)棲息地協(xié)變量和3個(gè)探測(cè)協(xié)變量,并通過(guò)評(píng)估棲息地協(xié)變量對(duì)雉類(lèi)生存的影響,提出對(duì)雉類(lèi)實(shí)施有效保護(hù)的措施;SUTHERLAND等[34-35]將滅絕(extinction)和定居(colonization)這2個(gè)協(xié)變量引入占域模型,并將其廣泛地應(yīng)用于集合種群動(dòng)態(tài)的分析。

3.2 模型選擇

占域模型被用于估算占域率和探測(cè)率,棲息地協(xié)變量和探測(cè)協(xié)變量會(huì)影響這些變量的估算[36],應(yīng)分析評(píng)估這些協(xié)變量對(duì)模型占域估算的影響程度[37]。這可通過(guò)每個(gè)模型的赤池信息量準(zhǔn)則值(Akaike information criterion,AIC,CAI)和模型權(quán)重(model weight)與零模型(null model)進(jìn)行比較[38-39]。首先,假設(shè)樣點(diǎn)占域率恒定,即ψ(.),允許物種探測(cè)率隨調(diào)查時(shí)間和棲息地協(xié)變量的變化而變化,即P(t)和P(Cov),分別進(jìn)行分析;然后,假設(shè)物種探測(cè)率恒定,即P(.),占域率隨棲息地協(xié)變量的變化而變化,即ψ(Cov),進(jìn)行分析。這些分析功能都能用Presence軟件實(shí)現(xiàn)[40]。每個(gè)物種的占域模型根據(jù)其ΔCAI值和模型權(quán)重進(jìn)行評(píng)估。模型權(quán)重在0～1之間,并且所有模型權(quán)重之和為1,模型權(quán)重越大,表示該模型是真實(shí)模型的可能性就越大。ΔCAI值越小的模型就越接近于最優(yōu)模型。有研究證明ΔCAI≤2的模型都被認(rèn)為是等效模型[41-42],可用于模型的參數(shù)估算。因此,根據(jù)ΔCAI值篩選出ΔCAI≤2的所有等效模型進(jìn)行下一步的模型估算。

4 模型探測(cè)率的異質(zhì)性

探測(cè)率是當(dāng)目標(biāo)物種在一個(gè)樣點(diǎn)出現(xiàn)時(shí),該物種在該樣點(diǎn)的一次調(diào)查中被探測(cè)到的概率,若要得到探測(cè)率,可通過(guò)調(diào)查物種個(gè)體、辨聽(tīng)叫聲、定位跟蹤、領(lǐng)地標(biāo)記等手段獲得出現(xiàn)/未出現(xiàn)數(shù)據(jù)來(lái)估算。但是對(duì)于大多數(shù)物種來(lái)說(shuō),由于在調(diào)查過(guò)程中會(huì)出現(xiàn)各種不可避免的因素,包括調(diào)查者經(jīng)驗(yàn)、天氣狀況、調(diào)查時(shí)間等,使得每次調(diào)查的探測(cè)率都可能不一樣,即探測(cè)率的異質(zhì)性[43-45]。探測(cè)率的異質(zhì)性會(huì)直接影響占域率估算的準(zhǔn)確性,為了獲得可信的占域率,須對(duì)其異質(zhì)性進(jìn)行合理的解釋[46]。在野外調(diào)查中很多物種都不能被準(zhǔn)確地探測(cè)到,因?yàn)榇蠖鄶?shù)出現(xiàn)在樣點(diǎn)的物種在調(diào)查期間并不總是恰好能被探測(cè)到[47-51]。對(duì)于大多數(shù)類(lèi)群而言,尤其是廣布的、隱蔽性的物種,探測(cè)率總是小于1且不恒定[52],即使是不運(yùn)動(dòng)的生物體(如植物)也是一樣[53-54],而且探測(cè)率在時(shí)間上可能會(huì)隨季節(jié)和環(huán)境條件的不同而有所差異,在空間上可能會(huì)隨特定樣點(diǎn)棲息地特征不同而有所差異[55-56]。如果調(diào)查到60%的目標(biāo)物種,一種情況可能是該物種出現(xiàn)在整個(gè)區(qū)域,探測(cè)率為0.6;第2種情況可能是該物種只出現(xiàn)在60%的區(qū)域,探測(cè)率為1;第3種情況可能是探測(cè)率為0.6和1之間的任何一個(gè)數(shù)值。探測(cè)率的異質(zhì)性往往會(huì)導(dǎo)致占域率估計(jì)值偏低(negatively biased)。為了減少占域率估算的誤差,可適當(dāng)增加目標(biāo)區(qū)域的樣點(diǎn)數(shù)、增加每個(gè)樣點(diǎn)的重復(fù)調(diào)查次數(shù),以提高探測(cè)率,使其更接近于1。

5 模型的應(yīng)用

迄今,已有越來(lái)越多的國(guó)外學(xué)者在研究占域模型并將其應(yīng)用于多個(gè)生態(tài)學(xué)領(lǐng)域[57-59]。MACKENZIE等[60]也進(jìn)一步詳細(xì)介紹了占域模型在各種情形下的應(yīng)用。BAILEY等[25]在4—6月中旬對(duì)美國(guó)大霧山國(guó)家公園的陸棲蠑螈進(jìn)行調(diào)查,共獲得39個(gè)樣點(diǎn)的探測(cè)/未探測(cè)數(shù)據(jù),得到的真實(shí)占域率為0.26。但筆者考慮到其占域率可能受干擾因素的影響,另外,探測(cè)率也可能隨環(huán)境條件的差異,如降雨和溫度的不同而不同,故將是否有干擾(Idist)作為棲息地協(xié)變量、將是否有干擾和調(diào)查時(shí)間(t)作為探測(cè)協(xié)變量進(jìn)行綜合分析后得到8個(gè)候選模型。根據(jù)ΔCAI值選出ψ(Idist)P(Idist+t)、ψ(Idist)P(t)、ψ(Idist)P(Idist)和ψ(Idist)P(.)4個(gè)最佳模型用于數(shù)據(jù)分析,結(jié)果表明在之前受過(guò)干擾的樣點(diǎn)和未受干擾的樣點(diǎn)占域估算值分別為0.19和0.7,探測(cè)率也隨不同調(diào)查時(shí)間和是否受過(guò)干擾而不同。ALBANESE等[61]在2014—2015年期間調(diào)查喬治亞州西南部的一種魚(yú)類(lèi)(bluenose shiner),共調(diào)查了39個(gè)樣點(diǎn),獲得實(shí)際占域率為13%,估算其占域率和探測(cè)率,并引入棲息地協(xié)變量將其與基本模型進(jìn)行比較,結(jié)果表明引入?yún)f(xié)變量的模型ΔCAI值最小,能更好地用于描述探測(cè)率和占域率,更好地用于解釋不完全的探測(cè)。WEIR等[62]用北美10個(gè)州的7 a數(shù)據(jù)(2001—2007年)描述了兩棲動(dòng)物第1個(gè)多年占域趨勢(shì)；用多季節(jié)占域模型分析樣點(diǎn)被物種占據(jù)的動(dòng)態(tài)過(guò)程,其中包括滅絕和定居2個(gè)參數(shù),并引入調(diào)查時(shí)間、日期和氣溫3個(gè)協(xié)變量,根據(jù)ΔCAI選擇最佳模型估算占域趨勢(shì),結(jié)果顯示有2個(gè)物種呈下降趨勢(shì),春雨蛙(Pseudacriscrucifer)在6個(gè)州呈顯著下降趨勢(shì),美國(guó)蟾蜍(Bufoamericanus)在4個(gè)州呈顯著下降趨勢(shì),牛蛙(Ranacatesbeiana)在4個(gè)州呈顯著上升趨勢(shì)。IRVIN等[63]在2008、2009年的5—7月對(duì)342個(gè)樣點(diǎn)草蜢沙鹀(Grasshoppersparrow)每年實(shí)施3次重復(fù)調(diào)查,記錄探測(cè)史和調(diào)查時(shí)間、溫度、風(fēng)速、觀測(cè)者4個(gè)探測(cè)協(xié)變量,探測(cè)到的實(shí)際占域率為35%,并且由ΔCAI≤2得到影響探測(cè)率的最主要因素是一天中的調(diào)查時(shí)間和風(fēng)速。占域狀態(tài)不僅依賴(lài)于棲息環(huán)境,還與種群密度有關(guān)。通過(guò)分析影響草蜢沙鹀選擇和占據(jù)繁殖棲息地的因素,評(píng)估其空間尺度上棲息地的使用,從而加強(qiáng)對(duì)草蜢沙鹀所偏好棲息地的有效保護(hù)。ADAMS等[35]利用占域模型首次分析了美國(guó)監(jiān)測(cè)得到的兩棲動(dòng)物多年數(shù)據(jù),應(yīng)用多季節(jié)占域模型分析了兩棲動(dòng)物的年度占域率變化,發(fā)現(xiàn)所有兩棲動(dòng)物占域率在2002—2011年期間年均下降3.7%,被列入紅色名錄的物種平均每年下降11.6%,所有這些結(jié)果都顯示兩棲動(dòng)物種群呈衰減趨勢(shì),同時(shí),評(píng)估了兩棲動(dòng)物嚴(yán)重衰減的原因。ANDREW-TILKER[64]用紅外相機(jī)方法在2012年7月到2013年1月對(duì)中南半島熱帶森林4種受威脅的哺乳動(dòng)物進(jìn)行調(diào)查,并用占域模型分析物種出現(xiàn)的數(shù)據(jù),得到麂、獼猴、野豬和鬣羚的真實(shí)占域率分別為0.58、0.55、0.38和0.30,探測(cè)率分別為0.12、0.17、0.14和0.17。為了更好地解釋不完美的探測(cè)率,研究者將森林類(lèi)型和離村莊的距離作為樣點(diǎn)協(xié)變量應(yīng)用于模型,根據(jù)CAI值和模型權(quán)重確定了影響這些物種分布的因素,并根據(jù)最佳模型估算這些物種在不同森林類(lèi)型及離村莊不同距離的占域,從而為這些受威脅物種提供更有針對(duì)性的保護(hù)措施。NAGY-REIS等[65]調(diào)查了大西洋森林遺跡里的伶猴(Callicebusnigrifrons)和白耳狨(Callithrixaurita),用占域模型估算其占域率和探測(cè)率。

但是,該模型目前在國(guó)內(nèi)的研究并不多見(jiàn),一些學(xué)者將其與紅外相機(jī)技術(shù)相結(jié)合獲得一些研究結(jié)果。李勤等[3]將相機(jī)陷阱調(diào)查方法與占域模型分析方法相結(jié)合,研究野生動(dòng)物種群的空間分布,解決了有關(guān)地理范圍、復(fù)合種群動(dòng)態(tài)、資源選擇、生境選擇和種間關(guān)系等問(wèn)題。肖治術(shù)等[66]將占域模型作為一種統(tǒng)計(jì)分析模型,處理野生動(dòng)物多樣性的監(jiān)測(cè)數(shù)據(jù)。李晟等[4]將物種的占有率和探測(cè)率用作紅外相機(jī)調(diào)查或監(jiān)測(cè)項(xiàng)目中的重要參數(shù),為野生動(dòng)物調(diào)查與保護(hù)管理提供了更為便捷有效的統(tǒng)計(jì)分析方法。然而,占域模型在我國(guó)還未得到充分的推廣和應(yīng)用。

6 討論與展望

綜合上述國(guó)外學(xué)者目前對(duì)占域模型的研究成果,以及我國(guó)部分學(xué)者對(duì)其初步的探討及應(yīng)用,筆者認(rèn)為該模型具有巨大的應(yīng)用潛力。(1)占域模型所要求的數(shù)據(jù)類(lèi)型相對(duì)簡(jiǎn)單,只需要記錄每個(gè)樣點(diǎn)在每次調(diào)查中目標(biāo)物種是否出現(xiàn),而并不需要清楚地記錄每個(gè)樣點(diǎn)目標(biāo)物種的種群數(shù)量,因此,在很大程度上減少了野外調(diào)查中的資源投入(人力、物力、資金等)。這一點(diǎn)明顯優(yōu)于之前的各種調(diào)查方法,如標(biāo)志重捕法。占域率可作為評(píng)估動(dòng)物種群變化的一個(gè)度量標(biāo)準(zhǔn)[27],且只需花費(fèi)更少的努力就能收集到合適的探測(cè)/未探測(cè)數(shù)據(jù)。因此,這也是占域模型在實(shí)際應(yīng)用中最大的優(yōu)勢(shì)。(2)占域模型可與其他調(diào)查方法結(jié)合起來(lái)使用,如用紅外相機(jī)技術(shù)取得出現(xiàn)/未出現(xiàn)數(shù)據(jù),并記錄樣點(diǎn)棲息地特征和天氣狀況,然后,將其導(dǎo)入占域模型進(jìn)行分析,便可估計(jì)預(yù)測(cè)目標(biāo)物種在未來(lái)的分布變化、種群動(dòng)態(tài)趨勢(shì)、最適生存生境以及干擾程度對(duì)物種生存的影響等,從而為物種保護(hù)提供有針對(duì)性的措施。盡管占域模型的優(yōu)勢(shì)尤為顯著,同時(shí)也得到了初步應(yīng)用并取得良好效果,但在實(shí)際應(yīng)用中仍然存在不確定性問(wèn)題,如不完全的探測(cè)(imperfect detection)會(huì)影響占域率的估算。為獲得更準(zhǔn)確的估算結(jié)果,應(yīng)該全面考慮占域模型可能出現(xiàn)的限制因素,如棲息地環(huán)境、天氣狀況、調(diào)查人員等,將其作為協(xié)變量用于解釋不完全的探測(cè)問(wèn)題。

最初由MACKENZIE等[1]提出的占域模型中,關(guān)于物種探測(cè)率進(jìn)一步延伸的問(wèn)題并沒(méi)有得到解決,在調(diào)查數(shù)據(jù)中一個(gè)物種實(shí)際出現(xiàn)了但并沒(méi)有被探測(cè)到而被記錄為“0”的問(wèn)題(false negatives)[67]。如一個(gè)樣點(diǎn)經(jīng)過(guò)重復(fù)調(diào)查,獲得數(shù)據(jù)為“0,0,0”,就會(huì)存在2種互相排斥的可能性去解釋這個(gè)結(jié)果,一種是該物種在這個(gè)樣點(diǎn)真實(shí)的不存在,所以調(diào)查結(jié)果是準(zhǔn)確的;另一種是該物種實(shí)際上存在于該樣點(diǎn),但是并沒(méi)有被探測(cè)到(false absence)。這個(gè)問(wèn)題在野外物種調(diào)查中普遍存在,使研究者不能獲得占域的無(wú)偏估計(jì)。在不同的環(huán)境下,即使是極其完全的模型方案設(shè)計(jì)和標(biāo)準(zhǔn)的調(diào)查計(jì)劃,不完全的探測(cè)仍然是數(shù)據(jù)分析中真實(shí)存在的一個(gè)問(wèn)題[68-70]。當(dāng)然,也有學(xué)者在MACKENZIE等提出的模型基礎(chǔ)上致力于研究開(kāi)發(fā)更靈活的模型來(lái)解釋這種假陰性的錯(cuò)誤[71-74]。如KéRY[46]解釋了占域模型中不完全的探測(cè),從而將其應(yīng)用于瑞士一種蜻蜓的分布預(yù)測(cè)上,結(jié)果證明占域分布模型分析得到的物種分布比傳統(tǒng)的廣義線性模型(generalized linear models,GLMs)更接近于真實(shí),同時(shí),還介紹了如何減少在物種分布中導(dǎo)致偏差估計(jì)的不完全的探測(cè)問(wèn)題。另外,在很多情況下由于各種不利的因素,一些樣點(diǎn)的調(diào)查可能無(wú)法實(shí)施,如調(diào)查人員在抵達(dá)樣點(diǎn)的途中遇到意外情況(汽車(chē)拋錨、突降暴雨、人員受傷等),受到外部條件的限制而被迫減少重復(fù)調(diào)查次數(shù),危險(xiǎn)地段(懸崖、滑坡等)難以到達(dá)。在上述情況下,就會(huì)導(dǎo)致模型分析中觀測(cè)值的缺失(missing observations)[75]。占域模型也適用于處理部分調(diào)查數(shù)據(jù)缺失的情況,但是樣點(diǎn)的數(shù)據(jù)采集必須有多次重復(fù)。如果每個(gè)樣點(diǎn)僅僅開(kāi)展過(guò)一次調(diào)查或無(wú)任何調(diào)查記錄,則不能用于模型估算。

關(guān)于占域模型的研究還在進(jìn)一步發(fā)展中,模型的功能也在逐步改進(jìn)和完善,預(yù)計(jì)其應(yīng)用也會(huì)越來(lái)越廣泛。今后,應(yīng)進(jìn)一步探討如何將占域模型應(yīng)用于我國(guó)生物多樣性調(diào)查和監(jiān)測(cè)計(jì)劃中,為生物多樣性保護(hù)和管理提供科學(xué)支撐。

[1] MACKENZIE D I,NICHOLS J D,LACHMAN G B,etal.Estimating Site Occupancy Rates When Detection Probabilities Are Less Than One[J].Ecology,2002,83(8):2248-2255.

[2] MACKENZIE D I,NICHOLS J D,HINES J E,etal.Estimating Site Occupancy,Colonization,and Local Extinction When a Species Is Detected Imperfectly[J].Ecology,2008,84(8):2200-2207.

[3] 李勤,鄔建國(guó),寇曉軍,等.相機(jī)陷阱在野生動(dòng)物種群生態(tài)學(xué)中的應(yīng)用[J].應(yīng)用生態(tài)學(xué)報(bào),2013,24(4):947-955.[LI Qin,WU Jian-guo,KOU Xiao-jun,etal.Applications of Camera Trap in Wildlife Population Ecology[J].Chinese Journal of Applied Ecology,2013,24(4):947-955.]

[4] 李晟,王大軍,肖治術(shù),等.紅外相機(jī)技術(shù)在我國(guó)野生動(dòng)物研究與保護(hù)中的應(yīng)用與前景[J].生物多樣性,2014,22(6):685-695.[LI Sheng,WANG Da-jun,XIAO Zhi-shu,etal.Camera-Trapping in Wildlife Research and Conservation in China:Review and Outlook[J].Biodiversity Science,2014,22(6):685-695.]

[5] DE SOLLA S R,SHIROSE L J,FERNIE K J,etal.Effect of Sampling Effort and Species Detectability on Volunteer Based Anuran Monitoring Programs[J].Biological Conservation,2005,121(4):585-594.

[6] CROSSLAND M R,MACKENZIE D I,HOLZAPFEL S.Assessment of Site-Occupancy Modeling as a Technique to Monitor Hochstetter′s Frog(Leiopelmabocbstetteri) Populations[J].DOC Research & Development,2005,218:223.

[7] O′CONNELL A F,BAILEY L L.Inference for Occupancy and Occupancy Dynamics[M]∥Camera Traps in Animal Ecology.Japan:Springer,2011:191-204.

[8] STANTON R A,THOMPSON F R,KESLER D C.Site Occupancy of Brown-Headed Nuthatches Varies With Habitat Restoration and Range-Limit Context[J].The Journal of Wildlife Management,2015,79(6):917-926.

[9] GOTELLI N J,ENGSTROM R T.Predicting Species Occurrences:Issues of Accuracy and Scale[J].The Auk,2003,120(4):1199-1200.

[10]HANSKI I.Inferences From Ecological Incidence Functions[J].The American Naturalist,1992,139(3):657-662.

[11]FISHER R N,SHAFFER H B.The Decline of Amphibians in California′s Great Central Valley[J].Conservation Biology,1996,10(5):1387-1397.

[12]VAN BUSKIRK J.Local and Landscape Influence on Amphibian Occurrence and Abundance[J].Ecology,2005,86(7):1936-1947.

[13]HAMES R S,ROSENBERG K V,LOWE J D,etal.Site Reoccupation in Fragmented Landscapes:Testing Predictions of Metapopulation Theory[J].Journal of Animal Ecology,2001,70(2):182-190.

[14]BARBRAUD C,NICHOLS J D,HINES J E,etal.Estimating Rates of Local Extinction and Colonization in Colonial Species and an Extension to the Metapopulation and Community Levels[J].Oikos,2003,101(1):113-126.

[16]LANDE R.Extinction Thresholds in Demographic Models of Territorial Populations[J].The American Naturalist,1987,130(4):624-635.

[17]LANDE R.Demographic Models of the Northern Spotted Owl (Strixoccidentaliscaurina)[J].Oecologia,1988,75(4):601-607.

[18]HANSKI I.A Practical Model of Metapopulation Dynamics[J].Journal of Animal Ecology,1994,63(1):151-162.

[19]MANLEY P N,ZIELINSKI W J,SCHLESINGER M D,etal.Evaluation of a Multiple-Species Approach to Monitoring Species at the Ecoregional Scale[J].Ecological Applications,2004,14(1):296-310.

[20]BAILEY L L,HINES J E,NICHOLS J D,etal.Sampling Design Trade-Offs in Occupancy Studies With Imperfect Detection:Examples and Software[J].Ecological Applications,2007,17(1):281-290.

[21]STANLEY T R,ROYLE J A.Estimating Site Occupancy and Abundance Using Indirect Detection Indices[J].The Journal of Wildlife Management,2005,69(3):874-883.

[22]BAILEY L L,SIMONS T R,POLLOCK K H.Estimating Site Occupancy and Species Detection Probability Parameters for Terrestrial Salamanders[J].Ecological Applications,2004,14(3):692-702.

[23]GUILLERA-ARROITA G.Impact of Sampling With Replacement in Occupancy Studies With Spatial Replication[J].Methods in Ecology and Evolution,2011,2(4):401-406.

[24]MACKENZIE D I,ROYLE J A.Designing Occupancy Studies:General Advice and Allocating Survey Effort[J].Journal of Applied Ecology,2005,42(6):1105-1114.

[25]BAILEY L,ADAMS M J.Occupancy Models to Study Wildlife[R].[s. l.]:Geological Survey(US),2005.

[26]MACKENZIE D I,BAILEY L L.Assessing the Fit of Site-Occupancy Models[J].Journal of Agricultural,Biological,and Environmental Statistics,2004,9(3):300-318.

[27]GU W D,SWIHART R K.Absent or Undetected? Effects of Non-Detection of Species Occurrence on Wildlife-Habitat Models[J].Biological Conservation,2004,116(2):195-203.

[28]LI S,MCSHEA W J,WANG D J,etal.Gauging the Impact of Management Expertise on the Distribution of Large Mammals Across Protected Areas[J].Diversity and Distributions,2012,18(12):1166-1176.

[29]LELE S R,MORENO M,BAYNE E.Dealing With Detection Error in Site Occupancy Surveys:What Can We Do With a Single Survey?[J].Journal of Plant Ecology,2012,5(1):22-31.

[30]MARTIN J,MCINTYRE C L,HINES J E,etal.Dynamic Multistate Site Occupancy Models to Evaluate Hypotheses Relevant to Conservation of Golden Eagles in Denali National Park,Alaska[J].Biological Conservation,2009,142(11):2726-2731.

[31]GOOCH M M,HEUPEL A M,PRICE S J,etal.The Effects of Survey Protocol on Detection Probabilities and Site Occupancy Estimates of Summer Breeding Anurans[J].Applied Herpetology,2006,3(2):129-142.

[32]DORAZIO R M,RODRGUEZ D T.A Gibbs Sampler for Bayesian Analysis of Site-Occupancy Data[J].Methods in Ecology and Evolution,2012,3(6):1093-1098.

[33]LI S,MCSHEA W J,WANG D J,etal.The Use of Infrared-Triggered Cameras for Surveying Phasianids in Sichuan Province,China[J].Ibis,2010,152(2):299-309.

[34]SUTHERLAND C S,ELSTON D A,LAMBIN X.A Demographic,Spatially Explicit Patch Occupancy Model of Metapopulation Dynamics and Persistence[J].Ecology,2014,95(11):3149-3160.

[35]ADAMS M J,MILLER D A,MUTHS E,etal.Trends in Amphibian Occupancy in the United States[J].PLoS One,2013,8(5):e64347.

[36]FAUTEUX D,MAZEROLLE M J,IMBEAU L,etal.Site Occupancy and Spatial Co-Occurrence of Boreal Small Mammals Are Favored by Late-Decay Woody Debris[J].Canadian Journal of Forest Research,2013,43(5):419-427.

[37]O′CONNELL A F JR,TALANCY N W,BAILEY L L,etal.Estimating Site Occupancy and Detection Probability Parameters for Meso- and Large Mammals in a Coastal Ecosystem[J].The Journal of Wildlife Management,2009,70(6):1625-1633.

[38]SIGNORELLI L,BASTOS R P,MARCO P D JR,etal.Landscape Context Affects Site Occupancy of Pond-Breeding Anurans Across a Disturbance Gradient in the Brazilian Cerrado[J].Landscape Ecology,2016,31(9):1997-2012.

[39]AKAIKE H.Information Theory and an Extension of the Maximum Likelihood Principle[M]∥Breakthroughs in Statistics.New York,USA:Springer,1992:610-624.

[40]BURNHAM K P,ANDERSON D R.Model Selection and Inference[M].New York,USA:Springer,1998:319.

[41]BURNHAN K P,ANDERSON D R.Model Selection and Multimodel Inference:A Practical Information-Theoretic Approach[M].New York,USA:Springer Science & Business Media,2002:455-484.

[42]MCSHEA W J,STEWART C,PETERSON L,etal.The Importance of Secondary Forest Blocks for Terrestrial Mammals Within an Acacia/ Secondary Forest Matrix in Sarawak,Malaysia[J].Biological Conservation,2009,142(12):3108-3119.

[43]ROYLE J A.Site Occupancy Models With Heterogeneous Detection Probabilities[J].Biometrics,2006,62(1):97-102.

[44]ALTWEGG R,WHEELER M,ERNI B.Climate and the Range Dynamics of Species With Imperfect Detection[J].Biology Letters,2008,4(5):581-584.

[46]KéRY M.Chapter 20-Nonstandard GLMMs 1:Site-Occupancy Species Distribution Model[M]∥Introduction to WinBUGS for Ecologists.Salt Lake City,UT,USA:Academic Press,2010:237-252.

[47]ANDRESEN L,EVERATT K T,SOMERS M J.Use of Site Occupancy Models for Targeted Monitoring of the Cheetah[J].Journal of Zoology,2014,292(3):212-220.

[48]KéRY M,GARDNER B,MONNERAT C.Predicting Species Distributions From Checklist Data Using Site-Occupancy Models[J].Journal of Biogeography,2010,37(10):1851-1862.

[49]YOCCOZ N G,NICHOLS J D,BOULINIER T.Monitoring of Biological Diversity in Space and Time[J].Trends in Ecology & Evolution,2001,16(8):446-453.

[50]KéRY M.Inferring the Absence of a Species:A Case Study of Snakes[J].The Journal of Wildlife Management,2002,66(2):330-338.

[51]WILLIAMS B K,NICHOLS J D,CONROY M J.Analysis and Management of Animal Populations:Modeling Estimation,and Decision Making[J].Journal of Wildlife Management,2002,67:1172-1181.

[52]KéRY M,SCHMIDT B R.Imperfect Detection and Its Consequences for Monitoring in Conservation[J].Community Ecology,2008,9(2):207-216.

[53]KéRY M,SPILLMANN J H,TRUONG C,etal.How Biased Are Estimates of Extinction Probability in Revisitation Studies?[J].Journal of Ecology,2006,94(5):980-986.

[54]CHEN G K,KéRY M,ZHANG J L,etal.Factors Affecting Detection Probability in Plant Distribution Studies[J].Journal of Ecology,2009,97(6):1383-1389.

[55]NUPP T E,SWIHART R K.Effect of Forest Patch Area on Population Attributes of White-Footed Mice (Peromyscusleucopus) in Fragmented Landscapes[J].Canadian Journal of Zoology,1996,74(3):467-472.

[56]ODELL E A,KNIGHT R L.Songbird and Medium-Sized Mammal Communities Associated With Exurban Development in Pitkin County,Colorado[J].Conservation Biology,2001,15(4):1143-1150.

[57]GERMAINE S S,COSENTINO B L.Screening Model for Determining Likelihood of Site Occupancy by Oregon Spotted Frogs (Ranapretiosa)in Washington State[R].Washington DC,USA:Washington State Department of Transportation,2004.

[58]FUKAYA K,ROYLE J A,OKUDA T,etal.A Multistate Dynamic Site Occupancy Model for Spatially Aggregated Sessile Communities[J].Methods in Ecology and Evolution,2016.DOI:10.1111/2041-210X.12690.

[59]ELMORE S A,HUYVAERT K P,BAILEY L L,etal.Multi-Scale Occupancy Approach to EstimateToxoplasmagondiiPrevalence and Detection Probability in Tissues:An Application and Guide for Field Sampling[J].International Journal for Parasitology,2016,46(9):563-570.

[60]MACKENZIE D I,NICHOLS J D,ROYLE J A,etal.Occupancy Estimation and Modeling:Inferring Patterns and Dynamics of Species Occurrence[M].Burlington,MA,USA:Academic Press,2005.

[61]ALBANESE B,PETERSON J T,FREEMAN B J,etal.Accounting for Incomplete Detection When Estimating Site Occupancy of Bluenose Shiner (Pteronotropiswelaka) in Southwest Georgia[J].Southeastern Naturalist,2007,6(4):657-668.

[62]WEIR L A,FISKE I J,ROYLE J A.Trends in Anuran Occupancy From Northeastern States of the North American Amphibian Monitoring Program[J].Herpetological Conservation and Biology,2009,4(3):389-402.

[63]IRVIN E,DUREN K R,BULER J J,etal.A Multi-Scale Occupancy Model for the Grasshopper Sparrow in the Mid-Atlantic[J].The Journal of Wildlife Management,2013,77(8):1564-1571.

[64]ANDREW-TILKER B S.Estimating Site Occupancy for Four Threatened Mammals in Southeastern Laos[D].Austin,USA:The University of Texas at Austin,2014.

[65]NAGY-REIS M B,NICHOLS J D,HINES J E,etal.Estimating Occupancy and Detection Probability of Callicebus Nigrifrons and Callithrix Aurita From Call Surveys in a Large Atlantic Forest Remnant[C]∥Meeting of the American Society of Primatologists.[s. l.]:[s. n.],2015:72-73.

[66]肖治術(shù),李欣海,王學(xué)志,等.探討我國(guó)森林野生動(dòng)物紅外相機(jī)監(jiān)測(cè)規(guī)范[J].生物多樣性,2014,22(6):704-711.[XIAO Zhi-shu,LI Xin-hai,WANG Xue-zhi,etal.Developing Camera-Trapping Protocols for Wildlife Monitoring in Chinese Forests[J].Biodiversity Science,2014,22(6):704-711.]

[67]ROYLE J A,LINK W A.Generalized Site Occupancy Models Allowing for False Positive and False Negative Errors[J].Ecology,2006,87(4):835-841.

[68]BROWN W S,KéRY M,HINES J E.Survival of Timber Rattlesnakes (Crotalushorridus) Estimated by Capture-Recapture Models in Relation to Age,Sex,Color Morph,Time,and Birthplace[J].Copeia,2007(3):656-671.

[69]HOCHACHKA W M,FIEDLER W.Trends in Trap Ability and Stop-Over Duration Can Confound Interpretations of Population Trajectories From Long-Term Migration Ringing Studies[J].Journal of Ornithology,2008,149(3):375-391.

[70]ROTA C T,WIKLE C K,KAYS R W,etal.A Two-Species Occupancy Model Accommodating Simultaneous Spatial and Interspecific Dependence[J].Ecology,2016,97(1):48-53.

[71]BAYLEY P B,PETERSON J T.An Approach to Estimate Probability of Presence and Richness of Fish Species[J].Transactions of the American Fisheries Society,2001,130(4):620-633.

[72]ROYLE J A,NICHOLS J D.Estimating Abundance From Repeated Presence-Absence Data or Point Counts[J].Ecology,2003,84(3):777-790.

[73]TYRE A J,TENHUMBERG B,FIELD S A,etal.Improving Precision and Reducing Bias in Biological Surveys:Estimating False-Negative Error Rates[J].Ecological Applications,2003,13(6):1790-1801.

[74]FISKE I J,ROYLE J A,GROSS K.Inference for Finite-Sample Trajectories in Dynamic Multi-State Site-Occupancy Models Using Hidden Markov Model Smoothing[J].Environmental and Ecological Statistics,2014,21(2):313-328.

[75]WEIR L A,ROYLE J A,NANJAPPA P,etal.Modeling Anuran Detection and Site Occupancy on North American Amphibian Monitoring Program (NAAMP) Routes in Maryland[J].Journal of Herpetology,2009,39(4):627-639.

(責(zé)任編輯： 李祥敏)

Site Occupancy Model: Principles and Research Progress.

WANYa-qiong1,GUOWei-bo2,WUJun2,LILi1,XUHai-gen2

(1.School of Life Sciences, Nanjing Normal University, Nanjing 210046, China; 2.Nanjing Institute of Environmental Sciences, Ministry of Environmental Protection, Nanjing 210042, China)

The site occupancy model, developed by Mackenzie in 2002, has been used to estimate proportion of an area occupied by target species for further efforts to estimate abundance of the species, predict range of its distribution and learn its community structure. A review is oriented to introduce basic principles of the occupancy model, including data type, basic parameters, model selection and application scope, and also summarize progresses of the researches of the model and its application. However, so far, the studies on the model and its application are mainly credited to foreign scholars, though in recent years, some researchers in China have also set out to try the model in the field of surveys and monitoring. On such a basis, some problems that may pop up in the actual application of the model have been analyzed, such as introduction of covariates, heterogeneity of detectivity, imperfectness of observations, lack of observations, etc. and the causes of these problems have been explained. In the end, proposals for further perfecting the model and applying it to biodiversity surveying and monitoring programs in China have been brought forth, and hence to provide some effective technical support for biodiversity protection.

site occupancy model; occupancy rate; detectivity; covariate

2016-07-29

環(huán)境保護(hù)部生物多樣性保護(hù)專(zhuān)項(xiàng)(2111101)

Q14

A

1673-4831(2017)08-0673-07

10.11934/j.issn.1673-4831.2017.08.001

萬(wàn)雅瓊(1992—),女,湖北宜昌人,碩士生,主要研究方向?yàn)樯锒鄻有员Ｗo(hù)。E-mail:qiong1273692100@163.com

①通信作者E-mail: xhg@nies.org