物種分布模型

Species distribution models (SDM; for review and definition see, e.g., Peterson et al., 2011) are a dominant paradigm to quantify the relationship between environmental dynamics and several manifestations of species biogeography. These statistical approaches pushed an emerging body of research describing the global distribution of species, addressing niche-based questions, supporting biodiversity conservation and ecosystem-based management, as well as infering the likely anthropogenic pressures leading to population turnover and extinction.

物種分布模型(SDM；有關(guān)審查和定義，請參見例如Peterson等，2011)是量化環(huán)境動態(tài)與物種生物地理學(xué)幾種表現(xiàn)之間關(guān)系的主要范例。 這些統(tǒng)計方法推動了一個新興的研究機(jī)構(gòu)，它描述了物種的全球分布，解決了基于生態(tài)位的問題，支持生物多樣性保護(hù)和基于生態(tài)系統(tǒng)的管理，并推斷出可能導(dǎo)致人口流動和滅絕的人為壓力。

Spatial autocorrelation (SA) is a common challenge while modelling the distribution and abundance of species. This phenomenon, likely present in most ecological datasets, denotes the situation where the values of variables sampled at nearby locations are not independent due to correlation with values at nearby locations (i.e., the value of a predictor variable at a given site can be partially predicted by the values at neighbouring sites).

在對物種的分布和豐富度進(jìn)行建模時，空間自相關(guān)(SA)是一個普遍的挑戰(zhàn)。 這種現(xiàn)象很可能出現(xiàn)在大多數(shù)生態(tài)數(shù)據(jù)集中，表示這樣一種情況：由于與附近位置的值相關(guān)，因此在附近位置采樣的變量的值不是獨(dú)立的(即，可以對給定位置的預(yù)測變量的值進(jìn)行部分預(yù)測)根據(jù)相鄰站點(diǎn)的值)。

Accounting for SA has not received much attention in applied SDM studies, however, when present, it may result in poorly specified models and inappropriate spatial inference and prediction. Recent studies proposed to incorporate SA into the actual models while predicting distributions (coined ‘spatial models’; Dormann, 2007), however, this approach does not allow to transfer models to new independent data (e.g., temporal and spatial transferability).

在應(yīng)用SDM研究中，對SA的會計處理并未引起太多關(guān)注，但是，如果存在SA會計，可能會導(dǎo)致指定的模型不正確以及空間推斷和預(yù)測不適當(dāng)。 最近的研究提出在預(yù)測分布的同時將SA合并到實(shí)際模型中(硬幣化的“空間模型”； Dormann，2007年)，但是，這種方法不允許將模型轉(zhuǎn)移到新的獨(dú)立數(shù)據(jù)中(例如，時間和空間轉(zhuǎn)移性)。

I propose a straightforward approach to reduce the effect of SA in SDM (see also Boavida et al., 2016 for more details). I use a simple example bellow focused on a brown algae species capable of producing marine forests and a set of environmental predictors known to largely explain its distribution.

我提出了一種直接的方法來減少SA在SDM中的影響(更多信息，另請參閱Boavida等人，2016)。 我用一個簡單的例子來說明波紋管，它著重于能夠生產(chǎn)海洋森林的褐藻物種，以及一組已知的環(huán)境預(yù)測因子，可以在很大程度上解釋其分布。

Get the R code: Reducing spatial autocorrelationhttps://github.com/jorgeassis/spatialAutocorrelationFig. Initial set of occurrence records with potential negative effect of spatial autocorrelation.圖：初始記錄集，具有空間自相關(guān)的潛在負(fù)面影響。

1. A correlogram is produced to assess the correlation of each variable predictor within a range of geographic distances.

1.產(chǎn)生相關(guān)圖以評估地理距離范圍內(nèi)每個變量預(yù)測變量的相關(guān)性。

2. For each distance class, a linear model tests the effect of correlation with geographic distance. This finds the minimum non-significant autocorrelated distance.

2.對于每個距離類別，線性模型測試與地理距離的相關(guān)性影響。 這將找到最小的不重要的自相關(guān)距離。

Fig. Correlogram of a variable predictor within a range of distances (Open circles: non-significant correlation).圖。在距離范圍內(nèi)的變量預(yù)測變量的相關(guān)圖(空心圓：非顯著相關(guān)性)。

3. The average of the minimum non-significant distances found per variable is used to prune the occurrence records, by leaving only one record within such distance.

3.每個變量找到的最小非有效距離的平均值用于修剪出現(xiàn)記錄，方法是在該距離內(nèi)僅保留一個記錄。

Fig. Minimum and average (dashed line) non-significant autocorrelated distances of variable predictors.圖。可變預(yù)測變量的最小和平均(虛線)非重要自相關(guān)距離。 Fig. Final pruned dataset of occurrence records with reduced potential effect of spatial autocorrelation.圖。最終修剪的事件記錄數(shù)據(jù)集，其空間自相關(guān)的潛在影響降低。 Get the R code: Reducing spatial autocorrelationhttps://github.com/jorgeassis/spatialAutocorrelation

引用文獻(xiàn) (Literature cited)

C. F. Dormann (2007) Effects of incorporating spatial autocorrelation into the analysis of species distribution data. Global Ecology and Biogeography. 16, 129–138.

CF Dormann(2007)將空間自相關(guān)納入物種分布數(shù)據(jù)分析的影響。 全球生態(tài)與生物地理。 16、129–138。

Peterson, A.T., Soberon, J., Pearson, R.G., Anderson, R.P., Martinez-Meyer, E., Nakamura, M. &Araujo, M.B. (2011) Ecological niches and geographic distributions. Monographs in Population Biology, 314 pp. Vol. 49. Princeton University Press, Princeton.

Peterson，AT，Soberon，J.，Pearson，RG，Anderson，RP，Martinez-Meyer，E.，Nakamura，M.＆Araujo，MB(2011)生態(tài)位和地理分布。 專著《人口生物學(xué)》 ，第314頁，第一卷。 49.普林斯頓大學(xué)出版社，普林斯頓。

Boavida, J., Assis, J., Silva, I. et al. (2016) Overlooked habitat of a vulnerable gorgonian revealed in the Mediterranean and Eastern Atlantic by ecological niche modelling. Scientific Reports. 6, 36460.

Boavida，J.，Assis，J.，Silva，I。 等。 (2016)通過生態(tài)位建模在地中海和東大西洋發(fā)現(xiàn)了易受攻擊的高哥人的棲息地。 科學(xué)報告。 6，36460。

翻譯自: https://medium.com/themarinedatascientist/reducing-spatial-autocorrelation-in-species-distribution-models-fe84d4269cee

物種分布模型

總結(jié)

以上是生活随笔為你收集整理的物种分布模型_减少物种分布建模中的空间自相关的全部內(nèi)容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網(wǎng)站內(nèi)容還不錯，歡迎將生活随笔推薦給好友。