Global Environmental Change 58 (2019) 101974
Contents lists available at ScienceDirect
Global Environmental Change journal homepage: www.elsevier.com/locate/gloenvcha
Indigenous perceptions of climate anomalies in Malaysian Borneo a,⁎
b
c
d
b
b
c
T. van Gevelt , H. Abok , M.M. Bennett , S.D. Fam , F. George , N. Kulathuramaiyer , C.T. Low , T. Zamane
T
a
Department of Politics and Public Administration, University of Hong Kong, Hong Kong SAR Institute of Social Informatics and Technological Innovations, Universiti Malaysia Sarawak, Malaysia Department of Geography, University of Hong Kong, Hong Kong SAR d Fenner School of Environment and Society, Australian National University, Australia e School of Computing and Creative Media, University College of Technology Sarawak, Malaysia b c
ARTICLE INFO
ABSTRACT
Keywords: Climate change Climate anomalies Indigenous communities Perceptions Adaptation Sarawak
Local perceptions of climate anomalies influence adaptation behaviour. Specifically, perceptions that are more accurate and homogenous at the community-level are more likely to facilitate the collective action required to adapt to the local effects of climate anomalies experienced by many indigenous communities. We combine primary data on perceptions of climate anomalies from 200 individuals in six Penan villages in Sarawak, Malaysia with instrumental climate data. We find that perceptions of climate anomalies vary substantially in terms of occurrence and magnitude, and do not generally correlate with instrumental climate data. We operationalise the Penan forest sign language (Oroo’) as a measure of traditional ecological knowledge (TEK) and find only weak evidence of a systematic statistical association with perceptions of climate anomalies among our sampled respondents. Our findings suggest caution in advancing adaptation strategies in indigenous communities that are predominantly premised on TEK. Instead, our findings suggest that in designing adaptation measures, indigenous communities may benefit by engaging in forums where community members and external stakeholders can come together, share their perceptions and observations of climate change, and reach a collective consensus on the community-level effects of climate change and pathways towards adaptation.
1. Introduction Many indigenous communities are among the most vulnerable to the impacts of climate change (Salick and Ross, 2009; Ford et al., 2018). With livelihoods that are often highly dependent on the environment, the increased frequency and severity of climate anomalies can significantly affect indigenous communities (Pyahala et al., 2016; Savo et al., 2016; van Gevelt, 2019). Recognising that many indigenous communities have continuously adapted to environmental change over their histories, there is an emerging consensus that empowering indigenous communities to leverage their traditional ecological knowledge (TEK) may be key to their successful adaptation to the local impacts of climate change (Bridges and McClatchey, 2009; Byg and Salick, 2009; Turner and Clifton, 2009). At the same time, there is a growing literature exploring local perceptions of climate change and the pathways through which perceptions influence adaptation behaviour. This literature provides compelling evidence suggesting that local perceptions of climate change – usually measured as extremes or anomalies – shape adaptation
⁎
behaviour through a number of different mechanisms, such as a Bayesian updating process or through salience effects (Hansen et al., 2012; Howe et al., 2012; Deryugina, 2013; Howe and Leisworowitz, 2013; Lee et al., 2015; Demski et al., 2017; Zanocco et al., 2018). One of the key messages emerging from this literature is the importance of understanding local perceptions of climate change in order to devise appropriate climate adaptation policy responses (Larcom et al., 2019). Combining the insights from these two strands of literature, we suggest that a detailed understanding of indigenous perceptions of climate change is essential in order to assess the potential for indigenous communities – with the appropriate enabling environment – to adapt to the local impacts of climate change (Berkes and Jolly, 2001; Alessa et al., 2008; Boissiere et al., 2013). Specifically, we submit that there are at least three major questions that need to be addressed in further detail. Firstly, to what extent are perceptions of climate change homogenous within and among contiguous indigenous communities? This is important given that collective action is required to adapt to the effects of climate change at the community-level (Adger, 2003; Ostrom, 2010). Secondly, to what extent are indigenous perceptions broadly
Corresponding author at: 8.04, Jockey Club Tower, Centennial Campus, Pokfulam Road, Hong Kong, Hong Kong SAR. E-mail address:
[email protected] (T. van Gevelt).
https://doi.org/10.1016/j.gloenvcha.2019.101974 Received 29 April 2019; Received in revised form 17 July 2019; Accepted 21 August 2019 0959-3780/ © 2019 Elsevier Ltd. All rights reserved.
Global Environmental Change 58 (2019) 101974
T. van Gevelt, et al.
accurate? This is important in order to understand to what extent perceptions may be affected by memory illusions (Daw, 2010), change blindness (Simons and Rensik, 2005; Alessa et al., 2008) or shifting baseline syndrome (Papworth et al., 2009; Fernandez-Llamazares et al., 2015). Thirdly, how does TEK affect perceptions of climate change? It is important to unravel the ‘black box’ of TEK and better understand how it affects indigenous perceptions of climate change (Pyahala et al., 2016). This is particularly important given the dynamic nature of TEK and the difficulties that many indigenous communities face in transmitting their TEK to younger generations (van Gevelt, 2019). While there is a substantial body of literature examining indigenous perceptions of climate change, review articles by Pyahala et al. (2016) and Savo et al. (2016) highlight a number of significant gaps in our understanding. Among these gaps is a geographical bias of studies, with the majority of studies being focused on indigenous communities in deserts and polar regions. There is comparatively less work undertaken in less ‘extreme’ environments, such as tropical rainforests, where we may expect the effects of climate change to be more subtle (Corlett and Primack, 2011). A second significant shortcoming highlighted by Pyahala et al. (2016) is a lack of methodological rigour. For example, a substantial proportion of reviewed studies that make use of primary data – both qualitative and quantitative – do not state their methodological approach (e.g. sample size, questions asked, sample time-frame) in enough detail to contextualise or assess the robustness of findings. We contribute to the literature by engaging with the three main questions mentioned above. Firstly, we use a novel co-designed data collection instrument to understand perceptions of climate anomalies from 200 individuals in six Penan communities in Sarawak, Malaysia. We do this to understand the extent to which perceptions are homogenous within and among contiguous indigenous communities. Secondly, we estimate to what extent indigenous perceptions correspond with instrumental climate data (a highly qualified measure of the accuracy of perceptions). We do this by comparing perceptions of climate anomalies with a statistical measure of climate anomalies calculated using gridded climate data. Thirdly, we use Poisson regression analysis to isolate the marginal effect of varying levels of TEK on perceptions of climate anomalies. The remainder of our paper is as follows. In Section 2, we provide an overview of climate change and indigenous communities in Sarawak, Malaysia, with a focus on the Penan. In Section 3, we present our data and empirical methods. In Section 4, we present our main results. In Section 5, we discuss our results and their significance in terms of their contribution to the literature and to climate change adaptation policy. We conclude in Section 6.
in relation to climate change. Focusing on the Kenyah Badeng, one of Sarawak's indigenous groups, they use ethnobiological methods to document traditional practices used by the Kenyah Badeng to predict the climate. While there is little systematic understanding of Dayak perceptions of climate change, documented anecdotal evidence gathered from indigenous community participants at the eBorneo Knowledge Fair in 2015 suggests that there is a broad consensus among indigenous communities that climate change is affecting rainfall patterns resulting in increasingly frequent and severe cases of flooding and drought, and making the temperature more erratic and unpredictable (Harris, 2017). The Penan are one of the main indigenous groups in Sarawak. Traditionally forest nomads, the Penan are famed for their intimate knowledge of the Borneo rainforest. For example, Koizumi and Momose (2007) estimate that the Penan have over 200 sources of food in the rainforest, as well as over 300 different types of construction materials and over 90 different kinds of medicinal herbs. Koizumi and Momose (2007) and Zaman et al. (2013) suggest that the Penan's knowledge and understanding of the Borneo rainforest is unique and unparalleled among Sarawak's indigenous groups. The unparalleled knowledge and understanding of the rainforest by the Penan is widely acknowledged by other Dayak groups, such as the Iban and Kelabit (Hansen, 1988). The centrality of the rainforest to the Penan is reflected in many ways. Perhaps most illustrative is the concept of Toro. Effectively a ‘journey’ that younger males must undertake to be considered Penan, Toro sees Penan elders transmit their intimate knowledge of the forest and their values of forest stewardship to young males so that they can gain the skills and knowledge necessary to forage and hunt for food, erect accommodation and locate potable water resources in the rainforest (Siew et al., 2013; Zaman et al., 2013; van Gevelt, 2019). The importance of Toro is arguably best summed up by the Penan adage: ‘if you do not know the forest, you are not a Penan.’ As of 2010, national records indicate that 77% of Sarawak's 16,000 Penan were settled in permanent villages, with 20% maintaining seminomadic lifestyles and 3% continuing to be nomadic (Lyndon et al., 2013). Among the settled and semi-nomadic Penan, the dominant livelihood activity now tends to be agriculture, primarily rice paddy farming, although the Penan continue to rely deeply on products from the rainforest for a substantial proportion of their food, as well as for medicine, construction materials and for non-timber forest products of high commercial value (Donovan and Puri, 2004; Siew et al., 2013). In addition to the challenges inherent in transitioning from nomadic to settled livelihoods, the Penan face numerous challenges as a result of modernisation and economic development in Sarawak. This is most visible in the significant migration pressures placed upon the youth who tend to move to the cities of Miri and Kuching due to better economic opportunities and the allure of urban life (Brosius, 2006). There is anecdotal evidence that both the transition to a mostly settled livelihood and the pressures of modernisation and outmigration are leading to the loss of the Penan's TEK. For example, Zaman et al. (2013) and van Gevelt (2019) note that the Penan youth are increasingly disinterested in TEK.
2. Climate change and indigenous communities in Sarawak Approximately one million people (around 40% of Sarawak's population) are indigenous.1 Collectively known as Dayak, many of Sarawak's indigenous population continue to live in villages deep in the interior despite substantial out-migration to nearby towns and cities (Brosius, 2006; Minority Rights Group International, 2018). Many Dayak communities face common problems of low economic, health and education outcomes, disenfranchisement from the political process, and the encroachment of customary land for logging and the construction of mega-dams (Sovacool and Valentine, 2011; Sovacool and Bulan, 2013; Lasimbang, 2015; Fam, 2017; van Gevelt et al., 2017). Extant studies on the Dayak have largely focused on the role of TEK in agro-forestry and forest management (Jessup and Vayda, 1988; Joshi et al., 2004; Weihreter, 2014; Siahaya et al., 2016). To the best of our knowledge, only Garay-Barayazarra and Puri (2011) have studied TEK
3. Methods 3.1. Study sites Our study sites consisted of six Penan villages located in the Totoh Apoh area of the Baram region of Sarawak: Long Win, Long Latei, Long Leng, Long Kerangan, Long Jenalong and Long Urang (see Fig. 1). Recent infrastructure developments mean that all six villages are now accessible by 4WD with the journey taking around one hour from the nearest town, Long Bedian, which itself is now only a four-hour drive on logging roads from Miri, the nearest city. Populations in our study sites ranged from 40 households (150 individuals) in Long Win to 89 households (340 individuals) in Long Latei.
1 The legal definition of indigeneity in the context of Malaysia and Sarawak has undergone numerous shifts in the years following British colonial rule in the 19th Century. For a brief history of the term, see Nelson et al. (2016).
2
Global Environmental Change 58 (2019) 101974
T. van Gevelt, et al.
Fig. 1. Study sites.
The shift to semi-nomadic and permanent lifestyles occurred in the 1950s for Long Win and Long Leng, and in the 1970s for Long Latei, Long Jenalong, Long Kerangan and Long Urang. The shift away from nomadic lifestyles was driven largely by the Borneo Evangelical Mission (now known as the Sidang Ingil Borneo), a Protestant denomination founded by Australian missionaries in 1928. The Penan were introduced to Christianity in the early 1950s by members of the Kenyah indigenous group, who had themselves come into direct contact with missionaries in the 1950s. Many Penan were responsive to Christianity and this led to change in some significant elements of their traditional belief system and led to their shift to semi-nomadic and permanent lifestyles.2 While initial contact with Christianity was through the Kenyah in the Upper Baram region, the Penan benefited from missionary efforts over the following decades. These included, for example, teaching the Penan how to cultivate food crops. All six villages have a village church under the Sidang Ingil Borneo denomination. Today, the village Church plays an important role in village governance with the Church leadership working closely with the village leader to settle any disputes that arise among villagers (Chen, 2016). In terms of infrastructure, all six villages are recipients of the state utility's Sarawak Alternative Rural Electrification Scheme (SARES) programme and have been provided with solar home systems for lighting and basic energy service needs. There is no phone connectivity
in the villages, but each village has government-provided water tanks. Except for Long Urang, all of the villages have kindergartens. The main livelihood activities of individuals in all six villages are agriculture and hunting-gathering from the forest, with some women also making handicrafts from rattan which they sell in Long Bedian. As is the case with most indigenous villages in Sarawak, all six Penan villages have experienced high levels of migration among their youths to nearby cities, such as Miri and Kuching, due to limited economic opportunities. 3.2. Data We use a combination of rich primary data and gridded climate data to engage with our three research questions: (1) Are perceptions of climate anomalies homogenous? (2) Do perceptions align with instrumental climate data? And (3) What is the marginal effect of varying levels of TEK on perceptions? Starting with our first question, we use two novel co-designed data instruments to collect primary data on perceptions of rainfall and temperature anomalies from respondents in our six villages. For our second question, we generate corresponding statistical measures of rainfall and temperature anomalies using instrumental climate data. For our third question, we create and enumerate a survey questionnaire to generate an ordinal measure of TEK, as well as data on other variables that may affect perceptions. 3.2.1. Perceptions We worked with Penan elders in our pilot village (Long Luteng) to iteratively co-design two vehicles to measure perceptions of rainfall and temperature. We first worked with the elders to decide on time-periods to which to anchor our measure of perceptions. According to the elders, the livelihoods of the Penan revolve around two wet seasons. The first wet season begins in March and ends in July. The second wet season
2
Previously, a major reason for the Penan to maintain a nomadic lifestyle was due to several traditional beliefs. These included: (1) being fearful of death, which in practice meant that Penan communities did not stay in the same location where a member of the community had died; (2) not felling large trees so as not to anger the spirits that inhabited the trees, a constraint on clearing space for settlement; and (3) a belief in bird omenology, which governed their movements through the forest (Chen 2016). 3
Global Environmental Change 58 (2019) 101974
T. van Gevelt, et al.
begins in October and ends in December. We therefore constructed our perception data collection vehicles around the two wet seasons and two dry seasons to maximise the accuracy of respondent recall. In order to obtain a suitably representative number of data points, we elected to cover the twelve seasons occurring over a period of three years (2015–2017). Our iterative design process resulted in the use of river levels as a measure of rainfall. River levels were selected as the Penan have excellent recall ability concerning river levels3 as they use rivers as reference points for navigating the rainforest and continue to rely heavily on hunting and fishing for subsistence (Hansen, 1988; Weisshaidinger et al., 2012). Each village is located directly adjacent to a tributary of the Baram river of approximately the same width, depth and flow, meaning that changes in river levels should be highly salient to respondents. To measure perceptions of rainfall anomalies, respondents were shown seven identical bottles each with a different level of water to represent river levels (see Supplementary Fig. 1). For each season, respondents were asked to select a bottle to represent what they perceived to be the normal (long-term) rainfall level. For each season for the years 2015, 2016 and 2017, respondents were asked to select bottles corresponding to what they perceived rainfall levels to have been. These selected bottles were then compared with the bottles representing normal rainfall levels for each season. We define an individual to have perceived a rainfall anomaly if their perceptions of river levels in a given season were at least one level above or below their perception of the normal river level for that season. For temperature, our iterative design process resulted in the use of clothing as a measure of temperature. The Penan are famously averse to sunlight (Hansen, 1988; Chen, 2016). If the Penan are expected to face exposure to sunlight – which is the norm now that the majority of Penan have shifted to semi-nomadic livelihoods that revolve primarily around paddy agriculture – they will wear more clothing the more intense the sun is. As sunlight and temperature are highly correlated, we can expect the Penan to wear more clothes the higher the temperature is (van den Besselaar et al., 2015). Similar to our measure of perceptions for rainfall anomalies, respondents were shown seven different sets of clothing (outfits) on an increasing scale (e.g. from little to much clothing) (see Supplementary Figures 2 and 3). Respondents were asked to identify the outfit that they would normally wear for each season. For each season for the years 2015, 2016 and 2017, respondents were asked to select the outfit they actually wore during that season. We define an individual to have perceived a temperature anomaly if the outfit they wore in a given season was at least one level above or below their perception of the normal outfit worn for that season. We piloted both of our co-designed data collection vehicles in the Penan village of Long Luteng. We selected Long Luteng as it is similar to our six sampled villages in terms of population, geography and culture. Additionally, as Long Luteng is suitably distant from our sample sites that there is little chance of discussion among participants involved in co-designing our measures and respondents in our six villages. As part of our pilot study, we engaged in in-depth discussions with respondents to make sure that respondents felt that our two measures were both accurate and feasible ways in which respondents could recall rainfall and temperature. There are some potential limitations to our two community co-designed measures of perceptions. For our rainfall measure, there is the possibility that river levels may be affected by external events. For example, the presence of upstream logging may increase the flow rate and the soil can carry more sediment. This may raise the river bed and therefore affect river levels. Visual inspection of Landsat images across roughly 9-year intervals from 1988 to 2015 revealed that there were
few instances of upstream logging (see Supplementary Fig. 4). This was further confirmed by inspection of the Hansen et al. (2013) Version 1.6. forest cover loss data. While Marudi district experienced significant logging between 2000 and 2017 downstream of the six villages, upstream, the minor amounts of logging that occurred were unlikely to have dramatically affected river levels (see Supplementary Fig. 5). For our temperature measure, although sunlight and temperature are highly correlated – instances of cloud cover (prevalent in the interior of Borneo) may mean that there will be days or relatively short stretches of time where there may be little sunlight, but temperatures may be high (Cheuk et al., 2012). As our measure of perceptions is seasonal as opposed to daily or weekly, it is unlikely that this is an issue. An additional concern may arise from possible interaction between rainfall and temperature anomalies. For example, if a day is warmer than normal and also experiences higher than normal levels of rainfall, it is possible that individuals may perceive the day to be cooler than it actually is. As our unit of time is seasonal rather than daily, it is likely that this will not systematically bias results. 3.2.2. Climate anomalies We use thirty years (1988–2017) of gridded spatial rainfall and temperature data from the CRU TS v4.02 database to generate our measures of climate anomalies. The CRU TS database is commonly used in similar studies and interpolates data from station records to provide monthly rainfall and temperature data from 1901 to 2017 at a resolution of 0.5 by 0.5° (Ramirez-Villegas and Challinor, 2012; Harris et al., 2014). Given the sparsity of reliable weather stations in Borneo, our climate data are interpolated from 14 weather stations located in a 10 by 10° grid centred on northern Borneo.4 We construct our statistical measures of rainfall and temperature anomalies per season as defined by Penan elders in Long Luteng. Consistent with our data collection vehicle for perceptions of climate anomalies, we construct seasonal measures of rainfall and temperature anomalies for 2015, 2016 and 2017. Using rainfall as an example, we follow Howe et al. (2012), McCright et al. (2014), Salehyan and Hendrix (2014), Randell and Gray (2016) and Larcom et al. (2019) and construct our measure according to:
a i, t =
Xi, t
µi i
(1)
where ai, t indicates whether there has been a seasonal rainfall anomaly in village i during period t. Xi, t is rainfall in village i during period t, μi represents the mean of average rainfall over a period of 30 years in village i, and σi is the standard deviation over a period of 30 years in village i. When overlaying our six villages to the 0.5 by 0.5° gridded spatial climate data from the CRU TS v4.02 dataset, all six villages are located within the same grid. We define a village to have experienced a climate anomaly if rainfall during period t is at least one absolute standard deviation away from the long-term mean. 3.2.3. Traditional ecological knowledge We use Oroo’ signs – a unique sign language of the Penan – as a measure of traditional ecological knowledge among our respondents. Oroo’ signs are made using sticks, twigs, branches and folded leaves. They are erected and positioned in different ways in the rainforest to indicate different messages. Oroo’ signs are able to transmit a wide variety of messages to other Penan deep in the rainforest. Traditionally, 4 This is in-line with other studies in the general region. For example, Boussiere et al. (2013) noted that the CRU TS data used in their study in Papua, Indonesia were interpolated from 11 weather stations. While our data are interpolated from relatively few weather stations and warrant careful interpretation, we note that in their analysis of the robustness of weather station networks, Ramirez-Villegas and Challinor (2012) found the sensitivities of weather station networks to data loss to be low in South-East Asia.
3 The intimate relationship between many Penan communities and the river is demonstrated by village names. For example, Long Win means ‘the River Win’ in the Penan language.
4
Global Environmental Change 58 (2019) 101974
T. van Gevelt, et al.
Oroo’ signs were used extensively by groups of travelling Penan to communicate with other groups across time and space. Among the Penan, individuals with an extensive vocabulary of Oroo’ signs are largely acknowledged to be those with a deep knowledge and understanding of the forest (Zaman and Winschiers-Theophilus, 2015; Zaman et al., 2019). Together with Penan elders and youths in our pilot village, we assembled a set of 10 Oroo’ signs ranging from more common signs to less common signs (see Supplementary Fig. S6). As part of our data collection process, we showed photographs of each of the 10 Oroo’ signs to our respondents and asked them to provide the meaning of each sign. We used this data to create a measure (ranging from 0 to 10) of TEK, with respondents being allocated one point for each correct answer.
with our statistical measure of anomalies. 3.3.2. Poisson regression analysis We use a Poisson model to estimate the marginal effect of TEK on the alignment of perceptions with instrumental climate data (a heavily qualified measure of accuracy). We do this by estimating variants of the following Poisson model:
ci, s =
0
+
1 Ti
+ Xi
+ Vi +
i
(2)
where ci, s is a count variable that represents the number of times individual i’s perceptions of climate anomalies in season s align with the instrumental climate data. Ti represents our measure of traditional ecological knowledge (on a scale of 0–10). Xi is a vector of individuallevel variables, including: age, gender, education, livelihood strategy, salience of climate change, and an asset index. Vi is a categorical variable denoting which village individual i is located in and captures village-level differences. We use robust standard errors clustered at the village-level to account for overdispersion in our data. We define whether an individual's perceptions of climate anomalies is broadly aligned with the instrumental climate data if there has been a climate anomaly where rainfall or temperature is at least 1 standard deviation away from the long-term average, and individual responses of river levels and clothing are at least 1 level away (in the same direction) from the self-reported long-term average.
3.2.4. Control variables We also collected data on a number of control variables which are likely to be associated with individual perceptions of climate anomalies (Dove et al., 2007; Deressa et al., 2009; Howe and Leiserowitz, 2013; Howe et al., 2014; Rudiak-Gould, 2014; Larcom et al., 2019). These include age, gender, education, livelihood strategy, an asset index, and a binary measure capturing the salience of climate change to the individual. A detailed description of explanatory variables is presented in Supplementary Table 1. 3.2.5. Data enumeration We enumerated our co-designed data collection vehicles to measure perceptions of rainfall and temperature, our TEK questionnaire, and a semi-structured questionnaire to collect data on control variables to 200 households across the six villages during June to August 2018. Given the small size of Penan settlements in Sarawak, we endeavoured to interview as many adults as possible in each village to ensure a sufficient sample size to undertake our statistical analysis. We spent two weeks in each village and surveyed every individual adult who was present during that period. While we achieved a 100% response rate in surveying every individual adult who was present, we note that we were unable to enumerate our data collection instrument to community members who were away from their respective villages for the duration of the two-week period.5
3.3.3. Robustness checks We undertake three robustness tests. Firstly, we account for possible mismatches concerning the time horizon that respondents may have used when considering seasonal deviations and the time horizon used to calculate our measures of rainfall and temperature anomalies. We do this by varying the number of years of rainfall and temperature data used to calculate the long-term averages and standard deviations of our rainfall and temperature anomalies and present our main results using long-term averages based on time horizons of 5, 10 and 20 years (McCright et al., 2014). Secondly, we vary our definition of a climate anomaly that we use in generating our statistical measures. We do this by defining there to have been a climate anomaly if rainfall or temperature deviate from their long-term means by at least two standard deviations. This is to account for the possibility that a one standard deviation may be too nuanced of an anomaly for respondents to recognise. Thirdly, we also account for the possibility that respondents may not consider a one-level change in perceptions as sufficient to correspond to the experience of a climate anomaly. We account for this by varying our definition of when we consider a respondent to have perceived a rainfall or temperature anomaly and re-estimate our results. Results from our three robustness tests are presented in the supplementary materials and discussed in Section 4.3.
3.3. Data analysis We used two main analytical approaches to engage with our three research questions. For our first two questions, our analysis is largely based on the generation and careful study of several key descriptive statistics constructed both from our primary data on perceptions of climate anomalies and our statistical construction of corresponding measures using instrumental climate data. For our third question, we formally estimate the marginal effect of varying levels of TEK on the alignment of perceptions with instrumental climate data using a Poisson model. We undertake three further robustness checks to test the validity of our results.
4. Results 4.1. Descriptive statistics
3.3.1. Descriptive statistics We explore a number of descriptive statistics that we generate using our primary data on perceptions of anomalies and our statistical measures of climate anomalies. We begin by first understanding the individual characteristics of our respondents. We then present our data on perceptions of climate anomalies to see if perceptions are homogenous. In our efforts to see if perceptions of climate anomalies correspond with instrumental climate data, we first present data representing our statistical measures of anomalies. We then combine our two sources of data to see the extent to which perceptions of climate anomalies align
4.1.1. Individual characteristics Table 1 presents descriptive statistics for traditional ecological knowledge and our individual-level control variables. We find the average TEK score, as measured by proficiency of Oroo’ signs, is 3.06 with a standard deviation of around 3. Six individuals scored perfect scores with a further 18 scoring at least eight out of ten. Notably, 59 individuals scored 0 out of 10. Turning to our individual-level control variables, the mean age is 38 years old with half of our respondents being female. 76 of our respondents have no formal education, 66 respondents have primary education and 58 have secondary education or above. For 182 of our respondents, the main livelihood strategy is farming. For our measure of salience, 71 individuals have experienced what they consider are direct impacts attributable to climate anomalies within the last three years.
5 According to village elders, while these individuals are considered community members, they reside away from the village for most of the year to gain an education or to be employed in urban areas.
5
Global Environmental Change 58 (2019) 101974
T. van Gevelt, et al.
decreases, increases and no change. The diversity of perceptions is best illustrated through an example. If we take the second wet season (W2) in 2017 as an example, we can see that 57 individuals stated that there was no deviation from the long-term mean and 82 individuals stated that rainfall was lower than their perceived long-term averages. If we turn to Supplementary Table 2, we see that from the 61 individuals who stated that there was a positive deviation from the long-term mean, 31 individuals perceived only a one-level increase, 24 individuals perceived a two-level increase and six individuals perceived a three-level increase. From the 82 individuals who perceived a negative deviation from the long-term mean, 43 perceived a one-level decrease, 17 perceived a two-level decrease, six perceived a three-level decrease and four perceived a four-level decrease. Turning to temperature in Fig. 3, we continue to see a wide range of perceptions, although the majority of individuals perceived aboveaverage temperatures in every season. Taking the same approach as in our analysis of rainfall, we focus on the first wet season (W1) of 2016 as an illustrative example. While over half of our sample (104 individuals) perceived an increase in temperature of at least one level, 41 perceived
Table 1 Descriptive statistics. Variable
Mean
Std. Dev.
Min.
Max.
TEK Age Gender Education Livelihood Asset index Salience of Climate Change N = 200
3.06 38.31 0.50 0.910 0.09 0.000 0.36
3.02 13.276 0.50 0.816 0.29 1 0.48
0 19 0 0 0 −3.97 0
10 78 1 2 1 1.89 1
4.1.2. Perceptions of climate anomalies Figs. 2 and 3 present respondent perceptions of deviations from their self-reported long-term means by season for rainfall and temperature, respectively.6 Supplementary Table 2 disaggregates these results and presents the number of levels by which individuals perceived rainfall and temperature to deviate. Starting with rainfall in Fig. 2, we can see that there are many individuals for each season who perceive
Fig. 2. Deviations from perceptions of long-term average rainfall.
Fig. 3. Deviations from perceptions of long-term average temperature.
6 Supplementary Tables 3 and 4 present respondent perceptions of deviations disaggregated by village. Supplementary Table 5 presents the results of a series of Chi-squared tests to test for systematic differences in perceptions by village. Our results suggest that there were systematic differences in perceptions for
(footnote continued) three seasons for perceptions of rainfall (at the 5% significance level) and for one season for perceptions of temperature (at the 5% significance level). 6
Global Environmental Change 58 (2019) 101974
T. van Gevelt, et al.
direction. Looking at rainfall, we see that the proportion of individuals with broadly aligned perceptions ranges from 0.19 to 0.36. Our results suggest that perceptions during seasons where there was a statistical climate anomaly were generally more aligned than for seasons where there was no statistical anomaly, with between 30 and 36 percent of our respondents having perceived changes broadly consistent with our anomaly measure. For temperature, we note a similar pattern where the proportion of individuals with broadly aligned perceptions is notably higher for seasons where there were statistical climate anomalies (0.37–0.52) than for seasons where there were no statistical anomalies (0.25–0.26).
Table 2 Rainfall and temperature anomalies. 2015 D1 Rainfall Temperature a
0 0
2016 W1
D2
a
a
1 1
1 1
W2 0 1
D1 0 1
2017 W1
D2
0 1
1 1
a
W2
D1
W1
D2
W2
0 1
0 1
0 0
1 0
1 0
Denotes a negative deviation from the long-term mean.
no deviation from the long-term mean and 55 perceived a decrease in temperature of at least one level. As seen in Supplementary Table 2, among the 104 individuals who perceived an increase in temperature, 43 perceived an increase of one-level, 30 perceived an increase of two levels, 18 perceived an increase of three levels, 11 perceived an increase of four-levels and two perceived an increase of five levels. Among the 55 who perceived a decrease in temperature, 30 perceived a decrease of one level, 16 perceived a decrease of two-levels, five perceived a decrease of three-levels, three perceived a decrease of four-levels and one perceived a decrease of five-levels.
4.2. Poisson regression estimates Table 4 presents estimations of the marginal effects of our measure of TEK on the alignment of perceptions. Columns (1–3) show our estimations for aligned perceptions of rainfall anomalies and columns (4–6) show our results for aligned perceptions of temperature anomalies. Starting with column (1), we include only our measure of TEK and do not find it to be statistically significant. Somewhat surprisingly, however, we find that our measure of traditional ecological knowledge has a negative coefficient. In column (2), we include individual-level control variables and continue to find our measure of traditional ecological knowledge to be statistically insignificant. In column (3), we include our village-level categorical variable. Our measure of traditional ecological knowledge continues to be statistically insignificant. Turning now to temperature, in column (4) we find our measure of traditional ecological knowledge to be statistically significant at the 1% level with a positive coefficient. In column (5), our measure of traditional ecological knowledge continues to be statistically significant when individual-level controls are included, albeit now at the 5% significance level and with a reduced coefficient. In our preferred specification (column 6), with both individual-level and village-level controls included, our measure of traditional ecological continues to be statistically significant at the 5% significance level with a coefficient of 0.072. This suggests that every additional Oroo’ sign known by the average respondent is associated with approximately 0.072 more seasons where perceptions of temperature anomalies are aligned with the instrumental data. Put another way, an individual who identified all ten Oroo’ signs correctly is likely to have perceptions that corresponded with the instrumental climate data 6% more frequently than an individual who identified no Oroo’ signs correctly.
4.1.3. Statistical measures of climate anomalies Table 2 presents our statistical measures of absolute rainfall and temperature anomalies according to dry and wet seasons over the threeyear period of 2015–2017. A value of one indicates that there was an anomaly during the given time period. We define an anomaly as at least a one standard deviation from the long-term mean (see Supplementary Tables 6 and 7 for more detailed information on the magnitudes of anomalies). We can see that rainfall deviated from the long-term mean by at least one standard deviation in five of the 12 available seasons. Three of these deviations were negative and two were positive. Turning to temperature, we note that temperature deviated from the long-term mean by at least one standard deviation in eight out the 12 available seasons. All temperature deviations were positive. 4.1.4. Alignment of perceptions Table 3 presents shows the extent to which perceptions of climate anomalies align with instrumental climate data. We define an individual's perception to broadly correspond with instrumental climate data if we find that there has been a statistical climate anomaly of at least 1 standard deviation and the individual has stated that there has been at least a one-level change in river levels or clothing in the same Table 3 Alignment of perceptions. Magnitude
2015 D1
Rainfall Temperature a
0.26 0.26
2016 W1
D2 a
0.34 0.50a
W2 a
0.30 0.39a
0.23 0.39a
2017
D1
W1
0.28 0.37a
0.25 0.52a
D2
W2 a
0.32 0.43a
D1
0.36 0.41a
W1
0.19 0.41a
0.25 0.25
D2
W2 a
0.31a 0.26
0.36 0.25
Denotes season with anomaly.
Table 4 Poisson regression estimations: average marginal effects.a Variables
(1)
(2)
(3)
(4)
TEK
−0.020 (0.020) No No −369.910 200
−0.023 (0.019) Yes No −366.497 200
−0.019 (0.024) Yes Yes −364.823 200
0.109 (0.036) No No −409.365 200
Individual-level controls Village-level controls Log pseudolikelihood N = 200
*P < 0.10, **P < 0.05 and ***P < 0.01. a Columns 1–3 are for rainfall perceptions; columns 4–6 are for temperature perceptions. 7
(5) ⁎⁎⁎
(6)
0.064 (0.032) Yes No −405.356 200 ⁎⁎
0.072⁎⁎ (0.031) Yes Yes −403.028 200
Global Environmental Change 58 (2019) 101974
T. van Gevelt, et al.
4.3. Robustness checks
results using a threshold of a two-level deviation for perceptions and a one standard deviation threshold for construction of our statistical climate anomaly measures. We now find TEK to be weakly statistically significant in columns (1) and (2) with a negative coefficient for rainfall. In column (3), our preferred specification, we no longer find our measure of TEK to be statistically significant. Turning to temperature, we no longer find our measure of TEK to be statistically significant in columns (4) to (6). Additionally, the coefficient of TEK is now negative in our preferred specification (column 6). Supplementary Table 20 presents Poisson model results using a threshold of a two-level deviation for perceptions and a two standard deviation threshold for our measures of statistical climate anomalies. We no longer find our measure of TEK to be statistically significant for any of our estimations (columns 1–6). Notably, the coefficients of TEK for our three estimations (columns 4–6) concerning perceptions of temperature anomalies are now positive.
To account for the possibility that respondents were selecting their average river levels and clothing for each season according to time horizons shorter than the 30-year period used in the calculation of our anomaly measures, we calculate anomaly measures based on 5-, 10and 20-year periods. As seen in Supplementary Table 8, varying the time period leads to some changes in the seasons in which a climate anomaly was experienced according to the instrumental data. For example, using a 5-, 10- or 20-year period results in a climate anomaly in the first dry season in 2016 whereas no climate anomaly is calculated for this season when using a 30-year period. Despite some variation in the occurrence of climate anomalies, we find that proportions of respondents with aligned perceptions for any new periods with climate anomalies are similar to those estimated over a 30-year period (see Supplementary Table 9). In Supplementary Tables 10–12, we re-estimate our Poisson models using our 5-, 10- and 20-year periods, respectively. Our results are broadly similar to our original estimations. For our second robustness check, we account for the possibility that our statistical measure of climate anomalies may be too nuanced for respondents to recognise. As can be seen in Supplementary Table 13, when using a threshold of two standard deviations, we no longer find any season where there has been a rainfall anomaly. For temperature, we find four consecutive seasons where there has been a positive temperature anomaly of at least two standard deviations. As can be seen in Supplementary Table 14, the proportion of respondents with aligned perceptions of rainfall anomalies ranges from 0.185 to 0.355. For temperature anomalies, the proportion of respondents with aligned perceptions for seasons without a temperature anomaly range from 0.230 to 0.320. For seasons with a temperature anomaly of at least two standard deviations, the proportion of respondents with aligned perceptions range from 0.370 to 0.520. Supplementary Table 15 presents the average marginal effects of TEK on the frequency of aligned perceptions when using our twostandard deviation threshold. In columns (1–3), we do not find TEK to be statistically significantly associated with aligned perceptions of rainfall anomalies. However, we do now find the coefficient to be positive in our preferred specification (column 3). In column (4), we continue to find TEK to be positively associated with the frequency of aligned perceptions at the 1% significance level. In column (5), our coefficient is slightly reduced, and statistical significance is now at the 10% level. In our preferred specification in column (6), we no longer find our measure of TEK to be statistically significant. Similarly, it is possible that respondents may have considered a onelevel change in perceptions as too subtle to correspond to the experience of a climate anomaly. As such, we change our definition of when we consider a respondent to have perceived a rainfall or temperature anomaly to a two-level deviation from the long-term average. As seen in Supplementary Table 16, using this threshold results in far more respondents perceiving no change for rainfall or temperature. Importantly, we still see significant deviations in terms of individuals perceiving deviations both below and above their self-reported longterm averages. Supplementary Table 17 presents data on the proportion of respondents with perceptions that correspond to our statistical measures of anomalies, calculated using a one standard deviation threshold. As a result of an increase in the number of respondents perceiving no change under our two-level deviation threshold, we now see the proportion of respondents with aligned perceptions for seasons without statistical climate anomalies to be substantially higher, from 0.54 to 0.60 for rainfall and 0.63 to 0.76 for temperature. For seasons with a statistical climate anomaly, we find the proportion of aligned perceptions to have decreased, from 0.10 to 0.19 for rainfall and 0.18 to 0.31 for temperature. We find similar results when using a statistical climate anomaly threshold of two standard deviations in Supplementary Table 18. Supplementary Table 19 presents re-estimated Poisson model
5. Discussion By focusing on a geographically understudied area and combining novel primary data on perceptions of climate anomalies with instrumental climate data, our study makes three important contributions to the literature on local perceptions and adaptation to climate change. Firstly, our results suggest that perceptions of rainfall and temperature anomalies vary substantially both in terms of whether individuals perceive anomalies to have occurred and in terms of the magnitude of the anomaly. We find this variation to exist among and within villages. Our results contrast with many extant studies that have found perceptions of different indicators of climate change to be broadly homogenous (Habiba et al., 2012; Bola et al., 2014; Gioli et al., 2014). We suggest that this may be as our study focuses on climate anomalies – significant deviations from the perceived ‘norm’ – rather than long-term trends. Given that many of the impacts of climate change are felt as a result of the increased frequency and severity of climate extremes, an understanding of how indigenous communities perceive such anomalies is necessary in learning how to adapt to the local impacts of climate change. While a detailed understanding of individual perception formation is outside the scope of this study, anecdotal evidence from informal conversations with respondents suggests that while all households within (and among) villages experienced the negative impacts of climate anomalies to physical assets (e.g. damaged houses, equipment) and livelihoods (e.g. destroyed crops and water supply shortages), these effects may have been more severe for different households at different points in time. This may have led to individuals anchoring their perceptions of climate anomalies for a given season to its impact (or lack of) on their assets and livelihoods. Secondly, given the heterogeneity of perceptions of climate anomalies in our sample, our results suggest that a substantial proportion of individual perceptions of climate anomalies do not correspond with instrumental climate data. Indeed, comparing individual perceptions of rainfall and temperature anomalies with corresponding statistical measures of anomalies based on gridded rainfall and temperature data suggests that only between 30 and 50 percent of individuals perceived climate anomalies in line with the instrumental climate data. Interestingly, our results suggest that respondents were more likely to over- rather than underestimate the presence climate anomalies. These findings are robust to different time-horizons and thresholds for defining climate anomalies. While many extant studies in the literature find indigenous perceptions of climate change to generally be in-line with instrumental climate data (e.g. Hageback et al., 2005; Klintenberg et al., 2007; West et al., 2008; Bryan et al., 2009; Byg and Salick, 2009; Valdivia et al., 2010; Chaudhary and Bawa, 2011; Kalanda-Joshua et al., 2011; McNeeley and Shulski, 2011; Habiba et al., 2012; Ashraf and Routray, 2013; Gioli et al., 2014; Nkomwa et al., 2014; Savo et al., 2016), there are a number of studies that find that indigenous perceptions and instrumental climate data diverge to some 8
Global Environmental Change 58 (2019) 101974
T. van Gevelt, et al.
degree (e.g. Meze-Hausken, 2004; Gearheard et al., 2006; Marin, 2010; Speranza et al., 2010; Boissiere et al., 2013; Bola et al., 2014). In studies where perceptions diverge from instrumental climate data, much of the divergence is attributed to the possibility that the instrumental and individual observations may not be operating at the same scale or even be measuring the same climate phenomena (Gearheard et al., 2006; Marin, 2010; Savo et al., 2016). It is possible that divergences among our sampled respondents may be at least partially attributable to discrepancies in scale and measurement scope for a number of potential reasons. Firstly, it is possible that the resolution of our gridded climate data and its interpolation from a relatively small number of weather stations, does not capture differences in rainfall and temperature anomalies that occur at the village-level.7 Secondly, despite the efforts taken during the co-design and piloting of our data collection instruments, there is a chance that there are substantial differences in the constructs being compared. This is as while our instrumental climate data measures rainfall and temperature, the individual-level data can be seen as abstractions of rainfall and temperature and may therefore not be necessarily directly related to them. Finally, it is possible that respondents may be anchoring their perceptions to a particular event (e.g. high rainfall and flooding over a relatively short timeframe) rather than holistically surveying the entire season. Our third contribution to the literature is an attempt to unravel the ‘black box’ of TEK by isolating the marginal effect of varying levels of TEK on rainfall and temperature anomaly perceptions. Using Poisson regression analysis, we find only weak evidence of a systematic statistical association between our measure of TEK and the alignment of perceptions to instrumental climate data (a heavily qualified measure of an accurate perception of a climate anomaly). When we do find evidence of a statistical association, we find the marginal effect of TEK to be relatively small. This is a provocative result given the important climate insights that a number of extant studies have attributed to TEK (e.g. Alessa et al., 2008; Speranza et al., 2010; Kalanda-Joshua et al., 2011; Boissiere et al., 2013; Gioli et al., 2014; Savo et al., 2016). While we are unable to identify the precise explanations underpinning our findings, there are a number of potential reasons as to why we only find weak evidence of a systematic statistical association. Firstly, our findings may reflect the possibility that Penan TEK is primarily associated with cultural values and identity rather than helping to understand the current environment and associated changes. This is as the Penan TEK is largely based on a rich historical understanding of the forest ecosystem and traditional practices. Secondly, it is possible that our measure of TEK – based on the Oroo’ forest sign language – does not correspond with the kind of TEK related to understanding changes in the climate. While we maintain that Oroo’ signs are a suitable proxy for Penan TEK in general, it is possible that an indicator based around phenology and seasonality may result in different findings (GarayBarayazarra and Puri, 2011). Thirdly, it is possible that our results are dependent on the validity of our definition of aligned perceptions, which itself rests on a number of previously discussed assumptions inherent in the generation of our statistical measures of anomalies and its comparison with our individual-level measures of perceptions. Taken together, our findings point towards several implications for adaptation among the Penan and indigenous communities more generally. That we find substantial heterogeneity in perceptions of climate anomalies both within and among villages suggests that Penan communities may currently lack a shared understanding of climate
anomalies and face difficulties in achieving the collective action often required to adapt to climate change at the local-level (Adger, 2003; Ostrom, 2010). The importance of a shared understanding is arguably especially important in the Penan context given their egalitarian governance structure, where decisions and actions of significance tend to only be made under unanimous agreement among all adults in the community (Chen, 2016). Indeed, in all six of our field sites, anecdotal evidence suggests that current adaptation efforts are so far limited to a minority of households who have acted independently by relocating houses and crops away from riverbanks as a response to flooding caused by heavy rainfall. We suggest that our six Penan communities may benefit from forums where community members and external stakeholders (e.g. climate (social) scientists) can come together, share their perceptions, and reach a collective understanding on the local impacts of climate change. This could then pave the way for a discussion of feasible adaptation strategies and the necessary capacity building and two-way knowledge transfers required to implement these adaptation strategies. While such a model of engagement is not new for the Penan and has been used to great effectiveness in a number of areas (e.g. agricultural extension services and cataloguing traditional knowledge), devising appropriate adaptation strategies will require Penan leadership and an acceptance by external stakeholders of the customary land rights of the Penan and the importance of their traditional culture (Chen, 2016; Weisshaidinger et al., 2012; van Gevelt et al., 2017; Zaman et al., 2019). This is likely to manifest itself by constraining the full set of adaptation options available, as the Penan, for example, are reluctant to make any physical changes to the environment (e.g. flood defences) due to a traditional belief structure and philosophy emphasising harmony over conflict with both humans and the environment (Chen, 2016). While such a process of engagement will likely require a sustained and concerted effort by both community members and external stakeholders, our finding that respondents tend to over-estimate rather than under-estimate the presence of climate anomalies may well suggest that many among the Penan perceive significant changes in the climate and may engage in such forums. More generally, our attempt to systematically understand the relationship between TEK and anomaly perceptions indicates that the relationship between TEK and perceptions of climate anomalies may not be straightforward, at least in tropical rainforests. Indeed, we suggest that our finding of only a weak statistical association between TEK and perceptions may be at least partially explained by the effects of climate change being more subtle than in more extreme environments, such as deserts and polar regions (Corlett and Primack, 2011). While we are aware of our methodological limitations and the need for future research to understand the mechanisms through which TEK relates to perceptions of climate anomalies, we suggest, at least in tropical rainforests, a note of caution in advancing adaptation strategies that are predominantly premised on TEK. 6. Conclusion A detailed understanding of local perceptions of climate anomalies is key to understanding how to enable indigenous communities to adapt to the local impacts of climate change. We use a combination of novel primary data on rainfall and temperature anomalies, traditional ecological knowledge and instrumental climate data to engage with key questions in this area. We find that: (1) perceptions of climate anomalies among our respondents vary substantially in terms of both occurrence and magnitude; (2) perceptions of climate anomalies do not generally correspond with our statistical measures of climate anomalies; (3) there is only weak evidence for a systematic association between TEK and perceptions of climate anomalies. Notwithstanding the methodological limitations of our study, our findings make at least three important contributions to the evidence-base. Firstly, by focusing on Sarawak, Malaysia we present evidence from a relatively
7 As noted in footnote 6 and shown in Supplementary Table 5, there is some evidence to suggest that there were systematic differences in perceptions for three out of the twelve seasons for rainfall and one out of the twelve seasons for temperature among the six villages. This suggests that during these seasons, villages may not have experienced the same rainfall and temperature anomalies for at least part of the season.
9
Global Environmental Change 58 (2019) 101974
T. van Gevelt, et al.
underrepresented region. This is especially important given that almost 40% of Sarawak's population are indigenous and that adapting to climate change has a direct impact on their livelihoods and stewardship of a biodiversity hotspot. Secondly, we make a methodological contribution by focusing on climate anomalies rather than trends, by using a community co-designed data collection vehicle, and operationalising a measure of TEK. While we are aware of the limitations of our quantitative approach, we believe that the evidence-base benefits from a diversity of rigorous approaches from which stakeholders can draw out findings and lessons within their appropriate situational and methodological contexts. Finally, while remaining deeply cognisant of the long histories of failed external interventions and the critical importance of empowering indigenous people to design adaptation strategies for their communities, our results suggest a note of caution in relying disproportionately or solely on TEK in their planning.
Borneo Malaysia: 14 years on. Int. J. Innov. Manag. Technol. 3 (6), 682–687. Corlett, R., Primack, R., 2011. Tropical Rain Forests: An Ecological and Biogeographical Comparison. Wiley-Blackwell, Oxford, pp. 1–326. Daw, T., 2010. Shifting baselines and memory illusions: What should we worry about when inferring trends from resource user interviews? Animal Conservation 13, 534–535. Demski, C., Capstick, S., Pidgeon, N., Sposato, R., Spence, A., 2017. Experiences of extreme weather affects climate change mitigation and adaptation responses. Clim. Change 140, 149–164. Deressa, T., Hassan, R., Ringler, C., Alemu, T., Yesuf, M., 2009. Determinants of farmers’ choice of adaptation methods to climate change in the Nile Basin of Ethiopia. Glob. Environ. Change 19 (2), 248–255. Deryugina, T., 2013. How do people update? The effects of local weather fluctuations on believes about global warming. Clim. Change 118, 397–416. Donovan, D., Puri, R., 2004. Learning from traditional knowledge of non-timber forest products: Penan Benalui and the autoecology of Aquilaria in Indonesian Borneo. Ecol. Soc. 9 (3). Dove, M., Smith, D., Campos, M., Matthews, A., Rademacher, A., Rhee, S., Yoder, L., 2007. Globalisation and the construction of western and non-western knowledge. In: Sillitoe, P. (Ed.), Local Science vs. Global Science: Approaches to Indigenous Knowledge in International Development. Berghahn, New York and Oxford. Fam, S.D., 2017. China came, China built, China left? The Sarawakian experience with Chinese dam building. J. Curr. Chin. Aff. 46, 119–158. Fernandez-Llamazares, A., Diaz-Reviriego, I., Luz, A., Cabeza, M., Pyhala, A., ReyesGarcia, V., 2015. Rapid ecosystem change challenges the adaptive capacity of local environmental knowledge. Global Environ. Change 31, 272–284. Ford, J., Sherman, M., Berrang-Ford, L., Llanos, A., Carcamo, C., Harper, S., Lwasa, S., Namayna, D., Marcello, T., Maillet, M., Edge, V., 2018. Preparing for the health impacts of climate change in indigenous communities: the role of community-based adaptation. Global Environ. Change 49, 129–139. Garay-Barayazarra, G., Puri, R., 2011. Smelling the monsoon: Senses and traditional weather forecasting knowledge among the Kenyah Badeng farmers of Sarawak, Malaysia. Indian J. Tradit. Knowl. 10 (1), 21–30. Gearheard, S., Matumeak, W., Angutikjuaq, I., Maslanik, J., Huntington, H., Leavitt, J., Kagak, D., Tigullaraq, G., Barry, R., 2006. It's not that simple: A collaborative comparison of sea ice environments, their uses, observed changes, and adaptations in Barrow, Alaska, USA, and Clyde River, Nunavut, Canada. Ambio 35, 203–211. Gioli, G., Khan, T., Scheffran, J., 2014. Climatic and environmental change in the Karakoram: Making sense of community perceptions and adaptation strategies. Reg. Environ. Change 14, 1151–1162. Habiba, U., Shaw, R., Takeuchi, Y., 2012. Farmer's perception and adaptation practices to cope with drought: perspectives from North-western Bangladesh. Int. J. Disaster Risk Reduct. 1, 72–84. Hageback, J., Sundberg, J., Ostwald, M., Chen, D., Yun, X., Knutsson, P., 2005. Climate variability and land-use change in Danangou watershed, China – examples of smallscale farmers’ adaptation. Clim. Change 72, 189–212. Hansen, E., 1988. Stranger in the Forest: On Foot Across Borneo. Vintage Departures, New York, pp. 1–286. Hansen, J., Sato, M., Ruedy, R., 2012. Perception of climate change. Proc. Natl. Acad. Sci. 109, E2415–E2423. Hansen, M., Potapov, P., Moore, R., Hancher, M., Turubanova, S., Tyukavina, A., Thau, D., Stehman, S., Goetz, S., Loveland, T., Kommareddy, A., Egorov, A., Chini, L., Justice, C., Townshend, J., 2013. High-resolution global maps of 21st century forest cover change. Science 342, 850–853. Harris, R., 2017. The eBorneo Knowledge Fair: Community Engagement for Impact. UNIMAS Publisher, Kuching, pp. 1–154. Harris, I., Jones, P., Osborn, T., Lister, D., 2014. Updated high-resolution grids of monthly climatic observations – the CRU TS3.10 Dataset. Int. J. Climatol. 34, 623–642. Howe, P., Markowitz, E., Lee, T., Ko, C., Leiserowitz, A., 2012. Global perceptions of local temperature change. Nat. Clim. Change 3, 352–356. Howe, P., Thaker, J., Leiserowitz, A., 2014. Public perceptions of rainfall change in India. Clim. Change 127, 211–225. Howe, P., Leiserowitz, A., 2013. Who remembers a hot summer or a cold winter? The asymmetric effect of beliefs about global warming on perceptions of local climate conditions in the U.S. Global Environ. Change 23, 1488–1500. Jessup, T., Vayda, A., 1988. Dayaks and forests of Interior Borneo. Expedition 30 (1), 5–17. Joshi, L., Kusuma, W., Martua, S., Elok, M., 2004. Indigenous System and Ecological Knowledge among Dayak People in Kutai Barat, East Kalimantan - A Preliminary Report (2004 No. 3). pp. 1–22 Bogor, Indonesia. Kalanda-Joshua, M., Ngongondo, C., Chipeta, L., Mpembeka, F., 2011. Integrating indigenous knowledge with conventional science: enhancing localised climate and weather forecasts in Nessa, Mulanje, Malawi. Phys. Chem. Earth 36, 996–1003. Klintenberg, P., Seely, M., Crhistiansson, C., 2007. Local and national perceptions of environmental change in central Northern Namibia: do they correspond. J. Arid. Environ. 69, 506–525. Koizumi, M., Momose, K., 2007. Penan Benalui wild-plant use, classification and nomenclature. Curr. Anthropol. 48 (3), 454–459. Larcom, S., She, P.W., van Gevelt, T., 2019. The UK summer heatwave of 2018 and public concern over energy security. Nat. Clim. Change 9, 370–373. Lasimbang, J., 2015. 2016 Yearbook: indigenous peoples in Malaysia. In: International Work Group for Indigenous Affairs, pp. 273-279. Lee, T., Markowitz, E., Howe, P., Ko, C., Leiserowitz, A., 2015. Predictors of public climate change awareness and risk perception around the world. Nat. Clim. Change 5, 1014–1020. Lyndon, N., Er, A., Sivapalan, S., Ali, H., Rosniza, A., Azima, M., Junaidi, A., Fuad, M.,
Declaration of Competing Interest None. Acknowledgments We would like to thank Claudia Canales Holzeis, seminar participants at the Asia Research Institute, National University of Singapore and two anonymous referees for their helpful comments on an earlier draft. We are thankful to Ezra Uda for his help in designing the data collection protocols and to Christy Wong for her research assistance. We especially thank the communities of Long Luteng, Long Win, Long Latei, Long Leng, Long Kerangan, Long Jenalong and Long Urang for their hospitality and for their valuable time and effort. This work was supported by the University of Hong Kong's Seed Fund for Basic Research (Ref: 201711159040). Ethical approval was granted by the Human Research Ethics Committee at the University of Hong Kong (Ref: EA1710019). Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.gloenvcha.2019.101974. References Adger, W., 2003. Social capital, collective action and adaptation to climate change. Econ. Geogr. 79, 386–404. Alessa, L., Kliskey, A., Williams, P., Barton, M., 2008. Perception of change in freshwater in remote resource-dependent Arctic communities. Global Environ. Change 18, 153–164. Ashraf, M., Routray, J., 2013. Perception and understanding of drought and coping strategies of farming households in north-west Balochistan. Int. J. Disaster Risk Reduct. 5, 49–60. Berkes, F., Jolly, D., 2001. Adapting to climate change: social-ecological resilience in a Canadian Western Arctic Community. Conserv. Ecol. 5 (2). Boissiere, M., Locatelli, B., Sheil, D., Padmanaba, M., Sadjudin, E., 2013. Local perceptions of climate variability and change in tropical forests of Papua, Indonesia. Ecol. Soc. 18 (4). Bola, G, Mabiza, C., Goldin, J., Kujinga, K., Nhapi, I., Makurira, H., Mashauri, D., 2014. Coping with droughts and floods: a case study of Kanyemba, Mbire District, Zimbabwe. Phys. Chem. Earth 67, 180–186. Bridges, K., McClatchey, W., 2009. Living on the margin: ethnoecological insights from Marshall Islanders at Rongelapatoll. Global Environ. Change 19, 140–146. Brosius, J., 2006. What counts as local knowledge in global environmental assessments and conventions? In: Reid, W., Berkes, F., Wilbanks, T., Capistrano, D. (Eds.), Bridging Scales and Knowledge Systems: Concepts and Applications in Ecosystem Assessment. Island Press, Washington, pp. 129–144. Bryan, E., Deressa, T., Gbetibouo, G., Ringler, C., 2009. Adaptation to climate change in Ethiopia and South Africa: options and constraints. Environ. Sci. Policy 12, 413–426. Byg, A., Salick, J., 2009. Local perspectives on a global phenomenon. Climate change in Eastern Tibetan villages. Glob. Environ. Change 19, 156–166. Chaudhary, P., Bawa, K., 2011. Local perceptions of climate change validated by scientific evidence in the Himalayas. Biol. Lett. 7, 767–770. Chen, P.C.Y., 2016. The Penan: Forest Nomads of Sarawak in Transition. Self published, pp. 1–95 ISBN: 978-967-14242-0-9. Cheuk, S., Atang, A., Lo, M.C., 2012. Community attitudes towards the telecentre in Bario,
10
Global Environmental Change 58 (2019) 101974
T. van Gevelt, et al. Hussein, M., Helmi, A., 2013. The world-view of Penan community on quality of life. Asian Soc. Sci. 9 (14), 98–105. Marin, A., 2010. Riders under storms: contributions of nomadic herders’ observations to analysing climate change in Mongolia. Global Environ. Change 20, 162–176. McCright, A., Dunlap, R., Xiao, C., 2014. The impacts of temperature anomalies and political orientation on perceived winter warming. Nat. Clim. Change 4, 1077–1081. McNeeley, S., Shulski, M., 2011. Anatomy of a closing window: Vulnerability to changing seasonality in Interior Alaska. Global Environ. Change 21, 464–473. Meze-Hausken, E., 2004. Contrasting climate variability and meteorological drought with perceived drought and climate change in northern Ethiopia. Clim. Res. 27, 19–31. Minority Rights Group International (2018). Indigenous Peoples and Ethic Minorities in Sarawak. https://minorityrights.org/minorities/indigenous-peoples-and-ethnicminorities-in-sarawak/. Accessed: April 2019. Nelson, J., Muhammed, N., Rashid, R., 2016. Native customary rights: does it hold the future of Sarawak's Natives? J. For. Environ. Sci. 32 (1), 82–93. Nkomwa, E., Kalanda-Joshua, M., Ngongondo, C., Monjerezi, M., Chipungu, F., 2014. Assessing indigenous knowledge systems and climate change adaptation strategies in agriculture: a case study of Chagaka Village, Chikhwawa, Southern Malawi. Phys. Chem. Earth 67-69, 164–172. Ostrom, E., 2010. Polycentric systems for coping with collective action and global environmental change. Global Environ. Change 20, 550–557. Papworth, S., Rist, J., Coad, L., Milner-Gulland, E., 2009. Evidence for shifting baseline syndrome in conservation. Conserv. Lett. 2, 93–100. Pyhala, A., Fernandez-Llamazares, A., Lehvavirta, H., Byg, A., Ruiz-Mallen, I., Salpeteur, M., Thornton, T., 2016. Global environmental change: local perceptions, understandings and explanations. Ecol. Soc. 21 (3). Ramirez-Villegas, J., Challinor, A., 2012. Assessing relevant climate data for agricultural applications. Agric. For. Meteorol. 161, 26–45. Randell, H., Gray, C., 2016. Climate variability and educational attainment: evidence from rural Ethiopia. Global Environ. Change 41, 111–123. Rudiak-Gould, P., 2014. The influence of science communication on indigenous climate change perception: theoretical and practical implications. Hum. Ecol. 42, 75–86. Salehyan, I., Hendrix, C., 2014. Climate shocks and political violence. Global Environ. Change 28, 239–250. Salick, J., Ross, N., 2009. Traditional peoples and climate change. Global Environ. Change 19, 137–139. Savo, V., Lepofsky, D., Benner, J., Kohfeld, K., Bailey, J., Lertzman, K., 2016. Observations of climate change among subsistence-oriented communities around the world. Nat. Clim. Change 6, 462–474. Siahaya, M., Hutauruk, T., Aponno, H., Hatulesila, E., Afif, J., Mardhanie, B., 2016. Traditional ecological knowledge on shifting cultivation and forest management in East Borneo. Int. J. Biodivers. Sci. 12, 1–23. Siew, S.T., Yeo, A., Zaman, T., 2013. Participatory action research in software development: indigenous knowledge management systems case study. In: Kurosu, M. (Ed.), Human-Computer Interaction. Springer-Verlag, Berlin.
Simons, D., Rensink, R., 2005. Change blindness: Past, present and future. Trends Cogn. Sci. 9, 16–20. Sovacool, B., Bulan, L., 2013. They'll be damned: the sustainability implications of the Sarawak Corridor of Renewable Energy (SCORE) in Malaysia. Sustainability Sci. 8, 121–133. Sovacool, B., Valentine, S., 2011. Bending bamboo: Restructuring rural electrification in Sarawak, Malaysia. Energy Sustain. Dev. 15, 240–253. Speranza, C., Kiteme, B., Ambenje, P., Wiesmann, U., Malaki, S., 2010. Indigenous knowledge related to climate variability and change: Insights from droughts in semiarid areas of former Makueni District, Kenya. Clim. Change 100, 295–315. Turner, N., Clifton, H., 2009. It's so different today: Climate change and indigenous lifeways in British Columbia, Canada. Global Environ. Change 19, 180–190. Valdivia, C., Seth, A., Gilles, J., Garcia, M., Jimenez, E., Cusicanqui, J., Navia, F., Yucra, E., 2010. Adapting to climate change in Andean Ecosystems: Landscapes, capitals, and perceptions shaping rural livelihood strategies and linking knowledge systems. Ann. Assoc. Am. Geogr. 100, 818–834. van den Besselaar, E., Sanchez-Lorenzo, A., Wild, M., Klein Tank, A., de Laat, A., 2015. Relationship between sunshine duration and temperature trends across Europe since the second half of the twentieth century. J. Geophys. Res. 120, 10823–10836. van Gevelt, T., Canales Holzeis, C., George, F., Zaman, T., 2017. Indigenous community preferences for electricity services: Evidence from a choice experiment in Sarawak, Malaysia. Energy Policy 108, 102–110. van Gevelt, T., 2019. Indigenous communities, ICT and rural development: case studies in Tanzania and Sarawak, Malaysia. In: Anand, P.B. (Ed.), Handbook on BRICS and Emerging Markets. Oxford University Press, Oxford. Weihreter, E., 2014. Traditional Knowledge, Perceptions and Forest Conditions in a Dayak Mentebah Community, West Kalimantan, Indonesia (No. 146). pp. 1–38 Bogor Barat, Indonesia. Weisshaidinger, R., Petrus, S., Beckel, J., 2012. Resource Protection through Sustainable Agriculture in an Upper Baram Penan Village (Sarawak/Malaysia). Research Institute of Organic Agriculture (FiBL), Vienna, Austria, pp. 1–60. West, C., Roncoli, C., Ouattara, F., 2008. Local perceptions and regional climate trends on the central plateau of Burkina Faso. Land Degrad. Dev. 19, 289–304. Zaman, T., Yeo, A., Kulathuramaiyer, N., 2013. Augmenting indigenous knowledge management with information and communication technology. Int. J. Serv. Technol. Manage. 19, 137–148. Zaman, T., Winschiers-Theophilus, H., et al., 2015. Penan's Oroo’ Short Message Signs (PO-SMS): co-design of a digital jungle sign language application. In: Abascal, J. (Ed.), Human-Computer Interaction – INTERACT 2015. Lecture Notes in Computer Science, vol. 9297. Springer, pp. 489–504. Zaman, T., Jengan, G., Winschier-Teophilus, H., 2019. Oroo’ Language of the Forest: Ha’ Jin Tana. UNIMAS Publisher, Kuching. Zanocco, C., Boudet, H., Nilson, R., Satein, H., Whitley, H., Flora, J., 2018. Place, proximity and perceived harm: extreme weather events and views about climate change. Clim. Change 149, 349–365.
11