Enigmatic declines in bird numbers in lowland forest of eastern Ecuador may be a consequence of climate change

Study site

Research was conducted at Tiputini Biodiversity Station (TBS), Orellana Province, Ecuador (ca 0°37′S, 76°10′W, 190–270 m above sea level). TBS is located adjacent to Yasuní National Park on a tract of largely undisturbed (by human activities) lowland rain forest within Yasuní Biosphere Reserve, one of the most diverse regions of the world (Bass et al., 2010). The biosphere reserve encompasses Yasuní National Park (∼982,000 ha), the adjacent Waorani Ethnic Reserve (∼800,000 ha), and a 10 km buffer zone. TBS is located approximately 22 km east of an offshoot of the Maxus Road, ∼25 km east of the nearest Waorani settlement, ∼12 km from an oil access road to the northwest, ∼20 km south of the Napo River, and approximately 17–19 km from the nearest settlements along the Napo River. Thus, the station is surrounded by extensive areas of intact forest. The station and nearby areas are dominated by terra firme forest; várzea forest, palm swamps, and various successional habitats also are present. Mean annual precipitation at Yasuní Research Station, approximately 28 km west of TBS, is about 3,100 mm.

Two ca 100-ha plots (ca 1 km × 1 km each) were established in terra firme forest during 2001. Both plots are gridded (100 × 200-m grid lines) and marked with 1.5-m PVC tubes. The Harpia plot ranges from ∼201 to 233 m elevation and is characterized by more dissected upland forest. The Puma plot is flatter overall although elevation range is similar, from ∼209 to 235 m. Both areas experience partial inundation when small streams back up as the Tiputini River rises; Puma has more areas that fill with persistent standing water during the rainy season. Dominant vegetation on both plots is tall, evergreen forest.

Bird sampling

Analyses

Sample effort (number of mist-net hours, where one mist net open 1 h equals 1 mist-net-hour or 1 mnh) varied slightly among samples. Consequently, comparisons of numbers of captures were based on captures per 100 mnh. Similarly, observations were expressed as number of birds per 100 m or 1,000 m to standardize numbers by effort. Species were classified into foraging guilds using categories of Terborgh et al. (1990). We used correlation coefficients to examine patterns of change in overall abundance, abundances of different guilds, and abundances of individual species represented by at least 140 captures or observations (or at least 100 on a single plot). We used correlation coefficients to examine patterns of change in relation to year and to Southern Oscillation Index (SOI) values with and without a 1 or 2-year lag. We used annual SOI values and values from October to March, a period that encompasses the main breeding period at TBS and that overlaps the sampling period (January through March). We used 1- and 2-year time lags under the assumption that if changes in climate affected reproduction and/or survival, those effects might not be felt immediately but rather would accumulate over time. SOI values were obtained from National Weather Service, Climate Prediction Center, National Oceanic and Atmospheric Administration (http://www.cpc.ncep.noaa.gov/data/indices/). Comparisons between plots were based on correlations and paired t-tests as appropriate. Correlations are used to indicate strength of association, without necessarily attributing a P-value. All statistical analyses were run with Statistix 10.0 (Analytical Software, 2013).

Approval for this research was obtained from Non-Regulatory Animal Research Committee (018-10WEC) at University Florida and, previously, from IACUC, University of Missouri. Work at Tiputini Biodiversity Station was conducted in accordance with research permit number 13-IC-FAU-DFN (and subsequent renewals), Ministerio del Ambiente, Distrito Forestal Napo, Tena, Ecuador.

Overall summary

We accumulated 7,302 captures (2,780 recaptures) of 155 species in 14,862 mnh on Harpia and 7,598 captures (2,938 recaptures) of 154 species in 14,263 mnh on Puma; 178 species were captured in total. We did not use mist nets to sample birds on either plot during March 2013 or on Puma during 2014 because heavy rains reduced our opportunities for netting. We accumulated 20,391 records of 276 species on Harpia and 17,427 records of 275 species on Puma; 306 species were recorded on both plots combined.

All species combined

Annual capture rates (birds 100-mnh⁻¹) were highly correlated between plots (Table 1) but were somewhat higher overall on Puma (Harpia: mean = 48.7 ± 2.1 SE, CV = 15.8; Puma: 52.1 ± 1.9, CV = 13.7; paired t-test, t = 3.91, df = 13, P < 0.002). Although overall capture rates declined from 2001 to 2014 on both plots (Table 1), declines were most pronounced from 2009 to 2014 (r = − 0.47 and −0.23 from 2001 to 2009 on Harpia and Puma respectively; r = − 0.92 and −0.94 from 2009 to 2014 on Harpia and Puma, respectively) (Fig. 1A). Capture rates on Puma in 2009, for example, were as high as they were in 2001.

Overall observation rates also were correlated between plots (r = 0.93, P < 0.001) but showed no difference between plots (paired t-test, t = 0.9, df = 7, P = 0.40; Fig. 1B). Observation rates were correlated with capture rates on each plot (Harpia: r = 0.87, P < 0.005; Puma: r = 0.74, P < 0.05). Observation rates did not vary significantly from 2003 to 2006 (Fig. 1B) but were much lower in later years (correlations across all years combined Harpia: r = − 0.90, P < 0.005; Puma: r = − 0.93, P < 0.001); rates in 2014 were approximately half those observed during the first years of the study.

Foraging guilds

Capture rates of all foraging guilds examined (except arboreal omnivores on Puma) declined from 2001 to 2014 (Fig. 2 and Table 1) but the extent of the decline varied among guilds and over time. Arboreal and terrestrial insectivores showed the steepest declines, particularly from 2009 to 2014; declines were not significant from 2001 to 2009 but were subsequently. Army-ant followers also showed steeper declines after 2009, with substantial fluctuations during the first years of the study. Mixed-species flocks also were quite variable but capture rates were substantially less in 2013 than in previous years; capture rates increased somewhat in 2014.

In contrast, frugivores declined in captures somewhat from 2001 to 2014 but unlike most insectivores, declines were not pronounced from 2009 on. Similarly, capture rates of nectarivores declined over time but correlations were much weaker than for insectivores. Correlations in capture rates between plots were strongest for terrestrial insectivores and somewhat less so for arboreal insectivores and mixed-species flocks; other guilds showed very different patterns between plots.

Observations of most foraging guilds also declined over time (Fig. 3 and Table 1). Declines were particularly pronounced among most insectivores, arboreal omnivores, and arboreal granivores. Declines were not, however, as consistent among dead-leaf foragers, bark insectivores, terrestrial frugivores, and nectarivores. Both canopy and understory flocking species declined substantially whereas army-ant followers were more variable but nonetheless lower in later years. Patterns of observations were similar between the two plots for most guilds, with the exceptions of terrestrial frugivores, army-ant followers, and to a lesser extent, terrestrial granivores, bark insectivores, and nectarivores.

Individual species

We examined variation in 28 species represented by at least 140 captures. Although capture rates of most species declined over time (Fig. 4A), capture rates also varied from one year to the next, often substantially, and often showed different patterns between plots (Fig. 4B, see also Fig. S1 and Table S1). Declines in capture rates were particularly pronounced after about 2009 for many species (Fig. S1). Species that showed particularly pronounced declines on both plots included a frugivore (Lepidothrix coronata), terrestrial insectivores (Sclerurus caudacutus, Formicarius spp.), and understory insectivores (Willisornis poecilinota, Myrmoborus myotherinus); for these species, capture rates were correlated between plots (Table S1). In addition, several species showed large declines only on one plot: Automolus infuscatus (flocks) and Hylophilus naevius (understory insectivore) on Puma; Glyphorynchus spirurus (bark) and Conopophaga peruviana (ground) on Harpia. One understory flycatcher, Platyrinchus coronatus, declined on Harpia but increased on Puma. Thus, declines were, overall, most pronounced among insectivores.

We examined variation in 68 species or genera that were observed at least 140 times, or if less, at least 100 times on the plot examined (e.g., Grallaria dignissima was common on Puma but rarely seen on Harpia; the reverse was true for Lipaugus vociferans). Observations of most species declined over time and most were correlated between plots (Figs. 4C, 4D and Table S1). Species that showed significant declines included highly vocal and conspicuous species (e.g., Ramphastos), vocal but less conspicuous species (e.g., Campylorhynchus turdinus), and less vocal or conspicuous species (e.g., Lepidothrix coronata, Euphonia spp.) (Fig. S2).

Southern oscillation index

Annual SOI values were lowest from 2002 to 2004 and highest from 2008 to 2011 (Fig. 5). Values for October through March, which encompasses the main breeding period of many species, followed a generally similar trend. Monthly values (Fig. S3) also demonstrated the tendency for SOI values to be greater in the later years of this study. We have both observation data and netting data for eight years; those rates were negatively correlated with SOI values for Oct–March, with a 2-yr time lag (r = − 0.62, −0.71, −0.71, and −0.60 for captures on Harpia, captures on Puma, observations on Harpia, and observations on Puma, respectively). Correlations with annual SOI values, with a 2-yr time lag, were weaker but still negative (−0.50, −0.50, −0.64, and −0.67, respectively). Capture rates were not correlated with SOI values when all 14 years were included.

Methodological considerations

Capture rates of birds often decline over time if an area is repeatedly netted, but those declines typically are seen over short time periods, such as when the same net locations are sampled for several days in a row (J Blake, pers. obs., 1985–1994). In our case, the same net locations were used each year but each net was opened for only one morning in January and one in March. Recapture rates are reasonably high (∼40%) suggesting that avoidance of nets is not a significant factor. Further, although capture rates declined somewhat during the first few years, those declines were offset by subsequent increases which would not be expected if netting per se had a long-term negative impact. Further, substantial declines (by about 50%) also were seen in observation data, which included data from many species that are not captured in nets.

All observations were done by one of us (JGB), so variation among observers is not an issue or possible cause of declines. If, on the other hand, ability to detect birds declined over time (i.e., because of loss of hearing) that might account for some of the observed declines. Yet, we would expect such changes to be most pronounced among species with high frequency songs that often are more difficult to detect as one gets older. In this study, increases in abundance were noted in some species with high frequency songs (e.g., Platyrinchus coronatus) and such species are still readily detected by JGB both in tropical and temperate forests (J Blake, pers. obs., 2015). Further, many of the species that have shown declines are loud and conspicuous, so a gradual decline in ability to detect such species seems an unlikely explanation. If uncontrolled variables related to environmental conditions or to observer abilities affect detection of species, we might expect that changes in abundance over time would not show a consistent pattern of decline but would, instead, vary in a more random fashion; such variation was not observed.

Habitat effects

Declines are not likely to have occurred in response to habitat changes as there have been no substantial changes either at our study site or in surrounding regions during the course of this study. Tree falls and blowdowns are a normal and typical form of disturbance which might cause some small-scale shifts in spatial distribution of birds or other organisms. Yet, given that home ranges of most bird species in the area are larger than typical gaps, such disturbances would not be expected to cause large, consistent declines in bird populations at the scale of our study plots. Other forms of disturbance in eastern Ecuador are associated with oil exploration and extraction, which have increased in the general region of Yasuní (Finer et al., 2008). Such activities can have negative impacts on species as a consequence of road construction, noise, and other disturbances (Canaday, 1997). Yet, there has not been a substantial increase in such activity during the course of this study, at least in the area of our study plots. Similarly, roads may facilitate hunting by indigenous groups (e.g., Waorani) and hunting has caused declines in bird and mammal populations in some areas of Yasuní (Suárez et al., 2009). Although some hunting has occurred in the vicinity of the station (J Blake, pers. obs., 2014), the increase has not been substantial. Further, declines occurred in many species that are not targeted by hunters (e.g., Lepidothrix, Formicarius). Thus, we see no evidence that habitat loss or alteration, and associated human activities, may have caused the observed declines in bird numbers.

Resource abundance

Responses of species to ENSO events may vary with foraging behavior and diet (Wolfe & Ralph, 2009). We do not, however, have concurrent measures of resource abundance (e.g., fruit, insects) so we cannot directly evaluate the potential importance of resource availability on populations. Yet, some results suggest that changes in resources may have played a role in at least some of the observed declines. Although declines in abundance occurred across a broad range of species and trophic guilds, declines in abundance were particularly pronounced among many insectivorous groups. Terrestrial insectivores, for example, showed particularly pronounced declines in abundance; such insectivores also are among those most negatively affected by habitat fragmentation (Powell, Cordeiro & Stratford, 2015). Disappearance from small patches of habitat may be related to reductions in availability of leaf-litter insects, suggesting that such species are sensitive to loss of resources. Understory insectivores often are particularly vulnerable to disturbances (Powell, Cordeiro & Stratford, 2015) and many also declined in abundance in this study. Changes in precipitation can affect understory insectivores in various direct and indirect ways through effects on resource (insect) abundance (Powell, Cordeiro & Stratford, 2015), which might account for some of the observed declines. Declines were less general among most frugivores and nectarivores, suggesting that resources fluctuate in different ways. Lack of sufficient resources might lead to reductions in either nesting attempts or nesting success, with consequences for recruitment. When combined with typical survival rates of 3 to 5 years for many species in the understory (Blake & Loiselle, 2013), lack of recruitment could lead to subsequent declines in overall abundance.

Disease

Blood parasites such as Plasmodium and Haemoproteus (haemosporidia) and other diseases (e.g., West Nile virus) can be harmful to their hosts (Van Riper III et al., 1986; Merino et al., 2000) and, in some cases, can cause large-scale population declines and even extinction (e.g., Plasmodium relictum and Hawaiian native avifauna; Warner (1968); Van Riper III et al. (1986). In a previous study based on data collected from our study plots (Svensson-Coelho et al., 2013), overall prevalence of blood parasites among species ranged from a low of 5.6% in Pipra filicauda to 91.2% in Formicarius colma. All well-sampled species were infected, suggesting that infection is widespread among species at our site. Further, prevalence varied among years, from an overall prevalence of 14.5% in 2006 to a high of 33% in 2009; the increase in 2009 was widespread across species but was followed by a decline in overall prevalence in 2010. If the increase in infection rates had negative consequences for reproduction or survival, this might account for some of the declines. Although low level, chronic infections are generally considered benign, chronic malaria can, in some cases, have negative consequences for lifespan and reproductive success (Asghar et al., 2015). Several species with particularly high infection rates, such as Automolus infuscatus (59%), Formicarius analis (54%), and Formicarius colma (91%) (Svensson-Coelho et al., 2013), were among those species that showed some of the steepest declines in abundance. Lepidothrix coronata (16.3%) had the highest prevalence among manakins and also showed the greatest decline in abundance among members of that family. Yet, some species with relatively low infection rates, such as Pithys alba (6%) and Willisornis poecilinota (24%), also showed substantial declines in abundance. Increased levels of infection might make some individuals more susceptible to other factors, such as reduced availability of resources, leading to population declines. At present, we do not have any direct evidence, however, to link disease to declining numbers of any particular species.

Climate

Warming temperatures cause species to move to higher elevations (Forero-Medina et al., 2011; Anderson et al., 2013; Freeman & Freeman, 2014), which could lead to declines in abundance at lower elevations. If temperature increases in lowland forest are sufficient to affect reproduction and survival of many species, then populations could certainly decline, particularly given that there would be little opportunity to move to higher elevations or latitudes (see Colwell et al., 2008). Global temperature has apparently increased at a rate of ∼0.2 °C/decade since 1975 (Hansen et al., 2006) with an increase of ∼0.25°C/decade in lowland tropics (cited in Colwell et al., 2008). If correct, then we might expect to see increases of ∼0.12 to ∼0.15°C during 5–7 years (i.e., from 2008 to 2014). It is unlikely that such an increase would cause the rapid population declines seen in this study. Thus, we do not think that warming temperatures have been the cause of the observed changes.

What has changed during the course of this study has been the strength and frequency of La Niña events, as shown by variation in SOI. Particularly pronounced events occurred in the latter part of the study and show some correlation with population declines. Extreme La Niña events, such as that of 2010–2011, can bring excessive rains, which can cause widespread damage from flooding (e.g., Hoyos et al., 2013; see also Restrepo & Kjerfve, 2000) and which might also affect reproductive behavior of birds. Changes in rainfall associated with changes in climate might affect birds directly (through effects on reproductive success; Stiles, 1992) or indirectly through effects on resource abundance, body condition, predation, or susceptibility to disease. Reduction in reproductive success would lead to population declines as adult birds die and are not replaced with new recruits. In lowland tropical wet forests, such as our site, nest failure is already high (Ryder et al., 2008), so further reductions in nesting attempts or success might be expected to have noticeable effects on recruitment. Previous studies have noted changes in reproductive success, survival, and abundance associated with changes in global circulation patterns (e.g., LaManna et al., 2012; García-Pérez et al., 2014). In our area, stronger La Niña events are associated with increased rainfall during the period of time when many birds typically are nesting, and when our samples (nets and observations) occur. Given that observed declines in abundance of many species coincided with a period with considerably higher SOI values (La Niña events), it is at least plausible that this change in climate patterns could account for observed declines in bird populations. Although the mechanism through which climate change acts is yet unknown, if changes in climate are indeed driving population declines, then as conditions change in the future (if they do), we might expect to see an increase in bird populations over time.