Effects of Human Subsistence Activities on Forest Birds in Northern ...

Contributed Paper

Effects of Human Subsistence Activities on Forest Birds in Northern Kenya LUCA BORGHESIO Department of Biological Sciences (M/C 066), University of Illinois at Chicago, 845 W. Taylor Street, Chicago, IL 60607-7060, U.S.A. and Department of Zoology, National Museums of Kenya, Box 40658-00100, Nairobi, Kenya, email [email protected]

Abstract: Indigenous tribes and conservation biologists may have common goals and may be able to collaborate on the maintenance of biodiversity, but few researchers have evaluated the impacts and potential benefits of human subsistence activities. I studied the effects of subsistence activities (primarily wood collection) of nomadic pastoralists in 3 Afromontane forests of northern Kenya. In surveys of 404, 25-m-radius plots, I recorded vegetation structure and composition of the forest bird community. Plots with higher levels of human activity had significantly different vegetation structure, with more-open canopies, more grass, and fewer tree stems. Nectarivores (abundance +231%) and aerial insectivores (+66%) were more abundant in plots with more-intense wood collecting than in plots with less human activity, whereas abundance of forest specialists (−28%) decreased in plots with more-intense human activity. Abundance of 58% of the bird species either increased or decreased significantly in plots with more-intense human activity. Generally, the number of individuals of forest specialists decreased (6 of 7 species showed significant responses) and the number of individuals of edge and nonforest species increased with increasing human activity. Canonical correspondence analysis showed that an intensification of human activities would favor nectarivores, aerial insectivores, granivores, and omnivores and would negatively affect large-sized, ground-foraging species and arboreal frugivores. Subsistence human activities favored the invasion of forest by edge species at the expense of forest specialists; thus, further intensification of forest exploitation by local peoples is not recommended. At the same time, however, subsistence activities in northern Kenya forests appeared to increase the structural diversity of the vegetation and provided suitable habitat for part (but not all) of the forest avifauna, which suggests that subsistence human activities may have a role in the maintenance of bird diversity. Keywords: Afromontane forest, forest birds, nomadic pastoralists, subsistence human activities, wood collection Efecto de las Actividades Humanas de Subsistencia sobre Aves de Bosque en el Norte de Kenia

Resumen: Las tribus nativas y los bi´ologos de la conservaci´on pueden tener metas comunes y pueden colaborar en el mantenimiento de la biodiversidad, pero pocos investigadores han evaluado los impactos y beneficios potenciales de las actividades humanas de subsistencia. Estudi´e los efectos de las actividades de subsistencia (principalmente la colecta de le˜ na) de pastores n´ omadas en 3 bosques Afromontanos en el Norte de Kenia. Registr´e la estructura de la vegetaci´ on y la composici´ on de la comunidad de aves de bosque en 404 parcelas de 25 m de radio. La estructura de la vegetaci´ on en las parcelas con mayor nivel de actividad humana fue significativamente diferente, con el dosel m´ as abierto, m´ as pasto y menos tallos de a ´ rbol. Los nectar´ıvoros (abundancia +231%) e insect´ıvoros a´ereos (+66%) fueron m´ as abundantes en parcelas con colecta intensiva de le˜ na que en parcelas con menos actividad humana, mientras que la abundancia de especialistas de bosque (−28%) disminuy´ o en parcelas con mayor nivel de actividad humana. La abundancia de 58% de las especies de aves aument´ o o disminuy´ o significativamente en parcelas con mayor intensidad de actividad humana. En general, el n´ umero de individuos de especialistas de bosque disminuy´ o (6 de 7 especies mostraron respuestas significativas) y el n´ umero de individuos de especies de borde y no forestales increment´ o con el aumento de las actividades humanas. El an´ alisis de correspondencia can´ onica mostr´ o que la intensificaci´ on de las actividades humanas podr´ıa favorecer a los nectar´ıvoros, insect´ıvoros a´ereos, gran´ıvoros y omn´ıvoros y

Paper submitted April 12, 2007; revised manuscript accepted September 12, 2007.

384 Conservation Biology, Volume 22, No. 2, 384–394  C 2008 Society for Conservation Biology DOI: 10.1111/j.1523-1739.2007.00872.x

Borghesio

385

afectar´ıa negativamente a especies terrestres de talla grande y a frug´ıvoros arb´ oreos, Las actividades humanas de subsistencia favorecieron la invasi´ on del bosque por especies de borde a expensas de especialistas de bosque; por lo tanto, no se recomienda la intensificaci´ on de la explotaci´ on del bosque por los habitantes locales. Sin embargo, al mismo tiempo, las actividades de subsistencia en los bosques del norte de Kenia parecen haber incrementado la diversidad estructural de la vegetaci´ on y proporcionaron un h´ abitat adecuado para parte (pero no toda) de la avifauna del bosque, lo cual sugiere que las actividades humanas de subsistencia pueden tener un papel en el mantenimiento de la diversidad de aves.

Palabras Clave: actividades humanas de subsistencia, aves de bosque, bosque Afromontano, colecta de le˜na, pastores n´ omadas

Introduction Over large areas of the tropics, difficult access makes intensive, commercially oriented forest exploitation impossible (Matthews et al. 2000). Nomadic human tribes live in these regions, and the impact of their activities is of conservation relevance and has been much debated. Nomadic tribes practice shifting agriculture, graze their herds in the forest, and extract nontimber products— such as building poles, fuel wood, fruits, roots, honey, and bush meat—from the forest (Food and Agriculture Organization 1995). Sustainability is a central issue in the traditional use of forest resources because local peoples have few economic alternatives (Ticktin 2004). Moreover, many tropical forests have been inhabited for millennia, and the biota of many “wild” places may have adapted to this long-lasting interaction between humans and the environment (Gadgil et al. 1993). Indeed, some researchers have found high species diversity in habitats exploited by local tribes (Andrade & Rubio-Torgler 1994; Tsai et al. 2006), and grazing by herds kept by nomadic pastoralists may be essential to the maintenance of biodiversity in African ecosystems (Smith 1992; Brockington 2002). Indigenous tribes may contribute to conservation (Adams & McShane 1992; Berkes et al. 2000), but tests of this hypothesis are rare. Local people can also negatively affect the environment, and there are numerous examples of past and present cultures that caused widespread environmental degradation (Redman 1999; Terborgh 1999). Forests from which nontimber products are extracted can have lower species richness, higher tree mortality, and different species than forests that are less affected by people (Murali et al. 1996; Ganeshaiah et al. 1998). Vegetational changes can in turn influence other ecosystem components, such as populations of frugivorous birds (Moegenburg & Levey 2002). Livestock grazing can also negatively affect forest structure and diversity (Sagar & Singh 2004; Madhusudan 2005), but most grazing studies have focused on the competition between wild and domestic herbivores or on the effects of pastoralism on floral composition and vegetation structure (e.g., Madhusudan

2004; Sagar & Singh 2004). Analyses of the effects on other ecosystem components (such as birds) are few. The goals of conservationists may or may not coincide with those of local peoples and with their rights to exploit the environment where they live. In this respect, the situation in Kenya is emblematic. On one hand, many of the country’s forests are in poor shape, and forest resource use is intense (Wass 1995; Bennun & Njoroge 1999). On the other hand, forests are also important sources of revenue and dry-season grazing (Trench & Makee 1994; Wass 1995). Research on the effects of forest uses by local people is needed if a balance between their use of the forest and conservation goals is to be achieved. I studied an area of northern Kenya inhabited by nomadic tribesmen and explored the effects of their activities on the birds of the Afromontane forest. Focusing on increasing levels of intensity of human activity, I asked, how much do subsistence activities affect vegetation structure and the bird community? and can the consequences of intensification or reduction of use be forecasted? Study Areas Northern Kenya is a semiarid plateau with elevations of 500–1500 m, but on some isolated mountains increased rainfall allows the growth of forests (Herlocker 1979). I studied the birds of 3 such massifs: Mt. Nyiru (2◦ 08 N, 36◦ 51 E; elevation range 2300–2600 m; visited in February 2004; 126 plots surveyed), Mt. Marsabit (2◦ 17 N, 37◦ 57 E; 1200–1600 m; February–March 2000; 137 plots surveyed), and the Karissia Hills (1◦ 03 N, 36◦ 51 E; 1800– 2500 m; October 1999; 141 plots surveyed). Rainfall in the 3 sites is around 1000 mm/year, mainly occurring in late March through May and late October through December (Bake 1983). The 3 sites were visited in the dry season before the onset of rains. The forests of northern Kenya are part of the Afromontane center of endemism (White 1983). Trees such as Olea capensis, Olea europaea, Prunus africana, Cassipourea malosana, Teclea nobilis, and Dovyalis abyssinica are widespread (Bussmann 2002). The canopy is 17 to 25 m high and is interrupted by many

Conservation Biology Volume 22, No. 2, 2008

Subsistence Activities in Kenyan Forests

386

gaps and glades created by humans, treefalls, or large mammals (elephant [Loxodontha africana] and African buffalo [Syncerus caffer]). The bird fauna of the study sites is well known (Zimmerman et al. 1996; Borghesio & Laiolo 2004b). In our study we observed that Nyiru (32 species) and Marsabit (28 species) were less diverse than the larger, less-isolated Karissia Hills (69 species). Despite variation in species richness, however, the 3 sites were similar because the avifauna of the less-diverse sites was a nested subset of the richest site. The region is inhabited by nomadic pastoralists, but only the Samburu (in all areas) and Borana (in Marsabit) regularly exploit forest resources. Population density is estimated at 5 individuals/km2 at Nyiru, 17 at Marsabit, and 18 in the Karissia Hills districts, respectively (Central Bureau of Statistics 2001). Shepherds bring their herds in the forests in the dry season, but some large glades are inhabited permanently. Hunting is illegal and rarely practiced. Poachers target a few species of large mammals (elephant, African buffalo, bushbuck [Tragelaphus striatus]), but no bird. Species, Goats, sheep, cows, and donkeys graze in grassy glades and under continuous canopy, where animals feed on green leaves and young shoots. Shepherds collect wild honey in all months of the year, occasionally dropping large, unscalable trees. Moreover, fires used to scare away bees can run out of control and thus affect small (8 m high). Within 10-mradius plots centered on the census point, I counted the number of trees (woody plants taller than 3 m) in 3 classes

Conservation Biology Volume 22, No. 2, 2008

of diameter at breast height (dbh): 0–5 cm, 5–20 cm, >20 cm. I assessed 2 types of human activity on 2 levels (low and high). I estimated intensity of removal of fuel wood, pole wood, and nontimber products (e.g., barks, plant parts used for wooden utensils; hereafter referred to collectively to as wood collection) by counting cut woody stems and bark scars within 10 m of the observation point (0–5 cut stems, low-intensity wood collection; >5 stems, high-intensity wood collection). In each plot, I also estimated grazing intensity by observing teeth marks on plants in the grass layer and assessing the extent of grazed vegetation in the plot (0–20% grazed area, low-intensity grazing; >20%, high-intensity grazing). Nevertheless, because the intensity of grazing and wood collection were significantly and positively correlated (Spearman’s rho = 0.54, p 20 cm

1.05, 0.31 2.01, 0.16 12.3, < 0.0001 4.09, 0.044 0.033, 0.97

tion of general patterns in the bird community and the habitat where the species live. I used randomization (random reshuffling the rows [plots] of the bird data table; 999 permutations) to test the null hypothesis of no relationship between the 2 tables. The GLM analyses involved multiple testing of related data sets, which inflates chances of false positive (type I) statistical error. This type of error can be corrected for by adjusting significance levels with techniques (e.g., sequential Bonferroni) that constrain the results within a given level of experiment-wise probability. Nevertheless, this procedure produces conservative tests and the risk of not detecting patterns of potential conservation interest. This is not advisable in studies, such as mine, that focus on little-known processes that occur in poorly explored areas. In these cases, it is better to accept a higher chance of type I error in order to produce working hypotheses about the processes under study (Roback & Askins 2005). Thus, I present uncorrected probability levels. On the basis of this reasoning, to highlight responses at the species level, I selected the 43 bird species for which I had >10 observations from at least 1 site and used chi-squared tests to determine whether more or fewer observations than expected were found in plots with different intensities of wood collection. I calculated expected values from the proportion of plots surveyed in high- or lowimpact classes, under the null hypothesis of no influence of human activity on bird distribution among plots. This analysis runs the risk of type I statistical error if a few plots dominate the total sample and skew the results. Nevertheless, sample sizes were small at the species level, precluding the use of analysis of variance. Thus, I used chi-squared tests to generate preliminary hypotheses on the largest possible set of bird species because these hypotheses are highly relevant to conservation.

Conservation Biology Volume 22, No. 2, 2008

Subsistence Activities in Kenyan Forests

388

Results Effects on Habitat Structure Intensity of wood collection was negatively correlated to distance from forest edge (Sperman’s rho = −0.38, p < 0.001) and positively correlated to elevation (Spearman’s rho = 0.34, p < 0.001). After controlling for elevation and distance from edge, plots with more-intense wood collection were more open and had more cover in the grass and less in the subcanopy and canopy layers. Cover in the shrub layer did not change significantly (Table 1, Fig. 1). Densities of tree stems decreased with increasing wood collection, although smaller trees (< 20 cm diameter) were more affected than larger ones (Fig. 1). Effects on Birds I observed 2530 individuals of 77 species. At the guild level, plots with more-intense wood collection had significantly more individual forest generalists (average +30%), forest visitors (+82%), aerial insectivores (+66%), and nectarivores (+231%) than plots with less-intense wood collection. On the contrary, abundance of forest specialists decreased (−28%, Table 2, Fig. 2) with increasing intensity of wood collection. At the species level, 25 of 43 bird species showed significant or nearly significant, either positive or negative, responses to human activity (Table 3). Among forest specialists, the abundance of 6 species decreased and the abundance of 1 species increased in response to higher intensities of wood collection. Among forest visitors, the opposite was observed: 6 species of 8 increased in response to more-intense wood collection. Forest generalists exhibited more balanced responses (6 increased and 4 decreased) to increasing intensity of wood collection. The CCA extracted 2 axes that showed highly significant (randomization test, 0.001) trends of change in the bird community (Fig. 3). Axis 1 (eigenvalue: 0.067; explained variance: 5.7%) was negatively correlated with tree-stem density and vegetation cover in the canopy and subcanopy layers. On the other hand, vegetation cover in the grass and shrub layers and intensity of wood collection correlated positively to this axis. Thus, axis 1 was a gradient of increasing forest openness and human activity, ranging from dense arboreal vegetation with little or no human activity to progressively more open, grassy glades with intense human activity. Forest specialists, ground insectivores, and frugivores had low CCA axis 1 scores, whereas nectarivores were located on the positive end. Axis 2 (eigenvalue: 0.033, explained variance: 2.9%) represented a gradient of forest maturity, ranging from areas with large-sized trees and dense canopy to young forest patches rich in shrubs and undergrowth, but had little correlation with intensity of wood collection. Frugivores were located toward the negative side of axis 2, whereas forest visitors increased on the opposite end. Conservation Biology Volume 22, No. 2, 2008

Figure 1. Effects of wood collection on percent plant cover in 4 vegetation layers and on the number of tree stems in 3 classes of diameter at breast height (dbh) at 3 sites in northern Kenya. The x-axes show increasing intensity (from left to right) of wood collection. Data points are slightly staggered horizontally to improve readability. The graphs show back-transformed least-square means and 95% CI predicted by general linear models in the 3 sites, controlling for 2 covariates (distance from forest edge and elevation).

Discussion I focused on the effects of wood collection, a widespread type of forest use, whose intensity was also positively correlated to grazing by domestic livestock. Because I studied multiple areas within a large region, it is not surprising that I found much variability among sites, but my results highlighted some robust general patterns and contribute to the debate on whether local tribes can be an ally to conservation. In northern Kenya the answer appears to depend on finding a correct balance. Traditional human activities are a tool to be used in the proper

Borghesio

389

Table 2. General linear models of the effects wood collection on total bird abundance and the abundance of 3 habitat and 8 foraging guilds.

Response variable Model factor Weather Distance from edge Time of the day Elevation Shrub percent cover Site Wood collection Site × wood collection

df 1,393 1,393 1,393 1,393 1,393 2,393 1,393 2,393

total abundance 0.51, 0.47 14.5,