Short-rotation woody cropping (SRWC) refers to silvicultural systems designed to produce woody biomass using short harvest cycles (1–15 years), intensive silvicultural techniques, high-yielding varieties, and often coppice regeneration. Recent emphasis on alternatives to fossil fuels has spurred interest in producing SRWC on privately owned and intensively managed forests of North America. We examined potential bird and small mammal response at the stand level to conversion of existing, intensively managed forests to SRWCs using meta-analysis of existing studies. We found 257 effect sizes for birds (243 effect sizes) and mammals (14 effect sizes) from 8 studies involving Populus spp. plantations. Diversity and abundance of bird guilds were lower on short-rotation plantations compared with reference woodlands, while abundance of individual bird species was more variable and not consistently higher or lower on SRWC plantations. Shrub-associated birds were more abundant on SRWC plantations, but forest-associated and cavity-nesting birds were less abundant. Effects on birds appeared to decrease with age of the SRWC plantation, but plantation age was also confounded with variation in the type of reference forest used for comparison. Both guilds and species of mammals were less abundant on SRWC plantations. These conclusions are tentative because none of these studies directly compared SRWC plantations to intensively managed forests. Plantations of SRWCs could contribute to overall landscape diversity in forest-dominated landscapes by providing shrubby habitat structure for nonforest species. However, extensive conversion of mature or intensively managed forests to SRWC would likely decrease overall diversity, especially if they replace habitat types of high conservation value.
Short-rotation woody cropping (SRWC) refers to silvicultural systems designed to produce woody biomass using short harvest cycles (1–15 years), intensive silvicultural techniques (e.g., fertilization, irrigation and competition control), high-yielding varieties, and often coppice regeneration (Dickmann, 2006). Many species used for SRWC resprout vigorously after harvest. In North America, Populus spp. (poplars and cottonwood) and willow (Salix spp.) currently show promise, and most research on both production and biodiversity response to SRWCs has focused on these two species. Other candidate SRWCs include loblolly pine (Pinus taeda), alder (Alnus sp.), black locust (Robinia pseudoacacia), silver maple (Acer saccharinum), sycamore (Platanus occidentalis), Eucalyptus spp., and sweetgum (Liquidambar styraciflua), although much less is known about their biofuel potential and relationship to biodiversity (Phillips et al., 1995; Dickmann, 2006). SRWC systems may be profitable in the southeastern, midwestern, Pacific-northwest and boreal regions of North America (Weih, 2004; Dickmann, 2006; Dale et al., 2010).
Using SRWCs for bioenergy is not a new concept (e.g., Christian et al., 1994), but fluctuations in fossil fuel prices, social pressure to develop alternate energy sources, government incentives, and emerging carbon markets have renewed interest in SRWCs on privately owned and intensively managed forests of North America. Biodiversity response to SRWCs will depend in large part on what type of habitat is replaced, landscape context, and scale of analysis (Dale et al., 2010). For example, SWRCs that replace cropland will likely increase stand-level diversity, but if SRWCs replace mature forests, native grasslands or other habitat types with high local diversity, then stand-level diversity will likely decrease (Christian et al., 1994; Fletcher et al., 2010). Biodiversity response to these same practices may differ based on the scale of analysis and characteristics of the landscape analyzed.
Bioenergy feedstocks may be derived from forest management systems that vary in their intensity in terms of silvicultural inputs and implications for ecosystems (Burger, 2002). In this review, we compare measures of stand-level biodiversity in SRWCs to forest management settings. We leave discussion of how landscape-level conversions may influence biodiversity to others (e.g., Fletcher et al., 2010). Our analyses were restricted to intensively-managed forests in North America, with a focus on Populus species because information about biodiversity response to other species and other regions is lacking.
Materials and methods
We reviewed the literature for papers that compared biodiversity responses between short-rotation plantations of woody species (e.g., cottonwood, poplar) to other forest types. Response variables included diversity metrics (i.e., species richness, diversity or evenness), total abundance, abundance of taxa or groups of species (guilds), and abundance of individual species. We used Wildlife & Ecology Worldwide and USDA Forest Service TreeSearch databases, supplemented by Google Scholar queries to search for relevant studies. We searched for the following forestry and biodiversity terms (all possible pairwise combinations): willow, eucalyptus, poplar, cottonwood, SRWC, biofuel plantation, biodiversity, diversity, richness, wildlife, birds, avian, amphibians, reptiles, invertebrates, insects and mammals. We supplemented searches by examining bibliographies of articles for additional references.
Because responses can vary greatly among taxa and among species within taxa, we considered different biodiversity measures (e.g., diversity, abundance, richness) from the same study to be independent effects (Bender et al., 1998). For birds, we also considered studies that presented analyses of breeding and winter bird response as separate ‘studies’ because habitat conditions and the behavior, habitat requirements and composition of bird communities can be very different during those two seasons. When studies presented comparisons for a metric in consecutive years, we calculated the overall mean effect and standard deviation using the pooled variance.
We conducted all meta-analyses using meta-win software (Rosenberg et al., 2000). For each response, we calculated a response ratio which is the ratio of the experimental to control groups (Hedges et al., 1999). For each response, we coded the diversity response from the SRWC plantation treatment as the experimental group. Thus, if richness or another biodiversity measure was higher on plantations compared with the ‘control’ comparison (e.g. bottomland hardwoods), response ratios would be >1.00. Ratios <1.00 indicate a negative response to SRWCs. Because some species means were zero, we added 1 to all species means before calculating effect sizes. We used bootstrap confidence intervals and considered a combined effect to be significant if the confidence interval did not include 1.00 (Kalies et al., 2010).
Because each study compared different-aged plantations to different forest types (sometimes very different), each comparison represented a unique set of conditions. Also, some studies compared plantation metrics to more than one additional forest type, so these effect sizes would not be independent because they included data from the same SRWC plantation sites. Thus, we calculated separate effect sizes for each unique plantation (age) vs. reference forest type comparisons for bird metrics only (only one study produced mammal effect sizes). We used fixed effects models for these subgroups because they were comprised of effect sizes from a single study. Also, some studies produced estimates of species richness that were corrected for area sampled (e.g., rarefaction; Christian et al., 1997) while others did not provide information about area sampled. As a result, overall effects sizes likely include much variation and more focused meta-analyses are warranted. Although some of these sub-groups have small sample sizes (and come from one paper), summary effect sizes are still better than vote counting or other forms of summary methods (Borenstein et al., 2009).
Results and discussion
We found 243 effect sizes for birds from seven studies (Fig. 1, Table 1). Cumulative effect size estimates indicated diversity and abundance were lower on SRWC plantations compared with reference woodlands for birds, while abundance of individual species was more variable and not consistently higher or lower on SRWC plantations. Unfortunately, none of these studies directly compared SRWCs to traditional forest plantations, so we cannot predict whether increasing SRWC in these systems will increase or decrease overall forest biodiversity. However, we analyzed comparisons of SRWCs with other forest types, and this provided some evidence for preliminary predictions and guidance for future research. Although these seven studies spanned 26 years, they are still relevant because density and rotation recommendations for biomass applications of SRWC have not substantially changed (Dickmann, 2006).
Table 1. Summary of manipulative studies used in meta-analysis to compare response of bird and mammal communities between short rotation woody crops and other forest types
Species' responses were highly variable, and confidence intervals for response ratios of species' abundances always included 1.00 with one exception (6–9 years cottonwood plantations compared with bottomland hardwood; Fig. 2). Abundance of individual species as a group was not different from 1.00 because some species were less abundant on SRWC plantations (compared with forested lands) while others were more abundant on SRWC plantations. We further investigated species responses by calculating effect sizes for all species with ≥2 individual effect sizes. Seven species were statistically more abundant on SRWC plantations (95% confidence interval did not include 1.00), and seven species were statistically more abundant on reference forests (Table 2). Species that preferred SRWC plantations were species commonly associated with dense, shrubby habitat structure [ruby-throated hummingbird (Archilochus colubris), warbling vireo (Vireo gilvus), yellow-breasted chat (Icteria virens), eastern towhee (Pipilo erythrophthalmus), indigo bunting (Passerina cyanea), orchard oriole (Icterus spurius) and American goldfinch (Spinus tristis) following Poole (2005)].
Table 2. Summary effect sizes for species with ≥2 effect sizes for a meta-analysis comparing response of bird and mammal communities between short rotation woody crops and other forest types
# of Effect Sizes
95% Bootstrap CI
Confidence intervals that do not overlap 1.0 are in bold.
Species associated with reference forests tended to be mature forest associates and/or cavity-nesters [red-headed woodpecker (Melanerpes erythrocephalus), red-bellied woodpecker (Melanerpes carolinus), pileated woodpecker (Dryocopus pileatus), white-eyed vireo (Vireo griseus), American robin (Turdus migratorius), brown thrasher (Toxostoma rufum) and northern parula (Parula americana)]. They likely responded to the lack of vertical structure and canopy in SRWC plantations, especially at earlier stages of growth. SRWC plantations are typically harvested at early ages, and so never develop large stems that provide cavities for hole-nesting birds. Thus, it is not surprising that nearly 50% of species associated with reference woodlands were cavity-nesters. Widespread adoption of SRWC could reduce stand-level density of cavity-nesting species because cavity-nesting birds may nest only at very low densities in SRWC plantations (Twedt et al., 2001). However, nest site availability could be increased using supplemental nest boxes (Twedt & Henne-Kerr, 2001), retaining some large stems in SRWC harvest units, or emphasizing nest sites in less-intensively managed portions of the landscape.
Effect of reference forest habitat on birds
Avian diversity response varied with the type of reference forest used (Fig. 2). Poplar (1–8 years old) and cottonwood (3–9 years old) plantations were less diverse and had lower overall bird abundance than reference forests (Christian et al., 1997; Hanowski et al., 1997; Tomlinson, 1977; Twedt et al., 1999; Wilson & Twedt, 2003) (Fig. 2). Fourteen-year-old cottonwood plantations were also less diverse than bottomland hardwood reference forests (both thinned and unthinned; McComb & Nobel, 1980). In contrast, 14-year-old cottonwood forests had similar avian diversity and abundance compared with unmanaged mixed hardwoods, but higher diversity and abundance than upland hardwood reference forests (both thinned and unthinned; McComb & Nobel, 1980). Two- to 10-year-old cottonwoods were more diverse than similarly aged oak reforestation plantations (Twedt et al., 2002).
Cottonwood plantations may be less diverse than bottomland hardwoods (yet, more diverse than upland hardwoods) in these studies because bottomland hardwoods are located in wetter and more heterogeneous conditions which may result in more diverse communities than upland forests (Drumtra and Cooper, 1999; Heitmeyer et al., 1999). Based on comparisons that were available for our meta-analysis, short-rotation poplar and cottonwood plantations seem to have intermediate avian diversity among other forest types, although comparisons with SRWC plantation forests and studies of other taxa are still lacking.
Effects of age of short-rotation plantations on birds
Younger-aged SRWC plantations tended to be less diverse than reference forests (response ratios <1.00), but older SRWC plantations were relatively more diverse (higher response ratios). We caution that in our dataset, ages are confounded with different reference forest types (and, in some cases, original land cover). However, increasing diversity with age of SRWC plantations has been observed for willow and poplar plantations in Europe (e.g., Sage & Robertson, 1996; Berg, 2002; but see Moser et al., 2002 for the opposite response in mammals). As SRWC plantations age, they grow in height which tends to increase both the vertical structure and heterogeneity of the stands, increasing number of different nesting and foraging substrates. Similarly, Moser & Hilpp (2004) found that wintering owls were most commonly found in the interiors of relatively older poplar plantations. While this relationship makes ecological sense, more research is warranted about how age (i.e., height and heterogeneity) of SRWCs, as well as characteristics of the land they are planted on, influences biodiversity.
Diversity of birds during winter and migration
The extent to which SRWC plantations could provide migration habitat for birds is unknown. Two studies suggest that cottonwood (Wilson & Twedt, 2003) and poplar plantations (Christian et al., 1997) may have lower species richness during migration compared with reference habitat types (bottomland hardwoods and surrounding woodlands, respectively), although differences were not large (Wilson & Twedt, 2003). Additionally, cottonwood plantations supported a distinctly different bird community than bottomland hardwoods, suggesting that some cottonwood plantations in an otherwise forested matrix may enhance overall landscape diversity of birds during migration.
It is also unknown how SRWC plantations influence diversity of winter bird communities. Hamel et al. (2002) compared winter bird communities in former agricultural sites where young forests were established using four different afforestation techniques, including planting of cottonwood stem cuttings followed by underplanting of Nuttall oak (Quercus nuttallii) seedlings 2 years later. Cottonwood plantations had higher species richness than in sites afforested with different methods (natural succession with regeneration control, sown Nuttall oak acorns, planted Nuttall oak seedlings), but similar total number of birds. However, they did not compare cottonwood to other types of forest (e.g., pine plantations, bottomland hardwoods). In Oregon, hybrid poplar plantations in a heavily agricultural matrix may provide suitable winter habitat for owls because their structure approximates that of native forests and they are often proximate to abundant prey in agricultural habitats (Moser & Hilpp, 2004). Because of their dense vegetation and sometimes high abundance of small mammals and other prey species, SRWC plantations have the potential to provide suitable winter habitat for some birds that prefer dense, shrubby vegetation structure.
Effects of landscape context on bird response
In addition to SRWC plantation type and age/structure, and the forest type or land use being replaced, landscape context may contribute to patterns of bird richness and abundance. Poplar plantations in the Upper Midwest had higher bird diversity compared with croplands and pastures, and this difference was greater in landscapes dominated by agricultural, urban or grassland land covers compared with forested landscapes (Christian et al., 1997). But, poplar plantations had lower bird diversity than surrounding wooded areas in both landscapes. Nevertheless, landscape context could potentially mediate biodiversity response to SRWCs in other areas. We expect landscape context to be influential in some regions, although this has not yet been demonstrated for comparisons between SRWCs and other forest types. Further research is warranted to establish if, and to what extent, landscape context might mediate effects of converting production forests to SRWCs.
Bird population responses to SRWC
Higher densities of birds (or higher abundance and higher diversity) does not necessarily indicate superior (or high quality) habitat because densities can often be higher in suboptimal or sink habitats (e.g., Van Horne, 1983; Battin, 2004). Measuring nesting success, fecundity and survivorship in SRWC plantations compared with other forest types is necessary to accurately predict effects of expanding SRWCs in forest systems on bird populations and diversity. Only one study has examined nest success in cottonwood plantations (Twedt et al., 2001). For all species and nests combined, nest success was lower on cottonwood plantations compared with bottomland hardwood forests primarily due to higher nest predation and parasitism rates. However, nest success of five species that were abundant in both cottonwood and bottomland hardwoods did not differ between the two forest types. Thus, lower nest success in cottonwoods may be a function of species-specific differences in nesting habitat and life history, rather than a negative effect of SRWC forestry per se (Twedt et al., 2001). Twedt et al. (2001) did note, however, that fire ants (Solenopis sp.) were more common in cottonwood plantations and were implicated in >11% of nest failures (unlike reference bottomland hardwoods). Nesting success in SRWC plantations relative to other forest types are generally unknown and poorly understood.
Only two studies (Christian et al., 1997; Staten, 1977) provided effect sizes for small mammals. Diversity was higher on cottonwood plantations compared with bottomland hardwoods in Mississippi (Staten, 1977), but was lower on cottonwood plantations relative to surrounding wooded habitat in the upper Midwest US (Christian et al., 1997). The meta-analysis for diversity was close to 1.00 but with wide confidence intervals. Abundance and species' abundance were consistently lower in SWRC (Fig. 1). Diversity and abundance of mice (Christian et al., 1997; Staten, 1977) and shrews (Christian et al., 1997) were consistently lower on poplar plantations compared with surrounding woodlands. As with birds, these did not compare SRWCs to intensively managed forests, so predicting effects of converting intensively managed timber forests to SRWC plantations is difficult.
For small mammals, other studies provide additional, but contrasting, information that does little to clarify the situation. Poplar plantations in the midwestern US had lower winter densities of lagomorphs compared with surrounding forest, and other forest mammals were rarely detected using poplar plantations (Christian, 1997). In contrast, rabbit (Silvilagus spp.) densities were greater on cottonwood plantations compared with surrounding forests in the Mississippi Alluvial Valley (Staten, 1977; Wesley et al., 1981). Cottonwood plantations elsewhere in the MAV contained a similar number of species to reference forests, but total abundance was intermediate between bottomland hardwoods (lower than cottonwood) and upland hardwoods (higher that cottonwood; McComb & Nobel, 1980). Additionally, diversity and abundance of small mammals in SRWC plantations may decline with stand age (Moser et al., 2002), further complicating the task of understanding biodiversity response.
Little is known about how large mammals are affected by SRWC plantations. White-tailed deer (Odocoileus virginianus) appeared to use poplar plantations no differently than other forest types in the midwestern US (Christian, 1997), but were reported to favor cottonwood plantations during parturition in the southern US (Wigley et al., 1980; Wesley et al., 1981). Browse damage to SRWC plantations by ungulates has been reported (Christian, 1997) indicating that SRWC trees and other vegetation may be suitable forage, but this has not been quantified. Clearly, additional research is needed to guide management.
Because we found only eight studies that represented nine comparisons of different SRWC plantation types (different ages of cottonwood or poplar) to very different types of reference forest, our conclusions are tentative. Diversity and abundance of birds and mammals were lower on SRWC plantations compared with reference forests. Species responses were mixed – SRWC plantations favored shrub-associated birds while canopy/mature forest-associated birds and cavity-nesters typically favored reference forest. SRWC plantations could likely contribute to overall landscape diversity in forest-dominated landscapes by providing shrubby habitat structure for non-forest species. For example, recommendations for maximizing wildlife habitat in the Mississippi Alluvial Valley call for ≤5% of the landscape to be comprised of shrub/scrub habitats, which SRWC plantations could help provide (Wilson et al., 2007). However, extensive conversion of intensively managed forests to SRWC would likely decrease overall diversity, especially if they replace high conservation value habitats (Archaux & Martin, 2009).
Although most SRWCs will likely be established on agricultural land (e.g. Christian et al., 1994; Sage et al., 2006; Fletcher et al., 2010; Rowe et al, 2009), comparing diversity of SRWC to other types of forests is important for several reasons. First, SWRCs are sometimes planted on forested lands (e.g., Auclair and Bouvarel, 1992; Weih, 2004) and SRWC plantations may become a more prominent component of forested landscapes. Second, much of the agricultural land potentially used for SRWC may have been forested in the past, especially marginal cropland on which SRWC may be most profitable. Third, SRWCs are one of many options for afforestation efforts (e.g., Twedt, 2006). Fourth, SRWC plantations and other forest types are but two of many land cover options when croplands are converted (e.g. croplands converted to SRWCs are lands that can no longer be converted to forest) and in some regions marginal cropland is commonly converted to and from forest cover (Wear et al., 2007). Thus, it is impossible to fully understand how SWRCs will increase or decrease diversity at the landscape and regional scales without SRWC vs. forest comparisons.
Geographic limitations, empirical knowledge gaps and research needs
Burger (2002) considered SRWC systems to be intermediate to forest plantations and agrosystems in terms of management inputs and implications for biodiversity. It is unknown how SRWCs differ from traditional forest plantations because existing studies did not use intensively managed forests as reference. Such comparisons need to be made because intensively managed forests cover increasing large areas in many regions, provide habitat for species of concern, and support major components of biodiversity (e.g., Carnus et al., 2006; Stephens & Wagner, 2007; Brockerhoff et al., 2008). Future studies should also compare SRWC plantations to reference forests in a similar seral stage (e.g., Twedt et al., 2002) in addition to more mature, reference forests (the other studies we reviewed). These studies should be continued over the length of the rotation. Present information about diversity in SRWC plantations in the United States is limited to the upper Midwest and the Mississippi Alluvial Valley, and few studies exist for taxa other than birds. Rectifying this lack of basic information should be the top priority for research concerning diversity and SRWCs.
We thank J. Homyack, A. Kroll, and J. Martin for helpful reviews of this manuscript.