Logs and other woody debris on the forest floor are an essential habitat for a wide range of saproxylic arthropods [1,2,3], which are directly or indirectly dependent on dead and dying wood [4]. These include not only phloem and wood feeders but also their predators and species associated with wood-decaying fungi. Many non-saproxylic ground-dwelling arthropods benefit from dead wood as well. For instance, a number of studies have shown litter-dwelling arthropods and other invertebrates to be more numerous immediately next to dead wood than short distances away from it [5,6,7,8,9,10,11,12,13,14,15,16,17]. These results can be attributed variously to dead wood providing a relatively stable source of moisture [18,19,20], shelter from predators or adverse weather conditions, and an abundance of prey or other food items.

With demand for forest products continuing to rise (especially now with a growing interest in biofuel production, see [21,22]), understanding the role of dead wood in maintaining biodiversity, tree productivity and long-term forest sustainability is becoming more important in forest management. Of particular value are studies investigating the effects of dead wood manipulations on these interests. Several manipulative studies have demonstrated the positive relationship between dead wood and ground-dwelling arthropod diversity [23,24,25], but the question is far from resolved and has received little attention in many forest types.

The forests of the southeastern United States are among the most productive in North America with loblolly pine (Pinus taeda L.) being the single most important timber species in the region. One of the few efforts to measure arthropod response to dead wood manipulations made in loblolly pine plantations began on the Savannah River Site in South Carolina in 1996. In the first phase of that study, plots in which all dead wood was removed annually were compared to reference plots from which no dead wood was removed. After pitfall-sampling ground-dwelling arthropods in those plots for five years, Hanula et al. [26] found no differences in overall abundance or morphospecies richness. Overall arthropod diversity and evenness were significantly lower in removal plots than in reference plots, however, and several families differed in abundance between the two treatments. These differences were observed only in the first two full years of sampling, however. In the second phase of the study, beginning in 2001, arthropods were sampled in these same plots for four more years. Two new treatments were added in which sampling also took place; one involving major inputs of logs to the forest floor and the other involving major inputs of dead standing trees. Contrary to expectations, but largely consistent with the first phase of the study, there were no differences in total ground-dwelling arthropod abundance, richness, diversity or composition among treatments [27]. Only ground beetles (family Carabidae) exhibited a significant association with dead wood.

During 1998, a study was installed on four widely-separated USDA Long Term Soil Productivity (LTSP) sites in the Southeastern United States to better understand the importance of dead wood to nutrient cycling and biodiversity in pine forests. The responses of beetles, the most diverse arthropods on the forest floor, are reported here.

This study took place at the following four LTSP sites: Croatan National Forest, Craven Country, NC, USA; DeSoto National Forest, Jones Country, MS, USA; Kisatchie National Forest, Rapides Parish, LA, USA; Davy Crockett National Forest, Trinity Country, TX, USA. The forests were dominated by 50–60 yr pine [loblolly in LA and NC, slash (P. elliottii Engelm.) in MS, and loblolly and shortleaf (P. echinata Mill.) in TX] with an understory consisting of oak (Quercus sp.), hickory (Carya sp.), and sweetgum (Liquidambar styraciflua L.). For this experiment, the coarse woody debris was added to remnant mature pine stand, adjacent to the LTSP treatment plots. At each site, three of the mature stand plots were used as controls and contained only the existing woody debris already present or that which fell during the study. A 30 × 45 m rectangular area at each site was divided into six 15 × 15 m plots. The dead wood added to all of the plots came from the same 43-year-old loblolly pine stand in the Palustris Experimental Forest, Louisiana. Sections of wood were removed from the main trunk (i.e., “logs”, 30–35 cm dbh, 1 m long), limbs (2.5–10 cm mid-length diam, 0.5 m long), and twigs (0.5–1.5 cm diam, 0.25 m long) of loblolly pine trees felled for this purpose. Ten logs, 8 limbs, and 24 twigs were distributed across the three randomly selected dead wood plots at each site in April and May 1998. At each given site, the material was added all at once, on one date during the time frame mentioned above. At all sites, the amount of existing coarse woody debris and the vegetation on the ground were roughly the same. Our additions of material impacted the sites in a very similar manner, adding the above material to the baseline of that already onsite.

One pitfall trap (modified from [28]) was installed in each plot to sample ground-dwelling beetles. At each site, the trap was placed arbitrarily within the plot, in the immediate area of the added dead wood. Each trap consisted of a 15-cm-diameter plastic funnel fastened beneath a hole in the center of a 30 × 30 cm section of ~1.9 cm-thick plywood with bevelled edges. This assembly was positioned over a 2-liter plastic container, half-filled with a 1:1 mixture of propylene glycol and 95% ethanol, buried at ground level. Samples were collected approximately biweekly from May through September at all locations in 1998 and at all locations except for North Carolina in 1999. Additional samples were collected in all locations except for North Carolina in April and May of 2002.

Specimens not identified below family level (117 individuals were only identified to family level) were excluded from the dataset. After pooling all sampling periods for a given year, analyses of variance (ANOVA) were carried out for each year separately (due to differences in sampling intensity) to determine whether beetle abundance (log(x + 1)-transformed to achieve normality), richness, or Shannon’s diversity varied between treatments. Site location (= “State”) was included in the model as a blocking variable and was treated as a random effect. To provide additional information, separate analyses were performed on the following three groups: saproxylic (species directly or indirectly dependent on dead or dying wood), non-saproxylic ground beetles, and other species. These designations were based on familiarity with the taxa and information available in the literature. Voucher specimens have been deposited in the Louisiana State Arthropod Museum, Baton Rouge, Louisiana.

The final dataset consisted of 5172 beetle specimens belonging to 55 families and 378 taxa identified to genus or species (all of which are hereafter referred to as “species”, see Appendix). Of these, 47 (13%) were non-saproxylic ground beetles, 182 (48%) were saproxylic, and 149 (39%) were other species (Appendix). Texas traps yielded the most beetles in terms of both individuals (~44% of the total) and species (~60%) whereas Louisiana traps yielded the fewest (see Appendix; note that North Carolina traps were only sampled in 1998). State (block) was found to be a significant source of variation in most analyses of overall beetle richness, abundance and diversity in 1998 and 1999 but less so in 2002 (Table 1). The same was true when saproxylic species, ground beetles (which are non-saproxylic), and other species were analyzed separately (Table 1)

With respect to treatment differences, 2998 (58%) individuals and 302 (80%) species were collected from the dead wood plots compared to 2174 (42%) individuals and 261 (69%) species from the reference plots. Furthermore, 115 and 74 species were collected from the dead wood and reference plots, respectively. Overall, beetles were consistently more species rich, abundant, and diverse in the dead wood plots than in the reference plots during the entire study (Figure 1). Although these differences were greatest in 1998, significant differences were detected only in 1999 and only for diversity (Table 1). Similarly, only in 1999 were significant differences detected between treatments for saproxylic beetles, with them being more rich and diverse in dead wood plots than in references plots. No significant differences were detected in 2002 and non-saproxylic ground beetles and other beetle species, when analyzed separately, did not vary between treatments in any year (Table 1).

The large differences in mean beetle richness, abundance and diversity between treatments in 1998 (Figure 1) were not statistically significant due to strong interactions between state and treatment (Table 1). Most notably, the results from North Carolina were not consistent with those from the other states (Figure 2, Figure 3, Figure 4). After excluding North Carolina from the dataset, the differences in overall beetle richness (F_1,2 = 19.4, p < 0.05), saproxylic beetle richness (F_1,2 = 23.8, p < 0.05) and saproxylic beetle diversity (F_1,2 = 20.1, p < 0.05) observed in 1998 were found to be significant.

Despite conflicting results from North Carolina which we are unable to explain based on this study, beetles generally responded positively to the dead wood manipulations made in this study, particularly during the first year. Saproxylic species were more abundant, whereas non-saproxylic ground beetles and other species did not noticeably differ. The increased abundance of saproxylic species is not surprising as these organisms, by definition, are dependent on dying and dead wood [4]. Many were no doubt captured as they attempted to colonize the fresh wood added to the plots. It is noteworthy that the degree to which saproxylic species were more numerous in the dead wood plots declined after the first year. This is largely due to the fact that more species are associated with the phloem of freshly-killed loblolly pine wood than later decay stages [29]. Once the phloem is gone, which happens quickly, so too are these early colonists. Ulyshen and Hanula [29] provided a list of beetle species found to be specifically associated with freshly-killed loblolly pine in South Carolina. Of those, the following genera were collected only during 1998 (the year the wood was added to the plots) in the current study: the cerambycids Acanthocinus and Monochamus; the zopherids Colydium and Lasconotus; the curculionids Dendroctonus, Gnathotrichus, Ips, Orthotomicus, and Myoplatypus; the tenebrionid Corticeus, the staphylinid Myrmecocephalus, and the histerid Platysoma (results not shown). A wide variety of other species colonize loblolly pine after the phloem stage but not in such great numbers [29]. Furthermore, many of these species are likely to re-infest the same piece of wood until it no longer provides a suitable substrate [30], thereby reducing the likelihood of being captured by pitfall traps. While the ethanol used in the trap collection jars may have attracted insects, these effects would seem to have been equal between controls and treatments, and the added wood apparently overcame any confounding effects of using this preservative.

That ground beetles were not more strongly affected by the dead wood additions is somewhat surprising considering that many previous studies have shown this group to be positively associated with dead wood [26,27,31,32,33,34,35]. Our results, therefore, provide little support for the idea that epigeic predators are more likely than many non-saproxylic taxa to benefit from dead wood due to its positive effect on prey abundance [27]. It is possible however that the debris in the dead wood plots impeded carabid movement enough to reduce their capture rate, thereby masking a stronger beneficial effect. Furthermore, it appears that ground beetles were becoming increasingly numerous in the dead wood plots relative to the reference plots over the course of this study (see Figure 1). Although the differences were not statistically significant, ground beetle richness and diversity had non-overlapping standard errors in 2002. These findings suggest that the benefits of dead wood to these taxa may be somewhat delayed. The results from other studies are not consistent with this conclusion, however. Within the first year after sites were clearcut in Sweden, for example, Nittérus and Gunnarsson [34] collected significantly more ground beetles in pitfall traps placed under piles of slash compared to those placed out in the open. Perhaps dead wood represents a more important source of shelter from sunlight and other desiccating conditions in clearcuts than in forested plots such as those used in the current study.

While some degree of caution is advised in interpreting the statistical tests we report here (due to the relatively close proximity of the sites to one another), our results are largely consistent with those of Hanula et al. [26] and Ulyshen and Hanula [27]. These studies suggest that ground-dwelling arthropods (saproxylic species notwithstanding) are little-affected by manipulations of dead wood in loblolly pine forests. This conclusion, however, is in conflict with evidence that many ground-dwelling arthropod taxa are more numerous immediately next to loblolly pine logs than short distances away from them [15]. As suggested by Ulyshen and Hanula [27], this discrepancy may indicate that dead wood has a stronger effect on how ground-dwelling arthropods are spatially distributed (i.e., causing them to become more clumped) than on their abundance or species richness. Studies aimed at addressing this question would be of particular interest. More research is also needed to determine how manipulations of dead wood at larger scales and over longer time periods affect these organisms.