Complexity is costly: a meta‐analysis of parametric and non‐parametric methods for short‐term population forecasting

Short‐term forecasts based on time series of counts or survey data are widely used in population biology to provide advice concerning the management, harvest and conservation of natural populations. A common approach to produce these forecasts uses time‐series models, of different types, fit to time series of counts. Similar time‐series models are used in many other disciplines, however relative to the data available in these other disciplines, population data are often unusually short and noisy and models that perform well for data from other disciplines may not be appropriate for population data. In order to study the performance of time‐series forecasting models for natural animal population data, we assembled 2379 time series of vertebrate population indices from actual surveys. Our data were comprised of three vastly different types: highly variable (marine fish productivity), strongly cyclic (adult salmon counts), and small variance but long‐memory (bird and mammal counts). We tested the predictive performance of 49 different forecasting models grouped into three broad classes: autoregressive time‐series models, non‐linear regression‐type models and non‐parametric time‐series models. Low‐dimensional parametric autoregressive models gave the most accurate forecasts across a wide range of taxa; the most accurate model was one that simply treated the most recent observation as the forecast. More complex parametric and non‐parametric models performed worse, except when applied to highly cyclic species. Across taxa, certain life history characteristics were correlated with lower forecast error; specifically, we found that better forecasts were correlated with attributes of slow growing species: large maximum age and size for fishes and high trophic level for birds. Synthesis Evaluating the data support for multiple plausible models has been an integral focus of many ecological analyses. However, the most commonly used tools to quantify support have weighted models’ hindcasting and forecasting abilities. For many applications, predicting the past may be of little interest. Concentrating only on the future predictive performance of time series models, we performed a forecasting competition among many different kinds of statistical models, applying each to many different kinds of vertebrate time series of population abundance. Low‐dimensional (simple) models performed well overall, but more complex models did slightly better when applied to time series of cyclic species (e.g. salmon).     

Online Respondent-Driven Sampling for Studying Contact Patterns Relevant for the Spread of Close-Contact Pathogens: A Pilot Study in Thailand

Background

Information on social interactions is needed to understand the spread of airborne infections through a population. Previous studies mostly collected egocentric information of independent respondents with self-reported information about contacts. Respondent-driven sampling (RDS) is a sampling technique allowing respondents to recruit contacts from their social network. We explored the feasibility of webRDS for studying contact patterns relevant for the spread of respiratory pathogens.

Materials and Methods

We developed a webRDS system for facilitating and tracking recruitment by Facebook and email. One-day diary surveys were conducted by applying webRDS among a convenience sample of Thai students. Students were asked to record numbers of contacts at different settings and self-reported influenza-like-illness symptoms, and to recruit four contacts whom they had met in the previous week. Contacts were asked to do the same to create a network tree of socially connected individuals. Correlations between linked individuals were analysed to investigate assortativity within networks.

Results

We reached up to 6 waves of contacts of initial respondents, using only non-material incentives. Forty-four (23.0%) of the initially approached students recruited one or more contacts. In total 257 persons participated, of which 168 (65.4%) were recruited by others. Facebook was the most popular recruitment option (45.1%). Strong assortative mixing was seen by age, gender and education, indicating a tendency of respondents to connect to contacts with similar characteristics. Random mixing was seen by reported number of daily contacts.

Conclusions

Despite methodological challenges (e.g. clustering among respondents and their contacts), applying RDS provides new insights in mixing patterns relevant for close-contact infections in real-world networks. Such information increases our knowledge of the transmission of respiratory infections within populations and can be used to improve existing modelling approaches. It is worthwhile to further develop and explore webRDS for the detection of clusters of respiratory symptoms in social networks.     

Implementation of Web-Based Respondent-Driven Sampling among Men Who Have Sex with Men in Vietnam

Objective

Lack of representative data about hidden groups, like men who have sex with men (MSM), hinders an evidence-based response to the HIV epidemics. Respondent-driven sampling (RDS) was developed to overcome sampling challenges in studies of populations like MSM for which sampling frames are absent. Internet-based RDS (webRDS) can potentially circumvent limitations of the original RDS method. We aimed to implement and evaluate webRDS among a hidden population.

Methods and Design

This cross-sectional study took place 18 February to 12 April, 2011 among MSM in Vietnam. Inclusion criteria were men, aged 18 and above, who had ever had sex with another man and were living in Vietnam. Participants were invited by an MSM friend, logged in, and answered a survey. Participants could recruit up to four MSM friends. We evaluated the system by its success in generating sustained recruitment and the degree to which the sample compositions stabilized with increasing sample size.

Results

Twenty starting participants generated 676 participants over 24 recruitment waves. Analyses did not show evidence of bias due to ineligible participation. Estimated mean age was 22 years and 82% came from the two large metropolitan areas. 32 out of 63 provinces were represented. The median number of sexual partners during the last six months was two. The sample composition stabilized well for 16 out of 17 variables.

Conclusion

Results indicate that webRDS could be implemented at a low cost among Internet-using MSM in Vietnam. WebRDS may be a promising method for sampling of Internet-using MSM and other hidden groups.     

Predictors of Inconsistent Condom Use among a Hard to Reach Population of Young Women with Multiple Sexual Partners in Peri-Urban South Africa

Background

Evidence suggests that multiple concurrent sexual partnering may be a key driver of the high HIV prevalence among young women in South Africa. However, little is known about whether and to what extent women who have multiple sexual partners also engage in other high risk sexual behaviors such as inconsistent condom use. And yet, multiple concurrent sexual partnering is of little epidemiological relevance if all partners in these sexual networks use condoms consistently. This study assesses the prevalence of sexual risk behaviors and HIV, and predictors of inconsistent condom use among women aged 16–24 with multiple sexual partners in a peri-urban setting in South Africa.

Methods

We used Respondent Driven Sampling, a sampling strategy for hard-to-reach populations to recruit 259 women aged 16–24 in a bio-behavioral cross-sectional survey in the Western Cape province. Estimates of population proportions and 95% confidence intervals (CIs) were calculated using the Respondent-Driven Sampling Analysis Tool 5.6 (RDSAT). The primary outcome was inconsistent condom use in the past three months.

Results

Young women reported an average of 7 partners in the past 3 months and a high prevalence of sexual risk behaviors: concurrency (87%), transactional sex (91%) and age mixing (59%). Having >5 sexual partners in the last 3 months doubled the risk of unprotected sex (OR 2.43, CI 1.39–4.25). HIV prevalence was 4% among 16–19 year olds, increasing threefold (12%) at age 20–24.

Discussion

Multiple sexual partnering, where a high number of partners are acquired in a short space of time, is a fertile context for unprotected and risky sexual behavior. The young women featured in this survey present with a constellation of high-risk sexual behaviors that cluster to form a risk syndrome. Carefully tailored repeat bio-behavioral surveillance surveys are recommended for this sub-population.     

Microsatellite evolution in congeneric mammals: domestic and bighorn sheep

We compared genotypes at eight (AC)n microsatellite loci in domestic sheep (Ovis aries) and wild Rocky Mountain bighorn sheep (O. canadensis). The domestic sheep had greater genetic variation, higher allele-size variances, and larger allele sizes than the wild sheep. Accumulating evidence from higher taxonomic comparisons shows that these parameters are biased if microsatellite loci are selected in one taxon and used in another. Our results demonstrate similar biases between congeneric species. We compared standard measures of genetic variation, differentiation, and distance within and between species (H, D, FST) to newer measures based on allele-size variance (SW, SB, RST). The size-based distances better detected species-level divergence, but standard measures better distinguished allopatric populations. Empirical calibration of these measures at the subspecies level is needed to establish their useful ranges.      

Early events in the cellular formation of proparathyroid hormone

Early events in the cellular synthesis and subsequent transfer into membrane-limited compartments of pre-proparathyroid hormone (pre-proPTH) and proparathyroid hormone (proPTH) were investigated by electrophoretic analyses of newly synthesized proteins in subcellular fractions of parthyroid gland slices pulse-labeled for 0.5-5 min with [(35)S] methionine. During these short times of incubation, both pre-proPTH and proPTH were confined to the microsomal fraction. Labeled pre-proPTH and proPTH were detected in a 30-s interval between 0.5 and 1.0 min of incubation. The radioactivity in proPTH became relatively constant between 3 and 5 min, whereas the radioactivity in ProPTH increased markedly over this period. When corrected for the known content of methionine in the prohormone and the prohormone, we found four times as much radiolabeled prohormone as prehormone between 0.5 and 1.0 min of synthesis. Sequestration of labeled prohomrone into endoplasmic reticulum compartments was shown by treatment of the microsomal fraction with chymotrypsin and trypsin, which resulted in the degradation of the prehormone but not of the prohormones. Approximately 50 percent of pre-prohormone and 25 percent of prohormone were released from the microsomes by their extraction with 1.0 M KCl, whereas 80-90 percent of both was released by treatment with Triton X-100. These results in intact cells support the signal hypothesis proposed by Blobel and his co-workers in studies utilizing cell-free systems, inasmuch as the results indicate transfer of prohormone into the cisternal space of the rough endoplasmic reticulum concomitant with the growth of the nascent polypeptide chain. Appearance of membrane-sequestered proPTH takes place without entry of pre-proPTH into the cisternal space, suggesting that proteolytic removal of the leader peptide occurs during transfer of the polypeptide through the lipid bilayer. Further evidence in support of this process is that pre-proPTH is only partly extracted from the microsomes by treatment with 1.0 M KCl, suggesting that a substantial fraction of the nascent pre-proPTH is integrally inserted into the membranes before it is cleaved to form proPTH.